(Medicalstudyzone - Com) Paul's Fundamental Immunology PDF

Www.Medicalstudyzone.
com
Fundamental
Immunology
SEVENTH EDITION
Paul_FM_final.indd i 9/17/12 3:25 PM

Www.Medicalstudyzone.com
This PDF was created and uploaded by

www.medicalstudyzone.com which is one the
biggest free resources platform for medical
students and healthcare professionals. You can
access all medical Video Lectures, Books in PDF
Format or kindle Edition, Paid Medical Apps and
Softwares, Qbanks, Audio Lectures And Much
More Absolutely for Free By visiting our Website
https://medicalstudyzone.com all stuff are free
with no cost at all.
Furthermore You can also request a specific

Book In PDF Format OR Medical Video Lectures.
Paul_FM_final.indd ii 9/17/12 3:25 PM

Fundamental
Immunology
SEVENTH EDITION
EDITOR
William E. Paul, MD
Paul_FM_final.indd iii 9/17/12 3:25 PM

Acquisitions Editor: Julie Goolsby

Product Manager: Leanne Vandetty
Senior Manufacturing Manager: Benjamin Rivera
Marketing Manager: Kimberly Schonberger
Design Coordinator: Teresa Mallon
Production Service: Absolute Service, Inc.
© 2013 by LIPPINCOTT WILLIAMS & WILKINS, a WOLTERS KLUWER business
Two Commerce Square

2001 Market Street
Philadelphia, PA 19103 USA
LWW.com
6th Edition, © 2008 by LIPPINCOTT WILLIAMS & WILKINS, a WOLTERS KLUWER business
5th Edition, © 2003 by Lippincott Williams & Wilkins
4th Edition, © 1999 by Lippincott-Raven Publishers
All rights reserved. This book is protected by copyright. No part of this book may be reproduced
in any form by any means, including photocopying, or utilized by any information storage and
retrieval system without written permission from the copyright owner, except for brief quotations
embodied in critical articles and reviews. Materials appearing in this book prepared by individu-
als as part of their official duties as U.S. government employees are not covered by the above-
mentioned copyright.
Printed in China
Library of Congress Cataloging-in-Publication Data
Fundamental immunology / editor, William E. Paul. — 7th ed.

p. ; cm.
Includes bibliographical references and index.
ISBN 978-1-4511-1783-7 (alk. paper) — ISBN 1-4511-1783-3 (alk. paper)
I. Paul, William E.
[DNLM: 1. Immune System Phenomena. QW 540]
616.07’9—dc23
2012035820
Care has been taken to confi rm the accuracy of the information presented and to describe gener-
ally accepted practices. However, the authors, editors, and publisher are not responsible for errors
or omissions or for any consequences from application of the information in this book and make
no warranty, expressed or implied, with respect to the currency, completeness, or accuracy of the
contents of the publication. Application of the information in a particular situation remains the
professional responsibility of the practitioner.
The authors, editors, and publisher have exerted every effort to ensure that drug selection and
dosage set forth in this text are in accordance with current recommendations and practice at the
time of publication. However, in view of ongoing research, changes in government regulations, and
the constant flow of information relating to drug therapy and drug reactions, the reader is urged
to check the package insert for each drug for any change in indications and dosage and for added
warnings and precautions. This is particularly important when the recommended agent is a new
or infrequently employed drug.
Some drugs and medical devices presented in the publication have Food and Drug
Administration (FDA) clearance for limited use in restricted research settings. It is the responsibil-
ity of the health care provider to ascertain the FDA status of each drug or device planned for use
in their clinical practice.
To purchase additional copies of this book, call our customer service department at (800) 638-3030
or fax orders to (301) 223-2320. International customers should call (301) 223-2300.
Visit Lippincott Williams & Wilkins on the Internet: at LWW.com. Lippincott Williams & Wilkins
customer service representatives are available from 8:30 am to 6 pm, EST.
10 9 8 7 6 5 4 3 2 1
Paul_FM_final.indd iv 9/27/12 12:50 AM

For Charlotte, Gloria, Lucy, Jenna, Silvie, and Jake—and for Julien
Paul_FM_final.indd v 9/17/12 3:25 PM

Paul_FM_final.indd vi 9/17/12 3:25 PM

CHAPTER 36 COMPLEMENT | vii
CONTRIBUTING AUTHORS
Eitan M. Akirav, PhD Jean-Laurent Casanova, MD, PhD

Assistant Professor of Research Professor and Head of Laboratory
Department of Medicine St. Giles Laboratory of Human Genetics of Infectious
Stony Brook Medical School Diseases
Stony Brook, New York The Rockefeller University
Research Scientist Senior Attending Physician
Biomedical Research Core The Rockefeller University Hospital
Winthrop University Hospital New York, New York
Mineola, New York
Marco A. Cassatella, MD
Noelia Alonso Gonzalez, PhD Professor
Postdoctoral Fellow Pathology and Diagnostics
Department of Epidemiology and Public Health Verona University Medical School
Yale University Vice Director
New Haven, Connecticut General Pathology
Medical School
Yasmine Belkaid, PhD Verona, Italy
Chief, Mucosal Immunology Section
Laboratory of Parasitic Diseases Ameya S. Champhekar, PhD
National Institute of Allergy and Infectious Diseases Postdoctoral Scholar
National Institutes of Health Division of Biology
Bethesda, Maryland California Institute of Technology
Pasadena, California
Albert S. Bendelac, MD, PhD
Professor Akanksha Chaturvedi, PhD
Department of Pathology Postdoctoral Fellow
Howard Hughes Medical Institute Laboratory of Immunogenetics
University of Chicago NIH, NIAID
Chicago, Illinois Rockville, Maryland
Ira Berkower, MD, PhD Sangeeta S. Chavan, PhD

Lab Chief, Lab of Immunoregulation Associate Investigator
Division of Viral Products, Office of Vaccines Center for Biomedical Science
Center for Biologics Evaluation and Research The Feinstein Institute for Medical Research
US Food and Drug Administration Manhasset, New York
Bethesda, Maryland
Yueh-hsiu Chien, PhD
Sylvie Bertholet, PhD Professor
Unit Head Department of Microbiology and Immunology
Immunology Function Stanford University
Novartis Vaccines and Diagnostics Stanford, California
Siena, Italy
Mary Ellen Conley, MD
Jay A. Berzofsky, MD, PhD Professor
Branch Chief Department of Pediatrics
Vaccine Branch, Center for Cancer Research University of Tennessee
National Cancer Institute, National Institutes of Health Le Bonheur Children’s Hospital
Bethesda, Maryland Memphis, Tennessee
vii
Paul_FM_final.indd vii 9/17/12 3:25 PM

viii | CONTRIBUTING AUTHORS
Ken T. Coppieters Martin F. Flajnik, PhD

La Jolla Institute for Allergy and Immunology Professor
La Jolla, California Department of Microbiology and Immunology
University of Maryland Baltimore
Shane Crotty, PhD Baltimore, Maryland
Associate Professor
Division of Vaccine Discovery Sebastian Fugmann, PhD
La Jolla Institute for Allergy and Immunology Investigator
La Jolla, California Laboratory of Molecular Biology and Immunology
National Institute on Aging/NIH
Angel Davey, PhD Baltimore, Maryland
Postdoctoral Fellow
Laboratory of Immunogenetics Christopher C. Goodnow, BVSc, BScVet, PhD
NIH, NIAID Professor & NHMRC Australia Fellow
Rockville, Maryland Department of Immunology
John Curtin School of Medical Research
Mark M. Davis, PhD The Australian National University
Investigator, HHMI Canberra, Australian Capital Territory
Department of Microbiology and Immunology
Stanford University Siamon Gordon, FRS, MB, ChB, PhD
Stanford, California Emeritus Professor
Sir William Dunn School of Pathology
Ennio De Gregorio, PhD University of Oxford
Head Oxford, United Kingdom
Department of Immunology
Novartis Vaccines and Diagnostics Steven Greenberg, MD
Siena, Italy Senior Principal Scientist
Respiratory and Immunology Department
Charles A. Dinarello, MD Merck Research Laboratories
Professor of Medicine and Immunology Kenilworth, New Jersey
University of Colorado Denver Adjunct Associate Professor
Aurora, Colorado Department of Medicine
Professor of Experimental Medicine Columbia University College of Physicians and Surgeons
Department of Medicine New York, New York
Radboud University
Nijmegen, The Netherlands Neil S. Greenspan, MD, PhD
Professor
Anca Dorhoi Department of Pathology
Department of Immunology Case Western Reserve University
Max Planck Institute for Infection Biology Director, Histocompatibility and Immunogenetics
Berlin, Germany Laboratory
University Hospitals Case Medical Center
Ugo D’Oro, MD, PhD Cleveland, Ohio
Unit Head
Department of Immunology Function Ted H. Hansen, PhD
Novartis Vaccines and Diagnostics Professor
Siena, Italy Department of Pathology
Washington University School of Medicine
Louis Du Pasquier, PhD St. Louis, Missouri
Professor
Department of Zoology and Evolutionary Biology Richard R. Hardy, PhD
University of Basel Professor
Basel, Switzerland Fox Chase Cancer Center
Philadelphia, Pennsylvania
Hildegund C.J. Ertl, MD
Caspar Wistar Professor in Vaccine Research Dirk Homann
Professor and Program Leader, Immunology Program Department of Anesthesiology
Director, The Wistar Institute Vaccine Center Barbara Davis Center for Childhood Diabetes
The Wistar Institute Aurora, Colorado
Philadelphia, Pennsylvania University of Colorado, Denver
Paul_FM_final.indd viii 9/27/12 12:51 AM

CONTRIBUTING AUTHORS | ix
Marc K. Jenkins, PhD Judy Lieberman, MD, PhD

Distinguished McKnight Professor Senior Investigator
Department of Microbiology and Center for Immunology Immune Disease Institute
University of Minnesota Professor
Minneapolis, Minnesota Department of Pediatrics
Harvard Medical School
Susan M. Kaech, PhD Boston, Massachusetts
Associate Professor and HHMI Early Career Scientist
Department of Immunobiology Kang Liu, PhD
Yale University Assistant Professor
New Haven, Connecticut Department of Microbiology and Immunology
Columbia University Medical Center
Jannet Katz, DDS, PhD New York, New York
Professor
Department of Pediatric Dentistry Wanli Liu, PhD
University of Alabama at Birmingham Postdoctoral Fellow
Birmingham, Alabama Laboratory of Immunogenetics
NIH, NIAID
Stefan H.E. Kaufmann, PhD Rockville, Maryland
Director
Department of Immunology David H. Margulies, MD, PhD
Max Planck Institute for Infection Biology Chief, Molecular Biology Section
Professor Laboratory of Immunology
Charite University Hospital National Institute of Allergy and Infectious Diseases
Berlin, Germany National Institutes of Health
Bethesda, Maryland
Douglas S. Kwon, MD, PhD
Instructor in Medicine Edward E. Max, MD, PhD
Department of Infectious Disease Associate Director for Research
Harvard Medical School Office of Biotechnology Products
Associate Physician Center for Drugs Evaluation and Research
Ragon Institute of MGH, MIT and Harvard Food and Drug Administration
Massachusetts General Hospital Bethesda, Maryland
Boston, Massachusetts
James McCluskey, MD, FRACP, FRCPA, FAA
Michael J. Lenardo, MD Professor
Senior Investigator Department of Microbiology and Immunology
Laboratory of Immunology University of Melbourne
National Institutes of Health Melbourne, Victoria, Australia
Bethesda, Maryland
Michael G. McHeyzer-Williams, PhD
Warren J. Leonard, MD Professor
Chief, Lab of Molecular Immunology Department of Immunology & Microbial Science
Director, Immunology Center The Scripps Research Institute
National Heart, Lung, and Blood Institute La Jolla, California
National Institutes of Health
Bethesda, Maryland B. Paul Morgan, MB, PhD, MRCP, FRCPath
Dean and Head of School
Ian Paul Lewkowich, PhD School of Medicine
Assistant Professor Cardiff University
Department of Cellular and Molecular Immunology Honorary Consultant in Medical Biochemistry
Cincinnati Children’s Hospital Department of Medical Biochemistry & Immunology
Cincinnati, Ohio University Hospital of Wales
Cardiff, United Kingdom
Paul_FM_final.indd ix 9/17/12 3:25 PM

x | CONTRIBUTING AUTHORS
Philip M. Murphy, MD Luke A.J. O’Neill, PhD

Chief Professor of Biochemistry
Laboratory of Molecular Immunology School of Biochemistry and Immunology
National Institute of Allergy and Infectious Diseases Trinity Biomedical Sciences Institute
National Institutes of Health Trinity College
Bethesda, Maryland Dublin, Ireland
Moon H. Nahm, MD John J. O’Shea, MD

Professor Chief, Molecular Immunology and Inflammation Branch
Departments of Pathology and Microbiology Scientific Director, National Institute of Arthritis and
University of Alabama at Birmingham Musculoskeletal and Skin Diseases
Birmingham, Alabama National Institutes of Health
Bethesda, Maryland
Kannan Natarajan, PhD
Staff Scientist William E. Paul, MD
Molecular Biology Section, Laboratory of Immunology
National Institute of Allergy and Infectious Diseases Susan K. Pierce, PhD
National Institutes of Health Chief
Bethesda, Maryland Laboratory of Immunogenetics
NIH NIAID
Mihai G. Netea, MD Rockville, Maryland
Professor
Department of Medicine Carlo Pucillo
Radboud University Nijmegen Medical Center Department of Biomedical Science and Technology
Nijmegen, The Netherlands Lab of Immunology
University of Udine
Falk Nimmerjahn, PhD Udine, Italy
Professor for Immunology & Genetics
Chair of Institute of Genetics Rino Rappuoli, PhD
Department of Biology Global Head, Vaccines Research
University of Erlangen-Nuernberg Novartis Vaccines and Diagnostics
Erlangen, Germany Siena, Italy
Luigi D. Notarangelo, MD Jeffrey V. Ravetch, MD, PhD

Professor Theresa & Eugene Lang Professor
Departments of Pediatrics and Pathology Laboratory of Molecular Genetics & Immunology
Harvard Medical School The Rockefeller University
Jeffrey Modell Chair of Pediatric Immunology Research New York, New York
Division of Immunology
Children’s Hospital Boston Stephen T. Reece
Boston, Massachusetts Department of Immunology
Max Planck Institute for Infection Biology
Michel C. Nussenzweig, MD, PhD Berlin, Germany
Sherman Fairchild Investigator and HHMI Investigator
The Rockefeller University Eleanor M. Riley, PhD
New York, New York Professor of Immunology
Department of Immunology and Infection
Pamela S. Ohashi, PhD London School of Hygiene and Tropical Medicine
Professor London, United Kingdom
Department of Medical Biophysics and Immunology
University of Toronto Paul A. Roche, PhD
Senior Scientist National Institutes of Health
Immunotherapy Program National Cancer Institute
Ontario Cancer Institute Experimental Immunology Branch
Toronto, Ontario, Canada Bethesda, Maryland
Paul_FM_final.indd x 9/17/12 3:25 PM

CONTRIBUTING AUTHORS | xi
Jamie Rossjohn, PhD Alan Sher, PhD

Professor Laboratory of Parasitic Diseases
Department of Biochemistry and Molecular Biology National Institutes of Allergy and Infectious Disease
Monash University National Institutes of Health
Melbourne, Australia Bethesda, Maryland
Ellen V. Rothenberg, PhD Ethan M. Shevach, MD

Albert Billings Ruddock Professor of Biology Chief, Cellular Immunology Section
Division of Biology Laboratory of Immunology
California Institute of Technology National Institute of Allergy and Infectious Diseases
Pasadena, California National Institutes of Health
Bethesda, Maryland
Nancy H. Ruddle, PhD
Professor Emerita and Senior Research Scientist Andrew L. Snow, PhD
Department of Epidemiology and Public Health Assistant Professor
Yale University School of Medicine Department of Pharmacology
New Haven, Connecticut Uniformed Services University of the Health Sciences
David H. Sachs, MD Bethesda, Maryland
Paul S. Russell/Warner-Lambert Professor of Surgery
Harvard Medical School Hae Won Sohn, PhD
Director, Transplantation Biology Research Center Staff Scientist
Department of Surgery Laboratory of Immunogenetics
Massachusetts General Hospital NIH, NIAID
Boston, Massachusetts Rockville, Maryland
David L. Sacks, PhD Megan Sykes, MD

Laboratory of Parasitic Diseases Michael J. Friedlander Professor of Medicine
National Institutes of Allergy and Infectious Disease Professor of Microbiology & Immunology and Surgical
National Institutes of Health Sciences (in Surgery)
Bethesda, Maryland Columbia University
Director, Columbia Center for Translational Immunology
Takashi Saito, PhD Columbia University College of Physicians and Surgeons
Deputy Director New York, New York
Laboratory for Cell Signaling
RIKEN Research Center for Allergy and Immunology Nicola Tamassia
Yokohama, Kanagawa, Japan Department of Pathology and Diagnostics
Section of General Pathology
Patrizia Scapini Verona, Italy
Department of Pathology and Diagnostics
Section of General Pathology Kevin J. Tracey, MD
Verona, Italy Investigator
President
Stephen P. Schoenberger, PhD
Head, Center for Biomedical Science
Associate Professor
Feinstein Institute for Medical Research
Division of Developmental Immunology
Manhasset, New York
La Jolla Institute for Allergy and Immunology
La Jolla, California
Lucy A. Truman
Hans Schreiber, MD, PhD Postdoctoral Fellow
Professor Yale University
Department of Pathology New Haven, Connecticut
University of Chicago
Chicago, Illinois Matthias G. von Herrath, MD
Directory, Type 1 Diabetes Center
Harry W. Schroeder, Jr., MD, PhD Department of Diabetes/Immunology
Professor of Medicine, Microbiology and Genetics La Jolla Institute for Allergy & Immunology (LIAI)
Division of Clinical Immunology and Rheumatology La Jolla, California
University of Alabama at Birmingham
Birmingham, Alabama
Paul_FM_final.indd xi 9/17/12 3:25 PM

xii | CONTRIBUTING AUTHORS
David Wald, MD, PhD Kathryn J. Wood, DPhil

Assistant Professor Professor of Immunology
Department of Pathology Nuffield Department of Surgical Sciences
Case Western Reserve University University of Oxford
University Hospitals Case Medical Center Oxford, United Kingdom
Cleveland, Ohio
Thomas A. Wynn, PhD
Bruce D. Walker, MD Laboratory of Parasitic Diseases
Professor National Institutes of Allergy and Infectious Disease
Department of Infectious Disease National Institutes of Health
Harvard Medical School Bethesda, Maryland
Director
Ragon Institute of MGH, MIT, and Harvard Wayne M. Yokoyama, MD
Massachusetts General Hospital Investigator
Boston, Massachusetts Howard Hughes Medical Institute
Sam J. and Audrey Loew Professor of Medicine
Carl F. Ware, PhD
Rheumatology Division
Director and Professor Washington University School of Medicine in St. Louis
Department of Infectious & Inflammatory Diseases St. Louis, Missouri
Sanford-Burnham Medical Research Institute
La Jolla, California
Marsha Wills-Karp, PhD

Rieveschl Professor
Department of Pediatrics
University of Cincinnati
Director
Division of Immunobiology
Cincinnati Children’s Hospital Medical Center
Cincinnati, Ohio
Paul_FM_final.indd xii 9/17/12 3:25 PM

PREFACE
Immunology is the quintessential medical science. Indeed, Perhaps that will prove to be true of modern immunology
no branch of the medical sciences has improved the health of as well when it is looked at by a disinterested observer, but
people more than the application of immunologic principles for one in the midst, the pace of discovery seems to speed
to prevention of disease. Smallpox has been eliminated from up with each passing year. Endless fascination certainly de-
the planet as a natural infection, as has poliomyelitis from scribes my experience of immunology.
the western hemisphere. Hepatitis B vaccine has prevented I hope that this seventh edition will convey the dyna-
more cancers than any intervention other than smoking ces- mism and creativity of modern immunology and provide
sation. The human papilloma virus vaccine promises to cut the reader with a solid introduction to our field and a picture
strikingly the toll of cervical cancer. of much of the very latest that has been achieved. As with
The continued need for progress in immunology is clear. each of the previous editions, most of the chapters are en-
The epidemic of human immunodeficiency virus roars on. tirely new and not simple reworkings of the chapter in the
Glimmers of hope from vaccine trials have led to a redou- previous edition. In order to contain the seventh edition
bling of effort, and the struggle to design effective vaccines within one volume, a decision was made to cite references
for the great infectious scourges goes on, with encouraging in the printed text but only to provide the detailed citations
results but no breakthroughs yet. Highly effective therapeu- in the online version. However, the references will be linked
tic vaccines for cancers still elude us, but some immunologic to PubMed so cited information can be easily obtained. The
therapies for cancer have met with encouraging results. We electronic version can be accessed at www.fundamentalim-
hope for much more as we marshal the tools coming from munology.com.
the study of the innate immune system, regulatory T cells, As before, this edition begins with an introductory sec-
and lymphocyte differentiation and effector function. tion consisting of the chapters “The Immune System” and
Understanding the basis of inflammation and the cyto- “History of Immunology.” These give an overview of mod-
kine world has given us effective drugs to treat rheumatoid ern immunology and of its origins, and provide those new
arthritis and a growing number of other autoinflamma- to the field with the basis to go on to the subsequent chap-
tory/autoimmune diseases. The value of the interventions ters. This is followed by an “expanded introduction” pro-
based on this knowledge, such as the use of tumor necrosis vided by the sections “Organization and Evolution of the
factor, interleukin-6, and interleukin -1 blockers, is now Immune System,” “Immunoglobulins and B-Lymphocytes,”
established. The application of anti–CD20 in the treatment and “T-Lymphocytes.” These are followed by the two core
of autoimmune disorders shows great promise. Even more “basic” immunology sections: “The Intersection of Innate
promising strategies are on the horizon. and Adaptive Immunity” and “Induction, Regulation, and
Fundamental Immunology has the goal of aiding in the Effector Functions of the Immune Response.” The book con-
education of a new generation of immunologists who can cludes with sections devoted to the immune system’s role in
both probe more deeply into the organizing principles of the protection against pathogenic microorganisms, “Immunity
immune system and can translate this new information into to Infectious Agents,” and to how the immune system is
effective treatments and preventatives that will extend and involved in a variety of human disorders, “Immunologic
enlarge on the record of immunologic science in bettering Mechanisms in Disease.”
the lot of human kind. I repeat a word of caution that has been in the Preface
Were I beginning the task of preparing a comprehensive to each edition. Immunology is moving very fast. Each
text of immunology today, I might have titled it Immunology, of the chapters is written by an expert in the field, but in
Endless Fascination. Certainly that describes my own view of some areas there may be differences of opinion expressed by
this science over the 30 years that I have been working on the equally accomplished authors. I ask the reader to take note
seven editions of Fundamental Immunology. I had believed of the differences and to follow developments in the field.
that scientific progress was marked by periods of intense
creativity, during which new concepts were established, fol- William E. Paul
lowed by longer periods of consolidation, when work that Washington, DC
made important but anticipated advances would dominate.
xiii
Paul_FM_final.indd xiii 9/17/12 3:25 PM

Paul_FM_final.indd xiv 9/17/12 3:25 PM

ACKNOWLEDGMENTS
The preparation of the seventh edition required the ef- immeasurably more difficult. Frances DeStefano’s coun-
forts of many individuals. I particularly wish to thank each sel was of utmost importance in planning this edition and
of the authors. Their contributions, prepared in the midst when she had to leave the project, Julie Goolsby took over
of extremely busy schedules, are responsible for the value and played a key role in making important publication
of this book. Leanne Vandetty of Lippincott Williams & decisions. I wish to gratefully acknowledge the efforts of
Wilkins saw that the process of receiving, editing, and as- each of the members of the editorial and production staffs
sembling the chapters went as smoothly as possible. Without of Lippincott Williams & Wilkins who participated in the
her efforts, the completion of this edition would have been preparation of this edition.
xv
Paul_FM_final.indd xv 9/17/12 3:25 PM

Paul_FM_final.indd xvi 9/17/12 3:25 PM

CHAPTER 36 COMPLEMENT | xvii
CONTENTS
Contributing Authors vii Section IV. T-Lymphocytes

Preface xiii
Acknowledgments xv
11 T-Cell Antigen Receptors . . . . . . . . . . . . . . . . . . .279
Mark M. Davis, Yueh-Hsiu Chien
Section I. Introduction to Immunology

12 Mechanisms of T-Lymphocyte
1 The Immune System . . . . . . . . . . . . . . . . . . . . . . . . . .1 Signaling and Activation . . . . . . . . . . . . . . . . . . . .306
William E. Paul Takashi Saito
2 History of Immunology . . . . . . . . . . . . . . . . . . . . . . .22 13 T-Lymphocyte Developmental Biology . . . . . . . .325

Steven Greenberg Ellen V. Rothenberg, Ameya Champhekar
14 Peripheral T-Lymphocyte
Section II. Organization and Evolution of Responses and Function . . . . . . . . . . . . . . . . . . . .355
the Immune System Marc K. Jenkins
3 Lymphoid Tissues and Organs . . . . . . . . . . . . . . . .47

Eitan M. Akirav, Noelia Alonso-Gonzalez, Section V. The Intersection of Innate and
Lucy A. Truman, Nancy H. Ruddle
Adaptive Immunity
4 Evolution of the Immune System . . . . . . . . . . . . . .67 15 The Innate Immune System . . . . . . . . . . . . . . . . . .367
Martin F. Flajnik, Louis Du Pasquier
Luke A. J. O’Neill
Section III. Immunoglobulins and 16 Dendritic Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . .381

Kang Liu, Michel C. Nussenzweig
B-Lymphocytes
5 Immunoglobulins: Structure and Function . . . . .129 17 Natural Killer Cells . . . . . . . . . . . . . . . . . . . . . . . . .395
Wayne M. Yokoyama
Harry W. Schroeder Jr, David Wald, Neil S. Greenspan
6 Immunoglobulins: Molecular Genetics . . . . . . .150 18 CD1d–Restricted Natural Killer T Cells . . . . . . .432

Albert Bendelac
Edward E. Max, Sebastian Fugmann
7 Antigen-Antibody Interactions and 19 Macrophages and Phagocytosis . . . . . . . . . . . . .448

Monoclonal Antibodies . . . . . . . . . . . . . . . . . . . . .183 Siamon Gordon
Jay A. Berzofsky, Ira J. Berkower
20 Granulocytes and Mast Cells . . . . . . . . . . . . . . . .468
8 B-Lymphocyte Development and Biology . . . . .215 Patrizia Scapini, Nicola Tamassia,
Richard R. Hardy Carlo Pucillo, Marco A. Cassatella
9 B-Lymphocyte Receptors, Signaling 21 The Major Histocompatibility

Mechanisms, and Activation . . . . . . . . . . . . . . . .246 Complex and Its Proteins . . . . . . . . . . . . . . . . . . .487
Akanksha Chaturvedi, Angel Davey, Wanli Liu, Hae Won Sohn, David H. Margulies, Kannan Natarajan,
Susan K. Pierce Jamie Rossjohn, James McCluskey
10 B-Lymphocyte Responses . . . . . . . . . . . . . . . . . . .261 22 Antigen Processing and Presentation . . . . . . . .524

Michael McHeyzer-Williams Ted H. Hansen, Paul A. Roche
xvii
Paul_FM_final.indd xvii 9/17/12 3:25 PM

xviii | CONTENTS
Section VI. Induction, Regulation, 36 Complement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .863

and Effector Functions of the Immune B. Paul Morgan
Response 37 Cell-Mediated Cytotoxicity . . . . . . . . . . . . . . . . . .891

Judy Lieberman
23 Immunogenicity and Antigen Structure . . . . . . .539
Jay A. Berzofsky, Ira J. Berkower
Section VII. Immunity to Infectious Agents

24 Fc Receptors and Their Role in Immune
Regulation and Inflammation . . . . . . . . . . . . . . . .583 38 The Immune Response to Parasites . . . . . . . . . .910
Jeffrey V. Ravetch, Falk Nimmerjahn Thomas A. Wynn, David L. Sacks, Alan Sher, Eleanor M. Riley
25 Type I Cytokines and Interferons, and 39 Immunity to Viruses . . . . . . . . . . . . . . . . . . . . . . . .937

Their Receptors . . . . . . . . . . . . . . . . . . . . . . . . . . . .601 Hildegund C.J. Ertl
Warren J. Leonard
40 Immunity to Intracellular Bacteria . . . . . . . . . . .973
26 The Interleukin-1 Family . . . . . . . . . . . . . . . . . . . .639 Anca Dorhoi, Stephen T. Reece, Stefan H. E. Kaufmann
Charles A. Dinarello, Mihai G. Netea
41 Immunity to Extracellular Bacteria . . . . . . . . . .1001
27 Tumor Necrosis Factor–Related Moon H. Nahm, Jannet Katz
Cytokines in Immunity . . . . . . . . . . . . . . . . . . . . . .659
Carl F. Ware 42 Immunology of Human Immunodeficiency
Virus Infection . . . . . . . . . . . . . . . . . . . . . . . . . . . .1016
28 Chemokines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .681 Douglas S. Kwon, Bruce D. Walker
Philip M. Murphy
43 Vaccines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1032
29 Helper T-Cell Differentiation Ennio De Gregorio, Ugo D’Oro, Sylvie Bertholet, Rino Rappuoli
and Plasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .708
John J. O’Shea
Section VIII. Immunologic Mechanisms
30 Programmed Cell Death . . . . . . . . . . . . . . . . . . . . .718 in Disease
Andrew L. Snow, Michael J. Lenardo
44 Autoimmunity and Autoimmune Diseases . . . .1069
31 Immunologic Memory . . . . . . . . . . . . . . . . . . . . . .741 Ken T. Coppieters, Matthias G. von Herrath, Dirk Homann
Shane Crotty, Susan M. Kaech, Stephen P. Schoenberger
45 Immunologic Mechanisms of
32 Immunologic Tolerance . . . . . . . . . . . . . . . . . . . . .765 Allergic Disorders. . . . . . . . . . . . . . . . . . . . . . . . .1113
Christopher C. Goodnow, Pamela S. Ohashi Marsha Wills-Karp, Ian Lewkowich
33 Regulatory/Suppressor T Cells . . . . . . . . . . . . . . .795 46 Transplantation Immunology . . . . . . . . . . . . . . .1154

Megan Sykes, Kathryn Wood, David H. Sachs
Ethan M. Shevach
34 The Mucosal Immune System . . . . . . . . . . . . . . .833 47 Cancer Immunology . . . . . . . . . . . . . . . . . . . . . . .1200

Hans Schreiber
Yasmine Belkaid
35 Neurophysiologic Reflex 48 Inborn Errors of Immunity . . . . . . . . . . . . . . . . . .1235

Jean-Laurent Casanova, Mary Ellen Conley, Luigi D. Notarangelo
Mechanisms in Immunology . . . . . . . . . . . . . . . . .850
Sangeeta S. Chavan, Kevin J. Tracey
Index 1267
Paul_FM_final.indd xviii 9/17/12 3:25 PM

Nothing is as powerful as an idea whose time has come.

Paraphrased from Victor Hugo
Everything should be made as simple as possible, but not simpler.

A lbert Einstein
From my teachers I have learned much, from my colleagues still more, but from my students most of all.
The Talmud
Discovery consists of seeing what everybody has seen and thinking what nobody has thought.
A lbert Szent-Gyorgyi
. . . the clonal selection hypothesis . . . assumes that . . . there exist clones of mesenchymal cells, each carrying immunologically
reactive sites . . . complementary . . . to one (or possibly a small number) of potential antigenic determinants.
Frank Macfarlane Burnet
In the fields of observation, chance favors the prepared mind.

Louis Pasteur
In all things of nature there is something of the marvelous.

A ristotle
Paul_FM_final.indd xix 9/17/12 3:25 PM

Paul_FM_final.indd xx 9/17/12 3:25 PM

SECTION
I Introduction to Immunology
CHAPTER
1 The Immune System
William E. Paul
The immune system is a remarkable defense mechanism. It that can be found in the online version of Fundamental
makes rapid, specific, and protective responses against the Immunology.
myriad potentially pathogenic microorganisms that inhabit
the world in which we live. The tragic examples of acquired
immunodeficiency syndrome (AIDS) and the inherited se-
KEY CHARACTERISTICS OF THE IMMUNE SYSTEM
vere combined immunodeficiencies graphically illustrate Innate Immunity (Chapters 15, 17, 19, and 20)
the consequences of a nonfunctional adaptive immune sys- Powerful nonspecific defenses prevent or limit infections
tem. Patients with AIDS and children with severe combined by most potentially pathogenic microorganisms. The epi-
immunodeficiency often fall victim to infections that are of thelium provides both a physical barrier to the entry of
little or no consequence to those with normally functioning microbes and produces a variety of antimicrobial factors.
immune systems. The immune system also has a role in the Agents that penetrate the epithelium are met with mac-
rejection of tumors and, when dysregulated, may give rise to rophages and related cells possessing “microbial sensors”
a series of autoimmune diseases, including insulin-depen- that recognize key molecules characteristic of many mi-
dent diabetes mellitus, multiple sclerosis, rheumatoid ar- crobial agents. These “pattern recognition receptors”
thritis, systemic lupus erythematosus, inflammatory bowel include several families of molecules, of which the most
disease, among others. intensively studied are the toll-like receptors (TLRs) and
Fundamental Immunology has as its goal the authori- the nucleotide oligomerization domain–like receptors.
tative presentation of the basic elements of the immune Each TLR recognizes a distinct substance (or set of sub-
system, of the means through which the mechanisms of stances) associated with microbial agents; for example,
immunity act in a wide range of clinical conditions, in- TLR4 recognizes lipopolysaccharides, TLR3, double-
cluding recovery from infectious diseases, rejection of stranded ribonucleic acid, and TLR9, unmethylated CpG-
tumors, transplantation of tissue and organs, autoimmu- containing DNA. Because the recognized substances are
nity and other immunopathologic conditions, and allergy, generally indispensable to the infectious agent, microbial
and how the mechanisms of immunity can be marshaled sensors provide a highly efficient means to recognize po-
by vaccination to provide protection against microbial tential pathogens.
pathogens. The interaction of a TLR with its ligand induces a series of
The purpose of the opening chapter is to provide intracellular signaling events, of which activation of the NF-
readers with a general introduction to our current under- κ B system is particularly important. Macrophage activation
standing of the immune system. It should be of particular with enhancement of the cell’s phagocytic activity and the
importance for those with a limited background in im- induction of antimicrobial systems aid in the destruction of
munology, providing them with the preparation needed the pathogen. The induction of an inflammatory response
for subsequent chapters of the book. Rather than provid- as a result of the activation of the innate immune system
ing extensive references in this chapter, each of the sub- recruits other cell types, including neutrophils, to the site.
ject headings will indicate the chapters that deal in detail The innate system can provide an effective means to control
with the topic under discussion. Those chapters will not or eliminate pathogens. Indeed, life forms other than verte-
only provide an extended treatment of the topic but will brates rely on the innate immune system to allow them to
also furnish the reader with a comprehensive reference list deal with microbial infection.
1
Paul_CH01_final.indd 1 9/17/12 5:16 AM

2 | SECTION I INTRODUCTION TO IMMUNOLOGY
In vertebrates, the innate immune system also acts to The Immune System is Tolerant of Self-Antigens
recruit antigen-specific immune responses, not only by at- (Chapters 32 and 33)
tracting cells of the immune system to the site of infection, One of the most important features of the immune system
but also through the uptake of antigen by dendritic cells is its ability to discriminate between antigenic determinants
(DCs) and its transport by these cells to lymphoid tissues expressed on foreign substances, such as pathogenic mi-
where primary immune responses are initiated. Activated crobes, and potential antigenic determinants expressed by
DCs express cell surface costimulatory molecules and pro- the tissues of the host. The failure of the system to make
duce cytokines that can regulate the quality of the immune full-blown immune responses to self-antigens is referred
response so that it is most appropriate to combating the par- to as immunologic tolerance. Tolerance is a complex pro-
ticular infectious agent, be it a virus, bacterium, or parasite. cess that actually involves several distinct processes. One
element, perhaps the most important, is an active process
Primary Responses (Chapters 10 and 14) involving the elimination or inactivation of cells that can
recognize self-antigens. In addition, there are mechanisms
Primary immune responses are initiated when a foreign
through which cells that encounter antigens (such as self-
antigenic substance interacts with antigen-specific lym-
antigens) in the absence of cues from the innate immune
phocytes under appropriate circumstances. The response
system may fail to make a response, may make a minimal
generally consists of the production of antibody molecules
response, or may be inactivated through a process referred
specific for the antigenic determinants of the immunogen
to as anergy. Finally, a specialized set of T cells exist des-
and of the expansion and differentiation of antigen-specific
ignated regulatory cells that actively suppress responses
helper and effector T-lymphocytes. The latter include cy-
against self-antigens. Indeed, individuals who have muta-
tokine-producing cells and killer T cells, capable of lysing
tions in the key transcription factor Foxp3 expressed by the
infected cells. Generally, the combination of the innate
regulatory cells develop severe multiorgan autoimmunity
immune response and the primary adaptive response are
(Immunodysregulation polyendocrinopathy, enteropathy
sufficient to eradicate or to control the microbe. Indeed, the
X-linked syndrome). The critical necessity to control self-
most effective function of the immune system is to mount a
reactivity is clearly shown by this multilayered system that
response that eliminates the infectious agent from the body,
involves elimination, inactivation, and suppression.
so-called sterilizing immunity.
Immune Responses Against Self-Antigens can Result

Secondary Responses and Immunologic Memory in Autoimmune Diseases (Chapter 44)
(Chapters 10, 14, 29, and 31) Failure in establishing immunologic tolerance or un-
As a consequence of initial encounter with antigen, the usual presentations of self-antigens can give rise to tissue-
immunized individual develops a state of immunologic damaging immune responses directed against antigenic
memory. If the same (or a closely related) microorganism is determinants on host molecules. These can result in auto-
encountered again, a secondary response is made. This gen- immune diseases. As has already been mentioned, a group
erally consists of an antibody response that is more rapid, of extremely important diseases are caused by autoimmune
greater in magnitude, and composed of antibodies that bind responses or have major autoimmune components, includ-
to the antigen with greater affinity and are more effective ing systemic lupus erythematosus, rheumatoid arthritis,
in clearing the microbe from the body. A more rapid and insulin-dependent diabetes mellitus, multiple sclerosis, my-
more effective T-cell response also ensues. Thus, an initial asthenia gravis, and inflammatory bowel disease. Efforts
infection with a microorganism often initiates a state of im- to treat these diseases by modulating the autoimmune re-
munity in which the individual is protected against a second sponse are a major theme of contemporary medicine.
infection. In the majority of situations, protection is pro-
vided by high-affinity antibody molecules that rapidly clear
the re-introduced microbe. This is the basis of most licensed Acquired Immunodeficiency Syndrome is an Example
vaccines; the great power of vaccines is illustrated by the of a Disease Caused by a Virus That the Immune
elimination of smallpox from the world and by the complete System Generally Fails to Eliminate (Chapter 42)
control of polio in the western hemisphere. Immune responses against infectious agents do not always
lead to elimination of the pathogen. In some instances, a
chronic infection ensues in which the immune system adopts
The Immune Response is Highly Specific and the a variety of strategies to limit damage caused by the organ-
Antigenic Universe is Vast ism or by the immune response. Indeed, herpes viruses, such
The immune response is highly specific. Primary immuniza- as human cytomegalovirus, frequently are not eliminated by
tion with a given microorganism evokes antibodies and T cells immune responses and establish a chronic infection in which
that are specific for the antigenic determinants found on that the virus is controlled by immune responses. One of the most
microorganism but that generally fail to recognize (or recog- notable infectious diseases in which the immune response
nize only poorly) antigenic determinants expressed by unre- generally fails to eliminate the organism is AIDS, caused by
lated microbes. Indeed, the range of antigenic specificities that the human immunodeficiency virus (HIV). In this instance,
can be discriminated by the immune system is enormous. the principal infected cells are those of the immune system

CHAPTER 1 THE IMMUNE SYSTEM | 3
itself, leading to an eventual state in which the individual can infections. Some memory lymphocytes are “on patrol” in the
no longer mount protective immune responses against other tissues, scanning for reintroduction of their specific antigens.
microbial pathogens. Indeed, under the assault of HIV, con- Lymphocytes are also found in the central lymphoid organs,
trol of viruses such as cytomegalovirus is lost and they may the thymus, and bone marrow, where they undergo the de-
cause major tissue damage. velopmental steps that equip them to mediate the responses
of the mature immune system.
Individual lymphocytes are specialized in that they are
Major Principles of Immunity committed to respond to a limited set of structurally related
The major principles of the immune response are: antigens. This commitment exists before the first contact of
• Elimination of many microbial agents through the the immune system with a given antigen. It is expressed by
nonspecific protective mechanisms of the innate im- the presence on the lymphocyte’s surface of receptors specific
mune system for determinants (epitopes) of the antigen. Each lymphocyte
• Cues from the innate immune system inform the cells possesses a population of receptors, all of which have identi-
of the adaptive immune system as to whether it is ap- cal combining sites (this is a slight oversimplification as oc-
propriate to make a response and what type of response casionally T cells and less frequently B cells may express two
to make populations of receptors). One set, or clone, of lymphocytes
• Cells of the adaptive immune system display exquisitely differs from another clone in the structure of the combining
specific recognition of foreign antigens and mobilize region of its receptors and thus in the epitopes that it can
potent mechanisms for elimination of microbes bearing recognize. The ability of an organism to respond to virtually
such antigens any non–self-antigen is achieved by the existence of a very
• The immune system displays memory of its previous large number of different lymphocytes, each bearing recep-
responses tors specific for a distinct epitope. As a consequence, lym-
• Tolerance of self-antigens. phocytes are an enormously heterogeneous group of cells.
Based on reasonable assumptions as to the range of diversity
The remainder of this introductory chapter will describe that can be created in the genes encoding antigen-specific
briefly the molecular and cellular basis of the system and receptors, it is virtually certain that the number of distinct
how these central characteristics of the immune response combining sites on lymphocyte receptors of an adult human
may be explained. can be measured in the millions.
Lymphocytes differ from each other not only in the
specificity of their receptors but also in their functions. There
CELLS OF THE IMMUNE SYSTEM AND THEIR are two broad classes of lymphocytes: the B-lymphocytes,
SPECIFIC RECEPTORS AND PRODUCTS which are precursors of antibody-secreting cells, and the T
The immune system consists of several distinct cell types, (thymus-derived)-lymphocytes. T-lymphocytes express im-
each with important roles. The lymphocytes occupy central portant helper functions, such as the ability to aid in the devel-
stage because they are the cells that determine the specific- opment of specific types of immune responses, including the
ity of immunity. It is their response that orchestrates the ef- production of antibody by B cells, the increase in the microbi-
fector limbs of the immune system. Cells that interact with cidal activity of macrophages, and the recruitment of granu-
lymphocytes play critical parts both in the presentation of locytes to sites of infection. Other T-lymphocytes are involved
antigen and in the mediation of immunologic functions. in direct effector functions, such as the lysis of virus-infected
These cells include DCs and the closely related Langerhans cells or certain neoplastic cells. Regulatory T-lymphocytes
cells, monocyte/macrophages, natural killer (NK) cells, neu- have the capacity to suppress immune responses.
trophils, mast cells, basophils, and eosinophils. In addition,
a series of specialized epithelial and stromal cells provide the
anatomic environment in which immunity occurs, often by
B-LYMPHOCYTES AND ANTIBODY
secreting critical factors that regulate migration, growth and B-Lymphocyte Development (Chapter 8)
homeostasis, and gene activation in cells of the immune sys- B-lymphocytes derive from lymphoid progenitor cells,
tem. Such cells also play direct roles in the induction and which in turn are derived from hematopoietic stem cells
effector phases of the response. (Fig. 1.1). A detailed picture has been obtained of the molec-
The cells of the immune system are found in peripheral ular mechanisms through which committed early members
organized tissues, such as the spleen, lymph nodes, Peyer’s of the B lineage develop into mature B-lymphocytes. These
patches of the intestine, and tonsils, where primary immune events occur in the fetal liver and, in adult life, in the bone
responses generally occur (see Chapter 3). Many of the lym- marrow. Interaction with specialized stromal cells and their
phocytes comprise a recirculating pool of cells found in the products, including cytokines such as interleukin (IL)-7 and
blood and lymph, as well as in the lymph nodes and spleen, BAFF, are critical to the normal regulation of this process.
providing the means to deliver immunocompetent cells to The key events in B-cell development involve commit-
sites where they are needed and to allow immunity that is ment to the B lineage and repression of the capacity to
initiated locally to become generalized. Activated lympho- differentiate to cells of other lineages. In pro-B cells and
cytes acquire the capacity to enter nonlymphoid tissues pre-B cells, the genetic elements that encode the antigen-
where they can express effector functions and eradicate local specific receptors are assembled. These receptors are im-

FIG. 1.1. The patterns of gene expression, timing of gene rearrangement events, and capacity for self-replenishment and for rapid prolifera-
tion of developing B lymphocytes are indicated. (Adapted from Hardy RR, Hayakawa K. B cell development pathways. Ann Rev Immunol.
2001;19:595–621.)
munoglobulin (Ig) molecules specialized for expression on The H-chain V region is initially expressed in associa-
the cell surface. Igs are heterodimeric molecules consisting tion with the product of the μ C-region gene. Together, these
of heavy (H) and light (L) chains, both of which have vari- elements encode the μ IgH chain, which is used in Igs of the
able (V) regions, which are responsible for the binding of IgM class.
antigen and that differ in sequence from one Ig molecule The successful completion of the process of Ig gene rear-
to another (see Chapters 5, 6, and 7) (Fig. 1.2) and constant rangement and the expression of the resultant IgM on the
(C) regions. cell surface marks the transition between the pre-B– and
The genetic elements encoding the variable portions of B–cell states (see Fig. 1.1). The newly differentiated B cell
Ig H and L chains are not contiguous in germline DNA or initially expresses surface Ig solely of the IgM class. The cell
in the DNA of nonlymphoid cells (see Chapter 6) (Fig. 1.3). completes its maturation process by expressing on its sur-
In pro- and pre-B cells, these genetic elements are translo- face a second class of Ig composed of the same L chain and
cated to construct an expressible V-region gene. This process the same H chain V (VDJ) region but of a different H chain
involves a choice among a large set of potentially usable V, C region; this second Ig H chain is designated δ, and the Ig
diversity (D), and joining (J) elements in a combinatorial to which it contributes is designated IgD, so that the mature
manner and depends upon the recombinating activating naïve B cells express both IgM and IgD surface molecules
gene (RAG) proteins, RAG1 and RAG2. Such combinatorial that share the same V region.
translocation, together with the addition of diversity in the The differentiation process is controlled at several
course of the joining process, results in the generation of a steps by a system of checks that determines whether prior
very large number of distinct H and L chains. The pairing of steps have been successfully completed. These checks de-
H and L chains in a quasirandom manner further expands pend on the expression on the surface of the cell of appro-
the number of distinct Ig molecules that can be formed. priately constructed Ig or Ig-like molecules. For, example,

B-Lymphocyte Activation (Chapter 9)

A mature B cell can be activated by an encounter with an-
tigen-expressing epitopes that are recognized by its cell sur-
face Ig (Fig. 1.4). The activation process may be a direct one,
dependent on cross-linkage of membrane Ig molecules by
the antigen (cross-linkage–dependent B-cell activation), or an
indirect one, occurring most efficiently in the context of an
intimate interaction with a helper T cell, in a process often
referred to as cognate help.
Because each B cell bears membrane Ig molecules with iden-
tical variable regions, cross-linkage of the cell surface receptors
requires that the antigen express more than one copy of an epi-
tope complementary to the binding site of the receptor. This
requirement is fulfilled by antigens with repetitive epitopes.
Among these antigens are the capsular polysaccharides of many
medically important microorganisms such as pneumococci,
streptococci, and meningococci. Similar expression of multiple
identical epitopes on a single immunogenic particle is a property
of many viruses because they express multiple copies of envelope
proteins on their surface. Cross-linkage–dependent B-cell acti-
vation is a major protective immune response mounted against
these microbes. The binding of complement components (see
Chapter 36) to antigen or antigen/antibody complexes can in-
crease the magnitude of the cross-linkage–dependent B-cell ac-
tivation due to the action of a receptor for complement, which,
together with other molecules, increases the magnitude of a
FIG. 1.2. A schematic representation of an immunoglobulin mole- B-cell response to limiting amounts of antigen.
cule indicating the means through the variable regions and the CH1 Cognate help allows B cells to mount responses against
and CL regions of heavy and light chains pair with one another and antigens that cannot cross-link receptors and, at the same
how the CH2 and CH3 regions of the heavy chains pair. time, provides costimulatory signals that rescue B cells from
inactivation when they are stimulated by weak cross-linkage
in the period after a μ chain has been successfully as- events. Cognate help is dependent on the binding of antigen
sembled but before an L chain has been assembled, the μ by the B cell’s membrane Ig, the endocytosis of the antigen,
chain is expressed on the cell surface in association with and its fragmentation into peptides within the endosomal/
a surrogate L chain, consisting of VpreB and λ 5. Pre-B lysosomal compartment of the cell. Some of the resultant
cells that fail to express this μ /VpreB λ 5 complex do not peptides are loaded into a groove in a specialized set of
move forward to future differentiation states or do so very cell surface proteins, the class II major histocompatibility
inefficiently. complex (MHC) molecules (Fig. 1.5). The resultant class II/
FIG. 1.3. Organization and translocation of mouse immunoglobulin (Ig)H genes. IgH chains are encoded by four distinct genetic elements:
Igh-V (V), Igh-D (D), Igh-J (J) and Igh-C. The variable (V), diversity (D), and joining (J) genetic elements together specify the variable region
of the heavy chain. The Igh-C element specifies the constant (C) region. The same V region can be expressed in association with each of
the C regions (μ, δ, γ3,γ1,γ2b, γ2a, ε, and α). In the germline, the V, D, and J genes are far apart and there are multiple forms of each of these
genes. In the course of lymphocyte development, a VDJ gene complex is formed by translocation of individual V and D genes so that they lie
next to one of the J genes, with excision of the intervening genes. This VDJ complex is initially expressed with μ and δ C genes but may be
subsequently translocated so that it lies near one of the other C genes (e.g., γ1) and in that case leads to the expression of a VDJ γ1 chain.

A FIG. 1.4. Two Forms of B-Cell Activation. A: Cognate T cell/

B cell help. Resting B cells can bind antigens that
bear epitopes complementary to their cell surface Ig.
Even if the antigen cannot cross-link the receptor, it
will be endocytosed and enter late endosomes and ly-
sosomes where it will be degraded to peptides. Some
of these peptides will be loaded into class II major his-
tocompatibility complex molecules and brought to the
cell surface, where they can be recognized by CD4-
positive T cells that bear receptors specific for that
peptide/class II complex. This interaction allows an
activation ligand on the T cells (CD40 ligand) to bind
to its receptor on B cells (CD40) and to signal B-cell
activation. In addition, the T cells secrete several cy-
tokines that regulate the growth and differentiation
of the stimulated B cell. B: Cross-linkage–dependent
B cell activation. When B cells encounter antigens
that bear multiple copies of an epitope that can bind
to their surface immunoglobulin, the resultant cross-
linkage stimulates biochemical signals within the cell
leading to B-cell activation, growth, and differentia-
tion. In many instances, B-cell activation events may
B result from both pathways of stimulation.
FIG. 1.5. Illustration of the structure of the peptide

binding domain (α1 and β1) of a class II major histo-
compatibility complex molecule bound to an antigenic
peptide from influenza hemagglutinin (adapted by D.H.
Margulies from Stern LJ, Brown JH, Jardetzky TS, et
al. Crystal structure of the human class II MHC protein
HLA-DR1 complexed with an influenza virus peptide.
Nature. 1994;68[6468]:215–221.)

peptide complexes are expressed on the cell surface. As will occurs mainly in germinal centers. Both somatic hypermu-
be discussed subsequently, these complexes are the ligands tation and immunoglobulin class switching depend upon
for the antigen-specific receptors of a set of T cells desig- the action of activation-induced cytidine deaminase (AID)
nated CD4 T cells. CD4 T cells that have receptors specific that plays an important role in the breakage and repair of
for the class II/peptide complex expressed on the B-cell sur- DNA, which is essential for recombination events.
face recognize and interact with that B cell. That interac-
tion results in the activation of the B cell through the agency
of cell surface molecules expressed by the T cells (e.g., the B1 and Marginal Zone B-Lymphocytes
CD40 ligand [CD154]) and cytokines produced by the T cell (Chapters 8 and 10)
(see Fig. 1.4). The role of the B-cell receptor for antigen is B lymphocytes consist of at least three distinct populations:
to create the T-cell ligand on the surface of antigen-specific conventional B cells, B1 B cells, and marginal zone B cells.
B cells; activation of the B cell derives largely from the ac- B1 B cells were initially recognized because some express a
tion of the T cell. However, in many physiologic situations, cell-surface protein, CD5, not generally found on other B
receptor cross-linkage stimuli and cognate help synergize to cells. In the adult mouse, B1 B cells are found in relatively
yield more vigorous B-cell responses. Recently, it has been high frequency in the peritoneal cavity but are present at
shown that the association of ligands for TLRs with antigen low frequency in the spleen and lymph nodes. B1 B cells are
will strikingly enhance B cell responses. quite numerous in fetal and perinatal life and appear to be
self-renewing, in contrast to conventional B cells, in which
division and memory are antigen driven.
B-Lymphocyte Differentiation (Chapters 9 and 10) Marginal zone B cells are localized in a distinct anatomical
Activation of B cells prepares them to divide and to differ- region of the spleen (the marginal zone) that represents the
entiate either into antibody-secreting cells or into memory major antigen filtering and scavenging area. Like B1 B cells,
cells, so that there are more cells specific for the antigen marginal zone B cells express a repertoire biased toward bac-
used for immunization. Those cells that differentiate into terial cell wall constituents and senescent self-components.
antibody-secreting cells account for primary antibody re- Marginal zone and B1 B cells respond very rapidly to anti-
sponses. Some of these antibody-secreting cells migrate to genic challenge, likely independently of T cells. Uniquely,
the bone marrow where they may continue to produce anti- among all populations of B cells, marginal zone B cells are
body for an extended period of time and may have lifetimes dependent on Notch-2 signaling for their development.
very long. B1 B cells and marginal zone B cells are responsible for
Memory B cells give rise to antibody-secreting cells upon the secretion of the serum IgM that exists in nonimmunized
rechallenge of the individual. The hallmark of the antibody mice, often referred to as natural IgM. Among the antibodies
response to rechallenge (a secondary response) is that it is of found in such natural IgM are molecules that can combine
greater magnitude, occurs more promptly, is composed of an- with phosphatidylcholine (a component of pneumococcal
tibodies with higher affinity for the antigen, and is dominated cell walls) and with lipopolysaccharide and influenza virus.
by Igs expressing γ, α, or ε C regions (IgG, IgA, or IgE) rather B1 B cells also produce autoantibodies, although they are
than by IgM, which is the dominant Ig of the primary response. generally of low affinity and in most cases not pathogenic.
Division and differentiation of cells into antibody- There is evidence that B1 B cells are important in resistance
secreting cells is largely controlled by the interaction of the to several pathogens and may have a significant role in mu-
activated B cells with T cells expressing CD154 and by their cosal immunity.
stimulation by T-cell–derived cytokines.
The differentiation of activated B cells into memory cells
occurs in a specialized microenvironmental structure in the B-Lymphocyte Tolerance (Chapter 32)
spleen and lymph nodes: the germinal center. The increase in One of the central problems facing the immune system is that
antibody affinity also takes place within the germinal center. of being able to mount highly effective immune responses to
This process, designated affinity maturation, is dependent on the antigens of foreign, potentially pathogenic agents while
somatic hypermutation. The survival of B cells within the ignoring antigens associated with the host’s own tissues. The
germinal center depends on their capacity to bind antigen so mechanisms ensuring this failure to respond to self-antigens
that as the amount of antigen diminishes, B cells that have are complex and involve a series of strategies. Chief among
higher affinity receptors, either naturally or as a result of the them is elimination of cells capable of self-reactivity or the
hypermutation process, have a selective survival and growth inactivation of such cells. The encounter of immature, naïve
advantage. Thus, such cells come to dominate the population. B cells with antigens with repetitive epitopes capable of
The process through which a single H-chain V region can cross-linking membrane Ig can lead to elimination of the B
become expressed with genes encoding C regions other than cells, particularly if no T-cell help is provided at the time of
μ or δ is referred to as Ig class switching. It is dependent the encounter. This elimination of potentially self-reactive
on a gene translocation event through which the C-region cells is often referred to as clonal elimination. Many self-
genes between the genetic elements encoding the V region reactive cells, rather than dying upon encounter with self-
and the newly expressed C gene are excised, resulting in antigens, undergo a further round of Ig gene rearrangement.
the switched C gene being located in the position that the This receptor editing process allows a self-reactive cell to
Cμ gene formerly occupied (see Fig. 1.3). This process also substitute a new receptor and therefore to avoid elimination.

There are many self-antigens that are not encountered by regions (CDRs). The CDRs contain the amino acids that
the developing B-cell population or that do not have the ca- are the L chain’s contribution to the lining of the antibody’s
pacity to cross-link B-cell receptors to a sufficient degree to combining site. The three CDRs are interspersed among
elicit the receptor editing /clonal elimination process. Such four regions of much lower degree of variability, designated
cells, even when mature, may nonetheless be inactivated framework regions.
through a process that involves cross-linkage of receptors The H chains of Ig molecules are of several classes de-
without the receipt of critical costimulatory signals. These termined by their constant regions ( μ , δ, γ [of which there
inactivated cells may be retained in the body but are un- are several subclasses], α and ε). An assembled Ig mole-
responsive to antigen and are referred to as anergic. When cule, consisting of one or more units of two identical H
removed from the presence of the anergy-inducing stimulus, and L chains, derives its name from the constant region
anergic cells may regain responsiveness. of the H chain that it possesses. Thus, there are IgM, IgD,
IgG, IgA, and IgE antibodies. The H chains each consist
Immunoglobulins of a single amino terminal V region and three or four C
regions. In many H chains, a hinge region separates the
Structure (Chapter 5)
fi rst and second C regions and conveys flexibility to the
Igs are the antigen-specific membrane receptors and secreted
molecule, allowing the two combining sites of a single unit
products of B cells. They are members of a large family of
to move in relation to one another so as to promote the
proteins designated the Ig supergene family. Members of the
binding of a single antibody molecule to an antigen that
Ig supergene family have sequence homology, a common
has more than one copy of the same epitope. Such divalent
gene organization, and similarities in three-dimensional
binding to a single antigenic structure results in a great
structure. The latter is characterized by a structural ele-
gain in energy of interaction (see Chapter 7). The H-chain
ment referred to as the Ig fold, generally consisting of a set
V region, like that of the L chain, contains three CDRs lin-
of seven β -pleated sheets organized into two opposing layers
ing the combining site of the antibody and four framework
(Fig. 1.6). Many of the cell surface proteins that participate
regions.
in immunologic recognition processes, including the T-cell
The C region of each H-chain class conveys unique func-
receptor (TCR), the CD3 complex, and signaling molecules
tional attributes to the antibodies that possess it. Among the
associated with the B-cell receptor (Igα and Igβ), are mem-
distinct biologic functions of each class of antibody are the
bers of the Ig supergene family.
following:
The Igs themselves are constructed of a unit that consists
of two H chains and two L chains (see Fig. 1.2). The H and L • IgM antibodies are potent activators of the comple-
chains are composed of a series of domains, each consisting ment system (see Chapter 36).
of approximately 110 amino acids. • IgA antibodies are secreted into a variety of bodily
The L chains, of which there are two types (κ and λ), fluids and are principally responsible for immunity at
consist of two domains. The carboxy-terminal domain is mucosal surfaces (see Chapter 34).
essentially identical among L chains of a given type and is • IgE antibodies are bound by specific receptors (FcεRI)
referred to as the C region. As already discussed, the amino on basophils and mast cells. When cross-linked by
terminal domain varies from L chain to L chain and con- antigen, these IgE/FcεRI complexes cause the cells to
tributes to the binding site of antibody. Because of its vari- release a set of mediators responsible for allergic in-
ability, it is referred to as the V region. The variability of flammatory responses (see Chapter 45).
this region is largely concentrated in three segments, des- • IgD antibodies act virtually exclusively as membrane
ignated the hypervariable or complementarity-determining receptors for antigen.
FIG. 1.6. Schematic Drawing of the Variable

and Constant Domains of an Immunoglobulin
Light Chain Illustrating the “Immunoglobulin
Fold.” The β strands participating in the an-
tiparallel β-pleated sheets of each domain
are represented as arrows. The β strands
of the three-stranded sheets are shaded,
whereas those in the four-stranded sheets
are white. The intradomain disulfide bonds are
represented as black bars. Selected amino
acids are numbered with position 1 as the N
terminus. (Reprinted with permission from
Edmundson AB, Ely KR, Abola EE, Schiffer
M, Panagiotopoulous N. Rotational allomer-
ism and divergent evolution of domains in
immunoglobulin light chains. Biochemistry.
1975;14:3953–3961).

• IgG antibodies, made up of four subclasses in both this process is almost certainly not completely random, it
humans and mice, mediate a wide range of functions allows the formation of an exceedingly large number of dis-
including transplacental passage and opsonization of tinct Ig molecules, the majority of which will have individ-
antigens through binding of antigen/antibody com- ual specificities.
plexes to specialized Fc receptors on macrophages and The rearrangement events that result in the assembly
other cell types (see Chapters 19, 20, and 24). of expressible IgH and IgL chains occur in the course of
B-cell development in pro-B cells and pre-B cells, respec-
IgD, IgG, and IgE antibodies consist of a single unit of
tively (see Fig. 1.1). This process is regulated by the Ig
two H and L chains. IgM antibodies are constructed of five
products of the rearrangement events. The formation of a μ
or six such units, although they consist of a single unit when
chain signals the termination of rearrangement of H-chain
they act as membrane receptors. IgA antibodies may con-
gene elements and the onset of rearrangement of L-chain
sist of one or more units. The antibodies that are made up
gene elements, with κ rearrangements generally preceding
of more than a single unit generally contain an additional
λ rearrangements. One important consequence of this is
polypeptide chain, the J chain, that appears to play a role
that only a single expressible μ chain will be produced in
in the polymerization process. In addition, secreted IgA ex-
a given cell, as the fi rst expressible μ chain shuts off the
presses a chain, a secretory piece, that is derived from the re-
possibility of producing an expressible μ chain on the al-
ceptor for polymeric IgA, which plays a role in the transport
ternative chromosome. Comparable mechanisms exist to
of IgA through the cells lining the lumen of the gut.
ensure that only one L-chain gene is produced, leading
Each of the distinct Igs can exist as secreted antibodies
to the phenomenon known as allelic exclusion. Thus, the
and as membrane molecules. Antibodies and cell surface
product of only one of the two alternative allelic regions
receptors of the same class made by a specific cell have iden-
at both the H- and L-chain loci are expressed. The closely
tical structures except for differences in their carboxy-ter-
related phenomenon of L-chain isotype exclusion ensures
minal regions. Membrane Igs possess a hydrophobic region,
the production of either κ or λ chains in an individual cell,
spanning the membrane, and a short intracytoplasmic tail,
but not both. An obvious but critical consequence of allelic
both of which are lacking in the secretory form.
exclusion is that in most cases an individual B cell makes
antibodies, all of which have identical H- and L-chain V re-
Immunoglobulin Genetics (Chapter 6) gions, a central prediction of the clonal selection theory of
The components of the Ig H-chain gene have already been the immune response. During receptor editing, secondary
alluded to. To reiterate, the IgH chain gene of a mature rearrangements occur. Receptor editing is induced when
lymphocyte is derived from a set of genetic elements that the initial membrane Ig is capable of self-reactivity. As a
are separated from one another in the germline. The V re- consequence of the resultant secondary rearrangement,
gion is composed of three types of genetic elements: VH, D, Ig of a different specificity is expressed, usually no longer
and JH. More than 100 VH elements exist; there are more self-reactive.
than 10 D elements and a small number of JH elements (4 in
the mouse). An H-chain VH DJH gene is created by the trans-
location of one of the D elements on a given chromosome to Class Switching (Chapter 6)
one of the JH elements on that chromosome, generally with An individual B cell continues to express the same IgH-
the excision of the intervening DNA. This is followed by a chain V region as it matures but it can switch the IgH-chain
second translocation event in which one of the VH elements C region it uses (see Fig. 1.3). Thus, a cell that expresses re-
is brought into apposition with the assembled DJH element ceptors of the IgM and IgD classes may differentiate into a
to create the VH DJH (V region) gene (see Fig. 1.3). Although cell that expresses IgG, IgA, or IgE receptors and then into
it is likely that the choice of the VH, D, and JH elements that a cell secreting antibody of the same class as it expressed on
are assembled is not entirely random, the combinatorial the cell surface. This process allows the production of anti-
process allows the creation of a very large number of dis- bodies capable of mediating distinct biologic functions but
tinct H-chain V-region genes. Additional diversity is cre- that retain the same antigen-combining specificity. When
ated by the imprecision of the joining events and by the linked with the process of affinity maturation of antibodies,
deletion of nucleotides and addition of new, untemplated Ig class switching provides antibodies of extremely high ef-
nucleotides between D and JH and between VH and D, form- ficacy in preventing re-infection with microbial pathogens
ing N regions in these areas. This further increases the or in rapidly eliminating such pathogens. The associated
diversity of distinct IgH chains that can be generated from phenomena of class switching and affinity maturation ac-
the relatively modest amount of genetic information pres- count for the high degree of effectiveness of antibodies pro-
ent in the germline. duced in secondary immune responses.
The assembly of L-chain genes follows generally similar The process of class switching is known to involve a ge-
rules. However, L chains are assembled from VL and JL ele- netic recombination event between specialized switch (S)
ments only. Although there is junctional diversity, no N re- regions, containing repetitive sequences, that are located
gions exist for L chains. Additional diversity is provided by upstream of each C region genetic element (with the excep-
the existence of two classes of L chains, κ and λ. tion of the δ C region). Thus, the S region upstream of the μ
An Ig molecule is assembled by the pairing of an IgH- CH region gene (Sμ) recombines with an S region upstream
chain polypeptide with an IgL-chain polypeptide. Although of a more 3′ isotype, such as Sγ1, to create a chimeric S μ /Sγ1

This primary repertoire is sufficiently large so that most epi-

topes on foreign antigens will encounter B cells with com-
plementary receptors. Thus, if adequate T-cell help can be
generated, antibody responses can be made to a wide array
of foreign substances. Nonetheless, the antibody that is ini-
tially produced usually has a relatively low affi nity for the
antigen. This is partially compensated for by the fact that
IgM, the antibody initially made, is a pentamer. Through
multivalent binding, high avidities can be achieved even if
individual combining sites have only modest affinity (see
Chapter 7). In the course of T-cell–dependent B-cell stimu-
lation, particularly within the germinal center, a process of
somatic hypermutation is initiated that leads to a large num-
ber of mutational events, largely confined to the H-chain
FIG. 1.7. Immunoglobulin Class Switching. The process through and L-chain V-region genes and their immediately sur-
which a given VDJ gene in a stimulated B cell may switch the con- rounding introns.
stant region gene with which it is associated from μ to another,
During the process of somatic hypermutation, muta-
such as γ1, is illustrated. A recombination event occurs in which
DNA between a cleavage point in Sμ and one in Sγ1 forms a circular tional rates of 1 per 1,000 base pairs per generation may be
episome. This results in Cγ1 being located immediately downstream achieved. This implies that with each cell division close to
of the chimeric Sμ/γ1 region, in a position such that transcription one mutation will occur in either the H- or L-chain V region
initiating upstream of VDJ results in the formation of VDJCγ1 mRNA of an individual cell. Such a high rate of mutation creates an
and γ1 H-chain protein. enormous increase in antibody diversity. Although most of
these mutations will either not affect the affinity with which
the antibody binds its ligand or will lower that affinity, some
region and in the deletion of the intervening DNA (Fig. 1.7). will increase it. Thus, some B cells emerge that can bind an-
The genes encoding the C regions of the various γ chains (in tigen more avidly than the initial population of responding
the human γ1, γ 2, γ 3, and γ4; in the mouse γ1, γ 2a, γ 2b, and cells. Because there is an active process of apoptosis in the
γ 3), of the α chain, and of the ε chain are located 3′ of the germinal center from which B cells can be rescued by the
Cμ and Cδ genes. binding of antigen to their membrane receptors, cells with
The induction of the switching process is dependent the most avid receptors should have an advantage over other
on the action of a specialized set of B-cell stimulants. Of antigen-specific B cells and should come to dominate the
these, the most widely studied are CD154, expressed on the population of responding cells. Thus, upon rechallenge, the
surface of activated T cells, and the TLR ligands such as affinity of antibody produced will be greater than that in the
bacterial lipopolysaccharide. The targeting of the C region initial response. As time after immunization elapses, the af-
that will be expressed as a result of switching is largely de- finity of antibody produced will increase. This process leads
termined by cytokines. Thus, IL-4 determines that switch to the presence in immunized individuals of high-affinity
events in the human and mouse will be to the ε C region antibodies that are much more effective, on a weight basis,
and to the γ4 (human) or γ1 (mouse) C regions. In the in protecting against microbial agents and other antigen-
mouse, interferon-gamma (IFN-γ) determines switching bearing pathogens than was the antibody initially produced.
to γ 2a and transforming growth factor-beta (TGF-β) de- Together with antibody class switching, affi nity maturation
termines switching to α . A major goal is to understand the results in the increased effectiveness of antibody in prevent-
physiologic determination of the specificity of the switching reinfection with agents with which the individual has
ing process. Because cytokines are often the key controllers had a prior encounter.
of which Ig classes will represent the switched isotype, this
logically translates into asking what regulates the relative
amounts of particular cytokines that are produced by dif-
T-LYMPHOCYTES
ferent modes of immunization. T-lymphocytes constitute the second major class of lympho-
As already noted, both the switching process and somatic cytes. They derive from precursors in hematopoietic tissue,
hypermutation depend upon the AID. Mice and humans undergo differentiation in the thymus (hence the name thy-
that lack AID fail to undergo both immunoglobulin class mus-derived [T]-lymphocytes), and are then seeded to the
switching and somatic hypermutation. peripheral lymphoid tissue and to the recirculating pool of
lymphocytes (see Chapters 13 and 14). T cells are subdivided
into two distinct classes based on the cell surface receptors
Affinity Maturation and Somatic Hypermutation they express. The majority of T cells express TCRs consist-
(Chapters 6 and 10) ing of α and β chains. A second group of T cells express re-
The process of generation of diversity embodied in the con- ceptors made up of γ and δ chains. Among the α / β T cells
struction of the H- and L-chain V-region genes and of the are two important sublineages: those that express the core-
pairing of H and L chains creates a large number of distinct ceptor molecule CD4 (CD4 T cells) and those that express
antibody molecules, each expressed in an individual B cell. CD8 (CD8 T cells). These cells differ in how they recognize

antigen and mediate different types of regulatory and effec- develop into cytotoxic T-lymphocytes (CTLs) capable of ef-
tor functions. ficiently lysing target cells that express antigens recognized
CD4 T cells are the major helper cells of the immune sys- by the CTLs.
tem (see Chapter 29). Their helper function depends both on
cell surface molecules, such as CD154, induced upon these
cells when they are activated and on the wide array of cy- T-Lymphocyte Antigen Recognition
tokines they secrete upon stimulation. CD4 T cells tend to (Chapters 11, 21, and 22)
differentiate, as a consequence of priming, into cells that T cells differ from B cells in their mechanism of antigen rec-
principally secrete the cytokines IL-4, IL-13, IL-5 and IL-6 ognition. Ig, the B-cell’s receptor, binds to individual anti-
(TH2 cells) into cells that mainly produce IFN-γ and lympho- genic epitopes on the surface of native molecules, be they on
toxin (TH1 cells), or into cells that produce IL-17 and related cell surfaces or in solution. Antibody and B-cell receptors
cytokines (TH17 cells). TH2 cells are very effective in immunity evolved to bind to and to protect against microorganisms in
to helminthic parasites, TH1 cells are effective inducers of extracellular fluids.
cellular immune responses, involving enhancement in the By contrast, T cells invariably recognize cell-associated
microbicidal activity of monocytes and macrophages and molecules and mediate their functions by interacting with
consequent increased efficiency in lysing microorganisms and altering the behavior of such antigen-presenting cells
in intracellular vesicular compartments, while TH17 cells are (APCs). Indeed, the TCR does not recognize antigenic
efficient recruiters of granulocytes and other cells of the in- determinants on intact, undenatured molecules. Rather, it
flammatory system and play a major role in responses to ex- recognizes a complex consisting of a peptide, derived by in-
tracellular bacterial pathogens. CD4 T cells can also acquire tracellular proteolysis of the antigen, bound into a special-
the capacity to enter B-cell follicles and help B cells develop ized groove of a class II or class I MHC protein. Indeed, what
into antibody-producing cells and undergo immunoglobulin differentiates a CD4 T cell from a CD8 T cell is that the CD4
class switching and affinity maturation; the cells are referred T cells recognize peptide/class II complexes whereas the
to as T follicular helper (Tfh) cells. Another possible fate of CD8 T cells recognize peptide/class I complexes.
naïve CD4 T cells is to differentiate into induced regulatory The TCR’s ligand (i.e., the peptide/MHC protein complex)
T cells (iTregs). However, most Tregs develop as a independent is created within the APC. In general, class II MHC molecules
lineage of CD4 T cells. Tregs express the transcription factor bind peptides derived from proteins that have been taken up
Foxp3 and many express large amounts of the α chain of the by the APC through an endocytic process (Fig. 1.8). These
IL-2 receptor (CD25). endocytosed proteins are fragmented by proteolytic enzymes
T cells mediate important effector functions. Some of within the endosomal/lysosomal compartment. The result-
these are determined by the patterns of cytokines they se- ing peptides are loaded into class II MHC that traffic through
crete. These powerful molecules can be directly toxic to this compartment. Peptide-loaded class II molecules are then
target cells and can mobilize potent inflammatory mecha- expressed on the surface of the APC where they are available
nisms. In addition, T cells, particularly CD8 T cells, can to be bound by CD4 T cells that have TCRs capable of recog-
FIG. 1.8. Pathways of Antigen Processing. Exogenous

antigen (Ea) enters the cell via endocytosis and is trans-
ported from early endosomes into late endosomes or
prelysosomes, where it is fragmented and where re-
sulting peptides (Ea-derived peptides) may be loaded
into class II major histocompatibility complex (MHC)
molecules. The latter have been transported from the
rough endoplasmic reticulum (RER) through the Golgi
apparatus to the peptide-containing vesicles. Class II
MHC molecules/Ea-derived peptide complexes are then
transported to the cell surface, where they may be rec-
ognized by T-cell receptor expressed on CD4+ T cells.
Cytoplasmic antigens (Ca) are degraded in the cytoplasm
and then enter the RER through a peptide transporter. In
the RER, Ca-derived peptides are loaded into class I MHC
molecules that move through the Golgi apparatus into
secretory vesicles and are then expressed on the cell
surface where they may be recognized by CD8+ T cells
(Reprinted with permission from Paul WE. In: Gallin JI,
Goldstein, I, Snyderman, R, ed. Inflammation. New York:
Raven,;1992:776.)

nizing the expressed cell surface peptide/MHC protein com- family members. The TCR is associated with a set of trans-
plex. Thus, CD4 T cells are specialized to largely react with membrane proteins, collectively designated the CD3 com-
antigens derived from extracellular sources. plex, that play a critical role in signal transduction. The CD3
In contrast, class I MHC molecules are mainly loaded with complex consists of γ, δ (note that the CD3 γ and δ chains
peptides derived from internally synthesized proteins, such and the TCR γ and δ chains are distinct polypeptides that,
as viral gene products. These peptides are produced from unfortunately, have similar designations) and ε chains and
cytosolic proteins by proteolysis within the proteasome and is associated with a homodimer of two ζ chains or a het-
are translocated into the rough endoplasmic reticulum. Such erodimer of ζ and η chains. CD3 γ, δ, and ε consist of extra-
peptides, generally nine amino acids in length, are bound by cellular domains that are Ig supergene family members. The
class I MHC molecules. The complex is brought to the cell cytosolic domains of CD3 γ, δ, and ε and of ζ and η contain
surface, where it can be recognized by CD8 T cells expressing one or more copies of the immunoreceptor tyrosine–based
appropriate receptors. This property gives the T-cell system, activation motif (ITAM) (D/ExxYxxLxxxxxxxYxxL/I) that
particularly CD8 T cells, the ability to detect cells express- is found in a variety of chains associated with immune rec-
ing proteins that are different from, or produced in much ognition receptors. This motif appears to be important in
larger amounts than, those of cells of the remainder of the the signal transduction process and provides a site through
organism (e.g., viral antigens [whether internal, envelope, which protein tyrosine kinases can interact with these
or cell surface] or mutant antigens [such as active oncogene chains to propagate signaling events.
products]), even if these proteins, in their intact form, are The TCR chains are organized much like Ig chains. Their
neither expressed on the cell surface nor secreted. N-terminal portions are variable and their C-terminal por-
Although this division of class I–binding peptides being tions are constant. Furthermore, similar recombinational
derived from internally synthesized proteins and class II– mechanisms are used to assemble the V-region genes of the
binding peptides from imported proteins is generally cor- TCR chains. Thus, the V region of the TCR β chain is en-
rect, there are important exceptions to this rule that are coded by a gene constructed from three distinct genetic el-
central for the function of the immune system. The most ements (Vb , D, and Jb ) that are separated in the germline.
effective priming of naive CD8 T cells occurs in response to Although the relative numbers of Vb , D, and Jb genes differ
peptide/MHC-I complexes expressed by DCs and yet many from that for the comparable Ig H variable region elements,
viruses do not infect these cells but rather target other cell the strategies for creation of a very large number of distinct
types. Viral antigens produced by infected cells can be taken genes by combinatorial assembly are the same. Both junc-
up by specialized DCs and loaded into class I molecules in a tional diversity and N-region addition further diversify the
process referred to as cross-presentation. genes and their encoded products. TCR β has fewer V genes
than IgH but much more diversity centered on the D/J re-
gion, which encodes the equivalent of the third CDR of Igs.
T-Lymphocyte Receptors (Chapter 11) The α chain follows similar principles, except that it does
The TCR is a disulfide-linked heterodimer (Fig. 1.9). Its not use a D gene.
constituent chains (α and β, or γ and δ) are Ig supergene The genes for TCR γ and δ chains are assembled in a simi-
lar manner except that they have many fewer V genes from
which to choose. Indeed, γ /δ T cells in certain environments,
such as the skin and specific mucosal surfaces, are excep-
tionally homogeneous. It has been suggested that the TCRs
encoded by these essentially invariant γ and δ chains may
be specific for some antigen that signals microbial invasion
and that activation of γ /δ T cells through this mechanism
constitutes an initial response that aids the development of
the more sophisticated response of α / β T cells.
T-Lymphocyte Activation (Chapter 12)

T-cell activation is dependent on the interaction of the TCR/
CD3 complex with its cognate ligand, a peptide bound in the
groove of a class I or class II MHC molecule, on the surface
of a competent APC. Through the use of chimeric cell sur-
face molecules that possess cytosolic domains largely limited
to the ITAM signaling motif alluded to previously, it is clear
that cross-linkage of molecules containing such domains
FIG. 1.9. The T-Cell Antigen Receptor. Illustrated schematically is can generate some of the signals that result from TCR en-
the antigen binding subunit comprised of an αβ heterodimer and the
associated invariant CD3 and ζ chains. Acidic (−) and basic (+) resi- gagement. Nonetheless, the molecular events set in motion
dues located within the plasma membrane are indicated. The open by receptor engagement are complex ones. Among the earli-
rectangular boxes indicate motifs within the cytoplasmic domains est steps is the activation of tyrosine kinases leading to the
that interact with protein tyrosine kinases. tyrosine phosphorylation of a set of substrates that control

several signaling pathways. Current evidence indicates that CD80/86 has a second receptor on the T cell, CTLA-4,
early events in this process involve the src family tyrosine that is expressed later in the course of T-cell activation. The
kinases p56lck, associated with the cytosolic domains of the bulk of evidence indicates that the engagement of CTLA-4
CD4 and CD8 coreceptors, and p59fyn, and ZAP-70, a Syk by CD80/86 leads to a set of biochemical signals that termi-
family tyrosine kinase, that binds to the phosphorylated nate the T-cell response. Mice that are deficient in CTLA-4
ITAMs of the ζ chain. The protein tyrosine phosphatase expression develop fulminant autoimmune responses and
CD45, found on the surface of all T cells, also plays a critical anti–CTLA-4 antibodies are used as drugs to enhance anti-
role in T-cell activation. tumor immune responses.
A series of important substrates are tyrosine phosphory-
lated as a result of the action of the kinases associated with
T Lymphocyte Development (Chapter 13)
the TCR complex. These include a 1) set of adapter proteins
that link the TCR to the Ras pathway; 2) phospholipase Cγ1, Upon entry into the thymus, T-cell precursors do not express
the tyrosine phosphorylation of which increases its catalytic TCR chains, the CD3 complex, or the CD4 or CD8 molecules
activity and engages the inositol phospholipid metabolic (Fig. 1.10). Because these cells lack both CD4 and CD8, they
pathway, leading to elevation of intracellular free calcium are often referred to as double-negative cells. Thymocytes
concentration and activation protein kinase C; and 3) a series develop from this double-negative pool into cells that are
of other important enzymes that control cellular growth and both CD4 + and CD8 + (double-positive cells) and express
differentiation. Particularly important is the phosphorylation low levels of TCR and CD3 on their surface. In turn, dou-
of LAT, a molecule that acts as an organizing scaffold to which ble-positive cells further differentiate into relatively mature
a series of signaling intermediates bind and upon which they thymocytes that express either CD4 or CD8 (single-positive
become activated and control downstream signaling. cells) and high levels of the TCR/CD3 complex.
The recognition and early activation events result in the The expression of the TCR depends on complex rear-
reorganization of cell surface and cytosolic molecules on rangement processes that generate TCR α and β (or γ and
the T cell, and correspondingly, on the APC to produce a
structure, the immunological synapse. The apposition of
key interacting molecules involving a small segment of the
membranes of the two cells concentrates these interacting
molecules in a manner that both strengthens the interac-
tion between the cells and intensifies the signaling events. It
also creates a limited space into which cytokines may be se-
creted to influence the behavior of the interacting cells. The
formation of the immunological synapse is one mechanism
through which the recognition of relatively small numbers
of ligands by TCRs on a specific T cell can be converted into
a vigorous stimulatory process.
In general, normal T cells and cloned T-cell lines that are
stimulated only by TCR cross-linkage fail to give complete
responses. TCR engagement by itself may often lead to a re-
sponse in which the key T-cell–derived growth factor, IL-2,
is not produced and in which the cells enter a state of an-
ergy such that they are unresponsive or poorly responsive
to a subsequent competent stimulus (see Chapter 32). Full
responsiveness of a T cell requires, in addition to receptor
engagement, an accessory cell–delivered costimulatory sig-
nal. The engagement of CD28 on the T cell by CD80 and/
or CD86 on the APC (or the engagement of comparable FIG. 1.10. Development of `/a T Cells in the Thymus. Double-
ligand/receptor pairs on the two cells) provides potent co- negative T cells (4−8−) acquire CD4 and CD8 (4+8+) and then express
stimulatory activity. Inhibitors of this interaction mark- α/β TCRs, initially at low levels. Thereafter, the degree of expression
of T-cell receptiors increases and the cells differentiate into CD4 or
edly diminish antigen-specific T-cell activation in vivo and
CD8 cells and are then exported to the periphery. Once the T cells
in vitro, indicating that the CD80/86–CD28 interaction is have expressed receptors, their survival depends upon the recog-
physiologically important in T-cell activation (see Chapters nition of peptide/major histocompatibility complex (MHC) class I or
12 and 14). class II molecules with an affinity above some given threshold. Cells
The interaction of CD80/86 with CD28 increases cytokine that fail to do so undergo apoptosis. These cells have failed to be
production by the responding T cells. For the production of positively selected. Positive selection is associated with the differ-
IL-2, this increase appears to be mediated both by enhanc- entiation of 4+8+ cells into CD4 or CD8 cells. Positive selection involv-
ing peptide/class I MHC molecules leads to the development of CD8
ing the transcription of the IL-2 gene and by stabilizing IL-2 cells whereas positive selection involving peptide/class II MHC mol-
mRNA. These dual consequences of the CD80/86–CD28 in- ecules leads to the development of CD4 cells. If a T cell recognizes
teraction cause a striking increase in the production of IL-2 by a peptide/MHC complex with high affinity, it is also eliminated via
antigen-stimulated T cells. apoptosis (it is negatively selected).

δ) chains. Once TCR chains are expressed, these cells un- membrane Ig of B cells or do so inefficiently. B cells bind the
dergo two important selection processes within the thymus. antigen through their membrane Ig, and the complex un-
One, termed negative selection, is the deletion of cells that dergoes endocytosis. Within the endosomal and lysosomal
express receptors that bind with high affinity to complexes compartments, the antigen is fragmented into peptides by
of self-peptides with self-MHC molecules. This is a major proteolytic enzymes and one or more of the generated pep-
mechanism through which the T-cell compartment devel- tides are loaded into class II MHC molecules, which traffic
ops immunologic unresponsiveness to self antigens (see through this vesicular compartment. The resulting complex
Chapters 13 and 32). In addition, a second major selection of class II MHC molecule and bound peptide is exported to
process is positive selection, in which T cells with receptors the B-cell surface membrane. T cells with receptors specific
with “intermediate affinity” for self-peptides bound to self- for the peptide/class II molecular complex recognize that
MHC molecules are selected, thus forming the basis of the complex on the B cell.
T-cell repertoire for foreign peptides associated with self- B-cell activation depends not only on the binding of pep-
MHC molecules. T cells that are not positively selected are tide/class II MHC complexes on the B cell surface by the
eliminated in the thymic cortex by apoptosis. Similarly, T TCR but also on the interaction of T-cell CD154 with CD40
cells that are negatively selected as a result of high-affinity on the B cell. T cells do not constitutively express CD154;
binding to self-peptide/self-MHC complexes are also deleted rather, it is induced as a result of an interaction with an ac-
through apoptotic death. These two selection processes re- tivated APC that expresses a cognate antigen recognized by
sult in the development of a population of T cells that are bi- the TCR of the T cell. Furthermore, CD80/86 are generally
ased toward the recognition of peptides in association with expressed by activated but not resting B cells so that inter-
self-MHC molecules from which those cells that are poten- actions involving resting B cells and naïve T cells generally
tially autoreactive (capable of high-affinity binding of self- do not lead to efficient antibody production. By contrast, a
peptide/self-MHC complexes) have been purged. T cell already activated and expressing CD154 can interact
One important event in the development of T cells is with a resting B cell, leading to its upregulation of CD80/86
their differentiation from double-positive cells into CD4 + or and to a more productive T-cell/B-cell interaction with the
CD8 + single-positive cells. This process involves the inter- delivery of cognate help and the development of the B cell
action of double-positive thymocytes with peptide bound to into an antibody-producing cell. Similarly, activated B cells
class II or class I MHC molecules on accessory cells. Indeed, expressing large amounts of class II molecules and CD80/86
CD4 binds to monomorphic sites on class II molecules, can act as effective APCs and can participate with T cells in
whereas CD8 binds to comparable sites on class I molecules. efficient cognate help interactions. Cross-linkage of mem-
The capacity of the TCR and CD4 (or of the TCR and CD8) brane Ig on the B cell, even if inefficient, may synergize
to bind to a class II MHC (or a class I MHC) molecule on an with the CD154/CD40 interaction to yield vigorous B-cell
accessory cell leads either to the differentiation of double- activation.
positive thymocytes into CD4 + (or CD8+) single-positive T The subsequent events in the B-cell response program, in-
cells or to the selection of cells that have “stochastically” dif- cluding proliferation, Ig secretion, and class switching, either
ferentiated down the CD4 (or CD8) pathway. depend on or are enhanced by the actions of T-cell–derived
Less is understood about the differentiation of thymo- cytokines. Thus, B-cell proliferation and Ig secretion are en-
cytes that express TCRs composed of γ /δ chains. These hanced by the actions of several type I cytokines including
cells fail to express either CD4 or CD8. However, γ /δ cells IL-2 and IL-4. Ig class switching is dependent both on the ini-
are relatively numerous early in fetal life; this, together with tiation of competence for switching, which can be induced
their limited degree of heterogeneity, suggests that they may by the CD154/CD40 interaction, and on the targeting of par-
comprise a relatively primitive T-cell compartment. ticular C regions for switching, which is determined, in many
instances, by cytokines. The best studied example of this is
the role of IL-4 in determining switching to IgG1 and IgE in
T-Lymphocyte Functions (Chapters 14, 29, and 33) the mouse and to IgG4 and IgE in the human. Indeed, the
T cells mediate a wide range of immunologic functions. central role of IL-4 in the production of IgE is demonstrated
These include the capacity to help B cells develop into an- by the fact that mice that lack the IL-4 gene or the gene for the
tibody-producing cells, the capacity to increase the micro- IL-4 receptor α chain, as a result of gene knockouts, have a
bicidal action of monocytes/macrophages, the inhibition of marked defect in IgE production. Similarly, IFN-γ determines
certain types of immune responses, direct killing of target switching to IgG2a in the mouse. The relationship between
cells, and mobilization of the inflammatory response. In gen- TFH cells that produce IL-4 and TH2 cells and those that pro-
eral, these effects depend on their expression of specific cell- duce IFN-γ and TH1 cells is still uncertain.
surface molecules and the secretion of cytokines.
Induction of Cellular Immunity (Chapters 14 and 19)
T Cells that Help Antibody Responses (Chapter 13) T cells also may act to enhance the capacity of monocytes
Helper T cells, TFH cells, can stimulate B cells to make an- and macrophages to destroy intracellular microorganisms.
tibody responses to proteins and other T-cell–dependent In particular, IFN-γ enhances several mechanisms through
antigens. T-cell–dependent antigens are immunogens in which mononuclear phagocytes destroy intracellular bac-
which individual epitopes appear only once or only a limited teria and parasites, including the generation of nitric oxide
number of times so that they are unable to cross-link the and induction of TNF. TH1 cells are particularly effective in

enhancing microbicidal action because they produce IFN-γ. ligand on the surface of the CTL with fas on the surface of
By contrast, three of the major cytokines produced by TH2 the target cell initiates apoptosis in the target cell.
cells, IL-4, IL-13, and IL-10, block these activities; IL-4 and CTL-mediated lysis is a major mechanism for the destruc-
IL-13 induce an alternative gene activation program in mac- tion of virus-infected cells. If activated during the period in
rophages resulting in alternatively activated macrophages, which the virus is in its eclipse phase, CTLs may be capable
characterized (in the mouse) by the expression of arginase of eliminating virus and curing the host with relatively lim-
1 and chitinase. Thus, TH2 cells often oppose the action of ited cell destruction. On the other hand, vigorous CTL activ-
TH1 cells in inducing cellular immunity and in certain infec- ity after a virus has been widely disseminated may lead to
tions with microorganisms that are intracellular pathogens substantial tissue injury because of the large number of cells
of macrophages; a TH2-dominated response may be associ- that are killed by the action of the CTLs. Thus, in many in-
ated with failure to control the infection. fections, the disease is caused by the destruction of tissue by
CTLs rather than by the virus itself. One example is hepatitis
Regulatory T Cells (Chapter 33) B, in which much of the liver damage represents the attack of
There has been a longstanding interest in the capacity of T hepatitis B virus–specific CTLs on infected liver cells.
cells to diminish as well as to help immune responses. Cells It is usually observed that CTLs that have been induced
that mediate such effects are referred to as Tregs. Tregs may be as a result of a viral infection or intentional immunization
identified by their expression of Foxp3 and of CD25, the IL-2 must be reactivated in vitro through the recognition of anti-
receptor alpha chain. These cells inhibit the capacity of both gen on the target cell. This is particularly true if some inter-
CD4 and CD8 T cells to respond to their cognate antigens. val has elapsed between the time of infection or immuniza-
The mechanisms through which their suppressor function tion and the time of test. This has led to some question being
is mediated are still somewhat controversial. In some in- raised as to the importance of CTL immunity in protection
stances, it appears that cell/cell contact is essential for sup- against re-infection and how important CTL generation is
pression, whereas in other circumstances production of cy- in the long-term immunity induced by protective vaccines.
tokines by Tregs has been implicated in their ability to inhibit On the other hand, in active infections, such as seen in in-
responses. Evidence has been presented for both IL-10 and dividuals with HIV, CTL that can kill their target cells im-
TGF-β as mediators of inhibition. mediately are often seen. There is much evidence to suggest
Tregs have been particularly studied in the context of vari- that these cells play an active role in controlling the number
ous autoimmune conditions. In the absence of Tregs, conven- of HIV-positive T cells.
tional T cells cause several types of autoimmune responses,
including autoimmune gastritis and inflammatory bowel
disease. Tregs express cell surface receptors allowing them to CYTOKINES (CHAPTERS 25 TO 28)
recognize autoantigens; their responses to such recognition Many of the functions of cells of the immune system are
results in the suppression of responses by conventional T cells. mediated through the production of a set of small proteins
Whether the receptor repertoire of Tregs and the conventional referred to as cytokines. These proteins can now be divided
T cells are the same has not been fully determined, although into several families. They include the type I cytokines or
there is increasing evidence that Tregs derive from a thymic hematopoeitins that encompass many of the interleukins
CD4 T-cell population with relatively high affinity for self- (i.e., IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, IL-9, IL-11, IL-12, IL-
antigen. As noted previously, iTregs can be derived in the pe- 13, IL-15, IL-21, IL-23, and IL-27), as well as several hemato-
riphery from naive CD4 T-cell populations. This is seen when poietic growth factors; the type II cytokines, including the
naive cells are stimulated by their cognate ligands in the pres- interferons and IL-10; the TNF-related molecules, including
ence of TGF-β and IL-2. TNF, lymphotoxin, and Fas ligand; Ig superfamily members,
including IL-1, IL-18, IL-33, IL-36, and IL-37; and the che-
Cytotoxic T Cells (Chapter 37) mokines, a large family of molecules playing critical roles in
One of the most striking actions of T cells is the lysis of cells a wide variety of immune and inflammatory functions. IL-17
expressing specific antigens. Most cells with such cytotoxic and its congeners, including IL-25, constitute a structurally
activity are CD8 T cells that recognize peptides derived from unique set of cytokines.
proteins produced within the target cell and bound to class Many of the cytokines are T-cell products; their production
I MHC molecules expressed on the surface of the target cell. represents one of the means through which the wide variety of
However, CD4 T cells can express CTL activity, although functions of T cells are mediated. Most cytokines are not con-
in such cases the antigen recognized is a peptide associated stitutive products of the T cell. Rather, they are produced in re-
with a class II MHC molecule; often, such peptides derive sponse to T-cell activation, usually resulting from presentation
from exogenous antigens. of antigen to T cells by APCs in concert with the action of a co-
There are two major mechanisms of cytotoxicity. One stimulatory molecule, such as the interaction of CD80/86 with
involves the production by the CTL of perforin, a molecule CD28. Although cytokines are produced in small quantities,
that can insert into the membrane of target cells and pro- they are potent, binding to their receptors with equilibrium
mote the lysis of that cell. Perforin-mediated lysis is medi- constants of approximately 1010 M-1. In some instances, cyto-
ated a series of enzymes produced by activated CTLs, re- kines are directionally secreted into the immunological syn-
ferred to as granzymes. Many active CTLs also express large apse formed between a T cell and an APC. In such cases, the
amounts of fas ligand on their surface. The interaction of fas cytokine acts in a paracrine manner. Indeed, many cytokines

have limited action at a distance from the cell that produced encoding these molecules show an unprecedented degree of
them. This appears to be particularly true of many of the type polymorphism. This, together with their critical role in an-
I cytokines. However, other cytokines act by diffusion through tigen presentation, explains their central role as the target of
extracellular fluids and blood to target cells that are distant the immune responses leading to the rejection of organ and
from the producers. Among these are cytokines that have pro- tissue allografts.
inflammatory effects, such as IL-1, IL-6, and TNF, and the The MHC also includes other genes, particularly genes
chemokines, that play important roles in regulating the mi- for certain complement components. In addition, genes for
gration of lymphocytes and other cell types. the cytokines TNF-α and lymphotoxin (also designated
TNF-β) are found in the MHC.
Chemokines (Chapter 28)
A large family of small proteins that are chemotactic cyto- Class I MHC Molecules (Chapter 21)
kines (chemokines) have been described. While members Class I MHC molecules are membrane glycoproteins ex-
of this family have a variety of functions, perhaps the most pressed on most cells. They consist of an α chain of ap-
dramatic is their capacity to regulate leukocyte migration proximately 45,000 daltons noncovalently associated with
and thus to act as critical dynamic organizers of cell dis- α2-microglobulin, a 12,000-dalton molecule (Fig. 1.11). The
tribution in the immune and inflammatory responses. The gene for the α chain is encoded in the MHC, whereas that
receptors for chemokines are seven transmembrane-span- for β2-microglobulin is not. Both the α chain and β2-micro-
ning, G-protein coupled receptors. globulin are Ig supergene family members. The α chain is
The chemokines are subdivided based on the number highly polymorphic, with the polymorphisms found mainly
and positioning of their highly conserved cysteines. Among in the regions that constitute the binding sites for antigen-
chemokines with four conserved cysteines, the cysteines are derived peptides and that are contact sites for the TCR.
adjacent in one large group (the CC chemokines), whereas The class I α chain consists of three extracellular regions
in a second large group they are separated by one amino or domains, each of similar length, designated α1, α2, and
acid (CXC chemokines). There are also rare chemokines α3. In addition, α chains have a membrane-spanning do-
in which the cysteines are separated by three amino acids main and a short carboxy-terminal cytoplasmic tail. The
(CX3C) or in which there are only two conserved cysteines
(C chemokines).
Individual chemokines may signal through more than
one chemokine receptor, and individual receptors may in-
teract with more than one chemokine, producing a complex
set of chemokine/chemokine receptor pairs and providing
opportunities for exceedingly fine regulation of cellular
functions.
THE MAJOR HISTOCOMPATIBILITY COMPLEX AND

ANTIGEN PRESENTATION (CHAPTERS 21 AND 22)
The MHC has already been introduced in this chapter in the
discussion of T-cell recognition of antigen-derived peptides
bound to specialized grooves in class I and class II MHC
proteins. Indeed, the class I and class II MHC molecules are
essential to the process of T-cell recognition and response.
Nonetheless, they were first recognized not for this reason but
because of the dominant role that MHC class I and class II
proteins play in transplantation immunity (see Chapter 46).
When the genetic basis of transplantation rejection be-
tween mice of distinct inbred strains was sought, it was rec-
ognized that although multiple genetic regions contributed
to the rejection process, one region played a dominant role.
Differences at this region alone would cause prompt graft
rejection, whereas any other individual difference usually FIG. 1.11. Model of the Class I HLA-A2 Molecule. A schematic rep-
resulted in a slow rejection of foreign tissue. For this reason, resentation of the structure of the HLA-A2 class I major histocom-
the genetic region responsible for prompt graft rejection was patibility complex (MHC) molecule. The polymorphic α1 and α2 do-
termed the major histocompatibility complex. mains are at the top. They form a groove into which antigen-derived
peptides fit to form the peptide/MHC class I complex that is recog-
In all higher vertebrates that have been thoroughly nized by T-cell receptors of CD8 T cells. (Reprinted from Bjorkman
studied, a comparable MHC exists. The defining features of PJ, Saper MA, Samraoui B, Bennett WS, Strominger JL, Wiley DC.
the MHC are the transplantation antigens that it encodes. Structure of the human class II histocompatibility antigen, HLA-A2.
These are the class I and class II MHC molecules. The genes Nature. 1987;329:506–512).

crystal structure of class I molecules indicates that the α1 cause expression of class II molecules on many cell types that
and α2 domains form a site for the binding of peptides de- normally lack these cell surface molecules. Interferons also can
rived from antigens. This site is defi ned by a floor consisting cause striking upregulation in the expression of class I MHC
of β sheets and bounded by α-helical walls. The polymor- molecules. Thus, immunologically mediated inflammation
phisms of the class I molecule are mainly in these areas. may result in aberrant expression of class II MHC molecules
In the human, three loci encoding classical class I mol- and heightened expression of class I molecules. Such altered
ecules have been defined; these are designated human leu- expression of MHC molecules can allow cells that do not nor-
kocyte antigen (HLA)-A, HLA-B, and HLA-C. All display mally function as APCs for CD4 T cells to do so and enhances
high degrees of polymorphism. A similar situation exists the sensitivity of such cells to CD8 T cells. This has impor-
in the mouse. In addition, there are a series of genes that tant consequences for immunopathologic responses and for
encode class I–like molecules (class Ib molecules). Some of autoimmunity.
these have been shown to have antigen-presenting activity
for formylated peptides, suggesting that they may be spe-
cialized to present certain prokaryotic antigens. The class Ib Antigen Presentation (Chapter 22)
molecule CD1 has been shown to have antigen-presenting As already discussed, the function of class I and class II
function for mycobacterial lipids, providing a mechanism MHC molecules is to bind and present antigen-derived pep-
through which T cells specific for such molecules can be tides to T cells whose receptors can recognize the peptide/
generated. CD1d, presenting certain endogenous or exog- MHC complex that is generated. There are two major types
enous phospholipids, is recognized by a novel class of T cells of antigen-processing pathways, specialized to deal with dis-
(NK T cells) that produce large amounts of cytokines upon tinct classes of pathogens that the T cell system must con-
immediate stimulation. front (see Fig. 1.8).
Extracellular bacteria and extracellular proteins may enter
APCs by endocytosis or phagocytosis. Their antigens and the
Class II MHC Molecules (Chapter 21) antigens of bacteria that live within endosomes or lysosomes
Class II MHC molecules are heterodimeric membrane gly- are fragmented in these organelles, and peptides derived from
coproteins. Their constituent chains are designated α and the antigen are loaded into class II MHC molecules as these
β; both chains are immunoglobulin supergene family mem- proteins traverse the vesicular compartments in which the
bers, and both are encoded within the MHC. Each chain peptides are found. The loading of peptide is important in
consists of two extracellular domains (α1 and α2; β1 and stabilizing the structure of the class II MHC molecule. The
β2, respectively), a hydrophobic domain, and a short cyto- acidic pH of the compartments in which loading occurs fa-
plasmic segment. The overall conformation of class II MHC cilitates the loading process. Once the peptide-loaded class II
molecules appears to be quite similar to that of class I mol- molecules reaches neutral pH, such as at the cell surface, the
ecules. The peptide-binding site of the class II molecules is peptide/MHC complex is stable. Peptide dissociation from
contributed to by the α1 and β1 domains (see Fig. 1.5); it is such class II molecules is very slow, with a half-time mea-
within these domains that the majority of the polymorphic sured in hours. The peptide/class II complex is recognized
residues of class II molecules are found. by those CD4 T cells that have complementary receptors. As
A comparison of the three-dimensional structures of already pointed out, the specialization of CD4 T cells to rec-
class I and class II molecules indicates certain distinctive ognize peptide/class II complexes is partly due to the affin-
features that explain differences in the length of peptides ity of the CD4 molecule for monomorphic determinants on
that the two types of MHC molecules can bind. Class I mol- class II molecules. Obviously, this form of antigen processing
ecules generally bind peptides with a mean length of nine can only apply to cells that express class II MHC molecules.
amino acids, whereas class II molecules can bind substan- Indeed, APCs for CD4 T cells principally include cells that
tially larger peptides. normally express class II MHC molecules, namely DCs, B
In the mouse, class II MHC molecules are encoded by cells, and macrophages.
genes within the I region of the MHC. These molecules are T cells also can recognize proteins that are produced
often referred to as I region–associated (Ia) antigens. Two within the cell that presents the antigen. The major patho-
sets of class II molecules exist, designated I-A and I-E, re- gens recognized by this means are viruses and other obligate
spectively. The α and β chains of the I-A molecules (Aα and intracellular (nonendosomal/nonlysosomal) microbes that
Aβ) pair with one another, as do the α and β chains of I-E have infected cells. In addition, proteins that are unique to
(Eα and Eβ). tumors, such as mutant oncogenes, or are overexpressed in
In the human, there are three major sets of class II mol- tumors also can be recognized by T cells. Endogenously pro-
ecules, encoded in the DR, DQ, and DP regions of the HLA duced proteins are fragmented in the cytosol by proteases
complex. in the proteasome. The resultant peptides are transported
Class II molecules have a more restricted tissue distribu- into the rough endoplasmic reticulum through the action
tion than class I molecules. Class II molecules are found on B of a specialized transport system. These peptides are then
cells, DCs, epidermal Langerhans cells, macrophages, thymic available for loading into class I molecules. In contrast to
epithelial cells, and, in the human, activated T cells. Levels of the loading of class II molecules, which is facilitated by the
class II molecule expression are regulated in many cell types acid pH of the loading environment, the loading of class I
by interferons and in B cells by IL-4. Indeed, interferons can molecules is controlled by interaction of the class I α chain

with β2-microglobulin. Thus, the bond between peptide Indeed, this process also occurs within the thymus in the
and class I molecule is generally weak in the absence of generation of the T-cell repertoire, as already discussed. T
β2-microglobulin, and the binding of β2-microglobulin cells developing within the thymus undergo a positive selec-
strikingly stabilizes the complex. (Similarly, the binding of tion event in which those T cells capable of recognizing MHC
β2-microglobulin to the α chain is markedly enhanced by molecules displayed within the thymus are selected (and the
the presence of peptide in the α chain groove.) The peptide- remainder undergo programmed cell death). This leads to
loaded class I molecule is then brought to the cell surface. In the skewing of the population of T cells that emerges from
contrast to peptide-loaded class II molecules that are recog- the thymus so that the cells are specialized to respond to pep-
nized by CD4 T cells, peptide-loaded class I molecules are tides on self-MHC molecules. One of the unsolved enigmas of
recognized by CD8 T cells. This form of antigen process- positive selection within the thymus is how the vast array of
ing and presentation can be performed by virtually all cells T cells with receptors capable of reacting with a very large set
because, with a few exceptions, class I MHC molecules are of foreign peptides associated with self-MHC molecules are
universally expressed. chosen by self-MHC molecules that can only display self-pep-
Although the specialization of class I molecules to bind tides. It is believed that a high degree of cross-reactivity may
and present endogenously produced peptides and of class exist so that T cells selected to bind a given class I (or class II)
II molecules to bind and present peptides derived from ex- molecule plus a particular self-peptide can also bind a set of
ogenous antigens is generally correct, there are exceptions, other (foreign) peptides bound to the same MHC molecule.
many of which have physiologic importance. Particularly Furthermore, the affinity of an interaction required for
important is the capacity of some DCs to load peptides from positive selection in the thymus appears to be considerably
exogenous antigens into class I MHC molecules, allowing lower that that required for full activation of peripheral T
sensitization of CD8 T cells to the antigens of pathogens that cells. Thus, thymocytes selected by a given self-peptide/
infect cells other than DCs. self-MHC complex will generally not mount a full response
when they encounter the same peptide/MHC complex in
the periphery, although they will respond to a set of foreign
T-Lymphocyte Recognition of Peptide/MHC Complexes peptide/MHC complexes to which they bind with higher af-
Results in MHC-Restricted Recognition (Chapter 11) finity. Recognition of the self-peptide/self-MHC complex in
Before the biochemical nature of the interaction between an- the periphery nonetheless is important in sustaining the vi-
tigen-derived peptides and MHC molecules was recognized, it ability of resting T-lymphocytes.
was observed that T-cell responses displayed MHC-restricted Our modern understanding of T-cell recognition also aids
antigen recognition. Thus, if individual animals were primed in explaining the phenomenon of immune response gene con-
to a given antigen, their T cells would be able to recognize and trol of specific responses. In many situations, the capacity to
respond to that antigen only if the APCs that presented the recognize simple antigens can be found in only some members
antigen shared MHC molecules with the animal that had been of a species. In most such cases, the genes that determine the ca-
immunized. The antigen would not be recognized when pre- pacity to make these responses have been mapped to the MHC.
sented by APCs of an allogeneic MHC type. This can now be Such immune response gene control of immune responses is
explained by the fact that the TCR recognizes peptide bound based on the capacity of different class II MHC molecules (or
to an MHC molecule. MHC molecules display high degrees class I MHC molecules) to bind different sets of peptides. Thus,
of polymorphism, and this polymorphism is concentrated in for simple molecules, it is likely that peptides can be generated
the regions of the class I and class II molecules that interact that are only capable of binding to some of the polymorphic
with the peptide and that can bind to the TCR. Differences in MHC molecules of the species. Only individuals that possess
structure of the MHC molecules derived from different indi- those allelic forms of the MHC will be able to respond to those
viduals (or different inbred strains of mice) profoundly affect antigens. Based on this, some individuals are nonresponders
the recognition process. Two obvious explanations exist to ac- because of the failure to generate a peptide/MHC molecule
count for this. First, the structure of the grooves in different complex that can be recognized by the T-cell system.
class I or class II MHC molecules may determine that a differ- This mechanism also may explain the linkage of MHC type
ent range of peptides are bound or, even if the same peptide with susceptibility to various diseases. Many diseases show a
is bound, may change the conformation of the surface of the greater incidence in individuals of a given MHC type. These
peptide presented to the TCR. Second, polymorphic sites on include reactive arthritides, gluten-sensitive enteropathy, in-
the walls of the α-helices that are exposed to the TCR can sulin-dependent diabetes mellitus, and rheumatoid arthritis
either enhance or diminish binding of the whole complex, de- (see Chapter 44). One explanation is that the MHC type that
pending on their structure. Thus, priming an individual with is associated with increased incidence may convey altered re-
a given antigen on APCs that are syngeneic to the individual sponsiveness to antigens of agents that cause or exacerbate the
will elicit a response by T cells whose TCRs are specific for disease. Indeed, it appears that many of these diseases are due
a complex consisting of a peptide derived from the antigen to enhanced or inappropriate immune responses.
and the exposed polymorphic residues of the MHC molecule.
When the same antigen is used with APCs of different MHC
type, it is unlikely that the same peptide/MHC surface can be Antigen-Presenting Cells (Chapter 16)
formed, and thus the primed T cells are not likely to bind and T cells recognize peptide/MHC complexes on the surface of
respond to such stimulation. other cells. Such cells are often referred to as APCs. Although

effector cells can mediate their functions by recognizing Monocytes and Macrophages (Chapter 19)
such complexes on virtually any cell type, naïve cells are Cells of the monocyte/macrophage lineage play a central role
most efficiently activated by a set of specialized APC, the in immunity. One of the key goals of cellular immunity is to
DCs. DCs are a multimember family with distinctive loca- aid the macrophages in eliminating organisms that have es-
tions and functions. Among them are the plasmacytoid DCs tablished intracellular infections. In general, non-activated
that are the principal source of type I interferons in viral macrophages are inefficient in destroying intracellular mi-
infections. crobes. However, the production of IFN-γ and other media-
In general, in their immature form, DCs are resident in the tors by T cells can enhance the capacity of macrophages to
tissues where they are efficient at capturing and endocytos- eliminate such microorganisms. Several mechanisms exist
ing antigen. Their antigen capture activity is dependent upon for this purpose, including the development of reactive
expression of several surface receptors including Fc receptors, forms of oxygen, the development of nitric oxide, and the
receptors for heat shock proteins, and C-type lectins. If they induction of a series of proteolytic enzymes, as well as the
receive signals, such as various inflammatory stimuli, often induction of cytokine production. Macrophages can act as
mediated by TLRs, they downregulate the expression of these APCs and thus can enlist the “help” of activated, cytokine-
molecules but increase their expression of surface MHC mol- producing CD4 + T cells in regulating their function.
ecules and various costimualtory molecules such as CD80/86. Although macrophages function as APCs for attracting
In addition, such stimulation induces expression of chemo- activated T cells, they do not appear to be particularly ef-
kine receptors such as CCR2 and CCR7. The latter allow fective in the activation of naïve CD4 T cells. In instances in
stimulated DCs to follow signals from the chemokines SLC which they are the site of infection or have phagocytosed in-
and ELC, and to migrate into the T-cell zone of lymph nodes. fectious agents or their proteins, antigens from these agents
As part of the maturation process, they may also acquire the may be transferred to DCs. In such cases, the DCs would
capacity to produce cytokines and express surface molecules be the principal APCs that activate naïve or possibly resting
that can aid in determining the polarization of T-cell prim- memory CD4 T cells. Such activated T cells would then be
ing. This includes the production of IL-12, IL-23, IL-6, and available to help infected macrophages.
IL-10, and the expression of inducible costimulator ligand and
of Notch ligands. Interaction of naïve T cells with immature
DCs may induce a state of peripheral tolerance. Natural Killer Cells (Chapter 17)
NK cells play an important role in the immune system.
Indeed, in mice that lack mature T and B cells due to the scid
EFFECTOR MECHANISMS OF IMMUNITY mutation, the NK system appears to be highly active and to
The ultimate purpose of the immune system is to mount re- provide these animals a substantial measure of protection
sponses that protect the individual against infections with against infection. NK cells are closely related to T cells. They
pathogenic microorganisms by eliminating these microbes lack conventional TCR (or Ig) but express two classes of re-
or, where it is not possible to eliminate infection, to control ceptors. They have a set of activating receptors that allow
their spread and virulence. In addition, the immune system them to recognize features associated with virally infected
may play an important role in the control of the develop- cells or tumor cells. They also express receptors for MHC
ment and spread of some malignant tumors. The responses molecules that shut off their lytic activity. Thus, virally in-
that actually cause the destruction of the agents that initi- fected cells or tumor cells that escape the surveillance of cy-
ate these pathogenic states (e.g., bacteria, viruses, parasites, totoxic T cells by downregulating or shutting off expression
tumor cells) are collectively the effector mechanisms of of MHC molecules then become targets for efficient killing
the immune system. Several have already been alluded to. by NK cells because the cytotoxic activity of the latter cells is
Among them are the cytotoxic action of CTLs, which leads no longer shut off by the recognition of particular alleles of
to the destruction of cells harboring viruses and, in some MHC class I molecules.
circumstances, expressing tumor antigens. In some cases, In addition, NK cells express a receptor for the Fc portion
antibody can be directly protective by neutralizing determi- of IgG (Fc γRIII). Antibody-coated cells can be recognized
nants essential to a critical step through which the patho- by NK cells, and such cells can then be lysed. This process is
gen establishes or spreads an infectious process. However, in referred to as antibody-dependent cellular cytotoxicity.
most cases, the immune system mobilizes powerful nonspe- NK cells are efficient producers of IFN-γ. A variety of
cific mechanisms to mediate its effector function. stimuli, including recognition of virally infected cells and
tumor cells, cross-linkage of FcγRIII, and stimulation by the
cytokines IL-12 and IL-18, cause striking induction of IFN-γ
Effector Cells of the Immune Response
production by NK cells.
Among the cells that mediate important functions in the im-
mune system are cells of the monocyte/macrophage lineage,
NK cells, mast cells, basophils, eosinophils, and neutrophils. Mast Cells and Basophils (Chapters 20 and 45)
It is beyond the scope of this introductory chapter to present Mast cells and basophils play important roles in the induction
an extended discussion of each of these important cell types. of allergic inflammatory responses. They express cell surface
However, a brief mention of some of their actions will help in receptors for the Fc portions of IgE (Fc εRI) and for certain
understanding their critical functions in the immune response. classes of IgG (Fc γR). This enables them to bind antibody to

their surfaces, and when antigens capable of reacting with Lectin

that antibody are introduced, the resultant cross-linkage of Pathway
Fc εRI and/or Fc γR results in the prompt release of a series of
potent mediators such as histamine, serotonin, and a variety
of enzymes that play critical roles in initiating allergic and
anaphylactic-type responses. In addition, such stimulation
also causes these cells to produce a set of cytokines, includ-
ing IL-3, IL-4, IL-13, IL-5, IL-6, granulocyte-macrophage
colony-stimulating factor, and TNFα , that have important
late consequences in allergic inflammatory responses.
Granulocytes (Chapter 20)

Granulocytes have critical roles to play in a wide range of in-
flammatory situations. Rather than attempting an extended
discussion of these potent cells, it may be sufficient to say that
in their absence it is exceedingly difficult to clear infections
with extracellular bacteria and that the immune response plays
an important role in orchestrating the growth, differentiation,
and mobilization of these crucial cells. Recent work indicates
that TH17 cells are particularly important because of their role in FIG. 1.12. The Complement System. The classical pathway of comple-
recruiting granulocytes to sites of immune responses. ment activation, usually initiated by the aggregation of C1 by binding
to antigen/antibody complexes, resulting in the formation of an en-
zyme, a C3 convertase, that cleaves C3 into two fragments, C3b and
Eosinophils (Chapters 20 and 45) C3a. The classical pathway can also be initiated by the aggregation
Eosinophils are bone marrow–derived myeloid cells that com- of mannan-binding lectin as a result of binding sugars expressed
plete their late differentiation under the influence of IL-5. They in the capsules of many pathogenic microbes. The components of
migrate to tissue sites in response to the chemokine eotaxin the lectin pathway appear to mimic the function of C1qrs. The alter-
and as a result of their adhesion receptors. Because TH2 cells native pathway of complement activation provides a potent means
of activating complement without requiring antibody recognition of
can produce IL-5 and stimulate the production of eotaxin, eo- antigen. It results in the formation of a distinct C3 convertase. The
sinophil accumulation is often associated with TH2-mediated fragments formed by cleaving C3 have important biologic activities.
inflammation. Eosinophils store a series of proteins in their In addition, C3b, together with elements of the classical pathway
secondary granules including major basic protein, eosinophil (C4b, C2a) or the alternative pathway (Bb, properdin), form enzymes
cationic protein, and eosionphil peroxidase. When released, (C5 convertases) that cleave C5, the initial member of the terminal
these proteins are responsible for much of the damage that family of proteins. Cleavage of C5 leads to the formation of the mem-
eosinophils mediate both to helminthic parasites and to the brane attack complex that can result in the osmotic lysis of cells.
epithelium. Eosinophils have been implicated as important in
protective responses to helminths and in the tissue damage antigen with IgM or IgG antibody. This leads to the binding
seen in allergic inflammation in conditions such as asthma. of the first component of complement, C1, and its activation,
creating the C1 esterase that can cleave the next two compo-
The Complement System (Chapter 36) nents of the complement system, C4 and C2.
The complement system is a complex system of proteolytic C4 is a trimeric molecule, consisting of α , β, and γ
enzymes, regulatory and inflammatory proteins and pep- chains. C1 esterase cleaves the α chain, releasing C4b, which
tides, cell surface receptors, and proteins capable of causing binds to surfaces in the immediate vicinity of the antigen/
the lysis of cells. The system can be thought of as consist- antibody/C1 esterase complex. A single C1 esterase molecule
ing of three arrays of proteins. Two of these sets of proteins, will cause the deposition of multiple C4b molecules.
when engaged, lead to the activation of the third component C2 is a single polypeptide chain that binds to C4b and
of complement (C3) (Fig. 1.12). The activation of C3 releases is then proteolytically cleaved by C1 esterase, releasing C2b.
proteins that are critical for opsonization (preparation for The resulting complex of the residual portion of C2 (C2a)
phagocytosis) of bacteria and other particles, and engages the with C4b (C4b2a) is a serine protease whose substrate is
third set of proteins that insert into biologic membranes and C3. Cleavage of C3 by C4b2a (also referred to as the classi-
produce cell death through osmotic lysis. In addition, frag- cal pathway C3 convertase) results in the release of C3a and
ments generated from some of the complement components C3b. A single antigen/antibody complex and its associated
(e.g., C3a and C5a) have potent inflammatory activities. C1 esterase can lead to the production of a large number of
C3 convertases (i.e., C4b2a complexes) and thus to cleavage
The Classical Pathway of Complement Activation of a large number of C3 molecules.
The two activation systems for C3 are referred to as the The components of the classical pathway can be activated
classical pathway and the alternative pathway. The classi- by a distinct, non–antibody-dependent mechanism, termed
cal pathway is initiated by the formation of complexes of the lectin pathway. The mannose-binding lectin (MBL) is

activated by binding to (and being cross-linked by) repeti- pathway C5 convertase) or a complex of C3bBb with a pro-
tive sugar residues such as N-acetylglucosamine or man- tein designated properdin (the alternative pathway C5 con-
nose. The activation of MBL recruits the MBL-associated vertase). Cleaved C5, C5b, forms a complex with C6 and
serine proteases MASP-1 and MASP-2, which cleave C4 and then with C7, C8, and C9. This C5b/C9 complex behaves
C2 and lead to the formation of the classical pathway C3 as an integral membrane protein that is responsible for the
convertase. Because the capsules of several pathogenic mi- formation of complement-induced lesions in cell mem-
crobes can be bound by MBL, the lectin pathway provides an branes. Such lesions have a donut-like appearance, with C9
antibody-independent mechanism through which the com- molecules forming the ring of the donut.
plement system can be activated by foreign microorganisms. In addition to the role of the complement system in op-
sonization and in cell lysis, several of the fragments of com-
The Alternative Pathway of Complement Activation plement components formed during activation are potent
Although discovered more recently, the alternative pathway mediators of inflammation. C3a, the 9,000-dalton fragment
is the evolutionarily more ancient system of complement ac- released by the action of the C3 convertases, binds to recep-
tivation. Indeed, it, and the MBL activation of the classical tors on mast cells and basophils, resulting in the release of
pathway, can be regarded as important components of the histamine and other mediators of anaphylaxis. C3a is thus
innate immune system. The alternative pathway can be ac- termed an anaphylotoxin, as is C5a, the 11,000-dalton
tivated by a variety of agents such as insoluble yeast cell wall fragment released as a result of the action of the C5 con-
preparations and bacterial lipopolysaccharide. Antigen/an- vertases. C5a is also a chemoattractant for neutrophils and
tibody complexes also can activate the alternative pathway. monocytes.
The C3 convertase of the alternative pathway consists of a Finally, it is important to note that the process of activa-
complex of C3b (itself a product of cleavage of C3) bound to tion of the complement cascade is highly regulated. Several
the b fragment of the molecule factor B. C3bBb is produced regulatory proteins (e.g., C1 esterase inhibitor, decay ac-
by the action of the hydrolytic enzyme, factor D, that cleaves celerator factor, membrane cofactor protein) exist that
factor B; this cleavage only occurs when factor B has been function to prevent uncontrolled complement activation.
bound by C3b. Abnormalities in these regulatory proteins are often associ-
Apart from the importance of the alternative pathway in ated with clinical disorders such as hereditary angioedema
activating the complement system in response to nonspe- and paroxysmal nocturnal hemoglobinuria.
cific stimulants, it also can act to amplify the activity of the
classical pathway because the C3 convertase of the classical
system (C4b2a) provides a source of C3b that can strikingly
CONCLUSION
enhance formation of the alternative pathway convertase This introductory chapter should provide the reader with
(C3bBb) in the presence of factor D. an appreciation of the overall organization of the immune
system and of the properties of its key cellular and mo-
The Terminal Components of the Complement System lecular components. It should be obvious that the immune
C3b, formed from C3 by the action of the C3 convertases, system is highly complex, that it is capable of a wide range
possesses an internal thioester bond that can be cleaved to of effector functions, and that its activities are subject to
form a free sulfhydryl group. The latter can form a covalent potent, but only partially understood, regulatory processes.
bond with a variety of surface structures. C3b is recognized As the most versatile and powerful defense of higher or-
by receptors on various types of cells, including macro- ganisms, the immune system may provide the key to the
phages and B cells. The binding of C3b to antibody-coated development of effective means to treat and prevent a broad
bacteria is often an essential step for the phagocytosis of range of diseases. Indeed, the last two sections of this book
these microbes by macrophages. deal with immunity to infectious agents and immuno-
C3b is also essential to the engagement of the termi- logic mechanisms in disease. The introductory material
nal components of the complement system (C5 through provided here should be of aid to the uninitiated reader in
C9) to form the membrane attack complex that causes understanding the immunologic mechanisms brought into
cellular lysis. This process is initiated by the cleavage of play in a wide range of clinical conditions in which im-
C5, a 200,000-dalton two-chain molecule. The C5 conver- mune processes play a major role either in pathogenesis or
tases that catalyze this reaction are C4b2a3b (the classical in recovery.

CHAPTER
2 History of Immunology
Steven Greenberg
INTRODUCTION but it is generally recognized as a compilation of works by

There comes a time during every argument at which the two Hippocrates himself as well as his students and intellectual
opposing parties reach a critical juncture: either resolution heirs. One of his students, as well as son-in-law, Polybus, was
or impasse. Variations on this essentially Socratic theme credited as the author of De Natura Hominis (On the Nature
have played out in all spheres of human intellectual activity. of Man), the earliest known text describing the ancient Greek
The dialectic of science ranges from incremental and rela- conceptual basis for disease pathogenesis, as embodied in
tively harmonious shifts in key, to a few abruptly dissonant the four humors: black bile, yellow bile, phlegm, and blood.5
ones, taking the form of what Thomas Kuhn would refer to The body of man has in itself blood, phlegm, yellow
as “paradigm shifts.”1 Such is the case with immunology, a bile and black bile; these make up the nature of his
field distinguished by more than its fair share of paradigm body, although these he feels pain or enjoys health.
shifts. Arguably, its first dialectic, between the “cellularists” Now he enjoys the most perfect health when these ele-
and the “humoralists,” did not result in an early synthesis, ments are duly proportioned to one another in respect
but was characterized by partisan and often entrenched of compounding, power and bulk, and when they are
positions. Ultimately, the two parallel paths of cellular and perfectly mingled. Pain is felt when one of these ele-
chemical immunology converged, but it was not until the ments is in defect or excess, or is isolated in the body
latter half of the 20th century that the two paths became one. without being compounded with all the others.
How the paths were forged in the first place was an amalgam
of the cultural institutions of the time, the creative output The Greek view that disease arose from an imbalance of
of the scientists themselves, and the imperatives of devising the four humors did not supersede the miasma theory of
effective strategies to combat infection and contagion. disease, but was rather a more general view of disease causa-
tion, compared with the subset of apparently communicable
diseases best explained by miasma.
ANTECEDENTS TO THE GERM THEORY OF DISEASE The Romans developed and refined Greek concepts of
Ancient Theories of Disease Causation disease. Marcus Terentius Varro (116 to 27 bce), referred to
Religious beliefs in ancient Greece drew contrasts between as “the most learned of all Romans” by the Roman rhetori-
the sacred or the pure (katharos) and the polluted (miaros).2 cian Quintilian,6 was a prolific Roman scholar, estimated to
Pollution, or miasma, was blamed for many ancient trans- have written more than 600 volumes. During the civil war
gressions, from the petty and personal, to the gravest, most of the first century, he served as Pompey’s legate in Spain
famously embodied in the Oedipus myth. To remove the stain and fought at Pharsalus against Caesar but ultimately recon-
of miasma, the transgressor must undergo rites of purifica- ciled with Caesar, who appointed him director of the public
tion (catharsis). To the ancient Greeks in the age of Homer, library in 47 bce. Varro is perhaps best known for his only
these were deeply ingrained beliefs that were essentially re- complete extant work, Rerum Rusticarum Libri Tres (On
ligious in nature. Because miasma was viewed as a source Agricultural Topics), in which he so presciently anticipated
of suffering, it is not surprising that miasma was implicated the existence of disease-causing microbes that seemed to
in disease, as described by Hippocrates in his treatise “On Varro to be the immediate cause of diseases7:
Air, Water, and Places” in which miasma was associated with
Precautions must also be taken in the neighbourhood
“unhealthy vapors.”3,4 Hippocrates is credited with being the
of swamps, both for the reasons given, and because
first to recognize the potential of disease to arise from the
there are bred certain minute creatures which cannot
environment and not as a result of religious superstition.
be seen by the eyes, which float in the air and enter
Mal aria, which is Old Italian for “bad air,” was one of many
the body through the mouth and nose and there cause
diseases thought to be caused by miasma. The concept that
serious diseases.
miasma was the source of disease persisted through the mil-
lennia and was a leading theory of how contagious diseases Thus, Varro provided a mechanistic basis for disease that
were transmitted up until the time of Pasteur. was consistent with the prevailing belief of miasma as the
Much of what we know about the medicine of ancient source of illness.
Greece is codified in The Hippocratic Corpus, a collection The Greek medical tradition was carried on for gen-
of more than 60 volumes of text. Its authorship is disputed, erations and was ultimately passed on to Claudius Galinus
22

CHAPTER 2 HISTORY OF IMMUNOLOGY | 23
(Galen), the Greek expatriate who was its greatest explicator. It seemed inevitable that in seeking an explanation for why
Galen was born in Pergamum in Asia Minor in 130 ce. After some developed disease and others did not, many others
beginning his medical training in Pergamum at the behest would rely on a moralistic or religious view. In a remarkable
of his father, he traveled widely in pursuit of “postgraduate” passage from Galen’s On the Different Types of Fever, not only
medical training in Smyrna, Corinth, and Alexandria. He does he set forth an explanation of how disease is transmit-
returned to Pergamum and practiced surgery on gladiators, ted through the air, and using the same term as Fracastoro
which provided a unique opportunity to deepen his knowl- would, “seeds,” some 1300 years later, but he also blames a
edge of human anatomy and perfect his surgical technique.8 licentious lifestyle for susceptibility to the plague:
Following an outbreak of plague among the Roman troops
Suppose, for example, that the circumambient air
in Aquileia in 168 ce, he was summoned by the Emperor
carries certain seeds of plague, and that of the bod-
Marcus Aurelius and was appointed personal physician to
ies which share [breathe] it, some are full of various
his son, Commodus.8 Galen’s view of medicine was based on
residues which are soon to become putrefied in them-
Hippocrates’ Corpus. His output was prodigious more than
selves, while others are clean and free of such resi-
any other ancient author of medical texts. He distinguished
dues. Assume also that in the former there is a general
symptoms from diseases and offered explanations of the for-
blockage of their pores, a so-called plethora, and a
mer that were consistent with his interpretations of disease
life of ease devoted to gluttony, drink and sex, with
pathogenesis; thus, tertian fever was the result of an “imbal-
all their necessarily concomitant digestive disorders.
ance of yellow bile,” quartan was caused by “too much black
The others, which are clean and lack these residues, as
bile,” and quotidian by “an excess of phlegm.” Vomiting
well as being fine in themselves, have all a wholesome
was viewed as the body’s attempt to expel poisons, and the
transpiration through pores that are neither blocked
prescription of bleeding was to rid the body of “corrupt hu-
nor constricted; they take appropriate exercise and
mors.”9 Galen’s view of medicine remained the dominant
lead a temperate life. Assuming all this, which of these
one until the 17th century.
bodies is most likely to be affected by the rotting air
they inspire?
Early Concepts of Immunity This passage has been analyzed extensively by Nutton, who
The term “immunity” itself is derived from the Latin prac- questioned the extent to which “seed” is used metaphori-
tice of “exemption” from taxes or public service that normal cally12 ; if so, it is particularly apt.
citizens had to discharge, a favor bestowed by the emperor to Religious explanations for disease and immunity per-
meritorious individuals or entire communities.10 However, sisted throughout history. Particularly during the growth
the concept of immunity dates back at least as far as of Christianity during the Middle Ages, disease and sin
Thucydides, who described the plague of Athens of 430 bce were linked, though not inextricably; the great theologian
that was responsible for the death of more than a quarter of Thomas Aquinas provided this distinction between sin and
the Athenian population11: other causes of diseases13 :
Yet it was with those who had recovered from the dis- . . . we need to consider that sin consists of a disorder
ease that the sick and the dying found most compas- of the soul, just as physical disease consists of a disor-
sion. These knew what I had from experience and had der of the body. And so sin is a disease of the soul, as it
now no fear for themselves; the same man was never were, and pardon is for sin what healing is for disease.
attacked twice-never at least fatally. Yet for many, a disease as dire as the plague would continue
The traditional, religious view of the plague was that it is to be viewed as divine retribution for sin; for others, it was
the work of Apollo who was held responsible for earlier the result of astrological phenomena, while for still others,
plagues (eg, the plague on the Greek army in Troy because it was the result of a “conspiracy plotted by Jews to poison
the Greek general Agamemnon abducted the daughter of wells.”14 With the exception of the latter, which at least of-
Apollo’s priest, Chryses). The religiously inclined could fered a proximate physical cause of disease, regardless of the
take some refuge in appealing to Apollo’s son, Asklepios, lack of evidence, there was an abstract quality to these ex-
who was revered as the god of healing, as were his daugh- planations that were shrouded in belief and superstition but
ters Hygeia (Hygiene), Iaso (Remedy), Akæso (Healing), lacking in substance. Further theories of disease pathogen-
Aglæa (Healthy glow), and Panakeia (Cure-all). In contrast, esis would have to await the 16th century.
Thucydides characteristically did not offer facile religious
explanations for sickness or recovery. In fact, he described Fracastoro’s Seeds
the futility of the Athenians’ plight in stark terms11:
Girolamo Fracastoro (1478 to 1553) was an Italian who
Neither the fear of the gods nor laws of men awed would have met most definitions of a Renaissance scholar:
any man, not the former because they concluded it an accomplished physician, poet, mathematician, botanist,
was alike to worship or not worship from seeing that and astronomer. Educated by his father in Verona, and later
alike they all perished, nor the latter because no man at the University of Padua, he became an instructor in logic
expected that lives would last till he received punish- at the University of Padua in 1501 and in anatomy in 1502.
ment of his crimes by judgment. He left Padua in 1508 and returned to Verona, where he

dedicated himself to his studies and his medical practice. the microscopes of the their time, advanced the concept that
In 1546, he proposed that disease was caused by seminaria “all living things are composed of cells and cell products.”19
(“seeds”) that could be transmitted by three ways: direct Virchow took this one step further by declaring omnis cel-
contact from one person to another; through “fomites,” lula e cellula or “all cells develop only from existing cells.”
or articles of clothing or dirty linen; and through the air. Whether Virchow rejected the germ theory or not is a matter
Although Nutton12 has pointed out that Galen and Lucretius of debate, but it is more likely that Virchow’s underlying em-
also used the term “seeds” to describe the transmission of phasis was not on external causes, but disease mechanisms,
illness through the air (see previous discussion), Fracastoro as he wrote in 1885: “First the discovery of the parasite, then
was the first to make them the focal point for disease trans- the investigation of its etiology, then the question: how does
mission and to describe their predilection for certain organs. it give rise to the disease.”20 Although it is hard to escape the
Like the 10th century Arab physician Rhazes, who believed possibility that a certain degree of professional jealousy may
smallpox to have an affinity for blood, and specifically for have played a role in Virchow’s refusal to embrace the germ
the traces of menstrual blood that were believed to taint theory of disease, his viewpoint is one of but many examples
everyone in utero, as later suggested by Avicenna, Fracastoro of apparently strict dichotomies in science that would ulti-
offered the following explanation for immunity to smallpox: mately undergo revision and later synthesis. This is a theme
Following infection by smallpox seminaria, the menstrual that was to be recapitulated many times in the history of
blood would putrefy, rise to the surface, and force its way out immunology.
via the smallpox pustules.15 “Hence when this process has
taken place, the malady usually does not recur because the
infection has already been secreted in the previous attack.” The Conceptual Basis for the Germ Theory of Disease
Fracastoro’s seminaria theory remained influential for Between Leeuwenhoek’s technical breakthroughs in lens
nearly three centuries, in many ways serving as an early design in the late 17th century and the work of Pasteur
template for the germ theory of disease. and Koch in the late 19th century, several individuals en-
dorsed Fracastoro’s “seed” theory, which gained new rele-
THE GERM THEORY OF DISEASE AND vance when bacteria were first visualized. Among these was
DEVELOPMENT OF VACCINES Jacob Henle, a German pathologist who was later to become
Koch’s teacher. Henle wrote a treatise that not only laid out
Until the mid-19th century, the Galenic view of disease or- the germ theory of disease in great detail but also arguably
igins was the dominant one. Not only was the etiology of articulated an early version of what would later be known as
diseases misunderstood but also was the origin of life itself. “Koch’s postulates”21: “Before microscopic forms can be re-
The theory of spontaneous generation, which arose from garded as the cause of contagion in man, they must be con-
Aristotle, held that life originated spontaneously from in- stantly found in contagious material. They must be isolated
animate matter. The first experimental evidence against from it and their strength tested.” However, Henle’s essay
spontaneous generation came from Franceso Redi, the head was a theoretical one and Henle himself never provided
physician in the Medici court, who in 1668 provided early any experimental evidence in support of it. In the same
evidence against the theory.16* Nevertheless, no overarch- year, decades before Pasteur and Koch would even begin
ing theory was proposed to replace it. Neither the mindset to describe the germ theory of disease, Henry Holland, a
nor the necessary technology were available until the latter Scottish-trained physician to Queen Caroline, who traveled
part of the 17th century, when an apprentice in a dry goods extensively and was acquainted with the scientific luminar-
store, Anton van Leeuwenhoek, began a lifelong obsession ies of the time, including Davie, Gay Lussac, Berthollet, and
with grinding the perfect lens. Leeuwenhoek’s lenses were Laplace, wrote a treatise in which he stated,22
tiny but were ground with high degree of curvature, en-
abling him to visualize the hitherto undiscovered word of The question is, what weight we may attach to the
microbes. On September 17, 1683, Leeuwenhoek wrote a let- opinion that certain diseases, and especially some of
ter to the Royal Society, which was the first description of epidemic and contagious kind, are derived from minute
living, motile bacteria obtained from the plaque of his own forms of animal life, existing in the atmosphere under
teeth.17 Leeuwenhoek was not the fi rst to build a microscope particular circumstances; and capable, by application
(which was used by, among others, Redi), but his was far to the lining membranes or other parts, of acting as a
superior to existing multilensed or compound microscopes. virus on the human body.
Other scientists of the time, notably Robert Hooke, also ob-
served microorganisms, and it was Hooke who was the first In a footnote, he cited others, including Kircher, and par-
to publish the first image of a microorganism (the fungus ticularly Johannes Nyander, who wrote nearly a century
Mucor) in 1665.18 Some 150 years later, Dutrochet and then earlier23 :
later Schwann, Schleiden, and Virchow, taking advantage of . . . it may be an easy matter for very minute insects to
be the causes of diverse contagious diseases,” of which
*
Much later, Spallanzani and Pasteur provided key experimental evidence against
he included plague, measles, smallpox, and syphilis.
spontaneous generation, a belief that some scholars still held in the late 19th century,
when Pasteur, for example, showed that broth in swan-neck-bent, but not unbent or Nyander himself credits Lynceus Leuwenhoekius (“lynx-eyed”
broken flasks, failed to support microbial growth. or “keen-eyed” Leuwenhoek) as the first to have seen such

“animalcules.” Thus, it is fair to say that the germ theory of teacher, Jacob Henle (see previous discussion) and his
disease itself had been “germinating” for some time prior to contemporary, Edwin Klebs.30 However, neither Henle
its scientific proof by Pasteur and Koch. nor Klebs applied their theories to any practical benefit,
which is why Koch received credit for the postulates. Koch
was the fi rst to articulate then systematically apply them
Experimental Evidence for the Germ Theory to prove that Bacillus anthracis was the causative agent of
of Disease anthrax. Koch was a country physician living in Prussia,
The honor of the first experimental demonstration for whose scientific career began unceremoniously with a gift
the germ theory of disease may belong to two students of a microscope from his wife.31 His fi rst series of inves-
of Francesco Redi. In 1687, soon after Redi had offered tigations began with observing the blood of a dead cow
proof against spontaneous generation, two of his students, that succumbed to anthrax. Confi rming the observation
Bonomo and Cestoni, went on to observe the causative agent of Davaine and others before him, he observed fi lamen-
of scabies using the newly developed microscope and were tous bacteria in the blood. Not content merely with the
able to transfer disease from person to person.24 Perhaps the observation, he began a series of technically challenging
first demonstration of the bacterial pathogenesis of diseases experiments, necessitating the development of many novel
in animals and humans was by Casimir Davaine, a French techniques used in microbiology laboratories even today,
scientist who provided essentially the same evidence that such as the use of solid medium to grow individual clones
Pasteur and Koch would years later that anthrax was caused or colonies of bacteria. He proved that the fi lamentous bac-
by bacteridies. In 1865, Davaine was awarded the Prix Bréant teria were present only in infected animals and were ca-
by the Académie des Science for his work.25 pable of reproducing the disease when injected into healthy
Davaine’s compatriot, Louis Pasteur, was the consum- animals. This was the fi rst systematic application of the
mate experimentalist. A chemist by training, his interest eponymous postulates, which has since become the sine
in infectious disease began with his study of fermentation. qua non of disease causation by infectious agents. His work
He made important contributions to the science of fermen- was published in 1876, 32 the fi rst of many groundbreak-
tation and his work led to many practical benefits to the ing publications. Koch’s most profoundly important con-
beer and wine industry of France. His interest in microbes tribution to medicine was the discovery of the causative
began with his speculation that the same type of microbe agent of tuberculosis. The lecture at which he announced
required for fermentation was likely responsible for trans- his fi ndings, on March 24, 1882, described later by Ehrlich
mitting disease. In 1865, he was asked to investigate a dis- as “the most important experience of his scientific life,” is
ease called pébrine that affected the silk worm industry. considered by many to be the single most important lec-
Within a year, Pasteur had established that pébrine was ture in medical history. Koch described the invention of
caused by a microbe, which provided further proof for the novel staining methods to detect the tubercle bacillus and
germ theory of disease. Some 14 years later, his expertise presented tissue dissections from guinea pigs that were in-
was again sought out of economic interests, in this case by fected with tuberculous material from the lungs of infected
farmers whose poultry stocks were diminished by chicken apes and humans who had died from the disease. 33 For “his
cholera. In a famous example of scientific serendipity, his investigations and discoveries in regard to tuberculosis,”
assistant, Charles Chamberland, failed to inoculate chick- Koch was awarded the Nobel Prize in 1905.
ens with cultures of chicken cholera bacilli, as instructed by
Pasteur, but instead went on vacation. Upon returning sev-
eral weeks later, he inoculated chickens with the old bacte- The Unhealthy Rivalry Between Pasteur and
rial cultures, but the chickens didn’t die as expected. Rather Koch, and its Lasting Effects
than disregard the experiment as a failure, “chance had fa- The Franco-Prusssian war of 1870, the culmination of
vored the prepared mind” of Pasteur, who had his assistants years of tension between France and Prussia, resulted in a
inject fresh cholera into the same hens that had previously Prussian victory and unity among the German states under
been injected; now none of the hens became ill. Pasteur had King Wilhelm of Prussia. The Treaty of Frankfurt left a uni-
surmised that the bacterial cultures had become weakened fied Germany the city of Strasbourg as well as possession of
by extended culture.26 Thus began the use of attenuated Alsace and the northern part of Lorraine, which was thought
strains of microbes to immunize against disease, 27 and the to contribute to further resentment of the Germans by the
birth of the science of vaccination. The term “vaccine,” de- French and public support for World War I. It is against this
rived from the Latin vaccus for cow, was, coined by Pasteur backdrop that the relationship between Pasteur and Koch
in honor of Jenner. Attenuation as a strategy of developing must be viewed. Koch had served in the Prussian army,
vaccines would prove to be enormously valuable, leading to and Pasteur’s son was a conscript fighting for the French.
development of the first rabies vaccine by Pasteur himself, Furthermore, there was intense professional rivalry between
a vaccine against the viral causative agent of yellow fever the two, especially over their work on anthrax pathogenesis.
by Theiler,28 and the Bacillus Calmette-Guérin vaccine at According to a letter from Charles Ruel, former privat docent
Pasteur’s institute.29 at the University of Geneva, Koch was in the audience when
Although Robert Koch is credited as the originator of Pasteur spoke on attenuation and vaccination at the fourth
“Koch’s Postulates,” it may come as a surprise that the International Congress of Hygiene and Demography held in
essence of the postulates were fi rst formulated by Koch’s Geneva in September 1882. Pasteur spoke repeatedly about

German collected works (recueil Allemand). According to severe case of smallpox. While in Istanbul, Lady Montague
the letter34 : observed the practice of variolation. Determined not to have
her family suffer as she had, she directed the surgeon of the
Koch and his friend Prof. Lichtheim, were sitting
embassy to learn the technique and, in March 1718, to va-
side by side; they knew French only imperfectly and
riolate her 5-year-old son. After her return to England, she
both mistook the word pride (orgueil) for collec-
promoted the technique and had her surgeon variolate her
tion (recueil). They felt their self-respect profoundly
4-year-old daughter in the presence of the king’s physician.
wounded and interpreted the words German pride as
The surgeon, Charles Maitland, was given leave to perform
a grave insult.
what came to be known as the “Royal Experiment,” in which
This is but one ironic example of the level of rancor and he variolated six condemned prisoners who later survived.
misunderstanding between the two great men. It is said that By these and other experiments, the safety of the procedure
Pasteur and Koch underwent some form of reconciliation was established, and two of the king’s grandchildren were
later in their lives. Regardless, the aftermath of their rivalry, variolated on April 17, 1722. After this, the practice of vario-
in many ways personifying the bitter relations between lation spread rapidly throughout England in the 1740s and
France and Germany, had a lasting effect on the evolution then to the American colonies.35,36
of the nascent field of immunology. In the years to come, It is difficult to say what influence the English country
the intellectual heirs of Pasteur and Koch would reenact physician Edward Jenner (1749 to 1823) had on Pasteur’s
the lifelong competitive tensions that characterized their later discovery of attenuation of bacterial cultures and its
relationship. application to vaccination. Regardless, it is fair to say that
Jenner had a major influence on public health, as he was the
first to publish the development and use of a safe alternative
The Germ Theory of Disease: A Summation to variolation.37 Although Jenner is rightly celebrated for his
In many ways, the “germ theory of disease” really did not development of cowpox as a safe vaccine for smallpox, he
begin with Pasteur and Koch, but rather by their predeces- was not the first to make use of a relatively nonpathogenic
sors, Henle, Klebs, and Davaine, who in turn owed credit virus to induce immunity. Twenty years earlier, Benjamin
to Fracastoro, and ultimately to Galen and Varro, some Jesty, an English farmer, inoculated his wife and two sons
1,600 years earlier. What enabled Pasteur and Koch to firmly with material taken from the cowpox lesion of the udder of
establish the germ theory of disease began as a thought pro- a cow in his neighbor’s herd.36 In 1796, Jenner inoculated
cess that over time became distilled to something tangible: James Phipps, an 8-year-old boy, with material obtained
First, the idea that invisible “seeds” might propagate disease; from a cowpox lesion that appeared on the hand of a dairy-
second, the advent of an optically superior microscope, by maid (Fig. 2.1). Six weeks later, he inoculated the experi-
Leuwenhoek, which enabled scientists their first glimpses at mental subject with smallpox without producing disease.
the “minute creatures” postulated by Varro; third, the grow- Further studies by Jenner established the efficacy of his
ing evidence against spontaneous generation that began with vaccination procedure. For this feat, Jenner received a cash
Redi, thus opening the door for a new theory of disease; and prize of £30,000 and election to nearly all of the learned so-
finally, the inductive genius and careful experimental tech- cieties throughout Europe.38
niques of Pasteur and Koch.
EMERGENCE OF IMMUNOLOGY AS A DISCIPLINE
The Long History of Vaccination: The Cellularists versus the Humoralists
Success and the Unprepared Mind The international renown of Pasteur and Koch led to the
Vaccination did not originate with Pasteur; its practice had establishment of research institutes that bore their names.
been carried out for centuries without any fundamental Many talented young scientists were drawn to these institutes
understanding of its basis. Probably the earliest recorded whose research missions should have been complementary,
example of intentionally inducing immunity to an infectious but were not, and the partisan battle that began with Pasteur
disease was in the 10th century in China, where smallpox and Koch would soon be reenacted on a larger scale.
was endemic.35 The process of “variolation” involved expos-
ing healthy people to material from the lesions caused by the
disease, either by putting it under the skin, or, more often, Metchnikoff and the Birth of Cellular Immunology
inserting powdered scabs from smallpox pustules into the Elie Metchnikoff (1845 to 1916), an ambitious student at the
nose. Variolation was known and practiced frequently in the University of Kharkoff (“I have zeal and ability, I am natu-
Ottoman Empire, where it had been introduced by Circassian rally talented—I am ambitious to become a distinguished
traders in the 17th century.35 Unfortunately, because there investigator”),31 borrowed a professor’s microscope and
was no standardization of the inoculum, variolation occa- began a lifelong quest to understand the cellular basis for
sionally resulted in death or disfigurement from smallpox, immunity. A comparative zoologist by training, his early
thus limiting its acceptance. Variolation later became popular academic focus was on understanding the development of
in England, mainly due to the efforts of Lady Mary Wortley metazoans. Heavily influenced by Darwin’s publication of
Montague. Lady Montague was the wife of the British am- The Origin of Species in 1859,39 he viewed early embryo-
bassador to the Ottoman court who herself had contracted a logic development as a competition among specialized cell

FIG. 2.1. Cowpox Pustule on the Arm of Sarah Nelmes.

Reprinted with permission from Jenner37; courtesy
of Dr Jenner’s House: Birthplace of Vaccination,
Gloucestershire, UK.
types.40 Ontogeny was seen as a set of “interactions of cell that a Lutheran pastor, Johann August Ephraim Goeze, was
lineages with each other to limit self-replication by any one the first to describe phagocytosis by microscopic observa-
component in favour of the interests of the organism as a tions of cells derived from hay infusoria in 1777. Much later,
whole.”40 He focused his interest on an amoeboid marker cell German pathologists of the mid-19th century, including
of the mesoderm, which was dubbed “phagocyte” (devour- Lieberkühn, Henle, and Vogel, drew connections between
ing cell) by a contemporary of Metchnikoff’s, the Viennese “pus corpuscles” of wounds and blood corpuscles.42 Others
zoologist, Carl Claus. In a justly famous passage from Olga at the time, particularly the English physicians William
Metchnikoff’s biography of her husband, she describes the Addison and Augustus Waller, observed leukocyte migra-
observation that became the defi ning moment of his scien- tion through capillaries in response to local injury, and Ernst
tific career41: Haeckel, a German marine biologist who was later to oppose
Metchnikoff in an early theory of gastrulation, described
One day when the whole family had gone to a circus to
molluscan leukocytes ingesting India ink particles in 1862.43
see some extraordinary performing apes, I remained
A handful of scientists of the time made the concep-
alone with my microscope, observing the life in the
tual link between phagocytosis and host defense, and
mobile cells of a transparent star-fish larva, when
Metchnikoff himself cites a few: “When (the phagocytosis
a new thought suddenly flashed across my brain. It
theory) is once firmly established, it will be time enough
struck me that similar cells might serve in the defense
to determine each part taken in its foundation by workers
of the organism against intruders. Feeling that there
such as Panum, Gaule, Roser, etc. . .”44 However, it seems
was in this something of surpassing interest, I felt so
that none of these workers seized upon this concept and
excited that I began striding up and down the room
appreciated its importance to the degree that Metchnikoff
and even went to the seashore in order to collect my
had; certainly, none had further developed and tested what
thoughts. I said to myself that, if my supposition was
in retrospect was the correct interpretation of the defensive
true, a splinter introduced into the body of a star-fish
function of phagocytosis.
larva, devoid of blood-vessels or of a nervous system,
Following the key experiment in 1882, described pre-
should soon be surrounded by mobile cells as is to be
viously, Metchnikoff performed many others to test the
observed in a man who runs a splinter into his fin-
“phagocytosis theory,” such as the observation of infection
ger. This was no sooner said than done. There was a
in water fleas, which he viewed as a Darwinian struggle
small garden to our dwelling, in which we had a few
between pathogen and host45 :
days previously organised a “Christmas tree” for the
children on a little tangerine tree; I fetched from it a Once they have insinuated themselves into the organ-
few rose thorns and introduced them at once under ism’s inmost part, the spores cause an accumulation of
the skin of some beautiful star-fish larvae as transpar- the mobile cells round them, which correspond to the
ent as water. I was too excited to sleep that night in white corpuscles in human blood. A battle takes place
the expectation of the result of my experiment, and between the two elements.
very early the next morning I ascertained that it had
Drawing the analogy with host defense in higher organisms,
fully succeeded. That experiment formed the basis of
Metchnikoff described the killing of the spores by “mobile
the phagocyte theory, to the development of which I
cells,” thus ensuring immunity for the organism. His view of
devoted the next twenty-five years of my life.
phagocytosis expanded to encompass not only host defense
Contrary to popular belief, Metchnikoff was not the first but also organismal development, which he viewed in a te-
to visualize and describe phagocytosis, nor was he the first leologic context; he cited the dissolution of the tadpole’s tail
to suggest that it played a role in host defense. In a histori- by the “pervasion of phagocytes.”46 By the time he moved to
cal recount of the history of phagocytosis,42 Stossel argues Paris to become Chef de Service at the Pasteur Institute in

1888, Metchnikoff had already formulated and tested what basis.”53,54 It is difficult to know what to make of this pas-
had become, according to his view, a cornerstone of the sci- sage; it sounds vague and seems to state the obvious, yet it
ence of immune system. Perhaps what he had not appreciated does suggest a degree of flexibility that more intransigent
at the time was that his move to Paris would come to sig- proponents of the humoralist camp seemed to lack at the
nify to the Germans a complicit alliance with the French. He time. Regardless, scientific reconciliation would not be
thus had unwittingly entered a battle that not only had been forthcoming until many years later, although phagocytosis
fought in the political arena but in the laboratory as well. theory was granted a temporary reprieve by the British phy-
sician Almoth Wright, who demonstrated the phenomenon
of opsonization of bacteria. Wright was the first to describe
The Ascendance of Humoral Immunity a mechanism by which humoral and cellular components
Several related developments in the new field of immunol- of immunity cooperate to kill bacteria.55 Wright is possi-
ogy occurred at the end of the 19th century that would seal bly best known in his incarnation as Sir Colenso Ridgeon
the fate of cellular immunology for at least 50 years: The in Shaw’s “The Doctor’s Dilemma.” Ridgeon defined “opso-
discovery by Roux and Yersin47 that toxins alone derived nin” as “...what you butter the disease germs with to make
from diphtheria bacilli could reproduce the symptoms of your white blood corpuscles eat them.”56 In what was viewed
diphtheria; the discovery by von Behring and Kitasato in as a well-deserved but partly symbolic gesture, nevertheless,
189048 of humoral immunity to diphtheria and tetanus and Metchnikoff and Ehrlich, two exemplars of the opposing
the passive transfer of immunity to diphtheria in animals schools of immunity, were awarded the Nobel Prize in 1908
by von Behring and Wernicke in 189249 ; and the discovery “in recognition for their work on immunity.” It would not
of alexins (Greek for “without a name”) by Hans Buchner be until 40 years later that another cell type of the immune
in 189950 and Jules Bordet at about the same time.51 Alexin system would be first identified as being the source of anti-
was renamed “complement” by Ehrlich, as it “comple- body57 and 70 years later when dendritic cells (DCs) would
mented” the activity of antibodies. Indeed, the ability of be first identified as being the key phagocytic leukocyte
humoral components alone to lyse bacteria (the Pfeiffer responsible for initiating the immune response.58
phenomenon) or erythrocytes in the absence of phagocyto-
sis51 provided strong independent evidence supporting the
humoralists’ claims. The discovery of complement also had Paul Ehrlich: The Cellularist’s Humoralist
practical uses, as complement fi xation became the basis of Paul Ehrlich embodies a pivotal figure in the history of
a widely used serologic test for the diagnosis of syphilis, the immunology. Although he would make many practical dis-
so-called Wasserman test.52 The collective discoveries of the coveries in his long research career, his greatest contribution
“humoralists” would lead to the successful treatment of a to immunology, like Jerne’s over a half century later, was a
number of previously intractable diseases, such as diphthe- conceptual breakthrough that served to stimulate the field
ria. Indeed, the first Nobel Prize in physiology of medicine of immunology for years to come. Although I am tempted
was awarded to von Behring in 1901, “For his work on serum to consider his “side chain theory” of antibody formation
therapy, especially its application against diphtheria, by a paradigm shift, that would presume that there was a pre-
which he has opened a new road in the domain of medical existing paradigm to shift from, when in fact there was no
science and thereby placed in the hands of the physician a paradigm of antibody specificity to begin with.
victorious weapon against illness and death.” Bordet himself Ehrlich began his research career as a medical student.
would later be awarded the Nobel Prize “for his studies in One of his professors was the pathologist Wilhelm von
regard to immunity.” Among Bordet’s contributions was the Waldeyer, who introduced the young Ehrlich, who already
development of the complement fi xation test together with showed a strong interest in chemistry, to histologic methods
his brother-in-law, Octave Gengou. This formed the basis of for staining cells and tissues. Following further training in
what Bordet termed “serodiagnosis.” several medical schools, he was influenced by the chemist
Although the lines in the sand had already been drawn by von Bayer and the pathologists Cohnheim, Haidenhain, and
the mostly Prussian humoralists led by Ehrlich on the one Weigert (his cousin). He presented his thesis on histologic
hand and the cellularists led by Metchnikoff on the other, it staining in Leipzig at the age of 24, in which he was the first
was Metchnikoff who wrote a letter to von Behring, propos- to describe mast cells. As noted by Silverstein, the opening
ing a scientific “truce”53 : sentence of the thesis gave an inkling of his approach to
science59 :
I now believe . . . we can calmly work side by side. We
can mutually support one another, just like the phago- While in the modern histological literature, directions
cytes and antitoxins, since . . . the phagocytes receive on tintorial method are already so numerous, and still
considerable assistance from the antitoxic property, increase from day to day, yet their theoretical basis has
just as the phagocytes . . . render great assistance to had only a very negligible consideration.
the organism or respectively its antitoxic powers, since
The same year, he was appointed senior physician in the
they capture and destroy . . . bacteria.
Department of Internal Medicine at the Charité-Hospital in
In fact, von Behring did not seem rigidly against the cel- Berlin. The head of the clinic, Friedrich Frerichs, encour-
lular school, and Metchnikoff viewed him as supporting the aged Ehrlich to continue his histochemical investigations,
view that “active immunity requires some type of cellular which led to the identification of neutrophils, eosinophils,

and basophils as well as the diagnosis of his own case of pul- lucid, logical, and profound. Beginning with a consideration
monary tuberculosis.60 of a simple chemical bond, Ehrlich somehow ends up
It is notable that the term “side-chain” (seitenketten with an antibody-producing cell; hence, “the cellularist’s
in German) was a chemical term in use at Ehrlich’s time humoralist.”
meaning much the same as it does today. It is inescapable Not every biologist was enamored of Ehrlich’s model, and
to conclude that Ehrlich’s focus and interest in chemistry soon after he proposed it, it came under attack, most notably
would be the driving force behind his thinking about how by Jules Bordet. Bordet objected to Ehrlich’s insistence that
antibody is formed, and “selected for,” by antigen. The es- the specificity of the antigen–antibody reaction required
sence of Ehrlich’s side chain theory, first proposed in 1897 an irreversible bond, whereas Bordet, whose views seemed
(Fig. 2.2) is well articulated in Ehrlich’s Nobel lecture and is to be more deductively grounded in his immediate line of
paraphrased here61: investigation (eg, precipitin reactions and complement fi xa-
tion) thought that adsorptive interactions between antigen
1. “The relationship between toxin and antitoxin are
and antibody were sufficient to account for specificity.15,60
strictly specific—tetanus antitoxin neutralizes exclu-
Although Ehrlich’s side-chain theory provided a logically
sively tetanus toxin, diphtheria serum only diphtheria
consistent mechanism to account for antibody specificity, it
toxin . . .”
would later be criticized for failing to account for antibody
2. “For this reason it must be assumed that the (toxin and
diversity. That problem would not be conquered for another
antitoxin) enter into a chemical bond . . . fitting each
60 years in yet another “paradigm shift” when Talmage,
other ‘like lock and key.’”
Burnet, and Lederberg proposed the clonal selection theory.
3. “The group in the protoplasm, the cell receptor, must be
However, the concept of clonality was not yet conceived of
identical with the antitoxin which is contained in solu-
in 1897, and in any case, it is hard to envision how a clonal
tion in the serum of immunized animals.”
selection theory could have been developed without the
4. “As these receptors, which may be regarded as lateral
prior description by Ehrlich of antibody selection itself.
chains (“seitenketten”) of the protoplasm...become
occupied by the toxin, the relevant normal func-
tion of this group is eliminated...the deficiency is not IMMUNOLOGY BRANCHES OUT: BENEFICIARIES OF
merely exactly compensated, but made up to excess THE EARLY FOCUS ON HUMORAL IMMUNITY
(hyperregeneration).” “Man built most nobly when limitations were at
What Ehrlich proposed purely on theoretical grounds is their greatest.”
a brilliant argument based on a combination of inductive –Frank Lloyd Wright
and deductive reasoning. Characteristic of Ehrlich, it is
Immunology during the early part of the 20th century was
heavily influenced by the early victories of the humoralists.
Aside from the spectacular practical implications of the
work of von Behring, Bordet, and Ehrlich, among others,
the immunologist’s toolkit of the early 20th century immu-
nologist was very limited. Advanced microscopic techniques
were not yet available, and cell fractionation techniques had
not yet been invented. Given these limitations, it is hard to
envision how the gap between antibody and cell could have
been bridged any closer than Ehrlich had managed, at least
on a theoretical level. However, the focus on humoral im-
munity did allow for the rapid development of several fields,
most notably immunochemistry as well as hematology and
allergy. It would be some time before cellular immunology
would catch up.
Karl Landsteiner and the Birth of Transfusion

Medicine and Autoimmunity
As described by Silverstein, “No single individual contrib-
uted as importantly to so many different areas of immu-
nology as did Karl Landsteiner.”15 Landsteiner was born
in Baden in 1868 and attended Vienna Medical School. He
began his training in internal medicine and studied chem-
istry with Emil Fischer, who would receive the Nobel Prize
in chemistry in 1902. He transferred to the Department of
Pathological Anatomy, home to Erdheim, Billroth, Escherich,
FIG. 2.2. Ehrlich’s “Seitenkette” (Side Chains). From Ehrlich.171 and other accomplished scientists, where he remained until

1907.62 Landsteiner’s first major accomplishment was the against IgG in the sera of some patients with rheumatoid
discovery of the human ABO blood group antigens at the arthritis.74 This observation fundamentally changed the
turn of the century.63,64 His motivation was succinctly de- field of rheumatology.75
scribed in his Nobel lecture speech65 : Landsteiner received the Nobel Prize in 1930 “for his
discovery of the human blood groups.” Ironically, upon
. . . proteins in individual animal and plant species receiving the prize, it is said that he felt the prize should
differ and are characteristic of each species . . . The have been awarded for his work on haptens, which would
problem raised by the discovery of biochemical speci- play a great role in the development of the growing field of
ficity peculiar to a species . . . was to establish whether immunochemistry.
the differentiation extends beyond the species and
whether the individuals within a species show similar
though smaller differences. Discovery of Hypersensitivity: The “Other Work”
In 1901 to 1902, Paul Portier and Charles Richet were at-
His experiment was a simple one, in which he applied the tempting to raise tolerance in laboratory animals to acti-
sera of six heathy men, including himself, to red blood notoxin, an uncharacterized toxin derived from tentacles
cells of each, and noted that the sera of the men reacted of sea anemones. Their experiments appeared to be unsuc-
differently with each other—no serum reacted with that cessful, and, in some cases, it appeared that the animals
same individual’s red blood cells. At the end of the paper, actually became sensitized to the antigen. They repeated
Landsteiner noted that the results could account for the their studies using dogs and obtained completely unan-
variable clinical consequences of human blood transfu- ticipated results: All eight dogs collapsed and died within
sion—and thus was borne the discovery of the ABO red minutes after receiving a relatively small dose. First think-
blood cell antigens that would later become useful in blood ing the results were due to experimental error, they later
typing prior to transfusions. Many years later, Landsteiner realized that the sensitized animals had all been exposed
would discover the M, N, P isoantigens in 192766 and the Rh to antigen 14 to 23 days previously.76 They proposed the
system in 194067). name “aphylaxis” (against protection), a term later changed
Landsteiner’s next major contribution was the codiscovery to “anaphylaxis.”77 Richet continued his investigations on
with Donath of the first autoimmune disease, paroxysmal anaphylaxis and was awarded the Nobel Prize in 1913 for
cold hemoglobinurea,68 which challenged Ehrlich’s dictum his work. Soon after Portier and Richet made their seminal
that such a situation could not occur.69 discovery, Maurice Arthus was able to induce a localized
The organism possesses certain contrivances by form of anaphylaxis (swelling and ultimately gangrene) by
means of which the immunity reaction . . . is prevented repeated subcutaneous injections of horse serum, consid-
from acting against the organism’s own elements and ered fairly nontoxic.76 Yet, a third type of hypersensitivity
so giving rise to autotoxins . . . so that one might be was described by the pediatricians von Pirquet and Schick,78
justified in speaking of a “horror autotoxicus” of the who noted that vaccinated children occasionally developed
organism. fever, joint pains, rash, diarrhea, and hypotension. They
concluded that the clinical features of what is now called
The nature of the contrivances was thought to be of “serum sickness” were not a direct result of the injection of
“the greatest importance” by Ehrlich, who later stated70 : antiserum, but the outcome of “when antigen and antibody
“According to our present investigations either the disap- meet.” As von Pirquet later explained,79
pearance of receptors or the presence of autoantitoxin is
We are able to observe the effects of the toxic body
foremost among these contrivances.” Depending on how
formed when antigen and antibody meet, that is,
this statement is interpreted, it could be viewed as Ehrlich’s
the serum disease. We see that at the time when the
formulation of either clonal deletion or anti-idiotypes.
antibody arises, and therefore the antigen disappears,
Two years after the discovery of paroxysmal cold hemoglo-
symptoms of general disease occur. The supposed con-
binurea, an Italian ophthalmologist, who observed sympa-
nection is that these symptoms are due to toxic bodies
thetic ophthalmia, a disease in which damage to one eye
formed by this digestion of the allergen through the
leads to inflammation of the opposite eye, speculated that
antibody.
this disease was due to “autocytotoxins.”71 Autoimmunity
research was taken up by a few others, but perhaps owing Although von Pirquet does not precisely define what he
to either a misinterpretation of Ehrlich, or possibly due to meant by “toxic body,” some have interpreted this to mean
deference to his authority in the field, progress was slow. antigen–antibody complexes. It is possible that he was re-
It would not be years later, until the discovery of the role luctant to be more specific as he did not actually have a way
of sensitization of the newly discovered Rh antigen as an of observing or quantifying the complexes; it is also pos-
etiology of erythroblastosis fetalis,72,73 that autoimmunity sible that he did not have a ready explanation for how such a
became an active area of research for immunologists. The complex, if formed, could lead to the symptoms and signs of
fi rst time that autoimmunity was fi rst associated specifi- serum sickness. It would not be until many years later that
cally with arthritis was 1957, when Kunkel and colleagues Frank Dixon and colleagues would precisely delineate the
discovered what came to be called “rheumatoid fac- nature of the immune complexes.80 Nevertheless, it is clear
tor,” large complexes of immunoglobulin (Ig)M directed that von Pirquet and Schick viewed this phenomenon as

lying along a continuum with the normal immune response, Immunoglobulin Structure: The “Keys” to the Problem
or rather being a necessary component of it. As they later It was clear that further progress on defining the physico-
stated,81 chemical nature of the antigen–antibody reaction required
The conception that the antibodies, which should the development of specific tools that were unavailable at
protect against disease, are also responsible for the the turn of the century. Antibodies, whose chemical struc-
disease, sounds at first absurd. . . One forgets too tures were unknown at the time, were considered “colloids”
easily that the disease represents only a stage in the (from Greek kolla, glue), a suspension of particles suspended
development of immunity, and that the organism often in a continuous phase of another component. In 1924, the
attains the advantage of immunity only by means of Swedish chemist Svedberg designed a centrifuge based on
disease. a modified milk separator. The centrifuge could develop
a centrifugal field of up to 5000 g and enabled Svedberg to
It is von Pirquet and Schick who coined the term “allergy” measure the molecular mass of hemoglobin83 ; he was the
(from the Greek allos, other, and ergon, work).79 By high- first to determine the molecular mass of macroglobulins, de-
lighting the role of tissue injury in promoting immunity, rived from a patient with Waldenström macroglobulinemia,
they closed the loop that Metchnikoff had begun at the end which we now know as IgM. A student in Svendberg’s labo-
of the 19th century. This theme would be revisited and ex- ratory, Arne Tiselius, developed gel electrophoresis,84 which
panded upon in the latter part of the 20th century when it allowed for the separation of molecules based on charge and
was discovered that the inflammation that accompanied the size. These tools enabled Michael Heidelberger to estab-
innate immune response was a necessary prequel to the ac- lish the field of “immunochemistry.” Heidelberger devoted
quired immune response. nearly his entire research career pursuing the implications of
a simple but profound discovery he made with Oswald Avery
in 1923 that type-specific antigens of pneumococcus bacteria
THE LONG JOURNEY FROM THE DAWN OF are complex polysaccharides. Over the next three decades,
IMMUNOCHEMISTRY TO THE STRUCTURE OF this discovery enabled him to determine, for the first time,
IMMUNOGLOBULINS the exact weight and chemical composition of antibodies,
Challenges to Ehrlich: The Problem with the “Keys” antigens, and complement. He showed that antibodies are
multivalent proteins and used these insights to devise a sim-
One of the difficulties that Ehrlich faced was bridging
ple vaccine against pneumonia whose effectiveness was first
the gap between the conceptual basis of antibody speci-
proven in soldiers who fought in World War II.85
ficity and the actual basis of antibody specificity. If his
side-chain theory was correct, then every cell involved in
antibody production would be capable of reacting against The Chemical Nature of Immunoglobulin Molecules
every possible antigen it might encounter. Even without By 1950, it was appreciated that antibodies were proteins
considering the cellular origins of antibodies, Ehrlich’s of 150,000 molecular mass containing bivalent antigen-
critics, such as the Viennese Max von Gruber, raised the combining sites. Based on Porter’s observation that Igs could
question of how the astonishingly large number of different be split into smaller products by enzymes such as papain,
specificities of the antibody molecules themselves could yielding Fab fragments that bind antigen and Fc (crystal-
be generated.15 The size of the repertoire seemed impos- lizable) portions that do not, Edelman and Poulik further
sibly large if every “lock” had a unique “key.” Landsteiner defined Ig structure by hypothesizing that Bence-Jones pro-
became von Gruber’s assistant at the University of Vienna teins, derived from myeloma cells, were free light chains of
in 1896, and he inherited von Gruber’s critical view of Ig molecules.86 As this hypothesis appeared to be correct,
Ehrlich’s theory. Landsteiner’s early approach to this this provided a means of obtaining a ready supply of ho-
problem was to adopt Bordet’s “colloid” explanation of mogeneous Ig subunits. Reduction of disulfide bonds led to
the antigen–antibody interaction,15 which rejected cova- still different products, enabling them to propose that the Ig
lent interactions in favor of multiple weaker interactions, molecule consists of two light chains and two heavy chains,
a theme that was later taken up by Pauling, with some and that the antigen-binding site consisted of contributions
interesting consequences. Later, stimulated by the work from both heavy and light chains.86 Further refinement
by Obermeyer and Pick, who described chemical modi- of techniques in protein chemistry allowed Edelman and
fications of proteins, 82 Landsteiner used structurally re- Porter to work out the primary structure of Ig molecules, for
lated reactive chemicals to derivatize proteins. He showed which they received the Nobel Prize in 1972.
that antisera raised against one of the chemically modi-
fied proteins would react to varying degrees with proteins
modified by structurally similar, but not identical, reac- CONFRONTING THE SIZE OF THE REPERTOIRE
tive molecules. These results were interpreted as being The value of a large repertoire from which to select is ap-
incompatible with Ehrlich’s “lock and key” specificity preciated by any performing musician. Yet in the 1950s, the
but called for a more nuanced view of antigen–antibody repertoire problem was unsolved, leading to competing the-
specificity. It became immediately apparent that the size ories of how a large repertoire of diverse antibodies are gen-
of the repertoire could be greatly enhanced if one allowed erated. There were two mutually exclusive leading schools
for such graded degrees of binding affi nities. of thought: “instruction” theories and “selection” theories.

Instruction Theories of Antibody Diversity histocompatibility complex (MHC) (pMHC) has recently
15
As noted by Silverstein, the fi rst description of antigen been proposed, in which the initial encounter between TcR
as template was by Bail and Tsuda, who proposed in 1909 and pMHC is governed by strong interactions, followed by
that antigen persists after its encounter with antibody, and “scanning” the epitope and modest changes in conforma-
that by so doing, it leaves an impression on the antibody.87 tion of the TcR, leading to strengthening of the TcR–pMHC
This concept was further refi ned by Breinl and Haurowitz, association.90
who suggested that antigen is carried to the site of protein
formation, where it would serve as a template to instruct
antibody formation.88 Finally, in 1940, Linus Pauling pro- Jerne’s Natural Selection Theory of
vided a chemical explanation for antigen-directed instruc- Antibody Formation
tion: that “antibodies differ from normal serum globulin Instruction theories of antibody specificity persisted through
only the way the ends of the polypeptide chain is coiled” as the 1950s, when they were modified to include participation
a result of their amino acid sequence, and that “they have of enzymes to act as intermediaries between antigen and an-
accessible a very great many configurations with nearly tibody as well as the newly discovered structure of deoxyri-
the same stability.” Under the “influence” of antigen, they bonucleic acid (DNA). In 1955, Niels Jerne published a highly
“assume configurations complementary to surface regions influential paper in which he proposed a new theory of an-
of the antigen,” forming two active ends, and that after tibody formation that he described as the “natural selection
“freeing one end and the liberation of the central part of theory of antibody formation.”91
the chain, it folds up to form the central part of the an-
The antigen is solely a selective carrier of spontane-
tibody molecule, with two oppositely directed ends able
ously circulating antibody to a system of cells which
to attach themselves to two antigen molecules.” This in-
can reproduce this antibody. Globulin molecules
teraction would be further stabilized by weak interactions
are continuously being synthesized in an enormous
between antigen and antibody.89 Assuming a degree of de-
variety of different configurations...among which...
generacy in the initial antigen–antibody interaction, these
will be fractions possessing affinity toward any anti-
theories were consistent with the fi ndings of Landsteiner,
gen to which the animal can respond.
who showed that antigen–antibody interactions were not
absolutely specific, as envisioned by Ehrlich (Fig. 2.3). In Jerne referred to these preexisting antibodies “natural an-
retrospect, these theories had great chemical appeal; how- tibodies,” and went on to state that antigens selectively at-
ever, a central weakness is that if the initial interactions tach to those globulin molecules that happen to have a
between antigen and antibody are degenerate, that implies complementary configuration. According to Jerne, once the
that the interactions cannot be of high affi nity (otherwise interaction occurs, the antigen–antibody complexes may be
they would not be degenerate). Yet, if the initial interac- engulfed by a “phagocytic cell,” at which point the antigen
tions are weak, how would they occur in the fi rst place? is eliminated. The antibody within the phagocytic cell can
At some level, there has to be a certain degree of preex- remain or be transferred to another cell, which is the signal
isting antibody specificity, which implies a preexisting for reproduction of the same specific antibodies. More an-
repertoire. Although the instruction theories of antibody tibody is released into the circulation, resulting in a larger
selectivity helped explain some of the specificity of the percentage of specific circulating antibody. Jerne states that
antigen–antibody interactions, they could not account for “the reproduction need not be highly faithful; copying mis-
all of them. It is notable that the underlying basic principles takes will be harmless and may occasionally produce an
have been since invoked for other encounters in the im- improved fit.”91 These are remarkable concepts, as Jerne ap-
mune system. For example, a “bar code model” of interac- peared to have invoked Metchnikoff, a Darwinian interpre-
tions between the T-cell receptor (TcR) and peptide-major tation of antibody selectivity at the organismic level, as well
FIG. 2.3. Demonstration of Serologic

Specificity by Landsteiner and Scheer.
Reactions from immune serum for ani-
line with various azoproteins and with
unchanged horse serum reading after
15 minutes. (1) azoprotein from chicken
serum and aniline, (2) azoprotein from
horse serum and aniline, (3) azoprotein
from horse serum and para-toluidine,
(4) azoprotein from horse serum and
para-nitroaniline, (5) azoprotein from
horse serum and para-aminobenzoic
acid, (6) azoprotein from horse serum
and para-arsanilic acid, (7) unchanged
horse serum, (8) saline control. Modified
from Landsteiner and van der Scheer.486

as to have anticipated affinity maturation. It is easy to see in which there is “selective multiplication of a few species
why Jerne’s ideas were so influential. However, the central out of a diverse population” and antibody production. In a
problem, the lack of an explanation for the huge existing very succinct but remarkably insightful passage, he lays out
repertoire, which hampered the instruction theories of anti- the essence of the clonal selection theory 93 :
body diversity, as well as Ehrlich’s side-chain theory was still
As a working hypothesis it is tempting to consider that
unanswered. Jerne admitted this weakness and speculated
one of the multiplying units in the antibody response
that the “spontaneous production of globulin molecules of
is the cell itself. According to this hypothesis, only
a great variety of random specificities” may reside in a “.
those cells are selected for multiplication whose syn-
. . specialized lymphoid tissue, such as that of the thymus.”91
thesized product has affinity for the antigen injected.
Perhaps a more existential problem was that the flow of
information was from the existing preformed antibody to At about the same time, Burnet published his own paper
more antibody without any genetic intermediary. What was outlining the key aspects of the clonal selection theory 94
lacking was a specific mechanism to transfer information (Fig. 2.4). In a later interview, Talmage discussed how he
between a specific antibody and the specific synthesis of that provided a preprint of his review to Burnet before he pub-
same antibody. Jerne admitted to this problem and nomi- lished his landmark paper on the clonal selection theory,
nated ribonucleic acid (RNA) as a key template; he then but both he and Nossal stated that Burnet developed the
stated “ . . . a protein molecule may determine the order of theory independently. Talmage thought that the two papers
the nucleotides in the synthesis of RNA,” citing a paper that were similar in substance, but that “Burnet had the good
does not actually make this assertion.91 fortune to make an analogy of the idea to the clones that
What accounted for this conceptual block that had per- grow in bacteria and he gave it a very catchy name, ‘clonal
sisted for more than half a century? On one level, it was ig- selection.’”95 Among the ideas that led both immunologists
norance of the “fundamental dogma of molecular biology,” to propose the clonal selection theory was the known loga-
which had not yet been articulated.92 But on another level, rithmic rise in antibody titer during the primary immune
it perhaps can be traced back to the decisive victory of the response “as if it is a product of some multiplying sub-
humoralists over the cellularists. So long as the focus was stance.”95 Other experiments demonstrated that lymphoid
on the antigen–antibody interaction, there was no way to cells obtained from previously immunized rabbits that were
invoke a biologically plausible mechanism of generating a then transferred to x-radiated recipients were sufficient to
preexisting Ig repertoire and of selectively expanding a spe- induce a recall response in the recipients.96
cific portion of that repertoire. Although we now know that the clonal selection theory
is correct, it took more than 10 years for clonal selection to
be widely accepted, according to Talmage.95 Among the key
Development of the Clonal Selection Theory pieces of evidence to prove the theory was the demonstra-
In 1957, the American immunologist David Talmage pub- tion of antibody production from single cells, by Nossal and
lished a review whose focus was allergy; however, in the Lederberg,97 and the later finding that cognate antigen was
review, Talmage drew an analogy between natural selection, capable of aggregating all surface Ig on individual cells,98
FIG. 2.4. In this diagram, antigen C (Ag.C) is recognized

by clone “c,” triggering this clone, but not others, to pro-
liferate. Antibody against Ag.C (AB.C) is indicated at the
bottom. From Burnet.246 Copyright ©1959 Sir MacFarlane
Burnet. Reprinted with the permission of Cambridge
University Press.

which corroborated Nossal and Lederberg’s results. Finally, was followed by the demonstration by Weigert et al.109 that
in 1975, Köhler and Milstein demonstrated the production the Ig Vκ region in the mouse is encoded by multiple V, J,
of monoclonal antibodies from single clones of immortal- and C regions joined at the DNA level during differentiation
ized plasma cells.99 Along with Jerne, they shared the Nobel of individual lymphocytes. Finally, in 1980, Early et al.110
Prize in 1984. demonstrated how V, D, and J regions of the Ig locus could
recombine to generate a virtually unlimited combination of
antibody specificities.
The Structural Basis of Antibody Diversity: In the ensuing years, the molecular mechanisms govern-
The Dialectic Revisited ing VDJ recombination were uncovered. These involved
By the mid-1970s, the puzzle of antibody diversity was far recognition of conserved sequences flanking germline V,
from solved—many of the pieces were still missing. The D, and J segments, introduction of double-strand breaks,
conceptual basis for the clonal selection theory was laid potential loss or gain of nucleotides at the coding junctions,
in 1957 by redirecting the focus of investigation from im- and polymerization and ligation to complete the join-
munochemistry to cell biology. However, what was needed ing process. Many talented scientists contributed to these
to actually prove the theory was a delineation of the mo- discoveries, including Alt, Yancopoulos, Blackwell, and
lecular basis for the enormous size of the repertoire. Two Gellert.111 This culminated in the isolation of the recom-
lines of investigation converged to provide this evidence: the binase activating genes (RAG) by Baltimore’s group.112, 113
sequencing of Ig proteins followed by the sequencing of Ig The dominant view that emerged from these studies largely
genes. In 1965, Hischmann and Craig sequenced two Bence- favored the “germline-ists,” reinforced by Tonegawa’s re-
Jones proteins and found that there was conservation of the ceiving the Nobel Prize in 1984. However, in yet another
amino acid sequence at the C-termini but considerable di- example of a synthesis of two apparently contradictory ap-
versity at the N-termini.100 In the ensuing years, sequence proximations of the truth, evidence for somatic hypermuta-
data on a number of myeloma proteins became available, tion began to emerge.114,115 Its importance was established
and in 1970, Kabat and Wu applied statistical criteria to when it was causally linked to the generation of B cells
analyze sequencing data from 77 Ig chains. They identified with very high affi nity antibodies,116 a phenomenon termed
three regions within the 107 residues comprising the light “affi nity maturation.”117
chain variable region that demonstrated a still further de-
gree of variability (ie, hypervariability). They hypothesized
that these regions of extreme diversity represented the
SPECIALIZATION WITHIN THE IMMUNE SYSTEM
complementarity-determining residues and suggested by Division of Labor: Discovery of T and B Cells
analogy with prokaryotes that they arose through episomal The first person to demonstrate delayed type hypersensitiv-
incorporation into the light chain locus by a recombination ity may have been Robert Koch in 1882. On his quixotic pur-
event.101 If this, indeed, was the underlying explanation for suit of developing a vaccination against tuberculosis, he in-
antibody diversity, then there would have to be a very large jected himself with spent medium from cultures of human
number of episomal elements to account for a diverse reper- tubercle bacilli and noted a particularly severe reaction, in-
toire. This was a different view than the one taken by Dreyer cluding systemic effects.118 Although he was not successful
and Bennet 5 years earlier, who proposed that variable rein developing a vaccine against tuberculosis, he recognized
gion genes that had undergone duplication and spontaneous the diagnostic potential of this procedure. It was not until
mutation throughout evolution underwent a “genetic scram- 1942, and then later in 1945, that Landsteiner and Chase
bling” event, resulting in their juxtaposition to the constant demonstrated that cells from guinea pigs previously im-
regions of the Ig genes. They even suggested the involvement munized with Mycobacterium tuberculosis or hapten would
of enzymes involved in DNA repair as contributing to such transfer reactivity to a naïve recipient when challenged with
an event.102 Thus, two opposing viewpoints began to emerge the immunogen.119,120 This was the first demonstration that
to account for the generation of diversity (or GOD, as play- cells, rather than antibody, transmitted specific immunity, a
fully described by Richard Gershon): somatic mutation of finding that in some ways vindicated Metchnikoff’s cellular
a few genes, as suggested originally by Burnet,94 and sup- focus.
ported by Weigert and Cohn’s sequencing data of the mouse The identity of the cells mediating the transferred hyper-
λ light chain locus,103 or somatic recombination among sensitivity was unknown. Based on experiments performed
many duplicated genes within the germline, as proposed decades earlier, as Silverstein has noted,121 James Murphy
by Dreyer and Bennet102 and later refined by Edelman and at the Rockefeller Institute argued that lymphocytes were
Gally104 and Hood and Talmage.105 In the years that followed, important in the host resistance to tuberculous infection.
various teleologic arguments were proposed to support one Murphy used mice exposed to x-rays or splenectomized to
theory over the other. In 1976, at least a partial resolution manipulate lymphocyte numbers and showed that condi-
was provided by Hozumi and Tonegawa, who provided evi- tions that would have been predicted to deplete lympho-
dence that the Vκ and Cκ loci from embryonic DNA rear- cytes resulted in early death of the mice due to disseminated
range to form a contiguous polypeptide in mature lympho- tuberculosis.122 In earlier papers, Murphy also showed more
cytes.106 The advent of molecular cloning led to direct proof directly that lymphocytes were important in graft rejec-
that the variable and constant regions of the light chain gene tion in transplanted chick embryos. Why were these seem-
had undergone rearrangement at the DNA level.107,108 This ingly important observations ignored? Was it because the

experiments relied to a certain extent on inference, rather in appearance of the radiolabeled cells into the thoracic duct,
than direct proof that lymphocytes were key mediators of thus defining the continuous recirculation of lymphocytes
tuberculous immunity? Most likely, it was a combination from the lymphatics to the blood.135 In retrospect, these ex-
of events: The lingering vestiges of the confrontation be- periments were critically important in understanding how
tween the cellularists and the humoralists and the fact that lymphocytes are constantly patrolling the lymphatics, vigi-
there was little conceptual basis for understanding how a lantly on the lookout for antigen.
small innocuous-appearing cell type could participate in
immunity. TRANSPLANTATION BIOLOGY AND THE PURSUIT
In a completely independent line of investigation, it was OF IMMUNOLOGIC TOLERANCE
known for some time that certain strains of mice had a
high spontaneous rate of developing lymphocytic leukemia. The history of transplantation began many hundreds of years
In attempting to explain the finding that thymectomy of ago and was vigorously pursued by surgeons and tumor re-
2-month-old mice failed to develop leukemia, Jacques Miller searchers during the early part of the 20th century. These
found that neonatally thymectomized mice appeared ill and events are well summarized in several texts, notably Brent’s
revealed a marked deficiency of lymphocytes in blood and A History of Transplantation Immunology and Silverstein’s A
lymphoid tissues. Furthermore, these mice failed to reject al- History of Immunology. The influence of these early workers,
lografts or xenografts.123 However, not all areas within lym- particularly Carrel’s, on the surgical aspects of transplanta-
phoid organs were depleted of lymphocytes, consistent with tion is clear, but their impact on the field of transplantation
“thymic-dependent” (paracortical areas of the lymph nodes immunology was limited because the conceptual framework
and periarteriolar sheaths of the spleen) and “thymic-inde- of immunology was still rudimentary. Analogous to the role
pendent” regions (follicles and medullary cords). Miller con- that smallpox had in catalyzing early vaccine development,
cluded that the thymus was important for the development large-scale bombing campaigns in World War II resulted
of a subset of lymphocytes important in allograft rejection. in many civilian and military burn victims who required
The involvement of lymphoid cells in antibody produc- skin grafting, compelling surgeons to develop better tech-
tion was considered likely in the 1940s, mainly due to “guilt niques to avoid homograft rejection. The British zoologist
by association.” For example, Erich and Harris injected Peter Medawar was assigned to the War Wounds Committee
antigens, such as typhoid vaccine and sheep erythrocytes, of the Medical Research Council. In 1943, Medawar and
into the feet of rabbits, and then compared the formation Gibson136 published “The fate of skin homografts in man”
of antibody to histologic changes in the draining lymph based on a single burn victim who received multiple “pinch
nodes.124 Similar experiments were performed using intra- grafts” of skin. The authors concluded that autografts suc-
venous injection of antigen, and the appearance of antibody ceed, whereas allografts fail after an initial take, and that the
and plasma cells in the spleen appeared to be correlated.125 destruction of the foreign epidermis is brought about by a
However, actual proof for the involvement of B cells in an- mechanism of active immunization. Medawar returned to
tibody production was quite accidental.126 In 1952, Bruce Oxford University to study homograft rejection in laboratory
Glick, a young doctorate student at Ohio State University, animals and proved that this was an immunologic phenom-
was investigating the function of an obscure avian organ, enon. Medawar137 concluded that the mechanism by which
the “bursa of Fabricius.” He removed the organ, which did foreign skin is eliminated belongs to the category of “actively
not result in a discernable phenotype. A fellow graduate stu- acquired immune reactions.” Shortly after this publication,
dent asked to use one of Glick’s birds to develop an anti- Ray Owen138 published the provocative finding that dizy-
body against Salmonella and found that the bursectomized gotic twin calves, who share a common circulatory system in
chicken failed to make antibodies. This led to the publica- utero, exhibit chimaerism with respect to their twin’s eryth-
tion that eventually appeared in Poultry Science describing rocytes and fail to produce antibodies against each other’s
the role of the organ in the generation of bursa-derived or erythrocytes. This led Burnet and Fenner139 to predict that
“B” cells.127 As there was no anatomic equivalent of the bursa introduction of a foreign antigen early enough in life would
in mammals, an exhaustive search finally revealed the bone fail to elicit an immune response. Medawar reasoned that the
marrow origin of these cells. It was also found that thymus- successful exchange of skin grafts between dizygotic calves
derived cells, later named “T cells,” were needed to “help” would verify this hypothesis. Together with his postdoctoral
B cells produce antibody.128–130 These distinctions were fur- fellow Rupert Billingham, he performed a series of grafting
ther clarified when Cooper et al.131,132 showed that T cells experiments that provided direct support for this model.140
were required for delayed-type hypersensitivity and graft Mammals and birds never develop, or develop to only
versus host reaction. Thus was borne one of many central a limited degree, the power to react immunologically
dichotomies that Mazumdar133 has argued drives the field of against foreign homologous tissue cells to which they
immunology. have been exposed sufficiently early in fœtal life . . .
In 1957, Gowans134 cannulated the thoracic duct of rats this phenomenon is the exact inverse of “actively ac-
and measured the rate of flow of the lymph. He suggested quired immunity”, and we therefore propose to de-
that “the continuous entry of living lymphocytes into the scribe it as “actively acquired tolerance.”
blood may be essential for maintaining the output of lym-
phocytes from the thoracic duct.” He later showed that in- At the same time, Milan Hašek141 in Prague demonstrated
travenous transfusion of radiolabeled lymphocytes resulted successful parabiosis of chick embryos derived from two

different strains. Hašek’s stated motivation to perform the the 1950s, Dausset observed that patients who had received
experiment was to determine whether exchange of blood in many blood transfusions produced antibodies that could
the different chick strains induced a “mutual metabolic as- agglutinate white blood cells from donors but not the pa-
similation between parabionts,” rather than to induce a state tient’s own cells. Several of the patients produced antibodies
of immune tolerance. It has been suggested that Hašek’s real against the same antigen.145 Subsequent family studies indi-
motivation for the experiment was to advance the Lysenkoist cated that a genetically determined system, named “human
genetic doctrine to appease the local communist regime in leukocyte antigen” (HLA) system, was found to be the or-
Czechoslovakia.142 Regardless, the result has been reinter- tholog of H-2 in the mouse. Dissection of the human system
preted as an example of the induction of immunologic toler- would take many years, during which time transplantation
ance. In 1960, Medawar shared the Nobel Prize with Burnet surgeons made use of the emerging findings to assist in tis-
“for the discovery of acquired immunologic tolerance.” sue typing. In the 1960s, Baruj Benacerraf, an immunologist
working at New York University, noticed that some outbred
guinea pigs responded to simple antigens by developing de-
Mechanism versus Metaphor: “Self/Nonself” layed hypersensitivity reactions upon challenge while others
Discrimination did not, and that this was under genetic control. He termed
Medawar’s experiments and Burnet’s formulation of the these genes “immune response genes.” McDevitt and Sela
clonal selection theory occurred at a critical juncture in the observed similar genes in mice, and McDevitt showed that
history of the evolving field of immunology. Had their de- they are encoded in the MHC.146 In 1980, Benacerraf, Snell,
velopment been dyssynchronous, it is doubtful that much and Dausset shared the Nobel Prize “for their work on ge-
progress would have been made on either front. Indeed, netically determined structures of the cell surface that regu-
transplantation experiments had been performed earlier in lated immunologic reactions.”
the century with little insight as to why allografts failed. The Of course, the identification of the HLA region and the
viewpoint that Burnet espoused, that the function of the subsequent cloning of the genes encoded in this region
immune system is to distinguish “self” from “nonself,” has proved to be landmarks in the history of immunology.
proved to have enormous heuristic value ever since its for- Besides providing a molecular identity to the key orches-
mulation more than 60 years ago. Is this distinction mostly trators of antigen presentation to T cells, the identification
metaphorical, as suggested by Tauber,143 or does it reflect a of specific MHC alleles with autoimmune diseases has led
more concrete generative reality? To begin to address this to important insights into their pathogenesis. Ironically,
question, it is necessary to briefly review a parallel develop- the first recognized HLA association with human disease,
ment in immunology, the discovery of the components of HLA-B27, which was associated with the disease ankylos-
the immune system that define molecular self-hood. ing spondylitis,144,147 has generated perhaps more heat than
light, as there is no definitive mechanism that explains how
HLA-B27 is linked to disease pathogenesis, underscoring
Looking Under the Hood: The Discovery of the that the HLA locus, which took so many years to uncover at
Major Histocompatibility Complex the genetic level, still has much to teach us.
The clonal selection theory represented a conceptual break-
through in the history of immunology, but it did not explain
how lymphocytes actually recognize antigen. These insights The Discovery of Major Histocompatibility Complex
would eventually come from three sources over the span of Restriction as the Molecular Basis for
20 years: studies of the genetics of graft rejection in inbred “Self/Nonself” Discrimination
strains of mice by George Snell in the 1940s, studies of the It was known for several years that cooperation between
agglutination of white blood cells by sera from transfused T and B cells occurred in syngeneic or H-2–compatible ani-
patients by Jean Dausset in the 1950s, and studies of the im- mals.148–150 In 1972, Kindred and Shreffler151 showed that even
mune response to simple antigens in guinea pigs by Baruj in nude mice, cooperation between T and B cells required H-2
Benacerraf in the 1960s. Snell was interested in identify- compatibility; however, the exact role of that H-2 molecules
ing genes that controlled the ability of mice to resist tumor played in this process and the nature of the T- and B-cell in-
transplants. He pioneered the use of congenic mice, which teractions remained a mystery. A valuable clue was provided
are genetically identical except for a single region or locus. by experiments by Rosenthal and Shevach,152 who demon-
Snell observed that tumor grafts were accepted between strated that efficient presentation of antigen by antigen-
inbred mice but not between mice of different strains. The pulsed macrophages to T cells also required histocompat-
same was true for normal tissues. Snell termed the under- ibility matching. In 1974, Doherty and Zinkernagel sought to
lying genes “histocompatibility” genes. In collaboration understand the role of T cells in the immune response to viral
with Peter Gorer, who had prepared antisera that reacted meningitis. They theorized that it was the strength of the im-
with cells from one mouse strain but not another, Snell mune response that caused the fatal destruction of brain cells
established that the major histocompatibility locus cor- infected with this virus. To test this theory, they mixed virus-
responded to a reactivity that Gorer had designated anti- infected mouse cells with T lymphocytes from other infected
gen II, renamed locus histocompatibility 2 or H-2.144 Two mice. The T lymphocytes did destroy the virus-infected cells,
loci within this region, designated K and D, carried genes but only if the infected cells and the lymphocytes came from
specifying antigens involved in triggering graft rejection. In a genetically identical strain of mice. T lymphocytes would

ignore virus-infected cells that had been taken from another SPECIALIZATION WITHIN THE IMMUNE SYSTEM II.
strain of mice.153 Further experiments strongly suggested that
the same TcR that recognizes viral antigen also recognizes Discovery of B- and T-Cell Antigen Receptors
the MHC molecule.154 The implications of the Zinkernagel- The discovery of the B-cell receptor (BcR) for antigen began
Doherty experiment were profound. First, it established the with Ehrlich,171 when he proposed that cells that produced
principle of MHC restriction: T cells recognize antigen only “amboceptors” expressed them at their surfaces; in fact,
in the context of MHC molecules. Second, the experiment Ehrlich’s drawings of amboceptor-producing cells empha-
established that cytotoxic TcR-bearing cells must recognize sized this point (see Fig. 2.2). Evidence for the existence of
two separate signals on an infected cell before they can de- surface Ig was provided by indirect immunofluorescence
stroy it. One signal is a fragment of the invading virus that and autoradiography, in which immunoreactivity against
the cell displays on its surface and the other is a self-identify- a single class, IgM, was observed in 1970.172–175 The func-
ing tag from the cell’s MHC molecules; it was felt likely that tion of the BcR was uncovered by Rock and Lanzavecchia,
the same TcR probably recognizes both. Thus, the experi- who showed that MHC class II–restricted antigen presen-
ment pointed to the identity of the molecular structure that tation by hapten-specific B cells was enhanced 103- to 104-
constituted immunologic “self”—it is the MHC molecule— fold by specific binding, endocytosis, and loading of peptide
and therefore a virus-infected cell bearing MHC molecules antigen onto MHC class II molecules.176,177 In 1952, Colonel
was likely to constitute “altered” or “nonself.” Finally, the fact Ogden Bruton at the Walter Reed Army Hospital was caring
that MHC is highly polymorphic implies that any given al- for the 8-year-old son of a general. The boy had had recur-
lelic product will be capable of forming a different altered self rent pneumococcal infections, including bacteremia, but he
from other MHC allelic products; thus, the specific identity recovered with antibiotics. Bruton noted that the boy did
of the MHC molecule itself determines the strength of the not mount an antibody response to pneumococcal vaccina-
immune response. tion and upon testing his serum using Tiselius apparatus,
Although the Zinkernagel-Doherty experiment answered the electrophoretic pattern revealed a complete absence of
many questions, the steps between encounter of antigen by γ-globulins.178 Other cases appeared, and it was soon clear
an antigen-presenting cell (APC) and presentation of that that the defect was X-linked. This is often cited as the first
antigen to a T cell was somewhat of a “black box.” It had description of a primary immunodeficiency. It was not until
long been thought that intact antigen was presented to T 1993 that the molecular defect of agammaglobulinemia was
cells, but it was not until 1981 that Ziegler and Unanue155 uncovered. It was due to deficient expression of a tyrosine
showed that an antigen processing event was necessary for kinase, named Bruton tyrosine kinase.179,180 This discovery
I-region (MHC class II)-restricted antigen presentation to T highlights another function of the BcR, which is to help
cells. This appeared to require antigen processing in a lyso- provide signals for B-cell maturation. Without its expres-
somal-like compartment.156 Peptide loading is a complicated sion, B-cell development is blocked beyond the pre-B-cell to
affair, involving prior binding of MHC class II by an “in- immature B-cell stage.181
variant chain” to prevent premature loading of incompletely The discovery of the TcR for antigen was likened to the
folded proteins in the endoplasmic reticulum.157,158 Peptide hunt of the apocryphal Snark,182 according to Mak,183 except
loading requires proteolysis of the invariant chain.159,160 the TcR was finally captured after a hunt that lasted over
Mellman and colleagues would go on to show that the actual 20 years or more, depending on one’s perspective. The hunt
compartment in which antigen loading occurred is uniquely was characterized by extended periods of uncertainty, when
specialized for antigen presentation in B cells161 and DCs.162 it was not clear whether the same receptor bound to anti-
In a landmark paper, Unanue and colleagues purified MHC gen and MHC simultaneously, whether two separate TcR
class II molecules from 1011 B cells and showed a 1:1 bind- proteins bound MHC and antigen separately (the “intimacy
ing with peptide and MHC class II I-Ad, but not I-A k,163 thus model”), or whether one TcR molecule bound MHC mol-
demonstrating MHC restriction at the biochemical level. ecules altered by antigen. It was also not known what form
The processing events required for MHC class I restric- the antigen would actually take, and many considered it
tion seemed more elusive, as some of the molecular com- likely that the binding was strictly analogous to the bind-
ponents required for this were not known at the time. It ing of antibody and antigen, amounting to an “IgT-antigen”
was suspected that an intracellular proteolytic event was interaction.184 The existence of the TcR was deemed within
needed to process antigen, but that was not proven until reach when a clone-specific monoclonal antibody (mAb)
1994 when Rock et al.164 showed that proteasomal inhibitors against a murine lymphoma was isolated.185 Finally, in 1984,
blocked degradation of most cell proteins and subsequent using differential hybridization approaches, Davis and Mak
generation of peptides presented on MHC class I molecules. independently announced the cloning of the β -subunit of
The actual mechanism by which peptides generated in the the mouse and human TcR, respectively.186,187 This was soon
cytosol gained entry into the secretory compartment was followed by the cloning of the α-subunits of the mouse and
provided in 1990, when four groups announced the identi- human TcRs,188–191 and the identification of other subunits of
fication of a member of a family of ABC transporters, called a different type of TcR, the γδ TcR.192 Eventually, the crystal
“TAP,” which provided this function.165–168 Finally, in 1987, structures of pMHC combinations demonstrated that con-
Bjorkman et al.169,170 demonstrated that the antigen in ques- tact between TcR and pMHC was dominated by direct TcR-
tion was a peptide actually bound to the groove of the MHC MHC contacts, rather than TcR-peptide interactions.193–195
class I molecule. Beyond the initial recognition stage, signaling through the

TcR, as in the BcR, is a highly regulated process tuned to rec- flare upon further injection of fish extract.206 This prop-
ognize cues from pMHC that govern positive and negative erty was expected for a class of antibodies termed “reagins”
selection.196 There is little doubt that had Medawar survived, (from German reagieren, to react). IgE was identified as a
he would have appreciated how the clonal selection theory specific immunoglobulin class by Teruka and Kimishige
has remained a driving force underlying the molecular de- Ishizaka in 1966207,208 and was shown to be increased in sera
tails of antigen receptor signaling. from asthmatics.209 In 1993, Kinet and colleagues demon-
strated that its high affinity receptor, Fc εRI, was required
for anaphylaxis.210
Discovery of Distinct Immunoglobulin Classes IgD is the most recent Ig subclass to be recognized. It was
Ehrlich’s terminology for antibodies changed depending discovered from a patient with multiple myeloma.211 Surface
on the context; he sometimes referred to “amboceptors,” or IgD is coexpressed with IgM on mature B cells. The function
immune bodies, and at other times he used the more familiar of IgD is not well understood, although the abundance of
term “antikörper.” Regardless, he did not know the chemical IgM- IgD+ B cells in the human upper respiratory mucosa
makeup of antibodies, though he recognized that they must suggests that it is likely involved in mucosal immunity.212
contain separate binding sites for antigen and complement. Because IgA and IgE were isolated based on their pre-
Elucidation of the actual structure of antibodies would dominant location (IgA) or function (IgE), their role in
have to await the seminal work of Porter and Edelman in the immune response was relatively easy to decode, but the
the 1960s; however, the actual discovery of specific anti- relationships between the different Ig classes was unknown.
body classes, or isotypes, took place over many years. The It was discovered in 1963 that the early immune response
first class of antibodies to be discovered was IgG. In 1939, was initiated by the rapidly sedimenting (19S) antibody,
Tiselius and Kabat immunized rabbits with ovalbumin, later shown to be IgM, followed by the production of a
then absorbed a portion of the resulting antiserum with ov- more slowly sedimenting (7S) antibody, now known to be
albumin. When they applied samples of the unabsorbed and IgG.213,214 Although production of the different Ig classes was
antigen-absorbed antisera to electrophoresis, they observed initially interpreted as being due to the participation of dif-
a marked decrease in the amount of protein that migrated in ferent cells, it was later shown that single clones of B cells
the γ region of electrophoretic mobility, farthest away from were capable of producing multiple isotypes215–217 in a pro-
the fast-migrating albumin peak. They called the antibody cess called class switch recombination (CSR). The enzymol-
that corresponded to this fraction “γ-globulin.”197 ogy of CSR was worked out much later by Alt, Nussenzweig,
The next antibody class to be discovered was a result of Honjo, and others, who showed that it required DNA repair
an observation by the Swedish oncologist Jan Waldenström. enzymes218,219 and activation-induced cytidine deaminase
He described two patients with oronasal bleeding, lymph- (AID). AID is mutated in an autosomal recessive form of
adenopathy, low serum fibrinogen, and increased lymphoid the primary immunodeficiency, hyper-IgM type 2, in which
cells in the bone marrow.198 With the help of a colleague afflicted patients are deficient in CSR and somatic hypermu-
in Svedberg’s laboratory, he noted that serum from these tation, consistent with an important role for AID in both
patients contained an abnormally large amount of homo- processes.220–222
geneous globulin with sedimentation coefficients corre-
sponding to a molecular weight of more than one million.
Waldenström thought this corresponded to a preformed Discovery of Antibody Effector Functions:
large molecule, which became known as macroglobulin. Fc Receptors
Proof that a macroglobulin possessed antibody activity was Almoth Wright was the first to recognize the importance of
finally provided in 1967,199 which ultimately led to its cur- antibody-mediated effector functions other than neutraliza-
rent name, IgM. tion or complement-mediated lysis. His studies were ignored
IgA was discovered by Gugler et al.200 who isolated IgA for a long time, and it would take many years to appreciate
from human milk, and Heremans et al.201 who isolated IgA the importance of the “other end” of the antibody molecule.
from human serum. It was later found in high concentra- Since the 1970s, studies on the effector functions of the Fc
tions in all exocrine secretions202 and further characterized portion of IgG made extensive use of several in vitro model
by Tomasi and Zigelbaum,203 who suggested that IgA plays systems, often involving phagocytosis.223–232 The eventual
an essential role in mucosal immunity. It was not until 1984 cloning of receptors for the Fc portion of IgG (FcγRs) by
that the mechanism of secretion of IgA and IgM across epi- Ravetch and et al.233 revealed many similarities and some
thelial barriers was uncovered; it was shown to depend on differences, but these early studies did not reveal how the
interactions of an Ig-associated polypeptide, the J chain, receptors transduced signaling events, such as phagocyto-
with a glycoprotein on the surface of epithelial cells, referred sis. It was not until 1989 that Michael Reth234 noted a short
to as the “polymeric IgA receptor.”204 This receptor medi- tyrosine-containing sequence in common with several an-
ated transcytosis of secretory IgA and IgM. tigen receptors, including subunits of the TcR, the BcR,235
The discovery of IgE had a particularly long gestation and Fcγ and Fcε receptors236 that the signaling function of
period and is well summarized in a review.205 It began with FcγRs was understood in a larger context. Following recep-
a report by Prausnitz, in 1921, who injected his forearm tor engagement, tyrosine residues within this consensus
with serum from a coworker allergic to fish (and coauthor sequence, subsequently named “immunoreceptor tyrosine-
on the ensuing paper); this was followed by a wheal and based activation motif” (ITAM), become phosphorylated

by Src family tyrosine kinases and serve as docking sites for thymus-derived antigen-sensitive cell-bound antigen con-
Syk tyrosine kinase,237 or ZAP-70, in the case of the TcR. The taining at two separate sites, one for a receptor on the “an-
membrane-associated tyrosine kinases become activated and tiantigen-sensitive” cell and another bound to a “carrier”
phosphorylate substrates that further convey downstream antibody. If the second site on the antigen was occupied by
signals. Absence of ZAP-70 leads to a severe combined im- antibody, then that would deliver a signal to the “antianti-
munodeficiency disease (SCID).238 These studies uncovered gen-sensitive” cell to induce immunity. In contrast, if the
the central role of nonreceptor tyrosine kinases in the im- site remained unbound, the antigen would induce toler-
mune system, which appeared to explain in large part how ance. In this model, two cells were involved, but they were
antigen and Fc receptors evoke calcium signaling,239 degran- separated in time and space. Among the problems with the
ulation,240 phagocytosis,229,241,242 and antibody-mediated cel- model was its circularity: What would provide the stimulus
lular cytotoxicity.243 Furthermore, the ITAM-containing γ- for the production of the carrier antibody in the first place?
chain244 that is associated with FcγRs is required for immune Other theories were proposed to explain the phenomenon of
complex-mediated glomerulonephritis in mice.245 Thus, it alloreactivity. The most intriguing of these was provided by
is likely that the hypersensitivity phenomena originally ob- Lafferty and Cunningham,252 who proposed the existence of
served by von Pirquet at the beginning of the 20th century, an “antigen bridge,” which was now provided by another cell
and later attributed to immune complexes by Dixon et al.,80 to provide “Signal 2.” When MHC restriction was described
has a similar pathophysiology, at least in part. in 1977, this model was modified somewhat to include the
provision that the second signal was triggered by MHC on
the stimulator cell253 (Table 2.1).
THE HERMENEUTICS OF LYMPHOCYTE ACTIVATION
Evolution of Early Lymphocyte Activation Models Discovery of Costimulation: T Cells as Beneficiaries
By the late 1950s, it became clear that lymphocytes were By 1975, models for lymphocyte activation had matured,
highly adept at interpreting or translating environmental but many details remained sketchy, particularly the identity
cues and responding by maintaining a state of activation or of the elusive Signal 2. In 1987, Jenkins and Schwartz utilized
tolerance. The focus of immunology had shifted from iden- a system designed to “convert” Ag-specific T-cell clones into
tifying the relevant cell types involved in the immune re- suppressor cells. They used a model system in which T-cell
sponse to discovering what can be viewed as the “hermeneu- clones against a pigeon cytochrome peptide were incubated
tics” (a term derived from the Greek ε‘ ρμηνευ′ω, “translate” with ethylene carbodiimide-fi xed Ag-pulsed splenocytes as
or “interpret”) of lymphocyte activation. Based on the clonal APCs. They obtained the surprising result that the T-cell
selection theory, Burnet246 proposed that activation or toler- clones became unresponsive to subsequent restimulation
ance was the result of an antigen stimulating a single cell, with untreated APCs plus peptide, although they remained
depending on the age of the organism. This was in keeping viable and proliferated in response to interleukin (IL)-2.254
with experiments in which tolerance was induced by expos- This showed that Ag-specific unresponsiveness could be
ing antigen during fetal life or shortly thereafter.138,140,141 In induced in the absence of suppressor cells and that stimula-
1959, Lederberg247 modified this model to describe activation with cognate antigen alone was insufficient to induce
tion or tolerance as occuring depending on the age of the proliferation; the missing Signal 2 was defi ned as “costimu-
cell rather than the organism. lation” and ultimately identified as cluster of differentiation
If an antigen is introduced prior to the maturation of (CD)28-mediated signaling by Allison and colleagues.255
any antibody-forming cell, the hypersensitivity of such Since this seminal discovery, an entire family of CD28-like
cells, while still immature, to an antigen-antibody molecules in T cells and their cognate ligands on APCs has
reaction will eliminate specific cell types as the arise been identified,256,257 which includes an inhibitory member
by mutation, thereby inducing apparent tolerance to of the family, cytotoxic T-lymphocyte antigen-4.258 This has
that antigen. led to the clinical development of cytotoxic T-lymphocyte
antigen-4–Ig, an inhibitor of CD28,259 which was approved
However, the model could not account for several obser- for the treatment of rheumatoid arthritis by the U.S. Food
vations, among which were that the dosage and specific and Drug Administration in 2006 and by the European
form of the immunogen, such as a hapten, could also influ- Medicines Agency in 2007.
ence the outcome of the encounter. This led Talmage and
Pearlman248 to propose an alternative model in which anti-
gen alone would induce tolerance, but aggregated antigen, Discovery of the Mechanism of B-Cell Help
perhaps associated with complement, could induce an addi- It was known since the 1960s that T cells were needed to
tional nonspecific stimulus to trigger clonal expansion. The “help” B cells produce antibody, but the molecular nature
suggestion of the need for a second “nonspecific” stimulus, of that help was unknown.129,260 Experiments had suggested
and particularly complement, was remarkably prescient in that T-cell help required cell-to-cell contact, implying the
light of the finding that complement can provide exactly existence of the involvement of cell surface receptor–ligand
such a stimulus to amplify immunogenicity by 103- to 104- interactions. Agonistic antibodies against CD40 drove
fold.249 In 1968, and later revised in 1970, Bretscher and B cells into cell cycle, raising the possibility that a cognate
Cohn250,251 provided a variation on this model, in which a ligand on T helper cells participates in the mechanism of

TABLE 2.1 Different Models of Lymphocyte Activation

Basis of Immune Recognition Signals Components Conditions for Tolerance Authors
Self versus nonself 1 B cell/Ag Immature organism Burnet, 1959230
1 B cell/Ag Immature Ab-forming cell Lederberg, 1959231
2 B cell/Ag–Ab + complement Absence of bound Ab/ Talmage and Pearlman,
complement 1963232
2 Humoral Ag-sensitive responder Absence of carrier Ab or Bretscher and Cohn,
cell/bivalent Ag/carrier lack of 1968,234 1970235
antigen-sensitive thymus-
dependent cell
2 Signal 1: Ab-bound responder Absence of Signal 2 Lafferty and Cunningham,
cells + Ag 1975236
Signal 2: Stimulator cell + Ag
Noninfectious self versus 2 Signal 1: Lymphocyte + MHC/Ag Absence of Signal 2 Janeway, 1989390
infected nonself Signal 2: PAMP+ APC
Danger 3+ “Signal 0”: Alarm Absence of “Signal 0” Matzinger, 1994451
Signal 1: (Th cell-APC–Ag;
Tc-infected cells; B cell/Ag;
Signal 2: Help (Th to B) or
costimulation (APC to T)
Abbreviations: Ag, antigen; Ab, antibody; APC, antigen-presenting cells; MHC, major histocompatibility complex; PAMP, pathogen-associated molecular pattern; Th, T helper.
T-cell help. Using a CD40–Ig fusion protein, murine CD40L 1989, at about the same time that the ITAM consensus se-
was cloned.261,262 Use of a mAb against the human CD40L quence was identified, two groups showed that Lck was ca-
confirmed that expression was restricted to mantle and cen- pable of phosphorylating subunits of the CD3 complex and
trocytic zones of lymphoid follicles and the spleen periarte- the ζ subunits of the TcR.280,281 Following the discovery of
riolar lymphoid sheath in association with CD40 + B cells,263 another key tyrosine kinase associated with the ζ subunit of
which is the distribution that might be expected for a highly the TcR, ZAP-70, by Chan and Weiss,238 a complicated cell
localized signal to provide T-cell help to B cells. Indeed, signaling paradigm began to emerge in which “Signal 1,” de-
in 2000, the specific type of T cell that provided CD40L- livered by pMHC expressed on either APCs or virus-infected
mediated help to B cells, now called a T follicular helper cell, cells, is conveyed by a series of phosphorylation events,
was identified by its unique anatomic location and expres- leading to phosphorylation of phospholipase C-γ, trigger-
sion of a homing receptor, CXCR5.264,265 In 1993, four groups ing increases in cytosolic calcium and dephosphorylation
independently announced that defects in the CD40 ligand of a key transcription factor, NFAT. The elusive machin-
gene are responsible for X-linked hyper-IgM syndrome in ery of “store-operated calcium channels” required for cal-
which affected adults express elevated levels of IgM and cium-based signaling282 was finally elucidated in 2006.283–286
defects in class switching.266–269 Defects in this pathway in lymphocytes lead to a form of
SCID.286 Dephosphorylated NFAT, alone or in conjunction
with AP-1, translocates to the nucleus where it binds to sites
T-Cell Subsets and T-Cell Signaling Paradigms on the promoters of key genes such as IL-2, culminating in
The widespread use of mAbs since their discovery by Köhler T-cell proliferation.287 The initial interaction between TcR-
and Milstein resulted in the production of many antibod- bearing T cells and pMHC-bearing APCs induces a coop-
ies directed against T cells that had nonoverlapping pat- erative series of protein–protein interactions in a spatially
terns of expression. Much of this early work was done by delimited fashion (the “immunologic synapse”) that facili-
Schlossman, Reinherz, and colleagues.270–275 In most cases, tates signaling events.288 We now know that signal transduc-
the antibodies themselves identified T cells with specific tion by the TcR shares common elements, and in some cases
functions; and in 1987 and 1988, two groups found that identical kinases and substrates, with signaling by other
the molecules recognized by two of these mAbs, denoted by ITAM-containing receptors, such as the BcR, and Fc recep-
their cluster designations, CD8 and CD4, mediated adhe- tors or IgG and IgE.289–291 Recent data suggest that similar
sion to MHC class I and II, respectively.276,277 CD4 and CD8 mechanisms governing “immunologic synapses” in T cells
were shown to bind to nonpolymorphic regions of MHC may also be characteristic of “phagocytic synapses,”292 un-
molecules,277 suggesting that they might stabilize other- derscoring a further degree of conservation of ITAM-based
wise weak interactions between TcR and pMHC. Later ex- pathways. It is ironic that lymphocytes, which had appeared
periments demonstrated that the cytosolic domains of these morphologically uninteresting to so many biologists until
coreceptors were able to bind, and recruit, the nonreceptor the latter part of the 20th century, would share so many
Src family tyrosine kinase Lck.278,279 It was not known at the characteristics in common with their visually more intrigu-
time what the relevant substrates of Lck might be, but in ing cousins.

Cell to Cell Communication: Mice Have a Tale to Tell other species in which the technique has been used. In 2007,
The first cytokine to be characterized is credited to Isaacs Smithies, Capecchi, and Evans shared the Nobel Prize “for
and Lindenmann,293 who studied an influenza virus-induced their discoveries of principles for introducing specific gene
factor from chick chorioallantoic membranes that blocked modifications in mice by the use of embryonic stem cells.”
a second viral infection; this factor was called “interferon”
(IFN). It was not purified to homogeneity until 20 years
later.294,295 Type I IFNs, which include IFNα and IFN-β, and Chemokines
type II IFN, IFNγ, bind to distinct receptors. A major ad- If cytokines determine what cells do, then chemokines de-
vance in our understanding of cytokine signaling occurred termine where cells go. The first chemokines purified were
in 1992, when Schindler et al.296 demonstrated the tyrosine derived from platelets, although neither protein (platelet
phosphorylation and nuclear translocation of a complex of factor 4 [CXCL4]314,315 and β -thromboglobulin [CXCL7]316)
four proteins, named IFNα stimulated gene factor 3. IFNα was recognized to be chemotactic when first identified.
stimulated gene factor proteins were isolated from nuclear IL-8 (CXCL8) was the first chemokine to be purified and
extracts derived from over 1010 cells, followed by sequencing sequenced based on its chemotactic function for neutro-
and cloning of their genes; two of these proteins are mem- phils,317 and its receptors were cloned in 1991.318,319 The word
bers of what are now known as STAT proteins, which were “chemotaxis” was coined by Pfeffer in 1884, who observed
shown to be substrates for members of the Janus-associated spermatids migrating toward a pipette containing maleic
kinase family.297–303 The most common severe immunodefi- acid salts.320 It is not clear who first explicitly observed che-
ciency attributable to defective cytokine signaling is X-linked motaxis of leukocytes, although several scientists, notably
SCID, which is due to mutation in the common γc gene that Schultze, Lieberkühn, Davaine, and Wharton Jones de-
is shared by receptors for IL-2, IL-4, IL-7, IL-9, IL-15, and scribed amoeboid movement in leukocytes in the middle of
IL-21.304 Mutations in Janus-associated kinase 3 were found the 19th century.321 Since then, chemotaxis has been stud-
in two patients with a severe form of SCID indistinguishable ied in vitro using various techniques, including specialized
from X-linked SCID.305,306 Dominant-negative mutations chambers that facilitate this process.322
in STAT3, which is important in the signal transduction of The effects of chemokines on cells are complex, but a
IL-6 and IL-10, were found to be responsible for hyper-IgE major function is to orchestrate the process of diapedesis,
syndrome, a disease associated with highly elevated serum the process of transendothelial migration first described by
IgE, recurrent staphylococcal skin infections and pneumo- Dutrochet in 1824.323 In 1979, Hayward et al.324 described six
nia, and skeletal abnormalities.307,308 infants from two families whose umbilical cords were still
Much of our knowledge of this pathway, as well as many attached at 3 weeks of age. Five of these developed severe
other components of the immune system, is due to the use local and disseminated infections from which four died. The
of gene targeting in mice. In 1985, Smithies et al.309 intro- molecular defect was identified by Springer et al.325 in 1984,
duced a short DNA sequence from the human beta-globin who showed that leukocytes from patients with the disease
locus into an erythroleukemia cell line and were able to lacked all known β2 integrin adhesion receptors. This was
detect a specific exchange of the beta-globin gene with the eventually traced to absent or abnormal β2 subunits (CD18).
homologous sequence in about 1 in 103, demonstrating the Marlin and Springer326 later described the requirement of
feasibility of gene targeting. At the same time, Capecchi the β2 integrin, LFA-1, to bind its counter-receptor, ICAM-
introduced DNA directly into the nucleus of a cell using a 1, on endothelial surfaces for diapedesis to occur normally.
microelectrode. Capecchi noted that multiple copies of the The function of chemokines is to provide chemotactic gra-
introduced gene were integrated into the host cell’s chromo- dients as well as activate the integrins for tight adhesion
some through homologous recombination. These studies prior to the active participation in endothelial cells in dia-
established the potential for homologous recombination in pedesis.327,328 Interestingly, LFA-1 was shown to be required
somatic cells. The next major step was based on the ability of as an accessory molecule during interaction of T cells with
using blastocyst-derived embryonic stem cells to introduce their cognate targets.329
genes into the germline of the mouse.310,311 By injecting Chemokines are critical for initiating the primary im-
blastocysts with cultured embryonic stem cells that were mune response as APCs, such as DCs. DCs and naïve T
infected with a retrovirus, Evans and colleagues gener- cells both share a common chemokine receptor, CCR7.
ated chimeric mice in which retroviral DNA was detectable T cells migrate through high endothelial venules, which
in both somatic and germ-line cells.311 Eventually, Evans, secrete the chemokine SLC (CCL21). SLC is needed for
Smithies, and Capecchi refined these techniques, which integrin-mediated adhesion of the T cells.330 Mice lacking
led to the first knockout mouse, in which the gene encod- expression of SLC have defects in lymphocyte and DC local-
ing hypoxanthine guanine phosphoribosyl transferase was ization to the T-cell zone of secondary lymphoid organs.331
deleted. The resultant phenotypic resembled that of the The chemokine-directed “patrolling” of naïve T cells in the
Lesch-Nyhan syndrome, which is characterized by mental lymph nodes is probably the major mechanism by which the
retardation and self-mutilation.312,313 Today, gene targeting very few lymphocytes with the relevant TcR specificity re-
techniques are used worldwide to study the effects of dele- main there to interact with the antigen-loaded APCs during
tion or overexpression of genes important in immunity and an immune response. Thus, chemokines are essential com-
any disease that can be reproduced in the mouse and in ponents of the “clonal selection” of lymphocytes.

T-Cell Polarization: The Power of Dichotomy then serve to regulate the immune response. While appeal-
In 1986, Coffman and Mosmann at DNAX Research ing, efforts to prove the importance of this concept have
Institute were interested in determining factors that dif- achieved limited success. In 1970, Gershon and Kondo344
ferentiated different types of T helper cell lines. One such found that thymus-derived cells could specifically induce
line was capable of inducing a 100-fold increase in IgE secre- tolerance, and they and others spent many years trying to
tion from mouse splenocytes, which was potently blocked isolate antigen-specific T suppressor cells but were unable to
by IFNγ.332 When the team characterized other cloned T-cell do so. In the 1990s, the field shifted away from clonotypic
lines, it became clear that there were very specific patterns suppressor cells toward T cells secreting specific tolerogenic
of secretion: Some clones, which they called TH1 cells, pro- factors, such as IL-10–secreting “Tr1 cells” and transform-
duced IL-2 and IFNγ, while other clones, which they called ing growth factor-β –secreting “Th3 cells.”340,345 This was
TH2 cells, produced factors that stimulated B cells (B-cell– followed by the identification of cell surface markers of a
stimulating factor-1), mast cells, other T cells, and IgE and subset of CD4 + cells that were required to suppress auto-
IgG1 secretion. All clones produced IL-3.333 Further work, immunity in mice. These turned out to be CD25 and CD5
in collaboration with Paul, who first identified B-cell– in a CD4 + population of cells in the mouse, cells which
stimulating factor-1,334–336 clarified that B-cell–stimulating were later called regulatory T cells. In 2001, the FoxP3 gene
factor-1 was the same factor that stimulated mast cells and T was identified as the gene that was mutated in Scurfy mice,
cells, later renamed IL-4. Coffman and Mosmann337 defined which develop severe autoimmunity as a result of a single
the now well-established paradigm of polarized T helper gene mutation.346 Similarly, mutations in the human homo-
cell secretion; they demonstrated that IL-4 and IFNγ recip- log of FoxP3 were found to be associated with the disease
rocally inhibit the outgrowth of TH1 and TH2 cells, respec- called IPEX (immune dysregulation, polyendocrinophathy
tively. The later identification of transcription factors that enteropathy, X-linked syndrome).347–349 Analogous to the
are selectively expressed in these cells (GATA-3 and T-bet, capacity for T-bet and GATA3 to drive TH1 and TH2 differ-
respectively) and can even reprogram the cells to transdif- entiation, respectively, expression of FoxP3 was sufficient to
ferentiate toward the “opposite” phenotype further substan- drive the differentiation of T helper cells to a CD4 + CD25 +
tiated this dichotomy.338,339 These findings had profound im- regulatory T phenotype.350,351
plications for immunologists, who began to classify diseases Central tolerance, sometimes referred to as “negative
according to their predominant TH cytokine profiles. Thus, selection,” is a means of shaping the repertoire that was
tuberculosis and most bacterial and fungal infections pro- thought likely to exist following the discovery of the thy-
duced a TH1 pattern of cytokines, whereas asthma and other mus as an immune organ by Miller in 1961; however, proof
allergic diseases produced a TH2 pattern of cytokines. While for this was lacking. How could one prove the existence of
this reductionist view of cytokine production is no doubt an a phenomenon that predicted the absence of a cell, rather
oversimplification, this simple dichotomy has proved quite than its presence? It was not until 1987 that Kappler et al.352
robust and has helped guide the development of novel thera- used a mAb against a specific TcR Vβ segment to show that
peutics for various diseases. the mAb recognized immature thymocytes but not mature
There is, however, a danger to oversimplification340 — thymocytes or T cells, thus proving that negative selection
sometimes it is necessary to be a “fox” rather than a “hedge- occurs. Von Boehmer demonstrated a similar phenomenon
hog.”341 Some diseases, such as Crohn disease, an inflam- using a TcR transgenic mouse model.353 Negative selection
matory bowel disease that was thought to be driven by the of immature B cells was also demonstrated in the classic ex-
TH1-polarizing cytokine, IL-12, is more likely driven by a periments of Nossal and Pike,354 Sidman and Unanue,355 and
related cytokine, IL-23, with which IL-12 shares a common Raff et al.356
subunit; indeed, a genome-wide association study in 2006 Clearly, the field of immunoregulation is still evolving.
revealed a highly significant association of the gene encod- Novel mechanisms of immunoregulation will surely be dis-
ing a subunit of the IL-23 receptor and early-onset Crohn covered, and novel regulatory interactions will be uncov-
disease.342 Ustekinumab, a mAb against the IL-12/IL-23 p40 ered. The role of tolerogenic DCs,357,358 CD8 + T cells,359,360
common subunit was approved by the U.S. Food and Drug and myeloid-derived suppressor cells361–363 are likely to play
Administration for the treatment of moderate-to-severe important roles and some of these will likely find their way
Crohn disease in 2006 and for moderate-to-severe plaque to the clinic. Perhaps, however, we have made some progress
psoriasis in 2009. in support of Ehrlich’s contention364 :
The organism possesses certain contrivances by means
The Language of Immunoregulation: of which the immunity reaction, so easily produced by
Explaining the “Contrivances” all kinds of cells, is prevented from working against
the organism’s own elements.
The concept of immunoregulation is implicit in the term
“immunity” (exempt), and the history of how different im-
munologists have viewed immunoregulation, depending on Immune Subversion: The Case of Tumors
their perspectives, is worthy of a chapter in itself. Jerne343 The idea that the immune system is capable of respond-
hypothesized that antibodies can act as antigens and elicit ing to tumors dates back to Ehrlich,365 who predicted that
an immune response against their idiotypes, which would cancer would occur at a high frequency in the absence of

an immune response. This theme was recapitulated by oth- until levels fell below a certain level, thus CD47 served as
ers, notably Burnet,366 who argued that T cells would be an aging “clock” that signified a “don’t eat me” signal.381 In
prominent in what he termed “immune surveillance.” The 2010, in a remarkable study, expression levels of CD47 on
field enjoyed a resurgence in 1994 when Schreiber and col- non-Hodgkin lymphoma (NHL) cells were negatively cor-
leagues demonstrated that tumors expressing dominant- related with survival. Blocking anti-CD47 antibodies pref-
negative IFNγ receptors demonstrated enhanced tumorige- erentially enabled phagocytosis of NHL cells and synergized
nicity in syngeneic mice.367 This was supported by studies with rituximab. Treatment of human NHL-engrafted mice
using IFNγ receptor-deficient or STAT1-deficient mice.368 with an anti-CD47 antibody reduced lymphoma burden
Additional evidence for an active immune response against and improved survival, while combination treatment with
tumors is the spontaneous development of autoimmunity rituximab (anti-CD20) led to elimination of lymphoma and
in some individuals with tumors; for example, paraneo- cure.382 In these settings, CD47, rather than MHC proteins,
plastic cerebellar degeneration occurs in individuals with a serve as a marker of “self.” It is somewhat counterintuitive
cytotoxic T-cell response to a shared antigen on tumor cells that, at least in the case of NHL, tumor masquerading as self
and neuronal cells.369 The immune response to tumors is is one mechanism by which tumors evade the immune sys-
complex, with participation from cytotoxic T cells, natural tem. Ironically, downregulation of MHC class I molecules,
killer (NK) cells, DCs, and myeloid cells.370 The “counter- the classic marker of self, is yet another mechanism of how
response” by the tumors, which has been termed “escape” tumor evasion of immunity.370
or “evasion,” is equally complex.370 While it is beyond the
scope of this chapter to discuss any of these in detail, sev-
eral relevant points of historical interest should be noted. In Immune Hijacking: The Discovery of
1976, a postdoctoral fellow named Mike Bevan was study- Human Immunodeficiency Virus
ing alloreactive cytotoxic T-cell responses and was able to The first description of what would come to be known as
demonstrate that spleen cells from an H-2b /H-2d–restricted the “acquired immunodeficiency syndrome” appeared in
mouse primed with H-2b cells that differed in minor his- 1981.383 Initially described in five homosexual men, the dis-
tocompatibility loci contained increased numbers of both ease soon was apparent in Haitians, transfusion recipients,
H-2b – and H-2d–restricted cytotoxic CD8 T cells.371,372 Bevan infants, Africans, and female sexual contacts of infected
interpreted this as evidence of “cross-priming,” in which men.384 In just 2 years, Montagnier’s team published the
ingestion of cell-associated antigens by phagocytosis led to first paper demonstrating the presence of retroviral particles
cytosolic entry of antigens and loading onto MHC class I from diseased patients,385 and a year later, Gallo’s group pub-
molecules and then presentation to cytotoxic T cells. It was lished five papers in Science providing convincing evidence
completely novel and counterintuitive, but subsequent stud- that this retrovirus was the cause of acquired immunodefi-
ies validated this interpretation.373 Recent progress has been ciency syndrome.386–390 Barré-Sinoussi and Montagnier won
made in our understanding mechanisms by which antigens the Nobel Prize for their discovery of human immunodefi-
gain entry into the cytosol from phagosomes.374 These stud- ciency virus (HIV), along with zur Hausen, for demonstrat-
ies were based on key findings by Desjardins and colleagues ing that human papillomavirus can cause cervical cancer.
who first demonstrated recruitment of endoplasmic reticu- Many felt that Gallo should have been awarded the prize
lum membrane to nascent phagosomes.375,376 Mice that lack as well.
CD8α + DCs, a cell type that is adept at cross-priming, were Pneumonia due to Pneumocystis carinii (later re-named
unable to demonstrate cross-priming and were incapable of P. jerovici) was common in initial cases of acquired immu-
generating a cytotoxic response against West Nile virus and nodeficiency syndrome, but patients soon presented with
a highly immunogenic fibrosarcoma tumor.377 Thus, cross- an array of opportunistic infectious diseases, malignancies
priming is important in tumor immunity. (eg, Kaposi sarcoma and lymphoma), and even autoimmune
Tumor cells are capable of circumventing the host im- diseases. In 1986, Maddon et al.391 demonstrated that CD4
mune response in various ways, including downregulation is an essential receptor that mediates HIV-1 entry into lym-
of MHC class I molecules, although this would potentially phocytes; this was followed in 1996 by the demonstration of
render them susceptible to NK-mediated killing, accord- CXCR4 and CCR5 as coreceptors for HIV-1 entry.392 Since
ing to the “missing self” hypothesis.378 Tumors elaborate a then, many laboratories worldwide have dedicated their ef-
host of cytokines and growth factors that directly inhibit forts to uncovering HIV pathogenesis.393,394 One of the major
various components of the immune system.370 An additional challenges to the immune system as well as the development
means of tumor evasion specifically bears on the meaning of of an HIV-1 vaccine is the enormous plasticity of the viral se-
“ immunologic self.” Most cells in the body express a cell sur- quence due to the high error rate of reverse transcription.393
face protein, CD47, which is a ligand for a receptor present Intensive efforts in academia and the pharmaceuti-
on macrophages and DCs, SIRPα .379 SIRPα interacts with cal industry have resulted in highly active antiretroviral
tyrosine phosphatases via its cytosolic domain380 ; therefore, therapy based on targeting multiple steps of viral replica-
corecruitment of SIRPα to tyrosine kinase-coupled signal- tion, mostly focusing on reverse transcriptase and protease.
ing scaffolds is predicted to inhibit kinase-mediated signal- A disease that was once virtually 100% fatal is now man-
ing events. Lindberg and colleagues showed in 2000 that ageable, resulting in an 80% to 90% decrease in mortality
CD47 on erythrocytes in mice prevented their phagocytosis rates in the United States and Europe.393 Among the major

challenges facing immunologists and virologists is eliminat- mechanism to account for the observation that some host
ing the latent reservoir of virus in resting memory CD4 + immune cells recognize and kill virus-infected or neoplastic
lymphocytes, developing an effective HIV-1 vaccine, and cells lacking expression of MHC class I molecules. NK cells
providing treatment for the 90% of infected individuals also provide an early, immediate source of cytokines, allow-
worldwide who reside in developing countries and have poor ing them to rapidly respond to injury prior to maturation of
access to antiretroviral therapy. the primary immune response.405
The “modern era” of the study of innate immunity
began in 1989. In a characteristic example of his remark-
METCHNIKOFF’S LEGACY able insight, Janeway406 published an article in which he
The Rediscovery of Innate Immunity described a new model for immune recognition. Janeway
Arguably, the most important advance in immunology in argued that rather than distinguish “self ” from “nonself,”
the last 15 years has been the “rediscovery” of innate immu- the immune system had evolved to distinguish “noninfec-
nity. The closest term to “innate immunity” that was used at tious self from infected nonself ” (see Table 2.1). Janeway
the turn of the century was “natural immunity.” When the compared the innate and acquired immune systems. He
cellularists and the humoralists were debating the relative emphasized that the former evolved to respond rapidly to
importance of cells and soluble antikörper at the turn of the “pathogen associated molecular patterns” by as yet un-
century, much of the phenomena that Metchnikoff observed identified receptors in a nonclonal fashion. Janeway was
under the microscope represented different aspects of innate acutely aware of the significance of Freund’s discovery of
immunity. However, Metchnikoff distinguished enhanced adjuvant407 and the implications this had for the primary
immunity due to vaccination as a result of what we would immune response. What was remarkable about Janeway’s
call today an acquired immune response: “an agglutinative hypothesis is its intuitive appeal and its Darwinian flavor,
substance...in the...fluids of the body becomes much more in some ways analogous to Metchnikoff ’s view of phagocy-
developed in those of immunised animals.”395 He thought tosis. In addition, it had great predictive power. In a com-
that infections could lead to an acquired immune response, pletely independent line of investigation, Jules Hoffmann
but the nature of that response was to foster an enhanced became interested in exploring inflammatory pathways in
response in phagocytic cells.395 Drosophila. Hoffman was aware of Baltimore’s discovery of
nuclear factor κ-light-chain-enhancer of activated B cells
In certain infective diseases terminating fatally a very
(NF-κ B), a transcription factor, originally identified as a
marked phagocytosis is observed even in susceptible
lipopolysaccharide (LPS)-inducible transcription factor in
animals . . . The acquisition of immunity against mi-
B cells in 1986.408 Mice rendered deficient in NF-κ B demon-
croorganisms is, therefore, due not only to the change
strated multifocal defects in immune responses, including
from negative to positive chemiotaxis, but also to the
those to LPS.409,410 Based on his observation of similarities
perfecting of the phagocytic and digestive powers of
between the cytokine-induced activation cascade of NF-κ B
the leucocytes.
in mammals and the activation of the morphogen dorsal
In 1932, Fleming396 recounted the discovery of lysozyme, in Drosophila embryos, Hoffmann demonstrated that the
which he made in 1921, “because its importance in con- dorsoventral signaling pathway and an extracellular toll li-
nection with natural immunity does not seem to be gener- gand control expression of antifungal peptide gene expres-
ally appreciated.” Although the discovery has been cited sion. Mutations in the toll signaling pathway dramatically
to represent another example of scientific serendipity, re- reduced survival after fungal infection.411 Until that time,
sulting from accidental dripping of nasal secretions from there was only a limited understanding of pathogen rec-
Fleming himself onto a culture plate, 397 Fleming wrote ognition in eukaryotes; the only receptor that was known
that “cultures of nasal mucus were made from a person to participate in the recognition of LPS was CD14, a mol-
suffering an acute cold” when the bacterolytic phenom- ecule that is expressed predominantly on macrophages.412
enon due to lysozyme was observed.396 Fleming would go Shortly after publication of Hoffmann’s landmark study,
on to discover penicillin, which was a bona fide example Janeway and Medzhitov published an article in Nature that
of scientific serendipity, and receive the Nobel Prize for described the cloning of a human homolog of Drosophila
his discovery. toll. A constitutively active mutant of human toll trans-
Of course, lysozyme is but one of many innate immune fected into human cell lines induced the activation of NF-
molecules important in the early phase of host defense. κ B and the expression of the inflammatory cytokines IL-1,
Other cationic proteins include defensins, which are secreted IL-6, and IL-8 as well as the expression of the costimulatory
by leukocytes and epithelial cells, and chemokines, whose molecule B7.1 “which is required for the activation of naïve
tertiary but not primary structures are related to defensins. T cells.”413 The choice of these proteins was not accidental,
Some members of both classes have the dual role of recruit- and it was clear that Medzhitov and Janeway had Janeway’s
ing inflammatory cells to sites of infection and killing bacte- hypothesis of 1989 in mind. Meanwhile, Beutler et al.414
ria.398–401 Other examples of components of innate immunity were zeroing in on the genetic identification of a mutation
include “innate immune lymphocytes,” such as NK cells, in a mouse strain that was incapable of responding to LPS,
NK T cells, and γδ T cells. Although it is beyond the scope a component of the cell walls of all gram-negative bacteria.
of this chapter to describe these cell types except in passing, Beutler and Cerami were the fi rst to show that immuniza-
the discovery of NK cells in 1975402–404 provided a cellular tion against tumor necrosis factor, the prototypic cytokine

that is produced following administration of LPS, pro- In 1978, Segal et al.445,446 identified the molecular defect in
tected mice from lethal shock.414 Beutler demonstrated that X-linked chronic granulomatous disease as the absence of
the codominant Lpsd allele of C3H/HeJ mice corresponded cytochrome b. Since then, all the components of this multi-
to a missense mutation in the toll-like receptor-4 gene.415 protein enzyme complex have been identified.447–452 Of note
Since these publications, a family of innate immune recep- is that the oxidase responsible for X-linked chronic granu-
tors has been defi ned, each responding to different ligands lomatous disease, now referred to as NOX2, is but one of a
encountered by hosts during infection.416 Initial interac- family of oxidases that are widely expressed and have been
tions of innate immune receptors on APCs with their li- implicated in various disease.453 We now know that there
gands encounters are necessary for the upregulation of are multiple reactive oxygen species, many of which com-
costimulatory molecules, without which tolerance would bine with other reactive molecules, such as nitric oxide, to
occur.417 Janeway’s prediction in 1989 was borne out, pro- generate reactive nitrogen species. Nathan and colleagues
viding strong empirical evidence in support of his model demonstrated that mice deficient in inducible nitric oxide
for immune recognition (see Table. 2.1). In 2012, Beutler synthase proved highly susceptible Mycobacterium tubercu-
and Hoffman, together with Steinman, received the Nobel losis, resembling wild-type littermates immunosuppressed
Prize. by high-dose glucocorticoids.454
The field of innate immunity has exploded since the key In addition to its role in microbial killing, NOX2 is
observations of Hoffman, Beutler, Janeway, and Medzhitov. recruited to early phagosomes in DCs and mediates the
Many other receptors and components of the innate im- sustained production of low levels of reactive oxygen spe-
mune system have been discovered. These include compo- cies, causing active and maintained alkalinization of the
nents of the complement system,418 cell surface lectins,419 phagosomal lumen. DCs lacking NOX2 show enhanced
collectins such as lung surfactant proteins420 and mannose- phagosomal acidification and increased antigen degrada-
binding proteins,421 scavenger receptors,422,423 and pentrax- tion, resulting in impaired cross-priming.455
ins.424 Most of these participate in recognition of pathogens,
including fungi,425 bacteria,426 and viruses such as HIV.427
In some instances, the phagocytic cells themselves serve as Cytosolic Components of Innate Immunity
the source of opsonins, such as complement.428 A variety Signal Danger
of innate immune receptors also recognize apoptotic and There are many additional cytosolic proteins important in
necrotic cells.429,430 Secretion of opsonins, such as milk fat innate immunity. Of particular interest is the discovery of
globule-EGF factor 8, can enhance uptake of targets such as a family of pattern recognition receptors, most commonly
apoptotic cells,431 and the absence of this protein has been referred to as nucleotide oligomerization domain–like re-
linked to autoimmunity in mice.432 Similar results were ceptors.456 In 2000 and 2001, a number of genetic studies
found in other mouse strains engineered to be deficient linked defects in NLR genes to inflammatory diseases, in-
in clearance of apoptotic cells.433,434 The conceptual basis cluding Crohn disease,457 Blau syndrome,458 Muckle-Wells
of these experiments was built on earlier work of Savill, syndrome,459 and familial Mediterranean fever.460 In 2002,
Henson, and others who showed that the immunologi- Tschopp and colleagues described a multiprotein complex
cally “silent” disposal of apoptotic debris is an active pro- that they called the “inflammasome,” which included mem-
cess serving to divert self-antigens toward a nonphlogistic bers of the nucleotide oligomerization domain–like recep-
mode of phagocytosis.435–437 In the context of resolution tors family and caspase-1.461 They showed that activation of
of acute infections, a similar function is provided by the the inflammasome leads to generation of IL-1β, a cytokine
production of omega-3 polyunsaturated fatty-acid-derived that was first cloned by Dinarello and colleagues in 1984,462
“anti-inflammatory” lipids (“resolvins”), fi rst identified by representing one of the key “endogenous pyrogens” first de-
Serhan et al. in 2000.438,439 tected in 1953.463,464 In 2006, Tschopp showed that the NLRP3
inflammasome was activated by monosodium urate crystals,
implicating this pathway in gout.465 From a historical per-
Cytosolic Components of Innate Immunity: spective, gout is among the earliest diseases to be described.
Discovery of the Nicotinamide Adenine It was identified by the Egyptians in 2640 bce and was later
Dinucleotide Phosphate-Oxidase recognized by Hippocrates in the fifth century bc, who re-
In 1957, Good and colleagues described an X-linked dis- ferred to it as “the unwalkable disease.” Leeuwenhoek was
ease in which children succumbed to chronic suppurative the first to observe urate crystals from a tophus.466 Recently,
and granulomatous infi ltrations and chronic infections.440 an advisory panel of the U.S. Food and Drug Administration
Neutrophils isolated from these children showed decreased recommended against approval of canakinumab, a human-
bactericidal activity, although they demonstrated normal ized mAb against IL-1, for the treatment of gout; although it
phagocytosis.441 However, they showed decreased hydrogen was effective, it was not deemed safe due to increased risk of
peroxide production and hexose monophosphate shunt serious infections.
activity.442 In 1974, Curnutte et al.443 identified defective In 1994, Matzinger467 proposed a new model of im-
superoxide anion production in children with this syn- mune recognition. She proposed that APCs are activated
drome who also failed to reduce the dye, nitroblue tetrazo- by danger/alarm signals from injured cells (see Table 2.1).
lium.444 This simple test has been widely used to diagnose Although this was a purely theoretical model, since then,
what came to be called “chronic granulomatous disease.” there have been numerous instances in which injured or

damaged components of the host have been recognized to but neither alone, were needed for an antibody response,
trigger inflammation. As Matzinger explains,468 and Mosier475 concluded that antibody production required
“antigen phagocytosis by macrophages and macrophage
Although this may seem to be just one more step lymphocyte interactions,” although he did not know the
down the pathway of slowly increasingly complex cel- nature of the interactions. Further experiments established
lular interaction, this small step drops us off a cliff . . . that exceedingly few adherent cells were needed for anti-
in which the “foreignness” of a pathogen is not the body production, perhaps as few as 1 in 104 adherent cells.476
important feature that triggers a response, and “self- In the mid-1970s, Zanvil Cohn and a postdoctorate in his
ness” is no guarantee of tolerance. laboratory at Rockefeller University, Ralph Steinman, were
Thus, release of uric acid crystals can be viewed as an ex- characterizing a novel adherent population of cells from
ample of a “danger signal.” Similarly, release of intracellular spleen.477–479 The cells had an unusual “tree-like” morphol-
stores of adenosine triphosphate from dying cells as a trigger ogy, prompting the name “dendritic cell.”477 These cells
for signaling cellular injury, predicted in 1988,469 was shown proved difficult to purify, necessitating Steinman and Cohn
to be capable of activating the NLRP3 inflammasome.470 to develop a rather laborious density gradient technique for
Although Matzinger’s model shares features in common with cell purification.480 In 1978, they used this technique to show
Janeway’s, the emphasis on endogenous danger signals rather that DCs could stimulate a primary mixed leukocyte reac-
than direct engagement of pattern recognition receptors by tion—an index of lymphocyte proliferation—and showed
pathogens is novel. The fact that pathogens and “endogenous that DCs are at least 100 times more effective than B cells
danger signals” share common receptors and signaling path- and macrophages for this function.58 This landmark paper
ways is consistent with the idea that either exogenous or en- was the first demonstration of the unique capability of DCs
dogenous triggers of innate immune receptors accomplish to efficiently present antigen. Since then, many DC sub-
the same thing: to facilitate the primary immune response. sets have been described with critical functions in shaping
the immune response. For example, Banchereau and col-
leagues showed that one subtype of DC, a “plasmacytoid”
Cytosolic Components Maintain Safety: DC, is the principal IFNα-secreting cell in systemic lupus
The Role of Autophagy erythematosus with the capacity to induce plasma cell dif-
There are numerous examples of microbes that are ingested ferentiation.481,482 Other DC subsets can respond to cues
by phagocytosis but either remain viable within acidic or- from epithelial cells, which secrete the cytokine TSLP, and
ganelles or escape into the host cytoplasm. Macroautophagy shape T-cell polarization toward a TH2 phenotype.483,484 This
is an evolutionarily conserved process in which cytoplasmic may be a critical pathway in promoting allergic inflamma-
components are sequestered by a double membrane sac, tion in asthma.485 Since the initial discovery of DCs in 1973,
eventually acquiring endosomal and lysosomal character- Steinman and his many colleagues and collaborators have
istics.471 In 2004, Deretic and colleagues showed that IFNγ uncovered novel functions of DCs in triggering and shaping
induced autophagy in macrophages, which contributed to immunity as well as inducing tolerance. Many laboratories
suppression of intracellular survival of mycobacteria.472 In worldwide are using DCs in both capacities, and Steinman
the same year, Nakagawa et al.473 demonstrated that autoph- himself was the recipient of a DC-based vaccine against his
agy was necessary for killing of group A Streptococcus, which own pancreatic cancer. Steinman was awarded the Nobel
had escaped into the host cytoplasm. Since then, many pub- Prize, posthumously, in 2012—tragically, just days after he
lications have confirmed the importance of autophagy in succumbed to cancer. The Nobel was a fitting tribute to a
both innate and acquired immunity.471 consummate cellular immunologist whose scientific curios-
ity and pedigree could be traced to Metchnikoff.
The Renaissance of Cellular Immunology:
Discovery of Dendritic Cells ACKNOWLEDGMENTS
Although we tend to take for granted the concept that I would like to acknowledge my appreciation of my former
APCs are necessary to process antigen, this was not firmly mentor and good friend, Sam Silverstein, my former teachers,
established until 1967. Mosier used a method pioneered by Ben Pernis and Len Chess, and to my colleagues who helped
Mishell and Dutton for measuring in vitro antibody produc- make teaching a great source of joy to me: Ned Braunstein,
tion with sheep erythrocytes as antigen.474 Mosier separated Vinnie Butler, Mitch Cairo, Kathryn Calame, Steve Canfield,
mouse spleen cells into an adherent fraction and a nonad- Raph Clynes, Kathy Nickerson, Paul Rothman, Ale Pernis,
herent fraction. The adherent cells, which were phagocytic, Chris Schindler, and Bob Winchester. I am also indebted to
were deemed “macrophage rich,” and the nonadherent cells Arthur Silverstein, whom I have come to know only from his
were deemed “lymphocyte rich.” Both populations together, scholarship, but in whom I feel a kindred spirit.

Organization and Evolution of the

SECTION
II Immune System
CHAPTER
3 Lymphoid Tissues and Organs
Eitan M. Akirav • Noelia Alonso-Gonzalez • Lucy A. Truman • Nancy H. Ruddle
INTRODUCTION In this chapter, the structure, function, trafficking patterns,

The mammalian immune system defends against invading and developmental signals that regulate the hierarchy of lym-
pathogens, by both the innate and adaptive mechanisms. phoid organs will be described.
Although cells that can respond to pathogens are scattered in
tissues throughout the body, the optimal structures for the re- PRIMARY LYMPHOID ORGANS
sponse to antigens are organized, compartmentalized cellular The primary lymphoid organs are the sites where pre-B- and
aggregates that facilitate antigen concentration and presenta- pre-T-lymphocytes mature into naïve B and T cells in the
tion to a large repertoire of antigen-specific lymphocytes. The absence of foreign antigen. Each T cell or B cell expresses
primary lymphoid organs, the fetal liver, thymus, and bone a unique receptor that can recognize and respond to exog-
marrow, are the sites where diverse populations of naïve lym- enous antigen and, in most cases, discriminate between self-
phocytes mature to disperse throughout the body to await and foreign antigens. Naïve cells leave the primary lymphoid
foreign invaders. This remarkable differentiation process oc- organs having received and responded to developmental cues
curs in a foreign antigen-independent fashion. The secondary that result in the rearrangement of their genetic material to
lymphoid organs, including the lymph nodes, spleen, Peyer’s generate a repertoire capable of recognizing and respond-
patches, and other mucosal-associated lymphoid tissues ing to a wide variety of foreign antigens. In the course of
(MALTs), are discrete sites in which naïve, antigen-specific T- maturation in the primary lymphoid organs, the naïve lym-
and B-lymphocytes encounter invaders to generate an adap- phocytes express various chemokine receptors and adhesion
tive response. The lymph nodes and spleen have been consid- molecules that direct them to secondary lymphoid organs.
ered to be somewhat static structures, while, in fact, they are
responsive to environmental influences and undergo remark-
able changes in the course of antigenic challenge. Precise pro- Fetal Liver
grams control the development of the spleen, lymph nodes, The earliest lymphoid cell precursors derive from self-re-
Peyer’s patches, tonsils and adenoids, and (in the mouse and newing hematopoietic precursors called hematopoietic stem
rat) the nasal-associated lymphoid tissue (NALT). Somewhat cells (HSCs). During ontogeny, these cells occupy several
less anatomically restricted tissues that are even more sensi- niches. In the fetal mouse, the first wave of hematopoiesis
tive to the environment facilitate and include accumulations occurs in the yolk sac and aorta-gonad-mesonephros region
of lymphoid cells are organized, but less discretely defined: at E10.5.3 The placenta also contains HSC activity.4–6 Cells
the bronchus-associated lymphoid tissues (BALTs) and in- leave these tissues and migrate to the fetal liver, and then the
ducible lymphoid follicles (ILFs). Tertiary lymphoid organs, bone marrow and thymus and spleen under the influence
or more accurately, tertiary lymphoid tissues, are accumula- of chemokines and adhesion molecules (Fig. 3.1). The cells
tions of lymphoid cells that arise ectopically in sites that are in the fetal liver respond to CXCL12 (stromal cell derived
not anatomically restricted and are not regulated by devel- factor) and, in contrast to those in the bone marrow, also
opmental programs. Tertiary lymphoid tissues respond to respond to Steel factor7 (Table 3.1). The fetal liver is also the
environmental stimuli and arise during chronic inflamma- source of CD4 + CD3 − lymphoid tissue inducer cells that ex-
tion subsequent to microbial infection, graft rejection, auto- press lymphotoxin (LT) α (also called tumor necrosis factor
immunity, or cancer by the process of lymphoid neogenesis.1,2 [TNF] β) and LTβ. The requirement of inducer cells for the
47

48 | SECTION II ORGANIZATION AND EVOLUTION OF THE IMMUNE SYSTEM
FIG. 3.1. Embryonic and Postnatal Development of T and B Cells in Mouse Primary and Secondary Lymphoid Organs. Embryonic diagram
based on information summarized in Medvinsky et al.6
development of secondary lymphoid organs is described in (CD)5 B cells, predominately B1-B cells, and do not express
the following. Differentiation of HSCs to B cells occurs in terminal deoxynucleotidyl transferase and myosin-like light
the fetal liver, but does not require interleukin (IL)-7. B-cell chain.8,9 It is not known whether these differences are in-
development is described in detail elsewhere in this volume. trinsic to the cells or are due to differences in the cytokine
Of the several different subsets of B cells, those generated environment of fetal liver and bone marrow. The ligands for
by the fetal liver HSCs are somewhat limited and are of the P-selectin, E-selectin, and vascular cell adhesion molecule
B-1 subset.8 They do not give rise to cluster of differentiation (VCAM)-1 are required for the cells to leave the fetal liver
and home to the bone marrow.10,11
Chemokines Implicated in Lymphoid Bone Marrow

TABLE 3.1
Organ Development and Maintenance Functions
Standard Name Common Names Receptor
The bone marrow is the source of self-renewing populations
of stem cells. These cells include hematopoietic precurscor
CXCL12 SDF-1 CXCR4 cells, HSCs, and endothelial progenitor cells, which may
CXCL13 BCA-1, BLC CXCR5 derive from a single precursor.12 The adult bone marrow
CCL17 TARC CCR4 contains stem cells that can differentiate into adipocytes,
CCL19 ELC, MIP-3β CCR7 chondrocytes, osteocytes, and myoblasts.13 Collectively,
CCL20 MIP-3α CCR6 these cells are called hematopoietic/stem progenitor cells. In
CCL21 SLC, 6Ckine CCR7 the adult, cells leave the leave the bone marrow and seed the
CCL25 TECK CCR9 thymus where they undergo differentiation to naïve T cells.
CCL28 MEC CCR10 Additional factors enable the differentiation of immature B
BCA, B-cell chemoattractant; BLC, B lymphocyte chemoattractant; ELC, EBI-1 ligand cells from HSCs.
chemokine; MEC, mucosae as-associated epithelial chemokine; MIP, macrophage In addition to serving as a primary lymphoid organ where
inflammatory protein; SDF-1, stromal cell derived factor; SLC, secondary lymphoid
tissue chemokine; TARC, thymus and activation regulatory chemokine; B-cell differentiation and development occur, the bone mar-
TECK, thymus-expressed chemokine. row is also a home for antibody secreting cells.14 After B cells

CHAPTER 3 LYMPHOID TISSUES AND ORGANS | 49
have interacted with antigen in the secondary lymphoid or- HSC’s exit from the circulation into the marrow.23 CXCL12
gans, such as the lymph nodes, spleen, and Peyer’s patches, is also essential for the earliest stage of B-cell development
they enter the bloodstream and travel to the marrow. Thus, (pre-pro-B cells). Its receptor, CXCR4, is expressed on early
this organ not only serves as a primary lymphoid organ, but B cells and is downregulated in pre-B cells. CXCR4 remains
also as a reservoir for fully differentiated plasma cells. at low levels in immature B cells and mature B cells in sec-
ondary lymphoid organs, but is upregulated after B cells in-
Architecture: Cellular and Functional Niches teract with antigen and differentiate into plasma cells.14 This
The microenvironment of the bone marrow, contained in explains the propensity of antibody-secreting cells to return
the central cavity of bone, is a complex three-dimensional to the bone marrow. Once a pre-pro-B cell has interacted
structure, with cellular niches that influence B cells during with CXCL12, it moves on to a different cell expressing IL-7.
their development and later, as plasma cells, as they return In B-cell development, IL-7 acts later than CXCL12 in a nar-
to the bone marrow. The bone marrow has a rich blood sup- row window between pro-B cells and immature B cells in
ply with a nutrient artery that branches into ascending and the scheme proposed by Nagasawa.15
descending arteries further dividing into cortical capillaries,
periosteal capillaries, and endosteal capillaries, finally merg- Traffic In and Out: Chemokines and Adhesion Molecules
ing into a sinus.15 Previously, the prevailing understanding The extensive vascularization of the bone marrow allows
of B-cell differentiation in the bone marrow was that primi- entrance of hematopoietic precursors and plasma cells and
tive HSCs were located in close contact with the endosteum egress of mature cells. Hematopoietic progenitor recruit-
near osteoblasts (the “endosteal niche”). During the course ment to the bone marrow in the mouse is dependent on
of differentiation into mature B cells, they moved into the the interaction of a variety of chemokines, integrins, and
central region of the bone marrow cavity (the “vascular selectins, and their receptors, counter receptors, and vas-
niche”).16 The former niche was identified as the location of cular cell adhesion molecules. These include α4 integrin
HSC; the latter, as the site of B-cell differentiation.17 This an- (VLA4 or α4β1), VCAM-1,24 P-selectin glycoprotein ligand-1,
atomic concept has been challenged, as it has been reported E-selectin,25 α4β7, and mucosal addressin cell adhesion
that HSCs are found throughout the bone marrow. More molecule (MAdCAM-1).26 A small subpopulation of newly
recently, a reticular niche has been described that includes formed B cells in the bone marrow that expresses L-selectin
CXCL12 abundant reticular cells.18 Growth factors and cy- has been described,27 suggesting a mechanism for entrance
tokines produced by different stromal cells influence cells into lymph nodes (see subsequent discussion).
at different stages in their differentiation. Thus, it is more In addition to the acquisition of Ig expression that occurs
appropriate to consider functional or cellular, rather than in the bone marrow under the control of stromal cells, B cells
anatomical, niches.15,19 Once the HSCs differentiate into im- express various chemokine receptors and adhesion mol-
mature B cells expressing cell surface immunoglobulin (Ig) ecules in the process of differentiation. In a study of human
M, they undergo processes of negative selection and recep- bone marrow B cells, it was determined that pro-B and pre-B
tor editing, leave the bone marrow, and travel through the cells migrate toward CXCL12, but not toward a wide range of
blood stream to the secondary lymphoid organs where they other chemokines, including CCL19 and CCL21, though they
complete their differentiation. do express low levels of CCR7, the receptor for those ligands.
Several cytokines and chemokines influence B-cell dif- However, mature bone marrow B cells do respond to CCL 19
ferentiation in the bone marrow. Flt-3 ligand (also called and CCL 21,28 CXCL13 (indicating a functional CXCR5), and
Flk-2L) signals B-cell differentiation and growth and syner- CCL20 (MIP3α). On the other hand, CCR6 expression and
gizes with several other hematopoietic growth factors.20 Its responsiveness to CCL20 are only seen in mouse B cells after
receptor, Flt-3 (also known as Flk-2), expressed by primitive they have migrated into the periphery during the process of
HSCs, is a member of the class II tyrosine kinase family. In maturation and are in the circulating B cell pool.29
contrast to fetal liver, HSCs from adult bone marrow do not
respond to Steel factor. Chemokines contribute to B-cell dif-
ferentiation in the bone marrow and define the functional Thymus
niches. For excellent reviews of this topic, see Nagasawa,15 Functions
Heissig et al.,18 and Mazo et al.21 Many of these factors af- The thymus is defined as a primary lymphoid organ due to
fect other lymphoid cells, such as dendritic cells (DCs) and its inimitable role in T-cell development. Indeed, a mutant
T cells. Several of these factors, whose functions have been known as the “nude” (nu/nu) mouse, which lacks a normal
identified in gene deletion studies in mice, in morphologic thymus, is completely devoid of mature T cells.30 Although
analysis, and in cell culture studies, play roles in multiple some studies suggest that T cells can develop extrathymi-
aspects of lymphoid organ development; their activities, cally in organs such as the “cervical thymus”31 and the gut
though important in the bone marrow, are not limited to epithelium,32 the thymus remains the main site for T-cell
that organ. CXCL12, also known as stromal cell derived fac- maturation, education, and selection. T-cell precursors rep-
tor, is a chemokine that is crucial for recruitment of HSCs to resent more than 95% of total cells in the thymus and give
the bone marrow. It is widely expressed by osteoblasts, retic- rise to mature T cells. These cells are crucial components
ular cells,19 and endothelial cells.22 In fact, the interaction of of the adaptive immune system in that they are highly spe-
HSCs expressing CXCR4, the receptor for CXCL12, with that cific in their ability to recognize a nearly infi nite number of
chemokine on the endothelial surface is the first step in the antigens owing to their diverse repertoire.30,33

A B
FIG. 3.2. A: Thymic structure. Low power magnification of a hematoxylin-stained frozen section of a mouse thymus. The cortex is well popu-
lated with lymphocytes; the medulla is less tightly packed and thus stains less intensely. B: Diagram of mouse thymic cellular populations.
Precursor T cells enter through blood vessels at the corticomedullary junction. They progress to the medulla where they undergo differen-
tiation from double negative to double positive (DP) cells expressing T-cell receptors. Thymic stromal cells provide growth factors. DP cells
undergo positive selection, under the influence of cortical thymic epithelial cells. Single positive (SP) cluster of differentiation (CD)4 or CD8
cells migrate into the medulla where they undergo negative selection, mainly through autoimmune regulator expressing medullary thymic
epithelial cells and dendritic cells. SP cells, having undergone differentiation, exit through blood vessels (not shown) to the periphery.
The T-cell repertoire is shaped during development in the multilobed. Each lobe can be divided into three distinct re-
thymus by the processes of positive and negative selection. gions: capsule, cortex, and medulla. The latter two regions
Negative and positive selection are so stringent that nearly harbor thymocytes at various maturation stages. Although
95% of all T-cell precursors are deleted in the thymus.34 maturing T cells constitute the majority of cells in the thy-
Negative selection ensures that T cells, which are capable of mus, other cell types such as macrophages, DCs, B cells, and
recognizing the organism’s self-antigens presented on major epithelial cells are also present.36 Histologic analysis of the
histocompatibility complex (MHC) class I or II molecules thymus reveals a clear distinction between the thymic cor-
with high affinity, are eliminated prior to their export to the tex and medulla, which are separated by a corticomedullary
periphery. In contrast, positive selection, which requires a border. The thymic cortex appears darker and more densely
higher degree of T-cell receptor (TCR) avidity than negative populated with T-cell precursors, whereas, the medulla ap-
selection,35 allows for T cells recognizing self-antigen with pears considerably lighter and contains smaller numbers of
low to medium affinity to leave the thymus and eventually T cells relative to other cell types (Fig. 3.2A). Blood ves-
protect the organism against invading pathogens. sels and small blood capillaries are found throughout the
While negative selection is highly efficient in eliminating the thymus. The fact that T-cell progenitors are found in the
majority of self-reactive T cells, some of these cells do escape more highly vascularized corticomedullary border suggests
the thymus and exit to the periphery. These autoreactive cells that blood vessels in this region facilitate the entry of T-cell
impose a threat to various organs as their activation may result progenitors into the thymic parenchyma.37 In the thymic
in the development of autoimmune diseases. The immune sys- medulla, the close association between medullary thymic
tem has evolved several ways to prevent the activation of auto- epithelial cells and thymic blood vessels38 suggests that
reactive T cells in the periphery. One mechanism that protects these vessels may act as organizers of the medullary thy-
against autoimmunity is the generation of a T-cell population mic compartment. Lymphatic vessel distribution coincides
capable of suppressing activation of self-reactive T cells. These with that of blood vessels and capillaries.39 The majority of
thymus-derived protective T cells, known as regulatory cells lymphatic vessels are located in the thymic medulla, though
(Treg), recognize self-antigens with relatively high affinity. some can also be found in the cortex. The role of lymphatic
vessels in thymic function is unclear, although it has been
Architecture proposed that these vessels deliver extrathymic antigens
The thymus in the mouse consists of two symmetric lobes into the thymus or export mature T cells from the thymus
located above the heart, while in humans the thymus is into the circulation.

Cellular Composition and Functions corpuscles. Originally described by Arthur Hill Hassall in
T-cell precursors at various stages of differentiation repre- 1849, this structure consists of concentric stratified keratin-
sent the majority of cells in the thymus. Antigen-presenting izing epithelium. Hassall corpuscles are implicated in sev-
cells (APCs) of either hematopoietic or stromal origin medi- eral processes of the thymus,52 including the expression of
ate the education and selection of T cells in the thymus. The tissue-restricted antigens, such as insulin (a key antigen in
different stages in T-cell selection and maturation that take type 1 diabetes),53 Igs, and fi laggrin (key antigens in rheu-
place in distinct regions of the thymus (Fig. 3.2B) are dis- matoid arthritis),54,55 and serving as a prominent site for
cussed at length elsewhere in this volume. Briefly, lymphoid T-cell apoptosis.56
progenitors enter the thymus at the corticomedullary bor- DCs are professional APCs that are found in the thymus.
der.37 Following their entry, these cells, identified as double Thymic DCs can be divided into two distinct populations.
negative (DN) T cells due to a lack of expression of the cell The first population originates from a thymocyte precur-
surface molecules CD4 and CD8, undergo four maturation sor,57,58 whereas the second population is derived from par-
steps termed DN1 to DN4, which are distinguished by the tially mature peripheral DCs that continuously enter the
expression of two additional cell surface molecules: CD25 thymus from the circulation.59 DCs do not appear to be
and CD44.40,41 DN3 cells migrate to the subcapsular zone involved in positive selection, but do contribute to negative
while rearranging their TCRβ chain and expressing it in selection.60–63 They have also been implicated in the selec-
combination with a surrogate α chain. Those cells that have tion of Tregs in humans,64 and recent reports in the mouse
successfully rearranged the genes for α and β chains of the suggest that peripheral DCs can migrate to the thymus and
TCR become double positive (DP) cells, and express both serve as efficient inducers of Tregs.65 Activation of thymic
CD4 and CD8 surface markers. In the cortex, DP cells un- DCs can be mediated in part by the IL-7–like cytokine, thy-
dergo negative and positive selection.42,43 Positively selected mic stromal lymphopoietin. In the human thymus, Hassall
DP cells further differentiate into single positive (SP) cells corpuscles produce thymic stromal lymphopoietin, and thy-
expressing either CD4 or CD8. Following their differentia- mic stromal lymphopoietin–activated DCs can alter the fate
tion, SP cells relocate to the medulla where they mature and of self-reactive T cells from deletion to positive selection of
undergo further rounds of deletion. SP cells that do survive Tregs.66 More specifically, thymic stromal lymphopoietin ex-
are then exported out of the thymus.44 DN precursors can pression in a subset of DCs, plasmacytoid DCs, can induce
give rise to an additional T-cell population expressing the the generation of particularly potent Tregs from CD4 + CD8 −
γδTCR. These T cells are distinct from TCRαβ T cells in CD25 − thymocytes.67 These findings highlight the heteroge-
their tissue distribution and recognition of antigens. TCRαβ neity of DCs and emphasize their ability to fulfi ll different
DP cells control the development of TCRγδ cells via the pro- roles during T-cell development in the thymus.
duction of LTβ.45 Treg development in the human thymus oc- Macrophages and B cells are additional hematopoietic-
curs at the DP stage allowing for the production of CD4 and derived professional APCs in the thymus. In contrast to
CD8 positive Tregs.46 DCs, thymic macrophages are located throughout the thy-
The thymic parenchyma consists of a complex three- mus and do not play a significant role in T-cell selection.68 B
dimensional structure supported by thymic epithelial cells cells are detected in human and mouse thymus at relatively
(TECs). TECs in the thymic cortex and thymic medulla are low numbers69,70 and are characterized by the expression of
phenotypically and functionally distinct and support differ- the cell surface molecule CD5. They are capable of produc-
ent stages of T-cell maturation. Several cell surface marking antibodies of several different isotypes.71 It has been sug-
ers are used to distinguish medullary TECs (mTECs) from gested that thymic B cells induce negative selection in devel-
cortical TECs (cTECs) in the mouse. Among these mark- oping thymocytes,69 although B cell–deficient mice show a
ers are the cytokeratins K5 and K8, the adhesion molecule limited T-cell repertoire when compared with normal mice,
Ep-CAM, and the glycoprotein Ly-51. K8 and Ly-51 are ex- also suggesting a role in positive selection.72 Recent data
pressed by cTECs, while K5 and Ep-CAM are expressed by consistent with a role for B cells in thymic negative selection
mTECs.47,48 TECs are unique in that they express MHC II have emerged. Mice whose B cells were experimentally ma-
constitutively, similarly to professional APCs. The role of nipulated to express a myelin oligodendrocyte glycoprotein
TECs in T-cell selection was recently elucidated by their ex- peptide show deletion of myelin oligodendrocyte glycopro-
pression of the transcription factor, autoimmune regulator tein–specific T effector cells.73 One mechanism for the effect
(AIRE). This transcription factor plays an important role of B cells on T-cell selection is suggested by the observation
in T-cell selection and prevention of autoimmunity, as il- that B cells play a role in the control of tissue-restricted anti-
lustrated by the fact that humans with a mutated form of the gens. Studies done in mice deficient in B cells showed fewer
AIRE gene exhibit polyendocrine autoimmunity due to in- thymic epithelial cells with reduced expression of insulin
adequate T-cell selection.49,50 It has become clear that AIRE and myelin oligodendrocyte glycoprotein, native antigens of
controls the expression of certain tissue-restricted antigens the pancreas and brain, respectively.74
in the thymus, such as insulin, a protein that is unique to
the β cells of the islets of Langerhans in the pancreas.48,51 Traffic In and Out: Adhesion Molecules and Chemokines
The expression of tissue-restricted antigens in the thymus Adhesion molecules mediate the extravasation of leukocytes
facilitates the negative selection of maturing T cells, which from blood and lymphatic vessels into the thymus. These
would otherwise be allowed to leave the thymus. The thymic molecules also play an important role in facilitating lym-
medulla includes distinct structures also known as Hassall phocyte homing into various regions of the thymus. Indeed,

thymic blood vessels express the adhesion molecules inter- which repel SP cells out of thymus,84 and CCL19, which pro-
cellular adhesion molecule (ICAM)-1, VCAM-1, CD34, pe- motes T-cell emigration from the thymus of newborn mice.85
ripheral node addressin (PNAd) and vascular adhesion pro- The chemoattractant, sphingosine 1-phosphate (S1P), is an
tein-1.75 The expression of high levels of ICAM-1, VCAM-1, additional mediator of T-cell emigration. SP T cell express a
and vascular adhesion protein-1 on venules near the cortico- S1P receptor (S1P1) and are attracted to the high levels of S1P
medullary border suggests that these molecules may play a present in the serum promoting their egress.86,87
role in the recruitment of thymocyte progenitors. More spe- While the role of different thymic compartments and
cifically, vascular adhesion protein-1, which is restricted to chemokines in the maturation of naïve SP T cells has been
the venules surrounding the sites of progenitor homing, can extensively studied, the thymic regions and chemokines that
mediate the extravasation of leukocytes. The regional distri- control the selection of regulatory T cells remain largely un-
bution of these adhesion molecules further illustrates the im- known. It may be that chemokines produced by both cTECs
portance of a distinct anatomical separation between cortex and thymic DCs play a role in regulatory T-cell selection,
and medulla and represents their individual functions. albeit during different stages of maturation.
Chemokine-mediated T-cell migration and traffic to and
within the thymus is crucial for normal T-cell selection. The Development
compartmentalization of the thymus is orchestrated by a The initial development of the thymus at midgestation in the
milieu of chemokines. The thymic cortex is involved in the mouse is independent of vascularization or bone marrow–
maturation of DN cells to DP cells and by positively select- derived cells. In the mouse, the thymus rudiment is first evi-
ing DP T cells capable of recognizing MHC:peptide com- dent on day 11 of gestation as it evolves from the endoderm
plexes and eliminating the cells that are not. The thymic me- of the third pharyngeal pouch.88 This gives rise to the thy-
dulla acts as a site of negative selection of SP T cells based on mic lobes as well as to the parathyroid gland. On day 12.5 of
TCR recognition avidity to self-antigens and possibly posi- gestation, a separation of the primordium is observed and
tive selection of Treg. Various chemokines that are produced by day 13.5 a distinct thymus is apparent. Evidence has sug-
by the thymic cortex and medulla allow T cells expressing gested that not only the pharyngeal pouch endoderm but
different chemokine receptors to home to specific regions also the ectoderm may also be involved in thymic develop-
of the thymus. This differential expression of chemokines ment.89,90 More specifically, it was suggested that the pha-
is complemented by the fact that T cells at different matu- ryngeal pouch ectoderm contributes to the development of
ration states express different chemokine receptors. During cTECs whereas the endoderm contributes solely to the de-
development, the entry of lymphoid progenitors into the velopment of mTECs. This “dual origin” model was mainly
thymus is highly dependent on CCL21 and CCL25, which supported by histological data as well as data collected from
bind the chemokine receptors CCR7 and CCR9, respectively. thymi of nu/nu mice.91–93 More recently, it was shown that
Mice lacking CCR7 or CCL2176,77 show a transient delay in both thymic cortex and medulla are derived solely from the
thymus colonization by lymphocytes (day 14.5), and this pharyngeal pouch endoderm94 ; although the endoderm and
delay is further extended (day 17.5) in mice lacking CCR9.78 ectoderm are found in close proximity between gestational
Lymphoid progenitors enter the thymus at the corticome- day 10.5 and 11, only the endoderm actively contributes to
dullary border and commence their migration outwards to- thymic development.95 These findings support the so-called
ward the subcapsular region of the cortex as DN3 cells. The single origin model of thymus development.
expression of the chemokine receptors CXCR4 and CCR7 The contribution of lymphocytes to the normal develop-
by DN cells is important in directing cell migration.79,80 In ment of thymic cortex and medulla is well recognized. In
the subcapsular region, DN thymocytes that have success- models of T-cell deficiency, cTEC and mTEC development
fully rearranged their TCRαβ chains progress to the DP cell is halted at different stages depending on the stage of T-cell
stage. Positively selected DP cells move inwards toward the arrest. If T-cell development is arrested at the DN stage, as
thymic medulla for further differentiation into SP cells. The in the recombination activating gene–deficient mouse, the
ligands for CCR7 are crucial in mediating the migration of thymic medulla is greatly reduced while the thymic cortex
positively selected DP cells into the medulla as illustrated by remains unaffected.96,97 A more severe phenotype is ob-
the fact that a deficiency in CCR7 or its ligands, CCL19 or served in transgenic mice that overexpress the human CD3
CCL21, prevents DP cell relocation from the cortex to the signaling molecule. In these mice, T-cell arrest occurs earlier
medulla resulting in abnormal central tolerance.81,82 This than in recombination activating gene knockout mice, lead-
abnormal tolerance is associated with a reduction in thymic ing to a loss of both cortex and medulla and to a shift from
B-cell numbers and reduced expression of tissue-restricted a three-dimensional to a two-dimensional structure of the
antigens.74 Interestingly, antigen itself can control T-cell thymic epithelia.97
migration speed from the cortex to the medulla. In the pres- Recently, LTα and LTβ have been identified as master
ence of a negative selection ligand, T cells slow down consid- regulators of mTEC development and expression of tis-
erably and are limited to a confined zone 30 μ in diameter sue-restricted antigens. LTα and LTβ are members of the
allowing for a more prolonged selection and induction of TNF superfamily and mediate the processes of secondary
developmental arrest.83 lymphoid organ development and inflammation.1,98 In the
The export of positively selected SP T cells out of the thy- absence of LT, tissue-restricted antigen and AIRE expres-
mus is also dependent on chemokines. Chemokines involved sion are reduced and certain mTEC subpopulations fail to
in T-cell emigration are CXCL12 and its receptor CXCR4, develop.99,100 T cells, owing to their vast numbers, appear

to be the main source for LT production in the thymus45 ; nodes. Though most lymph nodes are classified as periph-
however, additional resident cells also serve as a source of eral lymph nodes, a few (cervical, mesenteric, and sacral),
LT. Recently, resident thymic B cells were identified as the termed mucosal nodes, express a slightly different comple-
highest source of LTα and LTβ on a per cell basis,74 revealing ment of endothelial adhesion molecules, cooperate with the
yet another aspect of thymic B cells in thymus development mucosal system, and are regulated somewhat differently in
and function. development (see following discussion). Although all lymph
The development of the thymus in humans closely fol- nodes are vascularized, and thus can receive antigens from
lows the model of thymic development in the mouse and the bloodstream, they are also served by a rich lymphatic
bird. Similar to the mouse, human thymic colonization by vessel system and are thus particularly effective in mounting
hematopoietic stem cells occurs relatively early, at week 8.2 responses to antigens that are present in tissues. These an-
of gestation. During this stage, the thymic medulla and cor- tigens may be derived from foreign invaders that are trans-
tex are organized, suggesting that thymocytes are required ported by APCs or can be derived from self-antigens. Thus,
for normal thymic development. Between gestation week 9.5 lymph nodes extend the role of the primary lymphoid or-
and 10, the first signs of thymocyte negative selection are gans and discriminate between dangerous foreign antigens
evident, and by gestation week 10 to 12.75 the gradual onset and benign self-antigens. This capacity relies on the APCs
of positive selection is detected.101 and their state of activation in the lymph node, and the rec-
ognition capacity of the naïve T and B cells.
Lymph nodes can also function as niches for generating
SECONDARY LYMPHOID ORGANS peripheral tolerance, an additional mechanism to minimize
Naïve cells express their receptors for specific antigen, leave the effects of those self-reactive T-cells that escape central
primary lymphoid organs, circulate through the blood- tolerance in the thymus.102 DCs constitutively sample self-
stream, migrate into the tissues, and lodge in secondary antigens and migrate to draining lymph nodes even in the
lymphoid organs. The frequency of naïve cells specific for an steady state.103–105 Because most self-antigen–bearing DCs in
individual antigen is quite low (estimates range from 1 in 105 lymph nodes are immature106 and have low levels of costim-
to 1 in 106). Thus, the chance that an individual T or B cell ulatory molecules, they are not effective at activating naïve
will encounter its specific antigen in the circulation is rather cells. They regulate self-reactive T cells by inducing anergy,
low. Secondary lymphoid organs are strategically located in clonal deletion, and/or expanding Tregs.105–109 Several groups
anatomically distinct sites where foreign antigen and APCs have recently described populations of cells in secondary
efficiently concentrate and activate rare antigen-specific lym- lymphoid organs that can present tissue-restricted antigens
phocytes, thus leading to the initiation of adaptive immune to CD8 cells.110–114 These include two different populations
responses and generation of long-lived protective immunity. of AIRE-expressing stromal cells, DCs, and AIRE-negative
These organs include highly organized, compartmentalized, lymphatic endothelial cells. These cells may play a role in
and mostly encapsulated tissues such as lymph nodes, spleen, self-tolerance, or perhaps they can prime for autoimmunity
appendix, tonsils, murine NALTs, and Peyer’s patches. Naïve in inflammatory conditions.
cells are also primed in less discrete tissues throughout the
body, including the BALTs, cryptopatches, and ILF. Structure and Organization
A collagen capsule surrounds the highly compartmentalized
lymph node (Fig. 3.3). The cortical region includes discrete
Lymph Nodes clusters called primary follicles consisting of densely packed
Lymph nodes are bean-shaped structures dispersed along naïve B cells and follicular DCs (FDCs). After B cells en-
lymphatic vessels. The lymphatic vessel system plays im- counter their cognate antigen, they are activated. They then
portant roles in tissue fluid balance, fat transport, and the proliferate secondary follicles and germinal centers develop.
immune response. In contradistinction to blood vessels, T cells and DCs distribute in the paracortex. Macrophages
which form a closed recirculating system, lymphatic vessels reside in the subcapsular zone and in the medullary area,
comprise a blind-end, unidirectional transportation system. and those in the subcapsular zone appear to be particularly
The absorbing lymphatic vessels, or lymphatic capillaries, adept at presenting antigen-antibody complexes to B cells.115
remove interstitial fluid and macromolecules from extra- Follicular DCs, a population of mesenchymal origin, sup-
cellular spaces and transport the collected lymph through port B-cell follicles or germinal centers under stimulation.1
the primary collector. The collected lymph and its cellular Plasma cells are also concentrated in the medulla as they
contents are transported into the thoracic duct and returned prepare to leave the lymph node and circulate to the bone
back to blood circulation. In humans, lymph collected from marrow. A network composed of reticular fibers, fibrous
the entire lower body region, the left head, and left arm re- extracellular matrix bundles, and another mesenchymal
gion accumulates in the thoracic duct and returns to blood cell population, the fibroblastic reticular cells, supports the
circulation via the left subclavian vein; lymph collected entire lymph node.116 Compartmentalization of cells in the
from right head and right arm region returns to blood via lymph node is orchestrated by lymphoid chemokines CCL19,
the right subclavian vein. CCL21, and CXCL13. Stromal cells in the paracortical region
Lymph nodes, usually embedded in fat, are located at produce CCL19 and the protein is transported to the surface
vascular junctions, and are served by lymphatic vessels of high endothelial venules (HEVs).117 CCL21 is encoded by
that bring in antigen, and connect them to other lymph several genes118 ; CCL21-leu is expressed by lymphatic vessels

FIG. 3.3. A: Lymph node structure and functional regions. The lymph node is divided into an outer cortex and inner medulla surrounded
by a capsule and lymphatic sinus. The cortex includes B cells and follicular dendritic cells. The paracortical region includes T cells and
dendritic cells. Macrophages are found in the subcapsular sinus and medullary cord. Lymphocytes enter into the lymph node through an
artery at the hilus region, and into the parenchyma through high endothelial venules (HEVs) expressing peripheral node addressin and
CCL19 and CCL21. Stromal cells also produce these chemokines. T cells and dendritic cells (DCs) are directed to the paracortical region
by CCL19 and CCL21. B cells are directed to the cortex by CXCL13. After interaction with antigen and T cells, germinal centers develop.
Antigen and DCs drain into the lymph node from the tissues through afferent lymphatic vessels. Antigen continues to percolate through
the node via a conduit system. Activated cells leave through efferent lymphatic vessels. Diagram of the blood network of a rat lymph node
(right) is adapted from Anderson and Anderson.125 Note the arteriovenous communications, the venous sphincters, and cells leaving the
HEVs into the parenchyma. B: Lymph node compartmentalization. Immunofluorescent staining of B cells (anti-B220, green) in follicles in
the cortex and T cells (anti-cluster of differentiation 3, red) in the paracortical area. The medulla is unstained.
outside the lymph node119 ; CCL21-ser is made by stromal node or leaving it all together.121,122 S1P1 facilitates lymphocyte
cells and HEVs in the lymph node. CCL19 and CCL21 recruit egress from lymph nodes as they move toward the ligand S1P
CCR7-expressing cells across the HEVs to the paracortical in the lymph,123,124 as discussed in more detail subsequently.
region. CXCL13, produced by stromal cells in the B-cell fol-
licles, attracts CXCR5 expressing B cells.120 After naïve T and Lymph Node Vasculature: Blood Vessels and
B cells encounter antigen, they undergo extensive changes in Lymphatic Vessels
expression of chemokine receptors and adhesion molecules Soluble antigen and APCs enter into lymph nodes via affer-
that result in their movement to different areas of the lymph ent lymphatic vessels. After surveying antigens in the lymph

nodes, lymphocytes leave those organs via efferent lym- adhesion to mesenteric lymph node HEVs.34,136,138 The selec-
phatic vessels that can connect to the next lymph node in the tive adhesion of naïve lymphocytes to HEVs is at least partially
chain and finally return to blood circulation.125–127 In this controlled by the differentially expressed adhesion mol-
manner, HEVs and lymphatic vessels maintain lymphocyte ecules in different lymphoid organs. PNAd rapidly replaces
homeostasis during the steady state. MAdCAM-1 after birth in mouse peripheral lymph nodes,139
Blood endothelial cells play a crucial role in lymphocyte but is expressed in mucosal lymph nodes together with
trafficking in the lymph node. One or two arteries enter MAdCAM-1, the ligand for the integrin α4β7.
the lymph node at the hilus. These arteries branch and pass Lymphatic vessels also play critical roles in the immune
through the medulla area, enter the cortex, and sometimes response. The collected lymph and cell contents enter the
continue in the subcapsular area. Beneath the subcapsular lymph node via several afferent lymphatic vessels and fi lter
sinus, the branching capillaries form loops and some of them through the node where they again are concentrated in the
become arteriovenous communications. Arteriovenous medullary sinus. Lymphatic vessels are concentrated in the
communications become HEVs in the cortex area and oc- subcapsular sinus and medullary area.125 Factors from affer-
casionally extend from the subcapsular sinus to the medulla. ent lymph can be either transported deep into the lymph
HEVs constitute a specialized postcapillary network in the node cortex or move via the subcapsular sinus and leave the
lymph node, playing a critical role in lymphocyte recircula- lymph node through efferent lymphatic vessels.116 In this
tion. Each main HEV trunk receives three to five branches manner, soluble antigen and APCs from peripheral tissues
lined with high endothelial cells and two or three branches are efficiently concentrated in the draining lymph node and
lined with flat endothelial cells. The luminal diameters of initiate an adaptive immune response. Lymphocytes can
HEVs progressively increase from cortex to medulla. Finally, also enter lymph nodes through afferent lymphatic vessels
HEVs merge into segmental veins in the medulla area and and leave via efferent vessels and move to the next lymph
join larger veins in the hilus125 (see Fig. 3.3A). Intravital mi- node in the chain. This is accomplished in part through a
croscopy has revealed that the entire venular tree consists gradient of S1P, which is at a high concentration in the lymph
of five branching orders with the higher orders in the para- and low concentration in the lymph node. Lymphocytes
cortex and the lower orders located in the medulla and hilus in the lymph node downregulate this receptor (S1P1) and
areas. Only the higher order venules, located in the T-cell then upregulate it as they prepare to leave and migrate to-
area, are specialized into HEVs and are recognized by the ward the higher concentration in the efferent lymph.123,124
monoclonal antibody MECA-79.126,127 Recirculating naive Human and murine lymphatic vessels express several char-
lymphocytes leave the bloodstream via HEVs, specialized acteristic markers: lymphatic vessel endothelial hyaluronan
vessels with a high cuboidal endothelium, and migrate receptor-1 (LYVE-1), Prox-1, podoplanin, CCL21, the vas-
into the lymph node parenchyma. PNAd, defined by the cular endothelial growth factor receptor-3 (VEGFR-3), and
MECA-79 antigen, is an L-selectin ligand, and is a charac- neuropilin-2.140
teristic HEV adhesion molecule. PNAd is composed of a
variety of core glycoproteins, including GlyCAM-1, CD34, Conduit System
Sgp200, and podocalyxin; these proteins must be sialylated, A conduit system in the lymph nodes physically connects the
sulfated, and fucosylated to become functional L-selectin lymphatic sinus with the walls of blood vessels and enables
ligands. The several enzymes that mediate these posttrans- the incoming factor(s) from lymph to move rapidly deep into
lational modification events include FucT-IV, FucT-VII, and the paracortical area.116,125 The conduit system consists of four
GlcNAc6ST2 (also called HEC-6ST, LSST, GST-3, HEC- layers: 1) a core of type I and type III collagen bundles; 2) a
GlcNAc6ST, gene name Chst4),128–132 which with the excep- microfibrillar zone composed largely of fibrillins; 3) a base-
tion of a population of cells in the intestine133,134 is uniquely ment membrane abundant with laminins 8 and 10, perlecan,
expressed in high endothelial cells. Together with another and type IV collagen that provides a supportive structure;
sulfotransferase, GlcNAC6ST1, that is expressed at sites in and 4) fibroblastic reticular cells that embrace the entire
addition to HEVs, it sulfates glycoproteins in the Golgi appa- conduit system.141,142 This conduit system enables incoming
ratus135 to generate the MECA-79 epitope. PNAd, expressed lymph to penetrate deep into the T-cell area. A special sub-
on the endothelial surface, slows down (tethers) naïve set of immature DCs, called conduit-associated DCs, can take
L-selectinhi lymphocytes in their progress through the blood up and process antigens moving along the conduit.142 In this
vessels. After this initial interaction, CCL19 and CCL21 on manner, the conduit system probably provides a physical sup-
the HEVs are instrumental in activating the lymphocyte port for rapid initiation of adaptive immune responses after
integrin lymphocyte function–associated antigen (LFA)-1. immunization.
This results in tight binding of LFA-1 to ICAM-1 on the HEV,
facilitating diapedesis of lymphocytes that migrate between The Intimate Relationship Between Lymph and High
or through the endothelial cells toward the chemokines lo- Endothelial Venules
cated in the paracortical region (T cells, DCs) or cortex (B Incoming lymph is necessary for the maintenance of HEV
cells). Data suggest that the HEV-lymphocyte interaction is phenotype and function. After afferent lymphatic vessels
not random in different lymphoid tissues in the mouse.136,137 are severed, dramatic changes occur in HEVs. These include
T-lymphocytes adhere preferentially to peripheral lymph flattening of the endothelium, a decrease in the uptake of
35
node HEVs, B cells prefer to adhere to HEVs in Peyer’s S-sulphate143,144 (a functional marker of GlcNAC6ST-2),
patches, and T and B cells exhibit an intermediate pattern of a reduction of lymphocyte adherence to the vessels,145–147

and decreased expression of PNAd and the HEV genes, separation of lymphatic endothelial cells from venous endo-
Glycam-1 and Fuc-TVII.146–148 An increase in MAdCAM-1 thelium requires a Syk/SLP-76 signal155 that is provided by
expression146 suggests that these events are not simply due platelets.156,157 Thus, during early lymphangiogenesis, some
to a general downregulation of blood vessel gene expression endothelial cells express both blood vessel and lymphatic
and indicates that continual accumulation of afferent lymph vessel markers, indicating the close association of these two
factor(s) in lymph nodes is necessary for HEV maintenance. vascular systems. Mesenchymal lymphangioblasts may also
The nature of such lymph factor(s) is unknown. contribute to early lymphangiogenesis.158 The lymphatic ve-
Topographic relations between HEVs and the lymphatic nous junction remains in adults in only limited regions but
sinus have been described in the rat mesenteric lymph nodes. plays an essential role in connecting the function of the two
Some HEVs are located in the medulla area and positioned vascular systems.
closely in relation to the lymphatic sinus.125 Occasionally, Studies in the mouse have taken advantage of transgenic,
HEVs are separated from an adjacent lymphatic sinus by knockout, and imaging studies to provide a mechanistic
only a thin layer of collagen bundles. The closely apposed understanding of the process of lymph node development.
HEV and lymphatic sinus in the medulla provide the physi- Although lymph sacs per se do not appear to be essential
cal support for the intimate relationship between lymph and for the initiation of lymph nodes, a lymphatic vessel system
HEV. However, most HEVs are located in the paracortical is necessary for the later development and cellular popula-
area and are separated from the lymphatic sinus by lympho- tion of lymph nodes.159 Furthermore, an interaction between
cytes. The conduit system physically connects the subcapsu- mesenchymal cells and primitive endothelial cells is appar-
lar sinus and HEVs and allows incoming lymph factor(s) to ent in the early lymph node anlage.160
migrate rapidly to the wall of HEVs.125,149,150 Low-molecular-
weight fluorescent tracers (below 70 kD) move rapidly via Cytokines, Chemokines, and Transcription Factors in
the conduit system and lead directly to the wall of HEVs. Lymphoid Organogenesis
In this manner, low-molecular-weight tracers can migrate The TNF-/LT-receptor family members play key roles in
with lymph within minutes to HEVs and enter the HEV secondary lymphoid organ development. LTα3, signaling
lumen.116 Lymph-borne chemokines likely adopt this route through TNFRI and TNFRII, and membrane bound LTα1β2,
to regulate HEV function. IL-8 administration via afferent signaling through the LTβ receptor (LTβR) have been impli-
lymph increases lymphocyte HEV transmigration within cated. Mice deficient in LTα lack all lymph nodes and Peyer’s
minutes.125,151 In addition, lymph-borne chemokines, such as patches, and exhibit a disorganized spleen and severely dis-
MIP-1α , also enter the conduit system and move rapidly to organized NALT (see subsequent discussion).161–163 Mice de-
HEVs.116 These data suggest that lymph factor(s) can quickly ficient in LTβ lack peripheral lymph nodes but retain mesen-
access and regulate HEVs. teric, sacral, and cervical lymph nodes.164–166 Ltb r- / - mice
have a phenotype similar to that of Lta- / - mice.167 LTβR is
Development also recognized by LTβ -related ligand, LIGHT, which also
Despite their distinctions in the adult with regard to binds to the herpes virus entry mediator. LIGHT-LTβR sig-
morphology and expressed genes, a close association of blood naling does not appear to play an essential role during lym-
vessels and lymphatic vessels is seen during embryogenesis. phoid organogenesis, as no significant defect is observed in
The generation of embryonic lymphatic vessels from preex- Light- / - mice168 ; however, mice doubly deficient in LIGHT
isting veins in pig embryos was first described in the early and LTβ have fewer mesenteric lymph nodes than mice de-
1900s by Sabin and has recently been molecularly defined.140 ficient in LTβ alone, indicating a cooperative effect of the
A variety of transcription factors have been identified that two LTβR ligands. Treatment of pregnant females with an
contribute to the lymphatic specification and maintenance inhibitory soluble protein of LTβR (LTβR and human IgG Fc
of the lymphatic vessels phenotype.152 As early as mouse E9, fusion protein, LTβR-Ig) inhibits most lymph nodes in the
expression of Sox18 and CoupTFII is apparent in the cardi- developing embryos, depending on the time of administra-
nal vein. These induce Prox1 in the dorsolateral side of the tion. Mesenteric lymph nodes are not inhibited by this treat-
vein resulting in polarization and lymphatic-biased endo- ment. These studies indicate that individual lymph nodes
thelial cells. Prox1 generates a feedback signal for the further differ in the nature and time of cytokine signaling during
maintenance of budding and migration of endothelial cells. development.169 Several additional cytokine and chemokine-
Targets of Prox1 include Prox1 itself, podoplanin, VEGFR3, receptor pairs are crucial for lymphoid organogenesis. Mice
integrin-α , and Nrp-2. Prox1 expression in blood endothe- deficient in IL-7 or IL-7R,170,171 or TRANCE or TRANCER,
lial cells also represses expression of markers of those cells. also called RANKL and RANK,172,173 exhibit defects in lymph
NfatC1, Foxc2, and Tbx1 are additional transcription fac- node development. CXCR5- or CXCL13-deficient mice174,175
tors that are expressed later in lymphatic vessel development lack some lymph nodes and almost all Peyer’s patches.98 The
and contribute to patterning, pericyte covering of lym- relative importance of the different cytokines in the devel-
phatic vessels, lymphatic vessel valves, and lymphatic ves- opment of individual lymphoid organs has been recently
sel maturation. At E11.5-12.0, CCL21 is expressed in these summarized.176
lymphatic-biased endothelial cells, as is VEGFR-3, which The NF-κ B signaling pathways, downstream of the TNF
is reduced in blood endothelial cells. The endothelial cells family receptors, play important roles in lymphoid organ
expressing LYVE-1, Prox1, VEGFR-3, and CCL21 become ir- development.177 The alternative pathway, characterized by
reversibly committed toward a lymphatic pathway.153,154 The NF-κ B–inducing kinase (NIK) and IKKα is particularly

important. aly/aly mice, which have a point mutation in Nik, HEV gene expression.198,199 Recent data suggest that the
lack all lymph nodes and Peyer’s patches.178,179 In these mice, actual physical presence of lymphocytes in the close vicin-
LTβR, but not TNFR, mediated signaling between NIK and ity of HEVs contributes in important ways to the physical
members of the TRAF family appears to be disrupted.179–181 cobblestone appearance of these vessels.200
LTβR signaling induces gene expression via both the classi-
cal and alternative NF-κ B pathways in mouse embryo fibro- Changes in Lymph Nodes after Immunization
blasts. The classical pathway mediated by p50:p65 heterodi- Lymph nodes undergo dramatic changes and remodeling
mers induces expression of proinflammatory genes (Vcam-1, after immunization. Early after a variety of immunogenic
Mip1b, Mip2). Intraperitoneal injection of an agonistic LTβR exposures, such as skin painting with oxazolone, injection
antibody induces splenic chemokines (CCL19, CCL21, and of ovalbumin or sheep red blood cells in adjuvant, or bac-
CXCL13) and requires NIK activity and subsequent p100 terial or viral infection, remodeling occurs. This remodel-
processing.182 IKKα is a critical component in alternative ing is apparent as a complex kinetics of changes in lymph
NF-κ B pathway. Mice with a mutated form of the Ikka gene flow, lymph cell content, blood flow, HEV gene expression,
have reduced HEV expression of HEC-6ST (GlcNAc6ST2) and lymphatic vessels.199 Afferent lymph flow and lymph cell
and GlyCAM-1, further confirming that the LTβR signal content increase soon after initial inflammation, and even-
regulates HEVs through the alternative NFκ B pathway.183 tually return to preimmunization levels.201–205 Lymph node
Several other signaling pathways that contribute to lym- lymphangiogenesis occurs, which eventually resolves.199,206
phoid organogenesis include the helix-loop-helix transcrip- Blood flow and lymphocyte migration into lymph nodes
tion factor inhibitor (Id2) and retinoid acid-related orphan peak at 72 to 96 hours,166,204,207,208 accompanied by an increase
receptors (RORs) RORγ and RORγ t.184–186 in HEV number and dilation,209,210 accounting for the signif-
Studies of lymph node anlage formation reveal the icant lymph node enlargement apparent at 72 to 96 hours.207
mechanisms by which cytokines trigger and coordinate Efferent lymph flow also increases soon after immunization,
lymphoid organogenesis. There is likely an initiating fac- but lymph cell content in the efferent lymph drops during
tor, though this has not been definitively identified. In the the first several hours, indicating the first wave of accumula-
case of Peyer’s patches, a cell producing the receptor tyro- tion of lymphocytes in the draining lymph node. The cell
sine kinase has been implicated.187,188 Recent data indicate content of the efferent lymph later increases and peaks at
that the earliest stages of lymph node development appear 72 to 96 hours.211 These events, taken together, contribute
to be dependent on retinoic acid that may be produced by to the significant enlargement of the draining lymph node
nerves in the vicinity of the developing node.189 Circulating at day 4 after immunization. During the early times after
CD4 + CD3 − CD45 + RORγ t + hematopoietic progenitor cells immunization, despite the increase in the number of HEVs,
called lymphoid tissue–inducer cells, derived from fetal the expression of genes that contribute to L-selectin ligand,
liver progenitors,98,185,190–192 provide crucial signals in lymph including Fuct-vii, Glycam-1, Sgp200, and Chst4, is initially
node organogenesis. These lymphoid tissue–inducer cells downregulated followed by a recovery.148,199,212 However,
accumulate in the developing lymph node, forming clus- despite the downregulation of chemokines CCL21 and
ters with resident stromal organizer cells, to initiate a cas- CXCL12, some genes such as those encoding CXCL9, CCL3,
cade of intracellular and intercellular events that lead to the and E-selectin are upregulated,213 as is MAdCAM-1,199 sug-
maturation of the primordial lymph node.1,98 During this gesting a reversion to an immature phenotype before the
early step, a positive feedback loop involves several signal- eventual recovery of the mature phenotype.
ing pathways, including LTαβ /LTβR, IL-7R/IL-7, CXCR5/ In rodents, HEV maturation is coincident with the con-
CXCL13, and RANK/RANKL, are expressed on the lym- tinuing development of the lymph node and population of
phoid tissue–inducer cells and the stromal organizer cells. lymphocytes after birth, indicating that the mature HEV
The prolonged interaction between lymphoid tissue– phenotype relies on the lymph node microenvironment.139,197
inducer cells and stromal organizer cells promotes the de- Plasticity of the mature lymph node is also seen after injec-
velopment of HEVs, which support the entry of naïve lym- tion of LTβR-Ig. LTβR-Ig treatment reduces lymph node cel-
phocytes.193 It is unclear how HEVs differentiate from the lularity, reverts the HEV phenotype to the immature state,
flat blood vessels during early lymphoid organogenesis. At inhibits FDC function, and disrupts immune responses to
birth, HEVs of all lymph nodes express MAdCAM-1, which foreign antigens.198,199,214
is replaced in the first few days in peripheral lymph nodes
by PNAd.194 Both MAdCAM-1 and PNAd are expressed in
mucosal lymph nodes. LTα alone can induce MAdCAM-1, Spleen
but PNAd requires LTαβ.2,195,196 In the remaining mesen- Function
teric lymph nodes of Ltb- / - mice, PNAd expression is im- The spleen is a large reddish organ located beneath the
paired,196 indicating that optimal lymph node HEV PNAd diaphragm close to the stomach and the pancreas. It is
expression requires LTα1β2 signaling through the LTβR the main fi lter of the blood and integrates the innate and
and the alternative NF-κ B pathway.183 Because the matura- adaptive responses. Its structure in well-defined different
tion of HEVs is coincident with further development of the compartments, red pulp and white pulp, determines a
lymph node,139,197 the homing of LT-expressing lymphocytes variety of functions. The red pulp is a source of hemato-
most likely contributes to HEV maturation. Continual sig- poiesis in the embryo that can continue in adult life under
naling through the LTβR is necessary for maintenance of stress. Clearance of blood-borne damaged platelets, aged

erythrocytes, and dead cells also occurs in that region. An be divided into two areas: red pulp and white pulp. A tran-
important function, consequent to its destruction of effete sit area, the marginal zone (MZ), surrounds the white pulp
erythrocytes, is its role in iron recycling. In some species, (Fig. 3.4). These compartments are anatomically and func-
such as horses or dogs, the spleen stores erythrocytes that tionally distinct. The red pulp, in its activities as a hematog-
can be released after stress. The white pulp and marginal enous organ, removes damaged cells and acts as a site for
zone constitute the highly organized lymphoid compart- iron storage and turnover. The white pulp is an organized
ment of the spleen. Due to this organized lymphoid struc- lymphoid structure. The complex structure of the spleen is
ture and the special vasculature and circulation, the spleen directly related to the complexity of the vasculature of this
is a crucial site for blood-borne antigen clearance and organ. The afferent splenic artery branches into central ar-
presentation to T and B cells. The spleen is crucial in defense terioles, surrounded by white pulp areas, and end in cords
against blood-borne pathogens and contributes most signif- in the red pulp. Blood then collects in venous sinuses that
icantly to defense against bacterial and fungi infections.215 determine the engulfment of erythrocytes by red pulp mac-
rophages. Finally, the sinuses empty into the efferent splenic
Architecture and Cellular Composition vein. Some small arterial branches end in the MZ, which
A fibrous capsule and trabeculae of fibrous connective tis- demarcates the red and white pulp. The MZ structure dif-
sue maintain the structure of the spleen. The general prin- fers somewhat between humans and rodents. A perifol-
ciples of splenic organization are similar across species, licular zone surrounds the human MZ, which consists of
though specifics may differ.216–220 Classically, the spleen can inner and outer marginal zones, whereas the rodent MZ is
FIG. 3.4. Organization of the Spleen White Pulp. Immunofluorescence staining of a white pulp unit in the mouse
spleen. T cells (anti- cluster of differentiation [CD]4+anti-CD8+, red) are localized around the central arteriole.
B cells (anti-immunoglobulin M, green) are localized in follicles around the T-cell area, surrounded by a layer of
metalophilic macrophages (labeled with monoclonal antibody-1) and a more peripheral layer of marginal zone
macrophages (labeled with monoclonal antibody ERTR9, orange). The marginal zone is located between the met-
alophilic macrophage and the marginal zone macrophage layers (not shown).357 From Chaplin358 with permission.

a single structure and has no perifollicular zone. The MZ is lymph node.234 This conduit differs from that in the lymph
the transition between the innate and acquired immune sys- node with regard to the identity of the transported mole-
tems and is an important transit area for cells reaching the cules by the fact that it contacts the blood rather than the
white pulp. It also contains a large number of resident cells. lymphatic system.
There are two specialized macrophage populations: the
MZ macrophages and MZ metallophilic macrophages. MZ Traffic In and Out: Chemokines and Adhesion Molecules
macrophages have a high phagocytic activity and are phe- Cell trafficking in the spleen is similar to that which occurs
notypically different from other macrophage populations in the lymph node in some respects, and differs in others.
in the spleen as exemplified by the expression of SIGNR1 The MZ is the main transit area for blood cells that enter the
and macrophage-associated receptor (MARCO) with col- white pulp. This process seems to be controlled by marginal
lagenous structure, molecules implicated in the recognition sinus lining cells235 and chemokines, involving signaling
of pathogens. SIGNR1, a C-type lectin in the mouse that is through G-couple protein receptors.236 Stromal cell–pro-
a homologue of human DC–specific intercellular adhesion duced lymphoid chemokines CXCL13, CCL19, and CCL21
molecule,221 is also found in medullary and subcapsular mac- control the migration of lymphocytes and their organization
rophages in the lymph node. MZ macrophages are crucial in the various compartments in the white pulp. CXCL13 is
Marco for the capture of a wide variety of pathogens, includ- essential to positioning of B cells in the follicles, whereas T
ing yeast, bacteria, and viruses.222,223 MARCO, expressed cells respond to CCL19 and CCL21. After the immune re-
constitutively on MZ macrophages,224 is a class A scavenger sponse, germinal center B cells differentiate into plasma
pattern recognition receptor. In addition to the recognition cells in the white pulp, then migrate to the red pulp or re-
of blood-borne pathogens, MARCO interplays with MZ B turn the bloodstream to migrate to peripheral tissues. In the
cells, modulating their migration to the white pulp.225 MZ migration to the red pulp, plasma cells upregulate CXCR4,
metallophilic macrophages are located at the inner bor- a receptor that binds the chemokine CXCL12, which is ex-
der of the MZ, in contact with marginal sinus lining cells, pressed in the red pulp.
stromal cells that express MadCAM-1.226 MZ metallophilic The role of adhesion molecules with regard to lymphocyte
macrophages differ from the MZ macrophages and red pulp traffic in the spleen is not fully well understood. Lymphocytes
macrophages in that they express the sialic acid-binding enter into the white pulp through the marginal sinus and cells
Ig-like lectin sialoadhesin (Siglec-1), which binds to oligo- lining that structure express MAdCAM-1 and ICAM-1.226,237
saccharide ligands present on many cells.227 Little is known MadCAM-1 expression in these cells seems to be involved
about the specific roles of this population of macrophages in splenic structure development and therefore migration of
during the immune response; however, the phenotype of B and T cells.235 However, treatment with anti-MAdCAM-1
mice deficient in sialoadhesin implies a role of MZ metallo- or anti-α1β7 antibodies does not totally inhibit homing to
philic macrophages in T-cell activation.228,229 Recent studies the spleen, suggesting that this ligand-receptor pair is not
confirm that MZ metallophilic macrophages are essential the only receptor required for lymphocyte entrance into the
for cross-presentation of blood-borne adenovirus antigens white pulp. Early studies indicating that treatment with an
to splenic CD8 DCs, activating cytotoxic T-lymphocytes.230 antibody that blocks the αL-integrin chain of LFA-1 inhib-
The MZ also contains a specialized subset of B cells that dif- its homing by only 20%, and that lymphocytes deficient in
fer phenotypically and functionally from follicular B cells; LFA-1 enter the white pulp,237 suggested that LFA-1 is not
they can be considered as a bridge between the innate and absolutely essential for entry of all cells into the white pulp.
adaptive immune systems. They express higher levels of IgM However, both α4β1 and LFA-1 integrins are necessary for
low levels of IgD and the molecule CD1d. These B cells bind B-cell retention in the MZ,238 indicating a role for VCAM-1
antigen in the MZ directly and/or through interactions with and ICAM-1 in cell trafficking in spleen. Once lymphocytes
MZ macrophages231 and migrate to the white pulp, where have encountered antigen, they most likely undergo changes
they present blood-borne antigens.232 The organization of in chemokine receptor and adhesion molecule expression
the remainder of the white pulp of the spleen is similar to similar to those noted in the lymph nodes, leave the white
that of the other secondary lymphoid organs with compart- pulp and enter the red pulp (plasma cells) or the circula-
mentalized B- and T-cell areas. The white pulp consists of a tion. The mechanisms by which lymphocytes leave the white
central arteriole that is surrounded by T cells, also known as pulp are unclear. The observation of channels that bridge
the periarteriolar lymphoid sheath, which is surrounded by into the MZ239 suggests one route of egress of lymphocytes.
B-cell follicles (see Fig. 3.4). T cells interact with DCs and B In the lymph nodes, the process of egress of activated lym-
cells. B cells migrate to the follicles where they interact with phocytes through the efferent lymphatic vessels is mediated
FDCs. At the T:B-cell border, T cells, especially T follicular by upregulation the expression of S1P1. In the spleen, S1P1 is
helper cells,233 interact with B cells. Germinal centers are the required for B-cell localization in the MZ, and the interplay
site of somatic hypermutation and Ig class switching. Plasma with CXCR5 regulates the constant trafficking of these cells
cells are found mainly in the red pulp. between the MZ and the white pulp,232,240 as MZ B cells from
The spleen does not have an afferent lymphatic system, S1P1-deficient mice are not found in the MZ but are found
and initial antigen transport must occur through the blood in the follicles.
vasculature. A conduit system has been described that al- Although it is still unclear which molecules are involved
lows antigens and chemokines to be transported through in the egress of lymphocytes from the spleen, and splenic
the white pulp in a manner similar to that described for the egress cannot be compared to egress in the lymph nodes,

recent studies attribute an important role to the transcrip- fetal spleen development, when segregation between white
tion factor Nkx2.3 in the regulation of spleen vasculature and red pulp begins, but essential for maintenance of this
and homing of lymphocytes.241,242 structure during postnatal development of the white pulp.248
Tnf- / - and Tnfr1- / - mice show defects similar to lympho-
Development toxin deficient mice, with the exception of MZ B cells. T
Because the spleen has characteristics of both a hemato- cells, B cells, and CD4 + CD3 − cells produce the cytokines
poietic organ and a secondary lymphoid organ, the genes necessary for maintenance of splenic architecture.249,250
that regulate its development include, in addition to LT,
others that are concerned with patterning and hematopoi- Plasticity after Virus Infection
esis. Several genes that contribute to the development of the Though much is known regarding changes in the lymph
spleen are also crucial for normal development of other non- node after immunization, the spleen has not been studied
lymphoid organs. The first sign of splenic development in as extensively in this regard. However, after infection with
the mouse occurs at E10.5-11. Segregation of red and white cytomegalovirus, white pulp T:B compartmentalization is
pulp starts to be regulated during embryogenesis and con- disrupted.251 The spleens of Lta- / - deficient mice exhibit a
tinues after birth. Some structures, such as the MZ, develop marked reduction in expression of CCL21-ser. This is even
during the 3 first weeks after birth. At E10.5-11, progenitor further reduced in cytomegalovirus infection, indicating
cells begin to condense within the dorsal mesogastrium, that, in the adult, LT-independent pathways can contribute
adjacent to the stomach and dorsal pancreas. The spleen to maintenance of expression of lymphoid chemokines.
and pancreas are so intimately associated that it is difficult
to distinguish between them at these early stages. In fact,
many genes that affect splenic development also contribute Mucosal-Associated Lymphoid Tissues
to pancreatic development. Several of these are homeobox General Features
genes and transcription factors that are expressed in the The MALT covers all mucosal surfaces not only in the gut,
splenopancreatic mesenchyme at E10.5.215 The effects of just but also the oropharyngeal and lacrimal mucosae, the nasal
one of these genes, Hox11 (now called Tlx1215), are on the and bronchial airways, and the genitourinary tracts. MALT
spleen. The others affect multiple organs. Tlx1 was origi- protects a huge surface area and contains approximately half
nally described as an oncogene in T-cell childhood acute of the lymphocytes of the entire immune system.252 MALT
leukemias involving the (10;14) translocation breakpoint.243 is considered to be the body’s gatekeeper because it is in
Tlx1-deficient mice are asplenic,244 and the product of this intimate contact with the commensal flora at the mucosal
gene is a cell survival factor.245 Several additional genes that surfaces.
are important for development of lymph nodes, NALT, and Although the location of some mucosal lymphoid tissues
Peyer’s patches also contribute to splenic development. Mice like the palatine tonsils and appendix are fixed, all MALT
deficient in members of the LT/TNF ligand receptor family are somewhat plastic because of their constant exposure to
and their downstream signaling molecules in the classical environmental antigens that induces them to change and re-
and alternative pathways exhibit defects in splenic develop- model. Tonsils, adenoids, and their equivalent in rodents, the
ment. However, none of these molecules is necessary for the NALT, have a fixed location, whereas the location, number,
early splenic anlagen, as mice deficient in any of the che- and size of Peyer’s patches in the small intestine vary accord-
mokines or cytokines retain a spleen. Recent studies address ing to antigen exposure. The BALT in the lung and ILFs in
a role for Glce, an enzyme that modifies heparan sulfate, the colon are the most plastic MALTs, and their number and
in early lymphoid tissue morphogenesis, as fetal spleen in location are subject to change by environmental influences.
Glce deficient mice exhibit a reduced size.246 The changes in The mucosal epithelium that surrounds MALT is popu-
lymphoid tissue organization in mice deficient in LT/TNF li- lated by a dense network of DCs, plasma cells, and intraepi-
gand receptor family members are due in part to a reduction thelial lymphocytes that helps to maintain the epithelial
or near absence of lymphoid chemokines (CXCL13, CCL19, barrier. These intraepithelial leukocytes provide retinoic
CCL21).247 Lymphoid chemokine messenger ribonucleic acid and cytokines like IL-10 and TGF-β that condition the
acid (mRNA) expression is reduced in the spleens of mice gut to become tolerant of antigens produced by the harm-
deficient in TNFR1, TNF, LTα , or LTβ, though CXCL12 less commensal bacteria.253,254 They are also responsible for
mRNA levels are normal. However, treatment with an ago- inducing and maintaining tolerance to food antigens and
nistic LTβR antibody induces expression of the lymphoid commensal bacteria.255
chemokines and CXCL12, suggesting that signaling through The large MALT structures like the tonsils, NALT, and
both the classical and alternative NF-κ B pathways are re- Peyer’s patches share many features with lymph nodes. Like
sponsible for organization and maintenance of splenic white lymph nodes, MALT contains HEVs for the entry of naïve
pulp architecture. Mice deficient in LTα exhibit a disorga- lymphocytes, and stromal cells that secrete chemokines
nized white pulp with loss of T- and B-cell compartmental- that direct lymphocyte traffic in the MALT (CCL19, CCL21
ization, MZ macrophages, metallophilic macrophages, MZ and its receptor CCR7256 ; CCL25 and CCL28 and their re-
B cells, MAdCAM-1 sinus lining cells, and germinal centers. ceptors CCR9 and CCR10257; and CXCL13 and its receptor
Ltβ - / - mice exhibit similar characteristics except that the CXCR5258,259). MALT also contains separate T- and B-cell
disorganization is somewhat less pronounced. However, compartments with B-cell follicles, FDCs, germinal centers,
these lymphoid cytokines seem to be dispensable during and interfollicular T cells and DCs.260

MALT differs importantly from other secondary lym- whereas lymphocytes leaving the Peyer’s patch express α4β7
phoid organs with regard to its capsules and afferent lym- and migrate to the gut mucosa. These “preferred pathways”
phatic vessels. Unlike the spleen and lymph nodes, which for effector cells coming out of the different MALT has led
are surrounded by a dense fibrous capsule, MALT often has to the suggestions that MALT is compartmentalized and cells
no capsule, with the exception of the tonsil that has a par- circling in the BALT are in a separate compartment from
tial capsule that separates it from the pharyngeal constrictor the GALT. It is more likely that the nature of antigen influ-
muscles. Because MALT can sample antigens directly at the ences the homing of effector cells. There needs to be flexibil-
epithelial surface, it has no afferent lymphatics. MALT epi- ity, and there is much overlap and redundancy in the system.
thelium, especially that of the Peyer’s patch, is replaced by For example, after the removal of the tonsils and adenoids in
specialized lymphoepithelial cells called microfold (M) cells, humans, or the NALT in the mouse, the cervical lymph node
which because of their high transcytotic capacity, transport can act as an inductive site.268 Likewise, the mouse BALT can
antigens to the underlying lymphoid tissue. Tonsils and ad- mount an immune response in the absence of secondary lym-
enoids are covered by a squamous epithelium and have few phoid organs.269,270
M cells. They can sample surface antigens using epithelial
DCs that push dendrites through the epithelial surface.261,262 Tonsils and Adenoids
M cells have not been characterized in human BALT.263 Waldeyer’s ring is a group of lymphoid tissues encircling the
The proximity of MALT to the epithelium is an impor- wall of the throat that are the fi rst defense against patho-
tant point, because it helps differentiate MALT from tertiary gens entering through the mouth or nose.271 Humans have
lymphoid organs (see following discussion). MALT can be several tonsils with indistinct borders: one pharyngeal ton-
defined as organized lymphoid structures in the mucosa sil (adenoid), two tubal, two palatine, and one lingual. In
that are in direct contact with the epithelium. Using these addition to T cells, the tonsils contain a large complement
criteria, a lymphoid aggregate in the submucosa that is not of B cells, many of which are positive for IgA.271 The tonsil
directly in contact with the epithelium (ie, below the lamina epithelium makes the polymeric IgA receptor, or secretory
propria) is more correctly defined as a tertiary lymphoid component,272 crucial for transport of IgA dimers across the
organ.264 epithelium. Secreted IgA provides an early form of defense
against pathogens and toxins. Overall, the adenoids produce
The Mucosal-Associated Lymphoid Tissue Immune more secreted IgA than the tonsils.
Response: Inductive and Effector Sites The common cold virus uses ICAM-1 as receptor to in-
MALT can simultaneously be a site for the induction of an vade the nasal mucosa. ICAM-1 is also an expressed HEV
immune response and an effector organ. The MALT surface in tonsils.271 As noted previously, HEVs are specialized ves-
containing M cells and DCs efficiently samples and trans- sels that express adhesion molecules that allow naïve lym-
ports antigens across the epithelium. Immediately below phocytes enter the tonsil. Naïve L-selectinhi lymphocytes
this single-cell epithelium is a dense network of DCs that adhere to PNAd on HEVs.258,259 L-selectinhi cells also bind
can extend dendrites through the epithelium to grasp an- to MAdCAM-1 predominantly found in the gut HEVs.
tigen to present to lymphocytes. Two distinct types of DC Considering the “mucosal name” MAdCAM-1, it may be
have been identified in MALT: CD103 + DC and CX3CR1+ surprising to discover that tonsil HEVs express MAdCAM-1
DC.265 Both types of DC produce tolerogenic responses to only weakly or not at all.273 In mice, the HEVs also manufac-
commensal bacteria by enhancing the differentiation of ture the chemokine CCL21 that can recruit naïve T cells and
Foxp3 + Tregs and inhibiting that of inflammatory Th17 cells. mature DCs that express CCR7. Human tonsil HEVs do not
CD103 + DCs express αE integrin, make retinoic acid, induce manufacture CCL21 but are able to display chemokines that
Tregs, and increase T-cell expression of the two gut hom- have been made and secreted by fibroblasts.274 During ton-
ing receptors CCR9 and α4β7. CD103 + DCs exit MALT via sillitis, inflammatory chemokines, cytokines, and adhesion
the lymphatics and present antigen in the draining lymph molecules such as CCL19,256 VCAM-1, and E- and P-selectin
nodes. CX3CR1+ DCs that express the chemokine receptor are upregulated.165
for fractalkine do not migrate to the lymph node. CX3CR1+
DCs are longer-lived mucosal resident DCs that produce less
retinoic acid than CD103 + DCs.265 Nasal-Associated Lymphoid Tissue
MALT has three effector roles: local antibody secretion, The NALTs are a pair of lymphoid organs above the soft
systemic antibody secretion, and effector lymphocyte disper- palate in mice and rats that are considered analogous to
sal. MALT plays an important role in defense against patho- Waldeyer’s ring.275 Despite being anatomically separate from
gens by generating cells that migrate to other sites. MALT can the genitourinary tract, NALT has an important “effec-
dispatch lymphocytes to lymph nodes, the spleen, and plasma tor” role concerning the generation of immune responses in
cells to the bone marrow. MALT also sends lymphocytes to the genitourinary tract.276 After nasal immunization with
other MALT effector sites including the salivary and lacrimal human papillomavirus 16 or ovalbumin, human papillo-
glands, the lactating breast,266 and the vagina. There are some mavirus 16–specific or ovalbumin-specific IgA is detected
differences between the effector cells made at different MALT in vaginal washes,163,277,278 and cytotoxic T cells are found in
sites. NALT-derived B cells express CCR7 and do not home vaginal draining lymph nodes.267 However, the NALT itself
back to the gut. Instead, cells induced in the NALT home to is clearly an inductive site for both humoral and cellular im-
the salivary glands and the vaginal mucosa lymph nodes,267 mune responses,279 and supports class switching to IgA.280

The stromal cells in murine NALT produce both CCL19 and Peyer’s Patches
CXCL13, whereas, in contrast to lymph nodes, only HEVs Peyer’s patches are lymphoid aggregates aligned on the
transcribe CCL21 mRNA.163 NALT HEVs express high levels antimesenteric border of the small intestine (Fig. 3.5A).
of luminal and abluminal PNAd and HEC-6ST.163 Similar to They are present in most species, though their number
the tonsils, the major homing receptor-ligand pair in NALT is and location vary. They are dome shaped and are covered
L-selectin and PNAd, rather than MAdCAM-1 as might have by a specialized epithelium that lacks surface microvilli
been expected of this mucosal-associated tissue.281 and goblet cells; they have numerous M cells. Because
Peyer’s patches lack afferent lymphatic vessels, these M
Development cells are critical transporters of antigen. Pathogens, in-
Human NALT and tonsils appear early in fetal develop- cluding human immunodeficiency virus and salmonella
ment, are cellular, and have primary follicles. After birth, take advantage of the M cells to facilitate their own in-
secondary follicles and germinal centers appear in response vasion.293,294 Below the epithelium is a diffuse area, the
to bacterial antigens.266 By contrast, the NALT in the mouse subepithelial dome, divided into three to six parts. Peyer’s
and the rat is hypocellular at birth and undergoes dramatic patch organization is similar to that of the lymph node
changes after weaning, strongly suggesting that bacterial (see Fig. 3.5). Peyer’s patches contain 6 to 12 basally lo-
colonization is important for NALT development in these cated germinal centers, but the B-cell areas are larger with
species. Id2 is required for initiation of the rodent NALT,282 a T:B cell ratio of 0.2; CD4 + cells predominate over CD8 +
although RORγ T, LTα , or LTβ are not required for this cells.295 Although M cells transport antigen across the
step.162 The expansion of rodent NALT at weaning includes epithelial barrier, they are not believed to have a crucial
changes in the expression of LTα and LTβ and lymphoid role in processing or presenting antigen. Peyer’s patches
chemokines, leading to T- and B-cell compartmentalization are populated by several subsets of DCs that can carry out
and HEV maturation.162,163 The alternative NF-κ B pathway these functions.296,297
is required for the expression of chemokines and the HEV
genes glycam-1 and chst4 in NALT.183 LTα , LTβ, IL7R, and Trafficking In and Out
the NIK signaling pathways are required for NALT organi- Cells enter Peyer’s patches through HEVs. In contrast to
zation and function, and mice that are deficient in these cy- peripheral lymph nodes and the NALT, the HEVs of Peyer’s
tokines have a hypocellular NALT.163,183,283,284 patches express MAdCAM-1, 298 and homing depends on
the interaction of MAdCAM-1 with the integrin α4β7 on
Bronchus-Associated Lymphoid Tissue lymphocytes. Luminal PNAd is rarely found in Peyer’s
BALT is less organized but more responsive to environ- patch HEVs; only occasional abluminal expression is de-
mental antigens than the NALT. The number of lymphoid tected. This pattern is identical to that seen in lymph node
aggregates varies depending on the level of microbial expo- HEVs in chst4- / - mice.128 However, lymphocytes from
sure and germ-free pigs have no BALT.252,285 BALT is found mice that lack both L-selectin ligand and α4β7 home less
commonly in rabbits and rats, is less frequent in guinea pigs efficiently to Peyer’s patches than do lymphocytes from
and pigs, and is absent in cats.264 BALT is not a prominent mice that lack only one or the other of the ligands, 299,300
structure in the laboratory mouse, and its presence varies by suggesting that an L-selectin ligand does contribute to
strain and age.286 homing to Peyer’s patches. Even though P-selectin is ex-
An inducible form of the BALT (iBALT) has been depressed only weakly on HEVs in Peyer’s patches, cells
scribed in Lta- /- mice after infection with influenza.287 from mice deficient in P-selectin show reduced rolling
The term iBALT is misleading because these aggregates have and adhesion in vivo. This suggests that P-selectin, as
no contact with the bronchial lumen and cannot sample in- well as MAdCAM-1, is important for homing to Peyer’s
spired antigens. Therefore, iBALT should be considered as a patches. 299
tertiary lymphoid organ.288 Splenectomized, lethally irradi- Lymphoid chemokines CCL19, CCL21, and CXCL13 are
ated LTα− / − mice reconstituted with normal bone marrow found in the T- and B-cell areas301 (see Fig. 3.5B). Stromal
have no lymphoid organs. However, these mice can generate cells make CXCL12 mRNA, but the protein is also found
immunological memory in the iBALT.269,270 on HEVs, presumably transported in a manner similar to
BALT is rare in normal adult human lungs, but tertiary CCL19.117 The CXCL16/CXCR6 chemokine/receptor pair
lymphoid organs are common in chronic pulmonary dis- is important in Peyer’s patches but not in lymph nodes.
eases.283 BALT is frequent in children and is found in fe- Even under germ-free conditions, CXCL16 is constitutively
tuses after infections in utero.289,290 Like NALT, the HEVs in produced by the dome epithelium.295 CD4 + and CD8 +
human BALT express PNAd and not MAdCAM-1.291 cells that express CXCR6 are found in Peyer’s patches
and require CXCL16 to home to the subepithelial dome.
CCL20 and its receptor CCR6 are also important in Peyer’s
Gut-Associated Lymphoid Tissue patches. CCL20 is made by the intestinal epithelium and
Gut-associated lymphoid tissue is the largest immune system plays a crucial role in dendritic cell trafficking to Peyer’s
in the body. The gut-associated lymphoid tissue in the small in- patches.302 Peyer’s patch HEVs express CCL25, the ligand
testine includes the Peyer’s patches, smaller isolated lymphoid for CCR9, which has been defi ned as a mucosal homing
follicles, and cryptopatches. In the large intestine, there is the chemokine receptor for intraepithelial lymphocytes and
appendix, caecal patches, and lymphoglandular complexes.292 plasma cells.78

FIG. 3.5. Organization of Peyer’s Patches. A: Photomicrograph of

Peyer’s patch. Peyer’s patches are located in the intestine near in-
testinal villi. The follicle associated epithelium (FAE) is in contact
with the gut lumen. M cells (not shown) in the FAE transport anti-
gen into the subepithelial dome, populated by dendritic cells and
T cells. The interfollicular T-cell–rich region surrounds the B-cell
follicle and germinal center. Courtesy of A. Iwasaki (Yale University
School of Medicine, New Haven, CT). B: In situ hybridization of che-
mokine messenger ribonucleic acids (mRNAs). On the left, CCL21
mRNA defines the T-cell zone and high endothelial venules; on the
A right, CXCL13 defines the B-cell zone.
Development Lymphocytes influence the maintenance of Peyer’s

Several cytokines and chemokines have been shown to be patches. Mice that lack mature T and B cells have either un-
crucial for Peyer’s patch development. Ltα- /- and Ltβ - /- detectable or small Peyer’s patches that lack follicles and ger-
mice completely lack Peyer’s patches,161,303 as do Cxcr5- / - minal centers.311,312 B cell–deficient mice retain some M cells,
mice.174,175,301 Because mice deficient in the LTβR also lack suggesting that T cells may regulate M-cell maintenance.
Peyer’s patches,304 these effects are mediated in large part When B cell–deficient mice are reconstituted by a membrane
through LTα1β2. However, some Tnfr1- / - mice lack orga- IgM transgene, M cells are found at levels comparable to
nized Peyer’s patches, suggesting that either LTα3 or TNFα those of normal mice,311 indicating that cells in addition to
also influences generation or later stages in maintenance of CD4 + CD3 + lymphoid tissue–inducer cells are crucial for the
Peyer’s patches.305 Mice deficient in IL-7, RORγ t, Id2, NIK, and maintenance of mature, functioning Peyer’s patches.
factors in the classical and alternative NFκ B pathways170,177,304
lack Peyer’s patches, although RANKL is not required.172
A model for the embryonic development of Peyer’s Cryptopatches and Isolated Lymphoid Follicles
patches306–308 provided the framework for studies in that The small intestine of a mouse contains 100 to 200 isolated
organ. The first sign of Peyer’s patch development at E15.5 in lymphoid follicles and more numerous cryptopatches. The
the mouse is the appearance of regions on the small intestine cryptopatches in the lamina propria containing lymphoid
that stain positively with antibodies to ICAM-1, VCAM-1, tissue–inducer cells and DCs can be precursors of ILFs. The
and LTβR. The cells in these aggregates are called the lym- cryptopatches are composed of lin− c-kit + cells, DCs, and
phoid tissue organizer cells and express CXCL13, CCL19, VCAM-1+ stromal cells with few or no mature T and B cells.
CCL21, and IL-7. At E17.5, clusters of IL-7R + CD4 + CD3 + The cryptopatch cells express RORγ t, IL-7R, and CCR6,313
inducer cells are found. These express LTα and LTβ, CXCR5 and their development is dependent on IL-7, CCR6, and its
and CCR7, Id2, and RORγ. They also express α4β1 integ- receptor CCL20. Cryptopatches are quite plastic, and al-
rin activated by CXCR5309 that allows interaction with the though the LT family is necessary for their development,
VCAM-1+ organizer cells. At E 18.5, mature T and B cells they can be restored by administration of wild-type bone
enter through HEVs, CCL20 is produced by the FAE, and marrow to adult Lta- / - mice.
DCs expressing CCR7 and CCR6 are found. By day 4, the After mice are exposed to microbes or during some forms
typical microarchitecture is apparent, with M cells and T- of autoimmunity, cryptopatches give rise to ILFs,314,315 which
and B-cell compartmentalization.310 resemble small Peyer’s patches and usually only have a single

dome. ILFs require B cells that express LT, but not T cells,
for their formation.316. In contrast to lymph nodes, they also
require the TNFR1.317 The location of ILFs is antigen driven,
and they resolve completely after mice are treated with anti-
biotics or cytokine inhibitors.121,185,314,316,317
Appendix
Both rabbits and humans have an appendix,318 a vestigial
organ distal to the ileocecal junction that divides the small
intestine from the colon. The appendix has epithelial M cells
and dense lymphoid accumulations and germinal centers
that are similar to Peyer’s patches. Similar to the tonsils and
adenoids, the appendix involutes and regresses after puberty.
Colon Lymphoglandular Complexes

The colon has the highest bacterial load of any of the
MALTs, and the distribution of the MALT is more random
than in the small intestine. Lymphoglandular complexes FIG. 3.6. Tertiary Lymphoid Tissue. Schematic diagram of Sjögren
are smaller than Peyer’s patches. As their name suggests, syndrome salivary gland. Note similarities with the organization of
the lymphoid cells of the lymphoglandular complexes sur- a lymph node.
round the mucus glands.292
TERTIARY LYMPHOID TISSUES and many other solid tumors.326 The tendency for tertiary
lymphoid tissues to develop into lymphomas has also been
Similarities to Secondary Lymphoid Organs noted in many studies.327
Tertiary lymphoid tissues, also termed tertiary lymphoid or- Ectopic lymphoid tissues are also apparent in several ani-
gans, are ectopic accumulations of lymphoid cells that arise in mal models. The pancreatic infiltrates in early stages of dia-
nonlymphoid organs during chronic inflammation through betes in the nonobese diabetic mouse express lymphoid che-
a process termed “lymphoid neogenesis” (also “lymphoid mokines328 and HEVs expressing MAdCAM-1,329,330 PNAd,
neo-organogenesis”).2 The iBALT could be considered as either and HEC-6ST.135 The brain in experimental autoimmune
a secondary or tertiary lymphoid tissue. This semantic issue encephalomyelitis, a model of multiple sclerosis, has HEVs,
epitomizes the plasticity of all lymphoid organs and the fact CCL19, CCL21, CXCL13, and FDCs.331–333 The thyroid in the
that lymphoid organ regulation represents a continuum from BioBreeding (BB) rat has T- and B-cell compartmentalization
ontogeny through chronic inflammation. A notable difference and DCs,334 and the gut in autoimmune gastritis in the
between secondary lymphoid organs and tertiary lymphoid mouse has HEVs and CXCL13.335 Atherosclerotic plaques of
tissues is the fact that the latter can arise in almost any organ apoprotein-E– deficient mice exhibit a marked increase in T
in the adult. Nonetheless, tertiary lymphoid tissues exhibit re- and B cells, expression of CCL19 and CCL21, and PNAd +
markable morphologic, cellular, chemokine, and vasculature HEVs.336–338 Lymphoid neogenesis also occurs in chronic
similarities to secondary lymphoid organs. They exhibit T- mouse heart allograft rejection.339,340 Tertiary lymphoid tis-
and B-cell compartmentalization; naïve T and B cells; DCs; sues have been induced in several transgenic mouse models
FDCs; germinal centers; plasma cells; lymphoid chemokines with the use of tissue-specific promoters driving the expres-
CCL19, CCL21, and CXCL131; HEVs135,196,319; and conduits.320 sion of inflammatory cytokines or lymphoid chemokines.1
Lymphatic vessels have also been noted in tertiary lymphoid These mouse models, in addition to serving as examples of
tissues in chronic graft rejection321,322 and mouse models of human disease, have provided valuable insight into the regu-
Sjögren syndrome134 (Fig. 3.6) and Type 1 diabetes.323 Although lation of secondary lymphoid organ development.
it is not completely clear whether they function as afferent and/
or efferent vessels, the fact that the tertiary lymphoid tissues From Chronic Inflammation to Organized Lymphoid
respond to S1P inhibitors suggest that their lymphatic vessels Microenvironments
function similarly to those in lymph nodes.323 Data generated from analyzing the cellular and molecular re-
Lymphoid neogenesis has been noted in humans in au- quirements for secondary lymphoid organ development have
toimmunity, microbial infection, and chronic allograft provided a paradigm for understanding the development of
rejection. A few examples are shown in Table 3.2 and are tertiary lymphoid tissues in chronic inflammation. This par-
described in more detail in Drayton et al.1 These accumu- adigm proposes that the processes and molecules governing
lations also occur in atherosclerotic plaques with FDCs, secondary lymphoid organs are also the basis of tertiary lym-
organized B-cell follicles, HEVs (HECA-452), and CCL19 phoid tissue development,2 and informs understanding of
and CCL21 in addition to those chemokines more often both secondary and tertiary lymphoid tissues. The physiologic
associated with acute inflammation.324 Tertiary lymphoid event(s) that precipitate lymphoid neogenesis remain unclear.
tissues have also been noted in non–small-cell lung cancer325 Data obtained from experiments in knockout and transgenic

TABLE 3.2 Lymphoid Neogenesis in Human Autoimmunity, Infectious Diseases, and Graft Rejection
Disease Affected Tissue Characteristics
Autoimmunity
Rheumatoid arthritis Synovial membrane T cells and B cells, plasma cells, GCs, FDCs, CXCL13, CCL21,
HEVs (PNAd, HEC-6ST)
Sjögren syndrome Salivary glands T cells and B cells, plasma cells, GCs, FDCs, CCL21, CXCL12,
CXCL13 HEVs (PNAd)
Myasthenia gravis Thymus T cells and B cells, GCs, FDCs
Hashimoto thyroiditis Thyroid T cells and B cells, GCs, FDCs CCL21, CXCL13, CXCL12,
plasma cells, HEVs (PNAd)
Grave disease Thymus T cells and B cells, GCs, FDCs, CCL21 CXCL13, CXCL12,
HEVs (PNAd)
Multiple sclerosis Brain Lymphatic capillaries, B-cell follicles and centroblasts, GCs,
CCL19, CCL21, CXCL12, CXCL13
Ulcerative colitis Colon CXCL13
Inflammatory bowel disease Bowel T cell-B compartments, LVs, HECA-452+ HEV
(Crohn disease)
Psoriatic arthritis Joint T cells and B cells, CXCL13, CCL19, CCL21, HEVs (PNAd)
Infectious Diseases
Borrelia burgdorferi/Lyme disease Joints T and B cells, FDCs, HEVs
Borrelia burgdorferi/ Central nervous system/ CXCL13
neuroborreliosis cerebrospinal fluid
Hepatitis C virus Liver T-cell and B-cell compartments, MAdCAM-1
Bartonella henselae/cat Granuloma CXCL13
scratch disease
Graft Rejection
Organ
Heart GCs
Kidney GCs, LVs
FDC, follicular dendritic cell; GC, germinal center; HEV, high endothelial venule; LV, lymphatic vessel; MAdCAM-1, mucosal addressin cell adhesion molecule; PNAd, peripheral
node addressin. Original references are in Drayton et al.1
mice and clinical observations indicate that cooperative activi- some tertiary lymphoid tissues are accompanied by tissue
ties of TNF/LT family members and the lymphoid chemokines damage. However, it is likely that tertiary lymphoid tissues
play central roles in this process. in chronic inflammation have roles in addition to tissue de-
Inflammation is a localized response to tissue injury, irri- struction. In the case of microbial infection, it is likely that
tation, or infection often marked by tissue damage. Acute in- lymphoid neogenesis occurs as a way to sequester pathogens
flammation is an early innate immune response that is gener- and to prevent their access to the other parts of the body.
ally short-lived and self-limiting. However, in some situations, This may represent a primitive form of immunity. Local an-
acute inflammation transitions to a chronic inflammatory re- tigen presentation within the tertiary lymphoid tissue itself
sponse that is long-lived and self-perpetuating. Tertiary lym- likely prevents bacteremia or viremia. Enhanced long-term
phoid tissues arise under conditions of constitutive cytokine survival has been noted for those individuals whose lung or
and/or chemokine expression, but the precise signal(s) that breast tumors included tertiary lymphoid tissues.325,326 The
initiates their development is unknown. By integrating studies propensity for tertiary lymphoid tissues to develop into lym-
of lymphoid neogenesis in human pathologies and in animal phomas, as noted previously, the ability of lymphatic vessels
models, it is becoming clear that at least three critical events to serve as conduits for tumor metastasis, and the ability of
promote tertiary lymphoid tissue formation: inflammatory (eg, tertiary lymphoid tissues to serve as sites of prion accumula-
TNF/LT) cytokine expression, lymphoid chemokine production are obvious manifestations of detrimental functions.342
tion by stromal cells, and HEV development. It is not known if Furthermore, the development of tertiary lymphoid tissues
lymphoid tissue–inducer cells are necessary for lymphoid neo- in autoimmunity may perpetuate clinical disease through
genesis, though such cells have been noted in the mouse models epitope spreading.
of lymphoid neogenesis,341 and lymphoid tissue–inducer-like Data from human and mouse studies provide compel-
cells have been described in several instances. ling evidence that tertiary lymphoid tissues are permissive
microenvironments for the induction of antigen-specific
Functions of Tertiary Lymphoid Tissues immune responses. Extensive immunohistochemical analy-
Ectopic accumulation of lymphoid cells has been consid- ses of tertiary lymphoid tissues in autoimmunity and other
ered the hallmark of destructive inflammation. Indeed, chronic inflammatory states have established the presence

of germinal centers and FDC networks in these tissues; autoreactive T cells and therefore is the result of the presen-
several groups have demonstrated that tertiary lymphoid tation of new antigens.353 Data generated in murine mod-
tissue germinal centers can support B-cell differentiation. els of central nervous system inflammation support the
Microdissection of discrete lymphocytic foci and subse- possibility that intermolecular and intramolecular epitope
quent deoxyribonucleic acid sequence analysis of germinal spreading occur in tertiary lymphoid tissues in the central
center B cells from the inflamed synovial tissue of patients nervous system.350 More recently, it has become apparent
with rheumatoid arthritis revealed a restricted number of that Tregs can populate tertiary lymphoid tissues,338 again
Vκ gene rearrangements, a result consistent with oligoclonal suggesting that manipulating the immune response at the
B-cell expansion in the synovial tissue.343 Somatic hyper- local site is an avenue of control.
mutation is apparent in synovial germinal center B cells.344
Furthermore, synovial B cells exhibit a limited number of Plasticity and Adaptability of Tertiary Lymphoid Tissues
heavy and light chain gene rearrangements consistent with Tertiary lymphoid tissues are the most plastic and adaptable
local clonal expansion of these cells. The molecular analysis of the lymphoid tissues. First, lymphoid neogenesis can be in-
of tertiary lymphoid tissue germinal centers from the sali- duced by a variety of stimuli. Their nimbleness in this regard
vary glands of patients with primary Sjögren syndrome or suggests that they might represent the most primitive tis-
the thymus of patients with myasthenia gravis demonstrates sues in the immune system. Tertiary lymphoid tissues can be
oligoclonal B-cell proliferation in these tissues in addition to “turned off” (ie, resolve) upon removal of the initial stimulus
somatic hypermutation of Ig variable genes.345–347 Together, or after therapeutic intervention. The destruction of the islets
these studies indicate that tertiary lymphoid tissue germinal of Langerhans β cells in type I diabetes mellitus is an example
centers in several autoimmune pathologies can support an- of a situation in which removal of the antigen stimulus is ac-
tigen-driven clonal expansion and extensive diversification. companied by tertiary lymphoid tissue resolution. Antibiotic
Another important hallmark of antigen-driven B-cell treatment results in the resolution of tertiary lymphoid tissues
responses is the terminal differentiation of activated B cells and even MALT lymphomas.354 Treatment has been shown to
into Ig-secreting plasma cells. Plasma cells have been detected resolve some established tertiary lymphoid tissues, reversing
in tertiary lymphoid tissues associated with germinal centers insulitis and protecting against diabetes in nonobese diabetic
in rheumatoid arthritis348 and in Sjögren syndrome. Mice ex- mice.355 Such treatment can also “turn off” established ter-
pressing LTα under the control of the rat insulin promoter tiary lymphoid tissues in a mouse model of collagen-induced
(RIPLTα) exhibit tertiary lymphoid tissue at the sites of trans- arthritis.356 These studies are similar to those described previ-
gene expression (pancreas, kidney, and skin). After immuni- ously regarding the plasticity of lymph nodes after mice are
zation with sheep red blood cells, evidence of isotype switch- immunized or treated with LTβR-Ig,198,199 again emphasizing
ing is apparent in these cellular infiltrates.2 Although the the commonality of these tissues.
presence of plasma cells in tertiary lymphoid tissues is consis-
tent with local antigen presentation, it is occasionally unclear CONCLUSION
whether these cells develop in the tertiary lymphoid tissues The immune system depends on a remarkable organization of
themselves or migrate from canonical secondary lymphoid tissues and cells. The organs have defined functions that in-
tissues. Nonetheless, taken together, these studies indicate clude generation of an immune repertoire (primary lymphoid
that tertiary lymphoid tissues in several human pathologies organs) and responding to antigen (secondary lymphoid or-
and animal models support antigen-driven B-cell differentia- gans and tertiary lymphoid tissues). Development of primary
tion marked by somatic hypermutation of Ig variable genes, and secondary lymphoid organs depends on precisely regu-
affinity maturation, isotype switching, and terminal differen- lated expression of cytokines, chemokines, and adhesion mol-
tiation into antibody-secreting plasma cells. ecules. Similar signals regulate the transition from inflamma-
T-cell priming occurs in tertiary lymphoid tissues as tion to tertiary lymphoid tissues. Chemokines and adhesion
suggested by the presence of isotype switched plasma cells molecules regulate trafficking in and out of lymphoid organs.
in the RIPLTα tertiary lymphoid tissues,2 accelerated graft Secondary lymphoid organs are remarkably plastic in their
rejection,349 and T-cell epitope spreading in the central ner- response to antigenic assault and adapt with changes in ex-
vous system during experimental allergic encephalomyeli- pression of chemokines and adhesion molecules to maximize
tis.350 The restricted TCR repertoire in a melanoma-associ- encounter of antigen with antigen-specific cells. Tertiary
ated tertiary lymphoid tissue351 further supports the concept lymphoid tissues, characteristic of many pathologic states,
that tertiary lymphoid tissues can act as priming sites. The may actually represent the most primitive form of lymphoid
demonstration of naïve T-cell proliferation in the islets of tissues in their even greater plasticity and ability to develop
nonobese diabetic mice after surgical removal of pancre- directly at the site of antigen exposure.
atic lymph nodes352 suggests, together with evidence noted
previously, that tertiary lymphoid tissues present antigen
to naïve cells at the local site and generate an immune re- ACKNOWLEDGMENTS
sponse. Determinant or epitope spreading, a phenomenon The authors thank Myriam Hill for assistance in figure
that arises in several autoimmune diseases, occurs when epi- preparation. This work was supported National Institutes
topes other than the inducing antigen become major targets of Health grants CA16885, DK57731, and HL098711 (NHR),
of an ongoing immune response. It is considered to occur a fellowship from Leukemia and Lymphoma (LAT) and a
subsequent to the tissue damage induced by the initiating Richard A. Gershon Fellowship (NAG).

CHAPTER
4 Evolution of the Immune System
Martin F. Flajnik • Louis Du Pasquier
INTRODUCTION GENERAL PRINCIPLES OF IMMUNE

Defense mechanisms are found in all living things, even bac- SYSTEM EVOLUTION
teria, where they are surprisingly elaborate. Although new The Common Ancestor Hypothesis
adaptive or adaptive-like (somatically generated) immune Figure 4.1 displays the extant animal phyla ranging from the
systems have been discovered in invertebrates and the jaw- single-celled protozoa to the metazoan protostome and deu-
less fish, adaptive immunity based upon immunoglobulin terostome lineages. It is often suggested or assumed by those
(Ig), T-cell receptors (TCRs), and the major histocompat- unfamiliar with thinking in evolutionary terms that mol-
ibility complex (MHC) is only present in jawed vertebrates ecules or mechanisms found in living protostomes, like the
(gnathostomes); because of clonal selection of lympho- well-studied arthropod Drosophila, are ancestral to similar
cytes, positive and negative selection in the thymus, MHC- molecules/mechanisms in mouse and human. While this is
regulated initiation of all adaptive responses, etc., the major true in some cases, one must realize that Drosophila and hu-
elements of the adaptive immune system in gnathostomes mans have taken just as long (over 900 million years) to evolve
are locked in a coevolving unit that arose in concert over from a common ancestral triploblastic coelomate (an animal
a short period of evolutionary time.1–3 In addition, a large with three germ layers and a mesoderm-lined body cavity, fea-
cast of supporting players, including a large array of cyto- tures shared by protostomes and deuterostomes) that looked
kines and chemokines, adhesion molecules, costimulatory nothing like a fly or a human, and thus Drosophila is not our
molecules, and well-defined primary and secondary lym- ancestor (ie, the manner by which flies and humans utilize
phoid tissues, evolved in the jawed vertebrates as well. This certain families of defense molecules may be quite different
scheme was superimposed onto an innate system inherited and both may be disparate from the common ancestor). Thus,
from invertebrates, from which many innate molecules and understanding of how model invertebrates and vertebrates
mechanisms were coopted for the initial phase of the adap- perform certain immune tasks is an important first step in
tive response and others for effector mechanisms at the our understanding of a particular mechanism, but we only
completion of adaptive responses. Over the last 10 years, we deduce what is primordial or derived when we have examined
have learned that various components of the innate immune similar immune mechanisms/molecules in species from a
system are also incredibly complex and locked as well in a wide range of phyla. We will touch upon each of the defense
coevolving unit.4,5 molecule families and will emphasize those which have been
In each group of organisms, one can detect a basic set conserved evolutionarily and those that have evolved rapidly.
of immune functions, and these are employed in differ-
ent ways in representative species. For example, we detect
that in the jawed vertebrates fine tuning, or adaptations, Rapid Evolution of Defense Mechanisms
or even degeneration of molecules/mechanisms in each Immune systems are often compared to the Red Queen in
group (Taxon) are observed, and not a steady progression Alice in Wonderland, (Red Queen’s Hypothesis7,8) who must
from fish to mammals as is documented for most other continually keep moving just to avoid falling behind. Because
physiologic systems.6 Given that all the canonical adap- of the perpetual conflict with pathogens, the immune system
tive immune system features are present in cartilaginous is in constant flux. This is exemplified by great differences in
fish and apparently none were lost (except in particular the immune systems of animals that are even within the same
groups of organisms that will be discussed), differential phylogenetic group (eg, mosquito [Anopheles] and fruitfly
utilization of defense molecules rather than sequential [Drosophila], both arthropods, or human and mouse, both
installation of new elements is observed. In this chapter, mammals). In fact, in contrast to what was believed previ-
there are only isolated cases of increasing complexity in ously, defense mechanisms are extremely diverse throughout
immune systems, but many examples of contractions/ the invertebrate phyla, and Kepler et al. have aptly and suc-
expansions of existing gene families; thus, “more or less cinctly described the situation in the title of a recent review:
of the same” rather than “more and more new features” “not homogeneous, not simple, not well understood.”9 In the
is the rule. We observe a bush growing from a short jawed vertebrate (or gnathostomes), the most rapidly evolving
stem rather than a tall tree with well-defined branches, system is an innate system, natural killer (NK) cell recogni-
and thus deducing the primitive, ancestral traits is not tion, governed by different classes (superfamilies) of recep-
clear cut. tors in mice and humans, and also extremely plastic even
67

650 MYA
Jawed vertebrates
Sharks Mammals LRR, TLR, NLR Cʹ, SRCR, TNF, IL-17, IgSF, NK, lectin, AMP, IgSF V-C1,
(58,000) Ig/TCR, RAG1/2, APOBEC/AID, MHC I/II, somatic DNA, Rel, TIR , PGRP
Jawless vertebrates LRR, TLR, lectins, Cʹ, IgSF V-C, APOBEC/AID, somatic DNA, VLR, Rel, TIR ,
Lamprey and Hagfish
(60) SRCR, IL-17, IL-8
Cephalochordates
900 MYA
Deuterostomes
Amphioxus LRR, TLR, lectins, Cʹ/TEP, PPO, IgSF V, VCBP, TNF, Rel, TIR , B1-3GRP
(25)
Urochordates Ciona, Botryllus LRR, TLR, lectins, PPO, TNF, Cʹ/TEP, IgSF V-C1, NK-like, "MHC", Rel, TIR ,
(3,000) B1-3GRP
Echinoderms Sea urchins

LRR, TLR (many!), Toll, NLR (many!), Cʹ/TEP, TNF, IL17 (many!),
(7,000) SRCR (many!), PGRP, β1-3GRP, lectin, PPO, IgSF V, RAG1/2, 185/333,
Rel, TIR , B1-3GRP
950 MYA
Platyhelmiths
Bilateria
Flatworms
(20,000)
Annelids Earthworms Hemolysins, TNF(?), NK-like, PPO, B1-3GRP

(9,000)
Mollusks Snails, oysters PPO, lectins, AMP, IgSF, MDM, FREP, somatic DNA, Rel, TIR , IL-17,
(70,000) B1-3GRP
Nemerteans Ribbon worms

(1,000)
Protostomes
Sipunculids Peanut worms NK-like

(250)
1300 MYA
Nematodes C. elegans Toll (but not immune), IgSF, lectins, AMP, PGRP, B1-3GRP
(20,000)
Arthropods Flies, mosquitos, LRR, Toll, lectins, PPO, IgSF, AMP, Cʹ/TEP, PGRP, GNBP, penaeidin,
(770,000) shrimps DSCAM (extensive diversity via splicing), Dicer, JAK/STAT, Rel, TIR
Cnidaria
Corals, hydra Cʹ, PPO, “MHC”, B1-3GRP, Toll, Rel, TIR , IL-1R-like
(10,000)
Porifera Sponges IL-1R-like, IgSF, lectins, PPO, “MHC”, Rel, TIR , B1-3GRP
(7,000)
1600 MYA
Protozoa
(50,000)
Plants LRR (NBS-LRR), TIR, convergent innate pathways

(250,000)
FIG. 4.1. Major Animal Groups and Immune Mechanisms/Molecules Described to Date in Each Group. The first box on the left in each row
describes the animal taxon and the approximate number of species in that group. The next box shows specific examples of species or sub-
groups. The third box lists molecules/mechanisms found in each group: underlined terms indicate somatic changes to antigen receptors or
secreted molecules. Figure modified from Hibino et al.48 and Flajnik and Du Pasquier.61 See Table 4.1 for definition of the acronyms.
within the same species, exemplified by the large number of of defense families and mechanisms. Klein11a has compared
killer immunoglobulin superfamily (KIR) haplotypes found this dichotomy to the two-headed god Janus, the major idea
in humans.10 Studies of Ig superfamily (SF) genes expressed being that certain basic mechanisms/functions are obliga-
in the nervous system and immune system showed defini- tory for immune systems to function, but they still must
tively that the immune system molecules evolve at a faster evolve rapidly to avoid pathogen subterfuge. For example,
rate.11 Finally, rapid evolution of immune system molecules MHC class I molecules have similar structure/function/
and mechanisms is the general rule, but molecules function- features in all gnathostomes, but even within groups of pri-
ing at different levels of immune defense (recognition, sig- mates class I genes are not orthologous (ie, they can be de-
naling, or effector) can evolve at widely varying rates. rived from totally different ancestral class I genes10). So, the
idea is to preserve vital immune functions but rapidly mod-
ify the gene or pathway to outwit the pathogen. Additionally,
Conservation of Defense Mechanisms certain features are conserved (eg, development and func-
While the immune system is the most rapidly evolving phys- tion of conventional αβ T cells), but a second, similar sys-
iologic system, nevertheless there is also deep conservation tem, can be exploited in very different ways in closely related

CHAPTER 4 EVOLUTION OF THE IMMUNE SYSTEM | 69
species (eg, the function[s] of γδ T cells). The “Janus para- response as one can track the emergence of new immune
digm,” therefore, can be quite useful when examining any mechanisms, as well as fine tuning of old ones, by exam-
pathway in the immune system. ining the paralogous syntenic groups of genes. Our view is
that these genomewide duplications were as crucial as the
“RAG transposon”18–20 in the development of the Ig/TCR/
Convergent Evolution MHC-based adaptive immunity. In addition, the common
Early on in the comparative study of immunity, it was as- ancestor of bony fish (teleosts) underwent a third round of
sumed that the same features appearing in different taxa genomewide duplication, which many believe to have played
proved that they were present in the common ancestor as well the major role in these fishes’ unique outlier status regarding
(ie, they were submitted to divergent evolution). While this immune system genetics and physiology.2,21
dictum still holds true and establishes one of the dogmas of
comparative immunology, later we discovered, because of the
aforementioned rapid evolution of immune systems, conver- Polymorphism
gence of similar functions has occurred in evolution (ie, the In addition to gene duplications, polymorphism also aug-
same function or even molecular conformation has arisen ments the diversity of immune recognition within a popu-
independently in different organisms, sometimes in species lation. It can be generated any time during the history of a
that are relatively closely related). While we will discuss sev- gene family of either receptor or effector molecules: MHC,
eral cases of convergent evolution throughout the chapter, for toll-like receptors (TLRs), Ig, TCRs, NK receptors (NKRs)
frame of reference, the NK cells, which use different recep- and related molecules, and antimicrobial peptides (AMPs)
tor families in primates and rodents to achieve precisely the are just a few examples. Polymorphism, either within the
same ends of recognizing polymorphic MHC class I mole- gene itself or in its regulatory elements, provides popula-
cules, is a striking example of convergence.12 Additionally, the tions with flexibility in function of the changing pathogenic
emergence of a lymphocyte-based somatic generation of two environment. This subject, central to the studies on MHC,
entirely different receptor families for the same function in leukocyte receptor complex (LRC), and NK cell complex
jawless and jawed vertebrates is another remarkable illustra- (NKC), is becoming well documented for immunity-related
tion of convergent evolution.13 Finally, in innate immunity, genes in insects as well. In Drosophila, polymorphism in
the cytosolic nucleotide-binding domain leucine-rich repeat regulatory networks is indeed expected as parasites often
(NLR) proteins, despite their striking similarity in deutero- target their elements.22 In humans, there are two major
stomes and plants, arose (at least) twice in evolution.14 NK cell haplotypes found in all subpopulations, which are
under “balancing selection.” In such a case, the polymor-
phisms presumably adopt a division of labor required for the
Multigene Families maintenance of the species: one haplotype is believed to be
Genes involved in immunity are often found in clusters, involved in protection from virus and the other perhaps for
with extensive contraction and expansion via so-called birth promoting reproduction.23
and death processes.15 It is well known that such gene clus-
ters can change rapidly over evolutionary time due to un-
equal recombination crossovers and gene conversion (and Somatic Generation of Diversity
not only in the immune system, but in any cis-duplicating Somatic modifications can take place at multiple levels to
gene family). Often, families of related immune genes— generate immune system diversity. Long believed to be the
especially those involved in recognition events—are found sole domain of jawed vertebrates, modifications at the de-
near the telomeres of chromosomes, presumably because oxyribonucleic acid (DNA) level via somatic hypermutation,
this further promotes gene-shuffling events. Nonclassical gene conversion, and rearrangement (primary and second-
MHC class I loci, NK receptors, and NLRs are conspicuous ary [eg, receptor editing]) irreversibly modify genes within
examples of this phenomenon, again believed to be a con- an individual. The well known V(D)J joining, class switch
sequence of the race against pathogens. We will discuss recombination (CSR), and somatic hypermutation (SHM)
many examples of how such multigene families have been are examples of this processes in the IgSF receptors of jawed
exploited in different species throughout the chapter. vertebrates, but modifications to genomic DNA can also
Gene duplication, either in the clusters mentioned above occur in the jawless fish and some invertebrates.7,13 The list
or as a consequence of en bloc duplications, certainly has of organisms undergoing such diversity of germline immune
been a major feature of immune system diversity and plas- genes will only grow as more organisms are examined and
ticity. The two types of duplications are not equivalent, the more genome and expressed sequence tags (EST) sequenc-
former being more taxon-specific and the latter (en bloc) ing projects are undertaken (see Fig. 4.1).
having a lasting impact on the entire system. It is now uni- Alternative splicing can be a source of tremendous diver-
versally accepted that two genome-wide duplications (the sity in some gene families encoding receptors involved in
so-called 2R hypothesis2,16 ; see Fig. 4.13) occurred early in immunity in insects and crustaceans. The Down syndrome
vertebrate history, tracking very well with the emergence of cell adhesion molecule (DSCAM) gene in several arthropods
the Ig/TCR/MHC-based adaptive immune system.2,17 This (described in detail in the following) was shown to gener-
theory forms the basis for much that will be discussed con- ate enormous diversity via ribonucleic acid (RNA) process-
cerning the evolution of the vertebrate adaptive immune ing.7,24 In the vertebrates, this mechanism is important in

determining the function of different molecules, best known TNF family, and certain other domains in immune recogni-
for the Igs (transmembrane [TM] versus secreted forms, tion (eg, scavenger receptor cysteine-rich [SRCR]). All of the
as well as inflammatory versus neutralizing forms in non- domains discussed in this chapter are found in Table 4.1 and
mammalian vertebrates). Further diversity can be obtained a few are displayed in Figure 4.2. As a means of introduction,
by the assembly of multichain receptors in which different two of these families, which constitute the “top two” quan-
components are combined. The classical example in the titatively, will be described in the following.
jawed vertebrate adaptive immune system is that of Ig light
(L) and heavy (H) chains of antibodies but similar combina-
Leucine-Rich Repeats
tion can occur with insect peptidoglycan-recognizing pro-
teins (PGRP), vertebrate TLRs, and many others. LRRs consist of 2 to 45 motifs of 20 to 30 amino acids in
The study of the evolution of immunity has resulted in length (XLXXLXLXXNXHXXHXXXXFXXLX) that fold
a fundamental appreciation of the heart of immunity, both into an arc shape (see Fig. 4.2).28 Both the concave and con-
innate and adaptive. Especially now, with studies in many vex parts of the domain have been shown to interact with
plants and in both invertebrate and vertebrate animals, we ligands. Molecular modeling suggests that the conserved
can see what features have been conserved and when they pattern LxxLxL is sufficient to impart the characteristic
arose. “Simple” genetic models such as Drosophila and horseshoe curvature to proteins with 20- to 30-residue re-
Candida elegans provide a glimpse into these elemental peats. LRRs are often flanked by cysteine-rich domains.
mechanisms and also allow us to remove the clouds that LRRs occur in proteins ranging from viruses to eukaryotes
surround studies of mouse and human, with so many in- and are found most famously in the toll/TLRs, as well as
terconnected pathways. As mentioned, examination of tyrosine kinase receptors, cell-adhesion molecules, resis-
the well-studied mammalian models in combination with tance (R) factors in plants found at the cell surface, and
studies of invertebrates allows us to deduce the condition in the cytosol, extracellular matrix (ECM)-binding glyco-
of the common ancestor. Interestingly, major pathways proteins (eg, peroxidasin), and are involved in a variety of
of defense known to all immunologists, such as the ones protein-protein interactions: signal transduction, cell ad-
involving TLR, JAK-signal transducer and activator of tran- hesion, DNA repair, recombination, transcription, RNA
scription (STAT), NOTCH, and tumor necrosis factor (TNF) processing, disease resistance, apoptosis, and the immune
pathways, clearly arose in an early animal ancestor and have response. LRR-containing proteins can be associated with
been perpetuated in derived fashions in all major taxa. We a variety of other domains, whether they are extracellular
shall see how these pathways are manipulated in different (LRR associated with IgSF or fibronectin [FN] type III) or
animals, always drawing upon the best-known mammalian intracellular (caterpillar family LRR associated with a va-
model as a foundation (whenever possible). riety of effector domains; see subsequent discussion). In
these chimeric molecules, the LRR moiety is involved in
recognition, most likely due to its extraordinarily malleable
MAJOR GENE FAMILIES INVOLVED IN IMMUNITY structure. There are at least six families of LRR proteins,
Defense molecules can be composed of a very large number characterized by different lengths and consensus sequences
of protein folds, some of which are clearly some used to a large of the repeats.29 Repeats from different LRR subfamilies
extent.25–27 Some of the most common families (Fig. 4.2) are never occur simultaneously and have most probably evolved
IgSF, leucine-rich repeats (LRRs), C-type lectins, and the independently in different organisms.
FIG. 4.2. Major Molecular Families Des-

cribed in the Text and Representatives
TCR of Each Family: Leucine-Rich Repeats
TLR NLR VLR (LRRs), Immunoglobulin Superfamily (IgSF),
(IgSF)
(LRR) (LRR) (LRR) C-type Peptidoglycan-Recognition Protein (PGRP),
PGRP B1-3GRP lectin a1-3 Glucan Recognition Protein (a1-
3GRP), and C-Type Lectins. Representative
structures are shown above for LRR (toll-
like receptor, nucleotide-binding domain
leucine-rich repeat [NLR], variable lympho-
cyte receptor), IgSF, PGRP, β1-3GRP, and
C-type lectin. Other acronyms are defined
in Table 4.1. For the NLR model, the echi-
TIR
noderms have N-terminal death domains,
DD NACHT whereas all other animals have caspase-
CARD recruitment domains. Figure modified from
Hibino et al.48

TABLE 4.1 Molecules and Abbreviations Found Throughout the Text

Acronym/Defense Molecule Full Name Function
AID Activation-induced cytidine deaminase SHM/gene conversion/CSR
APOBEC Apolipoprotein B mRNA editing enzyme Innate immunity (antiviral)
catalytic polypeptide
AGM Aorta/gonad/mesonephros Intraembryonic origin of hematopoietic cells
AMP Antimicrobial peptide Innate immunity (eg, defensins)
APAR Agnathan paired antigen receptor Similarities to Ig/TCR and NKRs
AVR Avirulence protein Pathogen effector recognized by plant NLR
Bf Factor B Enzyme of C′ cascade
B1-3GNP Beta 1-3 glucan-recognizing protein Binds to gram-negative bacteria
C′ Complement Innate/adaptive immunity
CARD Caspase-recruitment domain Domain in intracellular defense molecules
CATERPILLER or CLR CARD, transcription enhancer, R(purine)-binding, Apoptosis/immunity/inflammation
pyrin, lots of leucine repeats
CDR Complementarity-determining region Portion of Ig/TCR that binds to antigen
CSR Class switch recombination Adaptive humoral immunity modification
DD Death domain Cytosolic interacting domain
DSCAM Down syndrome cell adhesion molecule Insect immune (adaptive?) defense and
neuron specification
ECM Extracellular matrix
ETI Effector-triggered immunity Immunity in plants triggered by NLR
FBA F box–associated domain Intracellular domain
FcRN Fc receptor neonatal MHC-like FcR
FN3 Fibronectin type III repeat Domain found in many innate molecules
FREP Fibrinogen-related protein Mollusk (adaptive?) defense
FuHC Fusion histocompatibility Histocompatibility locus in tunicates
GALT Gut-associated lymphoid tissue
GPI Glycophosphatidylinositol Lipid linkage to cell membrane (eg, VLR)
Hemolysin Cell lysis
ICE Interleukin-converting enzyme IL-1β processing
Ig Immunoglobulin Adaptive immunity
IgSF Immunoglobulin superfamily Innate/adaptive immunity
IFN Interferon Innate (type I)/adaptive (type II) immunity
IMD Immune deficiency Insect innate defense
IRF Interferon regulatory factor Innate (transcription factor)
IRG Immunity-related GTPases Innate immunity
ITAM Immunoreceptor tyrosine-based activation motif Signaling motif for NK and antigen receptors
ITIM Immunoreceptor tyrosine-based inhibitory motif Signaling motif for NK and antigen receptors
JAK Janus kinase Signaling molecule associated with cytokine
receptors
KIR Killer IgSF receptor NK cell receptor
lectins For example, galectin, C-type, S-type Many (eg, NKRs, selectins)
LITR Leukocyte immune-type receptors Fish NK-like receptors of the IgSF
LMP Low-molecular-weight protein Proteasome subunit
LRC Leukocyte receptor complex Gene complex containing KIR and many IgSF
molecules
LRR Leucine-rich repeat Innate/adaptive immunity module
MAC Membrane-attack complex C′, pore-forming
MACPF MAC-perforin domain Potential pore former
MASP MBP-associated serine protease Lectin C′ pathway
MBP (or MBL) Mannose-binding protein (lectin) Lectin C′ pathway
MDM Mollusk defense molecule IgSF defense molecule
MHC Major histocompatibility complex T-cell recognition; innate immunity
MIF Macrophage inhibitory factor Innate immunity; inflammation
MyD88 (also dMyD88) (Drosophila) Myeloid differentiation primary TLR adaptor
response gene 88
NITR Novel immune-type receptors Teleost fish NK-like receptors of the IgSF
NK cell Natural killer cell Vertebrate innate cellular immunity
NKC Natural killer cell complex Gene complex with many C-type lectin
genes (especially NK cells)
NKR Natural killer cell receptor Receptor on NK cells
(continued)

TABLE 4.1 Molecules and Abbreviations Found Throughout the Text (Cont.)
Acronym/Defense Molecule Full Name Function
NALP NACHT leucine-rich repeat and PYD-containing Intracellular PRR
protein
NBD-LRR Nucleotide-binding domain LRR Motif of intracellular defense molecules
NFκB Nuclear factor-κB (Rel homology domain) Evolutionarily conserved transcription factor
NLR NACHT leucine-rich repeat protein Intracellular PRR
NOD Nucleotide oligomerization domain protein Intracellular PRR
NOS Nitric oxide synthase Intracellular killing innate defense molecule
PAMP Pathogen-associated molecular pattern Conserved target epitopes on pathogens
PCD Programmed cell death Many pathways
Penaedins Defense molecule in shrimp
PGRP Peptidoglycan-recognition protein Gram-positive bacteria defense family;
receptor and effector
PPO Propolyphenol oxidase Plant/invertebrate defense (melanization)
PRR Pattern-recognition receptor Recognize PAMP, innate/adaptive immunity
PMSB Proteasome subunit beta subunit Proteolytic member of 20S proteasome
Polμ DNA polymerase μ Error-prone polymerase (related to TdT)
PPO Prophenoloxidase Invertebrate defense molecule
PYD Pyrin domain Domain in intracellular defense molecules
RAG Recombination-activating gene Ig/TCR rearrangement
RFP-Y Restriction fragment polymorphism-Y Chicken nonclassical MHC gene cluster
RFX Regulatory factor X Transcription factor, class I regulation
RIG Retinoic acid-inducible gene Intracellular double-stranded RNA
recognition
RSS Recombination signal sequence DNA element next to Ig/TCR gene segments
necessary for RAG-mediated
rearrangement
RXR Retinoid X receptor Transcription factor encoded in MHC
SHM Somatic hypermutation Adaptive humoral immunity
SPE Spaezle-processing enzyme Insect defense molecule in toll cascade
SRCR Scavenger receptor cysteine-rich Innate immunity recognition molecule
TAK TGF-β activated kinase ubiquitin-dependent kinase of innate
pathways
TAP (and TAP-L) Transporter associated with antigen processing Rransports peptides from cytosol to ER lumen
TAPBP TAP-binding protein Tethers TAP to class I
TCR T-cell receptor Adaptive defense
TdT Terminal deoxynucleotidyl transferase Involved in Ig/TCR rearrangement
TEP Thioester-containing protein Opsonization (like C3)
TGF Transforming growth factor Immunosuppressive cytokine
TNF Tumor necrosis factor
UPD Unpaired Protostome cytokine induced by viral
infection
TLR Toll-like receptor Innate receptor on the cell surface or in
endosomes
TM Transmembrane
TNF Tumor necrosis factor Proinflammatory cytokine (and family)
TRIM Tripartite motif-containing proteins Large family of cytosolic innate defense
molecules
V-, C1-, C2-, I- Variable, constant 1 and 2, intermediate IgSF domain types
IgSF domain
VAV Guanine exchange factor, the “onc F” Encoded in MHC, involved in adaptive
proto-oncogene signaling pathways
VCBP Variable domain chitin binding Amphioxus defense molecule
VLR Variable lymphocyte receptor Agnathan adaptive defense molecule
WKRY Plant transcription factor used to upregulate
defense genes (analog of NF-κB)
XMIV Xenopus MHC-linked IgSF V region Xenopus MHC-linked NKR-like genes
XNC Xenopus nonclassical Xenopus class Ib cluster
185/333 Sea urchin defense molecule (Adaptive?) Defense
DNA, deoxyribonucleic acid; ER, endoplasmic reticulim; IL, interleukin; mRNA, messenger ribonucleic acid; RNA, ribonucleic acid.

LRR-containing proteins are involved in immunity from and G strands, so that in Igs the CDR3 are in the center of
plants to animals. The functions in the immune systems the binding site; for C domains, the other beta strand bear-
range from control of motility of hemocytes and lympho- ing the A, B, D, and E strands forms the interface.
cytes30 to specific recognition of antigens via a novel system The binding capacities of V domains in molecules besides
of gene rearrangement (the variable lymphocyte receptors Ig/TCR can reside in different areas of the molecule (inter-
[VLR] described in the following; see Fig. 4.9). LRRs can strand loops, A-A′ strand, F strand), while in Ig/TCR, CD8,
occur in soluble forms, the ECM, in the cytosol, or as TM and certain NKR, these regions are the targets for variation
forms, either integral membrane proteins or glycophospha- in shape and charge. The binding capacities can be modu-
tidylinositol (GPI)-anchored. The bottom line is that be- lated whether one domain acts as a single receptor unit (eg,
cause of its basic structure and malleability, the LRR module IgNAR) or whether it is associated with a contiguous domain
was locked in early in evolution as an ideal motif for recog- (eg, KIR, variable domain chitin-binding protein [VCBP])
nition of essentially any ligand. or with another polypeptide chain (eg, TCR, Ig). In the case
of a dimer, the binding capacity can again be modulated by
the presence or the absence in the G strand of a diglycine
Immunoglobulin Superfamily bulge, which can modify the space between the faces of the
IgSF domains are encountered in a very large number of mol- Ig domain. In several cases, the sites responsible for binding
ecules in the animal kingdom (see Fig. 4.2).31 They are found are known (KIR, Ig, TCR); in many other cases, they are not
intracellularly (eg, connectin) or as cell adhesion molecules, known but inferred from crystal structures and/or variabil-
many of which are in the nervous system (eg, the neural cell ity plots (leukocyte immune-type receptor [LITR], chicken
adhesion molecule, NCAM), coreceptors and costimulatory Ig-like receptor [CHIR], triggering receptor expressed on
molecules of the immune system (eg, cluster of differentia- myeloid cells [TREMs], DSCAM, hemolin).
tion [CD]79, CD80), molecules involved in antigen presen-
tation to lymphocytes (eg, class I molecules), certain classes
HEMATOPOIEIS AND CELL TYPES IN THE
cytokine receptors (eg, interleukin [IL]-1R), and of course
Ig (and TCR), where they were first characterized and were
INVERTEBRATES
bestowed with their name (Ig). They can be associated with Invertebrate Cell Types
other domains such as FN (eg, titin and FREP) and LRRs,32 Examples of conservation of fundamental mechanisms of ge-
or they can be the sole constitutive elements of the polypep- netic control of developmental pathway between protostomes
tide chain often associated to a transmembrane segment and and deuterostomes, even in the absence of homology of the
a cytoplasmic tail (or GPI-linked). The β barrel IgSF struc- cells or organ considered, are accumulating: the organization
ture was adopted independently in other families such as and expression of the homeotic gene clusters and eye forma-
cadherins, calycins, lipocalin, etc., and the super (or über) tion through the function of a complex of proteins includ-
family has hundreds of members and has been selected for ing Pax-6.34 The cell types involved, besides direct interaction
several different functions. These functions are somehow with the external layer of cells on the skin, or external tegu-
related, almost all involved in protein-ligand interactions. ments, have been specialized cells of mesodermal origin de-
The vertebrate lymphocyte surface can express 30 different voted to defense. This is true for all coelomates where effector
IgSF members simultaneously. cells have been identified, but recent data have shown that
IgSF domains are commonly classified according to dif- cnidarian diploblastic organisms that lack mesoderm also
ferent domain constitution in their β strands and loops.31,33 have many of the same genetic systems as the coelomates35
All conform to the stable shape of a β barrel consisting of (see Fig. 4.1). The cells can be circulating or sessile, and often
two interfacing β sheets, usually linked by a disulfide bridge. are found associated with the gut. Several morphologically
There are three types of domains: variable (V), and two distinct hemocyte types in insects cooperate in immune re-
types of constant (C1 and C2); the so-called I set domain is sponses: they attach to invading organisms and isolate them,
intermediate between the C1 and C2. The V domain is most trapping larger organisms in nodules or forming large multi-
complex with more strands (C′ and C″), which make up cellular capsules around them. Indirect evidence for the role
complementarity determining region (CDR) 2 in conven- of hemocytes in immune responses can be derived by con-
tional Igs and TCRs. C1 domains lack these strands entirely, trasting properties of such cells in healthy and parasitized an-
and C2/I domains have varying sizes in the C′/C″ region. V imals (ie, modifications in adherence and opsonic activity).
domains, either alone (eg, the new antigen receptor [NAR]) All animals show heterogeneity of the free circulating
or in association with another V domain (eg, Ig H/L), rec- cells, generically called hemocytes (arthropods), coelomo-
ognize the antigenic epitope and are therefore the most im- cytes, amebocytes (annelids, mollusks, and echinoderms),
portant elements for recognition. Domains with the typical or leukocytes (sipunculids). However, the repertoire of in-
V fold, whether belonging to the true V-set or the I-set, have sect “blood cells” is clearly less heterogeneous than that of
been found from sponges to insects (eg, amalgam, lachesinm vertebrates. Basically, three or four types of cell lineages
and fascicilin) and even in bacteria. The mollusk fibrinogen- can be identified in Drosophila (Fig. 4.336): plasmatacyte,
related proteins (FREPs, described in the following) have crystal cells, and lamellocytes, and an equivalent number
one or two V-like domains at their distal end, associated in Lepidoptera (butterflies). The functional roles they play
with a fibrinogen-like domain. For V domains, the interface consist of immune defense, disposing of apoptotic and
between dimers is the beta strand bearing the C, C′, C″, F, other debris, contributing to the ECM, and modeling of the

Cytokines
(UPD)
PGRP-LC
Serine
proteases/ Toll Domeless
serpins
Hemolymph Imd
PRRs dMyd88 JAK/STAT
Lymph glands (PGRP/
β1-3GRP)
Prohemocyte Fat body
Lamellocyte Plasmatocyte Crystal cell

– AMP
– Serine proteases
– TEPs
ENCAPSULATION PHAGOCYTOSIS COAGULATION MELANIZATION – Clotting factors
– DSCAM
FIG. 4.3. Types of Immune Responses and Cells in Insects, with Drosophila as the Prototype. Secreted defense molecules are made by fat
body cells in response to pathogens, which either act as direct effector molecules or feed back on hemocytes to stimulate their defense func-
tions. Unpaired (UPD) is produced after virus infection, stimulating defense molecule upregulation via the JAK/signal transducer and activator
of transcription pathway. Not shown is the RNAi pathway, also induced upon virus infection. Details on stimulation of the toll and immune
deficiency pathways are found in Figure 4.4. This figure was modified from Lemaitre and Hoffmann.36
nervous system. The immunity role encompasses phagocy- and involved in wound healing, whereas the ones de-
tosis, encapsulation, and sometimes production of effector rived from the splanchnopleura participate in immunity.
molecules (see Fig. 4.336). These roles all require recogni- Heterogeneity of annelid coelomocytes is not encountered in
tion of pathogen-associated molecular patterns (PAMPs) or primitive oligochaetes or in hirudinae (leeches). Phagocytic
self-derived defense molecules (ie, opsonization) at the cell coelomocytes show an acid phosphatase activity and a beta
surface.37–40 glucoronidase activity.41 The large coelomocytes and free
Only in a few organisms has the characterization of he- chloragocytes (eleocytes) in the typhlosole of Eisenia foetida
mocyte lineages gone beyond morphologic or basic physi- appear to produce the bacteriolytic and cytolytic factor
ologic functions. Among these free circulating cells are lysenin.42 From electron microscopy studies, macrophage-
always one or more types that can undergo phagocytosis. like cells seem to be involved in graft rejection. In the closely
Different cells participate in encapsulation, pinocytosis, related sipunculid phylum, two main cell types can be iden-
and nodule formation, and can upon stimulation produce tified in the blood: erythrocytes (a rare occurrence in in-
a great variety (within an individual and among species) vertebrates) and granular leukocytes. The latter are capable
of soluble effector molecules that may eliminate the patho- of cytotoxicity and even have dense granules reminiscent of
gen. In an attempt to integrate all of the data available in vertebrate “NK cells.”43
invertebrates, Hartenstein has proposed a unified nomen- Two developmental series have been described in mol-
clature of four basic types: prohemocytes, hyaline hemo- lusks, the hyaline and granular cells, but cephalopods seem
cytes (plasmatocytes or monocytes), granular hemocytes to have only one lineage. They participate in encapsula-
(granulocytes), and eleocytes (chloragocytes).37 These des- tion, with hemocytes adhering around the foreign body like
ignations will be found in the following description of the Drosophila lamellocytes. Phagocytosis is carried out by the
blood cell types. wandering granular cells, which resemble vertebrate mono-
Earthworm (annelid) coelom-tropic coelomocytes are cytes/macrophages. In oysters, electron microscopy revealed
called eleocytes. They contain glycogen and lipid and are different types of circulating hemocytes, including granular
considered of the same lineage as the chloragocytes involved hemocytes resembling the granulocytes of sipunculus men-
in the production of immune effector molecules such as feti- tioned previously.44,45 In crustaceans, the situation is similar
din or lysenin. The phagocytic cells of annelids are appar- to that in mollusks, with three main populations identified
ently granular “leukocytes” derived from the somatopleura based again on the presence of granules in the cytoplasm.

The hyaline cells are involved in the clotting process and The granular form is likely to be involved in postphagocytic
the granular cells in phagocytosis, encapsulation, and the activity, like in earthworms.49 Adoptive transfer of alloim-
prophenoloxidase (PPO) pathway. The hematopoietic organ munity in the solitary tunicate Styela can be achieved via
is located on the dorsal and dorsolateral regions of the lymphocyte-like cells.
stomach.39 Crustacean hemocytes can now be cultured and In Amphioxus, cells with phagocytic capacity have been
their response to virus can be examined,46 and markers of identified in the coelom with a morphology resembling
the three hematopoietic lineages are available.47 more the phagocytic echinoderm cells than urochordate
In insects, the so-called prohemocytes are believed to blood cells, a fact that is consistent with the new systematic
be stem cells. They are only found in the embryonic head positions of amphioxus and echinoderms.50 Both free cells
mesoderm and the larval lymph glands but not in the he- and the lining of the perivisceral coelom are able to phago-
molymph. However, prohemocytes are frequent in both the cytose bacteria. Cells with the morphologic appearance of
hemolymph and hematopoietic organs of the lepidopteran lymphocytes and expression of lymphocyte-specific genes
Bombyx (silkmoth). Plasmatocytes of Drosophila have a were detected in this species, the earliest identification of
phagocytic function. This type of hemocyte is equiva- such cells in phylogeny.51
lent to the granulocytes of Bombyx, which play a key role
in phagocytosis in larvae. Lamellocytes seem to be unique
to Drosophila, but they are probably the equivalent of the Hematopoiesis in the Invertebrates
lepidopteran plasmacytoid cells. Their precursors reside The history of the hemocytes is associated with that of the
in the larval lymph gland, where they differentiate in re- mesoderm among triploblastic organisms. The bilaterian
sponse to macroscopic pathogens, following a brief phase of ancestor was most likely a small acoelomate or pseudocoe-
mitosis linked to the presence of the pathogens and under lomate worm similar to extant platyhelminths (flatworms)
hormonal control via ecdysone. The transcription factors (see Fig. 4.1). A specialized vascular system or respiratory
(GATA, Friend-of-GATA, and Runx family proteins) and system was probably lacking, although cells specialized for
signal transduction pathways (toll/NF-κ B, Serrate/Notch, transport and excretions were likely present because they
and JAK/STAT) that are required for specification and pro- exist in most extant bilaterian phyla. One can further as-
liferation of blood cells during normal hematopoiesis, as sume that groups of mesoderm cells in the bilaterian ances-
well as during hematopoietic proliferation that accompanies tor could have formed epithelial structures lining internal
immune challenge, have been conserved throughout evolu- tubules or cavities (splanchnopleura). In coelomates, the
tion. The specific differentiation of lamellocytes requires mesoderm transforms into an epithelial sac, the walls of
the transcription factor Collier. The mammalian early which attach to the ectoderm (somatopleura) and the inner
B-cell factor, an ortholog of Collier, is involved in B-cell organs (splanchnopleura). Blood vessels are formed by tubu-
differentiation in mice. The Drosophila crystal cells are re- lar clefts bounded by the splanchnopleura. Excretory neph-
sponsible for melanization through the PPO system (see rocytes are integrated into those vascular walls, which also
subsequent discussion). In silkworm oenocytoids, crystal- gives rise to blood cells circulating within the blood vessels
like inclusions are also found, but they disappear later after (the pronephros of anurans and head kidney of teleost fish
bleeding.36,37,40 are important hematopoietic organs in vertebrates). Thus,
Echinoderm coelomocytes express a diversity of effector further evolutionary changes separated the three systems,
functions, but no studies of lineages have been performed. but there was a close original connection between them.
In echinoderms, the number of different coelomocytes may The origin of hemocytes has been investigated mainly
vary according to the particular family. The sea urchin is in arthropods. When examining principles that govern he-
endowed with at least four cell types, only one of which only matopoietic pathways, similarities have been observed with
is phagocytic and corresponds to the bladder or fi liform vertebrates, raising interesting evolutionary issues.37,40 In
forms. Another type is described as the round vibrating jawed vertebrates, the yolk sac or its equivalent gives rise
cell involved in clotting. Pigment cells (red spherule cells) to blood precursors that are primarily erythroid in nature
have been detected ingesting bacteria; the morphology of (but see the following: recent data suggest that B1 cells and
phagocytic cells can vary enormously, precluding any easy macrophages are also derived from this embryonic tissue).
classification.48 In succession, definitive hematopoiesis occurs in the aorta/
In tunicates, amoeboid cells circulate in the blood and gonad/mesonephros (AGM) region of the embryo, en-
are involved in a large number of processes, such as clotting, compassing all of the different cell types and multipotent
excretion, nutrition budding, and immunity. Large num- progenitors (although this is controversial). Like in the ver-
bers of blood cells are present (average of 107 per mL) in the tebrates, hematopoiesis in insects is biphasic. One phase
blood of ascidians such as Ciona. Hemoblasts are considered occurs in the embryo and the other during larval develop-
to be undifferentiated cells, perhaps the equivalent of the ment. Additionally, these waves occur in distinct locations of
prohemocytes of arthropods or the neoblasts of annelids. the embryonic head mesoderm and the larval lymph gland.
Blood cells in ascidians proliferate in the connective tissue In the early embryo expression of the GATA factor, serpent
next to the atrium. The pharyngeal hematopoietic nodule of (Srp) can be detected in the head mesoderm. This GATA
this animal contains a large number of hyaline and granular family of zinc-finger transcription factors is conserved
cells called “leukocytes” with supposed intermediary forms from yeast to vertebrates where they are involved in various
of differentiation between blast and granular mature types. aspects of hematopoiesis. Blood cell formation in the head

follows Srp expression, whereas in the lymph gland there is of hemocytes upon stimulation is an unresolved issue in the
a long delay between Srp expression and the appearance of invertebrates; clearly, clonal selection resulting in extensive
the lymph gland–derived hemocytes.38 Hematopoiesis in the proliferation is not the rule. The turnover of cell populations
head mesoderm and yolk sac may be related evolutionarily. has been the object of numerous, often unconvincing ex-
A further similarity occurs at the AGM/lymph gland level in periments. Still, new data have emerged, and it is clear that
Drosophila. The lymph gland develops from a part of lateral in several invertebrates, proliferation occurs in certain cell
mesoderm that also gives rise to vascular and excretory cells, types following encounters with pathogens. Very little cell
much like the vertebrate AGM. The conserved relationship proliferation occurs in the circulation of crayfish, but cells
between blood precursors and vascular and excretory sys- in the hematopoietic tissue divide after an injection of the
tems is intriguing. PAMP β1-3-glucan. New cells in the circulation developed
into functional synthetic germinal centers (GCs) and GCs
expressing the proPO transcript. RUNT protein expression
Hematopoiesis and Transcription Factors in the was upregulated prior to release of hemocytes. In contrast,
Vertebrates proPO was expressed in these cells only after their release
As mentioned previously, transcription factors of the into the circulation.54
family PAX 2/5/8; GATA 1, 2, 3; ets/erg; and runt domain– By contrast to the study of transcription factors that
containing factors have been cloned in several invertebrates. regulate hematopoiesis, relatively little is known about cyto-
One plausible model to explain the genesis of true lymphokines that drive hematopoiesis among invertebrates. It was
cytes in vertebrates is that closely related members of tran- reported that differentiation and growth of hematopoietic
scription factor families are the result of a relatively late stem cells in vitro from crayfish required the factor astakine,
divergence in lineage pathways followed by specialization of which contains a prokineticin domain55 ; prokineticins are
duplicated genes.52 These duplications could be those that involved in vertebrate hematopoiesis, another case of con-
apparently occurred during the history of chordates (see servation during the evolution of growth factors and blood
MHC and “Origins” section2). Within deuterostomes, the cell development.
generation of true GATA 2 and 3 probably occurred after Parasitization of Drosophila by the wasp Leptopilina bou-
echinoderms diverged from the chordate branch and the lardi leads to an increase in the number of both lamellocytes
GATA, ets, early B-cell factor, and Pax5-dependent pathways and crystal cells in the Drosophila larval lymph gland. This
of T-/B-cell differentiation are thus specific to vertebrates. is partially due to a limited burst of mitosis, suggesting that
It is already known that lampreys express a member of the both cell division and differentiation of lymph gland hemo-
purine box 1/spleen focus–forming virus integration-B gene cytes are required for encapsulation. In genetic backgrounds
family that is critically and specifically involved in jawed where ecdysone levels are low (ecdysoneless), the encapsula-
vertebrate lymphocyte differentiation. Expression has been tion response is compromised and mitotic amplification is
detected in the gut, which may be related to the fundamen- absent. This ecdysone-dependent regulation of hematopoie-
tal nature of “gut-associated lymphoid tissue (GALT)” as a sis is similar to the role of mammalian steroid hormones such
lymphoid cell–producing organ. as glucocorticoids that regulate transcription and influence
In vertebrates, the generation of T-, B-, and NK lympho- proliferation and differentiation of hematopoietic cells.56
cyte lineages from pluripotent hematopoietic stem cells de-
pends on the early and tissue-specific expression of Ikaros Phagocytosis
(and related loci), which by means of alternative splicing To obtain phagocytosis at the site of microorganism inva-
produces a variety of zinc-finger DNA-binding transcription sion implies recruitment of cells via chemoattraction. In
factors. The orthologs of Ikaros, Aiolos, Helios, and Eos have vertebrates, this can be done by several categories of mol-
been identified in the skate Raja eglanteria, where two of the ecules such as proinflammatory chemokines/cytokines
four Ikaros family members are expressed in their special- or the complement fragments C3a and C5a (as mentioned
ized hematopoietic tissues (epigonal and Leydig’s organs; in the following section, C3a fragments as we know from
see subsequent discussion) like in mammals.52 In lower deu- mammals may be found in tunicates but not other nonver-
terostomes, single genes that seem to be related to the ances- tebrates; yet, C3 may be cleaved in different ways in the in-
tor of the Ikaros and Ets family of transcription factors exist, vertebrates). C3b, mannose-binding lectin (MBL), and many
further suggesting that the division of labor between the other lectins can function as opsonins, and recent studies of
family members in the jawed vertebrates was a result of en the PGRPs, thioester-containing proteins (TEPs), DSCAMs,
bloc duplications.52,53 The conservation of Ikaros structure and eater have added to this repertoire.36 Ingestion follows
and expression reinforces its role as a master switch of he- phagocytosis, and then killing occurs by an oxidative mech-
matopoiesis. We discuss this topic further in the lymphoid anism with the production of reactive oxygen radicals and
tissues section. nitric oxide. These mechanisms are conserved in phylogeny,
and other basic mechanisms are being examined in more de-
tail now in protozoan models.57 Signaling pathways in com-
Responses of Hemocytes mon between vertebrates and the protozoon Dictyostelium
In this section, we simply touch on classical and specific re- include involvement of cyclic AMPs, integrins, and perhaps
sponses in the invertebrates, responses that are more univer- mitogen-activated protein (MAP) kinase cascades. Unique
sal are found in the innate immunity section. Proliferation to all jawed vertebrates studied to date, the activation of

phagocytes also leads to upregulation of the antigen pro- Of the over 1 million described species of animals (see
cessing machinery, costimulatory molecules, and proin- Fig. 4.1), approximately 95% are invertebrates represent-
flammatory cytokines that can enhance adaptive immunity. ing 33 phyla, some with one species (Placozoa, Cycliophora)
and others with over 1 million (Arthropoda). Because they
have major differences in body plans, development, size,
INNATE IMMUNE RESPONSES habitat, etc., wildly different types of immune systems in
Immune responses are often subdivided into recogni- diverse species should be expected. Early studies of inverte-
tion, signaling, and effector phases, which are subjected to brate immunology reached no consensus of how immunity
different pressures, defined by whether orthology is main- should be examined, but because vertebrate cellular adap-
tained and the relative divergence rates of the genes re- tive immunity was often defined (indeed, was discovered for
sponsible for the various phases. Recognition molecules are T cells) through transplantation reactions, attempts to re-
from evolutionarily conserved families, but as described veal specific memory by allograft rejection were often used.
previously, their genes are subjected to rapid duplication/ After many unsuccessful attempts to demonstrate memory
deletion so that orthology is rarely preserved. By contrast, of such responses (see the following) and after extensive
signaling pathways can be conserved (see Fig. 4.5), despite molecular studies, a consensus was reached that an inverte-
the fact that the genes are often divergent in sequence. brate adaptive immune system involving somatic generation
Effector molecules can either be extremely conserved (eg, of antigen receptors and their clonal expression was highly
reactive oxygen intermediates) or extremely divergent to the unlikely. However, the term “innate” is rigid and masks the
point of being species-specific (eg, AMP). Here, we break the possibility of other somatic alterations of invertebrate im-
immune response down into these three phases, beginning mune system molecules, as will be discussed.61,62 We will
with the recognition phase. categorize the molecules based on their location within
Initiation of an immune reaction can theoretically involve the cell.
either the recognition of nonself, altered self, or the absence
of self. Nonself-recognition can take place with receptors
(pattern recognition receptors [PRRs]) that detect PAMPs, Intracellular Recognition
which were originally defined by Janeway and colleagues Nucleotide-Binding Domain Leucine-Rich Repeat (NLR)
as evolutionarily conserved epitopes displayed by patho- One major group of intracellular sensors in animals and
gens but not host cells.58,59 The second mode, altered self, is plants is the NLR family (see Figs. 4.2 and 4.4).14,63 Each of
typified by molecules that are induced in self-cells during the family members has a central NB/NACHT (nucleotide-
infections and recognized by conserved defense molecules, binding domain) and C-terminal LRR used for recognition,
similar to the SOS systems mentioned in the MHC section, and a unique N-terminal domain. The subfamilies are de-
or by peptide presentation on MHC molecules. A third fined by their N-terminal domains, coiled-coil and toll-IL-1
mechanism, “am I still myself,” depends upon recognition receptor (TIR) in plants, and baculoviral inhibitory repeat,
of self-tags and their changes in expression60 (eg, NK recog- caspase-recruitment domain (CARD), pyrin domain (PYD),
nition of self-MHC molecules through KIR and C-type lec- and activation domain in animals (see Fig. 4.4). Thus far, the
tins). These latter two mechanisms have not been described NLRs have been found in deuterostomes but not protostomes
in the invertebrates for immune defense against pathogens, (see Fig. 4.1), which is surprising considering that plants
but it would not be surprising if they were revealed in the have intracellular defense proteins with a similar structure,
future, considering the new features of invertebrate immune and seem to have been derived via convergent evolution.14
systems that have been discovered recently and the usage of
this mode of recognition in many invertebrate histocompat-
ibility systems. PLANT NLR ANIMAL NLR
Whether the invader is related to its host (cells from
individuals of the same species or cells from a parasitoid) CARD
or are very distant from the host (fungi and bacteria in CC PYD
TIR NB-ARC LRR BIR NACHT LRR
metazoa), there are different principles of recognition. Yet
PAMP determinants have been identified on very different
organisms—sugars such as β1-3 glucan of fungi, lipopoly-
saccharide (LPS) and peptidoglycans of bacteria, phospho-
glycan of some parasites, and especially nucleic acids of • Detection of plant-specific • Cooperation with TLR to
affectors (Avr) upregulate proinflammatory
bacteria and viruses—and they can trigger similar cascades • Disease resistance or cytokines of the IL-1 family
of events. The foreign ligand can be bound by a molecule cell death • Pyroptosis (cell death)
in solution that initiates an effector proteolytic cascade (eg, • Upregulation of Class I
clotting or the complement cascade). On the other hand, a (NLRCS) and Class II (CIITA)
proteolytic cascade can be initiated and result in the pro-
FIG. 4.4. Structure and Major Functions of Nucleotide-Binding
duction of a self-ligand that interacts with a cell surface or Domain Leucine-Rich Repeats in Plants and Animals.14 Although
endosomic or cytosolic receptor. In this way, there need not the structures are quite similar, and recognition can be analogous,
be a great diversity of cell surface receptors, especially in the these families seem to have arisen via convergent evolution (see
absence of clonal selection. text for more details).

The specificity of plant NLR depends principally on licase domain, and a regulatory domain. Ligands bind to the
the LRRs, and these are targets for diversifying selection, regulatory domain, inducing a conformational change lead-
as described previously for multigene families. Plant NLR ing to interaction with the adaptor protein MAVS (or IPS-1)
can recognize pathogen effectors (pathogen-derived aviru- and ultimately to the induction of type I interferons (IFNs).
lence factors), or viral and fungal PAMPs directly via the A third member of the RLR family is LPG2, which lacks the
LRR domains, or via modifications of a host target that CARD domain; this molecule was originally believed to be
interacts with the N-terminal domains, altered self if you a negative regulator of RIG-I/MDA5-induced signaling, but
will (see Fig. 4.4). This type of activation in plants is termed that has been called into question.
effector-triggered immunity, which is specific of the NLRs This family is found in all of the vertebrates and in lower
(see Fig. 4.4). A host of downstream effectors are generated, deuterostomes, such as amphioxus and echinoderms.63
some involved in defense but others activating cell death Somewhat surprisingly, the RLR family is only mildly ex-
pathways.14,64 panded in sea urchins (12 members). While there is no re-
Best described in mammals, the nucleotide oligomeriza- port of bonafide RLR family members in protostomes but
tion domain (NOD)/NLR recognizes PAMPs such as pep- RLR activity is present,69 the cnidarian sea anemone has
tidoglycan and induces an autophagy-mediated destruction been reported to have a RLR homologue,70 again showing
of intracellular pathogens as well as production of proin- the importance of studying this taxon for the emergence of
flammatory cytokines; however, it remains controversial immune-related molecules.
whether there is direct or indirect recognition of the PAMPs
(similar to some responses in plants). Polymorphisms in the Cytosolic Deoxyribonucelic Acid (DNA) Sensors
NOD proteins are associated with inflammatory bowel dis- There are four mechanisms of cytosolic DNA sensing, three
eases. The NLRP and NAIP NLRs are activated by in ways of them, the DNA-dependent activator of IFN-regulatory
that are not well understood by various PAMPs or danger- factor, IFI16, and RNA polymerase III (which converts viral
associated molecular patterns and form inflammasomes, DNA into RNA recognized by RIG-I), induce type I IFN
best known for the activation of caspase 1 and the processing production through the intermediate STING, a protein as-
of pro–IL-1beta (or mature IL-18) for release from cells.14,65 sociated with the endoplasmic reticulum (ER).71 In addition
The founding member of the family, CIITA, has long been to being an intermediate in IFN upregulation (through IFN
known to upregulate class II genes (and associated genes, regulatory factor-3), STING is also a PRR in its own right,
such as cathepsins and invariant chain), and its function is responding to the PAMP cyclic dinucleotides produced by
somewhat outside the norm. Another member, NLRC5, has intracellular bacteria like Listeria ; this suggests that STING
been shown to upregulate MHC class I expression, but the was originally a PRR, and then was co-opted by several
mechanism is unknown.66 While the shuttling of the verte- other PRR sensors to induce effector functions.72 IFI16 is
brate NLRs CIITA and NLRC5 to the nucleus seems to be a part of the AIM2-like receptor family; the founding mem-
derived characteristic, movement of plant NLRs into the nu- ber, AIM2, like the inflammasome, activates caspase-1 to
cleus to activate transcription occurs, either directly or after process pro–IL-1beta.
recruitment of transcription factors like WRKY described These new molecules/mechanisms have so far only been
in the following. studied in mammals, but it would be surprising if they were
NLRs are expressed by echinoderm coelomocytes, again not operative (at least) in other vertebrates as a way to com-
representing a highly diversified family48 (> 200 members, bat DNA viruses. To date, they have not been found in the
similar to the TLRs and SRCRs). As mentioned, it is surpris- sea urchin or jawless fish databases.
ing that these genes do not seem to be represented in proto-
stomes, and thus the emergence of the family in plants and Tripartite Motifs
deuterostomes occurred through convergent evolution.67 On Tripartite motif (TRIM) proteins belong to a family induced
the contrary, in the vertebrates a search of the Danio rerio by type I and II IFNs, with 68 members in the human ge-
(and other teleosts) database have yielded a large number of nome. TRIMs are involved in resistance against pathogens
NLR sequences, more similar to the situation in plants.63 In in mammals, especially lentivirus (eg, human Trim 5α is
humans, most NLR genes are encoded in clusters on chro- a retroviral restriction factor with activity against human
mosomes 11p15, 16p12, and 19q13, where six sequences are immunodeficiency virus).73,74 The activity of proteasomes,
found in a single telomeric region. responsible for cytosolic protein degradation, has been im-
plicated in the TRIM5α-dependent attenuation of retroviral
Rig-I-Like Receptors (RLR) reverse transcription. TRIMs contain an N-terminal moiety
The retinoic acid-inducible gene (RIG)-I is an intracel- composed of three modules: RING (with an E3 ubiquitin-
lular defense molecule that is unrelated to the NOD pro- ase activity)-Bbox-coiled xoil motif followed by different
teins, with N-terminal CARD and C-terminal helicase C-terminal domains. TRIMs fit into two major categories by
domains.68 With the helicase domain, RIG-1 binds to an the function of their C-terminal domain: Category 1 with
uncapped 5′ phosphate group, which is diagnostic of viral a PHD, MATH, ARF, FNIII, exoII, or NHL domains, and
RNAs. RIG-I also recognizes short double-stranded RNAs, Category 2 with a B 30-2 domain shared with butyrophilins
while a second member of this family MDA5 recognizes and other proteins and essential for ligand binding.75 The
long double-stranded RNAs. These molecules contain two tertiary structure of TRIM21 revealed two binding pockets
CARD domains at the N-terminus, a DEXDc domain, a he- in the B30.2 domain formed by six variable loops.76

Despite reports to the contrary, the TRIM family is an- some are involved in bacterial defense through direct bind-
cient.77 The family has been greatly diversified in vertebrates ing and others through the melanization reaction.85 Some
and in a taxon-specific manner, as observed for many mul- C-type lectins are encoded in the NKC, including the Ly49
tigenic immune families.77 The zebrafish genome harbors a and NKG2 families, as well as CD94 and several other mem-
striking diversity of a subset of Category 2 TRIMs not en- bers of the family are central to NK-cell function in mam-
countered in mammals, called finTRIM, with 84 genes dis- mals. A molecule resembling CD94 but unlikely to be an
tributed in clusters on different chromosomes. This subset, ortholog (see the following) has been detected on a subset
specific of teleosts, is overexpressed after virus infection of hemocytes in Botryllus and Ciona, the functions of which
in the trout. In the B30.2 domain, residues under positive are unknown.86 Another large gene family that is implicated
selection are concentrated within a viral recognition motif in the response of the sea urchin to immune challenge in-
first recognized in mammalian Trim 5α .78 cludes 100 small C-type lectins,48 consistent with the enor-
Finally, trim genes encoding Category 2 proteins are pref- mous expansion of several immune defense families in this
erentially located in the vicinity of MHC or MHC gene para- animal. We describe other functions of C-type lectins in the
logs both in fish and human, suggesting that they may have NK cell sections.
been part of the ancestral MHC.79 The B30.2 domains most
closely related to finTRIM are found among NLRs, indicat- Scavenger Receptors
ing that the evolution of TRIMs and NLRs was intertwined The SRCR superfamily is an ancient (from sponges to chor-
by exon shuffling.80 Exon shuffling was likely responsible dates) and highly conserved group of cell surface and/or
for the presence of the B30.2 domain in butyrophilin and secreted proteins, some of which are involved in the develop-
TRIM genes where it was perhaps favored by the proximity ment of the immune system as well as the regulation of both
of gene in the MHC. It has been argued that during evolu- innate and adaptive immune responses; they are especially
tion the combination of SPRY and PRY motifs that build up well known for their function in macrophages.87 Group B
the B30.2 domain were selected and maintained for immune SRCR domains usually contain eight regularly spaced cyste-
defense.81 ines that allow the formation of a well-defined intradomain
disulfide-bond pattern. Scavenger receptors are best known
P47 GTPases for their housekeeping function of taking up lipids modified
Among IFN-inducible immunity-related genes with an in- by oxidation or acetylation, but they have many other func-
teresting evolutionary history, immunity-related GTPases tions as well, such as uptake of apoptotic bodies (eg, cro-
(IRG/p47 in mouse) function as cell-autonomous resistance quemort in Drosophila of the CD36 subfamily88).
factors by disrupting the vacuolar membrane surrounding SRCRs have been studied mainly in the coelomocytes of
parasites (eg, toxoplasma).82 The IRG system studied pri- echinoderms. Within a few hours after bacterial injection,
marily in mice (absent in humans83) is present throughout sea urchin coelomocytes upregulate a variety of genes in-
mammals but the number, type, and diversity of genes differ cluding an extremely diverse family of SRCRs.48,89 A very
greatly even between closely related species, one of the com- large number of SRCR domains are present (approximately
mon themes in immunity described previously. 1,200), but each individual may express different groups
Concerning the evolutionary origin of the IRGs, the of SRCR genes at different levels (and even with differen-
homologs of zebrafish and pufferfish seem to form two tial splicing). To assume that they are all involved in de-
teleost-specific groups, another common theme in this fense is premature, as SRCR genes can be both up- and
chapter. Their putative promoter regions suggest an expres- downregulated after infection with bacteria. As mentioned,
sion regulated by an IFN. Homology searches failed to find this high level of gene duplication is a general rule in the
any convincing ancestral form to the vertebrate IRG pro- echinoderms.
teins in the genomes of invertebrates, but in phylogenetic In mammals, the SRCR family as a whole is also poorly
trees vertebrate IRGs clusters with some families of bacte- defined but is involved in endocytosis, phagocytosis, and
rial GTPases. Thus, IRGs may be derived from a prokaryotic adhesion, and some members acts as PRRs that bind to LPS
GTPase acquired by a horizontal transfer subsequent to the or other bacterial components. SRCRs are widespread in the
appearance of eukaryotes.82 human genome and participate as domains in the struc-
ture of numerous receptors (eg, S4D-SRCRB, CD6, CD5-L,
CD163), but without showing the high level of duplication
Integral Membrane (and Sometimes Secreted) seen in the echinoderm families.87
Proteins
C-Type Lectins Down Syndrome Cell Adhesion Molecule
Lectins were originally defined by their ability to bind car- DSCAM in Drosophila and other arthropods was described
bohydrates in a calcium-dependent manner (how C-type originally by neurobiologists as an axon-guidance protein,
lectins got their name84) and some have been described dependent upon a large number of isoforms (> 30,000)
previously (and throughout the chapter). They are found in generated by alternative splicing for the IgSF domains and the
many phyla in both the deuterostome and protostome lin- transmembrane segment. DSCAM is also involved in insect
eages in both membrane and/or secreted forms (eg, MBL immunity, expressed in cells of the hematopoietic lineage,
described in the following). A large number of C-type lec- and clearly capable of binding to bacteria; like in the ner-
tins have been uncovered in the mosquito genome, and vous system, a large number of splice variants are generated,

clearly different from the ones expressed in neurons.24 In Peptodoglycan-Recognizing Protein and
Drosophila, the DSCAM gene is composed of 115 exons, 95 of a1-3 Glucan Receptors
which encode alternative possibilities for splicing of exons 4, PGRPs are found in a wide range of organisms but have
6, and 9. The molecule consists of 10 IgSF domains and 6 FN been best studied in insects, where they are classified into
domains, and present as either a membrane or soluble form, short (S) and long (L) forms. S forms are soluble and found
presumably generated by proteolysis of the membrane form. in the hemolymph, cuticle, and fat-body cells.94 L forms are
Each cell expresses only a fraction of the isoform repertoire. mainly expressed in hemocytes as integral membrane pro-
Knock out (RNAi) and anti-DSCAM treatment signifi- teins where their fi nal structure depends on combinatorial
cantly suppresses phagocytosis, at least in Drosophila. Soluble association of different isoforms, modulated by alternative
DSCAM constructs with different exon combinations were splicing. We provide a short description here, but delve
found to have differential pathogen-binding properties.90 In more deeply in the discussion of the insect toll and im-
addition, suppression of DSCAM in mosquitoes results in mune deficiency (IMD) pathways subsequently (Fig. 4.5).36
an impaired immunity to Plasmodium ; exposure of hemo- The expression of insect PGRPs is often upregulated by
cytes to different pathogens in culture gives rise to specific exposure to bacteria. PGRPs can activate the toll or IMD
modifications and selection of alternative splicing patterns. signal transduction pathways (see the following) or in-
A similar finding was made in crustaceans, in which partic- duce proteolytic cascades that generate AMPs, melaniza-
ular DSCAM isoforms were induced in response to different tion, or induce phagocytosis. PGRPs directly kill bacteria
pathogens in one species91 and epitope II was under selec- by inducing a suicide mechanism, fi rst demonstrated to be
tion in a study in Daphnia.92 The diversification of DSCAM activated by a type of unfolded protein (stress) response
seems to be specific of arthropods as neither flatworm nor in prokaryotes.95 Besides their defense functions, insect
sea urchin nor vertebrate DSCAM are diversified. The verte- PGRPs expressed in the gut are believed to promote ho-
brate DSCAM has only two forms, using two alternate TM meostasis with commensal bacteria (also discussed briefly
exons. The cytoplasmic tail can also be modified by alter- in the following). Both soluble and transmembrane forms
native splicing that could change its signaling properties by are present in sea urchins, some with potential catalytic
modulation of tyrosine-based motifs.93 Human DSCAM is function.48
duplicated on chromosomes 21 and 11, but does not appear In vertebrates, PGRPs are all secreted and have direct
to be involved in immunity. microbicidal activity. Best studied in zebrafish, PGRPs are
A VERTEBRATES INSECTS PLANTS B VERTEBRATES INSECTS
PAMP Gram+ PGRP PAMP TNF Gram (–) DAP-type

DAMP Fungi β1-3GRP DAMP PGN
Spaezle SPE PGR
(proteolytic
enzyme)
TLR Toll LRR (not homologous TNFR PGRP-LC
to Toll/TLR)
MyD88 Pelle dMyD88 TRADD
RIP1
IRAKs
Tube Kinase ?
(others)
MAP
MAP kinase (dicots) FADD kinase TAK1/TAB1 dTAK1/dTAB2
Cactus NEMO K63 Kenny K63
Ikβ NF-kB
(REL) Ikkβ IRD5
p50 p65 (REL) DIF Gene WRKY Caspase 8 JNK
transcription factors Ankyrin
Ikβ NF-κB REL
Apoptosis AP1 p50 p65 (REL) RELISH
Gene Gene Gene Gene Gene Gene

regulation regulation regulation regulation regulation regulation
–Proinflammatory –Drosomycin (fungi) –Multiple defense genes –cJUN –Proinflammatory –Diptericin
cytokines –Defensin (gram+) cytokines –Cercropin
–MHC, costimulation
FIG. 4.5. Comparison of Immune Response Induction, Intracellular Pathways, and Immune Outcome in Insect (eg, Drosophila), Vertebrates
(eg, human), and Plants (a Composite of Pathways in Monocots and Dicots). Note that the initiation of the response and the outcome(s) in
insects and vertebrates are quite different for both the toll/toll-like receptor pathway (left, A) and the immune deficiency/tumor necrosis fac-
tor pathways (right, B), but the intracellular signaling pathways are well conserved evolutionarily in insect and vertebrate (details in the text).
Note as well that plants use similar molecules for recognition (leucine-rich repeat–containing molecules) and have similar intracellular path-
ways with kinase cascades, but all of the molecules of recognition, signaling, and effector are derived by convergent evolution as compared
to animals. This figure was modified from Beutler et al.152

expressed in many tissues such as gills, skin, and intestine, achieved at this receptor level but rather in solution via
providing immune defense. They are expressed before the other intermediates (see the following). C. elegans has only
development of adaptive immunity and likely provide an one toll receptor, and rather than being antimicrobial re-
important protective role.96 The human PGRP genes are sponses, it promotes avoidance of a flatworm pathogen upon
found on the MHC chromosomal paralogs 1q21 and 19q13/ engagement; the signaling mechanism for the C. elegans
p13. All detected splice-variant isoforms bind to bacteria TOL (toll) is not evolutionarily conserved (it clearly does
and peptidoglycan. Like the fish molecules, mammalian not induce the NF-κ B pathway) and is under investiga-
PGRPs are also positioned at epithelial surfaces and pro- tion.67 A toll/TLR gene is present in the sea anemone, a cni-
mote intestinal homeostasis by discriminating somewhat darian, but not in other cnidarians such as hydra or coral,
between commensal (eg, lactobacilli) and pathogenic bac- which nevertheless have TIR domains associated with other
teria. Knockout mice have increased pathogenic bacteria on molecules.35 A TIR domain of the toll-receptor types was
mucosal surfaces that induce colitis after injury in the dex- detected in sponges, but, like IL-1R in vertebrates, it is as-
tran sulphate sodium autoimmune assay.97 sociated with a receptor with three IgSF domains.103 Plants
β1-3 glucan receptor proteins (β1-3GRPs, formerly known do not have toll/TLR per se, but do have LRR-containing
as gram-negative binding proteins (GNBPs) are related to bac- transmembrane sensors that function in a similar fashion102
terial β1-3 glucanases.98 They are found in insects and other (see Fig. 4.5). In summary, toll/TIR arose before the split of
arthropods where they bind bacteria, fungal β-1, 3-glucans, protostomes and deuterostomes, but has been lost in some
LPS, and/or bacterial lipoteichoic acid (without necessarily invertebrate groups and has been recruited to perform mul-
showing glucanase activity) . An ortholog is present in the tiple functions.
sea urchins, but not in vertebrates to date. Drosophila GNBP1 The arsenal of TLRs in vertebrates is endowed with spe-
together with PGRP-SA are required to activate the toll path- cific and diverse capacities. Each vertebrate TLR has its range
way in response to infection.36 of specificities and, in addition, combinations of different
TLR can create different binding specificities (eg, the asso-
Toll and Toll-like Receptors ciation of TLR2 with TLR6 or TLR1 and 2104 or even TLR2
The toll receptors were originally described in Drosophila homodimers in regulatory T cells105). This divergence in
as genes involved in early development, specifically in recognition function is well illustrated by the phylogenetic
dorsoventral patterning. Later, they were also shown to analysis of the toll and toll-related receptors in different
be essential sensors of infection, initiating antimicrobial phyla such as arthropods and vertebrates. Toll and related
responses.36,99 This family was then revealed to be a major proteins from insects and mammals cluster separately in
force in innate immunity in the vertebrates as well.100,101 the analysis, indicating independent generation of the major
As mentioned previously, across the metazoa structurally families in protostomes and vertebrates.102 Consistent with
closely related members of the toll family range from not the expansion of SRCR and NLR genes in sea urchins, hun-
being involved in immunity (in C. elegans and apparently dreds of TLRs also were found in this species.48 TLRs of the
in the horseshoe crab), to being the equivalent of a cytokine protostome type are in small numbers (3 members) while the
receptor (in Drosophila), to being PRR in the vertebrates vertebrate type has been enormously amplified, all within
and invertebrates.102 Six spaetzle-like and eight toll-like a single family (222 members) and most without introns.
molecules have been identified in Drosophila, but only one Vertebrate TLRs do not diverge rapidly and evolve at about
or two of them are clearly immunity-related.36,102 In jawed the same rate, and while there have been some duplications
vertebrates, they belong to a multigene family of PRR spe- in amphibians and fish, they are not greatly expanded like in
cific for diverse PAMPs and exhibiting different tissue dis- the echinoderms (no more than approximately 20 genes in
tributions and subcellular locations.27 In humans, many are any species).
on chromosome 4p and q (TLR 2, 1, 6, 10) but the others are
distributed on chromosomes 9, 1, 3, and X.
Ectodomains of TLRs comprise 19 to 25 tandem repeats Signaling Through Innate Surface
of LRR motifs made of 20 to 29 aa capped by characteris- Recognition Molecules
tic N- and C-terminal sequences. All of the toll receptors Four pathways of innate immunity triggering have con-
are homologous and appear similar in domain constitution served elements in eukaryotes: the toll/TLRs, the TNF-α /
among all animals. They also share the TIR domain, which IMD receptors, the intracellular NOD, and the JAK/STAT.
is the intracellular segment shared with the IL-1/-18/-33 re- Although toll receptors have been found in almost all
ceptors of vertebrates, as well as other molecules in plants. triploblastic coelomates, most of the work and the elucida-
TIR domains associate with Myd88 to initiate signaling cas- tion of pathways have been accomplished in Drosophila and
cades culminating in the activation of NFκ B/Rel (see the Anopheles.36 The diversity of AMPs that can be produced
following) (see Fig. 4.5). via the toll/IMD pathways is substantial, and as described
In Drosophila, the toll dimer is triggered by an interac- previously is classified in several categories depending upon
tion with the unique ligand spaetzle, which is the product the type of pathogen recognized (eg, gram (+), drosocin,
of a series of proteolytic cascades, with the most critical en- gram (−), diptericin; fungal, drosomycin) with differ-
zyme identified (spaetzle-processing enzyme). Activation ent effector functions (see Fig. 4.5). Insect antimicrobial
of the cascades triggers the production of antimicrobial molecules were originally discovered by the late Hans
peptides (see Fig. 4.5). The specificity of recognition is not G. Boman and colleagues in 1981, a seminal finding that

heralded the molecular analyses of innate immunity in the system arose early in the animal kingdom110 ; plants employ
invertebrates.106 a different system of transcriptional activators, the WRKY
molecules, which are activated directly by the mitogen-acti-
Toll and Immune Deficiency Pathways vated protein kinase cascades, similar to transcription factor
As described, invertebrate toll receptors are homologous to found in animals, AP1.
the vertebrate TLR, in the sense that they are integral mem-
brane LRR-containing proteins (see Fig. 4.5). Drosophila toll
is activated after it binds spaetzle, the product of a proteo- Extracellular Soluble Receptor with Effector Cascade
lytic cascade activated in solution after the interaction of Proteolytic cascades are initiated immediately following
molecules produced by fungi or gram-positive bacteria with interaction of foreign material bound by preformed pro-
GNBP and PGRP.107 The TIR cytoplasmic domain of the toll teins in solution, and this principle is conserved throughout
receptor then interacts with MyD88 (itself having a TIR do- evolution. Indeed, the proteolytic cascade upstream of pro-
main) followed by Tube and Pelle, leading to activation of duction of the toll ligand spaetzle resembles the complement
the homologous NF-κ B system (Cactus or Diff) that then or clotting cascades. The PPO cascade of arthropods lead-
induces transcription of various defense peptides.36,99,108 ing to melanization and the genesis of antibacterial prod-
This is remarkably similar to the cascade of events following ucts described in the following is another example in which
activation of mammalian TLRs where after their interaction peptidoglycans on microbial surfaces initiate the cascade
with PAMPs at the cell surface, a cascade is induced through resulting in the degranulation of hemocytes.
TLR including MyD88, IRAK, TRAF, TAK1, to NF-κ B via
the IKK signalosome. Thus, infection-induced toll activa- The Complement System
tion in Drosophila and TLR-dependent activation in mam- The best-studied immune proteolytic cascade that is sur-
mals reveal a common ancestry in primitive coelomates (or prisingly well conserved in the animal kingdom is com-
previous), in which defense genes under the control of a plement111,112 (Fig. 4.6). In contrast to the other defense
common signaling pathway lead to activation of Rel family molecules that we have discussed, orthologous complement
transactivators. genes can be detected in all of the deuterostomes without a
The Drosophila IMD pathway is employed in responses great deal of expansion/contractions of the gene family. The
to gram-negative bacteria109 (see Fig. 4.5). After interac- three major functions of complement in jawed vertebrates
tion with the cell surface receptor PGRP-LC mentioned are 1) coating of pathogens to promote uptake by phagocytes
previously, in a cascade similar to the mammalian TNF-αR (opsonization); 2) initiation of inflammatory responses by
signaling pathway, Drosophila tak1, an IKK signalosome, stimulating smooth muscle contraction, vasodilatation, and
and a Relish-mediated (instead of Diff) NF-κ B step, results chemoattraction of leukocytes; and 3) lysis of pathogens via
in transcription of antibacterial peptides like diptericin. The membrane disruption. Additionally, in the vertebrates, C′ is
Drosophila intracellular pathway is similar to the mamma- vital for the removal of immune complexes as well as elicita-
lian TNF-α receptor cascade, which also progresses via a tion of adaptive humoral immunity. The focal point of com-
death domain Mekk3, the signalosome, and NF-κ B result- plement is C3, which lies at the intersection of the alternative,
ing in cytokine production. In both cases, a link to pathways classical, and lectin pathways of complement activation. It is
leading to programmed cell death is possible; overexpres- the only known immune recognition molecule (besides its
sion of Drosophila IMD leads to apoptosis. When the activa- homologue C4) that makes a covalent bond with biologic
tion of either the fly toll or IMD pathway is considered, they surfaces via a thioester linkage. C3 has a nonspecific recog-
are analogous to a mammalian cytokine/cytokine receptor nition function, and it interacts with many other proteins,
system (eg, TNF-α) in which a soluble self-molecule acti- including proteases, opsonic receptors, complement activa-
vates cells via a surface receptor. Fitting with the paradigm tors, and inhibitors. In the alternative pathway, C3 exposes
put forward on recognition, signaling, and effector phases of its thioester bond in solution, and in the presence of cell
the immune response, the diversity of external recognition surfaces lacking regulatory proteins that block C3 activation
systems is not matched by an equivalent diversity of intra- (by cleaving it into iC3b), it associates with the protease fac-
cellular signaling pathways.22 There are conserved signaling tor B (B or Bf). After binding to C3, B becomes susceptible
cascades coupled to the receptors, giving the impression of to cleavage by the spontaneously active factor D, resulting in
conservation of the innate immunity pathways; yet, these formation of the active protease Bb that in combination with
pathways are also used in development, so which is primor- the covalently attached C3 cleaves many molecules of C3 in
dial remains an open question. an amplification step. Another nonadaptive recognition sys-
Plants do not have toll/TLR, but do have transmembrane tem, the lectin pathway, starts with the MBL (or the lectin
LRR-containing microbial sensors, of which FLS2 that binds ficolin), which is a PRR of the collectin family that binds
to flagellin, is best characterized27 (see Fig. 4.5). These mol- mannose residues on the surface of pathogens and can act as
ecules do not have an intracellular TIR domain (note that an opsonin. MBL is analogous to C1q with its high-avidity
TIR domains exist in plants, but not associated with the TM binding to surfaces by multiple interaction sites through
sensors), but do recruit a kinase of a similar nature to the globular C-terminal domains, but apparently it is not ho-
toll/TLR kinases (the so-called non-RD kinases) to activate mologous to C1q. Like C1q, which associates with the serine
downstream mitogen-activated protein kinase cascades. proteases C1r and C1s, the MBL-associated serine proteases
However, as mentioned previously, the NFκ B transcriptional (MASPs) physically interact with MBL and not only activate

CLASSICAL LECTIN ALTERNATIVE

Ab-Ag MBL (Ficolin) C3
(IgM or IgG/Y)
C1qrs MBL-MASP [C3-H2O]B

(C1r cleaves C1s) (D cleaves B)
C3, C4, (a2m, TEP)

Thioester
C3: From Diploblastic C4

C4: Jawed vertebrate (Cleaved by C1s or MASP2)
a2m: All metazoa
TEP: Protostomes
[C3-H2O]Bb
Inflammation (Bb cleaves C3 in plasma)
C5a > C3a > C4a C4b C2 C3
Jawed vertebrates (C1s or MASP2 (Cleaved by Microbial surface;
cleaves C2) C2b or Bb) properdin (factor P)
Function
Opsonization stabilizes
C3b: From Diploblastic
C2a C3a
TEP: Protostomes
Cell Lysis
MAC (C5b-9): Jawed vertebrates C4b C2b C3b C3b Bb
MACPF: All animals (C3 convertase) (C3 convertase)
C3, C4, C5, a2m, TEP, CD109 (Factors H, I and

(CR1 inhibits) CR4 inhibit)
C3, a2m: Diploblastic
Homologous members
C4, C5, CD109: Jawed vertebrates

TEP: Arthropods C4b C2b C3b C5 C3b Bb C3b
(C5 convertase) (Cleaved by (C5 convertase)
C2, B
C2b or Bb)
C2: Amphibian, mammal
B: Diploblastic
C5a
C1r, C1s, MASP
C1r, C1s: Jawed vertebrates C5b
MASP: Deuterostomes
C1q, MBL
C1q: Jawed vertebrates
MBL: Deuterostomes MAC: C5b C6 C7 C8 (C9)n (CD59 inhibits)
FIG. 4.6. Evolution of the Complement System. The general pathways and appearance of the various components in the phylogenetic tree are
emphasized.
the classical pathway of complement by splitting of C4 and C3 and MBL (and ficolin) are vital players in the im-
C2 (the same function as C1s; MASP2 appears to be the ac- mediate innate immune response in vertebrates, and both
tive protease), but also can activate the alternative pathway have been described in nonvertebrate deuterostomes.48,112
in ways that are not understood and thus completely bypass Thus far, the best-studied invertebrate systems for investi-
the classical pathway. Indeed, MASP-1 and -2 are homologs gation of C3 evolution are the sea urchin and the ascidians
of C1r and C1s (see Fig. 4.6). Both C1q and MBL can promote Halocynthia and Ciona, in which C3 and B molecules and
the uptake of apoptotic bodies by phagocytes, via collectin genes have been analyzed in some detail. In contrast to the
receptors. Another lectin, ficolin, can also initiate the MASP very high levels of C3 found in the plasma of jawed verte-
pathway,113 and it would not be surprising if other activa- brates, sea urchin C3 is not expressed at high levels but is
tors were discovered in the future (eg, the ancient molecule induced in response to infection in coelomocytes.114 The C3
C-reactive protein is also capable of activating C′). Finally, opsonic function clearly has been identified, but so far ini-
the classical pathway, which is dependent upon antibody tiation of inflammatory or lytic responses (if they exist) has
molecules bound to a surface, results in the same potential not been obvious. Receptors involved in the opsonization in
effector outcomes described previously for the alternative echinoderms have not been identified, but in the ascidian
pathway. Novel molecules initiating this pathway are C1q, gene fragments related to the C3 integrin receptor CR3 were
C1r, C1s, C4, and C2, as well as specific negative regulatory identified, and antisera raised to one of the receptors inhib-
proteins such as C4-binding protein. ited C3-dependent opsonization.115

Hagfish and lamprey C3-like genes were thought to be the classical pathway in mammals, it is possible that some
ancestral C3/C4 genes because the sequence predicts two portion of this pathway exists in prejawed vertebrates.
processing sites (leading to a three-chain molecule), like Nevertheless, C4 and C2 genes have not been detected to
C4, but a C3-like properdin-binding site is clearly present.116 date in jawless fish or invertebrates. A bonafide C2 homo-
However, like C3 in other animals the hagfish protein is logue has only been identified to the level of amphibians,
composed of only two chains of 115 and 72 kDa, and sea although duplicate B genes were isolated from cartilagi-
urchin and ascidian C3 sequences predict only two chains nous fish and teleost fish that may function both in the
(one proteolytic processing site). The lamprey, but not sea classical and alternative pathways. The lytic or MAC path-
urchin C3, has a recognizable C3a fragment known from way, which is initiated by the cleavage of C5 into C5a and
gnathostomes to be involved in inflammation, so the role of C5b, also has not been described in taxa older than car-
complement in inflammation may be a vertebrate invention tilaginous fish. Thus, opsonization and perhaps the in-
(but see the following). duction of inflammatory responses were the primordial
TEPs have been isolated from Drosophila and the mos- functions of the lectin/complement pathways. However, a
quito Anopheles, as well as several other arthropods.117,118 complementary DNA clone for CD59, a molecule that in-
While the insect molecules function in a C3-like fashion hibits MAC formation in self-cells, was identified from a
(opsonization), phylogenetic analysis shows them to be hagfish library, and some of the terminal components of
more related to α2-macroglobulin (note that a few insects the pathway have been detected in lower deuterostomes
actually have molecules more related to C3). TEPs in insects with no described functions.116 Interestingly, proteins with
function as opsonins, binding to parasites and promoting the MAC/perforin domain have been detected throughout
their phagocytosis or encapsulation. The evolution multi- the animal kingdom,48 and some are even involved in cyto-
member TEP families in Drosophila and mosquito followed toxic reactions; however, it seems that only vertebrates have
independent evolutionary paths, perhaps as a result of spe- bonafide terminal C′ components that are highly evolved
cific adaptation to distinct ecological environments as de- for targeted destruction of cell membranes. The perfo-
scribed in the introduction. The Drosophila genome encodes rin gene itself also seems to have arisen in gnathostomes,
six TEPs (whereas there are 15 genes in Anopheles, again from an ancient MAC/perforin domain-containing gene,
consistent with the major expansion of many immune gene macrophage-expressed gene 1 protein, which dates back to
families in the mosquito), three of which are upregulated sponges; thus, cellular cytotoxic reactions in the inverte-
after an immune challenge. Mosquito TEPs are involved in brates described in the following must use novel cytotoxic
killing of parasites, and the reaction is regulated by LRR- effector molecules.126
containing molecules to avoid destruction of self-tissues; C3, C4, C5, and α2m (and TEP) are members of the same
thus, full-blown complement-like systems complete with small family. A cell-surface–expressed (GPI-linked) member
inhibitors have arisen independently in protostomes and of this family, CD109, has been shown to associate with the
deuterostomes.119,120 transforming growth factor (TGF)-β receptor and modulate
C3-like genes are present in cnidarians121,122 and in the its expression.127 The protease inhibitor, α2m, clearly present
horseshoe crab Limulus.123 Good phylogenetic support was in invertebrates (protostomes and deuterostomes) and ver-
obtained for their relationship to C3, as compared to other tebrates, is thought to be the oldest, but obviously this must
members of the thioester-containing family like the TEPs. be viewed with caution considering the data in cnidarians.
Thus the emergence of C3 as a defense molecule predates Along with its ability to bind to and inactivate proteases of
the split between protostomes and deuterostomes. A gene all known specificities through a “bait region,” it also has
resembling the proteolytic enzyme Bf was discovered in been shown to be opsonic in some situations. α2m, C3, C4,
these protostomes as well (and in sea anemones), suggesting and CD109 (as well as the TEPs) have internal thioester
that the fundamental system was in place a billion years ago sites, so this feature is primordial; C5 subsequently lost the
(see Fig. 4.1). The lack of C3 in many other protostomes site. The first divergence probably occurred between α2m
suggests that the ancestral gene was lost and replaced by and C3, with C5 and then C4 emerging later in the jawed
the TEPs.117,118 vertebrates.128 Consistent with Ohno’s vertebrate polyploidi-
In jawed vertebrates and some lower deuterostomes, cer- zation scheme is the fact that C3, C4, and C5 genes are lo-
tain species express more than one C3 gene, suggesting that cated on three of the four previously described paralogous
the innate system might compensate in animals that do not clusters in mammals, and this is also fits with the absence
optimally make use of their adaptive immune system.124 of classical (no antibody) and lytic (no MAC) pathways in
Changes in the amino acid composition of the C3-binding phyla older than cartilaginous fish.2 α2m is encoded at the
site are found that may somehow regulate the types of sur- border of the NKC in mice and human, and there are simi-
faces bound by the different isotypes.125 Likewise, in lower larities between these regions and the other MHC paralogs
chordates such as Ciona, C3 and other complement com- (see Fig. 4.13). The C3a and C5a receptors that promote the
ponents can be duplicated.115 Diversification of the carbo- inflammatory responses upon complement activation have
hydrate recognition domains has been observed also in the been identified in several vertebrates and (perhaps) some
Ciona MBP family (nine members). lower deuterostomes; like the chemokine receptors they are
Like Ig/TCR/MHC, the classical pathway and the ter- G-protein coupled receptors whose genes may also be found
minal pathway membrane-attack complex (MAC) appears on the ohnologs (C3aR, chr 12p13; C5aR, chr 19q13). If in-
fi rst in cartilaginous fish.112 However, as MBL can activate deed such receptors are found in the prejawed vertebrates

as suggested by recent pioneering experiments in Ciona and immunity is unlike any other effector so far described but
Styela, it will be interesting to determine whether they are illustrates the utility of LRR in many different types of mol-
involved in some type of inflammation, thought to be the ecules and processes. Pathogens bound by AMPs can be
domain of the vertebrates.129 phagocytosed or walled off by a barrier of flattened hemo-
cytes and ECM. The ECM forms a basement membrane that
Melanization (Prophenoloxidase Cascade) becomes stabilized partly through peroxidases that generate
A major defense system in invertebrates is the melaniza- tyrosine-tyrosine bonds. The combination of LRR and Ig
tion of pathogens and damaged tissues,130 popularized by structures suggests that peroxidasin may precisely mediate
poor Gregor in Kafka’s Metamorphosis, when the cock- adhesion of cells to the ECM.
roach Gregor undergoes a melanization reaction from an As mentioned, a large number of LRR-Ig–containing
apple thrown into his thorax by his father. The process proteins has been discovered, most of them playing roles
is controlled by the circulating enzymes PPO and phe- in embryologic development.32 Many LRR-Ig proteins are
nol oxidase. The system is activated by β1-3GRP, PGRP, encoded in paralogous regions in the vicinity of immune
LPS-binding proteins, and other proteins that can bind to genes, showing an ancient direct connection between the
various PAMPs (see Fig. 4.3). The complexes launch a cas- families (see the following).
cade of serine protease activities resulting in cleavage of the
pro-form of a prophenoloxidase-activating enzyme into Fibrinogen-Related Proteins
the active form that in turn activates the PPO into phenol FREPs are proteins that were first discovered in the hemo-
oxidase. This leads to the production of quinones and fi- lymph of snails with an IgSF moiety (one or two V-like do-
nally melanin. Melanization can completely inhibit parasite mains) and a fibrinogen domain. The fibrinogen domain is
growth, whereas concomitant with PPO activation, many found in a large number of defense molecules throughout
other immune reactions are initiated, such as the genera- the animal kingdom (eg, the ficolins). FREP gene expression
tion of factors with antimicrobial-, cytotoxic-, opsonic-, or is upregulated following exposure to the mulluscan para-
encapsulation-promoting activities. The presence of specific sites such as schistosomes133 ; a snail strain resistant to schis-
proteinase inhibitors (of the serpin family) prevents unnec- tosomes shows an upregulation of the FREP 2 and 4 genes
essary activation of the cascade and overproduction of toxic of up to 50-fold. The original discovery of FREPs followed
products. Phenoloxidase is the key enzyme responsible for the recovery of snail proteins that bound to worm antigens,
the catalysis of melanization. It is a marker of the PPO acti- and thus this is one case in which the correlation between
vating system, and it can be an immune effector by itself as an invertebrate receptor and its ligand is clear. However, it is
demonstrated in ascidians. It is therefore interesting to as- not known whether the IgSF or fibrinogen domain (or both)
sess its conservation within all metazoa. A survey of the dif- bind to the antigen or which effector functions are induced
ferent organisms revealed the presence of phenoloxidase in after FREP binding.
many deuterostome and protostome phyla, and related mol- FREP diversity is remarkable in that there are many poly-
ecules are also present in sponges. In arthropods, several morphic genes as well as alternate messenger RNA splicing
PPO genes are present in the genome (nine in Drosophila to generate the diversity. In addition, based on the num-
and Aedes). Some may have different “immune” functions ber of genes and alleles in individual snails, there appears
such as injury repair. Several components that would main- to be a somatic diversification mechanism that modifies
tain the role of melanization in immunity may be lacking in FREP genes, either via mutation or gene conversion in the
different phyla even if many elements are conserved, and so region that encodes the IgSF domains.134 Over 300 unique
far the best examples of melanization associated with im- sequences were found in 22 snails, consistent with a somatic
munity are still found almost exclusively among arthropods diversification mechanism. Currently, there is no mecha-
and to a lesser extent in annelids. Despite the presence of nism to account for the mutations, but data accumulate for
molecules involved the pathway, the PPO cascade per se somatic modifications and the use of FREPs in critical de-
does not exist in vertebrates. fense against pathogens.135
FREPs are also present in arthropods where they lack the
Ig domains,136 once again with an expansion of genes in the
Effector Molecules mosquito Anopheles gambiae as compared to Drosophila.
Peroxidasin RNAi studies have shown that subsets of FREP genes are
Among molecules containing LRR motifs, peroxidasin oc- vital for defense against malarial parasites, and different
cupies a special place because of its involvement in hemo- FREPs bind to bacteria with different affinities.137 Homo-
cyte biology in insects and because of its homology to the and heterodimers can form between different FREPs, and
LRR motifs in the agnathan VLR and Ig domains similar to multimers can be fashioned that likely increase the avidity
Ig itself. Drosophila peroxidasin is an assembly of a cysteine- of binding. In summary, this ancient family of defense mol-
rich motif, six LRR, and four IgSF domains.131 The molecule ecules has all of the attributes described in the introduction:
is conserved in vertebrates, although a role in immunity rapidly evolving multigene family, conservation in divergent
has not been reported. Another molecule called peroxinec- protostomic invertebrate phyla, somatic diversification via
tin, with similarity at the level of the peroxidase region, has alternative splicing, and perhaps an unknown mutational/
been described in crustaceans and shown to be associated gene conversion mechanism, as well as heterodimeric (and
with immunity via the PPO cascade.132 Its involvement in multimeric) association.

185/333 a link to the regulation of commensals in early deutero-

There are a number of additional expanded gene fami- stomes. It should be noted here, however, that recent data
lies in the sea urchin genome that encode proteins with suggest that “tolerance” of commensals occurs in the pro-
immune-related functions. The 185/333 genes were first tostomic invertebrates as well.
noted because they are highly upregulated in coelomocytes
after exposure to LPS, constituting up to 7% of the mes- Antimicrobial Peptides (Defensins)
senger RNAs in such cells.138 Subsequently, transcripts were Each metazoan taxon produces a variety of molecules
shown to be upregulated by many different types of PAMPs. with intrinsic antimicrobial activity,144,145 the majority of
The encoded proteins have no detectable similarity to any which fall into three major categories: defensins, cateli-
other gene family, but they are highly diversified and are cidins, and histatins. Even in species with very small ge-
produced (at least) by a subset of coelomocytes. nomes (such as the tunicate Oikopleura [60 Mb genome]),
There are estimated to be at least 50 185/333 genes in the selection pressures have been strong enough to lead to ex-
sea urchin genome.139 The gene is composed of two exons, pansion of the Phospholipase a2 family, with 128 mem-
one encoding the leader and the other encoding the mature bers.146 Some families are evolutionarily conserved but
protein. The second exon is made up of so-called elements generally they diverge rapidly and orthologous relation-
and repetitive sequences that are quite different from gene ships are not apparent. The best studied group of AMPs
to gene, which can to a large extent explain the diversity of is the defensins, which are amphipathic cationic proteins;
expressed loci. However, there are hints of RNA editing, a their positively charged surface allows them to associate
unique form of alternative splicing, somatic mutation (per- with negatively charged membranes (more common in
haps targeting cytosine residues), and other (perhaps) novel pathogens), and a hydrophobic surface that allows them to
mechanisms of diversity to explain the incredible number disrupt the membranes, either by disordering lipids or ac-
of different isoforms; additionally, like many other defense tually forming pores. Most of the molecules are proteins,
molecules, there is evidence of protein multimerization. but an antimicrobial lipid called squalamine, which also
Whatever the mechanism of diversity generation, it would is modeled to have hydrophobic and positively charged
be surprising if this family were not vital for defense in echi- surfaces, is found at very high levels in dogfish and lam-
noids. Furthermore, the presence of this unique multigene prey body f luids.147 Defensins can either be constitutively
family in sea urchins is consistent with the great expan- expressed (eg, in respiratory epithelia in mammals) or
sion of other immune gene families in this group, including inducible (eg, see the following for Drosophila and see
TLRs, SRCRs, and NLRs.7 Fig. 4.5). Certain responses that seem systemic, like the
production of Drosophila defensins, can also take place
Variable Domain Chitin-Binding Proteins locally in the damaged tissues themselves; otherwise, a
VCBP, fi rst discovered in Amphioxus but present in Ciona systemic response is initiated in organs distant from the
as well, consist of two Ig domains of the V type but site of infection such as the fat body in Drosophila where
with a different folding motif when compared to Ig or induction of bactericidal peptide expression occurs.36
TCR V domains140 followed by a chitin-binding domain. Defensins are the focus of great attention in commercially
The chitin-binding domain resembles chitinases found bred species such as oysters, mussels, and crustaceans.
throughout the animal kingdom, and like dedicated chitin- Besides their direct defense functions, in mammals de-
ases VCBP is usually expressed in the gut. Apparently, there fensins play other roles, such as chemotaxis and immune
are no cell-surface–expressed forms and thus all VCBPs regulation.
are likely to be secreted, effector molecules. In Amphioxus,
their diversity is enormous, apparently entirely because of Penaedins. One set of diverse AMPs is the penaedins,
polymorphism and polygeny, and not somatic alterations. present in crustaceans (shrimp). Penaedins are small an-
Each individual can carry up to five genes per haplotype, timicrobial peptides (5 to 7 kda) that bind to bacteria and
and in limited studies (11 individuals), no identical hap- fungi, and consist of a conserved leader peptide followed
lotype has been encountered.141,142 The general structure of by an N-terminal proline-rich domain and a C-terminal
the V domain is like that of the vertebrate rearranging anti- cysteine-rich domain.7,148 Most of the diversity is found in
gen receptors, but with some unusual properties, including the proline-rich domain,149 suggesting that it is most im-
packing in a “head-to-tail” dimeric fashion, totally unlike portant for recognition, but both domains are required for
Ig and TCR. VCBP diversity does not reside in the Ig/TCR recognition of bacteria and fungi. Four classes of penaedins,
CDR resides, but rather in the A, A′, and B strands, like in PEN2 to 5, are expressed by shrimp hemocytes. A great di-
DSCAM. versity of isoforms is generated, with substitutions and dele-
By contrast to the Amphioxus VCBP, there are only a few tions within the proline-rich domain, suggesting that this
nonpolymorphic Ciona VCBP genes, which are expressed domain recognizes ligand; nevertheless, both domains seem
by gut epithelium and amebocytes. Soluble VCBPs bind to be required for function. Like the VCBPs, each penaei-
to bacteria and induce opsonization in amebocytes. It is din class seems to be encoded by a unique gene and isoform
hypothesized that these molecules may perform a func- diversity is generated by polymorphism. Multiple copies of
tion similar to mucosal IgA in vertebrates, which provides penaedin genes are present in different species, and there is
a “firewall” protecting from invasion of intestinal bacteria rapid expansion and contraction even within closely related
and promoting homeostasis.143 If true, this would provide organisms.

Responses to Viruses in the Invertebrates (bonafide NK, see the following) of the jawed vertebrate im-
Compared to immunity to extracellular pathogens in the in- mune system. Similarly the term “NK cells” covers different
vertebrates, the study of responses to intracellular pathogens cell types and different functions. NK cells of vertebrates
like viruses is in its infancy.150 First discovered in plants and can “recognize” missing self-MHC class I, but also ligands
in C. elegans, the RNAi pathway of defense against viruses induced on stressed cells following virus infection, transfor-
is also operative in Drosophila and Anopheles.151 Double- mation or stress. They can also have an immunoregulatory
stranded RNAs (viral or otherwise) are recognized by the role by interactions with antigen-presenting cells (APCs).
enzyme Dicer 2, generating small interfering RNAs that Many of these features obviously profit from a comparative
can associate with complementary RNAs and induce their approach.
degradation. In natural killing, the common denominator is a sponta-
Viruses also induce an “IFN-like response” through a cy- neous reaction (ie, it does not require any [known] antigenic
tokine receptor (domeless) that is homologous to the IL-6 priming but only cell contact). Some form of natural killing
receptor and signals through the JAK/STAT pathway.152 can be observed from the earliest metazoans onwards. Some
After viral infection, unknown signals (RNA, perhaps) in- marine sponge and corals avoid fusion with one another
duce the production of cytokines of the unpaired family by mechanism of cytotoxic cells or induce apoptosis at the
that bind to domeless on neighboring cells and upregulate level of the teguments.154 Phenomena more similar to ver-
a large number of genes involved in defense (see Fig. 4.3). tebrate NK killing are observed in sipunculid worms where
Nothing is known about the effector pathways of these re- allorecognition among populations was shown to result in
sponses, but mutants of one of the induced genes results in killing of allogeneic erythrocytes by lymphocyte-like cells.155
increased viral load. It should be noted that this pathway is Similar cases can be also encountered in annelids and mol-
not cell autonomous, inconsistent with the IFN pathway in lusks,156 The role of IgSF and lectin receptors known to be
vertebrates. involved as NK cell receptors in vertebrates has not been ex-
In a paradigm-changing paper, foreign antigen was amined in these invertebrates, even though some candidate
shown to directly interact with Drosophila toll7, resulting homologs have been identified. Note that the mechanism(s)
in the induction of antiviral autophagy and inhibition of cannot be perforin-mediated, based on the recent bioin-
viral replication.153 Toll-7 interacted with the glycoprotein formatics analysis on MAC/perforin domains described
from vesicular stomatitis virus at the cell surface to initiate previously in the complement section.126
the response. Thus, the dual paradigm of indirect and di- When comparative morphology or function is not infor-
rect recognition by toll and TLR, respectively, must now be mative, searching for conservation of transcription factors
modified. There are several other tolls without functions or cell surface markers may be useful. Survey of IgSF genes
in insects that might be involved in immunity, perhaps via across databases have not yielded any promising candidates
such mechanisms. to date. Despite the presence of polymorphic IgSF members
In summary, arthropods (and presumably other inverte- of the receptor tyrosine kinase family in sponges, their role
brates) use an RNAi pathway as well as a signaling pathway in allorecognition or killing has not been demonstrated.157
to combat viruses.150,152 As opposed to the systemic plant As mentioned, similarities were found among the lectin
RNAi response, the same pathway in protostomes is rather families, especially in prochordates where cytotoxicity has
cell autonomous. This response was lost in the vertebrates, been reported and associated with a discrete population
presumably because 1) viruses have been able to effectively of hemocytes the granular amoebocytes. The urochordate
counter this response and render it ineffective; 2) there have genome (Botryllus, Halocynthia, Ciona) encodes many lec-
been remarkable evolutionary innovations in the vertebrate tins with or without typical carbohydrate recognition sig-
innate and adaptive immune systems to combat viruses; natures, among them a putative CD94 homolog has been
and 3) vertebrates use a Dicer pathway extensively to regu- cloned and its expression followed in Botryllus158 ; its pre-
late expression of their own genes. The discovery of a viral dicted sequence does not match well to vertebrate CD94.
immune response quite like a type I IFN response in ver- This gene is differentially regulated during allorecogni-
tebrates demonstrates that the three major signaling path- tion in Botryllus, and a subpopulation of blood cells has
ways of defense in Drosophila, toll, IMD, and JAK/STAT, the receptor on their surface in both Botryllus and Ciona.
are similar to the vertebrate TLR/IL-1R, TNF, and IL-6/IFN Phagocytosis is inhibited by an antiserum recognizing the
pathways, respectively (the fi rst two homologous). Finally, Ciona homologue.159
other insect tolls may interact directly with PAMPs to pro- Other C-type lection homologs of CD209 and CD69 are
mote antiviral defense such as autophagy or apoptosis. This linked to the CD94/L gene on Ciona chromosome 1 (L.D.P.,
is a field in which rapid progress will be made in the near personal observation). Could they be part of a “pre-NK
future. complex”? Interestingly, all the human homologs of those
lectin genes are present either in the NK complex on 12 p13
(CD94, CD69) or on an MHC paralog 19p13 (CD209). Taken
Natural Killing Activity Across Metazoa together with studies of the chicken MHC which encodes
The word “cytotoxicity” encompasses vastly different pro- some C-type lectins (see the following), the data suggest that
tocols of cell killing by of different cell types. It can be a conserved MHC-linked region containing several lectin
an effector function of cells of the adaptive (cytotoxic genes was present before the emergence of MHC class I and
T-lymphocyte [CTL] or NKT cells) or of the innate arm II genes.2

In addition, a number of genes encoding membrane proin antiviral immunity. NK cell recognition of virus-infected
teins with extracellular C-type lectin or immunoreceptor cells engages the activating KIR and Ly49 receptors and
tyrosine-based inhibitory motifs (ITIMs) and immunore- NKG2D in this process. Thus, viruses are hypothesized to
ceptor tyrosine-based activation motif s (ITAMs) (plus their supply the evolutionary pressure on diversification of NKRs.
associated signal transduction molecules) were identified in In fact, it has been shown in mice that inhibitory recep-
Ciona, which suggests that activating and inhibitory receptors can rapidly mutate into activating receptors when viral
tors have an MHC class I– and II–independent function and “decoy” class I molecules evolve to engage inhibitory recep-
an early evolutionary origin.160 The ligands of these Ciona tors.164 Generally speaking, inhibitory receptors are older
molecules are of great interest to uncover. and more conserved, whereas activating receptors evolve
more rapidly and can be derived from inhibitory receptors
Natural Killer and Natural Killer–Like Cells via mutations that result in loss of the ITIM.165
NK cells express both activating and inhibitory cell sur-
face receptors; in fact, the paradigm for positive and nega- Comparative Studies of Natural Killer Function. NK cells
tive signaling via such receptors began with these cells60,161; were detected in Xenopus by in vitro 51Cr-release assays.
however, activating and inhibitory receptors (often paired) Splenocyte effectors from early thymectomized frogs spon-
are conserved throughout vertebrates and invertebrate deu- taneously lyse allogeneic thymus tumor cell lines that lack
terostomes, and are expressed in hematopoietic cells of all MHC antigen expression.166 This activity is increased after
types. In NK cells, stimulation of the activating receptors, the injection of tumor cells or after treating the splenocytes
which associate with proteins having an intracellular ITAM in vitro with mitogens, suggesting lymphokine activation
(CD3ζ, DAP12, or DAP10 conserved at least to the level of of the killers. Splenocytes isolated with an anti-NK mono-
bony fish162), results in killing of target cells. Inhibitory clonal antibody (mAb) revealed large lymphoid cells with
signaling receptors all possess cytoplasmic ITIMs, which distinct pseudopodia. Immunohistology indicated that
recruit phosphatases and generally are dominant over the each anti-NK mAb routinely labeled cells within the gut
activating receptors. These receptors fall in two catego- epithelium but NK cells were difficult to visualize in spleen
ries IgSF and C-type lectin group V(II). In general, NKRs sections.167
recognize MHC class I molecules of either the classi- In amphibians, NK cell studies are especially interest-
cal or nonclassical type, the latter sometimes encoded by ing because of natural experiments done by nature (ie, the
viruses.163 absence or low levels of MHC classical class I during larval
life of some species like Xenopus).168 They are bonafide NK
General Evolution of Natural Killer Receptor Families. As cells, distinct from T cells, as they fail to express TCR Vβ
mentioned, NK cells in mammals can use different types transcripts. NK emerge in late larval life, 7 weeks postfer-
of receptors, even encoded by different gene families, IgSF tilization, which is about 2 weeks after the time when cell
(KIR) in primates and C-type lectins (Ly49) in rodents. surface class I can be detected. The proportion of splenic
A few receptors are conserved but most others are highly NK cells remains very low until 3 to 4 months of age, but
variable. Very few families show conservation of domains by 1 year there is a sizeable population. Therefore, NK cells
throughout the jawed vertebrates.163 When dealing with the fail to develop prior to MHC class I protein normal expres-
origin of these genes in invertebrates, one has to imagine sion (at least NK cells of the type that can be measured with
under what pressures they evolved. The question remains as these assays and with NK cell–specific mAbs) and do not
to whether NK cells, or NK-like cells, preceded the emer- contribute to the larval immune system, whereas they do
gence of T- and B-lymphocytes. provide an important backup for T cells in the adult frog by
As described in the introduction, NKRs are the most contributing to antitumor immunity.
rapidly evolving molecular component of the gnathostome NK cells have also been described in a number of teleost
immune system. Most ligands for these diverse NKRs are fish with the most in-depth studies in catfish, in which there
MHC class I molecules, or molecules of host or pathogen are clonal likes of cytotoxic cells,167 some that clearly lack
origin related to MHC class I. The KIR families are diver- TCR expression.169 A subset of the fish NK cell bears a high-
gent, as very few genes are conserved even between chim- affinity FcR that can be utilized for antibody-dependent cel-
panzees and humans, and there are different numbers of lular cytotoxicity.170 Other subsets of NK cells spontaneously
genes in KIR haplotypes within a species. Humans have two kill allogeneic targets. Further study of these cloned lines
KIR haplotypes A and B, one encoding a large number of may provide much needed information on NK function in
activating receptors and the other very few.23 It is speculated phylogeny.
that the haplotypes are under balancing selection within the
population both for defense against virus and for maternal/ Phylogeny of Natural Killer Lectins
fetal interactions. By contrast, CD94 and NKG2 receptors Besides the well-described Ly49 family of receptors in rodents
are conserved throughout mammals. So whereas receptors (and horse) and the CD94 and NKG2 families in all mam-
for polymorphic class I molecules are divergent, those for mals, other mammalian NKRs are of interest. Studies in
nonpolymorphic or stress-induced class I molecules are mammals have shown that some NKC-encoded lectin-like
relatively conserved (despite the fact that their ligands Qa1 receptors in the Nkrp-l family can recognize other lectin-like
and HLA-E are not orthologous). NK cells play other impor- molecules, termed Clr, also encoded in the NKC.171 Having
tant roles in other innate immune responses, for example linked loci encoding receptor-ligand pairs suggests a genetic

strategy to preserve this interaction; perhaps the CD94 ho- There are other bony fish–specific IgSF NKR families.
mologs of invertebrates are genetically linked to genes en- One family, the novel immune-type receptors (NITRs),179
coding their ligands. In addition, as described in the follow- have one or two Ig domains with a charged residue in the
ing, the close genetic linkage of receptor and ligand genes is TM and could therefore be associated (by analogy) to an
a common theme in “histocompatibility reactions” through- ITAM DAP12 equivalent, and indeed was shown to inter-
out the animal and plant kingdoms described below. act with mammalian DAP12.162 Like NKp30, the NITR
In chickens, a single gene similar to CD94/NKG2 is en- N-terminal V domain is also of the VJ type. NITRs were
coded in a region syntenic with the mammalian NKC.172 It originally believed to be part of the LRC, but this was
is linked to CD69, another C-type lectin also encoded in the shown to be unlikely upon further analyses. NITRs can be
NKC of mouse/human. Chickens and quail MHC encode expressed by cells of the hematopoietic lineage, presumably
two C-type lectin NKR B-lec and B-NK, the latter being lymphocytes. NITRs have been found in all bony fish, with
most similar to NKPR1.173 Other C-type lectins are found in rapid contraction and expansion of the gene family, with the
the RFP-y locus, one that is also most similar to NKRP1.174 majority of proteins predicted to have inhibitory ITIMs in
These linkages give credence to the idea that the NKC and their cytoplasmic tails. In zebrafish, there are many NITR
MHC were syntenic in early jawed vertebrates (see the fol- genes that group into 12 distinct families.180 An extreme
lowing). While C-type lectin genes with some similarity to level of allelic polymorphism is apparent, along with hap-
mammalian NKR have been detected in ectothermic ver- lotype variation and family-specific isoform complexity. By
tebrates, no convincing orthology or synteny to the NKC contrast, only 11 related genes encoding distinct structural
has been found to date. In mammals including marsupials, forms have been identified in the channel catfish, and the
NKG2D is conserved, and CD94/NKG2 is found in mam- relatively small number of genes allowed functional studies
mals and birds (as well as NKPR1).163 If such C-type lectin to be performed. Additionally, taking advantage of the abil-
NKR are found in the future in cold-blooded vertebrates, ity to grow clonal lines of catfish hematopoietic cells, one
they will have to be studied in functional assays. granular cell line lacking all markers of B/T cells was shown
Given the apparent lack of MHC class I and class II in to express several NITRs. Expressed NITRs were fused to
agnathans and their convergently acquired adaptive im- an ITAM-containing motif and transfected into a T-cell
mune system (see the following), it is difficult to envisage hybridomas line with a nuclear factor of activated T cells
how NK cells with receptors of any type might function in (NFAT) promoter, and the specificity for particular catfish
these animals. It should be mentioned, however, that se- MHC alleles was maintained.181 Subsequent crystal struc-
quence similarity might be difficult to detect for an MHC ture analysis showed the NITR V domains to form dimers,
peptide-binding region (PBR), given the rapid rate of evo- much the same as Ig/TCR heterodimeric V regions. Thus,
lution of this gene family. Furthermore, it would not be the sequence analysis (ITAM/ITIM), signaling properties,
shocking if there were NK cells with ligands encoded by involvement in cytotoxicity, and recognition of MHC mol-
other gene families—in mammals, ligands for some acti- ecules identify the NITRs as excellent candidates for NKRs
vating NKRs have not been identified. It would be of in- in bony fish. Furthermore, the work serves as a paradigm for
terest to study the non-VLR–expressing lymphocytes in study of potential NKRs when homology and/or conserved
agnathans (if such cells exist) for their killing potential or synteny are lacking or ambiguous.
gene expression. IgSF inhibitory receptors usually form larger families of
molecules in comparison to activating receptors. This func-
Phylogeny of Immunoglobulin Superfamily tion can be devoted to two distinct families of receptors,
Natural Killer Receptors giving another example of the extremely rapid evolution of
IgSF-activating receptors have been recognized from carti- these molecules. There are many ITIM-carrying IgSF inte-
laginous fish onwards with a convincing activating NKp44 gral membrane receptors across the classes of vertebrates,
homolog first found in carp (called NILT), but with no and they seem to have had independent histories as it is
functional data.175 Subsequently, this family was found in difficult to convincingly detect orthologous genes between
other bony and cartilaginous fish and definitive orthology species. This is especially true of multigene families in fish,
to mammalian NKp44 was shown. The activating receptor with members equipped with possible ITIMs, including
NKp30 is also conserved, with orthologs found to the level the teleost NITR and LITR, and bird CHIR and CD300L.163
of cartilaginous fish176 ; in addition, as described in the MHC Several members of these families can be expressed on NK
section, there are V(J) genes within the frog MHC called cells, but expression studies are in their infancy in fish. It
XMIV that are ancient homologues of NKp30 and may be is sometimes difficult to distinguish FcR families from NK
NKRs of both activating and inhibitory types.177 The ligand KIR-like domains, and both FcR and KIR seem to stem from
for NKp30 has recently been uncovered, the stress-induced a same lineage. Given the role of the FcR binding to bonafide
molecule B7H6.178 Interestingly, there is a perfect correla- antibodies and conferring specificities to cells of the innate
tion between the presence or absence of NKp30 and B7H6 arms of the immune system, it is likely that these molecules
in the vertebrate line, with both genes lacking in birds and will be restricted to jawed vertebrates. The KIR activity that
bony fish but present in all other gnathostome classes.176 can incorporate pathogen and virus recognition may be
Furthermore, phylogenetic trees suggest a close relationship more primitive, but the ancestry of KIR is not well under-
between NKp30 and NKp44, consistent with their ancient stood, and the bonafide KIR family seems to be restricted
origins within the vertebrate line. to primates.163

Genes encoding the classical FcRs on the long arm of a large family of the previously described teleost NITR, and
human chromosome 1 (1q21–23) are linked to other FcR-like the MHC-linked XMIV in Xenopus. These molecules are
genes. A large multigene family, which includes genes encod- examined in more detail in the conclusion. Many IgSF pro-
ing the Fc γR and the NK cell Ig-like receptors, is located in teins expressed in the immune systems are also expressed
the LRC (human chromosome 19q13). This region could in in nervous systems where the signaling cacscades may be
fact be paralogous to 1q23 and may even have been originally conserved. This selection of IgSF domains, in two different
associated with the MHC (see the following). These families systems in which homologies are found in molecules with
belong to a larger class of activating or inhibitory receptors. quite different functions, may reveal adaptation capacities
Their phylogenetic conservation in birds, amphibians, and and constraints exerted on surface receptors.187
bony fish suggests a biologic importance even though the
size of the families, their expression pattern, and the spe-
VERTEBRATE ADAPTIVE IMMUNITY
cific nature of the receptors vary greatly among species.182 In
several cases, a commitment to a task in the immune system Not long ago, it was believed that only jawed vertebrates had a
may not be conserved among homologous members, and true adaptive immune system. From the previous discussion
the evolutionary fate of the family will be probably affected. of invertebrate immune responses, clearly mechanisms exist
Comparison of key residues in the domains may suggest a to generate high levels of immune diversity—even at the so-
possible common involvement in MHC recognition for the matic level—one of the hallmarks of an adaptive response.7
two families recently discovered in birds (CHIR)183 and the As described in the introduction, many in the field agree
teleosts (LITR).184 Other families were generated within a that the boundary between innate and adaptive immunity is
single class or even within a single order of vertebrates (eg, artificial, and it may not be a useful dichotomy when study-
the KIR described previously). The relationships of KIR ing immune responses in diverse organisms.61,62 Despite this
with many other multigene families such as IpLITR or NITR reluctance to exclusively classify systems as innate or adap-
remain to be explored. What was the scenario that led to tive, some features clearly fall into the latter category, such as
the present mammalian situation? If the fish observation on clonal expansion of uncommitted lymphocytes and specific
potential MHC binding holds true, the IgSF type of receptor memory. These conditions are not fulfi lled for the DSCAM,
seems to be the most primitive NKR.163 FREP, 185/333, or any other invertebrate systems described
previously, but of course we must be open to new mecha-
Other Immunoglobulin Superfamily Families to nisms besides the conventional outlook of adaptive immu-
Explore Further nity. Additionally, such adaptive immunity arose in concert
In more primitive vertebrates, the physical or genetic linkage with the emergence of lymphocytes in the lower chordates,
of relatively large IgSF families is well documented in the te- as these cells clearly are the major players; we will discuss
leost NITR but not yet elucidated in the case of other interest- this in the conclusion.
ing families in prochordates like the VCBP. In the sea urchin
genome, many IgSF await a complete analysis and will cer- Immunoglobulins
tainly contribute to a better understanding of the evolution
and origin of Ig/TCR.48 Large families of LRR-IgSF in am- A typical Ig molecule is composed of four polypeptide chains
phioxus185 could perhaps represent interesting intermediaries (two heavy [H] and two light [L]) joined into a macromo-
in the genesis of either VLR in agnathans or Ig/TCR in gna- lecular complex via several disulfide bonds (Fig. 4.7). Each
thostomes (see Fig. 4.10). In hagfish, the discovery of leuko- chain is composed of a linear combination of IgSF domains,
cyte expressed receptors agnathan-paired receptors (APARs) and almost all molecules studied to date can be expressed in
revealed what might have been a precursor of Ig or TCR.186 secreted or transmembrane forms.
APARs resemble Ag receptors and are expressed in leukocytes
and predicted to encode a group of membrane glycoproteins Immunoglobulin Heavy Chain Isotypes
with organizations characteristic of paired Ig-like receptors. Like all other building blocks of the adaptive immune
Based on their transmembrane regions, APAR-A molecules system, Ig is present in all jawed vertebrates (see Fig. 4.7).
are likely to associate with an adaptor molecule with an ITAM Consistent with studies of most molecules of the immune
and function as activating receptors. In contrast, APAR-B system, the sequences of IgH chain C region genes are not
molecules with an ITIM are likely to function as inhibitory well conserved in evolution and insertions and deletions in
receptors Thus, the APAR gene family has features charac- loop segments occur more often in C than in V domains.
teristic of paired Ig-like receptors. APAR V domains have a As a consequence, relationships among non-μ isotypes (and
J region and are more closely related to those of TCR/B-cell even μ isotypes among divergent taxa) are difficult to es-
receptor (BCR) than any other V-type domain identified tablish.188 Despite these obstacles, the field has developed a
to date outside of jawed vertebrates. Thus, the extracellular working evolutionary tree among all of the isotypes.
domain of APAR may be descended from a VJ-type domain
postulated to have acquired recombination signal sequences Immunoglobulin M
(RSS) in a jawed vertebrate lineage (see Fig. 4.10). IgM is present in all jawed vertebrates and has been as-
In jawed vertebrates, three such receptor families with sumed to be the primordial Ig isotype. It is also the isotype
VJ-type domains have been identified: a small family of expressed earliest in development in all tetrapods; until
mammalian proteins known as signal-regulatory proteins, recently, it was believed to be the case in fish as well, but

A MAMMALIAN Ig ISOTYPES
IgM IgD IgG IgE IgA Camelid IgG
–Primordial Ig –Primordial Ig –Derived from –Derived from –First found –Derived form of
–Transmembrane form –Highly plastic in evolution ‘IgY-like’ ancestor ‘IgY-like’ ancestor in reptiles bona fide IgG
defines B cells –Lost in several species –First appeared in –Dimer in secretions, –’Single-chain V’
–Pentamer/hexamer in amphibians monomer in sera convergent with
–Known as ‘IgW’ in
most vertebrates –TM region conserved –Mucosal Igs in a similar form in
several fish
–Tetramer in teleosts in evolution; involved other vertebrate cartilaginous
–Secreted form associated fish IgNAR
–Secreted monomer in in memory via convergence
with granulocytes and
cartilaginous fish (e.g. IgX in frogs)
inflammation in fish and
(associated with human
affinity increase) –H chain lacks V region in
–Associated with J chain bony fish
in all but teleosts and
shark monomeric IgM
Camelid
B IgM IgD (IgW) IgG (IgY) IgE IgA IgG/IgNAR IgZ/T
MAMMALS or
(IgY) (IgY)
BIRDS
or
(IgY)
REPTILES
(IgY) (IgYʹ, IgF) IgX1
or
AMPHIBIANS
(IgZ/T)4
2 3
BONY FISH
5 6
7
or
CARTILAGINOUS
FISH (IgM1gj)8
100 IgE
IgMgj 400 IgY
400
100
IgG Camelid IgG
IgA IgM/IgD/W
300
IgX 350 400 IgNAR
IgZ/T
FIG. 4.7. Immunoglobulin (Ig) Isotypes in the Jawed Certebrates. A: Mammalian isotypes and their relationships to Igs in vertebrates from
other classes. Each oval represents an Ig superfamily C1 domain. IgD is shown in two forms, mouse (left) and human (right).2 B: Major Ig
isotypes in all vertebrates. The bottom panel displays the approximate divergence times of all isotypes. IgM/D/W was found at the inception
of adaptive immunity. 1, IgX is in the IgA column because it is preferentially expressed in the intestine, and IgA seems to have been derived
from an IgX ancestor; IgX seems to have been derived from both IgM and IgY ancestors; 2, secreted IgM in teleost fish is a tetramer, and
the transmembrane (TM) form only has three C domains; 3, the teleost fish IgD H chains incorporate the μC1 domain via alternative splicing
in the TM form, and the secretory form does not have a V region and does not associate with L chains; 4, the new bony fish isotype, IgZ/T,
may not be found in all fish species; 5, the secreted form of shark/skate IgM is present as a pentamer and monomer at approximately equal
levels; 6, the TM form of IgW has four C domains; 7, a major TM form of IgNAR has three C domains, and IgNAR is related to camelid IgG by
convergent evolution; 8, no TM form has been found (to date) for IgM1gj. The bottom panel displays the approximate divergence times of all
isotypes. IgM/D/W was found at the inception of adaptive immunity.

this view has changed (see the following). The secretory μ become 7S producers or whether there are lineages of 19S-
H chain is found in all vertebrates, usually consists of one V and 7S-producing B cells is an open question. One work-
and four C1 domains, and is heavily glycosylated. H chains ing hypothesis is that J chain expression is important for
associate with each other and with L chains through disul- regulating whether a B cell makes 19S or 7S Ig, but of course
fide bridges in most species, and IgM subunits form pen- that could be at the lineage level or the switch level (see the
tamers or hexamers in all vertebrate classes except teleost following).
fish, which form tetramers.189 The μ CH4 domain is most
evolutionarily conserved, especially in its C-terminal re- Immunoglobulin M1gj
gion, whereas the CH2 domain evolves at the fastest rate.188 Nurse shark Ginglymostoma cirratum expresses an IgM
There are several μ-specific residues in each of the four CH subclass in neonates.196 The V H gene underwent V-D-J re-
domains among vertebrates suggesting a continuous line arrangement in germ cells (“germline-joined” or “gj,” see
of evolution, which is supported by phylogenetic analyses. the following). Expression of H1gj is detected in primary and
Like TCR TM regions, μ TM regions are also well con- secondary lymphoid tissues early in life, but in adults only in
served among sharks, mammals, and amphibians, but the the primary lymphoid tissue, the epigonal organ (see the fol-
process by which the Ig TM messenger RNA is assembled lowing). H1gj associates covalently with L chains and is most
varies in different species. In all vertebrate classes except te- similar in sequence to IgM H chains, but like mammalian
leosts, the μ TM region is encoded by separate exons that IgG it has three rather than the typical four IgM constant
are spliced to a site on μ messenger RNA located approxi- domains; deletion of the ancestral IgM second domain thus
mately 30 basepairs from the end of the CH4-encoding exon. defines both IgG and IgM1gj. Because sharks are in the oldest
By contrast, splicing of teleost fish μ messenger RNA takes vertebrate class known to possess antibodies, unique or spe-
place at the end of CH3 exon.190 In holostean fish (gar and cialized antibodies expressed early in ontogeny in sharks
sturgeon), cryptic splice donor sites are found in the CH4 and other vertebrates were likely present at the inception of
sequence that could lead to conventional splicing, but in the the adaptive immune system. It is suggested that this isotype
bowfin there is another cryptic splice donor site in CH3.191 interacts either with a common determinant on pathogens
The TM region itself is interesting as it is the only one that or a self–waste product.
does not contain a residue capable of making an ionic bond
with the ITAM-containing molecule (in this case, Ig-α and Immunoglobulin New Antigen Receptor and New
Ig-β.) As mentioned, some modifications apparently related Antigen Receptor-T-Cell Receptor
to the particular environment were noticed in the Antarctic A dimer found in the serum of nurse sharks and so far re-
fish Trematomus bernacchii. There are two remarkable in- stricted to elasmobranchs, IgNAR is composed of two H
sertions, one at the V H-CH1 boundary and another at the chains each containing a V domain generated by rearrange-
CH2-CH3 boundary; the latter insertion results in a very long ment and five constant C1 domains.197 IgNAR was originally
CH2-CH3 hinge region. Rates of nonsynonymous substitu- found in sera, but TM forms exist as ccomplementary DNA
tions were high in the modified regions, suggesting strong and cell-surface staining is detected with specific mAbs. The
selection for these modifications. These unusual features single V resembles a fraction of camel/llama (camelid) IgG
(also unique glycosylation sites) may permit flexibility of that binds to antigen in a monovalent fashion with a single
this IgM at very low temperatures.192 V region, but it clearly was derived by convergent evolution.
It has been known for a long time that in all elasmo- In phylogenetic trees, NAR V domains cluster with TCR and
branchs, IgM is present at very high amounts in the plasma L chain V domains rather with that V H. A molecule with
of cartilaginous fish and that it is found in two forms: mul- similar characteristics has also been reported in ratfish, al-
timeric (19S) and monomeric (7S).193 It is unlikely that the though it was independently derived from an ancestral Ig
two forms are encoded by different gene clusters because like the camelid molecule emerged from bonafide IgG.198
1) peptide maps are identical; 2) early work by Clem found IgNAR V region genes accumulate a high frequency of so-
the sequences of the cysteine-containing tail of 19S and 7S matic mutations (see the following).
H chains to be identical; and 3) all identified germline VH The crystal structure of a Type I IgNAR V regions
families are represented for the 19S form.194 Although most showed that, in contrast to typical V regions, they lacked
studies (but not all) reported that 19S and 7S are not dif- CDR2 and had a connection between the two IgSF sheets
ferentially regulated during an immune response, in a re- much like an IgSF C domain.199 The domain wraps around
cent study, the 19S response wanes over time and a stable its antigen (hen egg lysozyme [HEL]), with the CDR3 pen-
7S titer is maintained for periods of up to 2 years after im- etrating into the active site of the enzyme. The structure of
munization.195 In addition, antigen-specific 7S antibodies a Type II V region has a disulfide bond between CDR1 and
observed late in the response have a higher binding strength CDR3 that forces the most diverse regions of the molecule
than those found early, suggesting a maturation of the re- to form raised loop, similar to what has been described for
sponse, also generally at odds with the previous literature. camelid V domains. In total, the differential placement of
Finally, when specific antibody titers were allowed to drop, disulfide bonds forces major changes in the orientations
a memory response was observed that was exclusively of of CDR1 and CDR3, and provides two major conforma-
the 7S class. This work has shown that a “switch” indeed tions for antigen binding.200 The structure of a Type II NAR
occurs in the course of an immune response; whether the bound to HEL showed that it also interacted with the active
“switch” is due to an induction of the 19S-producing cells to site of the enzyme.

While analyzing the TCRVδ repertoire in nurse sharks, monotremes with properties similar to NAR-TCR.202 In this
an entirely new form of this chain, which encodes three case, there are also two V domains, but in marsupials one
domains, V-V-C, was detected.201 The C is encoded by the (proximal to the membrane) is germline-joined, and only
single-copy Cδ gene, and the membrane-proximal V is en- the membrane-distal domain undergoes rearrangement. In
coded by a Vδ gene that rearranges to the DJδ elements. monotremes, both of the V regions undergo rearrangement,
The membrane distal V domain is encoded by a gene in like for the NAR-TCR. This new TCRδ locus is preferen-
the NAR family, found in a rearranging VDJ cluster typical tially expressed early in development. This type of TCR is
of all cartilaginous fish Ig clusters. The NAR-TCR V genes, described in more detail in the TCR section.
unlike IgNAR V genes which have three D segments, only
have a single D region in each cluster. The particular Vδ Immunoglobulin R/Immunoglobulin New Antigen
loci linked to each NAR-TCR gene—called NAR-TCR– Receptor/Immunoglobulin W/Immunoglobulin
supporting Vδ —encode a cysteine in CDR1 that likely X/Immunoglobulin D
makes a disulfide bridge with the NAR-TCR V domain. All elasmobranchs studied to date have another isotype
The J segment of the rearranged NAR-TCR V gene splices called IgW. It was probably discovered long ago in skates as
at the RNA level directly to the supporting Vδ segment, a non-IgM secreted isotype called IgR, but no protein se-
which has lost its leader exon (Fig. 4.8). This organization quence of this molecule has been obtained for confirma-
likely arose from an IgNAR V cluster that translocated tion.203 Subsequently, an Ig gene was discovered in skates en-
to the TCRδ locus upstream of a Vδ gene segment. After coding a three-domain molecule with an unusual secretory
modifications of the supporting Vδ genes, this entire V-V tail that was named IgX (not to be confused with another
gene set duplicated and diverged several times in different isotype with that name in amphibians).204,205 A high molecu-
species of sharks. About 25% of the expressed nurse shark lar weight (MW) species detected by northern blotting with
TCR δ repertoire is composed of this TCR (encoded by 15 an IgW probe suggested that there might be a longer form of
to 20 V-V genes in this species), and have proposed that the this isotype, subsequently shown to be true in the sandbar
typical γ /δ TCR acts as a scaffold upon which sits the single shark (IgW), nurse shark (IgNARC—it was so-named be-
chain NAR V. cause the C domains had highest similarity to IgNAR C do-
Our interpretation is that, true to the proposal that γ /δ mains), and skate.206,207 It was originally believed that sharks
TCRs interact with free antigen, the NAR V is providing only expressed the long (seven-domain) IgW form, but they
a binding site that can interact with antigen in a different were later shown to have both secretory forms; the reason
way than conventional heterodimeric Vs. Thus, this is the for the discrepancy was shark-to-shark variation in expres-
first case in which a particular V region family has been sion of the short form, for unknown reasons. The major IgW
shown to be associated with a BCR and TCR; in the case of TM form, like IgM, is composed of five domains, but vari-
the BCR, the function likely resides within the Fc portion of ants with three domains—like the secretory forms—were
IgNAR and for the TCR the function (cytokine secretion, also detected.208 Very little is known about the function of
killing) lies within the T cell itself. Interestingly, a second this isotype, as IgW-specific mAbs have not been generated
TCRδ chain locus has also been described in marsupials and as they have for IgM and IgNAR.
IgH amphibian Mammal Vn Dn Jn Cμ Cγ C others

TRANSLOCON
IgH bony fish Vn Dτn Jτn Cτ Dμn Jμn Cμ Cδ
IgH cartilaginous fish ( V D D J Cμ/x )(

n
V D D D J CNAR ) n
CLUSTER
IgL bony fish Mammal Vλ, κ, σn Jλ, κ, σn Cλ, κ, σ TRANSLOCON
IgL bony, cartilaginous fish ( Vλ, κ, σ Jλ, κ, σ Cλ, κ, σ )( n

Vσ’ Jσ’ Cσ’ ) n
CLUSTER:
Some germ-line-joined
TCRβ cartilaginous fish Mammal Vn Dn Jn C TRANSLOCON
TCRγ cartilaginous fish Mammal Vn Jn C TRANSLOCON OR CLUSTER
TCRα/δ cartilaginous fish Mammal Vαn Vδn Dδ1–3 Jδn Cδ Jαmany Cα TRANSLOCON
TCRμ marsupial
(form of δ)
( V D J ) n
V D J Cμ DOUBLE V (ONE REARRANGING):
Cluster + germline-joined
NAR-TCR cartilaginous fish
(form of δ)
( V D J Vδ ) n
Vn D Jn Cδ DOUBLE V (TWO REARRANGING):
Cluster + translocon
FIG. 4.8. Organization of Immunoglobulin (Ig) Heavy and Light (L) Chain and T-Cell Receptor (TCR) Genes in All Jawed Vertebrates. All cold-
blooded gnathostomes except bony fish have three L chain isotypes, κ, λ, and σ; mammals have κ and λ; and birds have only λ. RSS are found
at the 3′ end of V segments, 5′ end of J segments, and on both sides of D segments. Marsupial μ TCR is related to Ig in its V regions and TCRδ
in its C region (and is not related to the μ of IgM).

IgW was thought to be a dead-end isotype in the cartilag- found between the VH and Cζ /C μ exons (see Fig. 4.8).
inous fish, but a homologue was found in the lungfish.209 It IgZ/T is a five-domain H chain that associates with L chains
also is present in two secreted forms, one with eight domains (see Fig. 4.7). The authors of the zebrafish paper proposed
and the other, like in the elasmobranchs, with three do- that lymphocytes bearing IgZ may be B1-cell equivalents,
mains (unfortunately the secretory tail was not sequenced, but a preliminary VH repertoire analysis did not suggest
and the TM form was not studied.) An Ig isotype was found that unique sets of V regions are expressed on IgZ com-
in Xenopus tropicalis most related to the lungfish IgW.210,211 pared to IgM. IgT is not preferentially expressed over IgM
Computer searches of databases for the X. tropicalis genome early in trout development.
project uncovered a new isotype flanked by the IgM and IgX Recent work has shown that IgT is the mucosal Ig of trout,
genes at the IgH locus. The deduced amino acid sequence with preferential expression in the intestine and skin.218
obtained from the exons on the genomic scaffold suggests Mucosal immunization with the parasite Toxoplasma gondii
a nine-domain molecule. The N-terminal C domains and showed a preferential induction of specific IgT. While all
the TM regions are most similar to mouse and human IgD teleosts have lost the J chain in evolution,219 biochemical
regions, and its genomic location also suggest that it is an analysis of IgT showed that it contained a secretory piece
IgD equivalent. Thus, these new data reveal that like IgM, derived from the poly-Ig receptor (pIgR). Furthermore,
IgW/D is an isotype that was present at the emergence of all like mammalian dimeric IgA, IgT coats trout commensal
extant vertebrate taxa. bacteria, presumably acting as a “firewall” preventing com-
Catfish IgD was the first member of this Ig class found mensals from breeching the blood–brain barrier. In sum,
outside of mammals,212 but this was not well accepted until IgT, while displaying unique evolutionary features, has at-
the isotype was found in all vertebrate classes. It is found tained almost all of the characteristics of IgA via convergent
attached to FcR on a subset of myeloid cells, and the secreted evolution.
form does not contain a V region as the leader is spliced to the
CH2 domain in plasma cell RNA.213 This strongly suggests Other Isotypes Related to Immunoglobulin G,
that the IgD is acting as a PRR, perhaps interacting with a Immunoglobulin E, Immunoglobulin A, and the
conserved region of a pathogen leading to innate immunity. Class Switch
The work on catfish prompted a reappraisal on the role of se- Other isotypes consist of four C domains in nonmamma-
creted IgD in humans.214 Like the fish IgD, the human IgD is lian vertebrates including Xenopus IgY and IgX, non-μ iso-
bound to a subset of myeloid cells, which are activated upon types of Rana, IgY of axolotl, and IgA and IgY of birds.194,220
IgD cross-linking. The receptor on human cells has not been In Xenopus, IgY is thymus-dependent; IgM and IgX are
identified, but this new work argues for strong evolutionary not, although thymectomy impacts specific IgM antibody
conservation of the function of secreted IgD. production (ie, antigen-specific IgM can be produced, but
An interesting feature of the IgD/W locus is its high plas- there is neither an increase in affi nity after immunization
ticity in evolution, both in terms of the number of domains nor elicitation of plaque-forming cells). IgM and IgX plasma
in different fish species and the plethora of splice variants cells are abundant in the gut,221 whereas IgY is expressed pri-
found, at least in cartilaginous fish.194,215 In sharks, two of marily in spleen. Axolotl IgM is present in the serum early
the C domains were derived approximately 250 million during development and represents the bulk of specific an-
years ago by a tandem duplication event, and there was a tibody synthesis after antigenic challenge.222 In contrast to
Xenopus-specific, two-domain tandem duplication event as Xenopus IgY, the axolotl ortholog appears late in develop-
well. Within teleost fish, the number of C exons for this iso- ment and is relatively insensitive to immunization. From 1
type is different in various species and the secreted and TM to 7 months posthatching, axolotl IgY is present in the gut
forms are encoded by different loci in the catfish. In addition epithelium associated with a secretory component.223 IgY
to the splice variants previously described in the cartilagi- progressively disappears from the gut and is undetectable in
nous fish and lungfish, in teleosts the IgM C1 domain exon the serum of 9-month-old animals. Thus, axolotl IgY, like
is spliced into the IgD transmembrane transcript.212 Even Xenopus IgX and trout IgT, may be analogous to mamma-
in mammals, there are different numbers of C domains in lian IgA. Xenopus IgX seemed to be most similar to IgM,
different species, and even exons that have emerged quite but as more sequences have become available it does seem
recently in evolution (see Fig. 4.7). It is our impression that to be orthologous to IgA, obviously consistent with its as-
this is the Ig locus that evolution “plays with,” perhaps using sumed function (Criscitiello, personal communication). In
it for different functions in vertebrate taxa. The beginning addition, IgX was formerly believed to be derived from IgM,
of understanding the function of the secreted form of IgD but recent data suggest that the N-terminal C-domains were
is a triumph for comparative immunology; likely the evo- derived from an IgY ancestor and the C-terminal domains
lutionary approach will reveal the function of the TM form from IgM.224
as well. The TM and cytoplasmic domains of Xenopus TM IgY
share residues with avian IgY and mammalian IgG and IgE,
Immunoglobulin Z/T suggesting that mammalian/avian isotypes share a common
A third, novel bony fish isotype was uncovered in screens ancestry with amphibian IgY.225 This homology is especially
of the EST and genomic databases called IgZ in zebrafish 216 interesting because studies in mice suggested that the IgG
and IgT in trout.217 Its genomic organization parallels the cytoplasmic tail is the central molecular element promoting
TCR α /δ locus in that the IgZ/T D, J, and C elements are rapid memory responses of plasmablasts.226,227 The motif in

the IgY/G tail is also found in other receptors and seems to addition, a third L chain, σ, originally described in Xenopus
recruit PI3kinase, which amplifies the signal derived from as a dead-end isotype, is in the ancestral clade and is present
Ig-α and -β. Finally, another Xenopus isotype was discovered in all cold-blooded vertebrates.231
from the databases, called IgF or IgY′ (because it is closely Elasmobranchs have four L chain isotypes (Type I, II, III,
related to IgY, probably duplicating within the Xenopus lin- and IV), and the combined data suggest that they are pres-
eage). Like other isotypes described in this section, it only ent in all chodricthyans.231 Type IV is the ortholog of the
has two C domains, but rather than alternative splicing, the unusual Xenopus σ L chain, and Type I is a dead-end variant
gene is organized in such a manner.211 There has been no of this isotype called σ′ or σ-cart (for cartilaginous fish).
biochemical identification of IgF. Type II L chains are of the λ isotype and have been cloned
CSR results from fusing switch regions upstream of the from all groups of cartilaginous fish; all genes of this isotype
μ gene and another 3′ isotype gene, accompanied by the are “germline-joined.” The type III is clearly κ-like, at least
deletion of the intervening sequences. Among vertebrates, in the V region and RSS orientation (κ is the only antigen re-
mammalian and bird class switch regions are GC-rich ceptor gene in which the V is associated with a 12mer RSS).
and contain tandem repeats in which certain motifs, such These relationships are more noticeable in the V sequences,
as TGGGG, GGGGT, AGCT, and GGCT, are abundant. which have defining characteristics such as CDR length;
Because of the GC richness of these regions, transcription although, the C sequences also fall into the same clusters,
generates stable R loops, which provide single-stranded they do so with much less phylogenetic support. Different
substrates for activation-induced deaminase (AID) activ- elasmobranch species express the L chain isotypes prefer-
ity. The fi rst comparative studies on switch were done in entially (eg, κ in nurse sharks, σ in horned sharks, and λ
the amphibian Xenopus where the switch regions are not in sandbar sharks); this pattern of expression may be due
GC-rich but adenine-thymine (AT)-rich and cannot form to expansions/contractions of the different isotype genes in
R loops.228 Replacement in mouse of the switch region with various elasmobranch lineages.
a Xenopus switch region showed that it mediated efficient Almost all of the L chains in bony fish can be categorized
CSR and that the junctions were associated with the short as either κ or σ, despite the large number of gene expan-
palindromic AGCT motifs, already recognized as the main sions and contractions in this group.232 As expected, recent
component in Xenopus switch.229 As predicted from the data have shown that λ exists in some species, consistent
absence of R loops, the Xenopus switch region supported with the work in sharks.233 Genes in different species can
recombination in both orientations. The breakpoints be found either in the cluster type organization, translo-
were located in the AGCT palindrome-rich region of the con, or some intermediate type, and this organization is
switch box. Other motifs have been identified in the other especially well suited to allow for receptor editing, either to
switch regions, such as in IgX and chicken isotypes; all of generate the repertoire or as a consequence of binding to
these correspond to the DGYW hotspot consensus. AID- self-ligand.234,235
mediated deamination in the context of these motifs may mAb studies suggested the existence of three Xenopus
be the conserved major event in the initiation of CSR. As L chain isotypes of 25, 27, and 29 kD with heterogeneous
described in the following, until recently it was believed two-dimensional gel patterns and preferential association of
that CSR fi rst appeared in amphibians, but recent data have some L chain isotypes with IgY H chains.236 Indeed, three
uncovered a precursor to switch in cartilaginous fish (but Xenopus L chains genes have been isolated: ρ (now κ), σ,
not teleosts). Thus, we should add another conserved fea- and λ. Only one C gene is present in the ρ locus, and it en-
ture of adaptive immunity that appears to have arisen in codes the most abundant L chain. The V and J RSS are of the
the early gnathostomes.230 κ-type and the five identified J segments are nearly identi-
As mentioned, a single gene can encode different Ig cal. The locus is deleted, like mammalian κ, when the other
forms, like for duck IgY, cartilaginous fish IgW, and cam- isotype genes are rearranged. Southern hybridizations with
elid IgG loci. It has been suggested that the two avian IgY genomic DNA from different animals showed V L sequences
short and long forms could be the functional equivalents to be both diverse and polymorphic. The λ Xenopus laevis
of both IgE and IgG, respectively; the same may be true of L chain isotype predicted from the biochemical studies con-
the cartilaginous fish IgW short and long forms with two sists of six distinct V L families. In the σ locus, the J segment
and six domains, respectively. Also as mentioned, IgF or has an unusual replacement of the diglycine bulge by two
IgY′ also falls into this category but not through alternative serines. The Rana major L chain type has an unusual intra-
splicing.194,211 chain disulfide bridge that is seemingly precludes covalent
association of its H and L chains.237
Immunoglobulin Light Chains Two L chain types were identified in reptiles. Chickens
L chains can be classified phylogenetically not only by their and turkeys only express one L chain (λ) with a single
sequence similarity, but also by the orientation of their V functional V and J gene, and the manner by which di-
and J RSS, which differ for mammalian λ and κ. There has versity is generated is likely responsible for this unusual
been much debate regarding the affi liations of L chains in evolutionarily derived arrangement (see the following).
various vertebrates, but as sequences have accumulated, we Nonproductive rearrangements are not detected on the
have a grasp on their phylogenetic emergence. Contrary to unexpressed L chain allele, and thus there is a strong pres-
what was believed, κ and λ L chains emerged early in verte- sure to generate functional joints (see the following). Such a
brate evolution, in the cartilaginous fish or placoderms. In system probably rendered a second (or third) L chain locus

superfluous.236 Within mammals (marsupial Monodelphis In summary, much of the V H germline repertoire has
domestica), the Vλ repertoire is comprised of at least three been conserved over extremely long periods of vertebrate
diverse families related to distinct placental families, sug- evolution. The birds and some mammalian species that
gesting the divergence of these genes before the separation rely on GALT to generate Ig diversity (see the following) are
of metatherians and eutherians more than 100 million years exceptions with a reduced germline repertoire (at least ex-
ago. Opossum λJC sequences are phylogenetically clustered, pressed repertoire), but as will be seen in the following, gene
as if these gene duplications were recent and the complexity conversion and SHM compensate for this situation in for-
of the λ locus seems greater than that found at the H chain mation of the primary repertoire. Even in the cartilaginous
locus.238 fish where there is a single V H family, there is nevertheless
In summary, all vertebrate groups except the birds great heterogeneity in CDR1 and CDR2 sequences (as well as
have two to four L chain isotypes, all of which can be cat- hypermutation) that must boost diversity in the expressed
egorized as κ, λ , or σ (and in cartilaginous fish σ-cart).231 repertoire.
However, we are still at a loss to understand the significance
of possessing multiple isotypes as there is scant evidence The J Chain
that L chains have any effector functions. It has been sug- The joining (J) chain is a small polypeptide, expressed by
gested that different isotypes may provide distinct CDR plasma cells, that regulates polymer formation of IgA and
conformations in association with H chains, or there may IgM.244 J-chain incorporation into dimeric IgA and penta-
be L chain/H chain preferences that provide some advan- meric IgM endows these antibodies with the ability to be
tage that is not obvious.239,240 At least the differential CDR transported across epithelial cell barriers. J chain facilitates
sizes for σ as compared to κ /λ suggest that the former ra- creation of the binding site for pIgR (secretory component
tionale is plausible. The preferential association of certain in the Ig polymers), not only by regulating the polymeric
L chains with particular IgH isotypes is suggestive of dif- structure but apparently also by interacting directly with the
ferent functions as well, but there is no evidence for such receptor. Therefore, both the J chain and the pIgR/secretory
functions to date. component are key proteins in secretory immunity. J chain
complementary DNAs have been reported in all jawed ver-
VH Evolution tebrates except the teleost fish, which have lost it.245 The ex-
Diversity of the immune repertoire depends on the variety of istence of Xenopus J chain suggests that, unlike mouse IgM,
V segments inherited in the germline and upon the further Xenopus IgM forms hexamers with J chain; alternatively, the
diversification by rearrangement (CDR3 only) and SHM (all previous electron microscopy studies identified IgX as the
CDR). Early in life, the repertoire depends chiefly on the hexameric isotype (the ξ chain has a stop codon before the
inheritable genes as one finds little N-region diversity and Cys of CH4 domain and thus cannot make a covalent attach-
somatic mutation (with exceptions; see the following). A ment to J chain). The highest level of J chain expression was
central question is how antibody germline V genes diversify detected in frog and bird intestine, correlating well with a
CDR during evolution while they are subject to homogeniz- role for J chain in mucosal immunity (although obviously
ing forces operating in most multigene families. Perhaps en- not for IgX secretion). Elasmobranch J chain shows high
vironmental antigens have played a major role in shaping the similarity to the N-terminal half of J chains from other
germline repertoire and have selected some V H/V L germline vertebrates, but is divergent or even absent in the other re-
sequences used by neonates. VH families arose in a bony fish gions. This result suggests that the function of J chain may
lineage and have been conserved for hundreds of millions of be solely for IgM polymerization in elasmobranchs, and the
years.241 Conserved regions defining families are found on transporting function arose later in evolution; consistent
solvent-exposed faces of the V H, at some distance from the with this idea, Xenopus, but not shark, J chain is capable if
antibody-combining site. Phylogenetic analyses show clus- interacting with human IgA and pIgR.245 As mentioned pre-
tering of V H into groups A, B, C, D, and E. All cartilaginous viously, the loss of J chain in bony fish does not preclude
fish V H belong to the monophyletic group E; bony fish V H the interaction of IgT with pIgR and transport across epi-
genes to cluster into all Groups (one group [D] unique only thelia, implying a strong pressure to maintain a mucosal-
to them). By contrast, group C includes bony fish sequences specific Ig.218 There was a claim for the presence of J chain
as well as V H from all other classes except cartilaginous in many protostomes because a homologue was cloned in
fish. Another phylogenetic analysis classifies mammalian earthworms, but no J chain sequences have appeared in
V H genes in three “clans” (I, II, and III), which have coex- the numerous protostomes or deuterostome invertebrate
isted in the genome for > 400 million years. Only in carti- databases.246
laginous fish does it appear that V H gene families have been
subjected to concerted evolution that homogenized member
genes (except for the IgM1gj V region described previously). T-Cell Receptors
It has been debated whether Ig V genes could be under di- Ig, TCRs, and MHC class I and class II are all composed of IgSF
rect positive selection or not because these genes hypermu- domains. The membrane-proximal domains of each Ig/TCR/
tate somatically. However, several features (eg, codon bias) MHC chain are IgSF C1-set domains, while the N-terminal
and discovery of high replacement/silent ratios in germline domains of Ig and TCR proteins are V-set domains encoded
gene CDR codons indeed argue for positive selection during by genes generated via rearrangement of two or three gene
evolution.242,243 segments during ontogeny (the membrane-distal domains of

MHC are a special case; see the following). In all vertebrates amphibians (axolotl and Xenopus). In addition to the typi-
studied to date, TCRs are membrane-bound and never se- cal IgSF domain features, there are several conserved regions
creted, while almost all Ig proteins have transmembrane and among vertebrate TCRβ chains, especially at positions 81 to
secreted forms.247 86, probably involved in TCR dimerization. There are also
remarkable differences: the solvent-exposed segment 98 to
`/a Constant Domains 120 in mammals is absent in all nonmammalian vertebrates.
Genes encoding the two types of TCR, α / β (which accounts This loop has been shown in mouse TCR to be important
for all known MHC-restricted regulatory and effector func- for negative selection events in the thymus; perhaps the ab-
tions) and γ /δ (which to the best of our knowledge recognizes sence of this region in nonmammalian vertebrates results
antigens in an Ig-like manner and may play immunoregula- in subtle differences in tolerance induction as compared to
tory or homeostatic roles during certain infections), existed mouse/human.256 The number of Cβ genes varies in differ-
in the earliest jawed vertebrates. The α / β TCR is considered ent species.
rather “boring” evolutionarily, as all gnathostomes from the Like Cα , Cβ sequences are not well conserved in evolu-
basal cartilaginous fish appear to have this MHC restricted tion (eg, the X. laevis Cβ gene does not cross-hybridize with
form.248 As mentioned, although many IgSF members exist X. tropicalis genomic DNA, and catfish Cβ has only 41% to
in the invertebrates, thus far no bonafide Ig/TCR sequences 42% identity with other teleost Cβ and 26% identity with
(ie, IgSF genes generated by somatic rearrangements) have horned shark Cβ). Two different catfish Cβ complementary
been isolated from jawless vertebrates or invertebrates, al- DNA sequences were identified, suggesting the existence
though a renewed search in jawless vertebrates may be fruit- of either two loci or allotypes, as is found in mammals.257
ful (see the following). Indeed, the damselfish Cβ was shown to be encoded by
While TCR genes have been cloned from representa- two polymorphic genes, and this feature seems to extend
tives of most vertebrate classes, few biochemical data are to other teleosts. As the polymorphic sites are believed to
available, except in birds where α / β and γ /δ TCR have been interact with the associated CD3-signaling molecules, the
identified with mAbs.249 In amphibians, the Xenopus α / β authors suggested that signals might be transduced to T cells
TCR was coimmunoprecipitated with cross-reactive an- in different ways depending on the particular expressed Cβ
tibodies raised against human CD3ε chains250 (note that allele. The damselfish Cα gene seems to be encoded by poly-
this same antibody indentifies CD3 epsilon from many morphic alleles as well.258
divergent species, which is quite unusual, 251 in turtles).
α chains from diverse vertebrates are poorly conserved `/a T-Cell Receptor Variable Domains
and the structure of the C α IgSF domain itself is unusual: Because T-cell recognition is MHC-restricted, TCR V re-
only strands A, B, C, E, and F can be identified, although gions have been evolutionarily selected for different prop-
strands E and F are shorter than those of mammals and erties as compared to Ig; indeed, TCR V regions are much
strand D is absent; this modification has an important less similar to each other than are Ig V regions from other
role in TCR dimerization and subsequent signaling.252 The antigen receptor loci.257 Furthermore, TCR Vs, unlike IgV H,
lack of conservation in this extracellular domain as well have conserved CDR3 lengths, suggesting that there is a re-
as deletions found in bird and teleost fish TCR (especially stricted size for recognition of MHC-peptide complexes.259
in the connecting peptide) suggest that the coreceptor may α loci in all vertebrates examined have many J segments,
be structurally distinct from mammalian CD3 complex and consistent with the mammalian paradigm, the absence
components. of D and the large numbers of J segments favors the poten-
Pre-Tα , which associates with TCR β chains during thy- tial for receptor editing during thymic positive selection.260
mocyte development, has been identified only in warm- A number of Vβ gene families are another evolutionarily
blooded vertebrates.253 The pre-Tα protein has no V domain, conserved feature. There are 12 families in nurse sharks and
and its interaction with the TCRβ chain is unique, based 19 in Xenopus. In axolotls Vβs are classified into nine cat-
on recent structural studies.254 Unlike all of the other TCR egories each with 75% or more nucleotide identity; as only
chains, pre-Tα has a long cytoplasmic tail, which seems to 35 genes were cloned, there are probably more families, and
be important for T-cell differentiation. Interestingly, the several are related to mammalian Vβ genes (human Vβ13
pre-Tα gene is linked to the MHC in mammals, and more and Vβ20257).
phylogenetic studies should be performed to determine There is evidence that the TCR has coevolved with MHC,
whether this linkage group is ancient (see the following). displayed by the canonical types of interactions seen in crys-
The TM region and cytoplasmic tail of Cα are the most tal structures.261 Vβ sequences from many different verte-
conserved parts of the molecule. Cα and Cβ TM segments brates share residues in CDR2 that interact with class II and
in all species have the so-called conserved antigen receptor CD1 MHC molecules (Y-46 and Y-48, especially).262 It will
transmembrane motif (CART) motif, in which conserved be of interest to search for other conserved interaction sites
amino acids form an interacting surface with the CD3 com- as crystal structures become available.
plex.255 Besides CART, the opposite TM face with conserved Studies of several teleost α /δ loci suggests an organiza-
residues Ile-Lys-Leu interacts with other CD3 components. tion similar to mouse/human, but with more rearrange-
The cytoplasmic region is remarkably conserved among all ment by inversion and more gene segments than are found
vertebrates. TCR β genes have been sequenced from several in mammals.263 Pulsed field gel analysis suggested that the
species of cartilaginous and bony fish and two species of horned shark α and δ TCR loci are closely linked.264 There

has been a recent, major modification in our thinking of the have revealed that the γ /δ TCR can either be used for innate
α /δ locus that will be described in the following. recognition (eg, the cells in the skin of mice or the blood of
humans) or adaptively with an enormous repertoire in ways
The f/c TCR we do not understand—this will be discussed in more detail
Complementary DNA sequences from the skate have signifi- in the conclusion. Consistent with the potential of γδ TCR
cant identity with prototypic mammalian γ and δ TCR genes to interact with nominal antigen, SHM has been detected in
with extensive V region diversity, putative D segments in δ, the γ genes of the sandbar shark, although it is not known
and varying degrees of junctional diversity.248 In the nurse whether the mutations generate the repertoire or appear in
shark, there is a great diversity of the δ V regions, the highest mature T cells after immunization.271,272
among the TCR chains; consistent with this diversity is the
presence of the NAR-TCRδ, which is found in approximately
25% of the δ Vs.257 The NAR-TCR has been found in all car- Immunoglobulin Gene Organization
tilaginous fish to date, including holocephalins, whereas VH Regions
NAR and IgW have only been detected in elasmobranchs. A rearranged V H gene consists of a leader segment, encoded
Thus, it is possible that NAR (especially the V region) arose by a canonical split exon, followed by four framework re-
as a TCR and was transferred to an Ig cluster sometime in gions and three CDRs (see Fig. 4.8). Canonical V H CDR1
evolution. nucleotide sequences are conserved in all jawed vertebrates
In axolotls, Vδ diversity was diminished in thymecto- and serve as targets for SHM.242,243 A major germline differ-
mized animals and TCR δ chains are expressed by cells in ence is the lack of conserved octamers and TATA box in the
lymphoid organs, skin, and intestine.265 Chicken γ /δ T cells 5′ region of shark Vs. In all species, functional V genes are
were identified long ago. Expression is found in thymus, assembled by rearrangement and joining of germline V, D,
spleen, and a γ /δ T-cell line, but not in B cells or α / β T-cell and J elements. Cartilaginous fish H chains are encoded by
lines. Three V subfamilies, three J gene segments, and one large numbers of clusters (> 100 in horned shark and dog-
C gene were identified at the TCR γ locus. All Vγ subfamilies fish; approximately 15 in the nurse shark). For IgNAR, there
participate in rearrangement during the first wave of thy- are only four V genes/haploid genome and only a few IgW
mocyte development, and the γ repertoire diversifies from V genes are detected in nurse sharks, but a large number in
embryonic day 10 onwards with random V-J recombination, skates and dogfish.273 There are widely varying numbers of
nuclease activity, and P- and N-nucleotide addition.249 V genes in different species; importantly, the V H complex-
In ruminants and chickens (so-called GALT species; ity does not seem to limit diversity of the antibody reper-
see the following), the γ /δ repertoires are quite diverse, toire in any ectothermic vertebrate studied to date. There are
and there seems to be ligand-mediated selection of γ /δ actually fewer functional human V H (44 functional, 79 pseu-
cells during ontogeny. In sheep, where γ /δ TCR diversity is dogenes that fall into seven families) than in many ecto-
thymus-dependent and follows a developmentally regulated therms. Dynamic reorganization of the H chain V regions
progression, no invariant γ /δ TCRs are found.266 The degree seems to have occurred at least eight times between 133 and
of γ /δ expression is correlated with the evolution of the TCR 10 million years ago.241 Perhaps species that utilize somatic
V families in warm-blooded vertebrates. Indeed, mammals mutation/selection “optimally” rely less on germline diver-
can be classified into “γ /δ low” (humans and mice, in which sity and therefore fewer functional genes are required. Only
γ /δ T cells constitute limited portion of the T-cell popula- approximately 10% of Xenopus V H are pseudogenes in the
tion) and “γ /δ high” (chicken, sheep, cattle, and rabbits, in three families (V H1–3) that have been exhaustively studied;
which such γ /δ cells comprise up to 60% of T cells). TCR thus, Xenopus with fewer lymphocytes has a greater number
V genes form subgroups in phylogenetic analyses, and hu- of functional V H genes than humans.
mans and mice have representative loci in most subgroups
whereas the other species appear to have lost some.267 Thus, Chondrichthyan Germline-Joined Genes
γ /δ -low species have a high degree of TCR-V gene diversity, In all vertebrate species, functional Ig genes are assembled
while γ /δ -high species have limited diversity. Interestingly, by rearranging DNA segments scattered on the chromo-
this pattern is similar to that found for IgV H genes. some. However, in cartilaginous fish some V genes are the
Recent work in Xenopus has shown that the presence of products of V(D)J rearrangement in eggs/sperm.196,274,275
prototypic VH at the TCRδ locus, in addition to typical α and Type I L chain (σ) genes are all germline-joined in skates
δ V domains.268 Furthermore, consistent with the presence of but split in horned sharks, and the piecemeal germline joins
VH, the αδ locus is closely linked to IgH and IgL λ in the frog (eg, VD, VDD, VDDJ) found in many horned shark H chain
genome. These data suggest that these heterodimeric loci gene clusters and in nurse shark L chain Type I (σ) clusters
arose via a cis duplication early in vertebrate history.263 The strongly suggest that the germline-joining is a derived fea-
TCRμ locus in monotremes and marsupials202,269 can now be ture. Definitive proof came from a study of a germline-joined
extended to the chicken as well,270 although like Xenopus there nurse shark Type III (κ) L chain gene, shown by phyloge-
are chicken V H with a three-domain molecule as is found in netic analysis to have been joined within the last 10 million
monotremes, marsupials, and sharks. In total, all of the non- years236 ; this was followed by the identification of the nurse
placental mammals except bony fish show some sort of chi- shark germline-joined IgM1gj described previously.196 When
meric Ig/TCR locus, with TCR δ V domains that are more Ig- there is a mixture of joined and conventional genes, the split
like than TCR-like. Generally speaking, comparative studies genes are expressed in adults, while the joined genes are

expressed at significant levels only early in ontogeny. When that this is not a rule. As mentioned previously, two groups
all of the genes in a particular family are joined (eg, skate have suggested that the teleost gene organization seems to
Type I L chain genes and Type II [λ] L chains in all elas- promote receptor editing.234,281
mobranchs), they continue to be expressed into adult life at
high levels. In mammals, what may appear like germline- D Segments
rearranged V genes are in most cases processed pseudogenes D segments are always present in one of the two loci en-
(eg, pseudo Vκ on chromosome 22 in human or in mouse). coding an Ig/TCR heterodimer (IgH and TCR β,δ), and
However, it is possible that the surrogate L chain gene VpreB the pressure maintaining this asymmetry is unknown.
is the product of a germline-joining event in the line leading Cartilaginous fish usually have two D genes/H chain cluster,
to mammals.276 Additionally, as described previously, there and there are only minor variations among the clusters; one
is a marsupial germline-joined V that serves as the mem- RSS follows the 12-12 paradigm seen in tetrapods and the
brane-proximal domain of TCRμ.202 other is like the TCR 12-23. In teleosts, amphibians, and rep-
tiles where the organization of the H chain locus is similar
Organization of Rearranging Genes to humans, the number of Ds deduced from complementary
As mentioned, shark IgH-chain genes are structured into DNAs ranges from 10 to 16. Two germline D segments have
tens to hundreds of clusters, each consisting of V, D, J, and been identified in Xenopus, and their RSS follow the rules
C elements277; all evidence from studies in horned shark, defined in mammals. In birds, there are 15 very similar DH.
nurse shark, skate, and sandbar shark (and holocephalin rat- There are several reasons why D segments may have been
fish) suggest that V, D, and J genes rearrange only within one preserved throughout evolution. Incorporation of D seg-
cluster (but, as detailed in the following, switch may occur ments augments CDR3 diversity and size, obviously directly
between clusters, most likely as a consequence of antigen influencing the combining site.259 Three different Ds con-
stimulation). While there is extensive N-region diversity and tribute to IgNAR CDR3, and besides generating great di-
sometimes usage of two D segments (three Ds in IgNAR), versity, CDR3 length and amino acid composition fulfi lls
and there are V H subfamilies having substantial CDR1/2 special tertiary structure requirements: D-encoded cysteine
heterogeneity, diversity of the primary repertoire is lower residues bond with cysteine(s) in the body of the V domain,
than in other vertebrates as there is no (or infrequent) rear- thereby stabilizing a loop involved in the antigen-binding of
rangement between clusters. The special constitution of the this unusual monomeric receptor.197,200 A similar situation
shark H and L chain loci suggests an exclusion mechanism has been reached by convergence in the monomeric variant
similar to that of mouse TCR γ loci, also found in clusters. of camel IgG.198 Finally, rearrangement of one locus “locks
It appears that only one V H transcript is expressed in each it in” and allows the second locus to undergo receptor edit-
lymphocyte, consistent with isotypic exclusion, despite the ing, as is the case for negative selection of the B-cell reper-
many clusters (see the following).278,279 Bony fish (teleosts toire and positive selection of the T-cell repertoire. Finally,
and chondrosteans like the sturgeon), frogs, reptiles, and in mice, one of the TCRδ D segments encodes a section that
mammals have very similar architectures of their H chain interacts with the ligand, the nonclassical class I molecule T
locus—the so-called translocon configuration. As described lymphocyte antigen (TLA).282
previously, multiple families of V H genes, each consisting
of many apparently functional elements (1 to 30 per fam-
ily), are separated from a smaller number of genomic D and Agnathan Variable Lymphocyte Receptor Structure
J elements. The possibility of combinatorial rearrangement and Function
enables more diversification than is possible with the car- In the 1960s, jawless vertebrates (hagfish and lampreys;
tilaginous fish clusters for a given number of segments. In see Fig. 4.1) were reported to mount humoral responses to
birds, the organization is similar but all V genes except those foreign antigens. However, for anti–group A streptococ-
most 3′ to the D elements are pseudogenes (see the follow- cal antigens, the hagfish “antibodies” (at least a proportion
ing). One exception is the coelacanth (Latimeria), in which of them) were actually the complement component C3283!
V genes are immediately followed by D segments, and then Lamprey and hagfish were long known, however, to pos-
multiple J segments as are found in all tetrapods.280 sess cells resembling lymphocytes and plasma cells,284 with
L chain gene organization is variable. In elasmobranchs, expression of lymphocyte- or at least leukocyte-specific
the organization is the same (ie, in clusters) as the H-chain genes,285 but the quest for RAG or bonafide Ig/TCR/MHC
locus without the D segments. The prototypic horned shark genes was a complete failure. Reports of specific memory
Type I (σ) L chain has a cluster organization in which V, J, in allograft rejection and other specific humoral responses
and C segments are closely linked. As mentioned previously, were difficult to reconcile with absence of the RAG-based
bony fish L-chain genes have the shark cluster-type orga- rearranging machinery and the possibility of generating
nization, but some species have multiple V genes in some specific lymphocyte clones.
clusters, demonstrating that there is rapid evolution of not Our view of the jawless fish immune system was radically
only sequence but gene organization as well in this taxon.232 transformed when Pancer et al. prepared complementary
In Xenopus, there are multiple Vκ (ρ) presumably derived DNA libraries of naïve lymphocyte RNA subtracted from
from one family: five J and a single C gene segment. In carti- lymphocyte RNA derived from lamprey larvae (ammocoetes)
laginous fish and birds there has been coevolution of Ig gene that had been immunized to a bacterial/PAMP mixture, and
architecture for H and L loci, but the teleosts have shown found a highly diverse set of LRR sequences enriched in the

immunized animals.286 The clones were not only diverse in in the lamprey genome project and are expressed in a
sequence, but also in the number of LRR “cassettes” found lymphocyte-specific fashion288 ; in fact, the two members that
between invariant 5′ (LRR NT) and 3′ (LRR CT) cassettes. have been discovered, CDA1 and CDA2, are differentially ex-
Unlike what has been discussed previously for immune pressed in agnathan lymphocyte subsets (see the following).
genes many species (especially sea urchins), there were not a Because APOBEC family members are involved in repertoire
great number of the germline NT and CT cassettes, suggest- building in jawless and jawed vertebrates, mutation/gene
ing that a somatic recombination process, convergent with conversion may have predated RAG-mediated repertoire
Ig/TCR rearrangement, occurred in the developing lam- building of the repertoire (see the following).
prey lymphocytes (Figs. 4.9 and 4.10). The genomic orga- VLR homologs were found in two additional lamprey
nization of germline VLR (the original gene discovered was species and in hagfish, the only other order of living jawless
VLRB) have 5′ and 3′ LRR cassettes separated by an intron. vertebrates reviewed in Boehm et al.13 As in the sea lamprey,
Upstream and downstream of the invariant exons are a large the incomplete hagfish germline VLR generate somatically
number of LRRs, which become inserted between the 5′ and highly diverse repertoires. Interestingly, the Amphioxus
3′ cassettes during lymphocyte differentiation. Individual genome harbors a large number of intronless VLR-like se-
lymphocytes apparently express a uniquely rearranged VLR quences that could represent an alternative germline VLR
gene in monoallelic fashion. The potential VLR repertoire diversity akin to the echinoderm gene families (TLR, SRCR)
can be as great as 1014, vastly outnumbering the lymphocytes described previously; one could even speculate that they are
within an individual.287,288 related to the ancestral VLR before invasion of its analogous
Fragments of homology between the LRR cassettes allow “RAG transposon” (see the following).
for joining and then priming of synthesis of a copy of the There are three VLR types: A, B, and C. VLR-B cell-surface
particular transferred cassette.289 Because these small regions receptors and secreted molecules are analogous to the jawed
of homology are found throughout the cassettes, “hybrid” vertebrate membrane and secreted BCR, and it was proposed
LRR can also be formed, further enhancing the diversity that cell activation was similar to T-independent pathways
over a simple insertion of cassettes. The insertion of LRRs in jawed vertebrates (ie, either direct stimulation through
can occur at either end of the somatic recombinant, but it is the surface VLR or surface VLR stimulation in combina-
always 5′-to-3′ on the growing strand. This type of genomic tion with a PRR/PAMP interaction). It was suggested that
modification resembles the initial stage of gene conversion, the jawless vertebrate adaptive system might be dedicated
but because there is not a complete transfer of genetic ma- exclusively to humoral immunity, with the PAMPs provid-
terial between two homologous gene segments, but actually ing the second signal to activate lymphocytes.290 Thus, it was
an addition of sequence, the assembly of the mature VLR is proposed that the lamprey system was “B cell-centric” and
more similar to a recombinatiorial mechanism described in therefore focused on humoral immunity.
yeast called “copy choice.” The enzymology of “copy choice” This model has been disproved as subsequent work
has not been examined; because a gene conversion-like pro- demonstrated that the VLRA and VLRB were differen-
cess occurs, it is likely that an APOBEC family member is tially expressed (see Fig. 4.10). The former receptor could
involved. Indeed, APOBEC family members were detected not be found in plasma and was expressed exclusively as
LRR LRR LRR LRR1 LRR NT LRR CT STALK LRR

NT CT
Insertion of flanking FIG. 4.9. Genetics, Generation of Diversity,
LRR modules and Speculative Cell Biology of the
LRR LRR1 LRR LRR Hagfish Variable Lymphocyte Receptor
LRR LRR CP STALK (VLR) System.13 The top line shows an in-
NT CT
complete VLR gene (NT and CT, N-terminal
and C-terminal cassettes, respectively).
LRR-V
LRR-1 LRR-Ve During lymphocyte ontogeny, upstream
LRR-CP and downstream LRR cassettes are in-
LRR-CT serted between the NT and CT gene seg-
LRR-NT Lamprey lymphocyte ments, resulting in an intronless, mature
VLR GPI-linked VLR gene (second line). VLR proteins are
N term
C term
attached to the lymphocyte surface via a
glycophosphatidylinositol linkage, and may
Antigenic be released into the blood upon antigenic
stimulation stimulation (also see Fig. 4.10). Boxed is a
hagfish VLR protein, the first structure to be
elucidated by Kasahara and colleagues.499
Clipped VLR? Note the loop at the C-terminal leucine-
Soluble molecules rich repeat, which is mentioned in the text
as a prominent region inserting into anti-
Note: VLRA never secreted gens, like single-domain antibodies.293,294

JAWED VERTEBRATES JAWLESS VERTEBRATES
Cytokines Cytokines (IL-17)

Cytotoxicity Cytotoxicity?
Antigen Antigen
VLRB VLRA
TCR
Ig
B cell T cell B cell T cell
2R
RAG
IgSF IgSF LRR LRR
IgSF IgSF IgSF IgSF LRR LRR LRR LRR
IgSF IgSF IgSF IgSF LRR LRR LRR LRR

B-like cell T-like cell B-like cell T-like cell
Differential recruitment of receptors
VJ C1
1R
GPIb-α IgSF IgSF B-cell like T-cell like

(LRR) (VJ-type) (C1-type)
APOBEC diversification?
FIG. 4.10. Emergence of Variable Lymphocyte Receptor (VLR) and Immunoglobulin (Ig) Early in Vertebrate Evolution. T and B cells likely pre-
ceded the divergence of the leucine-rich repeat and Ig superfamily antigen receptors. The VLR is most similar to a molecule called GPIb-α288
present in Amphioxus, and Ig superfamily antigen receptors are derived from molecules with so-called VJ and C1 domains, probably in the
proto–major histocompatibility complex (see text for candidates). Both precursor genes were present in basal chordates. The RAG transpo-
son and 2R likely ignited the appearance of the Ig superfamily receptors (Ig/T-cell receptor) (see text for full explanation). This figure also
shows B and T cells today in jawless and jawed vertebrates, with secretion of IgM from gnathostome B cells and a VLR tetramer of dimers
from VLRB-secreting lamprey B cells. Figure modified from Flajnik and Kasahara.2
a cell-surface receptor, while, as described, the latter was cytotoxic mediators, are properties of the T cells themselves.
present as both a lymphocyte receptor and a secreted mole- It is not clear how VLRC fits into this scheme at the moment.
cule.291 When microarray expression analysis was performed As expected for members of this family, the framework,
on VLRA- and VLRB-positive cells, the patterns were con- or backbone, of the LRR is very similar between the cas-
sistent with expression profi les in gnathostome T cells and settes. Diversity between cassettes is concentrated in the
B cells, respectively. The lamprey adaptive scheme, therefore, concave surface, presumably the region coming in contact
parallels the situation in jawed vertebrates where Ig is found with antigen. The crystal structures of two VLR-B mole-
both as a cell-surface receptor on B cells and as a secreted cules have been determined, one specific for HEL292 and the
effector molecule in the serum, whereas the TCR is pres- other for H-trisaccharide.293 In both cases, the concave sur-
ent only as a cell-surface receptor on T cells; effector func- face makes contacts as well as a loop in the LRR-CT, which
tions in T cells, such as cytokine secretion or production of in the case of HEL inserts into the active site, similar to

what has been found for single-domain camelid and shark immune response, enumeration of antigen-binding Igs by
Vs.198 One VLR-A crystal structure was determined, also isoelectrofocusing (IEF), and idiotypic analysis. Sequences
specific for HEL.294 This VLRA was derived from a HEL- of Ig and TCR genes expressed over the course of a response
hyperimmunized adult lamprey, which apparently under- help to estimate diversity at another level, allowing stud-
went affinity maturation with induced molecules having ies of V genes diversified by gene conversion and/or so-
affinities in the picomolar range. The antigen is bound in matic mutation during a response in a precise way. In the
the region of the molecule that is most diverse among all following survey, we describe studies of specific antibody
VLRAs; surprisingly, the LLRNT among all VLRA are not synthesis, T-cell responsiveness (T-B collaboration, MHC
diverse, suggesting that this region might bind to a self– restriction).
“restricting element” and the rest of the binding site to the
true antigen. Cartilaginous Fish
These are early days, but the comparisons to date suggest Natural antibodies binding many antigens have been de-
that VLRA and VLRB recognize antigen in different ways. tected at surprisingly high levels in chondrichthyans and
Like gnathostome BCR, VLRB is expressed at the cell sur- in some teleosts. In older experiments, after immuniza-
face and is secreted as a pentamer of dimers, much like IgM tion the horned shark mounted a low-affinity 19S (penta-
(see Fig. 4.10). The affinities detected to date from immu- meric) IgM antibody response, which varied little among
nized animals are in the micromolar range, again similar individuals and did not increase in affinity after prolonged
to low-affinity pentameric IgM in gnathostomes. By con- immunization.298
trast, VLRA has never been detected in the sera or secreted The relative homogeneity and large number of V genes
in culture, fitting with its similarity to TCR. The affinity hindered SHM studies until a single unique reference
of VLRA for soluble antigen was high (in the nanomolar horned shark IgM V H gene was found. Mutations in this
range) for the HEL-immunized animal, which appeared to gene were slightly more frequent than those in Xenopus (see
undergo affinity maturation.294 Combining the information the following). This first study proved that SHM preceded
with the conserved and diverse regions of the VLRA bid- diversity obtained by combinatorial association of gene
ing site, perhaps VLRA recognizes antigen (budding virus?) segments in evolution.299 In contrast to mutations in the
associated with a self-protein (MHC analogue?) at the cell horned shark IgM V H genes, unusual patterns of somatic
surface. It remains to be seen whether VLRA clones will be mutation were detected in nurse shark IgNAR (see the fol-
found from other immunized animals, and the HEL experi- lowing) and Type II (lambda) germline-joined L chains.
ment was an artifact of the immunization schedule and not Half of the mutations (338/631) occur in tandem without
physiologic. the GC bias seen in Xenopus or horned shark H chain V
The independent development of two different strategies genes. Tandem mutations and point mutations that take
for receptor somatic diversification at the dawn of vertebrate place simultaneously were not generated by gene conver-
evolution approximately 500 million years ago reveals the sion as there are no repeated patterns or potential donor
magnitude of the selective pressure applied on to the im- genes.300,301 The germline-joined L chain genes can only di-
mune system and the emergence of individualized adaptive versify through SHM, perhaps like the hypothetical proto-
responses (see Fig. 4.10). Are VLR responsible for the graft typic V region gene prior to RAG-mediated rearrangement
rejection results seen in the old experiments? If so, this is (ie, SHM may have preceded gene rearrangement as the
most likely a “convergent MHC” coopted for the VLR sys- primordial somatic diversification mechanism)300 (see the
tem. Skin grafting experiments (or other tests of alloreactiv- following). Lastly, a reappraisal of mutation at H chain and
ity) should be repeated to study potential VLR involvement. other L chain loci in the nurse shark showed that the muta-
The VLR system is discussed below as well in the lymphoid tions were not so different from the L chains; the differ-
tissues section and the conclusion. ences from the previous work were the different shark spe-
cies and the analysis of all H chain loci in the species rather
than only one unusual locus.302
Adaptive Immune Responses in Gnathostomes As mentioned, the small number of IgNAR genes also
The quality of T-cell and B-cell responses depends on the made it possible to analyze SHM, and in the first experi-
heterogeneity and the diversity of the antigen receptor rep- ments, random complementary DNAs were examined.197,296
ertoires, and the ability to select cells in secondary lym- The mutation frequency was about 10 times that of Xenopus
phoid tissues. Because of the indefinite number of potential and horned shark IgM, and even higher than in most stud-
Ig/TCR V region sequences in most taxa, potential diver- ies in mammals. lt was difficult to establish a pattern for the
sity exceeds the number of available lymphocytes; further- mutations due to their high frequency and because they are
more, all jawed vertebrates are capable of AID-dependent often contiguous, like in the L chain gene study described
SHM or gene conversion after antigen stimulation. Yet, previously. Mutations even in randomly isolated clones ap-
while potential repertoires are diverse in all vertebrate peared to be under positive selection in IgNAR secretory
classes, and polymorphic MHC class I and II and TCR but not TM clones, strongly suggesting that mutations do
genes have been isolated from all classes, antibody diver- not generate the primary repertoire like in sheep but arise
sity in nonmammalian vertebrates is relatively low.295–297 only after antigenic stimulation.303,304 In total, the shark mu-
The expressed repertoire has been studied via structural tations seem quite mammalian-like but with unusual fea-
studies, affinity measurements during the maturation of the tures. Analysis of the mutations in noncoding DNA suggests

an AID-dependent process coupled with an error-prone adaptive immunity.257 In summary, it appears that all of the
polymerase.305 components of adaptive immunity in mammals occur in
Affinity maturation and memory generation can be de- the cartilaginous fish.
tected in sharks.195 Soon after immunization with HEL, an
IgM response can be detected, primarily of the pentameric Bony Fish
class. Over time, 7S IgM and IgNAR responses develop, There are high levels of low-affinity natural antibody (up
and the 7S antibodies have a higher binding strength than to 11% of total Ig) to nitrophenylacetate in some bony fish.
those of the 19S class. When titers were permitted to drop Natural antibodies in catfish have been correlated with resis-
to baseline (or close to baseline), a memory response was tance to virus infection or furonculosis. As a rule, and simi-
induced by immunization of antigen without adjuvant. lar to cartilaginous fish, little affinity maturation has been
However, unlike responses in higher vertebrates, the titers detected in fish, although some changes in fi ne specificities
do not increase over those in the primary response, sug- were noticed in the trout with a sensitive enzyme-linked im-
gesting a unique type of regulation of antigen-specific IgM munosorbant assay–based test.309 The mild increase in trout
and IgNAR. Nevertheless, these data strongly suggest that antibody affinity (similar to that found in Xenopus and shark
the hallmarks of an adaptive response occur in sharks. This IgNAR) is attributed to selection of either minor preexisting
type of response is reminiscent of IgA responses of intes- B-cell populations or somatic mutants. In partially-inbred
tinal lymphocytes in mice.306 In another study, a family of self-fertilized or gynogenetic trout, variability of specific re-
HEL-specific IgNAR clones was followed over time after im- sponses is even more restricted. Affinity measured by equi-
munization, and a 10-fold increase in the affinity of an al- librium dialysis was of the order of 2.0 × 10 −6 M for trini-
ready high-affinity germline clone (10 −9 M) was observed.307 trophenol (TNP)-specific antibodies. A large literature deals
These results suggest that affinity maturation, memory, and with vaccination attempts in teleost fish, due to their eco-
“switch” to the monomeric IgM isotype occurs, but it takes nomic importance. The availability of catfish B-cell, macro-
much longer to attain these adaptive hallmarks compared phage, and T-cell lines have been instrumental in analyses
to mammals, perhaps a paradigm for ectotherms (see the of antibody production.310 There are puzzling differences in
following). responses from different teleost groups, much like differ-
Recent work showed that isotype switch can occur in ences between urodeles and anurans (amphibians). Cod, for
sharks, despite many reports to the contrary.230 Previously, example, do not respond well to specific antigen and have
studies of nurse and horned sharks complementary DNAs very high levels of “natural antibodies”; recent studies show-
showed that the V and C regions were derived from the ing that cod have lost class II genes provides an explanation
same gene clusters. By contrast, in the new work, immu- for the poor humoral responses (see the following).
nized animals showed switching between IgM clusters and Like the sharks, isolation of TCR genes and the exis-
even between IgM and IgW clusters. It is not known whether tence of a polymorphic class I and class II molecules sug-
there is any functional significance to the switch, but one gest that antigen presentation is operative teleosts, but
might predict the switch could dictate a change from 19S unlike sharks, functional experiments examining mam-
to 7S antibodies (see the following), or a modification of malian-like T-APC interactions have been performed.
effector class in IgM and IgW. None of the classic hotspots TCR messenger RNAs are selectively expressed, and spe-
(RGYW) upstream of constant region genes in tetrapods cific TCR rearrangements have been detected in catfish
vital to switch appear to be targeted to initiate the switch clonal cell lines, which produce factor(s) with leukocyte
in sharks, but other repetitive elements were detected that growth–promoting activity reviewed in Miller et al.310
might play a role. Modifications of the trout T-cell repertoire during an acute
The role of T cells in shark immune responses has not viral infection (rhabdovirus) have also been followed.311,312
been studied in detail. No thymectomy experiments have In nonintentionally immunized trout, adaptation of the
been performed, and T cells have not been monitored dur- spectratyping technique for TCRβ CDR3 length revealed
ing an immune response. Shark mixed lymphocyte re- a polyclonal naïve T-cell repertoire. After an acute infec-
sponses (MLR) and graft rejection have been attempted, tion with viral hemorrhagic septicemia virus, CDR3 size
MLR with little success (probably for technical reasons) and profi les were skewed for several Vβ /Jβ combinations, cor-
grafts with the demonstration of a chronic type of rejec- responding to T-cell clonal expansions. Both “public” and
tion for which the genetics has not been analyzed. However, “private” T-cell expansions were detected in the infected
from the MHC and TCR studies, it is clear that all of the genetically identical individuals. The “public” response
molecular components are available for proper antigen pre- resulted in expansion of Vβ4/Jβ1-positive T cells that ap-
sentation in sharks and skates, and studies of splenic archi- peared fi rst in the primary response and were boosted dur-
tecture suggestive class II + dendritic cells in the white pulps ing the secondary response. Further work examined fi ne
argue for a prominent T-cell regulatory role in adaptive specificity of the viral T-cell response, which is a model for
immunity.308 Furthermore, an increase in binding strength studies in cold-blooded animals.
and memory response, as well as mutation and now switch, Recent results suggest that, despite the fact that IgM is the
also strongly suggests a T-cell involvement in humoral im- major Ig expressed in an immune response, high- and low-
munity. Finally, recent studies of the thymus have shown affinity IgM appears to be a function of the degree of polym-
that the architecture and expression of well-known mark- erization of the tetramer.309 It is suggested that high-affinity
ers are wholly consistent with a typical T-cell regulation of interactions of TM IgM on B cells modifies the enzymes

involved in disulfide-bond formation, resulting in modifi- tilapia, and they spontaneously kill a variety of xenoge-
cations of secreted IgM.313 In addition, expression of tran- neic targets, including certain fish parasites and traditional
scription factors in different subpopulations of trout B cells mammalian NK cell targets.
suggests that the typical populations of naïve, memory, and
long- and short-lived plasma cells exist in bony fish; these Amphibians
populations have been followed in blood, spleen, and head Differences in immune system features between urodele
kidney by measuring proliferation and B cell–specific tran- (axolotl) and anuran (Xenopus) amphibians, already dis-
scription factors.314–316 Such studies should prove very useful cussed for MHC and Ig complexity, are also seen in im-
for vaccination of large populations of teleosts. mune responses. Rarely is such divergence seen within one
Studies of SHM in teleost fish have been hampered by the vertebrate class (although the two groups diverged over
lack of good reference genes, but it has been clear that AID- 250 million years ago!). Urodeles express a very restricted
targeting motifs are found in Ig V genes.317,318 Furthermore, antibody repertoire in response to specific antigen that
AID is expressed in the spleen during immune responses, peaks at 40 days postimmunization, and is entirely of the
again consistent with its role in SHM.319 A recent study IgM class, even though the serum also contains IgY.323 They
examined mutations in a reference zebrafish L chain gene do not respond well to thymus-dependent antigens, which
and detected the typical mutational pattern described pre- may be due to lack of T-cell help, yet their expressed TCR di-
viously (without the tandems seen in sharks).320 AID from versity looks normal. A population of axolotl B cells prolif-
several bony fish species is capable of inducing CSR in mam- erates specifically in response to LPS and also secretes both
malian B cells, despite the fact that teleosts do not undergo IgM and IgY. Moreover, a distinct lymphocyte subpopula-
CSR 318 ; perhaps the new results that demonstrate switching tion proliferates significantly in response to the T cell mito-
in sharks may shed light on this conundrum.230 gens Con A. T cells from young axolotls (before 10 months)
As mentioned, species living in extreme cold develop do not have this functional ability. Axolotl T cells also can
adaptive structural differences in their Igs.192 At the level be stimulated with to SEA/SEB, known from mammalian
of global immune response, temperature exerts a great studies to be superantigens.324
influence in ectothermic vertebrates in general, low tem- Anuran larvae can respond specifically (with only 106
perature generally being immunosuppressive. Lowering the lymphocytes) to many antigens, with a modest affinity mat-
water temperature from 23° to 11°C over a 24-hour period uration of the IgM anti-dinitrophenol (DNP) response.325
suppresses both B- and T-cell functions of catfish for 3 to In adults, the number of different anti-DNP antibodies
5 weeks as assessed by in vitro responses.321 Virgin T cells are does not exceed 40, versus 500 in mammals. In secondary
most sensitive to this cold-induced suppression, a property responses, the peak of the response is about 10-fold higher
shared with mammals when tested appropriately. Fish have and is reached in 2 weeks; there are no major changes in
developed ways to adapt to the lack of fluidity of their B-cell affinity over this initial rise. Isogeneic Xenopus produce ho-
membranes by altering the composition of fatty acid by mogenous antibodies to DNP, xenogeneic red blood cells, or
using more oleic acid at low temperatures. After appropriate phosphorylcholine with identical or similar IEF spectrotypes
in vivo acclimation, catfish T cells are better able to cap cell and idiotypes, while outbred individuals differ.326 Both IEF
surface molecules at low assay temperatures than are B cells, spectrotypes and idiotypes are inheritable, suggesting that
suggesting that capping is not the low temperature-sensitive diversity is a reflection of the germline repertoire without a
step involved in T-cell immunosuppression in catfish. major contribution from somatic mutations. Thus, somatic
In the NK section, we briefly discussed fish cytotoxic mutations were followed during the course of an antigen-
cells.169 In vitro studies have now shown that leukocytes from specific immune response at the peak of the modest affinity
immunized fish specifically kill a variety of target cells (al- maturation.327 The V H genes, like their mammalian homo-
logeneic erythrocytes and lymphocytes, hapten-coupled au- logues, contain the sequence motifs that target hypermu-
tologous cells); fish CTL of the αβ (and perhaps γδ) lineages tation, as described previously. Of the 32 members of the
as well as NK cells were found (see previous discussion). V H1 family involved in the anti-DNP response, expression
Naïve catfish leukocytes spontaneously kill allogeneic cells of only 5 was detected, indicating that immunization was
and virally infected autologous cells without sensitization, being monitored. Few mutations were detected (average:
and allogeneic cytotoxic responses were greatly enhanced 1.6 mutations per gene; range: 1 to 5), and there was not a
by in vitro alloantigen stimulation.322 Cloned cytotoxic cells strong preference for mutations in CDR1 and 2 and virtually
contain granules and likely induce apoptosis in sensitive none in CDR3. Like in the horn shark IgM study noted pre-
targets via a putative perforin/granzyme or Fas/FasL-like viously (but not IgNAR or Type II L chains), the mutations
interactions. All catfish cytotoxic cell lines express a signal- were targeted to GC bases, and such a pattern has been sug-
transduction molecule with homology to the Fc γ chain of gested to be the first phase of the SHM phase in mouse/
mammals; this chain with an ITAM is an accessory mol- human; perhaps Xenopus has lost the second phase of the
ecule for several activating receptors on mammalian NK process that results in an evening of mutation frequency for
cells.161 Importantly, these cytotoxic cells do not express a all bases. While the mutation frequency was lower than in
marker for catfish nonspecific cytotoxic cells. As described mammalian B cells, the rates were only four- to sevenfold
previously, nonspecific cytotoxic cells have been found in less in Xenopus. Thus, there is no shortage of variants, and
other fish species, including trout, carp, damselfish, and the reasons for the low heterogeneity and poor affinity mat-

uration may be due to less than optimal selection of the mu- Reptiles
tants. Indeed, because of a relatively low ratio of replacement Lack of an increase in affinity and homogeneity of IEF spec-
to silent mutations in the CDRs, it was argued that there is trotypes suggest low-antibody heterogeneity in reptiles.
no effective mechanism for selecting mutants, which in turn In the turtle Pseudemys scripta, a number of genomic VH
might be related to the absence of GC in Xenopus. In sum- sequences, representing possibly four families, were iso-
mary, the data from hypermutation, complementary DNA lated, as was a genomic Cμ, all shown to be encoded at a
heterogeneity, and spectratype dominance suggests that in single locus. In northern hybridizations, the Cμ4 probe
the absence of refined modes of selection in late-developing detected two transcripts; of the four VH groups, only one
clones, B cells producing somatic mutants may be outcom- was expressed, and multiple bands indicated the presence
peted by antibodies generated earlier in the response. of at least two non-μ transcripts. Among 32 unique VDJ
Essential T-cell functions in anurans have been dem- rearrangements from one animal, there were 22 sequence
onstrated with in vitro assays for T-B collaboration and variants in FR4, suggesting either a large number of J seg-
MHC restriction, demonstrating the similarity of the role of ments or somatic modification.336 The latter interpretation is
MHC in Xenopus and mammals.328 Regulatory T cells have supported by point mutations found in FR3 and CDR3. For
been shown indirectly in hematopoeitic/thymic chimaeras T cells, there are no data on T effector function, but there
for control of CTL generation and in antibody responses. are studies on the behavior of T-cell population changes
Ig synthesis can be enhanced following late thymectomy due to seasonal and hormonal variations. Thymocytes
in axolotl or Xenopus, again implying a role for thymic- from the turtle Mauremys caspica proliferate in response to
dependent regulatory cells. Thymectomy early in life totally phytohemagglutinin (PHA) and ConA, and can kill tumor
prevents CSR from IgM to IgY, but not IgX synthesis; thus, target cells by both antibody-dependent cellular cytotoxic-
T cells are absolutely required for the switch to the “IgG-like ity–mediated and NK-mediated cytotoxicity. Proliferative
isotype” and for high-affinity IgM responses, but switch to responses to PHA and Con A were higher for both sexes in
the mucosal Ig can be T-independent. Switching can also be spring and for females in winter than in the other seasons.337
induced in tadpoles, although one must hyperimmunize an-
imals for this response, due to a paucity of T cells in larvae. Birds and Mammals
The switch is also temperature-dependent, and as described The poor increase in affinity of chicken anti-DNP and anti-
previously for channel catfish, ectotherm T cells are quite fluorescein antibodies again indicates lower heterogeneity in
temperature-sensitive. AID is expressed in lymphocytes in chickens. Few changes occur after immunization, even if one
the spleen as well as in secretory cells.329 However, consistent waits 1 year after several injections. Perhaps similar to the
with many previous studies, there is no evidence for a typi- trout study described previously, a restricted population of
cal GC response to date in amphibians or any other ecto- high-affinity antibodies was found only after immunization
thermic species. The continued expression of AID in plasma in complete Freund's adjuvant (CFA).338 Hyperconversion
cells (and, presumably, early in embryogenesis) is of interest and somatic mutation in Ig genes have been found in splenic
and deserves further study. GC B cells after immunization.339 The relatively poor affinity
Similar to studies in mammals, the chaperone gp96 maturation of the chicken response may be due to a balance
has been shown to shuttle peptides into cells making them between gene conversion and somatic mutation. Indeed,
targets for MHC-restricted CTL lysis.330 Immunization of modification of V genes with segments of DNA is not an op-
frogs with gp96 from a thymic tumor results in the elici- timal strategy for fine-tuning antibody responses.340 In the
tation of CTL that display antitumor activity. Elegant ex- rabbit, there is also conversion/mutation by B cells in GCs
periments with gp96 vaccination have also shown that CTL after immunization. Within mammals, large variations are
activity against minor histocompatibility antigens is MHC- found from marsupials with no obvious secondary response,
restricted. As mentioned previously, NK cells have been to mouse with 1,000-fold increases in affinity, but the basis
characterized in Xenopus with mAbs that recognize non-B/T for the relatively poor responses has not been established.
cells. Those cells kill MHC class I–negative target tumor cells In conclusion, although all vertebrates have a very large
but not class I–positive lymphocytes, and after thymectomy potential for generating diverse antibodies after immuniza-
these cells are enriched in the spleen.331 CD8 + cells express- tion, only some mammals studied to date make the most
ing TCR were isolated with the same mAb, suggesting the ex- of this potential. Perhaps pressures on the immune system
istence of amphibian NKT cells; expression of the mAb epi- of cold-blooded vertebrates have been less intense due to a
tope on cells is induced by phorbol 12-myristate 13-acetate stronger innate immunity, and architecture of their lym-
(PMA)/ionomycin, and is also detected in CTL when MHC- phoid system is not optimal for selecting somatic mutants,
dependent cytotoxicity is reduced.332 Robert and colleagues or the great rises in affinity detected in antihapten responses
have developed one of the best-defined model of innate and are not physiologically relevant.297 An immune system using
adaptive immunity to viral infection in cold-blooded ver- somatic diversification at its “best” is well adapted to species
tebrates, the ranavirus FV3 in Xenopus.70,333 Involvement of where the value of single individuals is important (ie, species
CD8 cytotoxic cells and humoral responses have been stud- with small progenies); has that been the condition for the
ied over the course of primary and secondary infections with creation and selection of somatic rearrangement and of the
this virus.334,335 The system is now primed for the study of optimal usage of somatic mutations? If this explanation pro-
other cell type involvement, ontogeny, repertoire, etc. vides a rationale for the utilization of somatic mechanisms

in generating a repertoire and improving it, it does not tell to generate the repertoires: the terminal deoxynucleotidyl
us why it works so well in certain species and not in others. transferase (TdT) enzyme that modifies boundaries of re-
Perhaps one key is the organization of secondary lymphoid arranging gene segments, and somatic mutations, found
organs. Likely a combination of factors (eg, quantitative exclusively in B cells usually introduced during immune re-
and qualitative effects such as endothermy, secondary lym- sponses. However, progression of rearrangement during B-
phoid tissues, mutation versus conversion, the hypermu- and T-cell development and diversification follow different
tation mechanism itself, rates of proliferation thogen and rules in different vertebrates.220 It is conceivable that species
lymphocyte], etc.) are at work in the regulation of antibody hatching early with just a few lymphocytes are under pres-
responses.296,297 Finally, significant differences have been de- sures to develop a rapid response and may not use the same
tected in the mutational mechanism and targeting in ecto- mechanisms as species protected by the mother’s uterine en-
therms (eg, GC-richness in amphibians, tandem mutations vironment. It is also possible that immune systems of species
in sharks, TCR loci mutation in sharks, etc.) that may enrich with few offspring are under stronger pressures than species
our understanding of the general mutational mechanism, in- that have many offspring, and this could be reflected in the
cluding the enzymes involved and the targeting to V genes.318 manner diversity is generated. Studies of B- and T-cell dif-
ferentiation have been performed in many vertebrates. RAG
and TdT genes have been cloned in representatives of all ver-
Lymphoid Tissues tebrate classes, probes that allow the monitoring of lympho-
In addition to the molecules and functions characteristic of cyte development (see the following). Reagents have become
adaptive immunity, primary (lymphocyte-generating) and available permitting a monitoring of T-cell appearance in the
secondary (immune response–generating) lymphoid tis- lymphoid organs of ectotherms (crossreactive anti-CD3 sera
sues also define the specific immune system341 (Fig. 4.11). mentioned previously or TCR probes), as well as mAbs and
The thymus is present in all jawed vertebrates, and recent gene probes specific for Ig H/L chains that allow examination
evidence suggests its origin at the dawn of vertebrate emer- of B cells. As a rule, the thymus is the first organ to become
gence.342 All animals have hematopoietic cell–generating lymphoid during development. Another emerging rule is
tissues, and outside of the so-called GALT species, B cells that development of the thymus-dependent MHC-restricted
develop in such bone marrow equivalents in all jawed verte- T-cell repertoire is similar in all species, and this is reflected in
brates. With the advent of clonal selection, the accumulation the evolution of TCR gene organization described previously
and segregation of T and B cells in specialized organs for an- as well as a core set of molecules that are required for thy-
tigen presentation became necessary and indeed the spleen mocyte migration and differentiation343; in contrast, B-cell
as such an organ is found in all jawed vertebrates, but not in repertoire generation differs dramatically among different
agnathans or invertebrates. species, at times even within the same class of vertebrates.220
All jawed vertebrate species rearrange their antigen recep-
tor genes somatically (except the case mentioned previously Cartilaginous Fish
for some shark germline-joined genes276). Besides rearrange- Like all other major adaptive immune system components,
ment, with combinatorial joining of gene segments and im- cartilaginous fish are the first in evolution to possess a pro-
precision of the joins, there are two other sources of diversity totypic thymus originating from pharyngeal pouches. As
AGNATHAN CARTILAGINOUS FISH BONY FISH AMPHIBIAN BIRD MAMMAL

Thymus Spleen Thymus
Thymoid Thymus Epigonal Head kidney
Fat column organ Thymus Spleen
Peyer’s
patches
Bursa
Typhlosole Bone
Leydig’s Spleen
marrow
gland Thymus Spleen Spleen Bone Bone
marrow marrow
GALT GALT GALT GALT GALTA GALTA
1° Thymoid 1° Epigonal 1° Head kidney 1° Bone marrow 1° Bone marrow 1° Bone marrow
1° Leydig’s 1° Thymus 1° Thymus 1° Bursa 1° Thymus
1° Thymus 2° Spleen 2° Spleen 1° Thymus 2° Spleen
2° Spleen 2° Spleen 2° Lymph nodes
2° Nodes (?) 2° Peyer’s patches
Germinal centers Germinal centers
FIG. 4.11. Evolution of Lymphoid Tissues in the Vertebrates. All jawed vertebrates have a thymus and a spleen with demarcated T- and B-cell
zones. Fish have different bone marrow equivalents (epigonal, Leydig’s gland, head kidney), and amphibians are the oldest group with lym-
phopoietic bone marrow (also the first to have a typical immunoglobulin [Ig]H chain class switch, although sharks seem to have switch as
well, despite the IgH cluster organization). Germinal centers are found only in warm-blooded vertebrates. A thymus equivalent (thymoid) has
recently been discovered in agnathans.375

in mammals, it has a distinct cortex/medulla structure, It was suggested, based on the paucity of rearranged gene
and TdT expression was detected in thymocytes with cross- segments at the nonexpressed IgH loci, that there are lim-
reactive antisera and more recently by northern blotting iting factors regulating accessibility to RAG proteins and a
and in situ hybridization, where it is found throughout the short time window for rearrangement of accessible loci in
cortex.257 Interestingly, unlike most other vertebrates, age is B cells.347
not an indicator as to the size of the cartilaginous fish thy- The architecture of cartilaginous fish Ig loci allows great-
mus; it can be small or large at any stage of development. est diversity only in CDR3 because the CDR2 and CDR1 are
GALT is also found in elasmobranchs, but lymphoid tissue always encoded in the germline and V segments do not com-
in the spiral valve (intestine) clearly does not have typi- bine with (D)J segments from other clusters. Yet the num-
cal secondary lymphoid tissue structure; the spleen is the ber of possible CDR3 is essentially limitless, and the number
only tissue with compartmentalization of cells into discrete of germline clusters is also high (at least 15 genes in each
T-cell and B-cell zones.308 The Leydig’s and epigonal organs species and as many as 100 genes; and usually three rear-
(associated with the gonads) are lymphopoietic and eryth- rangement events take place because two D segments are in
ropoietic, producing mainly granulocytes and lymphocytes, each cluster). Thus, the potential diversity is greater than the
and there is high RAG expression in these tissues (see the number of lymphocytes, the general rule for generation of
following). Lymphocytes form nodules in the epigonal diversity in the vertebrates.
organ, probably indicative of differentiative events. In addi-
tion, many plasma cells are found peppered throughout the Bony Fish
epigonal, fitting with the bone marrow connection. In all teleosts examined, the thymus is the primary organ
At hatching, when dogfish embryos are exposed to water- for T-lymphocyte generation and head kidney the primary
borne antigens, structural development of the lymphomy- organ for B-cell development. The teleost thymus gland
eloid tissues is well advanced.344 In the nurse shark, neonatal originates from the pharyngeal pouches and can be uni-,
spleen white pulp consists entirely of class II–negative bi-, or trilobed, depending on the species,348 and it is the
B cells; by 5 months after birth, T-cell zones appear adja- first organ to become lymphoid. The cortex/medulla ar-
cent to the B-cell zones. Both the B-cell and T-cell zones are chitecture is not as precise in other vertebrate species, but
vascularized, and no detectable marginal zone separates red the duality of the compartment is apparent and varies from
pulp from white pulp. Class II–positive dendritic-like cells species to species.349 The spleen contains the basic elements
are found throughout the white pulp.345 seen in other vertebrates—blood vessels, red pulp, and white
In the skate Raja eglanteria, Ig and TCR expression is pulp—but the distinction between red and white pulp is less
sharply upregulated relatively late in development (8 weeks) obvious (the white pulp being poorly developed). In spleen,
by quantitative polymerase chain reaction. IgM expression the ellipsoids, which are actually terminal capillaries, have a
is first detected in the spleen of young skates but IgW is ex- thin endothelial layer surrounded by fibrous reticulum and
pressed first in gonad, liver, Leydig’s organ, and thymus.346 an accumulation of cells, mainly macrophages. Lymphocyte
In adults, Leydig’s organ and spleen are sites of the highest accumulations are often seen in their vicinity, especially
IgM and IgW expression. In nurse sharks, IgM1gj and 19S during immune responses, which have been suggested to
IgM appears in the serum before 7S IgM and IgNAR, and be primitive GCs, but they are not homologous; as men-
this profi le is reflected in the lack of IgNAR + cells in the tioned, AID expression is found in cells319 during immune
spleen until 2 months after birth. RAG and TdT expression responses. Red pulp is rich in melanomacrophage centers,
in the thymus and epigonal organ of the nurse shark sug- groups of pigment-containing cells at bifurcations of large
gests that lymphopoiesis is ongoing in adult life. In contrast blood vessels, which may regulate immune responses. The
to most other vertebrates, N-region diversity is detected (al- other main lymphoid organ is head kidney, believed to
beit reduced by approximately 50%) in skate and nurse shark function as mammalian bone marrow. The transparent ze-
IgM and IgNAR CDR3 from the earliest stages analyzed, brafish is being developed as a new model to study T-cell
suggesting that a diverse repertoire is important for young differentiation.343
elasmobranchs.308 As mentioned previously, a subset of Ig In the sea bass Dicentrarchus labrax, a mAb detects dif-
genes is prerearranged in the germline of chondrichthyans, ferentiating T cells (perhaps pre-T cells) as well as mature
and many of those germline-joined genes are transcribed in T cells as evidenced by the presence of TCR messenger RNA
the embryo and hatchling, but not in the adult. This pat- in the sorted populations. Cells seem to migrate from sur-
tern fits with the expression of the nurse shark IgM1gj with its rounding mesenchyme and subsequently mature in the
germline-joined V region, and suggests that some germline- thymus like in all vertebrates studied so far. T cells appear
joined genes “take advantage” of their early transcriptional earlier in ontogeny (between 5 to 12 days after hatching)
edge and thus some clusters can be selected for specialized than cytoplasmic Ig + pre-B cells, which are detected only
tasks in early development. With many gene clusters, it is at 52 days posthatching. Adult levels of T and B cells are
not known how “clusteric exclusion” is achieved at the mo- reached between 137 to 145 days after hatching, which is
lecular level (and why the germline-joined gene expression quite a long time compared to young amphibians.350,351
is extinguished in adult life), but as mentioned previously, Teleost RAG1 differs from mammalian RAG1 genes by
studies in two cartilaginous fish species suggest that only the presence of an intron of 666 base pairs (an intron is also
one H chain cluster is expressed in each lymphocyte.278,347 found in the sea urchin RAG1 gene in a similar position352).

Compared with other RAG1 sequences, trout RAG1 has a lymphoid organ from histologic observation, but high RAG
minimum of 78% similarity for the complete sequence expression in this tissue suggests lymphopoietic activity.359
and 89% similarity in the conserved region (amino acid The maintenance of RAG expression throughout adult life
417-1042). RAG1 transcripts are detected starting at day 20 suggests that lymphocytes are continually produced.
after fertilization. Trout TdT is highly expressed within the Thymectomy decreases or abolishes allograft rejection
thymus and to a lesser extent in the pronephros beginning capacity, MLR and PHA responsiveness, IgY antibody syn-
at 20 days postfertilization, which correlates with the ap- thesis, and all antibody responses that increase in affinity to
pearance of lymphocytes in these two tissues.353 Because the classic thymus-dependent antigens.331,360 MLR reactivity ma-
H chain cluster is in the translocon configuration and there tures before the ability to mount IgY responses in primary
are many V H families, it is assumed that diversity is gener- responses. Thymectomy at 7 days of age delays allograft re-
ated in the mouse/human mode. jection and abrogates specific IgY responses, whereas later
As described previously, studies of mature B-cell activa- in life it only abrogates antibody responses. Thymectomy
tion and homing have been done in trout. Immunization of performed later greatly affects the pool of peripheral T cells,
animals results in the production of short-term Ig-secreting as monitored with mAbs specific for molecules such as CD8.
cells in the blood and spleen and long-lived plasma cells in Early thymectomy results in the complete absence of T cells,
the head kidney. Further analysis with B cell–specific tran- but lymphocytes with T-cell markers, perhaps correspond-
scription factors like PAX5 and BLIMP1 reinforced the ing to NKT cells, can still be detected. In Xenopus, thy-
functional studies and showed that the blood contains pri- mocytes induce weak graft-versus-host reactions, whereas
marily “resting” B cells and the head kidney both plasma splenic T cells are good helpers and strong graft-versus-host
cells and B-cell precursors.314,354 These last findings appear inducers. The thymus contains some IgM-producing B cells
to be true of the cartilaginous fish as well, with the epi- and memory cells poised to switch to IgY synthesis, and in
gonal organ as the head kidney primary lymphoid tissue vitro responses are downregulated by naïve thymus cells.
equivalent. Interestingly, recent studies have shown trout Xenopus B cells respond in vitro to low doses of LPS not by
B cells to be quite efficient at phagocytosis, raising questions proliferation, but rather by Ig synthesis, and also respond to
about myeloid/lymphoid lineage commitment in the ver- PMA. Old reports of B-cell proliferation can be attributed to
tebrates.355 It is predicted that B cells from all ectothermic contaminants in LPS preparations.328
vertebrates are capable of phagocytosis (shown for Xenopus), Urodele embryos initially produce five pairs of thymic
and the work was recently extended to B1 cells in mice.356 It buds, the first two of which disappear.284 This results in a
certainly will be of interest to study these innate characteris- three-lobe thymus in Ambystoma, but in Pleurodeles and
tics of B-cell subsets in the future, considering the early ap- Triturus it forms one lobe. No cortex-medulla boundary is
pearance of B1 cells and macrophages differentiating from present, and the thymus generally resembles a canonical
the yolk sac of mouse embryos.328,357 cortex. There are at least three types of stromal epithelial
cells. There is no lymphopoietic activity in axolotl bone
Amphibians marrow, and hematopoiesis takes place in the spleen and in
In anurans, the thymus develops from the dorsal epithelium the peripheral layer of the liver. The spleen is not clearly di-
of the visceral pouches (the number of pouches varies with vided into white and red pulp.
species) and is the first tissue to become lymphopoietic. It About 40% of TCR-β VDJ junctions in 2.5-month-old
is colonized from days 6/7 onward by precursors derived Ambystoma larvae have N-additions, compared to about
from lateral plate and ventral mesoderm through the head 73% in 10- to 25-month-old animals.247 These VDJ junc-
mesenchyme. Precursors proliferate in situ as the epithelium tions had approximately 30% defective rearrangements at
begins to express MHC class II molecules but not classical all stages of development, which could be due to the slow
class I molecules. By day 8, thymic cortex/medulla archi- rate of cell division in the axolotl lymphoid organs and the
tecture resembles that of other vertebrates.328 Amphibians large genome in this urodele. As mentioned previously,
possess a spleen with red and white pulp, GALT with no or- many axolotl CDRβ3 sequences, deduced from in-frame
ganized secondary lymphoid tissue, and many nodules (but VDJ rearrangements, are the same in animals of different
no lymph nodes), with lymphopoietic activity in the kidney, origins. In contrast, in Xenopus, rearrangement starts on
liver, mesentery, and gills. The general morphology of lym- day 5 after fertilization for the V H locus, and within 9 days
phoid organs varies greatly according to species and changes all V H families are used. V H1 rearranges first followed by
with the season. In Xenopus, splenic white pulp is delineated V H3, and by day 9/10 V H 2, 6, 9, and 10 begin being rear-
by a boundary layer, and the central arteriole of the white ranged and then V H 5, 7, 8, 11 on day 13. For VL, the κ locus
pulp follicle terminates in the red pulp perifollicular area, is the first to rearrange on day 7 (2 days after V H), a situation
a T-dependent zone. Anurans, like all ectothermic verte- similar to that found in mammals. During this early phase,
brates, lack GCs. In Bufo calamita, colloidal carbon particles B cells are present in the liver, where their number increases
injected via the lymph sac are trapped by red-pulp macro- to approximately 500 cells.325 Later in larval life, rearrange-
phages, which then move through the marginal zone to the ment resumes at metamorphosis, as suggested by the low
white pulp.358 Giant, ramified, nonphagocytic cells found in incidence of pre-B cells and by the reexpression of RAG dur-
both white and red pulp have been proposed to be dendritic ing the second histogenesis of the lymphoid system. T cells
cells. Xenopus bone marrow does not appear to be a major show a similar type of RAG expression/cell renewal during

ontogeny as the B cells, and the larval and adult Vβ T-cell increased gradually during successive stages and declined at
repertoires differ significantly. Even early in development, birth. In the alligator (Alligator mississippiensis), like in mam-
tadpoles express a highly variable TCR-β repertoire despite mals after glass-wool filtration, nonadherent peripheral blood
the small number of lymphocytes (8,000 to 10,000 splenic leukocytes (PBL) responded to PHA and not to LPS, whereas
T cells); little redundancy in TCR complementary DNA re- adherent cells were stimulated by LPS.
covered from young larvae implies that clone sizes must be
extremely small, unlike in axolotls. Birds
In Xenopus, no lymphoid organ apart from the thymus The thymus, which develops in chickens from the third and
is detectable until day 12 when the spleen appears and with fourth pharyngeal pouches, consists of two sets of seven
it the ability to respond to antigen. For B cells, until this lobes each with definitive cortex/medulla. The thymus be-
time no selection occurs as suggested by the random ratio of comes lymphoid around day 11 of incubation. Splenic ar-
productive/non-productive VDJ rearrangements (2:1). After chitecture is less differentiated than in mammals. It is not
day 12, this ratio becomes 1:1 (ie, the rearrangements have lymphopoietic during embryogenesis as RAG-positive cells
been selected). Complementary DNA sequences on days 10 are found mainly in yolk sac and blood. Birds are the first
to 12 (when the number of B cells increases from 80 to 500) vertebrate group where follicular GCs and T-dependent
are not redundant as if each sequence was represented by areas comprising the periarteriolar lymphatic sheath are
one cell.325 RAG expression together with the detection of encountered. Plasma cells are located in the red pulp. γ /δ
DNA rearrangement circles in the bone marrow suggests TCR+ T-lymphocytes are chiefly concentrated in sinu-
that rearrangement is ongoing throughout life and is not re- soids, whereas α / β T cells fi ll the periarteriolar lymphatic
stricted to an early period, like in birds and rabbits. Tadpole sheath.249 Lymph nodes seem to be present in water and
rearrangements are characterized by a lack of N-region di- shore birds but not in chickens and related fowl.
versity, like in mammals but not axolotls or shark/skate (see The bursa of Fabricius is a primary lymphoid organ unique
previous discussion), and thus very short CDR3.361 During to birds in which B cells are produced.363 It arises at day 5 of
ontogeny, TdT appears in significant amounts in thymus of development and involutes 4 weeks later. T-B heterogeneity
tadpoles at metamorphic climax, but little expression is de- is obviously well defined in birds (indeed, the “B” in B cell
tected at earlier stages, which correlates well with the pau- stands for bursa.) The effects of thymectomy—T- and B-cell
city of N-region addition in larval IgH chain sequences.362 collaboration and generation of MHC-restricted helper and
Studies of the ontogeny of the Xenopus immune system have killer cells—are very similar to mammals, the other class of
revealed a less efficient tadpole immune response (skin graft warm-blooded vertebrates.
rejection and Ig heterogeneity and affinity); the absence of During the embryonic period, chicken stem cells found
TdT expression during tadpole life fits well with the findings in yolk sac and blood rearrange their IgH and L V genes si-
of lower larval Ig (and perhaps TCR) diversity. multaneously over a very restricted period of time, and very
few cells colonize each bursal follicle (about 104 follicles).364
Reptiles Three weeks after hatching, these cells have differentiated
In all reptiles studied, the thymic cortex and medulla are in the bursa and then seed the secondary lymphoid tissues,
clearly separated. The spleen has well-defined white and after which time B cells are no longer be generated from
red pulp regions, but T- and B-cell zones have not been multipotent stem cells; thus, only approximately 2 × 104 pro-
delineated with precision.348 In Chrysemys scripta, white ductive Ig rearrangements occur in the life of the chicken.
pulp is composed of two lymphoid compartments: lym- When an antiserum to chicken IgM was administered in
phoid tissue surrounds both central arterioles and thick ovo to block this early bursal immigration, there were no
layers of reticular tissue called ellipsoids. Even after para- stem cells arising later in development that can colonize
typhoid vaccine injection, splenic GCs are not formed, as the bursa, and these chickens lacked B cells for their entire
in fish and amphibians. Splenic red pulp is composed of a lives.365 Although the general Ig locus architecture is similar
system of venous sinuses and cords. In Python reticulatus, to that of frogs and mammals, only one rearrangement is
dendritic cells involved in immune complex trapping have possible as there is only one functional V L or V H on each
been identified and may be related to mammalian follicular allele.366,367 Diversity is created during bursal ontogeny by
dendritic cells. GALT develops later than spleen during de- a hyperconversion mechanism in which a pool of pseudo-
velopment, and it appears to be a secondary lymphoid organ genes (25 ψL and approximately 80 ψH) act as donors and
(but does not seem to contain the equivalent of the bursa of the unique rearranged gene acts as an acceptor during a pro-
Fabricius). Lymph node–like structures, especially in snakes liferative phase in bursal follicles. For H chains, the situation
(Elaphe) and lizards (Gehyra), have been reported. is more complex as there are multiple D elements. During
Reptiles, the evolutionary precursors of both birds and ontogeny, selection of productive rearrangements parallels
mammals, are a pivotal group, but unfortunately the functional the selection of a single D reading frame, suggesting that
heterogeneity of reptile lymphocytes is poorly documented. the many D segments favor D-D joins to provide junctions
There seems to be T-/B-cell heterogeneity because an antithy- that are diversified by gene conversion; the hyperconversion
mocyte antiserum altered some T cell–dependent functions in mechanism can also modify Ds because most donor pseudo-
the viviparous lizard Chalcides ocellatus. Embryonic thymo- genes are fused VD segments. The gene conversion process
cytes responded in MLR at all stages, but ConA responsiveness requires AID, which is also required for SHM and CSR.368

Because diversification by gene conversion occurs after Ig The potential repertoire of Ig and TCR as well as VLR-
rearrangement and cellular entry into bursal follicles, and combining sites is enormous in all vertebrates. The poten-
there is only a single germline V H and V L expressed on all tial antigen receptor repertoire in all species for both T and
developing B cells, it was tempting to implicate a bursal li- B cells is far greater than could ever be expressed in an ani-
gand binding to cell surface IgM to initiate and sustain cel- mal because of cell number limitations. Not all species or all
lular proliferation and gene conversion. However, surface gene families use combinatorial joining for repertoire build-
expression of IgM devoid of V regions permitted the typical ing, but all species assemble V, (D), and J gene segments to
B-cell developmental progression, demonstrating that such generate their functional Ig genes during B-cell ontogeny,
receptor/ligand interactions are not required.369 Thus, cur- and the imprecision of this assembly creates great somatic
rently we know little of how cells enter the bursa, which sig- diversity. Thus, from this survey in various species, one
nals induce them to proliferate/convert, and how cells arrest could not predict that there would be major differences in
their development and seed the periphery.364 immune responses in representatives of different vertebrate
classes, and yet as mentioned previously the mouse/human
Generation of Diversity in Mammals antibody responses are superior to those in many taxa.
Mechanisms leading to the generation of repertoire diversity
vary among mammalian species.220 Categories can be made More on Evolution of the Thymus
depending upon the mode of B-cell development: rabbits, The discovery of a second (cervical) functional thymus
cattle, swine, and chickens, unlike fish, amphibians, rep- in mice has raised ontogenetic and immunologic ques-
tiles, and primates/rodents, use a single V H family, of which tions.342,374 Is this second thymus the result of an atavism, or
only a few members (sometimes only one) are functional. does it correspond to what is seen in human when a cervi-
To diversify their antibody repertoire, this group uses gene cal thymus can form under certain pathogenic conditions
conversion or hypermutation in hindgut follicles of GALT during the migration of the thymus to its final mediastinal
early in life (rather than bone marrow throughout life) and location? There are examples of cervical thymi in primitive
is therefore known as the “GALT group.” At the rabbit H mammals such as marsupials but also in some prosimians.
locus, as in the chicken, a single V H is expressed in most pe- An “extra” thymus also is reminiscent of the multiple thymi
ripheral B cells. During development, B cells that have re- encountered quite frequently in cold-blooded vertebrates.
arranged this particular V in the bone marrow (and other As described, the thymus is promiscuous with regard to
sites) migrate to the appendix where this rearranged gene its precise developmental origin.284 The thymus arises from
is diversified by gene conversion using upstream donor V pouches two to six in cartilaginous fish, from the second
segments.370,371 This development of B cells in rabbits is de- pouch in frogs, the second and third in reptiles, and from
pendent on the intestinal microflora, and efforts are being the third and/or fourth in bony fish, birds, and mammals.
made to define the potential bacterial superantigens in- The final number of thymi can also be variable. It ranges
volved, as well as binding sites on IgM necessary for the dif- from five pairs of organs in sharks, to four in caecilian am-
ferentiation.372 In ruminants, the ileal Peyer patches are the phibians, to three in urodeles, and finally to one in many
bursa-like primary B cell–generating tissues.373 Although teleost fish species, anurans, and many mammals. The thy-
bursa, appendix, and sheep ileal Peyer patches show mor- mus of reptiles varies in term of location and number of
phologic similarities, the mechanisms generating diversity lobes, reflecting variation in embryonic origin. The adult
are different: conversion in the chicken and hypermutation thymus may be found anywhere from the base of the heart
in sheep, and both in the rabbit. As described previously, to the neck. For example, in lizards and snakes, there are
most of the “GALT group” also appears to lack IgD; thus, two lobes on each side of the neck with no subdivision in
IgD might serve some purpose in repertoire development in lobules. The crocodilian thymus is an elongated chain-like
some groups of mammals and not others. structure, not unlike that of birds. In turtles, the thymus is
In summary, the organization of the lymphoid tissues is a pair of lobes divided in lobules at the bifurcation of the
perhaps the only element of the immune system that shows common carotid associated with the parathyroids. The mul-
increasing complexity that can be superimposed on the tilobed thymus in birds is not the equivalent of the multiple
vertebrate phylogenetic tree. The absence of primary and thymi found in sharks because the subdivision is secondary
secondary lymphoid tissues (thymus and spleen) is corre- to primary organogenesis. Regardless of the underlying on-
lated with the absence of a rearranging and/or hypermutat- togenetic mechanism leading to the development the cervi-
ing receptor family in other animals, with the exception of cal thymus in mice, the result is suggestive of the secondary
the agnathan VLR in which the relevant lymphoid tissues “thymus spreading” in birds, but it differs with regard to an
are being defined (see the following). While all jawed verte- uneven final size distribution. All marsupials, except koalas
brates have a true secondary lymphoid tissue (spleen), ecto- and some species of wombats that only have a cervical thy-
therms lack lymph nodes and organized GALT. In addition, mus, have a thoracic thymus similar to that of placentals.
while ectotherms clearly have B-cell zones resembling fol- Some marsupials (kangaroos and possums) also have a cer-
licles, and despite the clear ability for ectothermic B cells to vical thymus; their thoracic thymus derives from the third
undergo SMH and at least some degree of affinity matura- and fourth pharyngeal pouches, whereas the cervical thy-
tion, GCs with follicular dendritic cells are not formed after mus arises mainly from the ectoderm of the cervical sinus
immunization; clearly, this was a major advance in the evo- with some participation from the second and third pouch
lution of the vertebrate immune system. (like in reptiles).

The second mouse thymus seems to show primitive char- to be expressed by lymphocytes in close proximity to the
acteristics such as the existence of a single lobe as is found foxn1-positive pharyngeal epithelial cells.375 Furthermore,
in amphibians, a superficial location, and a position com- only in these “developing” lymphocytes, but not in mature
patible with an origin involving another pharyngeal pouch VLRA-positive cells, could a high percentage (approximately
(presumably the second) that would be a marsupial and 25%) of out-of-frame VLRA genes be detected, implying
therefore perhaps a reptilian character. Embryology helps in that cells were differentiating in this region. In summary,
determining the possible scenarios: During mouse ontogeny, this tissue in lampreys, which was christened the “thymoid,”
the canonical thymus anlage can be recognized, beginning 1) is derived from the pharyngeal epithelium, 2) expresses
on day 11.5 in development, as a group of Foxn1-expressing classical thymic epithelial markers such as foxn1 and notch
cells located ventrally in the third pharyngeal pouch. If the ligands, and 3) is associated with developing VLRA cells,
cervical thymus were derived from an independent anlage, based on expression of the APOBEC family member CDA1,
one would rather expect to detect Foxn1+ cells in the endo- out-of-frame VLRA gene sequences, and failure to respond
derm outside the third pouch. This argues for a common to activation signals (such as the T cell mitogen PHA) that
origin of thoracic and cervical thymi and against a second stimulate mature lymphocytes. In addition, consistent with
thymus anlage outside the third pouch, and, hence, against the high percentage of cells with a nonfunctional recep-
the above hypothesis that the cervical thymus represents an tor, many lymphocytes undergo apoptosis in the thymoid,
atavistic organ. which is also comparable to the situation in jawed verte-
brates. Much more work is necessary to understand this sys-
A Thymus in Jawless Fish?. Until recently, it was believed tem, but the basic finding is extraordinary.
that the thymus appeared in evolution with the emergence Because the thymoid is expressed at the tips of all of the
of adaptive immunity in the extinct placoderm lineage ap- gill filaments,375 thymectomy will not be possible in jaw-
proximately 500 million years ago. There has never been any less fish. Perhaps procedures will be developed to block the
controversy concerning the presence of a thymus in all liv- interactions between the VLRA cells and the pharyngeal
ing jawed vertebrates, and as described, its requirement for epithelium, or the development of the thymoid itself can be
T-cell differentiation is universal. The lack of a thymus in disrupted. Assuming that this tissue indeed is the thymic
the jawless fish was consistent with the non-Ig/TCR adap- equivalent in lampreys, what is the significance of having the
tive immunity in these animals, but this has changed in a VLRA cells develop in a unique organ? In gnathostomes, T
new study.375 cells recognize antigen in the form of peptides in association
The finding that lampreys have two types of lympho- with MHC class I or class II molecules. Because of the high
cytes heralded a renewed search for a thymus equivalent in levels of MHC polymorphism, as mentioned previously, T
these animals.291 Indeed, there was an extensive literature on cells are positively selected in the thymus for cells that rec-
this topic over the past century identifying accumulations ognize antigen in association with the thymic MHC. Despite
of cells in various cranial regions, suggesting that lympho- major effort, neither MHC molecules nor the specialized
cytes could be differentiating in these areas. However, it has proteins associated with antigen processing have been de-
always been clear that there is no specialized tissue with tected in the jawless fish, and thus if there is positive selec-
a defined cortex and medulla in lampreys or hagfish, as is tion it must be orchestrated by a convergent system of anti-
seen in all gnathostomes. This conclusion was not well sup- gen processing/presentation. In the same vein, perhaps there
ported because there were no molecular markers for either is an AIRE equivalent expressed by the thymoid that ensures
lymphocytes or thymic epithelium in jawless fish. Indeed, deletion of self-reactive clones376 ; if so, it would also imply
consistent with almost all other accumulated data, a recent that a convergent antigen presentation system will be discov-
study concluded that there was “no evidence for a thymus ered in lampreys. It will be of interest to reexamine differen-
in lampreys.”343(p189) In that study, the major transcription tiation of lymphocytes in the pharyngeal epithelium of basal
factor involved in the development of T cells described pre- chordates; perhaps this will lead us to an understanding of
viously, foxn1, was expressed by the pharyngeal epithelium, the origins of adaptive immunity in the vertebrates. Finally,
but the lack of expression of any known lymphocyte mark- what is the significance of T cells developing in association
ers made it unlikely that this region was truly the thymus with the pharyngeal epithelium? Is it because this area in the
equivalent; a similar transcription profi le was seen in the gill gill region is evolutionarily plastic or is there some relevance
epithelium of the model basal chordate Amphioxus, which to exposure of the thymoid to the external environment?
truly seems to lack an adaptive immune system.
The discovery of “T cells” in lampreys opened a new pan- Discovery of a New Proteasome Element Expressed in
orama with the expression of T cell–specific genes besides Thymus. Recently, a new form of the proteasome only ex-
the antigen receptors. As mentioned, Pancer and colleagues pressed in the thymus was discovered in mammals, called
discovered that lamprey lymphocytes express two APOBEC β5t. The β5t gene was generated via a cis duplication from
family member genes, CDA1 and CDA2, and suggested that LMPX (β5), and in its absence cytotoxic T cells are im-
they were involved in the rearrangement (and perhaps mu- paired.377 Like many other components of adaptive immu-
tational) events in the VLR genes.288 In the T/B split paper nity we have discussed, β5t is found in all gnathostomes,
described previously, these APOBEC family members beginning with the cartilaginous fish. As we will see in the
seemed to be expressed specifically in either VLRA (CDA1) following, the LMPX equivalent of the immunoproteasome,
or VLRB (CDA2) cells. In the new study, CDA1 was shown β5i or LMP7, is polymorphic in amphibians and bony fish,

suggesting that it is a lynchpin in formation of the immuno- molecules (see the following). Besides mammals, β2m genes
proteasome.378 It is interesting as well that both β5t and β5i have been cloned from representatives of all jawed vertebrate
(and all of the immunoproteasome elements) have been lost classes. The β2m gene is outside the MHC in all tetrapods
in birds (see the following). and bony fish tested, and is a single copy gene in all species
except cod and trout, in which it has undergone multiple
duplications. Based on the levels of similarity between the
The Major Histocompatibility Complex various domains of class I and class II, it was predicted that
T cells distinguish self from nonself through the presenta- β2m was originally encoded in the MHC; indeed, recent
tion of small peptides bound to MHC class I and class II work has shown that it is linked to the MHC in the nurse
molecules (ie, MHC restriction). The genetic restriction of shark, adjacent to the ring3 gene.380 Like TCR/Ig/class I/II,
T cell–APC collaboration, processing of antigen by profes- β2m has not been found in jawless vertebrates, so its origin
sional APCs, and T-cell education in the thymus described remains mysterious.
in mice hold true (or is assumed) for most jawed vertebrate
classes. For technical reasons, no MHC-regulated T-cell re- Classical and Neoclassical Class I and Class II
sponses have been documented in cartilaginous fish, but the Class Ia (classical) and class Ib (nonclassical) genes are found
identification of polymorphic class I and II and rearrang- in all of the major groups of jawed vertebrates. Class Ia genes
ing TCRα / β genes and segregated T- and B-cell zones in are defined by their ubiquitous expression, their presence in
spleens (see previous discussion) strongly suggest that func- the MHC proper, and by high polymorphism (in most spe-
tional analyses will reveal MHC restriction of adaptive re- cies). In addition, class Ia proteins almost always have eight
sponses.257 By contrast, urodele amphibians and teleost cod conserved residues at both ends of the PBR that interact with
are notorious for their poor immune responses (see the fol- “mainchain” atoms of bound peptides and constrain their
lowing), but biochemical and molecular evidence suggests size to eight or nine residues; this feature often distinguishes
that class II polymorphism is low in the axolotl and cod have class Ia from class Ib (see the following). Thus, tight bind-
lost the entire class II–based immune system. Recent results ing of peptides, a likely source of conformational changes in
in agnathans and other vertebrates have opened our eyes to class I allowing transport through the endoplasmic reticu-
new functions in MHC. lum and cell surface expression, is an evolutionarily con-
served trait.381 In nonmammalian vertebrates, one of these
Class I/II Structure Through Evolution residues at the C-terminus is lysine rather than tyrosine, the
The three-dimensional organizations of class I and class II functional significance of which is unclear.379
are remarkably similar: the two membrane-distal domains The class Ia/Ib distinction holds in most taxa: one to
of both molecules form a PBR composed of two antiparallel three polymorphic class Ia genes are expressed ubiqui-
α helices resting on a floor of eight β strands, and the two tously in most species, whereas other minimally polymor-
membrane-proximal domains are IgSF C1. Although se- phic or monomorphic class Ib genes are expressed in a
quence identity among class I and class II genes in vertebrates tissue-specific fashion. The class Ib genes can be split into
is low (like most other immune genes), the four extracellu- two major groups: one set that is most related to the class
lar domain organization and other conserved features are Ia genes within a taxon and thus recently derived, and one
likely to be found in the ancestral class I/II gene.379 An intra- group that is ancient and emerged early in evolution.381 In
chain disulfide bridge exists within the class I PBR α2 and mouse and human, the set most closely related to class Ia
class IIβ1 domains, but not the class I/II PBR α1 domains, genes are closely linked within the MHC. In nonmam-
and phylogenetic trees show that these respective domains malian vertebrates, however, this first set is found in gene
are most similar. Bony fish class II α1 domains, like class II clusters on the same chromosome as the MHC proper but
DMα molecules, do have a disulfide bridge. The exon/intron far enough away to segregate independently from MHC (eg,
structure of class I and class II extracellular domains is also chicken [Rfpy] and Xenopus [XNC ]). One Xenopus class
well conserved, but some teleosts have acquired an intron Ib gene is expressed specifically in the lung and thus likely
in the exon encoding the IgSF β2 domain. Other conserved has a specialized function, and another gene (XNC10) is
features of class I genes include a glycosylation site on the expressed by T cells and may serve as a ligand for uncon-
loop between the α1 and α2 domains important in biosyn- ventional subsets of T cells.382,383 Chicken Rfpy is associated
thesis (shared with class II β chains), a Tyr and one to three with resistance to pathogens, and the recent structure of one
Ser in the cytoplasmic regions that can be phosphorylated in of these class Ib genes has shown an unusual hydrophobic
mammals, as well as several stabilizing ionic bonds. Class II structure in the groove.384 Class Ib genes related to but un-
with its two TM regions differs from class I with only one; linked to the classical class I have also been found in bony
conserved residues in the class II α and β TM/cytoplasmic and cartilaginous fish, and a lineage of class II–linked class
regions facilitate dimerization. In summary, because se- I genes was discovered in bony fish.385 Thus, class Ib genes
quence similarity is very low among MHC genes in different that arise in each taxon seem to have true class I–like func-
taxa, these conserved features are important for function, tions, but perhaps have become specialized (sometimes the
biosynthesis, and maintenance of structure. distinction between class Ia and class Ib is blurry; see the
β2m was the second IgSF molecule (C1 type) ever to be following). The second set of older class Ib genes that pre-
identified, originally found at high levels in the urine of dates divergence of taxa can have very different functions.381
patients with kidney disease. It associates with most class I For example, the neonatal Fc receptor is involved in binding

and transport of IgG molecules across epithelia as well as immune protection during larval life; perhaps expression of
protecting them from degradation (the Brambell receptor). class Ia is limited to organs that undergo massive destruc-
Furthermore, molecules only described so far in mammals tion and remodeling at metamorphosis. Class II molecules
are composed only of a PBR without IgSF domains; these also change their distribution after metamorphosis and are
unusual class I molecules do not bind peptides but rather highly expressed by unstimulated T cells.391 Axolotl class II
are important for the regulation of NK- and T-cell function molecules are also regulated differentially during ontogeny,
during infection. The paradigm for these SOS responses is expressed in young animals on B cells, and then expanded
the MHC class I-related protein (MIC) and UL16-binding to all hematopoietic cells, including erythrocytes, later in
protein (ULBP) class Ib molecules, which clearly do not life.392 Changes in MHC expression are not correlated with
bind peptides. Some teleost class Ib genes that fall outside cryptic metamorphosis in axolotls, but class II expression by
the major cluster of fish class I genes may fit into this cat- erythrocytes is correlated to the switch from larval to adult
egory.385,386 Finally, molecules like CD1 bind nonpeptidic globins. Unlike Xenopus, class I transcripts isolated so far
antigens for presentation to innate-like NKT cells. The phy- are expressed early in ontogeny, from hatching onwards.
logenetic analysis predicts that CD1 and FcRN are old class I Carp class I and class II transcripts are detected in em-
genes (see the following on CD1); the age revealed by the bryos 1 day after fertilization and reach a plateau at day 14.
phylogenetic tree also correlates well with the hypothesis However, the suspected class Ia protein does not appear
that ancient duplication events predating the emergence of until week 13, whereas β2m can be detected several weeks
jawed vertebrates resulted in the appearance of CD1, FcRN, earlier.393 It was suggested that another class I molecule is ex-
and MHC-linked class I genes (see the following). Was the pressed during early development of the carp hematopoietic
original function of class I linked to antigen presentation system, perhaps one of the unusual nonclassical molecules
(peptidic or otherwise), induction of an SOS response, or to that groups outside the teleost cluster. Interestingly, class I is
housekeeping functions? We do not have the answer because expressed in the brain of young, but not adult fish, suggest-
class Ia and class Ib molecules are present in the oldest living that class I molecules may play a role in neurogenesis.394
ing gnathostomes. The discovery of class I–like genes in ani- As mentioned previously, the cod was recently shown to
mals derived from ancestors predating adaptive immunity have lost class II genes, as well as the invariant chain and
or in the jawless fish would help resolve this question, but CD4 genes.395 The apparent lack of the entire class II system
to date no recognizable MHC molecules or their kin have of antigen recognition correlates well with the inability of
been detected in prejawed vertebrates; this topic is especially cod to make antigen-specific antibody responses.396 The au-
of interest since the dichotomy of lymphocytes as well as a thors speculate that the large numbers of class I molecules in
thymus candidate in jawless fish were discovered.291,375 cod may somehow compensate for the loss of a class II sys-
Class II molecules also have nearly invariant and evolu- tem. Axolotls, which have very low class II polymorphism,397
tionarily conserved residues that bind to main-chain atoms also have a highly expanded class I system.398 It is possible
of peptides, but these are in the center of the groove.379 Thus, that in both cases there has been a “use it or lose it” scenario,
tight binding to main-chain peptide atoms occurs in the in which a major arm of the immune system was lost due to
center of the class II PBR, and peptides are free to protrude a low pathogen load.
from both ends. The only nonclassical class II molecules so
far identified are the previously mentioned DM molecules Major Histocompatibility Complex Gene Organization
that lack these residues and DO proteins. DM molecules so As class I and class II proteins are structurally similar, it is
far have been cloned only from tetrapods177 and have not been no surprise that their genes are linked, a primordial trait
detected in any fish species despite the large genomic and EST subsequently lost only in bony fish (Fig. 4.12).2 But why are
databases for the bony fish. Thus, they either had not emerged structurally unrelated class I processing genes, including the
at the time fish arose or were lost in the bony fish lineage; immune proteasome components lmp2 and lmp7 and the
phylogenetic trees suggest the latter, and would be consistent TAP and TAPASIN genes, also found in the MHC? There
with the rapid rate of genome evolution in the teleosts. DO are two possible scenarios: primordial linkage of ances-
class II molecules, believed to modulate DM function, have tral processing and presenting genes in the MHC, or later
only been detected in placental mammals.387 The invariant recruitment of either the processing/presenting genes into
chain is another protein found only in gnathostomes388 ; its re- a primordial MHC. Based on the presence of similar clus-
cent role in class I cross-presentation suggests that it may have ters of MHC genes on paralogous chromosomal regions in
been important for biosynthesis of both class I and class II.389 humans and mice, Kasahara et al.17,399 proposed that ances-
tors of class I, class II, proteasome, transporter, and class III
Class I/II Expression genes were already linked before the emergence of the adap-
In Xenopus species, immunocompetent larvae express high tive immune system. Genomewide duplications around the
levels of class II on APCs such as B cells, but express only time of the origin of vertebrates (the so-called 2-Round or
very low levels of class Ia molecules on all hematopoietic 2R hypothesis), as proposed by Ohno et al.,16 may have pro-
cells until metamorphosis.328 Expression of the immuno- vided the raw material from which the immune system genes
proteasome element lmp7 and all identified class Ib isotypes was assembled (see the following; Fig. 4.13). As for all other
is also very low.390 Larval skin and gut, organs with epithelia adaptive immune genes described so far, neither class I/II
in contact with the environment, appear to coexpress class I nor immunoproteasome/transporter associated with anti-
(transcripts) and class II. Such expression may provide gen processing (TAP) have been isolated from hagfish nor

HYPOTHETICAL JAWED VERTEBRATE ‘UR’ MHC NURSE SHARK
Bf/C2 (2 genes)*
TAPBP (IgVC1)*
COLL11A2*
NOTCH4*
LMP 7/X*
ZNF297*
MECL1*
LMP7χ*
CL IIα/β
B7-like*
HSPA1*
FABGL*
TNFSF*
RING3*
RING3*
LMP 2*
LMP 7*
LMP 2*
LMP 7*
CL IIαβ
RXRB*
PBX2*
BAT2*
PPT2*
TAP 1
TAP 2
TAP 1
TAP 2
NKR*
TNX*
CL Ib
CL Ib
CL Ia
β2m
β2m
CLIa
C4*
C4*
Bf*
– Class I, II, III genes linked
Class I/II Class III – True class I region
TELEOST
COL11A2*
LMP 2/7*
ZNF297*
SACM2L
RXRB-L*
MECL1*
CL II αβ
FABGL*
KNSL2*
TAPBP*
RING3*
LMP 2*
LMP 7*
SKII2W
CSK2B
RPS18
PBX2*
CL II β
CL II β
BAT2*
PPT2*
DAXX
TAP 2
TAP 1
LMP*
HKE4
CL Ia
CL Ia
CL Ia
DSP
C4*
Bf*
– Loss of class II gene linkage to class I by translocation or differential silencing after polyploidization
– Loss of class III immune genes
– True class I region
XENOPUS
XMIV (IgSFVJ) (At least 6 loci)

“Adaptive MHC”
Class Ib (XNC) (Many loci)

CL IIαβ (At least 4 loci)
HSPA1 (At least 2 loci)

TNFSF (3 loci)*
CTX (IgSFV)
RAGE (IgV)
Ly6 (2 loci)
COL11A2*
C5NK2B*
ZNF297*
NOTCH*
MECL1*
FABGL*
KNSL2*
TAPBP*
RING3*
LMP 2*
LMP 7*
DAXX*
TNXB*
PBX2*
LPAAT
BAT2*
RxRB*
TAP 1
TAP 2
DMα
PPT2
DMβ
CL Ia
AIF1
NEU
IKBL
C4*
C2*
Bf*
– Highly conserved MHC compared with humans

– Class II, class I, TAP, proteasome genes tightly linked (Adaptive MHC) and true class I region
– MECL1 in class III region
– IgSF(VJ) NK-like receptors in class III region
CHICKEN
BG (>10 loci IgV)
R-fpy*
CD1 (2 loci)
TAPBP**
BG (IgV)
RING3*
LECTIN
CL αβ
CL IIβ
CL IIβ
TAP 1
TAP 2
CL IIa
DMα
CL Ib
DMβ
DMB
CL Ia
CL Ia
NKR
C4*
Bf*
– Compaction of genome including MHC; loss of “non-essential” genes especially LMP from the genome
– Expansion of non-classical class I/II far from MHC proper
– CD1 genes linked to MHC
– C type lectin NK receptor genes linked to MHC
– TAP/Class I co-evolution demonstrated
HUMAN
“Inflammatory region”
BTN (8 loci, IgVC1)
CL Ib (ABC-like)
NEU (sialidase)
TNFSF (3 loci)*
HSPAI (3 loci)*
C4 (2 loci Cʹ)*
BTL II (IgVC1)
CL II (DM) αβ
CL II (DQ) αβ
CL II (DR) αβ
CL II (DP) αβ
NKP30 (IgV)
RAGE (IgVo)
CL Ia (A,B,C)
CL II (DO) α
CL II (DO) β
CL Ib (MIC)
MOG (IgV)
COL11A2*
G6 (3 loci)
NOTCH4*
ZNF297*
SACM2L
C2 (Cʹ)∗
KNSL2*
TAPBP*
RING3*
LMP 2*
LMP 7*
SKII2W
DAXX*
TNXB*
FABGL
Bf(Cʹ)*
CSK28
RXRB*
RPS18
PBX2*
LPAAT
BAT2*
PPT2*
TAP 1
TAP 2
HKE4
LST1
AIF1
KBL
Extended class II Class II Class II Class I
– Translocation of class I and subsequent expansion

– Translocation of MECL1 to chromosome 16
– IgSF NK Receptor (NKp30) in class III region
FIG. 4.12. Major Histocompatibility Complex (MHC) Evolution. White indicates that the gene is found in at least one other species besides
human; black designates a gene with known or inferred immune function; asterisk indicates an ancestral gene found on at least two paralo-
gous MHC regions in mammals; large box indicates at least two genes in the particular region; double slash indicates linkage far away on
the same chromosome; space between lines indicates that the genes are on different linkage groups. The teleost MHC is a composite of the
zebrafish, trout, Fugu, and medakafish maps. Extensively modified from Flajnik and Kasahara.498

(C-type lectin)
(proteasome)
(cathepsin)
(IgSF)
B2M
Proto-MHC RXR PSMB TAP TAPBP TRIM RFX CTS NOTCH VAV TNFSF JAK MHCI/II NKR BRD B7
chromosome
Two rounds whole
genome duplication
• LRC (Chr 19q13) KIR, IgSF, ‘leukocyte’ • (Chr1p11-32) NOTCH2, JAK1, VAV3, JUN, B7H4,
• NKC (Chr 12p11) NKG2, CD4, LAG3, RXRG, CTSK/S, TNFSF4, 6, 18, (TRIM)n
A2M (C3-like), TAPBP-L, TAP-L • MHC (Chr5p21) TAPBP, TAP, PSMB8/9, C4,
TNSF1–3, NOTCH4, RXRB, (TRIM)n
• (Chr 14q11) CTS, PSMB5, 11, 5t
• (Chr19p13) NOTCH3, C3, JAK3, VAV1,
• (Chr 11q12) CTSD, CTSC
TNSF4/7/9, PIAS4
• (Chr9q32) JAK2, CTSL, CTSL2, NOTCH1, DXRA,
PSMB7, C5, VAV-2, TNSF8, TRIM14, TRIM32
FIG. 4.13. Genomewide Duplication (2R) Model of Major Histocompatibility Complex (MHC) Evolution. Displayed are the MHC paralogous
regions and their human chromosomal designations, first identified by Kasahara and modified over the years.2,17 Note that such paralogous
regions have also been studied for chemokine receptors,477 tumor necrosis factor and tumor necrosis factor receptors,459 immunoglobulin su-
perfamily molecules involved in cell-cell interactions,442 and others. Note that the natural killer cell complex and leukocyte receptor complex,
respectively, are likely to have been in the Ur MHC based upon linkages seen today in various taxa. The box on the right indicates the original
MHC paralogous regions identified by Kasahara on chromosomes 1, 6, 9, and 19.17 The box on the left shows paralogous regions uncovered
in further analyses.500 Modified from Flajnik and Kasahara.2
lampreys, and all of these genes as well as other genes in- (ie, MHC syntenic groups found on different mammalian
volved in immunity could have emerged as a consequence chromosomes resulted from ancient block duplications), it is
of the duplications. Because class I genes are found on two expected that the physical association of ancestral class I, II,
or three of the clusters, class I–like molecules may have pre- and III genes predated the emergence of jawed vertebrates,
dated class II in evolution. Indeed, NK-like recognition of a and such syntenies in ectothermic vertebrates are not sur-
class I or class I–like molecule encoded in an ancestral link- prising. Indeed, linkage studies in nonvertebrates Amphioxus
age group may have been at the origin of the adaptive im- and Ciona do support an ancient linkage of class I, II, and III
mune system (see the following). genes.405 Taken together, the data reveal that lack of synteny
TAP is an interesting case in that the MHC-linked TAP of class I, class II, and class III genes in teleosts is a derived
1/2 genes were clearly not part of the Ur MHC, but rather character. Independent assortment of class I and class II may
they were “recruited” to the MHC early in evolution. TAPs allow these genes to evolve at different rates: in some tele-
are members of the very large ABC transporter family, osts, class Ia alleles form ancient, slowly evolving lineages,
and are most closely related to bacterial ABC transporters; whereas class II genes evolve at similar rates as mammalian
this suggests that TAP1/2 were derived from a bacterial, or MHC alleles.406,407
more likely a mitochondrial gene, via horizontal transfer.400 The chicken MHC, the B complex, is on a microchro-
Furthermore, a close homologue of TAP1/2, TAP-L (TAP- mosome, and intron sizes and intergenic distances are both
like), unlike all of the other Ig/TCR/MHC genes we have quite small so that the entire complex is only a few hundred
discussed, is present in jawless fish.401 The function of TAP-L kb as compared to over 4,000 kb in humans and Xenopus.408
is not known, but it may be involved in cross-presentation402 Class Ia (BF), class IIβ (BL and DM), and TAP genes are in
or “typical” presentation in jawless fish. In summary, the the MHC, but there is no evidence for immunoproteasome
TAP genes are not following any of the “rules” that seem to genes, and almost all class III genes have been deleted except
hold true for the other genes involved in adaptive immunity, for C4. The quail MHC is similar, although is somewhat
and this story will be exciting to monitor, especially in the expanded, especially in genes related to the C-type lectin
context of the agnathan VLR system. NK receptors.409 Although most class III genes are found
In all nonmammalian vertebrates, and even in marsupi- on other chromosomes, lmp2/7 and MECL1 genes are ac-
als, the immunoproteasome and TAP genes are closely linked tually absent from the genome; indeed, peptides bound to
to class I genes, not to class II, in a true “class I region.”2 This chicken class I molecules sometimes have C-terminal glu-
result is most striking in bony fish (Fugu, zebrafish, medaka, tamic acid or aspartic acid, which are rare after proteoly-
trout) because class I/lmp/TAP/TAPBP and class II are found sis by mammalian proteasomes containing lmp2 and lmp7.
on different chromosomes.403,404 The class III region, histori- Indeed, the recent crystal structure of a chicken class I mol-
cally defined by the innate immune genes such as Bf/C2, and ecule shows that the alleles can interact with a broader array
C4 are also present in the Xenopus and elasmobranch MHC, of peptides as compared to mammalian class I alleles; the
showing that the class III association of class I/II with such PBR is “broader” and one of the conserved peptide-binding
genes is ancient.177 If Kasahara’s interpretation is correct residues is noncanonical.410 To explain the correlation of

diseases with particular haplotypes, Kaufman proposed that because of the poor immunogenetic aspects of the reactions.
the chicken has a minimal essential MHC composed of only However, the genetic regions and mechanisms respon-
those genes absolutely required to remain in the complex. sible and molecular underpinnings of such alloreactions
This concept has been reinforced recently by an analysis of are now becoming known, and no (or little) resemblance
resistance to viral infection (Rous sarcoma virus) governed to the vertebrate MHC has been found even though re-
by classical class I molecules.411,412 Additionally, the presence gions homologous of the extended vertebrate MHC have
of polymorphic TAP alleles closely linked to particular class been identified in some organisms but without functional
I alleles has demonstrated coevolution of the transporters correlation.405,421
and peptide-binding molecules.413 Colonies of Porifera, Cnidaria, Bryozoa, and Tunicata
Surprisingly, CD1 genes are closely linked to the chicken (see Fig. 4.1) often compete for space and may develop his-
MHC.414,415 As mentioned previously, CD1 genes in mam- tocompatibility reactions in the zone of contact. In addi-
mals seem to be on one of the MHC paralogous regions tion, cell-lineage parasitism, in which the somatic and/or
and it was suggested that it arose as a consequence of the germ cell lineage of one partner replaces that of the other,
en bloc duplication. While this is still possible (ie, there was may ensue if colonies fuse into a chimera. Thus, effector
differential silencing of CD1 and other class I genes on the functions following allorecognition also protect the ge-
two paralogous chromosomes), the more likely scenario is netic integrity of the individual. In some species, fusion
that CD1 arose by gene duplication within the MHC itself in is apparently restricted to tissues of the same individual
an ancestor of warm-blooded vertebrates; no bonafide CD1 (complete matching), in other species such as bryozoans,
genes have been detected to date in any fish or amphibians, fusion also occurs between genetically distinct individuals
so the entire NKT system may be specific of warm-blooded if they share kinship (partial matching). Two divergent in-
vertebrates. vertebrate phyla have been studied in detail: Cnidaria and
In summary, in all animals except placental mam- Tunicata.
mals, classical class I genes map closely to the TAP, lmp,
and tapasin genes, suggesting that the processing, trans- Porifera
port, and presenting genes were in an original “class I Sponges (Porifera) and placoza are the phylogenetically
region.”416 The tight linkage of the functionally, but not oldest extant metazoan phyla. Sponges, whether marine
structurally, related genes strongly suggests that such genes or freshwater species, possess a sophisticated histocom-
coevolve within particular MHC haplotypes. Indeed, in patibility system.422 Elements of the sponge immune sys-
Xenopus there are biallelic lineages of class Ia, LMP7, and tem involved in these reactions have been analyzed at the
TAP, which are always found as a set in wild-caught ani- molecular level. Sponge cells associate in a species-specific
mals.417 Although teleosts underwent an explosive adaptive process through multivalent calcium-dependent interac-
radiation 100 million years ago and primordial syntenies tions of carbohydrate structures on a 200 kd extracellu-
have been lost in many cases, there are deep lineages of lar membrane-bound proteoglycan called “aggregation
class Ia genes in many species, also found for Xenopus and factor,” well studied in Microciona and Geodia .422 The
cartilaginous fish class Ia genes. A study in medaka sug- glycan moiety is involved in cell adhesion and exhibits
gests that divergent noncoding regions between the class differences in size and epitope content among individu-
I–processing and –presenting genes do not permit recom- als, suggesting the existence of allelic variants. Therefore,
bination between lineages, hence preserving the linkage strong carbohydrate-based cell adhesion evolved at the
disequilibrium.418 Nonaka et al. proposed that these deep very start of Metazoan history. Other genes involved in
proteasome/class I lineages emerged at the dawn of verte- these reactions include one that is similar to the verte-
brate immunity and may even have been maintained per- brate MHC-linked allograft infl ammatory factor and an-
haps by convergent evolution in certain groups.419 In (most) other to the T-cell transcription factor. AIF-1 and T-cell
eutherian mammals, the class I region is not closely linked transcription factor genes are upregulated in vivo after
to lmp/TAP and is very unstable, with rapid duplications/ tissue transplantation, and in vitro in mixed sponge cell
deletions expected in a multigene complex (see Fig. 4.12); reaction.423 Polymorphic IgSF molecules are found on the
the same class I instability extends to the non–MHC-linked surface of sponge cells, but their relationship to allorecog-
class Ib genes in Xenopus species.420 nition events (if one exists) is not clear.424
Allogeneic recognition in vitro led to apoptotic cell death
in one partner and survival in the other.425 The process is
EVOLUTION OF ALLORECOGNITION controlled by a differential expression of the proapoptotic
Histocompatibility Reactions in Invertebrates and prosurvival proteins that are characteristic for the ini-
Scrutinizing graft rejection within a species (allograft) tiation of apoptosis (caspase, MA3, ALG-2 protein) and the
across the animal kingdom demonstrated that allorecog- prevention of programmed cell death (2 Bcl-2 homology
nition was almost universal among metazoa. Scrutinizing proteins, FAIM-related polypeptide, and DAD-1–related
specific memory in the same systems was the major tool protein). In an apoptotic mixed cell combination, character-
to establish the universality or not of adaptive immunity. istic apoptotic genes were expressed, while in the nonapop-
Moreover, it fueled speculations the origins of vertebrate totic aggregates the cell-survival genes are upregulated.423
MHC. Most of these studies led to inconclusive results In another species, Microciona, allogeneic interactions also

induce cellular reactions involving gray cells (sponge im- alr2 level as nearly all sampled alleles encoded unique gene
munocytes) and finally apoptosis. Analogous (but most products.431
likely not homologous) to T-cell responses, the response is
inhibited by cyclosporin A.426 Urochordates
From observation of 50 pairs of larval grafts within one Compared to Cnidaria, tunicates shared a recent common
F1 progeny of the marine sponge Crambe, 75% could fuse, ancestor with the vertebrates (see Fig. 4.1), and thus might
a proportion suggesting that the genetic control depends on be expected to have a histocompatibility system more related
one locus and sharing of one haplotype results in fusion; to MHC. Allorecognition has been studied in both colonial
a 100% fusion between mother and offspring is consistent and solitary ascidians. In Botryllus, a colonial ascidian, extra
with this interpretation. Unfortunately, the issue is com- attention was afforded this system because the locus con-
plicated as individuals from a given mother may not have trolling histocompatibility was linked to (or was the same
the same father. Still, the few data available are consistent locus as) the locus controlling fertilization by preventing
with a single or at least major histocompatibility locus (or self-fertilization. For fusion, at least one histocompatibility
region), the “rule” in other invertebrate phyla described in locus must be shared between the colonies and for fertiliza-
the following. tion the sperm must be mismatched from the egg.432
Metamorphosis in Botryllus is followed by budding that
Cnidaria eventually gives rise to a large colony of asexually derived
The existence of highly polymorphic histocompatibility loci genetically identical individuals (zooids), united through a
was demonstrated long ago in various corals where apop- vascular network. At the periphery of the colony, the vas-
tosis induction was responsible for the death of partners in culature ends in ampullae, which are the sites of interac-
incompatible combinations,154 whereas in sea anemone ne- tion when two colonies meet during their expansion. The
matocyte discharge was induced between incompatible indi- interaction results in either fusion of the two ampullae to
viduals.427 The colonial cnidarian Hydractinia was the only form a single chimeric colony sharing a common blood
species to provide a model to analyze genetic control of such supply or a rejection reaction during which the interacting
reactions. In 1950, Hauenschild noticed that allorecognition ampullae are destroyed, thus preventing vascular fusion.
seemed to be under the control of a single genetic region Hemocytes (morula cells) are involved in the reaction.433
with multiple alleles.428 Colonies of Hydractinia encrust the Fusion or rejection is governed by a single highly polymor-
shells of hermit crabs, where they grow by elongation and phic (tens to hundreds of alleles) locus called the FuHC (for
branching of stolons. Embryos and larvae fuse indiscrimi- fusion/histocompatibility; 10 wild-type individuals col-
nately. However, when two or more larvae are recruited to lected from around the Monterey Bay area yielded 18 cFuHC
the same substratum, colonial forms may come into contact alleles). As mentioned, when two colonies share one or both
through their stolons. If the two colonies are histocompat- FuHC alleles, they will fuse; rejection occurs if no alleles
ible, stolon tips adhere and fuse, establishing gastrovascu- are in common. However, like in Hydractinia, intermediary
lar connections and a permanent genetic chimera. If tips pathways have been reported in the past and have not been
of incompatible colonies fail to adhere, they swell (hyper- entirely elucidated.
plastic stolons) with the migration of nematocysts, which The C-terminal region of the FuHC molecule consists of
discharge and damage the tissues. In addition, transitory a nectin-like segment with three Ig domains, showing most
fusions can also occur in a few cases. Similar to the genetics similarity to chicken Igsf4 and related members conserved in
of the sponge alloreactions previously mentioned (and in all vertebrates as well as to one of the Ciona nectins.434 Those
other invertebrates and plants), colonies fuse if they share members are found in a tetrad of paralogs in vertebrates but
one or two haplotypes, reject if they share no haplotypes, not linked to the tetrad of MHC paralogs (see the following).
and display transitory fusion if they share only one allele The N-terminal region contains an epidermal growth factor
at one haplotype and no alleles at the other. Examination (EGF) domain and some other unrecognizable regions not
of the polymorphic locus governing this reaction (alr) re- conserved among ascidians. Any two FuHC alleles differ by
vealed the involvement of two closely-linked polymorphic an average of 4% at the nucleotide level. Unlike MHC class
loci 1.7 cm apart (likely encoding receptor and ligand), alr1 I and class II, polymorphic residues in FuHC alleles are not
and alr2.429 Alr1 encodes a three IgSF domain surface mole- concentrated in particular regions, and alternative splicing
cule, not unlike other metazoan nectin-like molecules, with can generate a fragment devoid of the IgSF moiety.
a cytoplasmic tail equipped with multiple signaling motifs Some 200 kb away from FuHC is a second polymorphic
(including a ITAM-like motif). In the first domain, the mol- (and polygenic) locus, fester, which is inherited in distinct
ecule shows a high level of variability at particular residues, haplotypes.435 Diversified through extensive alternative
suggesting positive selection. Alr1 is embedded in a genetic splicing, with each individual expressing a unique reper-
region consisting of multiple IgSF members. Alr2 encodes toire of splice forms (each individual expressing up to three
another highly polymorphic gene with three IgSF domain splice variants in addition to the regular full-length prod-
IgSF with the distal V-like domain being the most variable, uct), it potentially exists as both membrane-bound and se-
and an ITIM-like motif in the cytoplasmic tail adjacent to creted forms, all expressed in tissues intimately associated
various phosphorylation motifs. Like the corals studied by with histocompatibility. After fester knockdown, the histo-
Theodor,430 a high level of allele diversity was found at the compatibility reaction is blocked at the stage of initiation as

if the colonies ignored each other. By contrast, when Fester invertebrates, perhaps driven by “selection operating on
was blocked with specific mAbs, fusion reactions were unan error-prone genetic system for self-recognition that is
affected, but rejection reactions were turned into fusions perhaps constrained by derivation from a gametic function
in an allele-dependent manner. These data combined with selected to reduce inbreeding.443 Furthermore, even in mam-
its genetic location suggest that Fester encodes the FuHC mals there is a large literature suggesting a selection both at
ligand. Fester contains a short consensus repeat (or sushi the mate-choice and pregnancy levels for preserving hetero-
domain) often found in vertebrate complement receptors. zygocity at the MHC.443a However, the possibility of a com-
Beside these two polymorphic products, another Botryllus mon genetic system, or linked systems, governing fusion
cell surface–expressed nonpolymorphic molecule related to and gametic compatibility awaits confirmation. Because
fester, uncle fester, is required for incompatibility reactions, animals that are neither sessile nor unable to disperse their
while fester seems to be required for allele discrimination gametes can possess alloimmune responses is inconsistent
and inhibition of killing.436 So, several independent path- with the general hypothesis. Moreover, inbreeding avoid-
ways seem to control the final outcome of the interaction ance can only explain the selection of histocompatibility al-
between individuals.437 It should be mentioned that a recent leles if the histocompatibility loci are genetically linked to
study, based essentially on the incongruence between his- a large fraction of its genome. This is inconsistent with the
tocompatibiity profi les and FuHC polymorphism, questions tight linkage of histocompatibility genes to a single major
the validity of these reports.438 locus, especially in invertebrates. So inbreeding avoidance
is unlikely to contribute significantly to the selection of
Solitary Ascidians histocompatibility alleles, although in jawed vertebrates,
In Halocynthia roretzi, a “polymorphism of color” has been selection for heterozygocity at the MHC itself has obvious
observed and histocompatibility reset by a mixed hemocyte advantages.
technique in vitro resulting in a melanization reaction likely Another hypothesis is that alloimmunity was selected be-
to involve the PPO cascade.439 Depending on the strains, the cause it avoided intraspecific parasitism and/or competition
percentage of positive reaction varied from approximately for attachment sites.444 Indeed, after fusion of compatible col-
55% to 70%, indicative of polymorphism. Grafting experi- onies, bloodborne germline or totipotent stem cells are trans-
ments had already shown the existence of allorecognition ferred between colonies, and can expand and differentiate in
in solitary ascidians, and investigation at the cellular level the newly arising, asexually derived individuals of the vascu-
had demonstrated the occurrence of cytotoxic cells in such lar partner.445 This can result in a situation where only one
organisms.440 In order to shed light on allorecognition in genotype is represented in the gametic output of the fused
urochordates and on the molecules involved in preventing individuals. The FuHC polymorphism in Botryllus could
self-fertilization, gonadal complementary DNAs of three ge- function to restrict to compatible individuals the vascular fu-
netically unrelated Ciona intestinalis individuals were com- sion and the germline parasitism. The high allelic polymor-
pared by suppression subtractive hybridization. This led to phism characteristic of all invertebrate recognition systems
the discovery of the highly polymorphic variable complement may have evolved in response to selection for fusion with self
receptor–like 1 gene coding for a transmembrane protein rather than kin. Fusion with self will allow the development
with several short consensus repeat domains (short con- of a colony that benefits from fusion while eliminating the
sensus repeat/complement control protein [CCP]), a motif possible cost of somatic cell parasitism, and thus would be
shared with the variable fester receptor of Botryllus described the raison d’etre of the allorecognition mechanisms.446
previously.441 Genes encoding variable complement receptor– However, most animals are not sessile and thus are nei-
like are in the same linkage group as a set of IgSF domains ther prone to intraspecific competition nor to compete
with homology to nectin (CD155/poliovirus receptor [PVR], for limited substrate attachment sites. Tunicates besides
cortical thymocyte protein [CTX]/junctional-adhesion mol- Botryllus have evolved differently (eg, in Diplosoma, large
ecule [JAM] family members, etc.) and other adhesion mol- numbers of chimaeras are encountered in nature). In addi-
ecules of which related members can be found also on one tion, the intraspecific competition hypothesis demands that
genetic region in vertebrate, the 19q 34 human chromosome individuals maintain expression of histocompatibility al-
segment with the extended LRC (see the following).442 leles, even when the expression of these alleles enables their
own destruction during intraspecific competition. These
considerations suggest that intraspecific competition might
The Meaning of Histocompatibility Reactions in affect histocompatibility allele frequencies in some organ-
the Invertebrates isms under certain conditions.
The association between allorecognition in Botryllus and Both previous hypotheses are essentially based on ob-
fertilization led to the proposal that histocompatibility servation made in colonial tunicates, but colonial living
systems were selected during evolution to avoid inbreed- has evolved several times independently. Urochordates are
ing. The hypothesis made sense in the case of sessile colo- most related to the vertebrates, but the data argue against
nial invertebrates that might have difficulty dispersing their any of the genes involved in their histocompatibility reac-
gametes and therefore are susceptible to inbreeding depres- tions being ancestral to MHC class I and class II. It is clear in
sion. Indeed, the partial matching mentioned previously is all of these studies that a strong pressure for polymorphism
a general characteristic of fusion compatibility in colonial is the rule, but the evidence from all of these fascinating

reactions suggests that such “pressures” select for similar fashion, might be found in the invertebrates. Indeed,
systems via convergence, at least at the level of the recep- IL*-1–like activities have been described for echinoderm
tor. The selecting molecular environment can perhaps show coelomocytes (either IL-1–like production by such cells, or
more conservation (see the folllowing). On the other hand, the ability of the cells to respond to mammalian IL-1), but
as described previously, genetic regions with homology to unfortunately no molecular data revealing the structures of
the vertebrate MHC have been detected in the invertebrates, the active invertebrate cytokine/cytokine receptor have been
yet are (most likely) unrelated to the reactions detailed here reported. In fact, no ortholog has been detected in the ge-
(see conclusion). nome projects from protostomic invertebrates (see Fig. 4.1),
As described, in all of these allorecognition systems, and only IL-17 and TNF homologues have been detected in
convergent mechanisms have been encountered to reach an nonvertebrate deuterostomes and protostomes,48,448,449 and
analogous end, and thus it is unlikely that a unique “cause” IL-8 in agnathans; thus, we may consider these as primor-
arose for selecting them in diverse organisms. However, dial cytokines related to the vertebrate versions. A molecule
in an attempt to find an ultimate and general explanation from earthworms capable of activating the prophenoloxi-
to the selection during evolution of highly polymorphic dase defense pathway cross-reacted with a mAb directed to
allorecognition systems, it has been suggested that pathogen mammalian TNF-α .450 However, this molecule had no ho-
and retroviruses are the force behind the selection of allo- mology to TNF-α upon sequencing.
polymorphism.447 This does not imply intimate phylogenetic IL-1 activity as measured by costimulation assays or as
relationships between the systems observed but rather em- a consequence of PAMP stimulation has been detected in
phasizes analogous solutions when facing similar pressures. all nonmammalian vertebrates. IL-1β upregulation has been
We discuss “connections” between vertebrate immunity and detected after treatment of macrophages with LPS, consis-
these histocompatibility reactions in the conclusion. tent with its inflammatory function in mammals. In addi-
tion, injection of gram-negative bacteria into trout induced
IL-1β expression in many tissues.451 Identity with the mam-
CYTOKINES AND CHEMOKINES malian IL-1β gene in all other species ranges from 28% to
Many cytokines/chemokines and their receptors, like most 40% (identity between mammalian IL-1α and IL-1β is
molecules of the immune system, evolve rapidly. However, about 25%). In nonmammalian species, IL-1β lacks the so-
consistent with the “Big Bang” theory, it is an emerging pic- called ICE cleavage site, important for function in mouse/
ture that the majority of cytokines and chemokines found human.452 IL-18 is an IL-1–related cytokine, and in contrast
in mouse and human are also found in the genome and EST to IL-1 seems more focused in its function of potentiating
projects of nonmammalian jawed vertebrates, best studied TH1 responses. IL-18 has been detected in birds and fish,
in chickens and certain bony fish, but now extending to the but the tissue distribution in fish seems to be expanded as
cartilaginous fish as well. This suggests that whatever the compared to mammals. Additionally, IL-18 in nonmam-
initiating pressures in the evolution of the Ig/TCR adaptive malian vertebrates contains the ICE cleavage site, unlike its
immune system, the network of cytokines and chemokines cousin IL-1β. Other IL-1–related cytokines, including IL-33
emerged (practically) full blown early in evolution. A pic- and the IL-1F series, have also been found in several jawed
ture starts to emerge, but the gestalt is far from clear; this vertebrates, but with little study of functional activity. This
is, in part, because the situation in mouse human is clouded will be of great interest to uncover, considering the IL-1
by the plasticity of T helper cell lineages, as well as an en- family members’ roles in regulating Th1, Th2, and Th17
tirely new subset of immune cells, the innate lymphoid cells (and others?) differentiation.453
(ILCs), that produce cytokines previously thought to be Both chicken IL-1β and the IL-1R were identified and
within the domain of adaptive lymphocytes. have been expressed as recombinant proteins.454 The IL-1R
homology to mammalian orthologs is quite high (61% iden-
tity), but the highest similarity is found in the cytoplasmic
The Proinflammatory Cytokines, Interleukin-1 domains. In addition, there are four blocks of high similarity
Family Members, Interleukin-6, Interleukin-8, and to the cytoplasmic tail of toll/TLR proteins, and IL-1R and
Tumor Necrosis Factor ` TLR use similar signal transduction cascades (see previous
IL-1 and related family members IL-18 and IL-33, IL-6, discussion).
TNF-α , and IL-8 (CXCL8) are the prototypic cytokines as- As mentioned, IL-8 (actually, the CXC chemokine
sociated with inflammatory responses, which are defined by CXCL8) has been identified in the jawless lamprey and
induction of vasodilatation and vascular permeability, and seems to be expressed specifically in B cells, whereas the
upregulation of innate immune system–specific molecules IL-8 receptor is found in T cells; this has been one of the
that have direct functions or that costimulate/attract T and mechanisms proposed to permit the cells to interact in a
B cells. Classically, many of these activities can be assayed in cognate fashion.13,291 IL-8 has also been found in various
supernatants from PAMP (eg, LPS)-stimulated phagocytes gnathostomes such as trout, flounder, and perhaps chicken;
by determining whether thymocytes are induced to prolifer- a chicken CXC chemokine called K60 clusters with IL-8 in
ate when one also adds suboptimal concentrations of T-cell phylogenetic trees and is upregulated in macrophages stimu-
mitogens. It was reasonable to hypothesize that such cyto- lated with LPS, IL-1β, and IFN. Interestingly, Marek disease
kines, which act both at a distance as well as in a cognate virus expresses an IL-8 homologue (v-IL-8), which may be

involved in inducing immune deviation.455 A molecule re- can maintain the growth of T-cell blasts, and the ortholog
lated to the IL-R was detected in the sea urchin genome, but has been detected in bony fish and birds. The chicken IL-2
the ligand has not been found48 ; becuase there are so many protein is only 24% identical to human IL-2 and only 70%
members of the IL-1 family, the search should continue. identical to a near cousin, the turkey.454 IL-15, a relative of
The TNF family in mammals includes the canonical IL-2, has also been cloned in the chicken. A candidate IL-2R
TNF-α , lymphotoxin (LT) α , and LTβ, all encoded at the in chicken was identified by a mAb recognizing a 50-kDa
distal end of the MHC class III region,403 as well as a large molecule only on stimulated T-cells (thus an IL-2Rα ho-
number of other members with diverse immune functions. mologue). This mAb blocks costimulation by IL-2–like
TNF-α is the best studied of these cytokines, and it is one molecules in chicken T-cell supernatants and also reduces
of the key regulators of innate and adaptive immunity. The the capacity of T-cell blasts to absorb IL-2–like activity from
other two cytokines have a more limited tissue distribution supernatants. IL-2 has been studied in the bony fish Fugu. In
and function, and are noted in their roles in lymphoid tissue both chicken and the deduced Fugu IL-2 protein, there is a
development, especially in the formation of splenic white second set of cysteine residues, which are found in IL-15 and
pulp and segregation of T and B cells in lymph nodes. In thus is a primordial feature.461
contrast to IL-1, TNF homologues have been detected in cni- As mentioned, γ C is the signaling subunit of the IL-2,
darians, protostomes (see previous discussion), sea urchin,48 IL-4, IL-7, IL-9, IL-15, and IL-21 receptors; absence of this
Ciona,448 and Amphioxus,456 consistent with the idea that such chain in mammals leads to major defects in lymphocyte de-
a multifunctional cytokine would predate the jawed verte- velopment (“boy in the bubble”). In fish, a γ C homologue
brate adaptive immune system. Homologues have also been was cloned in rainbow trout with unusually high identity
cloned from several teleost species, and TNF-α expression (44% to 46%) to mouse/human genes.462 IL-1β, but not LPS,
in leukocytes is upregulated within 4 hours after treatment upregulated the trout gene in macrophage cultures and a
with LPS, IL-1β, and PMA. While there is good phylogenetic fibroblast cell line. Since then, many other orthologs have
support for orthology of fish TNF-α to that of mammals, been uncovered in fish, often duplicated in various species.460
the other TNF genes seem to be teleost-specific duplicates IL-21, which is encoded adjacent to IL-2 in mammals,
rather than the LT genes.457 Conversely, in Xenopus, the has been found in ectothermic vertebrates.460 Because one of
three TNF family members described previously in mam- its major functions is the differentiation of T-follicular cells
mals are closely linked in the class III region of the MHC.177 and the GC response, it will be of interest to study it role in
This is a bit of a surprising result as all of these family mem- those animals.
bers seem to be lacking in chickens, consistent (according IL-7 is involved in lymphocyte differentiation as well as
to the authors) with a lack of lymph nodes in these animals; homeostasis of mature lymphocytes. Recently, mutations in
however, the unusual nature of the MHC in birds (eg, the Il-7R and downstream signaling molecules in the zebrafish
immunoproteasome genes are missing) rather is consistent have shown a block in T-cell development.463 These mutants
with a loss of TNF genes in these animals, which is indeed will permit future study of lymphocyte production in this
surprising, especially for TNF- α.458 crucial developmental model.
An excellent review detailing the evolution of all TNF
and TNF receptor family members demonstrated that all of
TH2 Cytokines: Interleukin-4, Interleukin-5,
the genes are found in paralogs on four chromosomes, con-
sistent with the 2R hypothesis. This work demonstrated that Interleukin-9, and Interleukin-13
the earliest member was linked to the proto MHC, and thus Reponses to extracellular pathogens, especially fi larial
it is not surprising that TNF is an ancient gene family.459 worms, are largely regulated by TH2 cytokines in mam-
IL-6 and its related members IL-11 and IL-31 are also mals. IL-4 is the most pleiotropic cytokine in this regard,
found in most gnathostomes examined, with few func- stimulating the production of neutralizing antibodies,
tional data. stimulation of eosinophils and mast cells, and an antago-
nism of TH1 responses. IL-4, IL-5, and IL-13 (and granulo-
cyte macrophage colony-stimulating factor) are encoded in
Interleukin-2 and the fC Family of Cytokines the so-called TH2 cytokine complex in mammals, and all
The family of cytokines that signals through the γ C receptor of the genes are coordinately upregulated after stimulation
includes IL-2, IL-4, IL-7, IL-9, IL-15, and IL-21. Similar to of a nearby locus-control region. This same “TH2 complex”
what was described for most molecules involved in adaptive is found in chickens but with a psuedogene for IL-5, but in
immunity, all of these cytokines (except perhaps IL-9) are Xenopus only the IL-4 gene has been identified to date. Fish
present in all gnathostomes studied.460 Costimulation assays have genes in which the ancestor of IL-4 and IL-13 seems to
of thymocytes, as described previously for IL-1, and perpet- be in a preduplicated state, syntenic with genes in the TH2
uation of T-cell lines with stimulated T-cell supernatants are complex.464 Again, duplicates of IL-4/-13 and their receptor
performed to detect IL-2 or “T-cell growth factor” activities. have been found in various fish.460,465
Unlike IL-1, IL-2–like factors generally stimulate cells only At this stage of study, it may be that lower vertebrates do
from the same species, and it is a “cognate” cytokine, meant not have “full-blown” TH2 responses, consistent with the
for release only between closely opposed cells, or as an au- lack of canonical allergic responses in ectothermic verte-
tocrine factor. From teleost fish to mammals, stimulated brates. On the other hand, we must be careful because cyto-
T-cell supernatants costimulate thymocyte proliferation or kine genes evolve rapidly and relying on synteny is not always

dependable, especially in the teleost fish. It will be of interest leost fish, type II IFN seems to have the same function as in
to see whether IL-4 in nonmammalian vertebrates will pro- mammals, suggesting that the TH1-type responses emerged
mote switching to isotypes that are incapable of promoting early in vertebrate evolution.466
inflammatory responses. In addition, further studies of this
family and its receptors in teleost and cartilaginous fish will
reveal whether this facet of the adaptive immune response is Heterodimeric Cytokines
indeed a relative newcomer. The IL-12 family are composed of heterodimers that share
chains (p19, p40, p35, and EB13) and include IL-12, IL-23,
IL-27, and IL-35. IL-12 is generally considered to be a pro-
Interferons inflammatory cytokine produced by APCs that promotes
IFNs, classically known for their antiviral properties, are di- a TH1 response after exposure to intracellular pathogens.
vided into three groups: type I, type II, and type III. Perhaps Consistent with the ancient derivation of TH1 responses, the
surprisingly, this entire group has only been uncovered two subunits of IL-12 p70 (p35 and p40) have been found in
in jawed vertebrates to date. However, as described previ- chickens and several teleost species.454,470 The p40 subunit of
ously, viral immunity in insects is partially regulated via a IL-12 can also associate with p19 to form the cytokine IL23,
JAK/STAT response.152 Types I (α and β) and III (λ) IFN which is involved in the perpetuation of TH17 cells. IL-27
are widely expressed in cells of many types and induces in- generally inhibits all TH subsets but expands regulatory T
hibition of viral replication in neighboring cells, as well as cells and is composed of p28 and EB13; IL-35 is a relatively
molecules of the innate immune system such as inducible new regulatory T–produced cytokine composed of p35 and
nitric oxide synthase (iNOS) and IFN regulatory factor-1.466 EB13. All of these chains have been detected in gnathostomes
In contrast, type II IFN (IFNγ or immune IFN) is synthe- and several studies have been done to examine their expres-
sized by activated T cells, activates macrophages, and up- sion at the RNA level; however, to date no reagents have been
regulates class I, class II, immune proteasome subunits, and prepared to the heterodimers for functional analysis.
TAP, and a large number of other genes.
Antiviral activity is detected in supernatants from virally
infected fish fibroblasts, epithelial cell lines, and leukocytes. Transforming Growth Factor-a and Interleukin-10
All of the biochemical properties of mammalian type I IFN TGF forms a large family with pleiotropic effects in many
(eg, acid-stable, temperature-resistant) are present in these developmental systems. For the immune system, TGF-β
fish supernatants, and the putative IFN reduces viral cyto- isoforms are best known for their capacity to suppress adap-
pathic effects in homologous cell lines infected with virus.466 tive immune responses (even across species barriers), al-
In vivo, passive transfer of serum from virally infected fish though they can also stimulate lymphocytes under certain
protects naïve fish from acute viral pathogenesis. There ap- conditions, especially in mucosal surfaces. TGF-β inhibits
pears to be two lineages of type I IFNs in fish that are specific macrophage activation in trout and growth of T-cell lines
to this group. In chickens there are up to 10 closely related, in Xenopus species. Recombinant Xenopus TGF-β, like the
intronless type I IFN genes.455 Sequence identity to human mammalian form, also can inhibit IL-2–like dependent
type I IFN ranges from 25% to 80%, with the apparent growth of splenic lymphoblasts.471 Four TGF-β isoforms
functional gene having highest similarity. Interestingly, bats were isolated from chickens, as opposed to three major
have greatly expanded their type I and III IFN genes and forms in mammals.455 The three major forms of the cytokine
receptors (as well as many other genes associated with viral have been isolated in several teleost species.470
immunity).467 These animals are highly infected with (and IL-10 is often considered along with TGF-β because it is
susceptible to) virus, and some IFN forms are constitutively mostly an immunosuppressive cytokine with multiple ef-
expressed at low levels. fects; both cytokines are expressed in subsets of “regulatory
While type I/III IFNs can be highly duplicated in some T” cells in mammals. This cytokine was originally discov-
vertebrates, this is not true of IFN receptors, and in some ered by its ability to suppress TH1 responses, but now is
cases many cytokines will used two to four receptors within known to have a much expanded role in immunity. IL-10 has
a species. How can such a system work and result in different been found in chickens and a large number of teleosts.455,470
readouts? It appears that even single amino acid changes in As usual, functional experiments have lagged behind the
the cytokine can induce differential conformational changes molecular work, but recombinant chicken IL-10 can block
in the receptor that can modify downstream signaling.468 IFNγ production by splenocytes. IL-10 is a class II cytokine,
This paradigm will be exciting to follow in a phylogenetic related to IFNγ, IL-19, IL-20, IL-24, and IL-26. IL-22, a cy-
context. tokine released by TH17 cells and ILCs in mammals, is up-
Type II or IFNγ has been cloned in most jawed verte- regulated in fish vaccinated with pathogens and is correlated
brates. The chicken gene is 35% identical to human type II with immune protection.472
IFN and only 15% identical to chicken type I IFN.454,469 Note that there have been very few “regulatory T” experi-
Recombinant chicken IFN stimulates nitric oxide produc- ments reported in the comparative literature, but many older
tion and class II expression by macrophages. The gene has experiments that suggest that such cells exist. Over 25 years
been cloned in many fish, and has been studied mostly in ago, experiments in Xenopus showed that graft rejection
the trout where it has been shown to upregulate CXCL10 and could be delayed when lymphocytes from metamorphosing
activate protein kinase C. Thus, at least in chickens and te- animals were adoptively transferred into adult frogs.473 This

result suggested that a “wave” of suppressor cells emerged genes in ectotherms because of some “missing pieces” in the
near the time of metamorphosis that evolved to protect databases (or our ability to find some of the genes)? Some
animals from autoimmunity when adult-specific molecules genes may have been lost, as described previously for the
were expressed. It is time to reexamine such experiments chicken—again, further studies of amphibians and cartilag-
with modern tools. In a recent experiment in zebrafish, im- inous fish should help in our understanding. In mammals,
munization of animals with central nervous system (CNS) the LT TNF family members are important in the develop-
antigens induced an autoimmune reaction. Overproduction ment of lymph nodes and splenic white pulp, and yet the
of foxp3 inhibited the production of IL-17 and IFNγ in these three genes are present in lymph node–less Xenopus —what
animals, ameliorating the disease.474 are their functions in ectotherms? If we can uncover these
other roles in nonmammalian vertebrates, it could be quite
informative to reinvestigate such functions in mammals.
Interleukin-17 Finally, chickens and fish have evolved their own paralogs of
TH17 cells in mammals are important for inflammatory some of the well-known cytokine genes; their study should
responses to extracellular bacteria via indirect stimulation reveal selection pressures on particular species that could
of neutrophils. Despite its late appearance in the immu- also be quite informative in our understanding of the gestalt
nologic literature, IL-17 appears to be one of the most an- of the cytokine network.
cient cytokines, with homologues in protostomes and lower
deuterostomes, and extensively amplified in sea urchins.
Even the different isoforms seem to be conserved among General Evolution of Chemokines and Their Receptors
the jawed vertebrates, which should initiate many new av- Chemokines and chemokine receptors are essential for many
enues of study.470 The family is composed of IL-17A/F, with aspects of the immune system, including inflammation, the
the A and F forms being the best studied to date. IL-17A/F differentiation of lymphoid tissues, trafficking of hemato-
genes have been found in all mammals studied, as well as poietic cells during ontogeny and immune responses, and
IL-17C/E. The most ancient IL-17 form is IL-17D, present even stimulation of cells under various conditions.476 With
in both oysters and agnathans.291,449 This form is expressed the advent of the genome and EST projects, a seminal paper
in lamprey T cells but not B cells, and has been proposed as showed that, consistent with the Big Bang theory of adap-
a helper factor for cognate interactions.13 Finally, like many tive immune system evolution, all of the major classes of
other immune genes, as many as 20 IL-17 genes have been chemokines and their receptors were present in the bony
found in sea urchins, again suggesting expanded innate im- fish lineage, and probably will be found in cartilaginous
munity in this species.475 fish as well when the genomes are completed.477 Recently,
the analysis was updated to include all of the vertebrates
in the databases.476 Similar to most of the immune-related
Cytokine Summary genes discussed throughout the chapter, bony fish have more
The isolation of nonmammalian cytokines and cytokine re- chemokines/receptors than any other vertebrate, including
ceptor genes has lagged behind molecular characterization amphibians and mammals.
of antigen receptors and MHC. However, with the advent Interestingly, no CC (so-called homeostatic) or CXC (so-
of the genome and EST projects, we are rapidly acquiring a called inflammatory) chemokines/receptors were detected
comparative view of this field, at least at the genetic and mo- in lower deuterostomes like Ciona and sea urchin. The che-
lecular level. Teleost fish and chickens have paved the way mokine receptors are member so the G protein–coupled
in this field, but cartilaginous fish and agnathan databases receptor family, and of course such molecules are found in
will soon be complete and provide the big picture. Already, all animals. However, no members of the specialized family
it seems that cytokines like TNF and IL-17 seem to be most (G protein–coupled receptor γ) to which chemokine recep-
primordial, with homologues in Ciona, Amphioxus, and tors belong were found in lower deuterostomes. The che-
sea urchins, as well as deeper homologues in certain pro- mokine receptor genes are found on four chromosomes in
tostomes. Conversely, none of the other so-called adaptive mammals, four of which contain the HOX genes, one of the
cytokines (or even type I/III IFNs) seem to be present in the gene families that provided evidence for the 2R hypothesis
lower deuterostomes, consistent with the Big Bang theory of early in the reawakening of this theory. The distribution of
adaptive immunity.1,2 If it is true that the jawless fish lack chemokine receptor genes on these chromosomes correlates
these genes as well, despite their convergent adaptive im- very well with 2R, and helps to account for the large-scale
mune system, one can point to the evolution of lymphoid gene expansion of this family in the jawed vertebrates. Thus
organs and the segregation of lymphocyte subsets into dis- far, very few chemokines/receptors (eg, IL-8 [CXCL8] and
crete areas to help to explain the explosion and recruitment CXCR4) have been detected in agnathans, but as mentioned
of these genes. We discuss this concept in more detail in the previously, agnathan T cells express IL-8R and B cells IL-8,
conclusion section, especially regarding the relationship of believed to be a factor in promoting the cognate interactions
adaptive lymphocytes with ILCs. between the cells.291 CXCL12, a chemokine vital for thymic
Despite our “big picture” knowledge of the emergence differentiation, is also present in lamprey, consistent with
of cytokines, obviously functional experiments have lagged new work on the agnathan thymus discussed previously.375
behind. Furthermore, is it true that TH2 responses are evo- In summary, it appears that a few chemokines/receptors
lutionary latecomers, or are our early attempts at finding the arose at the dawn of vertebrate adaptive immunity, and they

were greatly expanded by both en bloc and cis duplications class II do bind peptides, it would suggest class I arose first.
and became “full-blown” shortly after the 2R Big Bang.476 Again, genome scans of jawless fish and lower deuterostomes
Currently, this is our most complete story of the evolution should be informative on this point, but to date no luck! As
of a family of immune genes in deuterostomes. described previously, we should be diligent because with the
discovery of T cells and a thymus-like structure in lamprey,
one expects some sort of “MHC” in the agnathans291,375 ;
ORIGINS OF ADAPTIVE IMMUNITY to date, the only homologue we have uncovered to date is
The immune system of vertebrates is unique because the TAP-L, which may hold the key to antigen presentation in
antigen-specific receptor expressed by lymphocytes, which this group.401
initiates cascades leading to activation of the adaptive im- From the paralog data, genes encoding the complement
mune system, is not the product of a complete germline- components C3 and Bf, TNF superfamily members, the
inherited gene. Rather, receptors are generated somatically signaling molecule Vav, B7 family members, and protea-
during lymphocyte ontogeny from gene segments scattered some subunits among other genes (such as tripartite motif-
at a particular locus. As described previously, the recep- containing) should have been present in the proto-MHC,
tors in gnathostomes are IgSF members composed of V before emergence of the adaptive immune system (see
and C domains, with the C domains being of the rare “C1” Fig. 4.13). Some of these genes were found in the Amphioxus
type, which is shared by MHC class II and class I molecules. and Ciona “MHC” linkage groups, and C3 and Bf genes
There are many specific questions: 1) Did MHC class I or are linked in the sea urchin. A fifth paralogous region on
class II come first?; 2) What is the origin of the MHC PBR?; human chr 12p13 contains the α2-macroglobulin gene (re-
3) Was the MHC involved with innate immunity before call the C3/4/5 homologue), a tapasin homologue, the C3a-
the emergence of adaptive immunity?; 4) Did somatic rear- receptor, and this “complex” is linked to the NKC. Taken
rangement or somatic mutation come first to diversify an- together, the data suggest that the proto-MHC included vital
tigen receptors?; 5) Which of the extant antigen receptors nonhomologous genes of the innate immune system, which
in gnathostomes, α / β TCR, γ /δ TCR, or IgH/L (if any) re- perhaps were linked to allow coordinate regulation of ex-
sembles the primordial receptor?; and 6) How did two dif- pression. After the en bloc duplications, Pontarotti et al. has
ferent antigen receptor gene families emerge in evolution, suggested that “functional restraints upon the complex were
after the appearance of separate lymphocyte populations? relaxed” and hence the duplicated members could evolve
The answers to these questions are speculative, but deduc- new functions, including features indispensable to the adap-
tions can be made based upon the wealth of molecular and tive system.405 If indeed innate immunity genes were already
emerging functional data. We base many of our arguments linked to allow upregulation at times of infection, it is no
on the large-scale duplications that were revealed for MHC surprise that the adaptive immune system piggybacked on
by Kasahara17 in 1996. The remarkable syntenies of pauci- such a gene complex. A final point: why did the duplicate
copy genes on the paralogous regions, and the recent fi nding genes survive rather well over hundreds of millions of years;
that an animal that predated the duplications (Amphioxus cis duplicates have been shown to degenerate rapidly over
and Ciona) has only single copy genes in the same syntenic evolutionary time. Evidence suggests that duplicates aris-
group orthologous to the four mammalian copies, make it ing from polyploidy (and by inference large en bloc duplica-
clear the en bloc duplications were involved.405,421 Further tions) survive better than cis duplicates, most likely because
analysis of this region in model lower deuterostomes, as de- they cannot be inactivated by unequal crossovers; the ability
scribed in the histocompatibility section, will continue to of the genes to survive over very long periods, perhaps
be vital in our understanding of early immunity. Genetic combined with strong selection pressures, would allow for
analyses in protostome lineages have not been very infor- subspecialization.479
mative, but we must reevaluate once we piece together data
from more deuterostome species.
Origins of Rearranging Receptors
The Rearranging Machinery
Major Histocompatibility Complex Origins Most models propose that the generation of somatically re-
Class I and class II molecules have been found only on the arranging receptors occurred abruptly in evolution via the
jawed vertebrates—so far, painstaking screenings of the generation of the RAG machinery made of two lymphocyte-
lower deuterostome or lamprey/hagfish databases have specific proteins, RAG1 and RAG2. RAG genes have so far
yielded no indication of these proteins. Based upon phy- been isolated in all classes of jawed vertebrates and have
logenetic analyses and thermodynamic arguments, most been quite conserved. In every case examined, RAG1 and
investigators believe that class II preceded class I in evolu- RAG2 genes are closely linked and in opposite transcrip-
tion.478 However, as stated previously, class I is much more tional orientation. Some regions of RAG1 and RAG2 are
plastic than class II as there are many different types of class similar to bacterial recombinases or to molecules involved
I molecules, some that do not even bind to peptides. Because in DNA repair (eg, RAD16) or the regulation of gene expres-
class I genes may be on two or three MHC paralogs, and sion (such as rpt-1r). Similarities to prokaryotic proteins
they can have functions outside the immune system; this and the gene structure suggest that vertebrates acquired the
is evidence that the primordial “PBR” may not have even RAG machinery by horizontal transfer and transposition
bound to anything. If this is true, and because class I and from bacteria.19,480 Indeed, RAG genetic organization has

some transposon characteristics: the RSS are reminiscent the finding of the RAG genes in sea urchins challenges this
of sequences involved in targeting excision of transposons. notion at some level. However, it cannot be overemphasized
A class of transposons detected in several protostome and that SHM is also at the origins of the immune system; fur-
deuterostome invertebrate species that shows similarity to thermore, all evidence to date suggests a gradual evolution of
the catalytic domain of RAG1.481 It is believed that RSS were the SHM machinery (the AID/APOBEC family and associ-
derived from the terminal inverted repeats of this transpo- ated polymerases/mismatch repair proteins) rather than the
son, called transib. In invertebrate genomes there are many “hopeful monster” generated by the famous RAG transpo-
of such “RAG1 core regions,” but only one has an open read- son. Thus, diversity generated via SHM or gene conversion
ing frame throughout the core and shows similarity to ver- may have existed in an adaptive immune system prior to re-
tebrate RAG1 in other regions as well. Furthermore, a RAG2 arrangement, and V gene rearrangement was superimposed
homologue is adjacent to the urchin RAG1 gene, in a simi- onto this already existing system.296,300 Indeed, the presence
lar orientation as is found in gnathostomes.352 This finding of APOBEC family members in the jawless fish (CDA1 and
was a big surprise and suggests that both RAG genes were in CDA2, as described previously) suggest that diversity gener-
place approximaltey 100 million years before the origin of ated via mutation/conversion preceded rearrangement. The
the Ig/TCR system. Their tissue distribution and expression RAG-induced rearrangement break and subsequent repair
during ontogeny are not known; nor have any candidate provided something new and novel in gnathostomes, not
genes been recognized to date with RSS that might be rec- only diversity in sequence but heterogeneity in size300 ; this
ognized by the echinoderm homologues. It must be admit- was a remarkable innovation and likely indeed heralded the
ted that it is unclear how this new result fits into the puzzle sophistication of jawed vertebrate adaptive immunity.
of the origins of adaptive immunity. Because transib is so
often found in the animal kingdom, it is certainly part of Which Antigen Receptor First? Phylogenetic analyses have
the transposon. By contrast, RAG2 is now believed to have suggested that among IgSF receptors γ /δ TCR-like ancestor
been present in the genome, and recent data have shown it may have predated α /β TCR and Ig H/L.485 This would sug-
to interact with active chromatin.482,483 Thus, the new idea gest that direct antigen recognition, perhaps by a cell surface
is that RAG2 recruits RAG1 to open chromatin, and rag1 is receptor, arose first in evolution followed by a secreted mol-
the active enzyme in the major rearrangement events, which ecule and an MHC-restricted one. Hood and colleagues argue
is wholly consistent with the transposon hypothesis with that phylogenetic analyses over such large evolutionary dis-
RAG1 (ie, transib) being the genome “invader.” tances obscure true relationships among the antigen receptor
Another source of somatic antigen receptor diversity genes (eg, the relationships of the molecules in the phyloge-
shared by all gnathostomes characterized to date is a unique netic trees has to impose multiple loss/gain of D segments in
DNA polymerase, TdT, which diversifies CDR3 during Ig and the different antigen receptor families) and suggest a model
TCR gene rearrangement through the addition of nucleotides based upon genomic organization, not so different from the
in a template-independent fashion.484 Furthermore, as de- Kasahara model.263 They propose an alternative phylogeny in
tailed previously, its expression serves as an unambiguous which an ancestral chromosomal region with linked genes en-
developmental marker for the sites of lymphopoiesis. TdT has coding both chains of an ancestral antigen receptor heterodi-
been highly conserved in both sequence (> 70% amino acid mer, one having D segments the other not (eg, IgH/IgL). The
similarity, > 50 amino acid identity) and overall structure α and δ TCR loci are still closely linked in all vertebrates ana-
during vertebrate evolution. An amino acid alignment of all lyzed (human chromosome 14), and a pericentric inversion
known TdT sequences reveals that some, but not all, struc- is suggested to have separated the TCR β and γ loci (linked
tural motifs believed to be critical for TdT activity are par- on human chromosome 7). This model predicts that D seg-
ticularly well conserved in all vertebrates studied. TdT protein ments only emerged once, and also explains the existence of
alignments, and the crystal structure for rat β-polymerase, inverted V elements in both the TCR β and δ loci. This model
support the hypothesis that both evolved from a common does not predict which antigen receptor is oldest, but does
ancestral DNA repair gene. In addition, four protein kinase provide a “simple view” of receptor evolution, consistent with
C phosphorylation sites are conserved, and hence may be in- the Kasahara/Ohno model. Recent data have added to this
volved in TdT regulation. Homologues related to the ances- scenario. First, as mentioned previously, VH elements have
tor of polymerase β and TdT have been found in sea urchin been found at the TCR-δ locus in many vertebrates, includ-
and other lower deuterostomes. Thus, unlike RAG, TdT has ing sharks, amphibians, birds, and marsupial mammals.269 In
evolved by gradual evolution from a polymerase family and part, this is due to the close proximity of the IgH/L (lambda)
was recruited for immune system function. and TCR alpha/delta loci in many living vertebrates (especially
Xenopus268), consistent with the Hood hypothesis of ancient
Rearrangement or Somatic Hypermutation First? en bloc duplications of antigen receptor loci.268 Additionally,
Because all antigen receptor genes use somatic rearrange- the idea that most TCR-δ receptors do not recognize MHC-
ment of V genes to generate diversity in CDR3 regions as well restricted antigen suggests that they can continue to “borrow”
as to promote combinatorial diversity, there is no doubt that elements from Ig loci over evolutionary time.
this mechanism is at the heart of adaptive immunity. Indeed, Thus, an alternative scenario is that MHC-restricted anti-
as described, most investigators believe that the introduction gen recognition arose first, perhaps derived from a NKR with
of the tranposable element into a V gene was the driving force a “VJ” IgSF domain that recognized SOS proteins, like the
in the abrupt appearance of vertebrate adaptive immunity; ULBPs or MIC today. The rearrangement break induced by

RAG occurs in CDR3, which is in the center of the antigen see previous discussion). Were Ig and TCR “invented” in a
receptor–combining site, in perfect position to interact with class of vertebrates now extinct (eg, the placoderms, which
peptide bound to MHC, which could have been another inno- are more primitive than cartilaginous fish but more ad-
vation to detect foreign antigens within cells.486 Thus, such cells vanced than agnathans)? The discovery of the three tapasin
would be most like extant αβ T cells with an MHC-restricted paralogs all with C1 domains suggests an origin prior to the
antigen receptor encoded by linked genes. Once rearrange- full establishment of the vertebrate genome.33
ment was introduced, the combining site may have been V domains, either alone (eg, IgNAR) or in association with
“relaxed” so that it was no longer forced to be MHC-restricted another V domain (eg, Ig H/L), recognize the antigenic epitope
(see the section on TCR above). The genomewide duplication and are therefore the most important elements for recognition.
then would provide two types of TCR, one MHC-restricted For this reason, they will be the first to be traced back in meta-
and the other not, like γδ TCR today. One locus, according to zoan evolution by asking whether V domains exist in inverte-
the Hood model, then underwent an en bloc cis duplication to brates. Domains with the typical V fold, whether belonging to
give rise to the IgH/L loci. Ig, then, developed a new differen- the true V-set or the I-set, have been found from sponges to in-
tiation pathway with an alternative splicing mechanism for the sects (although not necessarily involved in immune reactions;
transmembrane and secreted forms of BCRs. the first ones were discovered by nonimmunologists among
For the origin of the rearranging receptors, IgSF lineages molecules involved in nervous system differentiation in inver-
have to be traced back through phylogeny as such receptors tebrates [eg, amalgam, lachesin, and fascicilin]). Invertebrates
generated by somatic rearrangement do not exist outside the also use IgSF members in immunity, but so far they are not
jawed vertebrates. In a quest for molecules related to elusive V domains, but more I- or C2-set (eg, mollusk defense mol-
ancestors, without focusing on genes expressed in the im- ecule, hemolin, DSCAM). However, the mollusk FREPs have
mune systems of various phyla (ie, structure is more im- a V domain at their distal end, associated with a fibrinogen-
portant than function in this case), the most homologous like domain. As described, they are involved in antiparasitic
sequences and gene architectures in the various metazoan reactions and form a multigenic family with polymorphism.135
phyla must be scrutinized. Similarly, as mentioned previously, prochordate VCBP has a
VJ domain, but it mediates innate immunity.143
V(J) and C1 Domains Besides searching for VC1-encoding genes in nonverte-
C1 domains are found in the antigen receptors, MHC class brates, surveying the human genome for such genes has been
I and class II, and very few other molecules (see Fig. 4.10). fruitful. Indeed, nonrearranging V-containing molecules, ei-
This IgSF domain is so far most prevalent in gnathostomes, ther VJ alone, VC1, or C1 alone, have been found in the human
as if C1 domains arose concurrently with the adaptive im- genome. Interestingly, many of them are present in the MHC
mune system and coevolved with it. What was the value of class III region (human chromosome 6p21) or its paralogs (see
the C1 domain, and why is it found almost exclusively in Fig. 4.12 and 4.13). Two MHC-linked gene segments stand out:
adaptive immune system–related molecules? All of these a single VJ, NKP30, and a gene containing a VJC1 core, tapa-
molecules interact with coreceptors such as CD3 (TCR), Igα sin (TAP-binding protein), involved in antigen processing.
and β chains (Ig on B cells), CD4, and CD8 (with MHC on NKP30, made of a single Ig domain of the VJ type, is an NK
opposing cells), and it is conceivable that in sections of IgSF cell–activating receptor, and it may offer a link to cell types
domain in which C1 differs from the C2 there is a specific in invertebrates. It could be a relative of an ancient receptor
region favoring interaction with other molecules. whose history is linked to the emergence of MHC class II and
The G strand of Ig/TCR V domains is encoded by the class I; unlike most NKRs, NKp30 is evolutionarily conserved,
J gene segment, separated from the V region-encoded A-F present from sharks to mammals.176 In order to resemble an
strands, and rearrangement is necessary in order to as- ancestor, the NKP30 V domain need only be associated with
semble a complete V gene. The primary structure of each a C1 domain. In fact, a C1 single-domain gene, pre–TCR-α, is
Ig/TCR chain bears hallmarks of the dimeric nature of the also encoded in the MHC.253 Besides Ig and TCR, tapasin is one
receptor in which they participate. A diglycine bulge (Gly- of the rare cases, if not the only other case, of a gene segment
X-Gly), present in all V domains, is thought either to be a with a VJC1 structure existing on several paralogous linkage
beneficial adaptation, or to promote dimer formation by groups (6, 9q33, 19q13). In other words, while this gene is re-
inducing a twist in the G strand that results in V domain lated to the rearranging receptor structure, it is undoubtedly
pairing that appropriately orients the CDR. Monitoring this very old, and probably predated the ancient block (genome-
feature, therefore, might reveal genes that had the ability to wide) duplications. It could have acted as a donor of C1 to a
form dimers similar to that of modern antigen-specific re- V domain–containing gene in the MHC class III region (like
ceptors. In V genes that do not somatically rearrange, the the XMIV), which then could have been the first substrate of
G stand is an integral part of the V exon. In other remotely the rearrangement. Another set of molecules with distal VJC1
related IgSF genes, introns have invaded the V domain exon segments, the signal-regulatory proteins (VJ C1 C1), and the
creating a variety of V gene families. Many examples of such poliovirus receptor (VJ C1 C2) could represent another group
events can be found in the history of the Ig superfamily, for linked to the history of the Ig and TCR (see the following).
example in the genes encoding CD8 and CTX.33,487 TREM1 and TREM2, receptors on monocytes/neutro-
As described, no Ig/TCR genes have been isolated from phils involved in inflammatory responses, are composed of
hagfish or lampreys, although there are some tantalizing single VJ domains whose genes are MHC-linked. Myelin oli-
molecules potentially related to their ancestors (eg, APAR, godendrocyte protein (MOG) and P0, two single V domains

involved in the synthesis of myelin sheath, are encoded in have antigen receptors derived from different gene families,
the MHC paralogous region on chromosome 1. Chicken BG, at first glance it seemed that the T-B split occurred twice in
which is related to MOG but probably has a different func- evolution. However, the gene expression profi les for the lam-
tion, is encoded in the chicken MHC. Butyrophilin, CD83, prey T and B cells are too similar to the patterns in gnathos-
and tapasin all have VJ domains, and butyrophilin also has tome lymphocyte populations to be derived by convergent
a C1-type domain. More distant relatives with VJC2-based evolution; furthermore, the development of lamprey T cells
architectures are also found in MHC (receptor for advanced in a thymus-like structure also suggests evolution from a
glycosylation end products, CTX, lectin-related genes; see common ancestor. Thus, it seems most likely that both the
Fig. 4.12), and some of these genes related to the rearrang- LRR and IgSF receptor families were present in the com-
ing receptor ancestors are found on several paralogs (whether mon ancestor,2,13 with the VLR system predominating early
the MHC paralogs or other), suggesting that the V-C1 core in vertebrate evolution (see Fig. 4.10). Such a scenario exists
was generated early in vertebrate evolution subsequent to today with NKRs in mammals, with different gene families
the emergence of the chordate superphylum (see Fig. 4.13). (C-type lectin or IgSF) predominating in different species.12
Among all these molecules, butyrophilin is perhaps not on Because APOBEC family members are most likely in-
the direct track to antigen-specific receptors. Its C domain, al- volved in VLR assembly and perhaps in mutation, there may
though proven to be C1 through its crystal structure, is more have been a primordial IgSF receptor that was also modified
like a C2 at the primary sequence level, and belongs to the somatically by an AID-like molecule, which functioned side-
CD80/86 family, rather than the TCR. Finally, tracing the VJ by-side with VLRs. The rearrangement break caused by the
NITR gene family previously described in evolution may lead RAG transposon in a primordial IgSF gene offered a great
us to an understanding of the original NKC/LRC/MHC, as advantage, as was detailed previously, and the IgSF receptors
well as identifying more candidate genes related to the ances- then could have supplanted VLRs in gnathostomes, while
tral gene invaded by the RAG transposon. maintaining the AID-dependent machinery to diversify it
Many invertebrate molecules not involved in immune further. It is difficult to imagine how a rearrangement break
responses are present as a distal V domain associated with in an antigen receptor gene generated by LRR insertion could
one or more C2-type domains. In the vertebrates many be advantageous; conversely, as mentioned previously, a break
molecules have retained this feature such as CD2 and CTX. in the CDR3 loop in the center of the binding site is in per-
Some members resemble “primitive” antigen receptors and fect position to interact with peptide bound to MHC. While
several of them map, for instance in humans, to at least two lampreys have a thymoid for development of their “T cells,”
tetrads of paralogous genes, the MHC on chromosomes 1, as described previously, there was a great leap in complexity
6. 9 (12), 19 and the LRC on chromosomes 1, 3, 11, 19 (21) of thymus structure in the gnathostomes, consistent with a
(see Fig. 4.13).160,442 Many form dimers and are expressed in new TCR/MHC system in which positive and negative selec-
lymphocytes, where they form a family of adhesion molecules. tion became a requirement for a functioning adaptive im-
The crystal structure analysis of a CTX-related molecule mune system. In addition, the general transcriptional and
(JAM) revealed a unique form of dimerization, suggesting cytokine/chemokine networks for T-cell (and B-cell) devel-
that the diversity of ligand binding and domain-interactions opment would have also been in place in the jawless verte-
used by different IgSF domain is extensive. Two JAM mole- brates; in gnathostomes, this network became more complex
cules form a U-shaped dimer with highly complementary in- in the Big Bang (eg, see the chemokine section in which the
teractions between the N-terminal domains. Two salt bridges leap in sophistication can be examined with precision).
are formed in a complementary manner by a novel dimer- In summary, a working hypothesis proposes that the di-
ization motif, R (V, I, L) E. The receptor for advanced gly- vergence of lymphocytes into T and B cells occurred early in
cosylation end products gene has a rather “generic” receptor vertebrate evolution (see Fig. 4.10). LRR- and IgSF-based an-
function as it recognizes aged cells exposing particular carbo- tigen receptors existed simultaneously as well at this time (in
hydrate motifs. CD47, another conserved IgSF member with a the same organism or organisms in the same epoch), with
CTX-like V domain, suggests an ancient function.488 Perhaps VLR perhaps predominating in early vertebrates. The RAG
such nonrearranging VJC1 genes that regulated cytotoxic- transposon insertion into an IgSF receptor (perhaps one al-
ity/phagocytosis were predecessors of the antigen receptors. ready diversifying by mutation or conversion), along with
Some molecules with VJ/C1 cores involved in cell-cell inter- a genomewide duplication, resulted in the Ig/TCR/MHC-
actions often serve as ports of entry for viruses. In a move based system of gnathostomes, with all of its accompanying
from “property to function,” an arms race consequence could sophistication in all aspects of adaptive immunity, including
result in a virus receptor developing into an immune recep- development of the thymus and spleen, complexity of cyto-
tor. The best examples of such molecules are in the CTX JAM kine and chemokine networks, and intimate interactions of
family, in which receptor interactions with viruses may trig- cells with primary and secondary lymphoid tissues.
ger an apoptosis, a primitive form of antiviral immunity.33,489
CONCLUSION
Emergence of T and B Cells Since the last edition of Fundamental Immunology was pub-
As described previously, agnathans have now been shown to lished 5 years ago, there have been several astonishing find-
have cells in the T and B lineages, similar to what is found in ings, as well as other interesting discoveries and integrations,
gnathostomes.291 Because the jawless and jawed vertebrates regarding the evolution of the structure and function of the

immune system. Furthermore, there have been many genome by infi ltrating hematopoietic cells; later, this was superim-
and EST projects that have allowed us to determine whether posed onto adaptive lymphocytes in the vertebrate lineage.
or not particular genes or gene families are present in differ- Consistent with this idea, sea urchins have an expanded
ent taxa. Here, we briefly describe these findings and discuss family of IL-17 genes,48 suggestive of a very important func-
their broad relevance and relationship to future studies. tion for this cytokine. So, such a system may have been the
A new system of defense in bacteria and archaebacte- forerunner of both the ILCs and adaptive lymphocytes, per-
ria against bacteriophage, CRISPR, has been elucidated.490 haps found in tissues directly exposed to the environment.
Short phage DNA sequences have been acquired by prokary- Additionally, we are poised for more seminal studies of
otic genomes to direct sequence-specific protection against mucosal immunity. Mammals have dedicated secondary
phages. These elements are transcribed into small RNAs lymphoid tissues in their mucosa, and thus it is often dif-
that recruit endonucleases to cleave the phage nucleic acid. ficult to tease apart the innate from adaptive responses in
This system is widespread and its study is in its infancy. these tissues, except in contrived knockout or transgenic
The discovery of this system fits well with studies of plants models. By contrast, ectothermic vertebrates have no Peyer
and animals over the last decade uncovering the plethora of patches or mesenteric lymph nodes, and thus are excellent
mechanisms used to detect foreign nucleic acids (eg, TLR, models to examine responses in the lamina propria, which
NLR, RIG, APOBEC, etc.), as described previously. It should are predominantly T-independent.493
be clear to all immunologists by now that all living things Lastly, studies of how the microbiota interact with the
must distinguish self from nonself, and have developed in- immune system have fi lled journals to the gills over the
tricate mechanisms to achieve this end. past 5 years. McFall-Ngai suggested that the driving force
Cytokines such as IFNγ, IL-2, and IL-4, and other adaptive for adaptive immunity was not so much to fight off invad-
cytokines (and type I/III IFNs) and their receptors are not ers, but rather to tolerate commensals,494 perhaps not so
present outside of jawed vertebrates, nor is the vast array of different from Weaver’s proposal.492 This framework of in-
chemokines seen in gnathostomes. However, at the moment it teractions between leukocytes, and between leukocytes and
appears that (almost) all of the cytokines/chemokines discov- commensals/pathogens, may be the key to understanding
ered in mammals are also present in basal jawed vertebrates: the genesis of adaptive immunity.
Big Bang indeed! IL-17, a cytokine garnering great attention A new defense system in Drosophila, important for viral
in basic immunology, and TNF, a central cytokine in both defense, has been uncovered. Previously, the recognition of
innate and adaptive immunity, seem to be the first cytokines pathogens by TLRs in the invertebrates was believed to be
to have emerged in evolution based on studies of lower deu- indirect, interacting with self-molecules similar to a cyto-
terostomes. If, as suggested from studies in mammals, IL-17 kine interaction. However, Drosophila Tol7 has now been
is important for responses to extracellular bacteria, nonverte- shown to interact directly with viruses, similar to TLRs in
brate deuterostomes may hold the key to understanding the vertebrates,153 but, as expected, the effector phase is differ-
physiology of the entire system (ie, in the absence of the com- ent from the canonical (historical) toll pathway. There are
peting TH1, TH2, and regulatory T cells), and one may be several other tolls in the protostomes without a described
able to study the dynamics of IL-17 development and function function that can be studied in a new light.
in lower deuterostomes and even protostomes. One of the great triumphs in the field of comparative im-
Additionally, over the past few years a new type of cell, munology was the discovery that similar mechanisms are
the ILC, has come to prominence.491 These cells are pre- used to recognize extracellular pathogens and initiate im-
dominantly found in epithelia and mucosa and are involved mune responses. It has been approximately 15 years since
in homeostasis and a first line of defense. As they do not the Drosophila toll/IMD systems and their relationship to
bear antigen receptors, ILCs respond via cytokine recep- vertebrate TLRs and TNF receptors has been uncovered.
tors or PRRs. ILCs express cytokines consistent with T-cell Over the past few years, the intracellular defense mechanism
subsets, and it is tempting to propose that they preceded such as NOD/NLR, RIG, helicase, etc., have been elucidated
adaptive T cells in evolution, and the cytokine-production in vertebrates, and we are beginning to piece together their
profi les were coopted by the antigen receptor–bearing cells. evolutionary histories. The amazing similarities between the
However, as mentioned, it appears that “adaptive” cytokines deuterostome and plant NOD/NLR systems will challenge
arose late in evolution, coincident with adaptive immunity. us to understand their recognition events for years to come.14
Thus, it seems more like that ILCs arose simultaneously The VLR system continues to amaze us.13 Most compara-
with the adaptive cells in the immunologic Big Bang, and tive immunologists never expected a unique antigen receptor
there has been adaptation over time. However, as IL-17 and system to be uncovered in jawless fish. To then realize that
TGF-β are old cytokines, it has been proposed that their op- the divergence of lymphocytes into two lineages occurred in
posing functions arose before adaptive immunity.492 TGF-β the ancestor of jawed and jawless fish was astonishing. Now
belongs to an ancient family of cytokines, and at least one to have uncovered a primitive type of thymus candidate in
of its functions in mammals is to antagonize proinflam- jawless fish is a total shock. All of these discoveries force us
matory responses (regulatory T differentiation, regulation to acknowledge that we should not be surprised by future
of IgA synthesis, etc.). One model suggests that sophisti- discoveries in the evolution of immunity, such as a “conver-
cated innate immunity arose in mucosal tissues, regulated gent MHC,” or an antigen presentation system not based on
by the pro- and anti-inflammatory properties of IL-17 and peptide presentation, or even a primordial type of adaptive
TGF-β, respectively, by their expression in gut epithelia or immune system in lower deuterostomes or protostomes.

Furthermore, from comparative studies, Boehm and multiple chromosomes in the vertebrates; good examples of
colleagues have proposed a simple model for the most fun- how it has aided in our understanding of immunity were
damental elements required for thymocyte differentiation, the study of IgSF members involved in cell-cell interaction/
which has been superimposed onto the lamprey thymus-like costimulation, and the evolution/emergence of chemokines
structures.375,495 Further study of this minimal network of and TNF family members. An emerging paradigm is the
transcription factors, chemokines, and cytokines could un- possibility that genes encoding NKC, LRC, MHC, as well
cover the origins of thymus differentiation in lower deutero- as the antigen receptors and costimulatory molecules (and
stomes that clearly have no T/B cells. other immune genes) were all linked at an early point in evo-
Certain groups of organisms are interesting to study for lution.2,498 In addition, while the agnathan adaptive system
different reasons. The genome projects in cnidarians dem- is similar to the Ig/TCR/MHC system of gnathostomes, the
onstrated that certain gene families were more ancient than great sophistication that we see in the latter is not present in
previously believed (from studies of the classical models lampreys and hagfish. Our best guess at the moment is that
Drosophila and C. elegans).35 Thus, this taxonomic group the RAG transposon, in combination with the genomewide
has become the new rock stars of comparative immunology, duplications, both superimposed onto an already existing
not only for their immune complexity in the absence of a adaptive framework, resulted in the Big Bang of adaptive
mesodermal germ layer, but also for elucidation of a histo- immunity (see Fig. 4.10).
compatibility reaction that has fascinated us for decades.496 With the genome projects and advances in molecular
Teleost fish have been shown again and again to have biology, we have made strides in understanding old prob-
rapidly evolving immune systems, often expanding gene lems in vertebrate adaptive immunity, such as 1) when the
families in unprecedented ways (eg, NLR, tripartite motif- L chains emerged and what the significance of more than
containing), generating specific paralogs and losing others one isotype is; 2) which antigen receptor (if any of the ex-
(eg, cytokines, chemokines, etc.), losing canonical syntenies tant ones) came first, and all of them came to evolve to their
found in all other vertebrates (eg, MHC class I and class present state; 3) when IgD emerged and what its function is
II), and having unique forms of ancient molecules (eg, tet- in different vertebrate phyla, especially the transmembrane
rameric IgM with no J chain). The third genomewide du- form; and 4) how γ /δ T cells recognize antigen. We propose
plication (3R) is believed to have been a major force in this that are two arms of the γ /δ T-cell lineage, one innate and
instability and rapid evolution.20 This group provides an the other adaptive, similar to B cells and α / β T cells.
opportunity to study the plasticity of immune systems in The long-awaited molecular mechanisms of histocom-
related organisms over a short period of evolutionary time. patibility in nonvertebrates have been at least partially un-
With the cod story, they even offer a natural class II knock- covered in plants, Botryllus, and Hydractinia. The common
out system395! themes thus far are that the genes encoding the polymor-
Amphibians with metamorphosis offer models to study phic receptors/ligands are closely linked and that the genes
two modalities of self-tolerance, metamorphosis, and poly- all appear to be unrelated to the vertebrate MHC. However,
ploidy. The larval immune system must somehow be sup- there are tantalizing links to immune gene clusters found
pressed with the appearance of adult self-determinants throughout the animal kingdom that suggest there may
during metamorphosis and the refurbishing of the im- be an underlying fundamental similarity in recognition,
mune system. The amphibian Xenopus speciates by poly- perhaps related to 1) the paucity of surface receptors avail-
ploidy, allowing an experimental system to examine the able because of multiple constraints resulting in sharing of
effects of whole-genome duplications on immune system receptor/ligands, and 2) the signaling cascades.
loci (silencing, deletion, gene conversion, conservation), a Despite the great variation in mechanisms of allorecog-
key issue given the importance attributed to the genome- nition across phyla, are there commonalities? Whether in
wide duplications in the evolution of immune systems in cnidarians or in tunicates, the polymorphic genetic regions
vertebrates.328,399 involved contains from one to many IgSF members resem-
Birds offer another case of rapid evolution, probably bling members of the nectin PVR/CD155 family. In Ciona,
because of attack by viruses. For the MHC, birds have lost these genes are not only closely linked in a single region but
immunoproteasomes and the thymic-specific proteasome, also are homologs found in the LRC. Could this region be
and have modified their class I groove to be more promiscu- the conserved and link aspects of immunity between ver-
ous in peptide binding. Chickens (but not ducks) have lost tebrates and invertebrates? The frequent presence of ITIM
the RIG-1 gene, apparently making them susceptible to in- and ITAM indeed suggest the conservation of receptors in-
fluenza.497 Clearly, birds have something fundamental and teracting among these molecules.160 It is clearly of ancient
unique to teach us about how pathogens (most likely viruses origin. With paralogs on four different chromosome re-
in this case) can force a remodeling of innate and adaptive gions in gnathostomes as well as an ancient connection to
immunity. MHC and NKC, this region was present before 2R. It is not
The genomewide duplications early in the evolution of surprising now to detect very different molecular families
the vertebrates (the 2R hypothesis) have been confirmed in involved in immune recognition events: LRR, lectins, IgSF.
a variety of studies, and they were indeed important in the The comparison of allorecogniton in invertebrates has per-
Big Bang emergence of the adaptive immune system. The 2R haps helped to focus our attention on genetic regions that
paradigm is useful for studying any gene family found on may have provided the “context” for leukocyte interactions.

SECTION
III Immunoglobulins and B Lymphocytes
CHAPTER
5 Immunoglobulins: Structure and Function
Harry W. Schroeder Jr • David Wald • Neil S. Greenspan
INTRODUCTION the chains contains a single amino-terminal V Ig domain

Immunoglobulins (Igs) are marked by a duality of struc- and one, three, or four carboxy-terminal C Ig domains. H
ture and function.1 In common with other members of the chains contain three or four C domains, whereas L chains
Ig superfamily,2 they provide the immune system with a contain only one. H chains with three C domains tend to
conserved set of effector molecules. These effectors can ac- include a spacer hinge region between the first (CH1) and
tivate and fi x complement and they can bind to Fc receptors second (CH2).3
on the surfaces of granulocytes, monocytes, platelets, and At the primary sequence level, Igs are marked by the in-
other components of the immune response. Both activa- terspersion of regions of impressive sequence variability with
tion of complement and binding to Fc receptors can con- regions of equally impressive sequence conservation. The V
tribute to the induction or maintenance of inflammation. domains demonstrate the greatest molecular heterogeneity,
They also provide the immune system with a polyclonal set with some regions including non-germline– encoded vari-
of diverse ligand binding sites, which allow Igs, as a popu- ability and others exhibiting extensive germline conserva-
lation, to recognize an almost unlimited array of self- and tion across 500 million years of evolution.4 The molecular
non–self-antigens, which may range from compounds as heterogeneity of the V domains permits the creation of
fundamental to life as deoxyribonucleic acid to manmade binding sites, or paratopes, which can discriminate between
molecules that could not have played a role in the evolution antigens that may differ by as little as one atom. Thus it is
of the immune system. the V domains that encode the receptor function and define
Differential splicing allows individual Ig molecules to the monovalent specificity of the antibody. The H chain CH1
serve as either membrane-bound receptors for the B cell domain, which is immediately adjacent to the V, associates
that allow antigen-specific activation or as soluble effectors, with the single L chain C domain. Together, the CH1 and
which act at a distance. In vivo, proper effector function re- CL domains provide a stable platform for the paired set of
quires more than just antigen-specific binding; it requires V H and V L domains, which create the antigen binding site.
successful neutralization of the offending antigen while The distal CH2 and CH3 domains, for those antibodies with a
avoiding potentially pathogenic self-reactivity. hinge, or the CH3 and CH4 domains, for those with an extra
The receptor and effector functions of each individual Ig (CH2) domain, typically encode the effector functions of
can be localized to a separate region or domain of the mol- soluble antibody. Each of these Ig CH domains is encoded by
ecule. Each variable (V) or constant (C) domain consists of a separate exon. Although the sequences of the individual
approximately 110 to 130 amino acids, averaging 12,000 to CH domains are constant within the individual (and nearly
13,000 kD. A typical light (L) chain will thus mass approxi- constant within a species), they can vary greatly across spe-
mately 25 kD, and a three C domain Cγ H chain with its cies boundaries. The carboxy-terminal CH domain encodes
hinge will mass approximately 55 kD. a secretory tail, which permits the antibody to exit the cell.
Also encoded within the germline sequence of each CH
gene are two membrane/cytoplasmic tail domain exons,
Immunoglobulins are Heterodimers termed M1 and M2. Alternative splicing removes the se-
Igs are heterodimeric proteins, consisting of two H and cretory sequence typically encoded by the terminal CH3 or
two L chains (Figs. 5.1 and 5.2). The eponymous Ig domain CH4 domain and replaces it with the peptides encoded by
serves as the basic building block for both chains. Each of the M1 and M2 exons, converting a secretory antibody to a
129

130 | SECTION III IMMUNOGLOBULINS AND B LYMPHOCYTES
FIG. 5.1. Model of a Secreted Immunoglobulin (Ig)G. The germline exonic derivation of the sequence is shown at the top, and the protein
structure is shown at the bottom. The location of the various cysteine residues that help hold both the individual domains and the various
Ig subunits together are illustrated. Papain digests IgG molecules above the cysteine residues in the hinge that holds the two H chains
together yielding two Fab molecules and an Fc, whereas pepsin digests below releasing an (Fab)′2 fragment and two individual Fcs (which
are typically degraded to smaller peptide fragments). The location of some allotypic variants is illustrated. CH2 domains can be variably
glycosylated, which can also affect Ig structure and effector function.
membrane-embedded receptor.5 Between species, the mem- set of Ig molecules, all binding the same antigen, but at dif-
brane/cytoplasmic tail region is the most highly conserved ferent epitopes.7
portion of the CH domains, which befits its role as a link
to the intracellular signal transduction pathways that ulti- Membrane and Secretory Immunoglobulin
mately regulate B-cell function.
Alternative splicing allows Igs to serve either as soluble
antibodies or as membrane-bound antigen receptors. In
Paratopes and Epitopes their role as antibodies, Igs are released into the circula-
tion from where they may traffic into the tissues and across
The immunoglobulin-antigen interaction takes place be-
mucosal surfaces. In their role as the B-cell antigen recep-
tween the paratope, the site on the Ig at which the antigen
tor, they are anchored to the membrane by means of their
binds, and the epitope, which is the site on the antigen that is
M1:M2 transmembrane domain. Soluble antibodies can
bound. It is important to appreciate that antibodies do not
also be pressed into service as heterologous cell surface
recognize antigens; they recognize epitopes borne on anti-
antigen receptors by means of their attachment to mem-
gens.6 This makes it possible for Igs to discriminate between
brane-bound Fc receptors.8 This permits the power of anti-
two closely related antigens, each of which can be viewed as a
body recognition to be extended to nonlymphoid cells, such
collection of epitopes. It also is one scenario that permits the
as Fc-expressing granulocytes, macrophages, and mast cells.
same antibody to bind divergent antigens that share equiva-
The major difference between these two forms of cell sur-
lent epitopes, a phenomenon referred to as cross-reactivity.
face receptors is that Igs as B-cell antigen receptors provide
It has been estimated that triggering of effector functions
a monoclonal receptor for each B cell, whereas antibodies
in solution typically requires aggregation of three or more
bound to Fc receptors endow the cell with a polyclonal set of
effector domains, and thus tends to involve the binding of
antigen recognition molecules. This gives greater flexibility
three or more epitopes.6 For antigens encoding repeating
and increases the power of the effector cells to recognize an-
epitope structures, such as polysaccharides or antigen ag-
tigens with multiple non-self–epitopes.
gregates, binding of a single polymeric Ig molecule carrying
multiple effector domains, such as pentameric IgM, can be
sufficient to induce effector function. For antigens encoding Isotypes and Idiotypes
diverse epitopes, which is more typical of monodisperse sin- Igs can also serve as antigens for other Igs. Immunization of
gle-domain molecules in solution, triggering of inflamma- heterologous species with monoclonal antibodies (or a re-
tory effector functions may require the binding of a diverse stricted set of Igs) has shown that Igs contain both common

CHAPTER 5 IMMUNOGLOBULINS: STRUCTURE AND FUNCTION | 131
FIG. 5.2. Ribbon Diagram of a Complete Immunoglobulin (Ig)G1 Crystal (1 hzh in the protein data bank [PDB] from Data
of Harris et al.202). The major regions of the Ig are illustrated. The heavy-chain constant regions (green) also include the
hinge (yellow) between the first two domains. Cγ2 is glycosylated (also seen in yellow). The heavy- and light-chain vari-
able regions (red and dark blue, respectively) are N terminal to the heavy- (green) and light-chain (light blue) constant
regions. Complementarity determining region loops in the heavy- and light-chain variable regions (yellow and white) are
illustrated as well.
and individual antigenic determinants. Epitopes recognized monoclonal IgM rheumatoid factors derived from individ-
within the V portion of the antibodies used for immuni- uals with mixed cryoglobulinemia, each of which can be
zation that identify individual determinants are termed linked to the use of individual V gene segments.9
idiotypes (Fig. 5.3), whereas epitopes specific for the constant
portion are termed isotypes. Recognition of these isotypes
first allowed grouping of Igs into recognized classes. Each Classes and Allotypes
class of Ig defines an individual set of C domains that cor- Each of the various classes and subclasses of Igs has its own
responds to a single H chain constant region gene. For ex- unique role to play in the immunologic defense of the indi-
ample, IgM utilizes μ H chain C domains and IgE utilizes ε vidual. For example, IgA is the major class of Ig present in all
C domains. external secretions. It is primarily responsible for protect-
Some V domain epitopes derive from the germline ing mucosal surfaces. IgG subclasses bind Fc receptors dif-
sequence of V gene exons. These shared epitopes, commonly ferently, and thus vary in effector function.10 Determinants
referred to as public idiotopes or cross-reactive idiotypes (see common to subsets of individuals within a species, yet dif-
Fig. 5.3), are, from a genetic perspective, isotypic because fering between other members of that species, are termed
they can be found on many Igs of different antigen-binding allotypes and define inherited polymorphisms that result
specificities that derive from the same germline V region. from allelic forms of immunoglobulin C (less commonly,
Examples include the cross-reactive idiotypes found on V) genes.11

Glycosylation
N-linked carbohydrates can be found in all constant do-
mains as well as in some variable domains.12 The structure
of the attached N-linked carbohydrate can vary greatly, de-
pending on the degree of processing. These carbohydrates
can play a major role in Ig function.13 For example, human
IgG molecules contain a conserved glycosylation site at Asn
297, which is buried between the CH2 domains.14 This oli-
gosaccharide structure is almost as large as the CH2 domain
itself. O-linked sugars are also present in some Igs.12 Human
IgA1, but not IgA2, possesses a 13 amino acid hinge region
that contains three to five O-linked carbohydrate moieties.15
A deficiency in proper processing of these O-glycans can
contribute to IgA nephropathy, which is a disease that is
characterized by the presence of IgA1-containing immune
complexes in the glomerular mesangium.16
A HISTORICAL PERSPECTIVE
The identification of Ig as a key component of the immune
response began in the 19th century. This section describes
the history of the identification of Ig and introduces funda-
mental terminology.
Antibodies and Antigens

Aristotle and his contemporaries attributed disease to an
imbalance of the four vital humors: the blood, the phlegm,
and the yellow and black biles.17 In 1890, Behring (later,
von Behring) and Kitasato reported the existence of an ac-
tivity in the blood that could neutralize diphtheria toxin.18
They showed that sera containing this humoral antitoxin
activity would protect other animals exposed to the same
toxin. Ehrlich, who was the first to describe how diphtheria
toxin and antitoxin interact,19 made glancing reference to
“Antikörper” in a 1891 paper describing discrimination
between two immune bodies, or substances.20 The term
antigen was first introduced by Deutsch in 1899. He later
explained that antigen is a contraction of “Antisomatogen + -
Immunkörperbildner,” or that which induces the produc-
FIG. 5.3. Electron micrographs (top; ×350,000) and interpretive tion of immune bodies (antibodies). Thus, the operational
diagrams (below) of murine mAb hybridoma group A carbohy- definition of antibody and antigen is a classic tautology.
drate (HGAC) 39 (specific for the cell wall polysaccharide of
Streptococcus pyogenes) in complex with anti-idiotypic mAb Fab
fragments. HGAC 39 is represented in the diagrams as an open Gamma Globulins
figure, and the Fab anti-Id probes are represented as solid figures. In 1939, Tiselius and Kabat immunized rabbits with
The Fab arms of the antibody targets and probes are drawn to in- ovalbumin and fractionated the immune serum by elec-
dicate their rotational orientation as planar (oval with open cen- trophoresis into albumin, alpha-goblulin, beta-globulin,
ter), intermediate (bone shape with or without central opening), or
and gamma-globulin fractions.21 Absorption of the serum
perpendicular (“dumbbell shaped”). Different complexes illustrate
the range of Fab-Fab angles made possible by segmental flexibility. against ovalbumin depleted the gamma-globulin fraction,
1, anti-IdI-1 Fab; 2, anti-IdI-2 Fab; 39, HGAC 39; 3a, anti-IdI-3a Fab; hence the terms immunoglobulin and IgG. “Sizing” col-
K, anti-Cκ Fab; X, anti-IdX Fab. IdI designates an individual idiotope, umns were used to separate Igs into those that were “heavy”
and IdX, a crossreactive idiotope. Antibody complexes were stained (IgM), “regular” (IgA, IgE, IgD, IgG), and “light” (light
with 2% uranyl formate as described by Roux et al.108 Reproduced chain dimers). Immunoelectrophoresis subsequently per-
from Proceedings of the National Academy of Sciences (from Roux mitted identification of the various Ig classes and subclasses.
et al.108 with permission).
Fab and Fc
In 1949, Porter first used papain to digest IgG molecules
into two types of fragments, termed Fab and Fc (Table 5.1).22

permitted rational analysis of antibody structure and

Definitions of Key Immunoglobulin function.25 Recognition of the unique nature of a mol-
TABLE 5.1
Structure Nomenclature ecule consisting of one extremely variable V domain and
one highly conserved C domain led to the then-heretical
Fc A constant region dimer lacking CH1 Dreyer-Bennett proposal of “two genes, one polypeptide,”26
Fab A light chain dimerized to VH-CH1 resulting from which was subsequently and spectacularly confi rmed by
papain cleavage; this is monomeric because Tonegawa.27
papain cuts above the hinge disulfide bond(s)
F(ab)’2 A dimer of Fab’ resulting from pepsin cleavage below
the hinge disulfides; this is bivalent and can
precipitate antigen
THE IMMUNOGLOBULIN DOMAIN
Fab’ A monomer resulting from mild reduction of F(ab)’2: The Ig domain is the core unit that defines members of
an Fab with part of the hinge the Ig superfamily (reviewed in Williams and Barclay2 and
Fd The heavy chain portion of Fab (VH-CH1) obtained Harpaz and Chothia28). This section describes the Ig do-
following reductive denaturation of Fab main in detail.
Fv The variable part of Fab: a VH-VL dimer
Fb The constant part of Fab: a CH1-CL dimer
pFc’ A CH3 dimer The Immunoglobulin Superfamily
From Carayannopoulos and Capra206 with permission.
Each Ig domain consists of two sandwiched β-pleated sheets
“pinned” together by a disulfide bridge between two con-
served cysteine residues (Fig. 5.4). The structure of the
Papain digested IgG into two Fab fragments, each of which β-pleated sheets in an Ig domain varies depending on the
could bind antigen, and a single Fc fragment. Nisonoff de- number and conformation of strands in each sheet. Two
veloped the use of pepsin to split IgG into an Fc fragment such structures, V and C, are typically found in Igs. C-type
and a single dimeric F(ab)2 that could cross-link antigens.23 domains, which are the most compact, have seven antiparal-
Edelman broke disulfide bonds in IgG and was the first to lel strands distributed as three strands in the fi rst sheet and
show that IgG consisted of two H and two L chains.24 four strands in the second. Each of these strands has been
given an alphabetical designation ranging from amino ter-
minal A to carboxy-terminal G. Side chains positioned to lie
Two Genes, One Polypeptide sandwiched between the two strands tend to be nonpolar in
The portion of the constant domain encoded by the Fc nature. This hydrophobic core helps maintain the stability
fragment was the fi rst to be sequenced and then analyzed of the structure to the point that V domains engineered to
at the structural level. It could be readily crystallized when replace the conserved cysteines with serine residues retain
chilled. The heterogeneity of the V domain precluded se- their ability to bind antigen. The residues that populate the
quence and crystallographic analysis of an intact Ig chain external surface of the Ig domain and the residues that form
until Bence-Jones myeloma proteins were identified as the loops that link strands can vary greatly in sequence.
clonal, isolated Ig light chains. These intact chains could These solvent exposed residues offer multiple targets for
be purified and obtained in large quantities, which fi nally docking with other molecules.
A C
FIG. 5.4. The Immunoglobulin Domain. A: A

typical variable domain structure. Note the
projection of the C-C’ strands and loop away
from the core. B: A typical compact constant
domain structure. C: The cysteines used to
pin the two b-sheets together are found in
the B and F strands. D: The folding pattern
for variable and constant domains.2 B D

FIG. 5.5. Sequence Conservation and Hypervariability within a Heavy Chain Variable (V) Domain. The primary
sequence the V domain can be divided into four regions of sequence conservation, termed framework regions
(FRs), and three regions of hypervariability, termed complementarity determining regions (CDRs). A schematic
of the genomic origin of the variable domain is shown at the top of the figure. The classic separation of the se-
quence into FR and CDR by Kabat et al.31 is shown below the gene structure. The letter designation for individual
β strands is given beneath the Chothia and Lesk nomenclature,32 which focused more on structure. The posi-
tions of each of the four invariant residues of the VH chain (FR1 Cys22, FR2 Trp36, FR3 Cys92, and FR4 Trp103) are
shown as darkened circles on the Chothia and Lesk model. The ImMunoGeneTics information system® designa-
tion has attempted to rationalize sequence variability with structure and is the current nomenclature of choice.33
The V Domain ImMunoGeneTics information system® (IMGT)33 (see

V-type domains add two additional antiparallel strands to Fig. 5.5). (For students of the Ig repertoire, IMGT maintains
the first sheet, creating a five strand–four strand distribu- an extremely useful website, http://www.imgt.org, which
tion. Domain stability results from the tight packing of al- contains a large database of sequences as well as a multi-
ternately inward-pointing residue side chains enriched for plicity of software tools.)
the presence of hydrophobic moieties to create a hydropho- The C and C’ strands that define a V domain form FR2.
bic domain core. The H and L variable domains are held to- These strands project away from the core of the molecule
gether primarily through noncovalent interaction between (see Figs. 5.4 and 5.6) where they take on a conserved struc-
the inner faces of the β sheets.29,30 ture that is parallel and opposite to the FR2 of the compan-
Early comparisons of the primary sequences of the ion V and adjacent to the FR4 of the complementary chain.
V domains of different antibodies identified four inter- Approximately 50% of the interdomain contacts in the
vals of relative sequence stability, termed framework re- hydrophobic core of the V domain are formed by contacts
gions (FRs), which were separated by three hypervariable between the FR2 of one chain and the FR4 of the comple-
intervals, termed complementarity determining regions mentary chain.30 Another 30% to 45% is contributed by
(CDRs) (Figs. 5.5 and 5.6).31 The exact location of these contacts between the CDR3 and the FR2 or CDR3 of the
intervals has been adjusted over the years, fi rst by a focus complementary domain. The overall interdomain contact
on the primary sequence, 31 then by a focus on the three- includes between 12 and 21 residues from the L chain V do-
dimensional structure, 32 and, more recently, by a consensus main and 16 to 22 residues from H chain V domain, most of
integration of the two approaches by the international which are contributed by the FR2, CDR3, and FR4 regions.

highly homologous scaffold. This section describes the char-

acteristics of the Fab domain, its component V domains,
and the paratope, which is the part of the Fab that actually
binds antigen.
Fab, Fv, and Fb

The antigen-binding fragment (Fab) is a heterodimer that
contains an L chain in its entirety and the V and CH1 por-
tions of the H chain (see Figs. 5.1 and 5.6). The Fab in turn
can be divided into a variable fragment (Fv) composed of
the V H and V L domains, and a constant fragment (Fb) com-
posed of the CL and CH1 domains (see Table 5.1). Single Fv
fragments can be produced in the laboratory through ge-
netic engineering techniques.35 They recapitulate the mon-
ovalent antigen binding characteristics of the original, par-
ent antibody. Other than minor allotypic differences, the
sequences of the constant domains do not vary for a given
H chain or L chain isotype. The eponymous V domain,
however, is quite variable.
Generation of Immunoglobulin Variable Domains by

Recombination
Ig V domain genes are assembled in an ordered fashion by
a series of recombination events.27,36–39 The elegant mecha-
FIG. 5.6. The Structure of an Fab. The antigen-binding site is formed
by the heavy (H) and light (L) chain B-C, C′-C″, and F-G loops. Each nisms used for the assembly of these genes and the Fvs they
loop encodes a separate complementarity determining region (CDR). create are fully discussed in Chapter 6. However, in order to
The location of CDRs H1, H2, H3, L1, L2, and L3 are shown. The oppos- understand the relationship between antibody structure and
ing H and L chain C-C′ strands and loop help stabilize the interaction function, a brief review is in order.
between VH and VL. This C-C′ structure is encoded by the second In BALB/c mice, Ig V assembly begins with the joining of
V framework region, FR2. The inclusion of this structure permits one of 13 diversity (DH) gene segments to one of four joining
the variable (V) domains to interact in a head-to-head fashion. The
E-F strands and loop are encoded by the FR3 region and lie directly
(JH) gene segments. This is followed by the joining of one
below the antigen-binding site. The A-B strands and loop encode of 110 functional variable (V H) gene segments.40–42 Each DH
FR1 and lie between the CH1 and CL domains and the rest of the Vs. gene segment has the potential to rearrange in any one of six
The beta sheet strands of the CH1 and CL domains rest crosswise to reading frames (RFs), three by deletion and three by inver-
each other. The illustration is modified.203 sion. Thus, these 127 gene segments can come together in
3.4 × 104 combinations. The numbers of gene segments can
vary quite widely between different mouse strains, but the
process of assembly and the multiplicative effect of combi-
There are approximately 40 crucial sequence sites that natorial diversity is the same.
influence variable domain inter- and intradomain interac- In the final assembled V domain, the V H gene segment
tions.32,34 Four of these sites are relatively invariant: the two encodes FRs H1 to H3 and CDRs H1 and H2 in their entirety
cysteines that form the disulphide bridge between the beta (see Figs. 5.1 and 5.5), and the JH encodes FR-H4. CDR-H3 is
sheets and two tryptophan (phenylalanine in Jκ) residues, created de novo in developing B cells by the joining process.
one near the beginning of the C strand and the second near CDR-H3 contains the DH gene segment in its entirety, as well
the beginning of the G strand, that pack against the bridge as portions of the V H and JH gene segments.
to add stability. Beyond these and other common core resi- After a functional H chain has been created, L chain
dues, Ig domains can vary widely in their primary amino assembly begins. Mice contain two L chain loci, κ and λ. In
acid sequences. However, a common secondary and tertiary C57BL/6 mice, the κ locus includes five Jκ gene segments
structure characteristic of the core Ig V domain tends to be and 140 Vκ gene segments, of which 4 and 73 have been
preserved. shown to be functional, respectively.43–45 This provides
292 combinations. There is only one Cκ .46 The BALB/c λ
locus contains three Vλ, three functional Jλ, and two or
FAB STRUCTURE AND FUNCTION three functional Cλ chains.47,48 The λ constant domains
Introduction are functionally indistinguishable from each other. Due to
It is the Fab domain that allows Ig to discriminate between gene organization, the λ repertoire provides at most seven
antigens. The Fab is individually manufactured to precise combinations. Each V L encodes FRs L1 to L3, CDR L1 and
specifications by individual developing B cells. It shows an L2, and two thirds of CDR-L3 (see Fig. 5.1). Each J L en-
amazing array of binding capabilities while maintaining a codes one-third of CDR-L3 and FR4 in its entirety. Any one

H chain can combine with any one L chain, thus 211 V, D, through mechanisms of gene conversion, and V gene seg-
and J gene segments can provide approximately 1 × 107 dif- ments are no exception. Sequence relationships allow group-
ferent H:L combinations. ing them into families and clans of sequences that share
At the V→D and D→J junctions, the potential for CDR- nucleotide homology,52 as well as structural features. Close
H3 diversity is amplified by imprecision in the site of joining, inspection of the V H gene repertoire has shown that these
allowing exonucleolytic loss as well as palindromic (P junc- family relationships reflect segmental gene conversion cou-
tion) gain of terminal VH, DH, or JH germline sequence. B cells pled with selection for function.4,53,54
that develop after birth express the enzyme terminal deoxy- Due to the need to maintain a common secondary and
nucleotidyl transferase (TdT) during the H chain rearrange- tertiary core Ig V domain structure capable of associating
ment process.27,36 TdT catalyzes the relatively random incor- randomly with a complementary V chain to form a stable Fv,
poration of non-germline– encoded nucleotides between V H the core sequence of FR2, which is encoded by the V H gene
and DH, and between DH and JH. Each three nucleotides of N segment, and the core sequence of FR4, which is encoded by
addition increase the potential diversity of CDR-H3 20-fold. the JH gene segment, are highly conserved among all Ig V do-
Thus sequences with nine nucleotides of N addition each be- mains. Conversely, the need to generate a diverse repertoire
tween the V→D and D→J junctions would enhance the po- of antigen-binding sites has led to extensive diversity in the
tential for diversity by (20) 6, or by 6 × 107; six-fold greater CDR-1 and CDR-2 intervals. One might presume that the
than the potential diversity provided by VDJ gene segment FR1 and FR3 intervals, which form the external surface of
combinations. These genomic gymnastics permit the length the antibody, would not be under any particular constraints,
of CDR-H3 to vary from 5 to 20 amino acids among develop- but sequence comparisons suggest otherwise.
ing B cells in BALB/c bone marrow.49 Together, imprecision Given the need to diversify the CDRs and the need to
in the site of VDJ joining and N addition provides the op- preserve FR2, it is not surprising that family identity, which
portunity to create nearly random CDR-H3 sequence, poten- might reflect ancestral relationships, can be assigned by the
tially freeing the CDR-H3 repertoire from germline sequence extent of FR1 and FR3 similarity.55 Of these, FR1 appears to be
constraints. Although a limited amount of N addition is ob- under the greatest constraints, with VH gene segments belong-
served between VL and JL in human,50,51 N addition in murine ing to different families both within and across species barri-
L chains is distinctly uncommon. Moreover, the length of ers exhibiting extensive similarities in FR1 sequence (Fig. 5.7).
CDR-L3 appears to be under relatively strict control, greatly Sequence similarities in FR1 and, to a lesser extent, FR3 allow
limiting the potential for somatic L chain junctional diver- grouping of human and murine VH families into three clans of
sity.48,50 Thus, CDR-H3 represents the greatest focus for the related sequences, presumably reflecting an early divergence
initial somatic diversification of the antibody repertoire. in sequence from a primordial VH gene sequence (Fig. 5.8).
Segmental Conservation and Diversity within the Constraints on the Sequence and Structure of
V Domain Variable-Encoded Complemenatarity Determining
Although the large numbers of V gene segments might give Regions
the impression of a smooth incremental range of available The antigen-binding site of an Ig is formed by the juxta-
diversity, multigene families are thought to evolve in concert position of the six hypervariable H and L chain V domain
FIG. 5.7. A comparison of two human Clan I VH sequences that belong to different VH families (modified4). Shown is a compari-
son of the deoxyribonucleic acid and amino acid sequences of the V5-51 and V1-2 gene segments. Each line depicts a diver-
gence in a nucleotide or amino acid at that position. Shown at the bottom is a replacement/silent site substitution analysis by
interval. Random mutation tends to exhibit an R/S ratio of 2.9. The smaller the ratio, the greater the preservation of sequence.
The intervals identified by the arrows predict the family and clan of origin.

FIG. 5.8. Evolutionary Relationships among Vertebrate VH Families. The sizes or relative placements of the evolutionary connecting lines are
not to scale. VH sequences from all mammalian species analyzed to date can be placed into one of these three clans (modified4).
intervals: CDRs-H1, -H2, and -H3; and CDRs-L1, -L2, and classes). It further suggests evolutionarily and structurally
-L3.31 The CDR sequences of V gene segments tend to be en- imposed restrictions operating to counteract the random
riched for codons where mutations maximize replacement diversification of these CDRs.
substitutions.56 This includes the RGYW motif that facili-
tates somatic hypermutation.57,58 While evolution appears to
favor CDR1 and CDR2 sequences that facilitate codon di- Diversity and Constraints on the Sequence and
versity, it also appears to preserve specific loop structures. Structure of CDR-H3
Although there is great variation in the sequence and The combination of VDJ assortment, variation in the site
size of these CDRs, it has been shown that five of them, of gene segment rearrangement, and N nucleotide addition
CDR-H3 being the notable exception, possess one of a makes CDR-H3 the most variable of the six hypervariable
small set of main-chain conformations termed canonical regions. In some cases, the sequence of CDR-H3 appears de-
structures.32,34,59–62 Each canonical structure is determined signed to provide optimal flexibility.65 Correspondingly, it
by the loop size and by the presence of certain residues at has been more difficult to assign canonical structures to the
key positions in both the loop and framework regions. It CDR-H3 loops similar to those observed for the V-encoded
can be calculated that the total number of possible com- CDRs. However, insight into a gradient of possible struc-
binations of canonical structures, or structure classes, tures has been gained.
is 300.63 However, only 10 of these combinations, or CDR-H3 can be separated into a base, which is adjacent
classes, are sufficient to describe the majority of human to the frameworks, and a loop. The base tends to be stabi-
and mouse Fab sequences. Among specific classes, the lized by two common residues, an arginine at Kabat posi-
lengths of CDR-H3 and -L1 appear to correlate with the tion 94 (IMGT 106) and an aspartic acid at Kabat 101 (IMGT
type of recognized antigen. Antibodies with short loops in 116).32,33 These form a salt bridge which, together with the
-H2 and -L1 appear to be preferentially specific for large adjacent residues, tends to create one of three backbone con-
antigens (proteins), whereas antibodies with long loops in formations, termed kinked, extrakinked, and extended.66
-H2 and -L1 appear to be preferentially specific for small In some sequences with kinked or extrakinked bases, it is
molecules (haptens).63 possible to predict whether an intact hydrogen-bond ladder
Given that the sequence and structure of the frame- may be formed within the loop of the CDR-H3 region, or
work regions, which define families, influences the canoni- whether the hydrogen-bond ladder is likely to be broken.67,68
cal structure of V H-encoded CDRs, it is not surprising that However, for many CDR-H3 sequences, especially those that
the structure repertoire of canonical structures is strongly are longer, current tools provide less than optimal predic-
associated with family and clan identity.64 This implies re- tions for the structures of individual CDR-H3s.
strictions to the random diversification of the hypervariable Despite the potential for totally random sequence pro-
loop structures (canonical structures) and their combina- vided by the introduction of N nucleotides, close inspec-
tions within the same V H gene segment (canonical structure tion has shown that the distribution of amino acids in the

CDR-H3 loop is enriched for tyrosine and glycine,31,69 and length can create an antigen-binding groove. Each species
relatively depleted of highly polar (charged) or nonpolar appears to prefer a specific range of CDR-H3 lengths.75 Long
(hydrophobic) amino acids, although the precise pattern CDR-H3s, which can create “knobs” at the center of the
depends on the species of origin.70 This pattern of amino antigen-binding site, are unevenly distributed between spe-
acid utilization is established early in B-cell development, cies and reflect both divergence in germline sequence and
prior to the expression of Ig on the surface of the cell somatic selection.
(Fig. 5.9)49,71–73 and reflects evolutionary conservation of
JH and DH gene segment sequences. In particular, although
the absolute sequence of the DH is not the same, the pattern The Antigen-Binding Site is the Product of a Nested
of amino acid usage by RF is highly conserved. Of the six Gradient of Regulated Diversity
potential RFs, RF1 by deletion is enriched for tyrosine and The tension between the need to conserve essential struc-
glycine. RF2 and RF3 by deletion are enriched for hydro- ture and the need to emphasize diversity in an environment
phobic amino acids, as they are by inversion. RF1 by inver- subject to unpredictable antigen challenge appears to create
sion tends to encode highly polar, often positively charged, a gradient of regulated diversity in the Fv. The most highly
amino acids.69 Various species use different mechanisms to conserved components of the Fv are FR2 and FR4, which
bias for use of RF1 by deletion, to limit use of hydropho- form the hydrophobic core of the VH:VL dimer (see Figs. 5.4
bic RFs, and to restrict or prevent use of RFs enriched for and 5.5). FR1, which in the H chain presents with three con-
charged amino acids. Forced rearrangement into RFs with served structures, helps form the ball and socket joint be-
charged amino acids yields an altered repertoire enriched tween the VH and CH1. FR3, which in the H chain defines the
for charge and depleted of tyrosine and glycine.71 family and provides 7 different structures in human versus
The distribution of CDR-H3 lengths can also be regu- 16 different structures in mouse, frames the antigen-binding
lated both as a function of differentiation and as a function site (Fig. 5.10). The V-encoded CDRs, -H1, -H2, -L1, -L2, and
of ontogeny.49,74 In association with long V-encoded CDRs, most of -L3, are programmed for diversity. However, con-
short CDR-H3s create an antigen-binding cavity at the cen- served residues within these CDRs, which interact with V
ter of the antigen-binding site, and CDRs of intermediate family–associated FR3 residues, constrain diversity within a
TdT-ko
B n = 67
40
C n = 162
35
30 D n = 84
25
Percent
E n = 83
20
F n = 73
15
10
5
0
Young Adult Repertoire

B n = 194
40
C n = 373
35
30 D n = 279
25
Percent
E n = 255
20
F n = 254
15
10
5
0
R K N D Q E H P Y W S T G A M C F L V I
Amino Acid
FIG. 5.9. Both in the absence or presence of N-addition, the preference for tyrosine and glycine in complementarity determining region-H3
begins early and intensifies with B-cell development. VH7183DJCμ transcripts were cloned and isolated from fractions B (pro-B cells) through
F (mature B cells) from the bone marrow of 8- to 10-week-old terminal deoxynucleotidyl transferase-sufficient and terminal deoxynucleotidyl
transferase-knockout BALB/c mice.73 The amino acids are arranged by relative hydrophobicity, as assessed by a normalized Kyte-Doolittle
scale.204,205 Use is reported as the percent of the sequenced population. The number of unique sequences per fraction is shown.

Binding of “Superantigens” to Nonclassic Variable

Domain Antigen-Binding Sites
Not all antigens bind to the paratope created by the classic
antigen-binding site. Antigens that can bind to public idiot-
opes on V domain frameworks81 and recognize large portions
of the available repertoire are termed superantigens. There
are indications that B cell superantigens influence the patho-
genesis of some common infections, such as those caused by
Staphylococcus aureus.
ANTIGEN-ANTIBODY INTERACTIONS
Technological advances in biomolecular structure determi-
nation, analysis of molecular dynamics, protein expression
and mutagenesis, and biophysical investigation of recep-
tor-ligand complex formation have facilitated significant
advances in the understanding of antigen-antibody inter-
actions. Particularly valuable has been the integration of
high-resolution structural data with thermodynamic and
kinetic analyses on a number of antigen-monoclonal an-
FIG. 5.10. Location and Generation of Complementarity Determining tibody complexes. In this section, some of the key insights
Region-H3. A: A cartoon of the classic antigen-binding site (modified4). arising from these studies will be reviewed.
Due to its central location, most antigens bound to the antibody will in-
teract with complementarity determining region-H3.
Molecular Flexibility
Like many other protein domains, V domains exhibit varying
degrees and modes of molecular flexibility. Evidence suggests
that some V modules (ie, VL-VH pairs) can adopt two or more
preferred range of canonical structures. CDR-H3, the focus conformations with meaningful frequencies in the unbound
of junctional diversity, lies at the center of the antigen-bind- state. Because these different conformational states can ex-
ing site. The conformation of its base tends to fit within three hibit distinguishable binding proclivities, molecular flexibil-
basic structures. The loop varies greatly in sequence, yet still ity provides monoclonal antibodies with a mechanism for
maintains a bias for the use of tyrosine and glycine. Thus, polyspecificity.82 Molecular flexibility can also play a role in
diversity increases with proximity to the tip of the antigen- binding a single ligand, which, in such instances, may be bet-
binding site but appears to be held within regulated limits. ter understood as a process of binding rather than as a simple
The extent and pattern of diversity in CDR-H3 can have a event.83,84 The extent of conformational adjustment by anti-
critical effect in the biologic function of Ig as a soluble effec- body or antigen required for complex formation can influ-
tor molecule. Absence of N addition, with its enhancement ence both the thermodynamics and kinetics of that process.
of tyrosine and glycine usage (see Fig. 5.9) facilitates B-cell
development, whereas enrichment for charged amino acids
impairs it.71,76,77 However, both the absence of N addition
Role of Water
and enrichment for charged amino acids impair immune re- A significant role for water molecules has become clear from
sponses and protection in vivo.71,76,78 the study of high-resolution structures and thermodynamic
analyses of antigen-antibody complexes. Water molecules
Somatic Hypermutation and Affinity Maturation exhibit a broad range of association times with protein sur-
faces. Thus some of the more tightly protein-associated of
Following exposure to antigen and T-cell help, the V domain these solvent molecules effectively behave as parts of the
genes of germinal center lymphocytes can undergo mutation protein.85 Water molecules found in the antigen-antibody
at a rate of up to 10−3 changes per base pair per cell cycle,79 interface, whether constitutively bound or newly recruited,
a process termed somatic hypermutation. Somatic hyper- can make important contributions to both the intrinsic (ie,
mutation allows affinity maturation of the antibody reper- monovalent) affinity of the complex and to the differential
toire in response to repeated immunization or exposure to affinities for different ligands (ie, specificity).86 Amino acid
antigen. Although affinity maturation often preserves the residues of antigen and antibody can interact, indirectly,
canonical structure of the CDR loops, the distribution of through hydrogen bonds to one or more water molecules.86
diversity appears to differ between the primary and antigen-
selected repertoire.80 In the primary repertoire, diversity is
focused at the center of the binding site in CDR-H3. With Thermodynamics and Antigen-Antibody Interactions
hypermutation, somatic diversity appears to spread to the Specific residues in the antibody V domains or the antigen
V-encoded CDRs in the next ring of the binding site (see Fig. can contribute to complex formation in different ways, and
5.10), enabling a more custom-tailored fit. some residues can contribute in multiple ways.87 In addition

to making van der Waals contact, residues can be important Hinges

due to contributions to the free energy of complex forma- Some Ig isotypes contain a structural element that does not
tion or to the differential free energy of complex forma- strictly correspond to the canonical structural motifs of Ig
tion for two or more different ligands. Some residues may superfamily V and C domains, which is termed the hinge.
contribute primarily to modulation of the association rate, Where it occurs, it is located between the C-terminus of the
the “relaxation” of the forming complex, or the dissociation CH1 and the N-terminus of the CH2 domains. In isotypes
rate.83,88 Other contact residues contribute minimally to the such as IgG and IgA, the hinge is encoded by one (or more)
energetics of complex formation, and yet other noncontact separate exon(s). In isotypes with four CH domains (ie, IgM
residues can be significant thermodynamic contributors.89,90 and IgE), the CH2 domain serves in place of a classical hinge.
Structural and thermodynamic/kinetic comparisons of The hinge, or the CH2 domain in Igs that lack a hinge,
antibodies with germline V domain sequences and with permits an Fab arm to engage in an angular motion rela-
somatically mutated V domain sequences have provided tive both to the other Fab arm and to its Fc stem (Figs. 5.3
new insights into the structural and energetic bases for af- and 5.11). This permits the two Fab arms to cover a range
fi nity maturation.89,90 Mutations in both contact and non- from maximal extension to an almost parallel alignment.
contact residues can have major consequences, positive or The range of motion of the Fab arms reflects the nature of
negative, for the affi nity with which an antibody binds an the hinge region, which in some C genes is rigid and in oth-
antigen. They can also favor tighter binding by enhancing ers, such as human IgA1, functions more as a tether for each
V domain rigidity, thereby reducing the entropic penalty individual Fab than as a support. This flexibility has major
associated with complex formation.90 Antibodies derived implications for antibody function, because it enables a bi-
from secondary or later responses that have incorporated valent antibody molecule to bind epitopes in a variety of
somatic mutations have been shown to exhibit less than relative spatial arrangements.
absolute specificity. Even these “mature” antibodies can Among human IgG subclasses, the most unusual hinge re-
bind multiple ligands when screened on libraries of pep- gion is that of IgG3. Unlike other human IgG hinge regions, the
tides or proteins,91,92 a lesson likely to be relevant to most IgG3 hinge is encoded by a quadruplicated hinge exon, making
biomolecules. it the longest hinge (62 amino acids) by far (Table 5.2). The
It has been routine to distinguish between interactions primary structure of this hinge has been divided into upper,
in which antibodies bind to protein antigens versus those core (or middle), and lower hinge regions with somewhat dif-
in which antibodies bind to carbohydrate antigens. Recent ferent functional associations. The upper hinge in particular
studies of antibodies that bind to human immunodefi- has been associated with the magnitude of segmental flexibil-
ciency virus-1 gp120 with high affinity and that exhibit po- ity as assessed by fluorescence emission anisotropy kinetics98
tent and broad neutralizing activity suggest that antibodies
can bind to epitopes composed of both peptide and glycan
elements.93,94
IMMUNOGLOBULIN “ELBOW JOINTS” AND

“HINGES”
The structure of the constant domains can affect antibody-
antigen interactions by influencing the range of molecular
flexibility permitted between the two Fabs. In this section,
the role of the Ig hinge will be discussed.
Elbow Joints
Individual Ig V and C domains tend to create rather rigid
dimers. However, the antibody molecule as a whole, which
consists of four or more such dimeric modules linked like
beads on a string, can be viewed as a paradigm of molecular
flexibility.95 Flexibility begins between the Fv and the Fb of
the Fab at what is termed the elbow bend or elbow angle.96
This reflects both a ball and socket interaction between the
FR1 of the H chain and the CH1 domain,97 and the identity
of the L chain.96 Five residues (three in V H and two in CH1)
that are highly conserved in both antibodies and T-cell re-
ceptors make the key contacts that constitute this “joint.” FIG. 5.11. Illustration of the Motions and Flexibility of the
Immunoglobulin. Axial and segmental flexibility are determined by
Elbow angles, assessed from crystal structures of homoge- the hinge. The switch peptide (elbow) also contributes flexibility to
neous Fab fragments, range from 130 degrees to 180 degrees. the Fab. The measure of the elbow angle is defined with respect to
λ light chains appear to permit an Fv to adopt a wider range the Fv and Fb axes of two-fold symmetry. From Carayannopoulos
of elbow angles than their κ chain counterparts. and Capra206 with permission.

TABLE 5.2 Properties of Hinges in Immunoglobulin G, A, and D

Genetic Hinge
Upper Hinge Middle Hinge Lower Hinge Configuration Susceptibility to
Ig Type Length Length Length (Amino Acids/Exon) Proteolysis Special Features
IgG1 4 10 6 15
IgG2 3 8 6 12
IgG3 12 49 6 17-15-15-15
IgG4 7 4 6 12
IgA1 1 23 2 19 High Heavy O-linked glycosylation
IgD 34-24 High Extensive charged amino acids;
heavy O-linked glycosylation
at N terminus
and with the magnitude of Fab-Fab flexibility by immunoelec- involved in inter-H chain disulfide bonds. Experiments in
tron microscopy.99 The core or middle hinge appears to serve, which IgG molecules were modified to eliminate the hinge
at least in part, a spacer function. The lower hinge functions region demonstrate that covalent linkage between the hinge
primarily to facilitate CH2-CH2 interactions. regions just “upstream” of the two CH2 domains is critical
Segmental flexibility of the Ig molecule, conferred mainly for the preservation of IgG effector function.111
by the hinge, permits or facilitates simultaneous binding
through two or more Fab arms. Such monogamous biva- HEAVY CHAIN STRUCTURE AND FUNCTION
lency or multivalency, which enhances overall binding,100,101
What might be termed the fundamental strategy of hu-
is a crucial factor permitting biosynthetically feasible an-
moral immunity is a two-step process that begins with the
tibody concentrations to offer adequate immunity against
identification by antibodies (of appropriate binding specific-
replicating pathogens. While it is more speculative why the
ity) of the molecules or molecular complexes that should be
Fab-Fc geometry needs to vary, it may have to do with opti-
eliminated. Following such identification (ie, noncovalent
mizing effector function activation when antigen is bound,
complex formation), antibodies can then trigger other mo-
such as maintaining antigen binding when Fc receptors are
lecular systems (eg, complement) or cells (eg, phagocytes)
simultaneously engaged.
to destroy or remove the antigenic material—guilt by as-
The attribution of flexibility control to the hinge is sup-
sociation at the molecular level. Thus, antigen specificity,
ported by protein engineering studies in which V domain–
determined primarily by the V domains in the Fab arms,
identical IgGs of different subclasses were analyzed.98,102 This
is physically and functionally linked to effector function,
basic conclusion is also supported by studies in which hinge
the activation of which is primarily attributable to the C
regions have been selectively mutated or swapped among
domains of the Fc region. The effector functions associated
human or mouse IgG subclasses.103–105 One of these studies in-
with the humoral immune response primarily involve either
dicated that structural variation among subclasses in the CH1
complement or Fc receptor–bearing cells, such as neutro-
domain also influenced segmental flexibility as assessed by
phils, macrophages, and mast cells. As might be expected,
nanosecond fluorescence polarization measurements.103 Early
therefore, the Fc contains sites for noncovalently interacting
suggestions that IgG subclass–related differences in activat-
with complement components, such as C1q, and with Fc re-
ing the classical pathway of complement were explained by
ceptors. This section focuses on the structure and function
differences in segmental flexibility 98 were not confirmed by
of the Fc regions of the various Ig classes and subclasses.
the studies in which mutant hinge regions were created.103–105
There are several types of molecular motion attribut-
able to the hinge region that contribute to overall segmental Structure and Function of the Fc
flexibility (see Fig. 5.11). These include flexing between Fab The necessity for interacting effectively with relatively con-
arms (motion toward or away from one another in the same served molecules such as C1q and Fc receptors provides a
plane), Fab arms moving in and out of the same plane, Fab selective basis for maintaining the primary structures of
arms rotating along their long axes, and Fab arms moving the Fc region, at least where changes would undermine
in or out of the same plane as the Fc region.106,107 The inter- such intermolecular contacts. The degeneracy possible in
Fab angles observed by electron microscopy range from 0 noncovalent molecular recognition events permits selected
degrees to 180 degrees.108,109 Similarly, Fab arm long-axis ro- primary sequence variations without catastrophic alterations
tations can extend up to 180 degrees.110 in function. However, there are allotypic differences in the
Another key role of the hinge is the maintenance of the H chain constant domains among both human and mouse
CH2-CH2 interaction (ie, effectively constraining molecular Igs. In some cases, these allotypic differences are associated
mobility within the Fc region itself). The lower hinge stabi- with variation in function, at least in vitro. For example,
lizes CH2-CH2 contacts by providing the key cysteine residues two recombinant V domain–identical IgG3 antibodies, of

different allotypes, exhibited differential abilities to bind C1q be created during somatic hypermutation.124 Glycosylation
or initiate antibody-dependent cell-mediated cytotoxicity.112 in the Fc region has been shown to be important for anti-
Nevertheless, the H chains of the human IgG subclasses are body half-life and effector functions.125–127 Glycosylation of
particularly well conserved, with > 90% identity of amino the Fc domain influences complement activation as Ig hypo-
acid sequences. In other mammalian species that have two glycosylation influences affinity for C1q as well as Ig bind-
or more IgG subclasses, they tend to exhibit less amino acid ing to the FcR, possibly due to its effects on Ig structure.128
sequence identity than the human IgG subclasses. Differential sialylation of the core Fc polysaccharide has
By convention, the H chain constant domains are num- recently been shown to have dramatic effects on the proin-
bered from N-terminal to C-terminal, with CH1 residing flammatory versus anti-inflammatory activity of IgG.129,130
in the Fab arm and the remaining two (IgG, IgA, IgD) or Furthermore, IgD N-linked glycans are necessary for IgD to
three (IgM, IgE) C domains (CH2, CH3, and, if relevant, bind to the IgD receptor on T cells.131
CH4) residing in the Fc. In IgM and IgE, the CH2 domain V domain glycosylation potentially affects the affinity for
largely plays the role of the hinge region. The C domains antigen, antibody half-life, antibody secretion, and organ
of different isotypes and from different species share sev- targeting. Interestingly, glycosylation has been shown to be
eral key structural features. In distinction from V domains, capable of both positively and negatively affecting antigen
which consist of four- and five-stranded β -pleated sheets binding.132–135 The biologic significance of Ig glycosylation
linked through an intrachain disulfide bond, C domains can be seen from studies demonstrating that IgG from pa-
consist of three- and four-stranded β -pleated sheets linked tients with rheumatoid arthritis is galactosylated to a lesser
through an intrachain disulfide bond (see Fig. 5.4). Across extent (termed IgG G0) than IgG from normal controls.
isotypes, amino acid sequence identity for CH domains is In some cases, hypogalactosylation correlates with disease
approximately 30%, while for subclasses (within an iso- activity.136 Hypogalactosylation of IgG has also been found
type) the amino acid sequence identity for CH domains is in to occur in other chronic inflammatory diseases such as
the range of 60% to 90%. Important physical and biologi- Crohn’s disease and systemic lupus erythematosus.137
cal properties of the human Ig isotypes are summarized in
Table 5.3.
Immunoglobulin M
As noted previously, the traditional and accepted func-
tional anatomy of Igs attributes antigen binding (both IgM is an isotype of firsts. It is ontogenetically primary,
specificity and affinity) to the V modules in the Fab arms being expressed first on developing B lineage cells. IgM is
and effector function activation to the Fc region. While this also the isotype that initially dominates the primary hu-
scheme is both well supported and appealing, there is con- moral immune response. It is probably, along with IgD, a
siderable (perhaps not widely appreciated) evidence that in phylogenetically primitive isotype in jawed vertebrates (an
some cases structural variations in H chain domains (ie, almost first) and may be the most phylogenetically stable
CH domains and hinge) can influence both the affinity of isotype.138
the antibody for antigen and the discrimination among IgM serves important immunological functions both on
antigens.113–118 Although these instances of C domain influ- the surfaces of B-lymphocytes and in the fluid phase in the
ence on ligand binding (through the V domains) primar- blood and in the mucosal secretions. On the cell surface,
ily involve multivalent antigens, there are also reports that IgM consists of two identical μ H chains and two identical
suggest C domain influence in instances of monovalent L chains (μ2L2). It is initially expressed on B lineage cells
recognition.119–121 Mechanisms for these effects in the con- in noncovalent association with surrogate L chains, and
text of binding multivalent antigens include isotype-related subsequently, following successful L chain gene rearrange-
differences in segmental flexibility as well as the tendency ment, with κ or λ light chains. On the mature B-cell surface,
for self-association. Results from a study comparing resis- IgM is noncovalently associated with two other polypeptide
tance to pneumococcal infection for IgG3-deficient and chains, Ig-α (CD79a) and Ig-β (CD79b).139–141 These integral
IgG3-producing mice are consistent with the notion that the membrane proteins serve to transduce signals when surface
cooperative binding permitted by murine IgG3 antibodies IgM binds to and is cross-linked by cognate antigen.
contributes to the effectiveness of humoral immunity.122 In the secreted form, IgM can consist of either pentamers
(μ2L2)5 or, less often, hexamers (μ2L2) 6.142 The μ2L2 mono-
mers of the pentameric form are linked one to another by
Fc Glycosylation disulfide bonds in the CH4 domains. Two of these monomers
All Igs contain N-linked oligosaccharides, and it is becom- are, on one side, disulfide bonded not to another μ chain but
ing increasingly clear that this glycosylation plays significant to a 15,000 Da-polypeptide, called J chain. J chain is also
roles in Ig structure and function. Though the type and ex- found in polymeric IgA. There may be multiple patterns of
tent of glycosylation varies among isotypes, an N-linked oli- such disulfide bonding, such that different cysteines partici-
gosaccharide on Asn 297 in the CH2 domain is conserved on pate in different monomeric units.143
all mammalian IgGs and homologous portions of IgM, IgD, Application of electron microscopy to polymeric IgM
and IgE. As the average serum IgG contains 2.8 oligosaccha- molecules has suggested that IgM can adopt two different
rides, there is often glycosylation present in the V domain as quaternary arrangements: star and staple.144,145 All of the
well.123 The consensus sequence for the V domain N-linked antigen-binding sites are arrayed in radial fashion, in the
oligosaccharides is not present in the germline, but it can same plane as one another and the Fc regions, in the star

Paul_CH05_final.indd 143
TABLE 5.3 Properties of Immunoglobulin Isotypes
Class or subclass properties IgM IgD IgG1 IgG2 IgG3 IgG4 IgA1 IgA2 IgE
a
Molecular weight of secreted form (kDa) 950(p) 175 150 150 160 150 160(m), 300(d) 160(m), 350(d) 190
Sedimentation coefficient 19S 7S 6.6S 7S 11S 8S
Functional valency 5 or 10 2 2 2 2 2 2 or 4 2 or 4 2
Interheavy disulphide bonds per monomer 1 1 2 4 11 2 2 2 1
Membrane Ig cytoplasmic region 3 3 28 28 28 28 14 14 28
Secreted Ig tailpiece 20 9 2 2 2 2 20 20 2
Other chain J chain (16 kDa) 9 — — — — J chain (16 kDa) secretory —
component (70 kDa)
N-glycosylation sites 5 3 1 1 2 1 2 4 7
O-glycosylation sites 0 7 0 0 0 0 8 0 0
Carbohydrate average (%) 10–12 9–14 2–3 2–3 2–3 2–3 7–11 7–11 12–13
Adult level range (age 16–60) in serum (mg/ml)b 0.25–3.1 0.03–0.4 5–12 2–6 0.5–1 0.2–1 1.4–4.2 0.2–0.5 0.0001–0.0002
Approximate % total Ig in adult serum 10 0.2 45–53 11–15 3–6 1–4 11–14 1–4 0.004
CHAPTER 5
Synthetic rate (mg/kg weight/day) 3.3 0.2 33 33 33 33 19–29 3.3–5.3 0.002

Biological half-life (days) 5–10 2–8 21–24 21–24 7–8 21–24 5–7 4–6 1–5
Transplacental transfer 0 0 ++ + ++ ++ 0 0 0
Complement activation classical pathway (Clq) ++++ 0 +++ + ++++ 0 0 0 0
Complement activation alternative pathway 0 0 0 0 0 0 + 0 0
Reactivity with protein A via Fc 0 0 ++ ++ +/− ++ 0 0 0
Allotypes — — G1m G2m G3m — — A2m Em
Biological properties Primary antibody Mature Placental transfer, secondary antibody Secretory Ig, binds pIgR Allergy and
response, some B-cell for most responses to pathogen, binds parasite

binding to pIgR, marker macrophages and other phagocytic reactivity,
some binding to cells by FcγR binds FcεR
phagocytes
Ig, immunoglobulin; d, dimer; m, monomer; p, pentamer.
a
Light chain molecular weight is 25 kDa.
b
Total = 9.5-21.7 mg/ml.
Compiled from Carayannopoulos and Capra,206 Lefranc and Lefranc,209 Kuby,210 and Janeway et al.211
IMMUNOGLOBULINS: STRUCTURE AND FUNCTION
|
143
9/17/12 5:27 AM
arrangement. In the staple form, the Fab arms bend out of properties (such as serum half-life) among these subclasses.
the plane of the Fc regions. It has been conjectured that the However, there are also crucial functional commonalities,
staple form is utilized in binding simultaneously to two or such as placental passage (see Table 5.3).
more epitopes on multivalent antigens, such as bacterial or IgG antibodies are the hallmark of immunological mem-
viral surfaces. ory in the humoral immune response. In addition to the
A major pathway through which soluble IgM mediates isotype switch from IgM to IgG in a secondary antibody re-
immunity or immunopathology is the activation of the sponse, somatic hypermutation can lead to affinity matura-
classical pathway of complement. On a per molecule basis, tion, a process by which the average affinity of antibody for
relative to other isotypes, IgM is highly active in activating the antigen eliciting the immune response can increase.
the classical pathway and can thereby effectively opsonize IgG antibodies contribute to immunity directly and
bacterial pathogens. In select cases (eg, Neisseria menin- through the activation of complement or FcR-bearing cells.
gitidis), binding of IgM to bacterial surfaces, followed by Important examples of immunity mediated directly through
complement activation, can cause direct lysis of the bacte- antibody binding include neutralization of toxins (eg, diph-
ria through the insertion of the membrane attack complex theria toxin) and viruses (eg, poliovirus). Medically impor-
into the bacterial membrane.146 IgM, like polymeric IgA, can tant examples of IgG-induced complement activation in-
reduce the effective number of colony- or plaque-forming clude immunity to encapsulated bacterial pathogens leading
units for, respectively, bacteria and viruses, through aggluti- either to opsonization and destruction within phagocytes
nation. Significant physical and biological properties of IgM (eg., Streptococcus pneumoniae) or to direct complement-
and the other Ig isotypes are shown in Table 5.3. mediated lysis (eg, Neisseria meningitidis). Activation of
FcR-bearing cells by IgG antibodies has also been implicated
in immunity to pathogens (eg, Cryptococcus neoformans).150
Immunoglobulin D The consensus view is that human IgG1 and IgG3 isotypes
IgD is primarily of interest in its membrane form, as the are effective activators of the classical complement pathway.
soluble form of IgD is found in relatively modest concen- While some older sources state that IgG2 and IgG4 are weak
trations in the blood and other body fluids. The cell surface or nonactivators of the classical complement pathway, more
form of IgD is found along with IgM on all mature, naïve recent evidence suggests that when epitope density is high,
B cells, where it appears capable of transducing activating IgG2 is effective in activating complement.151,152 One possi-
and tolerizing signals.147 As is true for IgM, the membrane ble source for the isotype-related variation in complement-
form of IgD associates noncovalently with Ig-α (CD79a) and activating ability is variation in affinity for C1q (IgG3 > IgG1
Ig-β (CD79b). Simultaneous cell surface expression of two H > IgG2 > IgG4), the portion of the first component in the
chain isotypes expressing the same VH domains and the same classical pathway that physically contacts the CH2 domains
L chains occurs via differential ribonucleic acid splicing.148 of antibodies. However, isotype-associated differences in
IgD exhibits greater sensitivity to proteolytic cleavage complement activation have also been found to occur at
than IgM, which is consistent with a relatively short serum steps of the cascade subsequent to the binding of C1q to an-
half-life of only 2.8 days. The relatively long hinge region is tibody.112,153 For example, in one study of chimeric mono-
a primary target for proteolysis. Relatively modest efforts clonal antibodies engineered to express identical V domains
have historically been devoted to investigating of the func- and representing all four human IgG subclasses, the IgG3
tional roles of IgD antibodies, especially in the secreted antibody fi xed C1q better than the IgG1 antibody, but the
form, in comparison to the other isotypes. However, new IgG1 molecule was more effective in mediating comple-
information suggests that IgD antibodies are produced in ment-dependent cell lysis than the IgG3 molecule.153 Thus,
the upper respiratory mucosa by an unanticipated mecha- it is probably not possible to rank the relative abilities of the
nism of class switch recombination and that these antibod- IgG subclasses to activate complement in a single absolute
ies participate in host defense against pathogens relevant hierarchy.
to this anatomical environment. In addition, secreted IgD The affinities of IgG subclasses for Fc receptors vary from
antibodies can bind to basophils, through a receptor that about 5 × 105 M−1 to about 108 M−1. Recent studies in the
is yet to be identified, and when these antibodies are cross- mouse suggest that the relative contributions of IgG sub-
linked by cognate antigen, the basophils release potent me- classes to various immunopathological processes depend on
diators that influence immune reactivity, inflammation, their relative affinities for the activating versus inhibiting
and pathogen viability.149 isoforms of FcR.154
A remarkable attribute of IgG (for three of the four sub-
classes) is its serum half-life of about 23 days. This property,
Immunoglobulin G attributable to the Fc region and its interaction with the neo-
IgG is the predominant isotype (approximately 70% to 75% natal Fc receptor (FcγRn), has been exploited for therapeutics
of the total Ig) in the blood and extravascular compart- through the genetic fusion of solubilized receptors, (eg, cyto-
ments. The four human IgG subclasses (IgG1, IgG2, IgG3, toxic T-lymphocyte antigen 4) to IgG Fc regions.155
and IgG4) are named in order of their relative serum con- A recently derived insight into the function of IgG4 anti-
centrations, with IgG1 the most prevalent and IgG4 the least. bodies relates to their apparently unique ability to spontane-
There are differences in effector functions (eg, complement ously dissociate into half-molecules and to form antibodies
activation and Fc receptor binding) and other biological composed of different half-molecules with distinct antigen

specificities (ie, bispecific antibodies that cannot cross-link some human mucosal secretions, such as those in the large
antigen and thus have anti-inflammatory activity).156 This intestine and the female genital tract, there is variation in
phenomenon may be of particular relevance to specific im- the relative proportion of IgA1 and IgA2 in different secre-
munotherapy in the setting of allergy.157 tions. The shorter hinge region of IgA2 confers increased
resistance to bacterial proteases that might be encountered
in the mucosal environment. The extended hinge region of
Immunoglobulin A IgA1 is believed to permit molecules of this isotype to ac-
While IgG is the clearly predominant isotype in the blood, commodate variable epitope spacings on multivalent anti-
IgA is the dominant Ig isotype in the mucosal secretions as gens. While the long hinge region of IgA1 molecules might
well as in breast milk and colostrum.158 be expected to confer relatively high susceptibility to pro-
In the blood, 10% to 15% of the Ig is IgA (vs. 65% to teolysis, relative protection against the activity of bacterial
75% IgG). Moreover, IgA has a shorter half-life than IgG in proteases is provided by heavy O-linked glycosylation in the
serum. The predominant form of IgA in human serum or hinge (Fig. 5.12). Nevertheless, the IgA1 hinge is uniquely
plasma is monomeric (ie, α2L2), but there are small quan- susceptible to IgA proteases produced by certain pathogenic
tities of dimers [(α2L2)2] and fewer still trimers and tetra- bacteria.159
mers. Secretory IgA consists of dimers and lesser amounts of Secretory IgA (S-IgA) has been shown to participate
trimers and tetramers associated with one joining (J) chain in immunity against a range of viral, bacterial, and para-
(distinct from the J region in the heavy chain V domain) sitic pathogens at mucosal surfaces. The relative absence of
and one SC chain (see following discussion). The latter is the functional complement and phagocytes in mucosal secre-
extracellular portion of the polymeric Ig receptor (pIgR), tions is consistent with a different mix of mechanisms for
which is expressed by mucosal epithelial cells and transfers mediating immune effects associated with S-IgA versus,
polymeric IgA or IgM from basolateral to apical surfaces, for example, serum IgG. Mechanisms associated with IgA
thereby providing most of the Ig content of the mucosal se- are less dependent on inflammation-producing molecules
cretions. The J chain is disulfide bonded to the tail pieces, or cells, such as inhibition of microbial adherence through
short C-terminal extensions of the CH3 domains, of the two V module-mediated specific binding to microbial adhesins,
IgA monomers of a dimer, while SC forms a disulfide bond agglutination of microbes, blocking of microbial receptors
to a cysteine in one CH2 domain of one monomer. for cell surface carbohydrates with IgA-associated glycans,
The two α H chain C region genes correspond to two IgA and mucus trapping (in which binding of S-IgA to bacteria
subclasses, IgA1 and IgA2. IgA1 is the predominant (> 80%) makes them more adherent to host-generated mucus). There
IgA subclass in the serum. While IgA2 is the major form in is also evidence that polymeric IgA can neutralize viruses
FIG. 5.12. A comparison of an x-ray and neutron-solution–scattering theoretical model (human immunoglobulin (Ig)A1) and
x-ray crystal (murine IgG1 and IgG2a) structures. Light chains (yellow), heavy chains (red and dark blue), and glycosylation
(light blue) are illustrated. The extended length of IgA1 over that of IgG can be seen along with extensive glycosylation that
characterizes this isotype. From Boehm et al.207 with permission.

in some circumstances by interfering with steps postattach- present at the lowest concentration of any of the Ig isotypes
ment, such as internalization. In some cases, posttrans- (with roughly five orders of magnitude less IgE than IgG)
lational modifications of viral surface molecules, related and has the shortest half-life. The unimpressive quantitative
to proteolytic events associated with epithelial cell transit, representation of IgE in the blood is related to the high af-
lead to IgA antibodies expressing protection-related anti- finity of IgE antibodies for Fc εRI, often referred to as the
gen specificities that have no parallel in the IgG pool.160,161 high-affinity Fc receptor for IgE. Fc εRI is expressed on mast
Some of the contributions of IgA to immunity are mediated cells, basophils, Langerhans cells, and eosinophils. Due to
through binding to FcαRI (CD89) on human neutrophils, the high affinity of Fc εRI for its IgE ligand, mast cells and
monocytes/macrophages, and eosinophils. For example, basophils are covered with relatively long-lived Fc εRI-IgE
cells bearing FcαRI on their plasma membranes can phago- complexes.
cytose IgA-antigen complexes. FcαRI binds to the IgA Fc re- The interaction between IgE and Fc εRI has an affinity
gion between the Cα2 and the Cα3 domains.162 Amino acid of ∼1010 L/M. It primarily involves contacts between Fc εRI
residues in the IgA Fc region critically involved in the inter- and amino acids in the CH3 domains, with some contribu-
action with CD89 are indicated in Figure 5.13. tions from amino acids in the CH2 domains. Although each
The ability of IgA to activate complement is controversial. IgE potentially has two sites for interacting with the FcεRI,
At present, the preponderance of evidence suggests that IgA the stoichiometry has been shown to be 1:1. Furthermore,
does not activate the classical complement pathway and only it has been suggested that IgE binds to Fc εRI in a kinked
weakly, and under some pathophysiological circumstances, conformation.164
activates the alternative complement pathway. However, Upon ligation with bivalent or multivalent antigens spe-
there is evidence suggesting that, in vitro at least, polymeric, cifically recognized by the bound IgE molecules, the FcεRI
but not monomeric, IgA can activate the complement path- molecules transduce signals that activate the mast cells or ba-
way dependent on mannose-binding lectin.163 sophils to secrete potent mediators of inflammation, such as
Additional properties of the polymeric forms of IgM and histamine. These mediators are responsible for the symptoms
IgA are considered in the following section. associated with asthma, allergic rhinitis, and anaphylaxis.
There is a second receptor for IgE. Fc εRII (CD23), a
type II membrane protein, is expressed on monocytes/
Immunoglobulin E macrophages, B-lymphocytes, natural killer cells, follicu-
IgE is best known for its association with hypersensitivity lar dendritic cells, Langerhans cells, eosinophils, activated
reactions and allergy, but this isotype is also of interest in epithelial cells, and platelets. It binds monomeric IgE with
the context of immunity to parasites. In the blood, IgE is an affinity of ∼107 L/M, roughly three orders of magnitude
lower than the affinity of Fc εRI for IgE. Functional conse-
quences of FcεRII-IgE interaction on macrophages include
secretion of mediators of immediate hypersensitivity as well
as cytokines and chemokines. There is also evidence sug-
gesting that the Fc εRII-IgE interaction can contribute to an-
tigen capture and presentation to both B and T cells.
Fc Receptor Immunoglobulin Interactions

There are four receptors that bind the Fc regions of IgG mol-
ecules in humans and five such receptors in mice. Three of
the human receptors are expressed primarily on hemato-
poietic cells directly involved in immune responses: FcγRI,
FcγRII, and FcγRIII. Among the FcγR involved in antibody
effector functions, there is important variation in affinity
for IgG molecules. FcγRI binds IgG with relatively high af-
finity, permitting the binding of monomeric IgG. In con-
trast, FcγRII and FcγRIII bind to IgG with relatively low
affinity. Consequently, these latter two receptors do not
bind significant quantities of monomeric IgG but preferen-
tially interact with IgG that has been effectively aggregated
through interaction with bivalent or multivalent antigens
(ie, immune complexes). Key amino acid residues involved
in the binding FcγRs and Fc εR1 with their corresponding Ig
can be seen in Figure 5.14. Allotypic variations in FcγRIIA
FIG. 5.13. Illustration of the Residues Essential for the Binding of immu-
noglobulin (Ig)A to Fc`R (CD89). Residues represent mutations made in
can also influence affinity for IgG ligands and subsequent
IgA CH regions as mapped on an Fcy fragment. Of the residues mutated, effector function.165
L465 and L266 were found to be important for binding to CD89. From There are different isoforms of FcγRII and FcγRIII. Of
Carayannopoulos et al.162 with permission. particular functional relevance, FcγRIIA is activating,

FIG. 5.14. Certain residues are conserved between Fcγ Rs and FcεRl as well as between immunoglobulin (Ig)G and IgE
that facilitate binding. Two sites participate: site 1 in (A) and site 2 in (B). Heavy lines indicate the highest number of
contacts and dashed lines indicate the least. Of considerable note are residues W87 and W110 in site 2 of the receptors
and P426 in the Ig that form a core “praline sandwich” in the interaction between Ig and receptor. From Garman et al.208
with permission.
whereas FcγRIIB is inhibiting. In mouse models, deficiency IgG across the placenta. It also plays a crucial role in protecting
of FcγRIIB can be associated with an autoimmune syn- IgG from proteolytic degradation, thereby prolonging serum
drome similar to human lupus.166 Studies in mice have also half-life. While FcγRI, FcγRII, and FcγRIII are members of
suggested that the effectiveness of antibodies of the various the immunoglobulin superfamily, FcγRn is structurally sim-
murine IgG subclasses in mediating FcγR-dependent effec- ilar to major histocompatibility complex class I molecules,
tor functions are correlated with the ratio of affi nities of an- including noncovalent association with β2-microglobulin.
tibodies of those subclasses for activating versus inhibiting The interaction between FcγRn and IgG involves amino acid
FcγR.154 residues in the CH2-CH3 interface167 and is pH-sensitive. This
FcγRIIIA is expressed on natural killer cells, whereas latter property is consistent with the ability of FcγRn to bind
FcγRIIIB is expressed on neutrophils. The former receptor IgG in acidic vesicular compartments and then release it into
is attached to the membrane by a standard transmembrane the neutral-pH environment of the blood. A crystallographic
polypeptide, whereas the latter is attached via a glycophos- structure of the rat FcγRn in a 1:1 complex with a heterodi-
pholipid tail. meric Fc (containing only one FcγRn-binding site) reveals
The fourth human receptor, the neonatal FcγR, or FcγRn that there are conformational changes in the Fc on binding
(sometimes referred to as the Brambell receptor), transports to FcγRn.168 The investigators also identified three titratable

salt bridges that confer pH-dependent binding of the IgG Fc mucosal epithelial cells.175,176 Though IgM can undergo tran-
to the FcγRn. scytosis to the mucosal secretions, its principal action is in
There are other FcR that interact with the non-IgG iso- the serum.
types: IgM, IgD, IgA, and IgE. The FcR for IgA and IgE The ability of IgA and IgM to multimerize is due to a tail-
are covered in the respective sections devoted to the cor- piece, an additional C-terminal segment of 18 amino acids
responding isotypes. In mice, an additional Fcα /μR binds in the secreted forms of the μ and α heavy chains. Tailpieces
both IgA and IgM, whereas in human these receptors are of both IgM and IgA contain a penultimate cysteine (residue
distinct.169,170 Functional attributes of these receptors and 575 in IgM and 495 in IgA) that forms two different disulfide
the FcδR are continuing to be studied. bonds important for multimer formation. In an Ig mono-
There are functional as well as structural parallels among meric unit containing two identical H chains, one cysteine
FcR for IgG and non-IgG isotypes. Some of the features of residue forms intermonomeric subunit bonds, whereas the
Fc-FcR interaction that are conserved across isotypes are il- remaining cysteine residue on the other heavy chain bonds
lustrated in Figure 5.14. to a cysteine on the J chain.177–180
Another cysteine residue, Cys414, also forms intermo-
nomeric subunit disulfide bonds in IgM, and this bond
Transmembrane and Cytoplasmic Domains is important in hexamer formation.181 Besides disulfide
Igs are expressed in both membrane and secreted forms. In con- bonds, the highly conserved glycan linked to Asn563 in IgM
trast to the secreted form, the membrane Ig contains a trans- (and the homologous region in IgA) is also important for
membrane and a cytoplasmic domain. The transmembrane multimerization.182
domain is a typical single-pass polypeptide segment consisting Domain-swapping experiments demonstrate that the
of 26 hydrophobic residues, extending from the C-terminus tailpiece regulates multimerization in the context of the spe-
end of the C-terminal CH domain and which forms an alpha- cific H chain. While addition of the α tailpiece to IgM has
helix followed by a variable number of basic amino acids. The little effect on IgM polymerization, the introduction of the μ
cytoplasmic portion of the Ig H chain ranges in length from 3 tailpiece to IgA leads to higher-order IgA polymers.183 Based
amino acids for IgG to 30 amino acids for IgE. on this finding, it has been proposed that IgM polymeriza-
Both membrane expression of Ig and the integrity of the tion is more efficient than IgA polymerization.
cytoplasmic domain are important in antibody function. As
the cytoplasmic domain is rather short, membrane-bound
IgM is not thought to “signal” directly, but through the as- The J Chain
sociated Igα and Igβ molecules.171 However, disruption of The J chain, an evolutionarily conserved 137 amino acid
either membrane expression of IgG1 in mice or the cytoplas- polypeptide produced by B-lymphocytes, functions to regu-
mic tail results in the failure to generate an effective IgG1 late multimer formation and to promote linkage of multi-
response and IgG1 memory.172 Mice lacking IgE membrane meric Ig to pIgR on epithelial cells. The J chain consists of
expression exhibit significant impairment in IgE responses a single domain in a beta barrel conformation and does not
and have extremely low levels of secreted IgE.173 For those show sequence similarity to Ig domains.184 It contains eight
Igs with longer membrane:cytoplasmic tails, membrane Ig cysteine residues that participate in disulfide bonds with
affects signaling beyond its association with Igα and Igβ. two tailpiece cysteines, as described previously, as well as
Specific residues in the transmembrane domain have been function to stabilize its own structure through intramolecu-
identified that are crucial for signal transduction while hav- lar bonds.183,185 The J chain influences the polymerization of
ing no effect on the association with Igα and Igβ.174 the multimers, as in the absence of J chain, IgA forms fewer
dimers and IgM forms fewer pentamers.186
The J chain exists in all polymeric forms of IgA and is
HIGHER ORDER STRUCTURE important in IgA polymerization and secretion across the
Many of the biological functions of IgA and IgM are depen- mucosa. J chain is not required for IgM polymers but is re-
dent on their ability to form multimeric structures. This quired for external secretion. While IgM pentamers contain
section will discuss the role of multimeric Ig in immune J chain, hexamers almost always lack it. The makeup of IgM
function. is biologically significant because IgM hexamers have about
20-fold greater complement-activating activity than IgM
pentamers. The presence of increased levels of hexameric
Dimers, Pentamers, and Hexamers IgM has been postulated to play a role in the pathogenesis
The majority of multimeric IgA exists as dimers and, less of Waldenström macroglobulinemia and cold agglutinin
commonly, trimers and tetramers, while IgM forms pen- disease.187
tamers and occasionally hexamers. The polymeric struc-
tures of these antibodies enhances their functional affi nity
(avidity) for antigen, is essential for their active transport Immunoglobulin Transport
(both IgA and IgM) across epithelial cells to mucosal secre- Transport of dimeric IgA and pentameric IgM to the mu-
tions, and in the case of IgM, enhances the activation of the cosal secretions occurs after binding to pIgR that is present
classical pathway of complement. Once multimerized, IgA on the basolateral surface of the lining epithelial cells. The
or IgM in a complex with J chain can bind to pIgR and cross J chain is essential for the secretion of IgA and IgM and,

as described previously, influences the polymeric structure propria and excrete them to the mucosal lumen (where they
of the Ig.175 pIgR is a transmembrane receptor synthesized can be removed from the body) by the same pIgR-mediated
by mucosal epithelial cells that contains seven domains in- transcytosis process.199,200 Some pathogens can exploit the
cluding five extracellular V-like domains, a transmembrane pIgR-mediated transcytosis process in reverse to penetrate
domain, and a cytoplasmic domain.188 Once bound to pIgR, the mucosa. For example, the pneumococcal adhesin, CbpA,
polymeric Ig is endocytosed and transported to the api- can bind pIgR at the epithelial apical surface, leading to bac-
cal surface of the cell. pIgR is then proteolytically cleaved terial penetration of the mucosa.201
between the fifth and sixth domains to release a complex
(termed secretory Ig) containing the H, L, and J chains, and
the SC, which represents the cleaved extracellular portion CONCLUSION
of pIgR.189,190 As pIgR undergoes constitutive transcytosis Igs are extremely versatile molecules that can carry out many
in the absence of polymeric Ig, free SC is also released into biological activities at the same time. The need to be able to
the mucosal secretions. SC has several biological functions, recognize unique antigen structures prior to any previous
including protecting the Ig from degradation by proteases exposure coupled with the need to maintain host cell recep-
and binding bacterial antigens such as the Clostridium dif- tor or complement recognition properties presents a truly
ficile toxin A.191,192 SC also functions to localize sIgA to the unique challenge for the system. As has been described, the
mucus layer to help protect against invasion by pathogens.193 system incorporates diversity within specific constraints.
Results from pIgR-null mice demonstrate that alternate The precise biological niches may differ, but the overall de-
pathways exist to transport polymeric Ig to the mucosal se- sign for these molecules is the same.
cretions as some secretory Ig still crosses the epithelial cells The flexibility and biologic properties of Igs have made
in the absence of pIgR.194 Results from pIgR-null mice also them a major focus of molecular engineering. Igs are being
demonstrate the importance of high levels of secretory anti- used as therapeutic agents, as well as for biotechnology ap-
bodies as these mice are more susceptible to mucosal infec- plications. These opportunities have led to a resurgence of
tions with pathogens such as Salmonella typhimurium and interest in the structure–function aspects of antibodies as
Streptococcus pneumoniae.195,196 we approach “designer antibodies.” Both the variable and
The epithelial transcytosis of polymeric Ig has several constant portions of these molecules are current substrates
biological implications. First by delivering the Ig to the mu- for engineering purposes, offering the potential for altering
cosal surface, it enables antibodies to bind to pathogenic both receptor and effector function. The study of antibodies
agents and prevent them from penetrating the mucosa, a began with the need to understand how sera could neutral-
process termed immune exclusion. Second, transcytosing ize toxins. It is likely that antibodies will continue to be a
antibody can neutralize viruses intracellularly.197,198 Finally, major focus for those who seek to take fundamental prin-
polymeric Ig can bind to antigens in the mucosal lamina ciples of protein chemistry to the bedside.

CHAPTER
6 Immunoglobulins: Molecular Genetics
Edward E. Max • Sebastian Fugmann
INTRODUCTION OVERVIEW OF IMMUNOGLOBULIN

To respond to a foreign molecule (antigen) on an invading GENE ASSEMBLY
pathogen, the “humoral” immune system generates antibod- In the 1960s, investigators determined the amino acid se-
ies, or immunoglobulins (Igs), that can bind specifically to the quences of Igs secreted by several mouse myelomas (clonal
offending antigen. Each antibody molecule is composed of two tumors of B-lymphocytes that secrete a single pure species
identical light (L) chains and two identical heavy (H) chains, all of Ig). The N-terminal domains of the L and H chains—
linked by disulfide bonds to form a symmetric Y-shaped tetra- each roughly 100 amino acids—were highly diverse between
mer. The ability of the immune system to generate an antigen- different myeloma proteins and were designated variable (V)
specific antibody depends on the fact that, before exposure to regions. In contrast, sequences of the remaining domains of
antigen, millions of naïve resting B cells circulate in the indi- the proteins were essentially identical for every myeloma Ig
vidual, each cell displaying on its membrane several thousand of a given class (and so they were designated constant [C] re-
identical copies of a single unique species of antibody; these gion domains). The advent of recombinant DNA technology
serve as B-cell receptors (BCRs) for that lymphocyte. Only a allowed comparisons of V region genes expressed in differ-
tiny fraction of the B cells express a BCR capable of binding to ent myelomas with the corresponding sequences in nonlym-
any particular antigen. When these B cells bind their antigen, phoid DNA (commonly referred to as “germline” DNA). It
they become activated to proliferate and mature into antibody- was found that each myeloma V gene is composed of several
secreting plasma cells, which manufacture large amounts segments that are separated in germline DNA; these germ-
of antibody specific for the activating antigen. To be able to line segments must undergo one or more DNA recombina-
generate antibodies against a universe of diverse pathogens, tion events to assemble a complete V region.1 For example,
this “clonal selection” mechanism for specific antibody secre- each complete Vκ gene from a myeloma or B-lymphocyte
tion requires an enormous diversity of Ig species expressed on encodes roughly 108 amino acids and is assembled by link-
naïve B cells prior to antigen exposure. Indeed, in the 1960s ing one of about 40 germline Vκ segments (encoding amino
the number of different antibody sequences in the repertoire acids 1 through 95) to one (of five) “joining” or Jκ segments
of typical mouse was estimated in the millions. To encode this encoding residues 96 to 108. Similarly, a complete Vλ gene
many sequences seemed to require an unreasonably high per- is assembled from one germline Vλ segment and one Jλ
centage of the mammalian genome (now estimated to contain segment. H chains are assembled from three segments; a di-
only about 30,000 genes). Understanding the genetic source of versity (D) segment is interposed between V H and JH. In de-
Ig diversity—Ig gene assembly—was the first major challenge veloping B cells, the germline gene segments are assembled
and achievement of the molecular biologic investigations of into functional V exons by a process named V(D)J recombi-
antibody genes, and this will be discussed first in this chapter. nation (Fig. 6.1).
A week or so after antigen administration, the B-cell re- V(D)J recombination is a “cut and paste” process in which
sponse changes in two ways that generally improve the protec- the DNA between two recombining V, D, or J gene segments
tive functions of antibodies. B cells initially express antibodies is excised from the chromosome, and the two remaining
of the IgM isotype, but cells that migrate into germinal centers DNA segments are joined together to reseal the DNA break.
receive T-cell-derived stimuli that can induce them to switch The two principal proteins executing the “cut” phase of this
to production of IgG, IgA, or IgE without changing their anti- process are encoded by the recombination activating genes
gen specificity; this switch results from a deoxyribonucleic acid (RAG)1 and 2. These proteins recognize unique sequences,
(DNA) recombination event known as class switch recombi- known as recombination signal sequences (RSSs), that flank
nation (CSR). In addition, over the course of an immune re- and mark each eligible V, D, and J gene segment (RSSs are
sponse, the affinity of antibody for antigen gradually improves described further in the following). After the RAG proteins
as a result of somatic hypermutation (SHM) of antibody genes, cut the DNA, the subsequent “joining” of the gene segments
coupled to selection for B cells expressing high-affinity anti- relies largely on ubiquitous DNA repair factors.
bodies. CSR and SHM are discussed later in this chapter.
In this chapter, well-established facts about Ig genes are
summarized concisely, while areas currently under investi- How Recombination Contributes to Diversity
gation are considered in more detail, with particular atten- V(D)J recombination contributes in several distinct ways to
tion to topics expected to interest immunologists. the diversity of antigen-binding specificities. First, there is
150

CHAPTER 6 IMMUNOGLOBULINS: MOLECULAR GENETICS | 151
FIG. 6.1. Variable Assembly Recombination. A: In the κ locus, a single recombination event joins a germline
Vκ region with one of the Jκ segments. B: In the immunoglobulin H locus, an initial recombination joins a
diversity segment to a JH segment. A second recombination completes the variable assembly by joining a VH
to DJH.
combinatorial diversity, as each Ig locus contains multiple crystallography reveal that the V L-JL junction and the V H-
V, D (in case of the IgH locus), and J segments that can be DH-JH junction each encode one of the three “complemen-
combined in many ways. The total number of theoretically tarity determining region” loops of L or H chain that can
possible combinations of V H, DH, JH, V L, and JL, is the mul- contact antigen; thus, this junctional diversity is directly
tiplication product of the numbers of possible H chains— functionally relevant for diversifying antigen binding.
about 40 (V H) × 27 (DH) × 6 (JH) or 6480 combinations in The imprecision of V(D)J recombination increases Ig di-
humans—times the number of possible L chain combina- versity, but at a cost. Because the precise boundaries between
tions (about 290), or almost 2 million. This repertoire is V, D, and J result from independent stochastic events, only
vastly larger than could be achieved by devoting the same about one-third of all recombination events maintain the
total lengths of DNA sequence to preassembled variable correct reading frame through the J segments. Gene rear-
region exons. Second, there is junctional diversity gener- rangements leading to functional Ig genes are often referred
ated by flexibility in the position of joining between gene to as “productive,” while out-of-frame rearrangements are
segments. This was initially recognized by comparisons of labeled “nonproductive.”
nucleotide sequences of various myeloma Vκ genes to their
germline Vκ and Jκ precursors. As shown in Figure 6.2A,
these comparisons revealed that the crossover point be- Function of Recombination Signal Sequences
tween sequence derived from a germline Vκ region and a Analysis of DNA sequences flanking the germline V, D, and
Jκ region could vary in different myelomas, increasing the J gene segments revealed highly similar sequence motifs that
diversity of amino acids around codons 95 and 96. H chain have subsequently been shown to define targets for V(D)J
VDJ exons exhibit this flexibility at both V-D and D-J junc- recombination: the RSSs, which serve as the recognition se-
tions, yielding striking variation in the lengths D region– quences for the V(D)J recombinase proteins RAG1 and RAG2,
derived segments, from zero to about 14 amino acids. And as mentioned previously. Notably, RSSs lie adjacent to L- and
additional junctional diversity is produced by the addition H-chain Ig gene segments and to T-cell–receptor (TCR)
of nucleotides not present in any germline elements: “N” gene elements throughout phylogeny. RSSs consist of a con-
and “P” nucleotides, discussed below. Importantly, the served seven base pairs (bps) long “heptamer” (consensus:
three-dimensional structures of Igs established by X-ray CACAGTG) and a nine bp long “nonamer” sequences

A B
FIG. 6.2. Vk-Jk Recombination at Single Base Resolution. A: The sequence of the recombined MOPC41 κ gene around the VJ junc-
tion is shown (center) with the sequences of the two germline precursors (Vκ41 and Jκ1) shown above and below. The germline
origins of the recombined gene are indicated by the vertical lines and the shading of the V-derived sequence. B: The consequences
of joining the same germline sequences (from part A) at four different positions are shown. Of the four alternative recombination
products illustrated, the top one is that actually found in MOPC41. The second example has a single nucleotide difference but no
change in encoded amino acid sequence. The third and fourth alternatives yield Arg or Pro at position 96; both of these amino acids
have been found at this position in sequenced mouse κ chains.
(consensus: ACAAAAACC) that are separated by less well- heptamer and nonamer are the major determinants of RSS
conserved spacers of either approximately 12 or 23 bp in function necessary for V(D)J recombination, increasing evi-
length (Fig. 6.3). Based on the spacer lengths, the two classes dence suggests that spacer sequences can modulate recom-
of RSSs are referred to as 12-RSSs and 23-RSSs, respectively. bination efficiencies of compatible gene segments (e.g., they
(Note that some laboratories use the term recombination sig- affect the non-random usage of human Vκ elements2).
nal instead of RSS in their publications.)
Recombination occurs almost exclusively between coding
sequences associated with RSSs of different spacer lengths, a THE THREE IMMUNOGLOBULIN GENE LOCI
requirement referred to as the “12/23-rule” (i.e., the recom- To understand the contribution of the germline V, D, J ele-
bination between two 12-RSSs [or two 23-RSSs] is “forbid- ment repertoire to Ig diversity, several laboratories under-
den” and does not occur in vivo). Within each gene locus, all took cloning and sequence analysis of individual V region
gene segments of one class (e.g., all Vs in the Igκ locus) carry genes from the IgH, Ig κ, and Igλ loci of human and mouse.
RSSs with the same spacer length. Thus the 12/23 rule drives More recently, the complete sequences of all human and
appropriate recombination events leading to functional VJ mouse Ig loci have been determined as part of the genome
and VDJ products, and prevents futile recombination events, sequencing projects for these two species (available online
such as between two V or two J gene segments. While the at www.ncbi.nlm.nih.gov, though annotation that describes
FIG. 6.3. Conserved Elements Flank Germline Variable (V), Diversity (D), and Joining (J) Region Genes. Conserved
heptamer and nonamer recombination signal sequences (RSSs) lie adjacent to V, D, and J coding sequences and are
important for targeting V(D)J recombination. The heptamer and nonamer elements are separated by spacer regions of
about 12 basepairs (bp) (thin lines) or 23 bp (thick lines). Depending on the locus, V regions may be flanked by 12 bp or
23 bp RSS; and similarly for J regions. But one of each type of element must be present for recombination to occur, a
requirement that prevents futile recombination events (e.g., J to J).

Overview of the Number of Variable, Diversity, and Joining Segments in

TABLE 6.1
Each of the Three Immunoglobulin Loci in Humans and Mice
Locus Species V D J
Functional Pseudogenes Functional Pseudogenes Functional Pseudogenes
IgH Mouse 110 85 10 4
Human 40 83 24 3 6 3
Igκ Mouse 95 45 4 1
Human 46 87 5
Igλ Mouse 3 3 1
Human 36 56 4–5 2–3
D, diversity; Ig, immunoglobulin; J, joining; V, variability.
Pseudogenes are recognized based on sequence defects that would preclude function (premature stop codons, defective recombination signal sequences,
defective splice sites). The numbers in this table are approximate, owing to variation between mouse strains and between individual humans.
function and refers to earlier literature is incomplete). It is segments (depending on mouse strain) plus additional V H
important to point out that Ig gene loci are not identical be- pseudogene segments (Table 6.1). All V H elements are in
tween individuals (humans) or between individual strains the same transcriptional orientation as the D, J H, and CH
of inbred mice. Several Internet resources are devoted to regions.3 The V H segments are classified into 16 distinct
providing convenient updated access to Ig germline gene families based on sequence similarity; V H elements within
sequences. The international ImMunoGeneTics database a family show more than 80% nucleotide sequence iden-
(http://imgt.org) includes a database for Ig and TCR genes tity. Elements of individual V H families are largely clustered
from a variety of species, and includes maps, sequences, lists together, though some interdigitation occurs (Fig. 6.4).
of chromosomal translocations, and multiple helpful links. The V H families can be grouped into three “clans” based
IgBLAST (www.ncbi.nlm.nih.gov/igblast/) is a service of the on sequence conservation primarily of their framework
National Center for Biotechnology Information and allows a regions (framework region 1, codons 6 to 24, and frame-
submitted sequence to be searched against known annotated work region 3, codons 67 to 85, respectively), which form
germline V, D, and J sequences. the more conserved structural backbone of the Ig variable
region. Importantly, these clans are conserved between
The Murine Immunoglobulin H Germline Variable, man, mouse, and frog, suggesting that their emergence in
the repertoire preceded the amphibian-reptile divergence.4
Diversity, and Joining Gene Segments
VH Segments Diversity and JH
The murine V H region extends over about 2.5 megabases About 50 kb downstream of the most 3′ V H element resides
on chromosome 12 and includes roughly 100 functional the murine D cluster spanning about 80 kb, depending on
FIG. 6.4. Maps of the Murine and Human VH Loci. The 15 known murine VH gene families are shown in their approximate map po-
sitions. Each rectangle represents a cluster of VH genes of the indicated family; the clan identification of the VH families is indi-
cated by the color of the rectangle: black for clan I, gray for clan II, and white for clan III. Although some interdigitation is shown
by overlapping families (e.g., the Q52 and 7183 families), the murine VH families are largely clustered. In contrast, all human
VH genes (vertical lines) of a prototypic haplotype are shown in the right panel; extensive interdigitation of families is apparent.

the mouse strain (see Table 6.1). Each D segment is flanked where the numbers in parentheses represent the number of
by 12 RSSs on both sides, so that the 12/23 rule ensures that nucleotides in each segment. As an interesting contrast, the
all assembled V genes carry a D element between their V and hinge region of the α gene is encoded contiguously with the
J segments (which are both flanked by 23 RSSs that prevent CH2 domain with no intervening intron, while the unusually
direct V to J rearrangements). long human γ 3 hinge is encoded by three or four hinge exons.
The murine D elements are classified into four fami-
lies: DSP2, DFL16, DST4, and DQ52. Although D regions Genomic Organization of the CH Region
could theoretically contribute to Ig diversity by being read Each B-lymphocyte initially produces IgM by expressing an
in all three frames, the mouse has evolved mechanisms that assembled variable region linked to C μ, but may use CSR
strongly favor one of them.5 Four functional germline JH se- (discussed later in this chapter) to replace Cμ with one of
quences reside about 0.7 kb downstream of the most 3′ D the several CH regions lying downstream, thereby allowing
region, DQ52. expression of IgG, IgA, or IgE (Fig. 6.5A). Eight murine CH
genes span about 200 kb of DNA on chromosome 12; these
genes were linked by contiguous clones in 1982. 8 Several
The Human Immunoglobulin H Germline Variable, γ pseudogenes lie within the clustered γ functional genes 9
Diversity, and Joining Gene Segments (Fig. 6.5B). The coding sequences of all CH genes are ori-
VH Segments ented in the same direction.
The human V H locus spans 1.1 Mb at the telomeric end The human CH genes were similarly cloned, and then
of chromosome 14 (14q32.33) (see Table 6.1). The human eventually completely linked by the Human Genome
germline V H segments—numbering roughly 40 to 45—fall Project. The human IgH locus contains a large duplication,
into seven families that, in contrast to the family clusters with two copies of a γ–γ–ε– α unit separated by a γ pseu-
characteristic of the murine locus, are extensively interdig- dogene (see Fig. 6.5B). One of the duplicated ε sequences
itated (see Fig. 6.4). Some human V H sequences are poly- is also a pseudogene, and a third closely homologous ε-
morphic owing to V H insertions or deletions in different related sequence—a “processed” pseudogene—is present on
allelic chromosomes. Twenty-four additional germline V H chromosome 9.
sequences have been mapped to chromosome 15 and 16 and The IgH locus has also been examined in several other
represent nonfunctional “orphans” that were apparently du- species besides mouse and human, and several notable dif-
plicated from the IgH locus on chromosome 14. 6 ferences have been observed. Rabbits, for example, have 13
Cα sequences and only a single Cγ gene10 ; this unusual ex-
Diversity and JH Regions pansion of genes contributing to mucosal immunity may be
Twenty-six human D elements are located in an ∼40 kb re- related to the peculiar habit of coprophagy in these animals.
gion about 20 kb downstream of VH6, the most 3′ of the VH In contrast to the multiplicity of rabbit Cα genes, pigs have
genes.7 This D cluster is comprised of four tandem duplica- only one Cα gene and eight Cγ genes. Camels are unusual in
tions of a 9.5 kb segment containing a representative of each having H chains that function in the absence of L chains.11
of six D families. The twenty-seventh D element—DHQ52—
is the only one showing sequence similarity to a mouse seg- Membrane versus Secreted Immunoglobulin
ment (DQ52) and shares a homologous location just 5′ to Igs are found either as secreted molecules in the serum or
JH1. In contrast to mice, humans use all reading frames of D as membrane-bound receptors. The membrane-bound
elements.7 One reading frame encodes primarily hydrophilic μ chains contain a C-terminal hydrophobic transmem-
residues, one encodes hydrophobic residues, and one includes brane domain consisting of 26 uncharged hydrophobic
frequent stop codons. Some D elements contain stop codons amino acids encoded by additional membrane exons, and
that can be removed by nuclease trimming during VDJ as- these residues anchor the protein in the cell membrane
sembly. As in mice, the human JH cluster is immediately lipid bilayer. The membrane (μm) and secreted (μs) forms
downstream of DHQ52. are derived from the same gene by alternative splicing
(Fig. 6.6). The same general gene structure has been found
Heavy Chain Constant Regions for other CH genes, suggesting that differential splicing ac-
counts for the two forms of all Ig isotypes.
Murine and human genomic clones containing C region
Early B cells make roughly similar quantities of both μm
H-chain (CH) genes include separate exons encoding the
and μs, whereas maturation to the plasma cell stage is associ-
∼100 to 110 amino acid Ig domains. These domains were
ated with strong predominance of μs production, facilitating
independently identified by internal homologies of amino
high-level secretion of circulating Ig. The balance between
acid sequences and by three-dimensional structural analysis
the two ribonucleic acid (RNA) splice forms of μ has been
(X-ray crystallography). The exons are separated from each
interpreted as a competition between splicing of the CH4
other by introns of roughly 0.1 to 0.3 kb. Thus, for example,
and M1 exons versus the cleavage/polyadenylation at the up-
the mouse γ 2b protein has three major domains (CH1, CH2,
stream μs poly(A) site. These processes are mutually exclusive
and CH3) with a small hinge domain between CH1 and
because CH4-M1 splice removes the μs poly(A) site, while
CH2. The gene structure may be summarized as follows:
cleavage at the μs poly(A) site removes the membrane exons.
CH1 - intron - hinge - intron - CH2 - intron - CH3 Cis-regulatory elements (and corresponding transacting
(292) (314) (64) (106) (328) (119) (322) RNA binding proteins) control the balance between these

FIG. 6.5. Deletional Isotype Switch Recombination. A: The expression of “downstream” heavy chain
genes is accomplished by a recombination event that replaces the Cμ gene with the appropriate
heavy chain constant gene (Cε is shown as an example), deleting the deoxyribonucleic acid between
the recombination breakpoints. B: The murine and human heavy chain constant region genes are dia-
grammed with the approximate intergene distance indicated below (in kb); various literature values
for these distances differ somewhat, possibly due to allelic polymorphisms. The human locus shows
a large duplication of γ γ εα sequences.
processes. They include a GU-rich element downstream for lymphocyte activation following contact with antigen.
of the μs poly(A) site,12 the polyadenylation factor cleav- Transduction is mediated by an associated protein dimer
age stimulator factor 64,13 and the U1A protein.14 These composed of the BCR components Igα and Igβ (CD79a and
factors likely function downstream of B-lymphocyte–in- CD79b) whose cytoplasmic domains contain immunorecep-
duced maturation protein-1 (BLIMP-1) whose expression tor tyrosine-based activation motifs similar to those found
in Ig-secreting plasma cells was also found to be critical for in the CD3 chains mediating TCR signaling. Additional
μs poly(A) site utilization.15 Cis-acting sequences affecting signaling is mediated by conserved tyrosines in the cyto-
the ratio of alternative splice forms have been described for plasmic tails of the IgG and IgE H chains, which serve as
other isotypes besides Cμ, particularly Cα .16 a phosphorylation-dependent docking sites for the signal-
Membrane Ig serves as the antigen-recognition coming adapter Grb2.17 Binding of Grb2 enhances BCR signaling
ponent of the BCR that is critical for initiating the signal and subsequent B-cell proliferation.
FIG. 6.6. Two Ribonucleic Acids (RNAs) Generated

from the m Gene by Alternative Processing. The top line
illustrates the exons of the μ gene (black rectangles) in
an expressed, rearranged μ gene. A primary transcript
including all the exons present in the deoxyribonucleic
acid can be processed as shown to yield either μs RNA
(containing a C-terminal “secreted” sequence) or μm
RNA (containing the two membrane exons).

Kappa Light Chain Genes Jk and Ck Elements

Murine Germline Vk Locus In comparison to the H chain genes, the organization of
The murine Vκ locus spans about 3.2 mb on chromosome the C region segments in the κ locus is relatively simple (see
618 and contains 20 Vκ families, some of which are shared Table 6.1). A single Cκ gene with a single exon and with no
by human and mouse (see Table 6.1). Vκ sequences within reported alternative splice products is found in both mouse
a single family are largely clustered together. Some Vκ ele- and human. While all five Jκ elements are functional in hu-
ments lie in the opposite orientation to that of the Jκ and Cκ mans, the third J element in mice has not been observed in
elements, and these Vκ segments undergo VJ recombination functional κ L chains.
by an inversion rather than deletion (Fig. 6.7). A few Vκ se- Apart from the typical Vκ-Jκ rearrangements, an addi-
quences have been localized to chromosome 16 and 19 and tional recombination event occurs uniquely in the κ locus.
are considered orphan genes. The event is mediated by V(D)J recombination utilizing a
23-RSS element—designated Recombining Sequence in the
Human Germline Vk Locus mouse 20 and Kappa Deleting Element in the human21—that
The human Vκ locus (see Table 6.1) lies on the short arm is positioned in an intergenic region downstream of Cκ ; the
of chromosome 2 (2p11-2) spanning ∼2 mb of DNA.19 The recombination results in the deletion of the Cκ exon. Hence,
locus includes a large inverted duplication, so that most Vκ Cκ fragments are undetectable on Southern blots of DNA
sequences exist in pairs with one copy lying in the cluster from λ-expressing human lymphoid cells, 22 as in most B
proximal to Jκ (and in the same orientation) and a second cells the Cκ genes are apparently deleted from both chro-
copy (inverted) in the distal cluster. The average sequence mosomes before Igλ gene rearrangement begins.
similarity between duplicates is 98.9%, suggesting the du-
plication occurred less than 5 million years ago. This is con-
sistent with the absence of such duplication in chimpanzees, Lambda Light Chain Genes
which diverged from the human lineage approximately 6 Murine l Locus
million years ago. Interestingly, about 5% of human alleles In laboratory mouse strains, only about 5% of the
also lack the distal duplication. B-lymphocytes utilize Igλ L chain, and the diversity of these
Outside the Igκ locus, at least 25 orphan Vκ segments L chains is meager due to the very small number of V region
have been identified in clusters on chromosome 1, 2, and 22. genes (Fig. 6.8). Complete sequence analysis23 of the murine
The orphan cluster located in the long arm of chromosome 2 locus revealed two V-J-C clusters (Vλ2-Vλx-Jλ2Cλ2-Jλ4Cλ4
was probably separated from the major locus—on the short and Vλ1-Jλ3Cλ3-Jλ1Cλ1) separated by about 110 kb. Each Jλ
arm of this chromosome—by a pericentric inversion (which is linked to its own Cλ region gene, but Jλ4 is nonfunctional.
must have occurred rather recently in evolution as it is ab- Recombination occurs largely within each cluster, although
sent from chimpanzee and gorilla). Vλ2Cλ1 products are occasionally observed. The ancestry of
the Vλx element is uncertain, as it is rather dissimilar to the
other Vλ segments; indeed, it resembles Vκ as much as Vλ. In
contrast to the Igκ locus, the Vλ segments are flanked by 23-
RSS and the Jλ gene segments by 12-RSS (see Fig 6.3).
Human l Locus
The human Vλ region was characterized by intensive clon-
ing, sequencing, and mapping of Vλ elements and ulti-
mately by the complete sequence analysis of 1 Mb covering
the entire locus 24 (see Table 6.1). Within the Vλ cluster lies
the human VpreB gene (discussed below), as well as several
genes and pseudogenes unrelated to the Igλ system.
λ L chains are much more abundant in man than in
mouse (about 40% of human L chains are λ versus about
5% in mouse). Four forms of human λ chains have been
classified serologically, with differences residing in a small
A B
number of amino acids in the C region. The serologic clas-
FIG. 6.7. The Same “Micro” Mechanism of Recombination can Join sification of Kern+ corresponds to a glycine at position
Vk and Jk by Deletion or Inversion, Depending on the Relative 152 versus a serine in Kern−. The Oz + designation corre-
Orientation of the Two Precursors in Germline DNA. A: When V sponds to a lysine at position 190 versus an arginine in the
coding sequence (shaded rectangle) and J coding sequence (white Oz− variant. Similarly, Mcg + λ chains (versus Mcg−) con-
rectangle) are oriented in the same 5′ → 3′ direction in germline DNA tain asparagine 112 (versus alanine), threonine 114 (versus
(as indicated by the internal arrowheads), the recombination yields serine) and lysine 163 (versus threonine).
a VJ coding joint plus a DNA circle containing the signal joint (ap-
posed triangles). B: If V is oriented in the opposite direction in germ-
Four functional human Jλ-Cλ segments and three pseu-
line DNA, an identical recombination reaction at the “micro” level dogenes are clustered within an approximately 33 kb region
(inside shaded circle) leaves the signal joint linked to the recombined of DNA (see Fig. 6.8) and the four major expressed human λ
VJ coding joint. isotypes correspond to the functional JCλ1, JCλ2, JCλ3, and

FIG. 6.8. Germline l Genes. The maps in this figure are schematic (i.e., not to scale). A: The murine λ gene
system includes four JC complexes and three V genes, as shown. B: The human λ locus includes multiple V
genes, of which only three are shown. The human VpreB “surrogate” light chain gene is located within the Vλ
cluster. The Cλ locus includes a segment of seven JC complexes plus three additional unlinked sequences. The
hatched JC complexes diagrammed above the seven linked λ sequences represent polymorphic variants with
additional duplications of the JC unit. The 14.1 sequence—the human λ5 “surrogate” light chain homolog—lies
downstream of the JC cluster. Exon 1 of the 14.1 gene is homologous to an exon upstream of Jλ1 (as indicated
by the unlabeled white rectangle).
JCλ7, with the latter encoding an isotype provisionally des- appears to be functional, 28 and a less similar VpreB3 has also
ignated Mcp.25 JCλ6 may be functional in some individuals, been described. Neither λ5 nor VpreB genes show evidence
and the common allele—which has a 4 bp insertion lead- of gene rearrangement in B or pre–B cells, and homologs
ing to a deletion of the C-terminal third of the Cλ region— have been found in every mammalian species examined.
can nevertheless undergo Vλ-Jλ recombination, encoding a The two SLC proteins form a L chain–like heterodimer that
truncated protein that can associate with H chains. A vari- is able to fulfill some functions of a true L chain, including as-
ety of polymorphic variants of the human λ locus have been sociation with μ H chains to permit surface μ expression prior
detected, apparently the result of gene duplication, as shown to the availability of κ or λ L chains. Thus, when a μ H chain
in Figure 6.8.26 Lastly, three Cλ-related sequences have been gene was transfected into an Ig-negative myeloma line, no
discovered near the major Jλ-Cλ cluster. One of these, des- surface μ expression was observed unless λ5 and VpreB genes
ignated λ14.1, represents the human homolog of the murine were also transfected.29 Surface μ chains are covalently linked
“surrogate” L chain λ5 (see following discussion). to the λ5 protein, while the VpreB1 protein is noncovalently as-
sociated. The expression of μ-SLC on the surface of pre–B cells
triggers the onset of Vκ-Jκ rearrangement, as discussed below.
l-Related “Surrogate” Light Chains In humans, three λ5-like sequences are located down-
Ig H chains cannot reach the cell surface without pairing stream of the human Cλ cluster on chromosome 22 (see
with Ig L chains. However, Ig μ H chains can be detected Fig. 6.8), but only one—designated 14.1—appears to be
on the surface of pre–B cells whose Ig κ and Igλ loci are still functional. The human VpreB homolog lies within the Vλ
in their germline configuration and thus do not produce L cluster30 in contrast to murine VpreB, which lies close up-
chains. In these cells, a “surrogate L chain” (SLC) composed stream of λ5.
of two smaller proteins, VpreB and λ5, facilitates the surface
expression of the μ H chain protein. The first component
(λ5) was identified as the product of a gene expressed exclu- V(D)J RECOMBINATION
sively in pre–B cells that showed high sequence similarity The mechanism by which germline variable region segments
to the J and C regions of the λ locus, 27 As four murine Cλ (V L and JL, or V H, D, and JH) are assembled in the DNA to
genes were already known, it was designated λ5. The sec- form a complete active V region has been pursued ever since
ond component of the SLC was identified as a gene residing Ig gene recombination was first discovered. In this section
about 4.7 kb upstream of λ5 in the mouse genome. Based we will address 1) the molecular mechanism of the reaction,
on its similarities to both Vλ and Vκ (and its expression in 2) the topology of the recombination events, 3) the compo-
pre–B cells), it was called VpreB1. A second, nearly identi- nents of the recombinase machinery, and 4) the regulation
cal sequence in the mouse genome is named VpreB2 and of that machinery during B-cell development.

Molecular Mechanism of V(D)J Recombination

Recombination Model Overview
A model for the detailed mechanism of the V(D)J recombi-
nation event must account for the observed features of the
recombination products and of their germline precursors.
In the germline precursors, the RSSs with their heptamer,
nonamer, and appropriate 12 or 23 bp spacers are neces-
sary and sufficient to create efficient recombination tar-
gets; model substrates in which RSSs fl ank DNA sequences
completely unrelated to Ig genes are competent to undergo
recombination. The model shown in Figure 6.9 will serve
as a framework for discussion of the recombination mech-
anism. The recombination is thought to begin with bind-
ing of the RAG1-RAG2 complex to the RSSs that flank the
two gene segments to be recombined. Simultaneous DNA
cleavage occurs precisely between the RSSs and the gene
segments. The two ends of the RSSs (frequently named
“signal ends”) are joined directly, forming “signal joints.”
In contrast, the ends of the gene segments (also referred to
as “coding ends”) are processed prior to joining and are
ultimately ligated together, giving rise to “coding joints”
and completing the recombination event.
Recombination Products: Coding Joints and

Signal Joints
In the recombination products, signal joints are typically di-
rect ligation products of the signal ends: the RSSs are joined
directly at the heptamers (“back-to-back”), and nucleotide
additions or deletions at these junctions are quite rare. The
properties of the coding joints, however, are more complex,
as the joining reaction at these DNA ends is “imprecise.”
The following features are frequently present:
1. Deletions: variable number of bases are deleted from
the ends of the coding regions (in comparison to the
“complete” sequence in the germline precursor)
2. Nongermline (“N”) nucleotides: random nucleotides
(with a bias toward G and C) are added by a template-
independent DNA polymerase (discussed below). The
sequence of these N nucleotides has no relationship to
the germline V, D, or J sequences. FIG. 6.9. Model for V Assembly Recombinations. All V assembly recombi-
3. Palindromic (“P”) nucleotides: the ends of the cod- nation reactions (in immunoglobulin and T-cell receptor genes) may pro-
ing gene segments are sealed by a DNA hairpin struc- ceed by a common mechanism, illustrated here by D-J recombination.
ture (see Fig. 6.9, and discussed in the following). The recombination signal sequences (RSSs) are included in triangles,
which is the conventionally used RSS graphic. Hairpin loops are cre-
“Opening” of these hairpins frequently occurs by nick-
ated on coding ends dependent on the action of the two Recombination
ing at some distance away from the hairpin tip lead- Activating Genes: RAG1 and RAG2. After the opening of the hairpin loops,
ing to single-stranded overhangs. Filling in of such the pictured diversity coding sequence shows the effects of “nibbling”
overhangs by DNA polymerases generates DNA palin- by exonuclease, but the joining coding sequence is spared and shows
dromes that mirror the nucleotides at the end of the V, P nucleotide generation; N region addition is pictured in this example as
D, or J segment.31 P nucleotides are generally only one occurring only on the diversity region end. In reality, exonuclease diges-
or two bps, but they can be longer, especially in mice tion and N nucleotide addition can occur on either (or both) ends. The
steps in the proposed mechanism are discussed in the text.
with the severe combined immunodeficiency defect
(SCID) disorder in which the opening of the hairpins
occurs in an aberrant manner. polymerase chain reaction (LM-PCR) to detect signal ends.
This technique involves ligating blunt double-stranded oli-
Recombination Intermediates: Blunt Signal Ends and gonucleotide linkers to blunt genomic DNA breaks, and then
Hairpin Coding Ends amplifying the ligation junctions between a primer in the li-
To study broken DNA ends as intermediates in V(D)J recom- gated oligonucleotide and a primer based on known sequence
bination, several laboratories employed ligation-mediated– from the ligated genomic DNA; amplification products can

then be cloned and sequenced. LM-PCR analyses of both In the first phase of V(D)J recombination, the DNA is cut at
TCR and Ig genes undergoing V(D)J recombination showed both gene segment–RSS boundaries that participate in the
the signal ends to be blunt double strand breaks (dsbs), reaction, thereby generating four DNA ends. In principle,
usually exactly at the heptamer border.32 Similar LM-PCR there are three possible topologies in which these DNA ends
experiments failed to detect the coding ends unless they can be rejoined:
were pretreated with mung bean nuclease, a single-strand- 1. “Signal and coding joints”: the standard reaction prod-
specific endonuclease that recognizes the distortion of DNA uct in which the two coding ends get joined generating
at a hairpin structure. Sequences of these LM-PCR products the assembled VJ gene and the 12-RSS/23-RSS signal
from coding ends suggested that the hairpins are precisely joint.
at the end of the coding elements, usually without loss or 2. “Open and shut joints”: the RSSs get ligated back to the
gain of a single nucleotide.33 By Southern blot analyses, cod- gene segments from which they were released. These
ing ends were found to have two properties suggestive of a joints are topologically identical to the starting DNAs,
hairpin-like structure: 1) resistance to exonuclease treat- but can be distinguished from them if nucleotides have
ment, and 2) doubling of the apparent length of restriction been added or deleted at the junctions.
fragments under denaturing electrophoresis conditions.34 3. “Hybrid joints”: joints in which the RSSs have traded
Hairpin ends represent V(D)J recombination intermedi- places so that the 23-RSS that was flanking the Vκ seg-
ates that, in wild-type cells, are opened at the hairpin tip (or ment is now linked to the Jκ segment, and vice versa.
a few nucleotides away from it) by the Artemis nuclease (dis-
“Hybrid” and “open and shut” joints have been observed
cussed below). P nucleotides result from opening the loop at
in transfected plasmids bearing artificial recombination
an asymmetric position (see Fig. 6.9); this model would ex-
substrates36 as well as in endogenous Ig loci in vivo.37
plain why P nucleotides are never observed at coding ends
that have been “nibbled” after opening of the hairpin. P nu- Secondary V(D)J Recombination
cleotide segments in the rare coding joints observed in SCID As discussed previously, imprecise joining of gene seg-
mice are unusually long and likely result from resolution of ments causes about two-thirds of all recombination prod-
hairpins by nicking enzymes that, unlike Artemis, do not ucts to be out-of-frame. Thus, a B-lymphocyte could end
focus on the area near the tip of hairpin loops but instead nick up with nonproductively rearranged Igκ genes on both al-
in variable positions in the double-stranded hairpin “stem.”34 leles. However, germline Vκ segments lying upstream of an
initial Vκ-Jκ recombination junction can recombine with
Jκ segments lying downstream of the junction, producing a
Topology of V(D)J Recombination “secondary” recombination event, as shown in Figure 6.10A.
Deletion versus Inversion
If a V segment and a J segment are both oriented in the
same direction, they can recombine by excising the DNA
between the coding sequences and ligating the two cod-
ing ends. Ligation of the two signal ends produces a DNA
circle that generally lacks replication origins and therefore
fails to replicate as cells divide after V(D)J recombination.
Such excision circles are therefore generally absent in mature
B-lymphocytes that have already undergone several rounds A
of proliferation after completing the Ig gene assembly. By
isolating circular DNA from cells actively undergoing Vκ-Jκ
rearrangement, it is possible to isolate and characterize the
circular molecules bearing signal joints.35
As mentioned previously, some germline Vκ genes are
oriented in the opposite direction from the Jκ-Cκ region. In
these cases, VJ recombination occurs by an inversion of the
DNA between the recombining V and J segments, leaving
both the VκJκ coding joint and the signal joint (formed by
ligating the RSSs) retained in the chromosome (see Fig. 6.7).
This demonstrates that the enzymatic machinery “sees” only
the DNA in the immediate vicinity of the recombination site B
and is insensitive to the topology of the DNA strands far
from this site. FIG. 6.10. Secondary Recombinations. A: In the κ light chain system,
a primary recombination can be followed by recombination between
Nonstandard Joints an upstream V and a downstream J. B: Analogous secondary recom-
binations can occur in the heavy chain system between upstream D
In addition to the canonical coding and signal joints, sev- and downstream J segments. After V(D)J recombination eliminates
eral “nonstandard” recombination joints have been docu- all 12-RSS signal elements from the chromosome, secondary recom-
mented, that, though not contributing to physiologic Ig gene bination can still occur between VH (23-RSS) and an internal hep-
assembly, represent tell-tale signs of a recombination event.36 tamer within the VH coding sequence of the VDJ unit.

Such secondary recombination also occurs in cells that have as expected, no measurable recombination occurred in this
assembled a productive Vκ-Jκ joint if the encoded antigen nonlymphoid cell. However, when either human or murine
binding domain recognizes an autoantigen. This type of sec- genomic DNA was transfected into these fibroblasts, a small
ondary recombination, known as “receptor editing,” is consid- fraction of recipient cells stably expressed recombinase
ered in more detail later in this chapter. activity, activating the selectable marker. This suggested that
In the IgH locus, secondary D-J rearrangements some- a single transfected genomic DNA fragment was able confer
times occur, but only until V H-DJH recombination removes recombinase activity in a fibroblast. (Presumably the fibro-
all unused upstream DH segments (Fig. 6.10B). VDJ rear- blast contained endogenous copies of the same genes, but
rangement eliminates all the 12-RSSs from the IgH locus their expression was repressed by mechanisms that could
that could pair with the 23-RSSs flanking the upstream V H not repress the transfected genes.) This active fragment was
elements. Sometimes, these V H segments do, however, re- cloned and turned out to contain two closely linked genes,
combine with an established VDJ unit, displacing most of designated RAG1 and RAG2, respectively. Both RAG1 and
the originally assembled V H element,38 a process sometimes RAG2 are essential for recombination; therefore, these genes
called V H replacement. Such events are mediated by cryptic would not have been discovered by this transfection tech-
RSSs (mainly a heptamer sequence) that is present near the nique if they had not been closely linked in the genome. The
3′ end of about 70% of all V H genes (see Fig. 6.10B). Such in- genes are notable for having no introns splitting up their
ternal cryptic RSSs are not generally found in L chain genes. open reading frame in most species, and for their opposite
As discussed previously for the L chain, secondary recombi- transcriptional orientation in all species examined.
nation represents a rescue mechanism for cells with nonpro- A crucial role for the RAG genes in V(D)J recombination
ductive rearrangements on both H chain chromosomes, and was supported by the conservation of these genes in all jawed
for cells whose encoded antibody recognizes an autoantigen. vertebrate species analyzed thus far, from shark through
man. RAG1 and RAG2 are expressed together in develop-
ing B and T cells, specifically at the stages at which V(D)J
The V(D)J Machinery recombinase activity is required for the assembly of Ig and
Since the discovery of V(D)J recombination as the process that TCR genes. Moreover, mouse strains in which either gene
assembles the germline antigen receptor gene segments into has been eliminated by homologous recombination (gene
functional genes, one major question was the identity of the “knockouts”) have no mature B or T cells, as the result of
enzymatic machinery catalyzing this complex set of reactions. their inability to initiate V(D)J recombination.41,42 Similarly,
Genetic and biochemical work by a large number of laborato- a subset of human patients with SCID syndrome character-
ries led to identification of a total of 13 different proteins that ized by the complete absence of T- or B-lymphocytes have
have been shown to be directly involved in V(D)J recombina- been found to have null mutations in RAG genes.43 Patients
tion: RAG1, RAG2, HMG1, Ku70, Ku80, DNA-PKcs, Artemis, with hypomorphic alleles often have a complex set of fea-
pol μ, pol λ, TdT, XRCC4, Cernunnos/XRCC4-like factor tures (oligoclonal T cells, hepatosplenomegaly, eosinophilia,
(XLF), and DNA ligase IV. The only lymphoid-specific fac- decreased serum Ig but elevated IgE) known as the Omenn
tors are RAG1, RAG2, and TdT; all others are ubiquitously ex- syndrome, which can also be caused by defects in other genes
pressed in all cell types, and this feature allows investigators to involved in V(D)J recombination. Interestingly, the same
study aspects of V(D)J recombination by ectopically express- RAG mutation in different patients can cause either Omenn
ing the RAG proteins in nonlymphoid cells. A recent biochem- syndrome or SCID, depending on unknown factors.44
ical tour de force study showed that coding joint formation RAG1 shows intrinsic binding affinity for the RSS nona-
could be recapitulated in vitro using artificial recombination mer sequence via its nonamer binding domain even in the
substrates and highly purified preparations of all 13 proteins.39 absence of RAG2. Exhaustive mutational analysis has re-
The respective coding joints showed all of the features typi- vealed that RAG1 contains the catalytic center of the RAG
cally observed in vivo (nucleotide deletion, N nucleotide, and complex, composed of three amino acids critical for all en-
P nucleotide addition), suggesting that most, if not all, of the zymatic activity: D600, D708, and E962.45,46 RAG2, on the
factors involved in the coding end processing steps of V(D)J other hand, serves as a regulatory cofactor; it has no intrin-
recombination have been identified. In contrast, signal joint sic binding affinity for RSSs, but once bound to RAG1 im-
formation was not observed. This step seems to require the re- proves the strength and specificity of RAG1 RSS contacts.47,48
moval of the RAG proteins after the cleavage reaction and is It is also enhances RAG activity on chromosomal substrates
likely to require additional factors as yet unidentified. and it restricts V(D)J recombination to the G0/G1 stage of
the cell cycle (both features are discussed below).
Recombination Activating Gene Proteins: Mediators of Attempts to determine the molecular role of the RAG
Early Steps in V(D)J Recombination proteins in cell-free recombination assays were initially ham-
A major advance in the investigation of V(D)J recombination pered by poor solubility of the proteins, but functional analy-
was the identification of two genes whose products are criti- ses of truncated RAG genes (using RAG expression vectors
cal for this process in the B and T cell lineages. In the pioneer- cotransfected into fibroblasts along with recombination sub-
ing experiments, Schatz and Baltimore40 stably transfected strate plasmids) revealed that surprisingly large segments of
fibroblasts with a construct containing a selectable marker both proteins could be deleted without eliminating recom-
whose expression was dependent on V(D)J recombination; binase activity, and some of the remaining core regions were

soluble and could be handled relatively easily in experiments. next to the primordial RAG2 gene) or both RAG1 and
This work allowed the demonstration that in a cell-free in RAG2. The transposase activity of RAGs, however,
vitro system, core regions of the two RAG proteins together seems to be almost completely suppressed in vivo, and
are capable of carrying out cleavage of substrate DNAs as well the C-terminus of RAG2 may have evolved to control
as hairpin formation on the coding end.49 this potentially deleterious activity.51,58–60
The RAG-mediated cleavage occurs in two steps: first a 2. VH replacement. As mentioned previously, recombina-
nick is introduced on the top strand between a gene seg- tion events can occur between a VH 23-RSS and cryp-
ment and the adjacent heptamer (see Fig. 6.9), then the tic RSS within rearranged VH coding sequences. An in
3′-hydroxyl group participates as the nucleophile in a direct vitro model suggests that in VH replacement, the RAG
transesterification reaction to attack the phosphodiester proteins nick both DNA strands without forming a
bond adjacent to the heptamer on the bottom strand (see hairpin coding end.61 Whether this is indeed a com-
Fig. 6.9), yielding a DNA hairpin structure on the coding pletely different activity is unclear.
end and a new 3′-hydroxyl group on the 3′ end of the bottom 3. Translocations at non-RSS sequences. The RAG com-
heptamer strand.50 After DNA cleavage, the RAG proteins plex also generates two nicks to cleave within the major
remain in a complex with the DNA ends and facilitate as- breakpoint region of the Bcl2 gene. This 150-bp segment
pects of the joining phase. Mutant forms of RAG1 or RAG2 is the target of a common RAG-catalyzed translocation
have been reported that are competent for cleavage but show between the IgH locus and the Bcl2 gene occurring in
impairment in coding or signal joint formation.51 most follicular lymphomas. In this segment, there are
While nicking can occur asynchronously at the 12-RSS no RSSs, and the RAG proteins recognize an unusual
and 23-RSS, hairpin formation is “coupled” and occurs sequence-dependent DNA conformation different
synchronously at both RSSs. In vitro, coupled cleavage re- from the normal B-form double helix.62
quires only the RAG proteins, HMG1/2 (discussed below) Although the “core” RAG proteins have been useful for
and Mg 2+ as the divalent metal ion in the reaction buffer. elucidating the molecular mechanism of the cleavage step
In vivo, DNA dsb formation at an individual RSS is danger- of V(D)J recombination in biochemical studies, it is clear
ous as it could give rise to translocations, and it is thought that the “noncore” portions of each protein confer impor-
that Mg 2+ promotes an optimal molecular “architecture” for tant functions, as expected from their sequence conserva-
controlled V(D)J recombination. In vivo experiments indeed tion across species. Broadly speaking, the “noncore” regions
suggest that RAG proteins may bind to and introduce a nick ensure regulated and efficient recombination on the physi-
at a single 12-RSS, but do not complete DNA cleavage until ological substrates (i.e., imperfect RSSs deviating from the
a matching 23-RSS is captured into the RAG-RSS complex.52 perfect consensus heptamer and nonamer) in the context
In addition to the “classical” activities of RAG proteins of chromatin. The functions of the “noncore” regions have
on DNA segments containing RSSs, these proteins can also largely been inferred by comparing V(D)J recombination
catalyze DNA strand cleavage on “nonstandard” substrates. products from cells expressing core RAG proteins versus
1. Transposition. In vitro, purified recombinant core full-length versions, and more recently by in vitro studies
RAG proteins can catalyze the excision and inser- using full-length RAG proteins that are now available for
tion of a DNA fragment with signal ends into foreign such analyses.
DNA, acting as a transposase.53,54 This property pro- The C-terminal region of RAG2 has multiple functions
vides additional support for the early speculation that and is important for achieving normal numbers of B- and
the V(D)J recombination system may have originated T-lymphocytes in vivo, 63 for the formation of precise sig-
by insertion of transposon-like DNA fragment encod- nal joints during IgH recombination, 64 and for protecting
ing RAG genes (and bearing RSSs at its ends) into a against RAG-mediated DNA transposition.51,65 These func-
primordial antigen receptor gene, thereby generat- tions are thought to be conferred at least in part, by a plant
ing a pair of separated V and J gene segments. This homeo domain (PHD) zinc finger fold that is formed by
model of the origin of V(D)J recombination is con- amino acids 414 to 487 in murine RAG2. This PHD domain
sistent with the many mechanistic similarities at the binds specifically to the tails of histone H3 that are trimeth-
molecular level between Ig gene rearrangements and ylated at lysine 4 (H3K4Me3), 66–68 a histone modification
transposition,55 and the recent identification of the that is associated with “open” chromatin and that is uniquely
Transib transposase family that shows striking se- present on “accessible” RSSs in Ig loci (discussed below). In
quence similarity to RAG1 and is widespread in in- vitro studies suggest that the binding of the RAG2 PHD do-
sect, echinoderm, helminth, coelenterate, and fungal main to histone tails causes a conformational change that
genomes.56 The recent finding of an apparent homo- increases the catalytic activity of the RAG complex. 69
log of the entire RAG1 and RAG2 gene locus in a sea Furthermore, the RAG2 C terminus regulates RAG2 pro-
urchin genome suggests that the two RAG genes may tein levels—and hence V(D)J recombinase activity—across
have entered the genome of a common ancestor of the cell cycle to prevent dsbs during DNA synthesis or mito-
all deuterostomes far earlier than the Ig-/TCR-based sis, when such breaks could lead to chromosomal deletions.32
adaptive immune system developed.57 It remains un- RAG1 protein and messenger RNA (mRNA) transcript levels
clear whether the primordial RAG transposon en- of both RAG genes vary little across the cell cycle, but phos-
coded solely RAG1 (which would then have integrated phorylation of RAG2 at Thr490 by the cyclin-dependent

kinase cdk2 mediates its destruction via ubiquitination and ubiquitous proteins that bind DNA in a non–sequence-spe-
proteasomal degradation during S phase.70 Mice expressing cific manner and to cause a local bend in DNA. The two
RAG2 with a T490A mutation (which cannot be phosphory- RAG proteins can form a stable signal complex with a 12-
lated) showed RAG2 protein and dsbs throughout the cell RSS, but efficient complex formation with a 23-RSS requires
cycle, demonstrating the importance of the RAG2 degrada- the addition of either HMG1 or HMG2.76 HMG1/2 appar-
tion signal in cell-cycle control of V(D)J recombination.71,72 ently stabilizes the bending of the 23-RSS that is induced by
The N-terminal noncore region of RAG1 is required in the RAG proteins themselves.77
vivo for optimal RAG1 activity and for the formation of
precise signal joints in D-J recombination.64 This region Nonhomologous End Joining Components
of RAG1 contains a RING finger domain that seems to be The RAG proteins are the essential lymphocyte-specific fac-
required for ubiquitination of several proteins, including tors in the DNA cleavage phase of V(D)J recombination, but
histone H3.73 DNA repair factors that are part of a DNA repair pathway
Apart from the obvious importance of the RAG proteins known as nonhomologous end joining (NHEJ) are essen-
in understanding the initial steps of V(D)J recombination, tial for the joining phase. NHEJ is the major pathway for
knowledge of these proteins and their genes has allowed repair of dsbs (such as those induced by ionizing radiation
two major technical advances that have opened the way to or reactive oxygen species) during the G0-G1 phases of the
many additional experiments. First, various nonlymphoid cell cycle. (In the S and G2 phases, the additional chroma-
cell lines with known defects in various DNA repair genes tid genome copy enables breaks to be repaired by homolo-
have been transfected with the RAG genes to identify genes gous recombination.) The six classical core components of
involved V(D)J recombination (these factors are described NHEJ are Ku70, Ku80, DNA-PKcs, XRCC4, DNA Ligase IV,
below). Second, availability of the RAG1 and RAG2 knock- Artemis, and Cernnunos/XLF, but additional proteins play
out mice has been instrumental in a large number of im- a role in some models of NHEJ.
munology studies. These mice completely lack functional B
cells or T cells, and are not “leaky” like SCID mice, which
develop some functional B and T cells, especially as the The DNA-PK Complex . The first gene for an NHEJ compo-
animals age. Thus the RAG-deficient mice can be used to nent to be recognized as participating in V(D)J recombina-
study the importance of the “innate” immune system (i.e., tion was the SCID gene. This gene was originally identified
responses that occur in the absence of antigen-specific lym- as being mutated in the scid mouse strain that is immunode-
phocytes) in particular immune responses. They can also ficient due to a marked impairment in V(D)J recombination
be used as recipients for various lymphocyte populations of both Ig and TCR genes. Lymphocytes from scid mice are
to explore the roles of different cell types. They can also be able to perform the RAG-mediated cleavage reaction, and
used as recipients for various lymphocyte populations to can also form signal joints, but are markedly defective in
explore the roles of different cell types. They can be trans- coding joint formation. Subsequently, it was found that the
fected with transgenes encoding specific Ig genes to study scid mutation also impairs NHEJ, causing radiosensitivity.
the roles of specific antibodies in B cell development and in The gene mutated in the scid mouse strain encodes DNA-
immune responses. Finally, they can be used in “RAG com- PKcs, a large protein (460 kD) with a kinase domain near
plementation” experiments designed to assess the pheno- its C terminus that is related to phosophoinositide-3-kinase
type—in lymphocytes—of various other gene knockouts.74 (PI3K). This kinase is DNA-dependent and represents the
In RAG complementation, embryonic stem cells in which catalytic subunit (hence “cs”) of a heterotrimer known as
the gene of interest has been knocked out by homologous the DNA-PK complex. The other components are Ku70 and
recombination are injected into homozygous RAG2 knock- Ku80 (also referred to as Ku86), which were originally iden-
out (RAG2−/−) blastocysts. This procedure yields chimeric tified as the autoantigens recognized by a patient antiserum
mice in which all B and T cells derive from the embryonic (Ku was the coded name of the patient, and the numbers
stem cells deleted for the gene of interest, as these are the refer to the approximate size of the proteins, 70 kD and 80
only source of intact RAG genes to support lymphocyte de- to 86 kD, respectively). Together, these two very abundant
velopment. Such animals can be made more easily than a proteins form a heterodimer that binds to the ends of dou-
knockout mouse line, and can be used to study the effect of ble-stranded DNA independent of the nucleotide sequence
gene deletion in lymphocytes independent of effects the de- of the DNA. The DNA-Ku complex can then recruit DNA-
letion may have in other cells. In particular, for cases where PKcs and activate autophosphorylation of this protein.78
the gene knockout causes embryonic lethality due to effects In vitro activation of DNA-PKcs was found to be efficient
on nonlymphoid cells, RAG complementation allows the se- when DNA ends either were at high concentration or, if at
lective knockout in lymphocytes to be studied in the back- low concentration, were on DNA fragments long enough to
ground normal gene expression in nonlymphoid cells. circularize readily. In contrast, when the DNA-PKcs was lo-
cated on the ends of DNA fragments too short to circular-
ize (and too dilute for efficient intermolecular interactions
High Mobility Group Proteins with other DNA ends), the DNA-PKcs activation was much
The search for RAG cofactors that stimulate cleavage activ- reduced. These observations suggest that kinase activation
ity in biochemical assays led to the identification of HMG1.75 can occur only after two DNA ends are brought together
HMG1 (and the closely related HMG2) are abundant and by DNA-PKcs in “synapsis.”79,80 Further phosphorylation of

DNA-PKcs inactivates the protein and may prepare it for and therefore V(D)J recombination requires a single-strand
removal once DNA ends have been sealed. endonuclease activity to cleave the hairpins. This activity
Ku genes are highly conserved through evolution, and is conferred by the protein named Artemis, which was dis-
homologs are even found encoded in the genome of some covered through positional cloning of the genetic defect in
bacteria, consistent with a function in general NHEJ not a group of human SCID patients with defects in V(D)J re-
restricted to V(D)J recombination. While mice with a tar- combination and increased radiation sensitivity.94 Patients
geted deletion of DNA-PKcs resemble the original scid with homozygous null mutations of Artemis survive (no
mutation (i.e., defective coding but functional signal joint embryonic lethality) and show sensitivity to γ irradiation
formation81,82), Ku70 and Ku80 mutant cell lines are defec- as well as defects in coding joints, while signal joint forma-
tive in both signal and coding joint formation, and Ku70- tion is normal. Hypomorphic Artemis mutations can cause
and Ku80-deficient mice exhibit a complete block in B- and features of the Omenn syndrome similar to those observed
T-cell development due to their inability to undergo V(D)J with hypomorphic RAG gene mutations.95 Purified recom-
recombination. 83–85 binant Artemis protein has an intrinsic exonuclease activity
in vitro; however, when complexed with DNA-PKcs in the
DNA Ligase IV and XRCC4 . An important role of activated presence of DNA ends, it gains a single-strand endonucle-
Ku-DNA-PKcs complex is to recruit the additional compo- ase activity and, in an ATP-dependent step, becomes phos-
nents of NHEJ. One such component is DNA ligase IV, which phorylated at multiple sites in the C-terminal region of the
is recruited to the Ku complex and activated by the protein protein.96,97 The Artemis endonuclease can cleave synthetic
XRCC4. 86,87 The evidence suggests that DNA ligase IV is the and RAG-generated hairpin ends as well as other single-
essential ligase that joins DNA ends in V(D)J recombina- stranded DNA near a transition to double-strand DNA.98
tion and NHEJ. Human patients with ligase IV deficiency
(characterized by hypomorphic alleles) have a severe phe- DNA Polymerase X Family Members. If a hairpin open-
notype including chromosomal instability, developmental ing leaves blunt ends or complementary sticky ends (like
and growth retardation, radiosensitivity, and immunodefi- the ends generated by many restriction enzymes), in vitro
ciency with a T–B–NK+ phenotype. 88 The rare DH-JH junc- joining experiments suggest that these ends can be joined
tions detected show extensive nucleotide deletion consistent by ligase IV without any additional processing.99 However,
with delayed ligation and prolonged exonuclease digestion.89 as Artemis probably opens most hairpins noncomplemen-
In mice, disruption of either the XRCC4 or the DNA ligase tary DNA overhangs, further processing of DNA ends gen-
IV gene causes embryonic lethality associated with neuronal erally occurs before ligation completes the recombination.
apoptosis. Crossing these mice with p53 mutants does not This processing may include further nuclease digestion
improve V(D)J recombination, but rescues the mice from (by Artemis or exonucleases) and apparently also involves
embryonic lethality, suggesting that neuronal cells may be variable DNA extension by three DNA polymerases—poly-
unusually susceptible to p53-triggered apoptosis induced by merase λ , polymerase μ, and terminal deoxynucleotidyl
normal low-level DNA damage during brain development; transferase (TdT)—all of which are members of the poly-
a similar mechanism may explain the severe human phe- merase X family. Interestingly, all three proteins contain a
notype.90 DNA ligase IV is the only NHEJ component abso- Brca1-C-terminus domain, which is thought to confer bind-
lutely required to join compatible sticky DNA ends in vitro, ing to Ku.100
though XRCC4 can stimulate this activity significantly. 87
Cernunnos/XRCC4-like Factor. The next NHEJ component Terminal Deoxynucleotidyl Transferase and N Regions.
was independently discovered by two laboratories. One group TdT, the primary source of untemplated “N region” addi-
used yeast two-hybrid screening to search for proteins inter- tions in VDJ junctions, is an enzyme uniquely expressed in
acting with XRCC4.91 The other group searched for the gene the thymus and bone marrow; in the B lineage, it is expressed
causing a syndrome of T+ B lymphocytopenia, increased ra- almost exclusively in pro-B cells. It catalyzes the nontem-
diosensitivity, and microcephaly in a Turkish family; these plated addition of nucleotides to the 3′ end of DNA strands.
investigators used functional cDNA rescue of a patient’s cell Though no template determines the nucleotides added, the
line from a radiomimetic drug to identify the gene.92 The enzyme adds dG residues preferentially, consistent with N
protein identified by both groups is a 299 amino acid nuclear region sequences observed in VDJ joints. Both TdT expres-
protein, which was named Cernunnos or XLF. The protein sion and N nucleotide addition are characteristically absent
has a predicted secondary structure similar to that of XRCC4, from fetal lymphocytes.101 N region addition is common in
to which it binds in cells93 as expected from its isolation via H chain genes (recombined in pro-B cells) but rare in mu-
two-hybrid screen. When Cernunnos/XLF-deficient fibro- rine L chain genes (recombined in pre-B cells), though per-
blasts were transfected with RAG genes and a recombination haps somewhat less rare in human.102 This is consistent with
substrate, imprecise signal joining was observed, similar to the observation that in mice the expression of a μ H chain
the defect in patients with hypomorphic DNA ligase IV mu- may downregulate TdT expression,103 contributing to the re-
tations. These experiments all suggest a role for Cernunnos/ duced level during the stage of L chain recombination.
XLF linked to the function of XRCC4 and ligase IV. Lymphocytes with engineered defects in their TdT genes
produced rearranged Ig V regions with almost no N addi-
Artemis. The coding ends generated by RAG cleavage can- tions. Conversely, when TdT expression was engineered in
not be directly ligated because of their hairpin structure, cells undergoing κ or λ L chain rearrangement, the level of

N nucleotide addition to these coding joints was dramati- are compatible with near normal V(D)J recombination.
cally increased. Furthermore, mice engineered to undergo Possibly, they participate in backup mechanisms to prevent
premature Vκ-Jκ joining in pro-B cells show an increased aberrant V(D)J recombination and thus translocations.
frequency of N region nucleotides in their recombined Vκ
genes.104 These results suggest that the low frequency of N Pax5/B-Cell–Specific Activator Protein. Pax5 (also known
region sequences in normal κ or λ recombinations is caused as B-cell–specific activator protein; BSAP) is a transcrip-
by the reduced levels of TdT at this stage of B-cell develop- tion factor required for normal B-cell development. Pax5-
ment (see following discussion). deficient mice are able to complete DJH recombination, but
The absence of N region addition in TdT mutant mice, V H to DJH recombination is impaired except for certain V H
as well as in normal fetal lymphocytes, is associated with an genes located proximal to the D regions. Interestingly, 94%
increase in the frequency of recombination junctions with of human and mouse V H coding genes were found to have
microhomologies. These are short stretches of nucleotides potential Pax5 binding sites. Surprisingly, Pax5 was found
that are present close to the end of both germline gene seg- to coimmunoprecipitate with RAG proteins, to potentiate
ments involved in the recombination event. These junctions in vitro cleavage of a V H gene RSS, and to enhance V H to
suggest a joining intermediate in which the complementary DJH recombination in RAG-transfected fibroblasts; the lat-
single-stranded regions from the two coding ends hybridize ter enhancement required intact Pax5 binding sites in the
to each other, much as “sticky ends” generated by restriction V H sequence.109
endonucleases can facilitate ligation of DNA fragments.
This alternative joining pathway may restrict the diversity
of neonatal antibodies; the resulting antibodies are possibly
REGULATION OF V(D)J RECOMBINATION IN B-CELL
enriched in specificities for commonly encountered patho- DEVELOPMENT
gens, or have broadened specificity, as has been reported for The expression of only one antigen binding specificity by
TCRs lacking N regions.105 Decreased N region nucleotides each B-lymphocyte is a crucial requirement of the clonal
and a high incidence of homology-mediated recombination selection model of the humoral immune response. Thus,
have also been found in the rare coding joints formed in the recombination events that occur between Ig gene seg-
Ku80−/− mice, consistent with a role for Ku in recruiting ments are carefully regulated so that most B cells express
TdT or supporting its action.106 only one L chain isotype, either Ig κ or Ig λ (isotype exclu-
sion), and use only one of the two alleles of H and L chain
Polymerase m and Polymerase l. Polymerase μ and poly- genes (allelic exclusion). These constraints ensure that
merase λ are ubiquitously expressed polymerases. Both each B cell expresses a single H2L2 combination. Current
readily fi ll in single-strand gaps in DNA and apparently par- evidence suggests that V(D)J recombination is controlled
ticipate in V(D)J recombination by fi lling in single-strand largely at two levels: regulation of the RAG protein activity
3′ overhangs generated by asymmetric hairpin opening. and regulation of accessibility of the germline V, D, and
Without this fi lling in, such overhangs might be resected J elements to the recombinase machinery. Both of these are
by nucleases. Indeed, when in vitro NHEJ reconstitution ex- controlled by the stage of B-cell development; conversely,
periments are performed using purified proteins and DNA the expression of Ig provides a signal critical for regulating
fragments with overhanging ends, the omission of poly- maturation of B cells. A brief scheme of B-cell development
merase μ or polymerase λ increases the deletional trimming is presented in the following as background.
at junctions.100 Similar excessive deletions at VDJ junctions B- and T-lymphocytes differentiate from pluripotent
are observed in mice lacking polymerase μ or polymerase hematopoietic stem cells in the fetal liver and bone mar-
λ. Remarkably, however, polymerase μ knockout mice show row (Fig. 6.11). The primordial lymphoid progenitor has
abnormalities only in their L chains,107 whereas the deletions the potential to differentiate into B- or T-lymphocytes or
in polymerase λ knockouts are restricted to their H chains.108 natural killer cells. Among the earliest markers that indicate
This selectivity may be explained by corresponding changes B-lineage specificity are the non-Ig components of the pre-
in the relative mRNA levels for these two polymerases at dif- BCR: Igα , Igβ, and λ5. CD19, which functions as a corecep-
ferent stages of B-cell development. tor in signal transduction, first appears in large proliferating
“pro-B” cells, which also express several other distinguish-
Other Participants in V(D)J Recombination ing surface markers including c-kit, B220, TdT, and CD43.
DNA Damage Response Factors. In eukaryotic cells, DNA RAG gene expression in pro-B cells initiates D to J rear-
breaks initiate signals that halt cell division, induce DNA rangements on both alleles. Subsequently, recombination
repair, and in some cases trigger apoptosis. Several pro- with germline V H elements occurs; if the recombination is
teins apart from NHEJ components can be detected at DNA “productive” (i.e., yielding an “in-frame” VDJ junction), a
breaks induced by V(D)J recombination or irradiation, μ H chain protein can be produced. This protein appears on
including γ-H2AX, a phosphorylated form of the histone the B-cell surface along with SLC in a pre-BCR (also named
H2AX; ATM, the product of the gene mutated in the disease μ-SLC) complex that also includes Igα and Igβ. As the re-
ataxia telangiectasia; Nbs1 (or nibrin), the product of the sulting large pre-B cells proliferate, RAG gene expression
gene mutated in Nijmegen breakage syndrome; and 53BP1, declines. After several rounds of division, the cells become
p53 binding protein 1. The importance of these proteins in smaller, stop dividing, turn up RAG gene expression once
V(D)J recombination is not clear because defects in all three more, undergo L chain recombination, and express surface

FIG. 6.11. Immunoglobulin (Ig) Gene Recombination in B-Cell Development. A simplified scheme of B-cell development is presented as a
background for discussion of Ig gene recombination. The stages occurring in the bone marrow versus in the periphery (e.g., lymph nodes,
spleen) are shown, along with the status of IgH and IgL genes at each stage. A graphic depicting the Ig-related proteins displayed on the
surface at each stage is presented; at the bottom, the stage-dependent expression of recombination activating genes and terminal deoxy-
nucleotidyl transferase—both important in V(D)J recombination—is schematically depicted.
IgM. These “immature B cells” again turn down RAG ex- in the same cell. An initial nonproductive rearrangement
pression. In these IgM + IgD− immature B cells, contact with would have no inhibitory effect, so recombination could
autoantigens may upregulate RAG expression again to facili- continue until a functional product resulted or until the cell
tate receptor editing (discussed in more detail below). When used up all its germline precursors.
immature B cells eventually also express surface IgD, they In pro-B cells, the first Ig gene rearrangements join D to
become “mature B cells” and migrate into the periphery, JH segments (commonly on both chromosomes), and this is
ready to be triggered by antigen exposure. followed by V H to DJH recombination. If the first V H to DJH
recombination in a pro-B cell produces a functional VDJ
gene, a functional μ H chain will be expressed on the cell
Allelic Exclusion and Regulated V(D)J Recombination surface paired with the SLCs. The expression of this pre-
The previous description of B-cell development serves as a BCR complex has been shown to have two consequences.
background to understand an explanation of allelic exclu- First, it blocks further H chain recombination by decreasing
sion that was first proposed by Alt and colleagues110 and RAG gene expression111 and by reducing target accessibility,
has been supported by subsequent experiments. According as reflected in decreased V H gene transcription.112 The latter
to this model the functional rearrangement of an L (or H) is important for rendering the IgH locus inaccessible dur-
chain gene in a particular B cell would inhibit further L (or ing subsequent rearrangement of the Ig κ and Igλ loci. If the
H) chain gene rearrangement in the same cell. If the inhibi- initial V H to DJH rearrangement is nonfunctional (e.g., out
tion occurred promptly after the first functional rearrange- of frame), subsequent V H to DJH recombination occurs on
ment, then two functional Igs could never be produced the other allele. If the VDJ recombination product on the

second chromosome is also nonproductive, then the cell has Parameters Affecting Recombinational Accessibility
reached a dead end and is eliminated by apoptosis.113 and Transcription
The second consequence of pre-BCR expression is the V(D)J recombination is triggered by RAG expression in the
initiation of Ig L chain recombination. This effect was origi- development of both B and T cells, yet Ig gene recombination
nally deduced from the rarity of κ-expressing cells without is largely confined to B cells (exception: early T cells typically
H chain gene rearrangement, suggesting that H chain ex- show D-JH recombination); TCR gene recombination is ex-
pression is required for κ recombination. As additional evi- clusive to T cells. A widely accepted explanation for this locus
dence, a functional μ gene introduced into early B-lineage specificity is provided by the “accessibility” model.120 This
cells can cause RAG gene expression and turn on transcrip- model proposes that only those gene segments programmed
tion of unrearranged Vκ genes. These are designated “sterile” for recombination at a given stage of B- and T-cell develop-
transcripts because they cannot encode a κ protein, but they ment are “accessible” to the RAG recombinase. One clue
are required for Vκ-Jκ recombination. When this recombi- suggesting this model was that susceptibility to recombina-
nation ensues, the possibilities for functional and nonpro- tion and transcription of germline gene elements seem to be
ductive Vκ-Jκ rearrangements resemble those discussed tightly correlated.120 For example, many germline V H genes
previously for the H chain. Expression of a functional κ are transcribed at the pre-B cell stage, just at the time when
chain that can associate with μ to form a surface-expressed these genes are targets for recombination; these transcripts—
IgM molecule results in the downregulation of RAG gene designated “sterile” like the Vκ transcripts mentioned previ-
expression and suppression of further κ rearrangements. By ously—are not seen in more mature B cells in which H chain
this mechanism, functional rearranged VκJ-Cκ transgenes recombination has been terminated. In support of the acces-
can suppress rearrangement of endogenous κ genes.114 sibility model, recombinant RAG proteins incubated with
Most B cells show isotypic exclusion (i.e., they express nuclei purified from pro-B cells (which generate sterile tran-
either κ or λ but not both). Furthermore, κ rearrange- scripts in the IgH locus) were found to cleave DNA at Ig JH
ment seems to occur before λ. Thus in normal and malig- RSSs, but not at TCRδ RSSs; conversely, in pro-T nuclei the
nant human B-lymphoid cells, κ-expressing cells gener- TCRδ RSS was cleaved, but not an Ig gene RSS.121
ally have their λ genes in germline configuration, while in One molecular correlate of accessibility is the epigenetic state
λ-expressing cells, κ genes are either rearranged (rarely) or of DNA in the nuclear chromatin. The minimal repeat unit of
deleted (most commonly) by recombination signal recom- chromatin is the nucleosome, which consists of eight core his-
bination events discussed previously in this chapter.22 The tones (two copies each of H2A, H2B, H3, and H4) with 146 bp
mechanisms that dictate the order of L chain recombination DNA wrapped around it. In vitro, RAG proteins are unable to
remain unknown. Plausible models include either the selec- bind to and cut DNA wrapped around nucleosomes,122,123 and
tive suppression of λ recombination until all options on the hence nucleosomes have to be shifted or removed (a process
Ig κ locus are exhausted or differences in the timing of the called chromatin remodeling) to allow access. An alternative
developmental programs controlling κ and λ accessibility. but not mutually exclusive approach to gain access is post-
translational modification of the histone tails, which regulates
Regulation of RAG Expression the tightness of DNA-nucleosome contacts. The following sec-
A complete explanation of RAG gene expression would tion provides an overview of how accessibility of the Ig gene
explain its lymphoid specificity, the two waves of RAG ex- loci for RAG activity is regulated by several distinct but inter-
pression (during IgH and IgL rearrangements) and the connected epigenetic mechanisms. We discuss a few impor-
autoantigen-induced upregulation associated with receptor tant examples for each mechanism and refer to comprehensive
editing. Although our current knowledge is still incomplete, review articles for an in-depth discussion.
several cis-regulatory elements that regulate RAG expres- Subnuclear Localization
sion have been characterized. Surprisingly, the elements In general, inactive genes tend to be located in the periph-
and mechanism for regulating expression during B- and ery of nuclei, while active genes are recruited to a more cen-
T-cell development are distinct. RAG1 and RAG2 are tran- tral nuclear location.124 It is unclear whether the location
scribed toward each other in opposite directions, driven per se dictates the chromatin state of a locus or whether
by promoters near the respective transcription start sites. the movement is a consequence of a locus being “opened.”
Three B-cell–specific enhancers—designated Erag, D3, and Fluorescence in situ hybridization (FISH) with large (∼100
Ep—have been reported, lying about 23 kb, 8 kb, and 1.6 kb) probes specific for Ig loci is routinely used to reveal the
kb, respectively, upstream of RAG2.115–117 The B-cell–spe- position of Ig gene loci and control genes in the nucleus. The
cific function of these regulatory regions is likely explained IgH and Igκ loci are located at the nuclear periphery in he-
by the intersecting specificities of transcription factors that matopoietic progenitors and pro-T cells, but move to central
interact with them, including Pax5, E2A, FoxP1, FoxO1, areas of the nucleus in pro-B cells.125 As only the IgH locus
NFATc1, and Ikaros. NFκ B, which binds at several loca- gets rearranged at this stage, the correlation of position with
tions in the RAG enhancers, and FoxO1 (binding to Erag) accessibility is not perfect.
were found to be important mediators of the upregulation
of RAG expression in cells undergoing receptor editing.118,119 Transcription
Regulation of RAG2 protein across the cell cycle has been Transcription typically occurs in “open” chromatin, and an
discussed previously in this chapter. emerging theme suggests that while some locus “opening” has

to precede transcription, transcription per se also positively regions of intervening DNA. Evidence for this model derives
reinforces this chromatin state. As mentioned previously, from FISH experiments showing greater compaction of the
transcription of individual elements within Ig gene loci cor- IgH locus in pro-B cells poised to undergo V(D)J recombi-
relates well with their availability for V(D)J recombination at nation than in their earlier hematopoietic progenitors.125 B
that stage. In early pro-B cells when D to JH recombination cells deficient in Pax5 are impaired in recombination of the
takes places a “μo” transcript starts upstream of DQ52 and most distal V H regions and show less movement of these re-
proceeds all the way through the JH elements126,127 ; and only gions towards the JH-C locus than normal cells.134 Data from
after DJH rearrangement do VH sterile transcripts appear.120 a recently developed high-resolution FISH method provide
It is currently unclear whether transcription per se is directly a detailed model of the three-dimensional structure of the
linked to recombination or whether the correlation is largely IgH locus, revealing rosette-like structures with central
mediated by similar requirements for gaining access to chro- hubs from which several loops extend.135
matinized DNA.
cis Mediators of Accessibility and Looping
Histone Modifications All previously described properties are dependent on cis-
Posttranslational modifications of histone tails are important regulatory elements within the Ig loci, including classic
epigenetic marks of the chromatin state (also referred to as the promoters and enhancers. Individual promoters are present
“histone code”). Distinct marks correlate well with actively upstream of all V elements in all Ig loci, while the down-
transcribed and inactive (or repressed) gene loci. As these pat- stream D and J elements share a smaller number of promot-
terns hold true for Ig loci as well, active marks (e.g., histone ers. Enhancers are present in each Ig locus, and they are
acetylation) and inactive marks (e.g., the methylation of lysine thought to confer the transcriptional activation of each locus
9 on histone H3, H3K9Me2) correlate well with both tran- at the appropriate stage of B-cell development. Murine κ
scription and recombination accessibility.128 Importantly, one and IgH loci have intronic enhancers in the intron between
particular mark for open chromatin, H3K4Me3, is directly J and C (iEκ and Eμ, respectively), and all three loci have
linked to the RAG recombinase. As discussed previously, his- enhancers downstream of C coding regions. (For example,
tone tails with this modification are recognized by the PHD downstream of the murine IgH locus is a complex of four
domain of RAG2; strikingly, the distribution of RAG2 in Ig enhancers, collectively known as the 3′regulatory region.)
loci matches exactly the pattern of H3K4Me3 (which mark ac- Promoters and enhancers were originally defined based on
cessible gene segments).129 Beyond facilitating the recruitment their role in regulating transcription, but these and other re-
of the recombinase, this histone modification also increases cently reported elements appear to play additional roles in
cleavage activity in vitro.69 Ig gene recombination. Transcriptional activation and the
correlated locus “opening” is mediated by the recruitment of
Methylation transcription factors that in turn recruit histone modifying
Most cytosine residues within CpG dinucleotides are and chromatin-remodeling enzymes. In addition, promoters
methylated in mammalian DNA, but genes that are actively and enhancers regulate transcription through the formation
expressed in a particular cell are generally relatively hypo- of DNA loops, some of which are critical determinants of
methylated in that cell type, implying that DNA methyla- the three-dimensional structure of Ig loci, thereby affecting
tion inhibits transcription. DNA methylation also seems to V(D)J recombination as well as CSR (as discussed below).
inhibit V(D)J recombination. The developmental matura- An enhancer may activate transcription of several genes
tion from pro-B to pre-B cells is associated with progression within a given gene locus, but its effects may be deleterious if it
from a κ locus that is largely methylated, nontranscribed, can activate other nearby genes requiring different patterns of
and nonrearranging to one that is hypomethylated, tran- expression. To prevent enhancer function beyond appropriate
scribed, and rearranging.130,131 Furthermore, methylation of domains, boundary elements known as insulators establish
artificial recombination substrates blocked V(D)J recom- borders between gene loci that are differentially regulated.
bination when transfected into a recombination competent The protein CCCTC binding factor (CTCF) is commonly
B-cell line132 ; V(D)J recombination of a transgenic construct found at insulators and also functions by creating DNA loops.
occurred only when it was unmethylated.133 Methylation and One such insulator apparently lies downstream of a complex
histone acetylation are interrelated; for example, the methyl- of enhancers at the 3′ end of the IgH locus, the 3′-regulatory
CpG-binding protein MeCP2 recruits histone deacetylases, region, where it may protect genes further downstream from
which reduce acetylation of histones. being regulated by the Ig enhancer elements.136 A recently dis-
covered regulatory element with insulator properties is the
Gene Localization in the Nucleus intergenic control region-1 (IGCR1), which participates in
While accessibility allows the RAG proteins to selectively CTCF–dependent looping between the Eμ and 3′ regulatory
bind to appropriate sets of RSSs at each developmental stage, region enhancers.137–139 Based on results of deletion of this re-
V(D)J recombination also requires that a pair of compatible gion, the IGCR1 element suppresses recombination of VH to
gene segments (and their RSSs) are in close physical proxim- D segments not already joined to JH, prevents VH to DJH re-
ity. This becomes a particularly daunting requirement for combination in T cells, and mediates the BCR-induced signal
gene segments > 1 mb apart in linear DNA sequence. Such that terminates sterile VH transcription and recombinational
distant segments are apparently brought close together in the accessibility of VH segments after a productive VDJ recombi-
nucleus, a process of “locus compaction” that loops out large nation leads to expression of a μ protein.

Late RAG Expression: Receptor Editing and mice provided evidence of receptor editing leading to vir-
Receptor Revision tually 100% λ L chain expression.143
Although the RAG genes are generally downregulated by
a signal mediated by the appearance of membrane IgM
at the end of the pre–B-cell stage, RAG gene expression IMMUNOGLOBULIN GENE ALTERATIONS IN
and V(D)J recombination can recur later during “receptor GERMINAL CENTERS
editing” of autoreactive B cells in the bone marrow. One Several days after exposure to an antigen, B cells accumulate
estimate suggests that about 25% of Igs are products of in local lymph nodes, gut-associated lymphoid tissue, and
receptor editing.140 After production of an initial Ig κ . pro- spleen, and begin additional maturation steps in germinal
tein, receptor editing by L chain rearrangement can occur centers (GCs). During the GC response, antigen-driven B cells
three ways: an initial VκJκ junction could be deleted by undergo cycles of proliferation and their Ig genes undergo two
recombination between an upstream V and downstream J unique alterations. 1) Lymphocytes switch from making IgM
on the same chromosome (“secondary recombination,” as to making a new H chain isotype by the process of CSR. This
discussed previously), VκJκ recombination could occur on process introduces dsbs upstream of Cμ in a specific repeti-
the other (allelic) copy of the κ locus, or V λ-Jλ recombina- tive noncoding DNA segment—the “switch region”—and in
tion could be activated. a similar switch region upstream of the new target Cx region;
Most replacement of productively rearranged L or H the DNA between the breaks is then deleted, and the ends of
chain genes likely serves to extinguish an antibody that the remaining chromosomal DNA are rejoined so that the as-
was autoreactive, thus complementing two other mecha- sembled VDJ region now lies upstream of the new Cx gene
nisms to silence autoantibodies: anergization and cell de- (Fig. 6.12). 2) In the other GC-associated gene alteration, the
letion by apoptosis. Early studies with transgenic autoan- affinity of the antibody for its antigen increases by a process
tibodies suggested that anergy or deletion were the main that introduces random mutations in VH and VL—somatic
fates of self-reactive B cells, but these conclusions may have hypermutation (SHM)—and then selects for B cells produc-
depended on the nonphysiologic inability of the cells to ing higher-affinity antibodies. For many years, CSR and SHM
silence the transgenic autoantibody by receptor editing. were considered to be unrelated processes, but several lines of
More recent studies involving autoantibody “knockin” evidence provided hints that they might share some mechanis-
genes—i.e., productive V[D]J recombined genes swapped tic features. First, while dsbs are expected intermediates for the
into the physiologic positions within the IgH or Ig κ gene DNA recombination underlying class switching, DNA breaks
loci using homologous recombination —have shown that in V regions were also detected accompanying SHM. Second
receptor editing is the major mechanism for B cell toler- (and conversely), in addition to the mutations occurring in V
ance.141,142 This conclusion was also supported by a study of regions associated with SHM, mutations were also observed
mice expressing a transgenic antibody against the murine surrounding the recombination junctions of CSR. Third, RNA
Cκ constant region, a model of a self-superantigen; these transcription was found to be required for both processes.
VDJ Iμ Sμ Cμ Iγ Sγ Cγ Iε Cε
Sε
unswitched
DNA + IL-4 + CD40 activation
germline transcript
FIG. 6.12. Switch Regions and Composite
Switch Junctions. The recombination
breakpoints in isotype switch recombi-
nation fall within repetitive “switch” (S)
regions. Stimuli that activate switch re-
combination (IL-4 and CD40 activation in
VDJ Cε the example shown) generally promote
Sμ Sε
transcription across the target S region,
chromosomal initiating just upstream at the “I” exon.
μ−ε product Recombination between Sμ and Sε pro-
duces two composite switch junctions:
+ Sε Sμ Cμ
an Sμ-Sε junction retained in chromo-
somal DNA, and a reciprocal Sε-Sμ junc-
reciprocal tion found in fractions of circular DNA.
Cγ
ε−μ circle Polymerase chain reaction amplification
Sγ across either composite junction can be
used to study switch recombination.

Activation-Induced Deaminase 2. Defective SHM and CSR are observed with inhibi-
A fourth and dramatic link between CSR and SHM was the tion or genetic inactivation of uracil-N-glycosylase
discovery that both processes require the protein known (UNG), an enzyme that removes uracil residues from
as activation-induced deaminase (AID). The gene encod- DNA,150,151 consistent with the idea that CSR and SHM
ing AID (known as Aicda, for activation induced c ytosine involve a step in which AID-produced uracil in DNA
deaminase) was discovered144 by a subtractive strategy de- is removed by UNG. Moreover, B cells from Ung−/−
signed to screen for transcripts that were expressed in a mu- mice accumulate uracils in V and in switch regions in
rine B-cell line when induced to undergo CSR, but that were an AID-dependent manner.152
not expressed in uninduced cells. AID is expressed almost 3. AID isolated from B-lymphocytes can deaminate cytidine
exclusively in GC B cells and in B cells activated in vitro, to uracil in single-strand DNA in vitro153–155 or in double-
though exceptions to this generalization will be discussed strand DNA that is being transcribed in vitro, presumably
later. Mice engineered with a targeted defect in the Aicda because transcription causes localized regions of single-
gene are completely deficient in CSR and SHM. The same strandedness.155–157
defects are seen in patients with a homozygous defect in the 4. AID is found to be associated with IgH genes in vivo in
human Aicda gene, a condition known as the hyper-IgM B cells undergoing CSR,158–160 as assessed by Chromatin
syndrome-2.145 These patients have elevated serum levels ImmunoPrecipitation (ChIP) using an anti-AID
of IgM because their B cells cannot undergo efficient CSR. antibody.
AID is not only necessary for CSR and SHM, but apparently 5. A DNA sequence motif WRC—where W is A or T
sufficient (in a mammalian cell at least), as overexpression (weak Watson-Crick pairs) and R (purine) is A or
of AID in fibroblasts can confer transcription-dependent G—that has been recognized as a hotspot target of
CSR of an artificial switch construct146 and transcription- SHM is also a preferred target for AID deamination in
dependent SHM of a transfected model mutation target vitro,154,157 and the hotspot preference can be altered by
DNA.147 These experiments suggest that AID is the only engineering amino acid changes in the segment of the
B-cell–specific protein required for SHM and CSR. AID is AID protein thought to recognize the hotspot target in
also required for somatic Ig gene conversion in those species DNA.161–163
(e.g., rabbit and chicken) that use that process to somatically In light of these and other observations, the DNA deamina-
diversify Ig genes. tion model for AID action is now almost universally accepted.
As translated from the cDNA, AID is a ∼24 kD 198 amino The AID protein is highly conserved from fish to human, with
acid protein that forms homodimers. AID shows 34% amino all species having approximately 200 amino acids and sharing
acid identity with the RNA editing enzyme APOBEC1, which sequence similarities throughout the protein. In all species,
catalyzes the deamination of a cytosine residue to uracil in AID contains a motif common to the active site of all cytidine
a specific position in the mRNA encoding apolipoprotein B. deaminases: H[A/V]E − X(24–36) − PCXXC. This motif is also
The human Apobec1 and Aicda genes are genetically linked, found in other members of the AID-APOBEC gene family, in-
both lying at chromosome 12p13. Three other APOBEC1- cluding APOBEC2, APOBEC3 (with several distinct paralogs in
related genes on other chromosomes have also been identi- human), and APOBEC4.164 AID is encoded in the five exons of
fied, but are not thought to participate in SHM or CSR. the Aicda gene.
Like APOBEC1, recombinant AID protein has a cyti- Structure-function relationships of AID have been
dine deaminase activity in vitro; it was initially proposed probed by examining cross-species sequence comparisons
that, by analogy with APOBEC1, AID functions by deami- and the effects of natural and engineered mutations in the
nating cytidines in specific RNAs to produce novel edited protein. Remarkably, mutations in the N-terminus of AID
transcripts encoding one or more proteins required for CSR impaired SHM but not CSR, whereas mutations or dele-
and SHM. However, no evidence of AID-dependent edited tion in the C-terminus selectively impaired CSR,165–167 sug-
RNAs has been reported, nor does AID deaminate RNA gesting the possibility that the N- and C-terminal regions
cytidines in vitro. Instead, current evidence indicates that of the protein contact specific cofactors required (respec-
AID acts on DNA, deaminating cytidines to uracil—in V tively) for SHM and CSR function. Other functional fea-
regions for SHM, and in switch regions for CSR. The result- tures of the protein include a dimerization domain, several
ing uracils would be then recognized as both Watson-Crick phosphorylatable residues, segments affecting nuclear lo-
mismatches and abnormal DNA bases by the cell’s genetic calization, and target sequences for association with other
surveillance machinery, triggering error-prone repair of V proteins.
regions to produce SHM, or DNA cleavage to mediate CSR. Although AID was initially recognized for its participa-
This DNA-deamination model for AID is consistent with tion in SHM and CSR in GC cells, more recent reports have
various properties reported for this protein in vitro or in reported several other roles for this protein, both beneficial
cells. and deleterious.
1. In transfected Escherichia coli cells, AID mutates cy- 1. Epigenetic methylation. Selective DNA methylation of
tidine to uracil in DNA, a result that would be unex- cytidines is an epigenetic mechanism that participates
pected by the RNA-deamination model, as bacteria in the regulation of transcription, and the methylation
presumably lack the specific mammalian RNA targets pattern of DNA is normally replicated when DNA is rep-
predicted by that model.148,149 licated prior to cell division. In primordial germ cells of

early embryogenesis, a global erasure of methylation oc- increased off-target mutations and increased c-myc-IgH
curs as part of the reprogramming to pluripotency, and translocations. Perhaps to protect against such effects, miR-
AID apparently participates in this process.168,169 155 is itself upregulated by the same signals that induce
2. Germline mutation. AID can deaminate methylcytosine Aicda transcription in B cells.
to thymine, which unlike uracil is a natural DNA base.
This reaction in germ cells may contribute to the most
frequent germline point mutation observed in mam- Posttranslational Regulation
mals: the transition from CpG to TpG.170 The level of AID activity in the nucleus is affected by its dis-
3. Genomic protection. AID and other APOBEC family tribution between cytoplasm and nucleus, by the protein
members protect cells against retroviral infections and half-life, and by phosphorylation at various positions in the
against the spread of endogenous retroviruses.171–173 protein.
4. Tolerance, apoptosis. AID appears to be important for AID was found to shuttle between nucleus and cyto-
the establishment of B-cell tolerance in humans and plasm.189 Most of the protein is cytoplasmic and only a small
mice, in that the low levels of AID expression observed fraction is in the nucleus, where it can act on DNA to cause
in immature and transitional B cells appear to be nec- CSR and SHM as well as “off-target” deleterious effects.
essary for suppressing the appearance of autoantibod- Movement between cytoplasm and nucleus is controlled by at
ies.174–176 This effect may be related to the apparent least three independent mechanisms: active nuclear import
requirement of AID for normal levels of apoptosis, mediated by an N-terminal nuclear localization signal,190 ac-
which contribute to the elimination of B cells express- tive nuclear export mediated by a C-terminal nuclear export
ing autoantibodies. signal,189,191 and a cytoplasmic retention mechanism appar-
5. Oncogenic mutations and translocations. Soon after AID ent when active import and export mechanisms are both
was discovered, it was observed that overexpression of blocked. Two proteins that may facilitate AID import inde-
the protein caused tumors in transgenic mice,177 and pendently of the nuclear localization signal are GANP192 and
even normal expression of AID in B cells contributes CTNNBL1.193 Nuclear-cytoplasmic distribution affects AID
to tumorigenic translocations and oncogenic muta- stability, because in the nucleus AID is targeted for ubiqui-
tions.178 A notable example of AID-stimulated trans- tination and proteasomal degradation, shortening its half-
location is the recurrent c-myc/Ig translocation seen in life.194 Conversely, in the cytoplasm, the chaperone protein
Burkitt lymphoma and murine myelomas. Other can- Hsp90 protects AID from proteasomal degradation.195
cers where AID plays a role through mutation include In addition to ubiquitination, AID is also subject to post-
intestinal and lung cancer. translational phosphorylation at several distinct residues
in the protein, including Ser3, Thr27, Ser38, Thr140, and
Based on studies of AID + / − heterozygous mice expressing Tyr184. Of these, the functional relevance of the Ser38 site
roughly 50% of normal AID, the level of AID activity seems is best understood. Phosphorylation of AID Ser38 is essen-
to be the limiting factor for both SHM and CSR, but is also tial to recruit replication protein A (RPA), a trimeric single-
limiting for oncogenic translocations and mutations.179,180 strand DNA binding protein159,196 that may be required to
Therefore, AID activity must be regulated robustly to bal- stabilize a single-strand DNA target for AID. Preparations of
ance between its multiple beneficial and deleterious actions. AID purified from transfected E. coli or from nonlymphoid
The regulation of AID activity is extremely complex, in- eukaryotic cells lack Ser38 phosphorylation; they can deam-
volving multiple levels of control, and is still incompletely inate single-strand DNA in vitro, but are inactive in an assay
understood. for deamination of double-strand DNA transcribed in vitro
by T7 polymerase. Activity in the latter assay requires AID
Transcriptional and Posttranscriptional Regulation of that is phosphorylated at Ser38 and associated with RPA. In
AID Expression the AID protein, Ser38 lies within a conserved phosphoryla-
Early studies showed that AID expression in B cells is up- tion site [RRXX(T/S)] for the cyclic-AMP dependent pro-
regulated by IL-4 and CD40 engagement. In more recent tein kinase A (PKA), and multiple experiments have con-
investigations, four DNA regions that regulate Aicda gene firmed that PKA phosphorylates AID at Ser38 in vitro and
expression have been identified. They include an upstream physiologically in vivo.196,197 The physiologic importance of
enhancer responsive to T-cell signals, a promoter, a regula- Ser38 phosphorylation is strongly supported by observations
tory region just downstream of exon 1, and an additional of the effect of mutating this serine to alanine. The AIDS38Α
enhancer downstream of the Aicda gene.181–186 protein has essentially normal activity in deaminating sin-
The microRNA miR-155 has been shown to suppress AID gle-stranded DNA in vitro, but severely reduced activity in
expression.187,188 MicroRNAs are short (21 to 23 nt) RNAs deaminating transcribed double-strand DNA in vitro.
that hybridize to complementary sites in numerous target Less is known about the other sites of AID phosphory-
RNAs, triggering their degradation or inhibiting their translation. At Thr140, alanine replacement causes more modest
lation. A target for miR-155 lies in the 3′ untranslated region inhibition of AID activity than Ser38Ala, preferentially af-
of the AID transcript. AID expression was increased in miR- fecting SHM in vivo, but without significant effect on in vitro
155–deficient mice and in mice whose Aicda gene was mu- catalytic activity.198 In contrast to Ser38 and Thr140, phos-
tated to disrupt the 3′UTR target sequence. In both cases, phorylation of Ser3 apparently suppresses in vivo AID ac-
consequences of the increase in AID expression included tivity; alanine replacement at this position increases SHM,

CSR, and oncogenic translocation, also without affecting in expressing IgM with germline (unmutated) versions of Ig
vitro catalytic activity.199 variable regions resulting from gene rearrangements that
Clearly, multiple mechanisms regulate the level of activ- occurred prior to immunization. Because of the diversity of
ity of AID in B-cell nuclei, but these mechanisms cannot available V H, D, JH, V L, and JL sequences as well as the impres-
explain how—within nuclei—AID acts on Ig V genes and sive recombinational potential described previously, some B
switch regions at a much higher frequency than on the rest cells will express Ig molecules capable of binding the anti-
of the genome. Targeting mechanisms differ somewhat be- gen with modest affinity. Antigen binding stimulates these B
tween SHM and CSR, and will be discussed in the context cells to proliferate and to move to lymphoid follicles, where
of those processes. they form GCs. In the GC environment, AID is expressed
and SHM machinery is activated, generating random mu-
tations in the Ig genes of stimulated GC B cells. Many of
Somatic Hypermutation these mutations reduce the affinity of the encoded antibody
The hypothesis that antibody genes inherited in the germline for antigen, 201 and some may induce autoantibody specifici-
might be subject to somatic mutation in lymphocytes during ties (i.e., the ability to bind to self-molecules202). As clear-
the life of an individual was suggested as an explanation for ance of the antigen lowers the antigen concentrations, only
the diversity of antibodies several years before recombinant the cells displaying high affinity for antigen will be stimu-
DNA technology became available to clarify the role of V(D)J lated effectively; cells displaying lower-affinity antibodies
recombination. Persuasive evidence for somatic mutation was or antibodies with affinity for self-antigens are subjected to
reported in the 1970s: analyses of Vλ1 amino acid sequences programmed cell death (“apoptosis”).203,204 The preferential
of murine myeloma antibodies showed many instances of proliferation of the high-affinity cells and their maturation
a particular prototype sequence, plus several variants con- to secreting plasma cells causes an increase in the average
taining unique single amino acid substitutions that could affinity of the antibodies in the serum. Some high-affinity
be explained by single nucleotide changes. The prototype cells become memory cells, persisting long after the initial
was interpreted as reflecting the germline sequence, with antigen exposure, ready to respond to a subsequent antigen
the variants arising by somatic mutation.200 Subsequent in- exposure with rapid production of high-affinity antibody.
vestigations at the DNA level revealed myeloma V region In this model, the driving force for affinity maturation—
sequences that deviated from their germline counterparts, analogous to natural selection in the evolution of species—is
verifying the principle of somatic mutation. selection for high-antibody affinity in the face of falling an-
Somatic mutations are much rarer in IgM than in anti- tigen concentration. The importance of this selective force
bodies with “switched” isotypes (IgG, IgA, and IgE) made is suggested by the observation that affinity maturation can
by B cells that have been exposed to AID in GCs, but they do be inhibited by repeated injection of antigen (which removes
occur; antibodies with “switched” isotypes without muta- the selective pressure for high affinity) 205 or by overexpres-
tions are also found. These observations suggest that though sion of the antiapoptotic protein Bcl-XL (which allows sur-
both SHM and CSR normally occur in GC B cells, the two vival of B cells expressing low affinity antibody).206
processes are unlinked.
Cellular Context of Somatic Mutation
Role of Hypermutation in Immune Responses Each GC appears to be populated by a small number of
To understand the role of SHM in the antibody response, antigen-specific founder B cells207 and an unusual Thy-1
several groups have studied the extent of Ig gene mutation negative T-cell population, also antigen-specific, 208 The
at different times after the immunization of mice. Studies of GC environment promotes contact between the B cell and
the responses to p-azophenylarsonate (Ars), phosphorylcho- follicular dendritic cells— which store, process, and pres-
line, influenza hemagglutinin, oxazalone, and several other ent antigen—and T-lymphocytes, which activate somatic
antigens have all indicated that the initial response after mutation in part via CD40-CD40Ligand interaction.209 In
primary immunization is established by antibodies show- a widely accepted model of GC function, SHM occurs in a
ing no somatic mutation. About 1 week after immunization, subpopulation of B cells known as centroblasts. These cells
mutated sequences begin to be observed, increasing during proliferate in the “dark zone” of the GC and bear character-
the next week or so. Booster immunizations yield sequences istic surface markers including IgD, CD38, and the recep-
showing additional mutations. tor for peanut agglutinin. Proliferating GC centroblasts give
Many hybridomas made late in the immune response pro- rise to nondividing centrocytes in the “light zone” of the
duce mutated antibodies with a higher antigen affinity than the GC; there centrocytes are programmed for apoptosis unless
unmutated (sometimes loosely called “germline”) antibodies they are rescued by follicular dendritic cell–presented anti-
made early after immunization. The shift to higher affinity is a gen and T-cell activation via CD40 engagement. Selection
phenomenon long recognized at the level of (polyclonal) anti- for high-affinity antibodies occurs because cells expressing
sera and has been termed “affinity maturation.” This phenom- high-affinity antibodies are most efficiently rescued from
enon can now be explained as the result of an “evolutionary” apoptosis. Surviving centrocytes may return to the dark
mechanism selecting antibodies of progressively higher affin- zone to undergo several successive cycles of mutation and
ity from the pool of randomly mutated V sequences. proliferation followed by selection. This model is supported
According to this model, at the time of initial anti- by direct observation (by two-photon microscopy) of B cells
gen exposure an animal has a set of naïve B-lymphocytes moving between light zone and dark zone.210 Migration of

B cells to follicles or GC zones is thought to be controlled analyzing abnormalities in SHM that are observed in vari-
by a complex system of chemotactic receptor /ligand pairs ous mutant B cells, Di Noia and Neuberger150 formulated a
including CXCR5/CXCL13, CXCR4/SDF1, CCR7/CCL9 (or model for SHM that has explained this question and has
CCL21), S1P1/sphingosine-1-phosphate, and EBI2/its lipid gained wide acceptance (Fig. 6.13). The model proposes that
ligand.211 Although certain features of this classical model after AID-catalyzed deamination creates a uracil residue in
have been challenged, the broad outlines have received the target DNA, the possible outcomes depend on how the
strong support from studies in which cells in the dark zone resulting mismatch is resolved. 1) The U:G mismatch may be
or light zone of individual GCs can be marked by a photo- replicated as described above, resulting in what are known
activatable green fluorescent protein (GFP) and then fol- as phase 1A mutations. 2) The uracil base may, before repli-
lowed for several hours using two-photon imaging as the cation, be excised by UNG, creating an abasic site. Normally,
cells migrate through a mounted lymph node.212 the creation of such abasic sites is the first step of the base
The notion that GC B cells compete to avoid apoptosis on excision repair pathway, in which subsequent steps remove
the basis of antigen affinity of their BCR has been has been the sugar-phosphate backbone, leaving a single nucleotide
supported by a study that directly compared caspase activa- gap that is then restored to a C:G bp by DNA polymerase
tion (an apoptosis marker) in B cells expressing transgenic β and DNA ligase. If the DNA replicates before the abasic
antibodies with higher versus lower affinity for the same an- site is repaired, the strand with the abasic site may directly
tigen; the lower-affinity cells were found to undergo a signifi- engage translesional polymerases (which are error-prone, as
cantly higher rate of apoptosis.213 The mechanism by which discussed below) to insert an unpaired nucleotide (i.e., any
high-affinity antigen binding selects for survival is not fully nucleotide) opposite the abasic lesion, leading to phase 1B
understood. One possibility is that a higher-affinity BCR mutations (see Fig. 6.13). 3) The original U:G mismatch—
directly delivers a stronger activation signal to the B cell, or possibly the abasic site created by UNG action—may be
inhibiting apoptosis. However, a second possibility is that a recognized by the mismatch repair (MMR) system of the
higher-affinity BCR could capture antigen more efficiently, cell. MMR triggers excision of the DNA strand for several
enabling stronger antigen presentation to GC T cells, which nucleotides surrounding the mismatch, and this strand is
could then deliver stronger survival signals to the B cell via then resynthesized by polymerase β, in the case of faithful
secreted or surface protein interactions. Consistent with the repair, or by error-prone polymerases, in the case of SHM.
second model, experiments using two-photon microscopy Error-prone repair inserts mispaired bases, which may be-
have documented that that B cells from mice immunized with come fi xed on one strand by replication, creating mutations
a fluorescent antigen can capture that antigen from follicu- (designated phase 2) from both A:T and G:C bps at some
lar dendritic cells in the GC.214 To test whether B-cell antigen distance from the position of the original U:G mismatch. If
capture might confer survival independently of the BCR, in- a mispaired base is recognized by MMR before replication,
vestigators engineered B cells in ovalbumin-primed mice to a new cycle of MMR may be initiated, extending mutation
deliver ovalbumin via a surface lectin instead of via the BCR; even farther from the initial deamination event.
they observed that the internalized ovalbumin antigen was
capable of conferring a B-cell survival advantage in the GC Role of Uracil-N-Glycosylase in Somatic Hypermutation.
in the absence of BCR engagement.212 However, effective BCR The UNG gene encodes two proteins that differ in their N
signaling can enhance antigen processing and presentation,215 termini as a result of alternative promoters that generate
so internalization/presentation and BCR signaling may work different initial coding exons. UNG1 is expressed in mito-
together to promote GC B-cell survival and mediate selection chondria, whereas UNG2 is nuclear. Hydrolytic deamina-
for high affinity. tion of cytosines to uracil occurs at a significant rate in all
Most experiments on SHM have focused on GC cells, eukaryotic and prokaryotic cells, and misincorporation of
as does the discussion in this chapter. However, T-cell– dUTP during replication further contributes to the load of
independent SHM may also occur in a population of less uracil in DNA. UNG2 plays a major role in mitigating this
mature cells, which may populate the splenic marginal zone load by initiating faithful base excision repair.
and which may increase the repertoire of circulating lym- In SHM, the faithful repair of uracil is somehow sub-
phocytes prior to antigen exposure, especially in young inverted to introduce mutations. Ung − / − mice and human
dividuals, 216,217 or perhaps function in tolerance induction, patients with a rare form of HyperIgM immunodeficiency
as discussed previously. Also, mice lacking histologically due to UNG mutations have similar immunologic phe-
detectable GCs as a result of lymphotoxin-α deficiency are notypes. First, as discussed in the following, Ung − / − in-
capable of SHM and affinity maturation.218 T-cell–indepen- dividuals are profoundly defective in CSR, as expected
dent antigens can induce a low level of SHM in B cells.219 if this process requires AID-catalyzed deamination of
cytidine followed by UNG-catalyzed removal of uracil.
Molecular Mechanism of Hypermutation Although the frequency of mutation is roughly normal in
AID deaminates cytidine to uracil, an analog of thymine. UNG-deficient individuals, mutations at C are almost ex-
Thus, if replicated before repair, an original C:G base pair clusively transitions of C to T (phase IA). As suggested by
would, in the daughter cell receiving the uracil-bearing DNA Figure 6.13, Ung − / − individuals would not create the aba-
strand, mutate to a T:A. But it was initially not clear how sic sites that lead to C → G and C → A mutations by rep-
cytidine deamination could affect A:T base pairs, which are lication, though some mutations of these kinds could be
targeted in 50% to 60% of mutations observed in SHM. By produced by MMR. Indeed, the frequencies of mutations

C
G T
A
AID
Phase IA
T
A T
A
Replication
MSH2-MSH6
U Exo I
T
A G A
G
POL η
UNG Ligase IV
C
G T
A
Replication
T
A GC C
G
G
T T APE1, MRN?
Phase II
A A
C G
Phase IB
G C
T
A
G
A
T T
A
GC A
T
G
C T
A
FIG. 6.13. Mechanistic Model of Reactions Triggered by Activation-Induced Deaminase (AID) in Somatic
Hypermutation and Class Switch Recombination (CSR). The graphic at the top depicts a small region in a VH gene
showing a targeted C:G basepair and a nearby T:A pair. After AID-catalyzed deamination of a C residue, the DNA may
be subjected to various modifications as shown, leading to mutations indicated by nucleotides in gray rectangles.
The model is described in the text. The Phase II mutations occur when mismatched bases incorporated by an error-
prone polymerase are replicated before they can be corrected by MMR. The same mechanism explains two switch-
region–associated events occurring during CSR: DNA strand breaks (see center panel at bottom) and mutations..
from A, T, and G in Ung − / − individuals are normal, effects on SHM and CSR, mutations in MMR genes, espe-
apparently resulting from MMR activated by the U:G mis- cially MSH2 and MLH1, underly hereditary nonpolyposis
match in many cells. colorectal cancer. Recent efforts in several laboratories have
led to in vitro reconstitution of mammalian MMR with puri-
Mismatch Repair in Somatic Hypermutation. MMR is a fied components, enabling powerful analysis of this complex
highly conserved mechanism that detects abnormalities in mechanism.220 In addition to the MutS and MutL proteins,
DNA, including mispaired nucleotides and abnormal bases, the system requires the following components: Exo1 to ex-
and repairs them. The eukaryotic MMR system has two main cise the gap; replication protein A, which binds to the single
components. MutS binds tightly and specifically to DNA strand DNA in the gapped region (and is known to bind to
defects and recruits MutL. The Mut complex then activates phosphorylated AID, as discussed previously); proliferating
a latent nuclease activity to remove a segment of the DNA cell nuclear antigen (PCNA), which promotes processivity
strand including the mismatched base; this gapped strand by encircling DNA in a sliding ring clamp and by recruiting
is then resynthesized by a DNA polymerase. Mammals have other components; DNA polymerase δ ; DNA ligase I; and
several MutS homologs, including three reported in somatic several other proteins.
cells—MSH2, MSH3, and MSH6—which exist in the cell as Knockouts of MSH2, MSH6, and Exo1 have been stud-
heterodimers. MSH2-MSH6 (also known as MutSα) is the ied by several laboratories and show consistent decrease in
most abundant form and is specialized for recognizing sin- B-cell SHM from A:T bps (with a corresponding increase
gle base gaps or mismatches, while MSH2-MSH3 (MutSβ) in the percentage of mutations from G:C bps).221,222 These
recognizes larger gaps and insertion/deletion loops. The results, in the context of the model of Figure 6.13, suggest
mammalian MutL proteins are heterodimers consisting of that MMR triggered by the MSH2-MSH6 heterodimer pri-
MLH1 paired with PMS1, PMS2, or MLH3. Apart from their marily introduces mutation at A:T bps. The predilection for

mutating A:T bps matches the activity of polymerase η, as in nonlymphoid cells is accurately repaired. One apparent
discussed in the next section, implying that this polymerase clue is that some TLS polymerases are upregulated in B cells
may be the one most frequently engaged by the MMR ma- undergoing SHM. In addition, when the sliding clamp pro-
chinery in repairing AID-generated lesions. Indeed, MSH2- tein PCNA is monoubiquitinated at lysine residue K164, it is
MSH6 is capable of binding to a U:G mismatch, MSH2 can known to bind and activate TLS polymerases including poly-
bind to polymerase η in cell extracts, and MSH2-MSH6 can merase η. When this ubiquitination is prevented, either by
stimulate the activity of polymerase η in vitro.223 knockout of the specific E3 ubiquitin ligase of PCNA or by a
K164R mutation at the ubiquitination site in PCNA, SHM is
Error-Prone Polymerases. Although avoidance of error is a abnormal. PCNA Κ164Ρ mice were found to have a dramatically
high priority for most DNA replication, error-prone DNA syn- decreased ability to mutate A:T sites during SHM, similar
thesis served an important function long before SHM evolved to the phenotype of the polymerase η knockout, as though
to mutate Ig genes; indeed, error-prone polymerases have been polymerase η requires monoubiquinated PCNA in order to
most thoroughly studied in E. coli and yeast. These polymerases participate in SHM.235 However, these clues do not fully ex-
are useful in all cells for replication of DNA containing focal plain why Ig genes in GC B cells are less faithfully repaired
lesions that would block replication by stringent high-fidelity than other genes in the same cells. This question is part of
polymerases. However, the tolerance of these error-prone poly- the larger issue of mutational targeting, as discussed below.
merases to abnormalities in DNA structure is accompanied by
tolerance of non–Watson-Crick basepairing in the replication Targeting and Distribution of Mutations
of normal DNA and by absence of DNA proofreading activ- The mutation rate of Ig genes in GC B cells undergoing SHM
ity. More than a dozen of these polymerases have evolved in may reach as high as 10−3 mutations/bp/generation, or about
eukaryotes, each specializing in different aspects of “transle- 10 6 times higher than the normal genomic mutation rate; 236
sion DNA synthesis” (TLS) and showing differing spectra of this elevated rate could be lethal to B cells if mutations were
infidelity in replication of normal DNA. Clear evidence for par- not carefully targeted specifically to Ig genes. Several other
ticipation of several TLS polymerases in SHM has come from genes highly expressed in GC B cells were also found to be
comparing the known in vitro activities of a particular poly- mutated at lower levels (e.g., Bcl6 and Cd95237), though sev-
merase with the SHM abnormalities seen in mice or humans eral other genes expressed in GC B cells at comparable levels
with mutations in the corresponding polymerase gene. are not mutated. Apparently, some of the features that target
Polymerase η characteristically inserts mismatched bases SHM to Ig genes may be shared by other genes.
opposite T nucleotides, and individuals lacking this poly- Recently, a genome-wide analysis has made it clear that
merase show the predicted abnormality: decreased mutations AID binds much more widely to the genome than was pre-
at A:T base pairs. This was first demonstrated in patients with viously appreciated. When chromatin from mouse B cells
xeroderma pigmentosum variant disease, whose defects in incubated with lipopolysaccharide (LPS) and IL-4 was ana-
polymerase η subject them to sunlight-induced skin cancers in lyzed by ChIP with an anti-AID antibody, 5910 genes were
addition to their abnormal SHM; a similar defect in mutations found by deep sequencing of the immunoprecipitated DNA
from A:T bps was subsequently shown for polymerase η knock- (ChIP-seq),160 including many previously described AID
out mice.224,225 Polymerase η binds to MSH2-MSH6, as men- targets. Although this is a large number, it is still a small mi-
tioned previously, and is upregulated in cells undergoing SHM. nority of the genome. As judged by the number of sequence
In addition to polymerase η, the TLS polymerase REV1 tags recovered, AID binding to the IgH locus was substan-
has been shown to participate in SHM. REV1 is known to tially higher than binding to any other region. A sample of
preferentially insert cytosine residues opposite uracil or abasic genes identified as AID-binding by ChIP-seq showed sig-
sites, and Rev1 knockout B cells were found to be impaired in nificant mutation frequency, whereas control genes did not.
C to G mutations, especially in the nontranscribed (coding) The AID-binding genes were associated with high levels of
strand.226 Polymerase ζ, 227,228 polymerase κ,229 and polymerase mRNA transcription (by RNA-seq), by a chromatin mark
θ230–232 knockouts have also been reported to show altered commonly associated with transcription—histone H3 tri-
SHM patterns, but these effects are smaller or not seen in all methylated on Lys4 (H3K4me3)—and by density of Pol II
circumstances. binding. However, only the IgH locus, with its uniquely
Error-prone repair apparently operates in competition high levels of AID occupancy and mutations, bound to RPA;
with faithful repair carried out by DNA polymerase β. The this binding was AID-dependent and was diminished in the
human B-cell line BL2 is known to undergo SHM, but in S38A and T140A mutants of AID described previously.
subclones of this line with higher SHM proficiency, poly- The difference between AID-targeting and SHM fre-
merase β levels were very low; overexpression of polymerase quency has been reinforced by analysis of SHM in panels of
β in a proficient subclone suppressed SHM activity.233 non-Ig genes isolated from murine Peyer patch B cells.238 All
Compared to wild-type B cells, polymerase β – deficient B the non-Ig genes were mutated at levels much lower than the
cells (developing in a wild-type recipient) showed increased 2.2×10 −2 mutations per bp detected near JH4, but substan-
switch region mutation after induction of CSR in vitro.234 tial numbers of mutations were found at Bcl6, Pim1, CD79B,
A fundamental question about the action of error-prone and several other previously reported targets of SHM. The
polymerases is how they are specifically engaged for SHM of mutations were almost completely AID-dependent, as in
Ig genes, given that most spontaneous cytosine deamination cells from Aicda −/ − mice, all genes showed very low mu-

tation frequencies that were barely above the sequencing in vitro only when the DNA is transcribed. One hypothesis
error rate. However, different genes varied widely in the way suggests that DNA near the advancing RNA polymerase
their mutation rates were affected by the double knockout complex becomes negatively supercoiled or partially sin-
Ung −/ − Msh2−/ −, in which only phase 1A mutations should gle-stranded, which might facilitate access of the DNA
be possible. At one extreme, c-myc, which is known to mu- to AID. 244 Another explanation, involving R loops, is de-
tate at a very low rate in GC B cells, showed a mutation rate scribed below. Evidence for patches of single-stranded
in the Ung −/ − Msh2−/ − cells almost 17-fold higher than in DNA in transcribed V regions undergoing SHM has been
the wild-type, as though almost all AID-induced deamina- obtained using a bisulfite technique.245
tion in the wild-type B cells had been faithfully repaired by In vivo dependence of SHM frequency on transcrip-
UNG- and Msh2-dependent mechanisms. In contrast, Bcl6 tional activity has been confi rmed by several fi ndings. In
mutations were only 1.3-fold higher in the double knockout, one study, SHM was studied in knockin mice with a pre-
suggesting that faithful repair of this gene was dramatically recombined V HDJ H region driven by either of two V region
less active. promoters; SHM rates in these knockin strains were highly
Thus SHM susceptibility depends not only on differ- correlated with transcription driven by the different pro-
ential AID targeting across the genome but also on dif- moters.246 In another study, a tetracycline-inducible GFP
ferential ratios of error-prone versus faithful repair. The reporter gene engineered with a stop codon was stably
distinction between these two variables (deamination and transfected into a B-cell line. The rate of reversion of the
repair) has only recently been recognized, so it was not stop codon by SHM (allowing GFP expression) was found
taken into account in many earlier studies described in to be directly related to the transcription rate, as regulated
the following. by a tetracycline analog. 247 Other studies examining SHM
in models where transcription was altered by mutating en-
RNA Transcription. A relationship between SHM and RNA hancers have supported the relationship between SHM and
transcription is suggested by the observation that unrear- enhancer-induced transcription. Some non-Ig enhanc-
ranged V H and Vκ genes are generally neither transcribed ers could support SHM in a stably transfected cell line, 248
nor mutated, but become susceptible to both processes when but others could not, leading to the suggestion that spe-
V(D)J recombination moves them close to their (respective) cific elements in enhancers might confer susceptibility to
intronic enhancers, Eμ and iEκ. In contrast, the λ locus SHM. One candidate element is the sequence CAGGTG,
lacks an enhancer between J and C; unrearranged Vλ re- which is a target for E-box transcription factors and which
gions are transcribed in B cells 239 and can be mutated.240 The is found in several Ig enhancers. When a murine κ en-
Ig coding sequences are apparently not specifically required hancer containing CAGGTG was linked to a GFP gene
to target SHM, as transgenic V coding sequence can be re- inactivated by a premature stop codon, stable transfec-
placed by a human β -globin gene or prokaryotic neo or gpt tants of chicken B-cell line DT40 were found to produce
gene without affecting the hypermutation rate downstream mutants that expressed GFP, but a single mutation in the
of the promoter. CAGGTG motif prevented SHM without changing tran-
Mutations around a given V region are distributed in scription. 249 Inactivation of the E-box transcription factor
a domain that begins roughly 100 to 200 bps downstream E2A in DT40 cells was found to reduce SHM of endogenous
of the promoter, extends for ∼1.5 to 2 kb downstream, and Ig genes without significant effects on Ig gene (or AID)
then tapers off long before the RNA transcriptional termi- transcription.250
nation. This distribution led to the hypothesis that after However, some conflicting conclusions have been drawn
transcriptional initiation, a “mutator factor” attaches to about enhancer-dependence of SHM depending on whether
the transcriptional machinery, attacks DNA as the tran- experiments investigated transgenes versus constructs
scription complex moves downstream, and eventually falls created in the endogenous context using homologous re-
off, so that further transcription proceeds without muta- combination (i.e., knockouts or knockins). For example,
tions.241 Consistent with this model, a VκJκ-Cκ transgene the downstream IgH enhancers HS3b and HS4 were found
bearing a second Vκ promoter engineered upstream of the important for SHM in transgenes251 but dispensable in the
Cκ region was found to incur mutations over a second do- context of endogenous genes.252
main extending into the Cκ region, in addition to the usual In one case, transcriptional enhancer activity has been
V region mutations.242 Conversely, the insertion of an ir- clearly separated from an associated SHM stimulating ac-
relevant 2 kb DNA fragment between a Vκ promoter and tivity. Just downstream of the chicken Ig λ 3′ enhancer lies
the leader (signal peptide) exon prevented mutation within a DNA segment whose deletion in the DT40 B-cell line se-
the Vκ transgene, which now apparently lay downstream verely impaired SHM without dramatically affecting tran-
of the mutational domain.243 (Mutations also occur in scription; this segment could confer SHM when inserted
domains surrounding repetitive switch regions upstream into a non-Ig locus.253,254
of CH genes. The frequency of these mutations is similar To summarize, it appears that transcription is necessary
to that in V regions, but the domains are larger, correlat- but not sufficient for targeting hypermutation, but the ad-
ing to some extent with the length of the switch region, ditional elements required for SHM have not been defi ned
as discussed in the following.) The key mutator factor is as of this writing (though E2A function may be among
apparently AID, which can deaminate double-strand DNA them).

Chromatin Marks. A parameter related to transcription sequences known as switch (or S) regions lying 5′ of each
of Ig genes (i.e., their context in chromatin) has also been CH region (except C δ). While most switch breakpoints fall
studied in relation to SHM targeting. Culture of a B-cell line in the S regions, some are in nearby nonrepetitive DNA.
under conditions that upregulate SHM (in the presence of The S region of the mouse μ gene, S μ , is located about 1 to
T cells and anti-IgM) led to increased acetylation of histone 2 kb 5′ to the C μ coding sequence and is composed of nu-
H3 and H4 at V region but not C region DNA, as assessed merous tandem repeats of sequences of the form (GAGCT)
by ChIP analysis 255 paralleling the distribution of mutations. n(GGGGT), where n is usually 2 to 5 but can range as high
The increased histone acetylation was not a consequence of as 17. All of the S regions of downstream isotypes include
AID action as it occurred when AID expression (and SHM) pentamers similar to GAGCT and GGGGT embedded in
was inhibited by antisense treatment; moreover, when cells larger repeat units rather than precisely tandemly repeated
were treated with the deacetylase inhibitor trichostatin A, as in S μ . In support of the critical role of S regions for CSR,
the C region was both acetylated and subjected to SHM. knockout of Sγ1 by homologous recombination essentially
Other histone modifications that may be correlated with abolished expression of IgG1 from that allele, 259 and mice
SHM are phosphorylation or monoubiquitination of histone with a deletion of S μ were also dramatically impaired in
H2B.256,257 CSR. 260,261
A switch recombination between, for example, μ and ε
Hotspot Focusing. Mutations occurring within V region genes produces a composite Sμ-Sε sequence (see Fig. 6.12).
genes expressed in vivo in B cells may be highly selected for From a comparison between the sequence of an Sμ-Sε com-
antigen-binding function of the expressed antibody. To ana- posite switch region and the sequences of the germline Sμ
lyze the spectrum of mutations produced by SHM unbiased and Sε, one can localize the exact recombination sites be-
by selection, investigators have studied nonproductively re- tween Sμ and Sε that occurred in each allele. Such compari-
arranged VDJ alleles or “passenger” transgenes engineered sons have indicated that there is no specific site, either in
with stop codons to prevent expression as a protein. In Sμ or in any other S region, where the recombination al-
these genes, mutational “hot spots” as well as “cold spots” ways occurs, although clusters of recombination sites have
have been recognized, apparently due to local sequence been reported at two specific regions within the tandem re-
features that may promote or suppress somatic mutation. peats of murine Sγ 3 region.262 Thus, unlike the enzymatic
The consensus sequence WRC (i.e., [A/T][G/A][C]) is the machinery of V-J recombination, the switch machinery can
most consistent hotspot for mutation, presumably reflect- break and join sequences in a broad target region, and as
ing the predilection of AID for in vitro deamination of this the recombination targets are in intronic DNA, there are
sequence. It is possible that evolution has concentrated mu- no reading frame complications. Often both IgH alleles in a
tational hot spot frequencies in CDR regions to enhance the single cell undergo switching to the same downstream iso-
potential for diversity generation in the parts of the protein type. Some alleles undergo sequential switching events; for
critical for antigen contact.258 example, a common pathway to IgE expression is an initial
The previous discussion has identified several factors μ → γ1 switch, followed by a CSR between the composite
influencing the targeting of SHM, a process in which a Sμ-Sγ1 switch region and the S ε region.263 Sγ1- Sε switching
unique triggering event—AID-dependent deamination of may even occur independently of Sμ.264 Although most CSR
DNA—is followed by a cascade of other events that depend occurs as a deletion within a single IgH allele, switching be-
on mechanisms common to most cells. All of these steps, tween two allelic chromosomes was detected at a frequency
including the AID-dependent trigger, can be affected by of roughly 7% to 10% in mouse and rabbit.265,266
biologic parameters common to all cells, such as transcrip- DNA fragments excised by switch recombination have
tion, enhancers, epigenetic state, DNA repair mechanisms, been cloned from fractions of circular DNA isolated from
etc. Many of these same mechanisms affect CSR as well as cells actively undergoing isotype switch recombination. Thus,
SHM. Recently, three additional biologic parameters have at least some of the excised DNA segments ligate their ends to
been found to influence AID-triggered events in CSR: RNA form “switch circles”; these contain composite switch junc-
polymerase stalling, the RNA splicing in spliceosomes, and tions that are in theory reciprocal to the composite switch
RNA degradation by exosomes. It is possible that these three junction retained on chromosomal DNA (see Fig. 6.12).
processes impinge on both SHM and CSR, but because they Because switch circles are not linked to centromeres and do
were discovered in the context of CSR, they are discussed in not apparently contain origins of replication, they are not ef-
the following section. ficiently replicated. Therefore, they are not found in cells that
have divided many times after switching (e.g., in myelomas
or hybridomas).
Heavy Chain Switch Many composite switch junction sequences show muta-
Switch Regions and Switch Junctions tions near the recombination breakpoint when compared to
As briefly mentioned previously in this chapter, isotype the corresponding germline switch sequences. Indeed, many
switching involves removal of C μ from downstream of features of CSR are shared by SHM (as indicated previously
the rearranged H chain VDJ gene and its replacement by in this chapter), including the requirements for transcrip-
a new downstream CH region. This occurs by a deletional tion, AID, UNG2, and MMR components for normal CSR
recombination—CSR—in which the recombinational and SHM. However, CSR is more complex in that it involves
breakpoints generally occur within G-rich repetitive DNA simultaneous targeting to two DNA regions (i.e., switch

regions of Cμ and the target Cx), it requires dsbs, and it activation. The relationship of CSR to cell division is sup-
consequently requires mechanisms to repair the breaks. ported by evidence that switching is linked to the cell cycle 271
AID can theoretically trigger single-strand breaks on both and to the number of cell divisions after stimulation, 272 a
DNA strands by the mechanisms discussed for SHM, in- phenomenon which may reflect cell division-related regu-
cluding cleavage by APE-1. If these breaks are close enough lation of AID expression.273 However, apart from activating
together, they can effectively form a dsb with staggered proliferation, CD40 has additional effects that may facilitate
ends, and if dsbs occur in two switch regions, the DNA re- switching, including upregulation of IL-4 responsiveness
pair mechanisms common to all cells may rejoin the broken and IL-4 receptor number, 274 upregulation of sterile “switch
ends to complete a CSR event. Indeed, B cells in which Sμ transcripts” (discussed below), and upregulation of AID
and Sγ1 were deleted and replaced by recognition sites for expression. Activated B cells also express CD40L, which
the yeast homing endonuclease I-SceI were capable of medi- can not only trigger CD40 signaling but also transduce a
ating some μ → γ1 CSR in the absence of AID if I-SceI was “reverse” signal affecting B-cell function.275 Independent of
expressed, 267 highlighting the participation of ubiquitous CD40, B-cell activation can independently be stimulated by
AID-independent repair mechanisms for CSR. These mech- TLR ligands and cytokines such as APRIL, another member
anisms depend on the NHEJ components described previ- of the TNF family.276
ously in this chapter plus backup participation by alternative
end-joining mechanisms discussed below. Regulation of Class Switch Recombination to Specific
Isotypes: Promoters, Enhancers, and Chromatin
Regulation of Isotype Switching: Proliferation and AID Different isotypes are known to predominate in different im-
Expression mune responses depending on the antigen, route of antigen
Isotype switching occurs physiologically in animals about administration, and several other parameters. These differ-
1 week after immunization with T-dependent antigens, at ent parameters act in part by influencing the cytokine milieu
about the same time that somatic mutation of Ig genes be- of the B cells. IL-4, for example, promotes the expression of
gins. Both processes normally occur in GCs of lymphoid IgE and IgG1, whereas TNF-β promotes switching to IgA.
organs, a location that facilitates interactions between These lymphokines have been proposed to act by making the
B cells, T cells, and follicular dendritic cells presenting an- target isotype “accessible” to switch recombinase machinery
tigen. As demonstrated by in vitro switching experiments, that may be largely non–isotype-specific. The accessibility
T cells promote switching by secretion of cytokines (espe- is associated with expression of a “sterile” or “germline”
cially IL-4 and transforming growth factor β) as well as by RNA transcript that initiates upstream of a target S region
cell-to-cell contact. (see Fig. 6.12) and extends through the target C region. The
A major component of the cell contact signal is mediated germline transcript is spliced so that a noncoding upstream
by an interaction between the B-cell surface marker CD40 exon known as an I (“intron”) region is joined to the first
and its ligand—designated CD40L, CD154, or gp39—which coding exon of the C region. (This contrasts with the “pro-
is expressed on activated T cells (primarily CD4 +). CD40 is ductive” transcript containing VDJ spliced to the C region.)
a member of the TNF–receptor family, while CD40L belongs The same experimental conditions—particularly the same
to the TNF–ligand family. The dependence of switching on cytokines—that favor the accumulation of germline tran-
the CD40-CD40L interaction is highlighted by the genetic scripts from a particular isotype generally also stimulate
disease known as the X-linked hyper-IgM syndrome-1, switch recombination to the same isotype. In many cases, the
which was found to be caused by a defect in the human gene signals transduced by the cytokine receptor have been eluci-
encoding the CD40L/gp39.268 Like AID-deficient patients dated. For example, IL-4 stimulates germline transcription
with hyper-IgM syndrome-2 described previously, patients by activating the transcription factor STAT6, which binds to
with X-linked hyper-IgM syndrome-1 have elevated concen- one of several IL-4 response motifs in the promoter region
trations of IgM in their serum and almost no Igs of other upstream of Iε and Iγ1. CD40 engagement also acts in part
isotypes. In addition, their antibodies fail to show affinity through NFκ B-mediated binding to I region promoters.277
maturation or evidence of B-cell memory responses. Similar Gene targeting experiments have shown that mouse
defects are seen in humans with mutations in their CD40 strains lacking the I region (and its promoter) of a particular
gene, an autosomal recessive disease designated hyper-IgM isotype do not switch to that isotype, reinforcing the idea
3.269 CSR impairment may also be caused by abnormal func- that sterile transcription is necessary for CSR.278 The low
tion of CD40 signaling pathway components including extent of sequence conservation of the I exons and the lack
IKKγ (also known as NEMO), NFκ B proteins, and C-Jun of consistent open reading frames suggest that these tran-
N-terminal kinase. Mouse strains with engineered defects scripts do not encode a functional protein. Indeed, the exact
in CD40 are defective in SHM and T-dependent CSR, but sequence of the I region may be irrelevant as an I region
respond with normal isotype switching to T-independent can be replaced by an unrelated sequence and still support
antigens. The T-independent switching pathway may be CSR.279 However, the transcribed exon upstream of the S re-
especially important for gut-associated switching to IgA.270 gion apparently needs a splice donor site allowing splicing
One role of the CD40 engagement is to induce B-cell pro- to the downstream C region, as a targeted construct lack-
liferation. Indeed, other proliferative stimuli (e.g., LPS or ing such a splice donor site was reportedly unable to sup-
IgM or IgD crosslinking) can support cytokine-induced iso- port CSR even though transcription through the S region
type switching in vitro in the absence of T cells and CD40 occurred.280

Apart from I region promoters, germline transcription heterochromatin protein 1, as binding to H3K9me-modified
and isotype switching are also regulated by IgH enhanc- chromatin. By coimmunoprecipitation, KAP1 was con-
ers. A combined deletion of the murine 3′-regulatory region firmed to bind to AID in vivo, and B-cell–specific conditional
enhancers HS3b and HS4 was found to cause a significant knockout of KAP1 was found to diminish AID binding to Sμ
impairment in switching to most isotypes, although switch- and to impair CSR efficiency by about 50%.
ing to IgG1 was unaffected (and IgA only moderately de-
creased).281 The diminished switching was associated with Switch Region Targeting of AID in Class Switch
diminished germline transcription of the same isotypes, Recombination: R Loops, Paused Polymerase II, and AGCT
suggesting that an important function of the enhancers A model in which AID loads onto an RNA polymerase com-
in CSR is to increase germline transcription. The relative plex and acts on DNA as the transcription complex travels
independence of γ1 from regulation by enhancers may be downstream was discussed previously in the context of AID
related to the putative locus control region associated with function in SHM, and a similar model apparently applies
that gene.282 in the context of CSR. The domain of AID susceptibility in
Enhancers are believed to function via physical inter- switch regions can be deduced from the distribution of C:G
actions between enhancer-bound proteins and promoter- → T:A mutations in B cells that are defective for both UNG
bound proteins, creating a DNA loop that brings enhancers and MMR, because (as shown in Fig 6.13) in these cells the
into close proximity to promoters. Such looping may have only AID-dependent mutations would be U:G mismatches
special significance in promoting DNA recombination be- resolved by replication to T:A. A study of the distribution
tween segments of DNA lying great linear distances apart of such mutations in clones from Ung−/−, Msh2−/− B cells
in the chromosome, as has been discussed in the context of found that a domain of mutations began about 150 bp 3′ of
V(D)J recombination. Chromosome conformation capture the Iμ transcription start site and extended 4 to 5 kb down-
experiments have shown that Eμ and 3′-regulatory region stream, with diminishing mutation frequency near Cμ.293 The
enhancers are in close proximity in mature resting B cells, location of AID binding in B-cell DNA can also be directly
and that when B cells are stimulated to switch to γ1 by LPS + determined by ChIP-seq analysis, as discussed previously.
IL4 (or to γ 3 by LPS alone), the corresponding I region pro- AID was found to bind to many loci that are transcribed in
moter moves close to the two enhancers.283,284 This looping B cells; indeed, the patterns of AID and RNA polymerase II
would bring Sμ, which lies just downstream of Eμ, close to binding detected by ChIP are very similar.160 RPA, however,
the Iγ1 (or Iγ 3) promoter, presumably facilitating recombi- was efficiently bound only at IgH switch regions, and this
nation between Sμ and the Sγ region. binding was inhibited by the Ser38Ala mutation that blocks
Another parameter correlated with regional transcrip- the critical Ser38 phosphorylation discussed previously.
tional regulation is the chromatin context of the genes, One likely consequence of germline transcription in fa-
including specific modifications of histone proteins, as cilitating CSR involves the formation of a stable RNA:DNA
discussed previously in this chapter. For example, acetyla- complex known as an R-loop. In this structure, RNA tran-
tion of histone H3 (H3Ac) is associated with DNA regions scribed from the template strand of a DNA molecule binds
of increased “accessibility” to transcription (as well as to tightly to that strand with Watson-Crick basepairing, dis-
experimental digestion by restriction enzymes), while tran- placing the other DNA strand, which forms a single-stranded
scriptional promoters tend to be marked with trimethyl- loop. In support of the R-loop model, cell-free transcription
ation of H3 at lysine4 (H3K4me3). In B cells stimulated to across G-rich switch regions was found to generate a stable
switch to γ1 by LPS + IL4 (or to γ 3 by LPS alone), the corre- association of the transcript RNA with the template DNA 294 ;
sponding I region and switch regions show increased H3Ac significantly, no substantial association occurred when the
and H3K4me3 marks.285,286 The importance of the H3K4me3 switch region was transcribed in reverse orientation, leading
mark for transcription and CSR is highlighted by evidence to a C-rich transcript, or when the transcribed template was
that preventing this mark—by B cell–specific knockout of a DNA fragment other than a switch region. The displaced
PTIP, a component of the machinery that catalyzes this DNA strand in S region R loops was susceptible to deamina-
modification—leads to impaired germline γ transcription tion by AID.155 Evidence that such R-loops form in vivo over
and defective CSR to γ isotypes.287 PTIP knockout B cells S regions has been obtained from bisulfite analysis of single-
show decreased DNA looping between the 3′ enhancer re- strand DNA regions in B cells.295,296
gion and Iγ region promoters, suggesting that PTIP contrib- The tightly bound RNA-DNA complex on the template
utes to this looping.288 Knockdown of other factors that are strand of an R-loop may explain both the high level of poly-
necessary to maintain H3K4me3 also reduce CSR efficiency, merase II accumulation and the AID-induced mutations
including the components of the complex known as FACT in the vicinity of switch region DNA.297 Progression of the
( facilitates chromatin t ranscription).289 polymerase II complex transcribing through the switch re-
Surprisingly, a histone modification generally associ- gion might be impeded by the RNA bound to the template
ated with gene silencing—trimethylation of H3 lysine 9 strand, leading to an accumulation of “stalled” polymerase
(H3K9me3)—has also been reported to mark switch regions II molecules; the “stalling” of AID molecules associated
targeted for CSR in both mouse and human B cells.290,291 with the transcription complex might prolong exposure
Recently, a screen for nuclear proteins associating with of the DNA to deamination by AID. Some support for this
AID in vitro identified KAP1 (KRAB domain–associated scheme was reported from an in vitro model in which AID-
protein 1),292 a transcriptional repressor that associates with triggered mutations in a transcribed DNA substrate were

increased when transcription was slowed by reducing the role. Replacement of a natural murine Sμ sequence with the
nucleotide triphosphate concentration.298 The importance AT-rich frog Sμ sequence, which cannot form an R-loop, sup-
of stalled polymerase II for CSR was reinforced by the results ported somewhat reduced but still substantial frequencies of
of a screen for factors whose knockdown by specific shRNAs CSR, and it functioned equally well in either orientation.
would inhibit CSR 299 : one protein recovered in this screen Significantly, in either orientation, the recombination junc-
was the murine homolog of suppressor of Ty 5 (Spt5), a tions were clustered in a portion of the S region that is rich
transcription elongation factor known to be associated with in repeats of the sequence AGCT, which is a special case of
stalled polymerase II. Spt5 knockdown decreased both CSR the WRC consensus sequence for AID targeting, being pres-
and switch region hypermutation, as well as AID occupancy ent on both strands as a self-complementary palindrome.
of the Sμ region, without affecting cellular levels of AID pro- Indeed, the AGCT-rich region was a good substrate for in
tein or germline transcripts. By coimmunoprecipitation ex- vitro deamination by AID when transcribed in association
periments, AID and Spt5 were found to associate. And genes with RPA. The AGCT motif is enriched in all mammalian S
with high Spt5 occupancy by ChIP analysis were found to be regions and may be particularly effective as a target for CSR
most susceptible to mutations induced by AID overexpres- because its presence in clusters on both strands promotes
sion. Polymerase II stalling has been extensively studied in closely spaced nicks on opposite strands, or even a dsb with
yeast and drosophila, and is found in many genes that re- a single base overhang if the cytosines in both strands of the
quire more rapid changes in expression than can be achieved same AGCT motif are targeted.303 The density of AGCT in
by modulating transcriptional initiation.300 Apparently, the S regions correlates with the location of switch junctions bet-
complex of proteins mediating polymerase II stalling has ter than the density of WRC or the boundaries of G-richness
been adapted in B cells for the special function of regulating or R-loops.296 These results all suggest that clusters of AGCT
CSR of IgH genes. may represent a target for CSR that evolved in amphibians,
The R-loop model explains how the “upper” nontran- with R-loop formation evolving later in mammals to further
scribed DNA strand would become single-stranded and ac- enhance AID accessibility to S region DNA.
cessible to AID, but it raises the question of how AID might
gain access to the “lower” template DNA strand, held in a DNA Breaks as Intermediates in Class Switch
tight RNA-DNA hybrid. This strand is known to be acces- Recombination
sible to AID because it undergoes C → T mutations in Ung − The recombination event that underlies isotype switch-
/− Msh2− /− double knockout B cells.293 A likely answer to ing includes DNA breaks and rejoining events that must
this question has come from an analysis of proteins bound involve both strands of DNA. Although the RAG-induced
to AID when mixed with a B cell extract plus in vitro–tran- DNA breaks that initiate V(D)J recombination occur at the
scribed switch region DNA.301 The complex of AID and corresponding position on the two strands (yielding a blunt
transcribed DNA was found to bind to components of the end and a hairpin), the nature of the ends in the initial CSR
multisubunit RNA exosome. The exosome is an evolution- cleavage is not so clear. An early compilation of switch junc-
arily conserved structure containing nine core proteins that tions304 found only infrequent instances of microhomology
can associate with RNA nucleases, leading to degradation at the junction (i.e., short sequence segments that are iden-
of RNA from template DNA; the exosome could thus po- tical in the unrearranged S sequences near the recombina-
tentially expose the template DNA strand to AID. Indeed, tion breakpoint). As these microhomology examples would
shRNA knockdown of one exosome component in the mu- be consistent with invasion of one DNA strand from Sμ
rine B cell line CH12F3 inhibited CSR. Moreover, exosome targeting a short homologous region in a downstream S re-
components were found to associate with AID in vivo by gion (or vice versa), the rarity of such junctions has been
immunoprecipitation experiments and were detected (by interpreted as an indication that CSR only rarely occurs by
ChIP) bound to switch region DNA in cells activated for strand invasion and instead usually proceeds by ligation of
CSR. Finally, in a deamination assay of a model switch re- blunt DNA ends. However, dsbs with staggered ends would
gion DNA transcribed in vitro by T7 polymerase, the addi- generally be the result of the widely accepted mechanism
tion of AID + RPA + PKA led to deamination of only the shown in Figure 6.14 (bottom left): a DNA break on one
nontemplate strand, but the further addition of exosome strand might result from AID-catalyzed cytosine deamina-
components led to deamination on the template strand as tion, removal of the resulting uracil by UNG, and single-
well. These experiments highlight the exosome as a likely strand cleavage 5′ to the abasic site by an endonuclease,
candidate for explaining AID action on the template strand probably apurinic-apyridinic endonuclease 1 (APE1). The
of R-loops, though an unhybridized “lower strand” may initial staggered ends could be converted to blunt ends
also be produced as a result of transcription-dependent su- through exonuclease trimming of a 5′ or 3′ single strand
percoiling or antisense transcription. overhang, through fi lling in of a shorter 3′ end by a DNA
However, the propensity for R-loop formation is not the polymerase, or through a combination of both processes.
only property of switch regions that facilitates CSR. When the Filling in by error-prone polymerases could explain the
Sγ1 region was inverted, it retained about 25% of the wild- mutations commonly observed around the switch junction,
type CSR activity.302 As this inverted, and now C-rich, DNA as mentioned previously.
segment could not form an R-loop, this result suggests that Evidence supporting initially staggered DNA breaks in
while the R-loop contributes to CSR, other features of the S CSR was reported from occasional switch junctions observed
region that are preserved in the inverted sequence also play a in a model CSR substrate designed with two oppositely

AID RNA pol

AID, bound to RNA polymerase,
deaminates Cs in DNA
When the CSR-associated dsbs in Sμ were examined by LM-
PCR in B cells from normal and AID − /− individuals, the dsbs
S region repeats
were significantly fewer in AID − /− B cells, though not com-
pletely absent. Furthermore, microscopic foci of the modified
Transcription through S region leads
histone γH2AX (which rapidly accumulates at dsbs) were found
to R loop formation and extensive at IgH genes (localized by FISH) in B cells undergoing CSR;
deamination of upper DNA strand
U U these foci were strikingly diminished in AID − /− B cells, consis-
U
tent with AID-dependence of dsbs in CSR.309 These foci, as well
as switch region dsbs, occur predominantly in the G1 phase of
RN
A the cell cycle.310 Like the off-target deamination by AID dis-
RNA is removed from “lower” template strand cussed previously in the context of SHM, off-target DNA breaks
by exosome, allowing AID to target this strand
U
are another potential consequence of AID. B cells from mice
U
U U overexpressing AID showed a high incidence of chromosomal
translocations and DNA breaks compared with Aicda− /− mice;
e
in the context of homozygous p53 deficiency (which allows cell
U
growth despite DNA damage), many of the mice developed
U U U U B-cell lymphomas.311 In a study designed to systematically
examine the locations of dsbs in mouse B cells stimulated to
U U
undergo CSR, ChIP was used to identify DNA associated with
UNG removes uracils;
U:G mismatch recognized
by MMR; Exo I clears DNA abasic sites cleaved by Nbs1, a protein that rapidly binds to dsbs in vivo, as discussed
patch on strand with U; APE1; closely spaced
nicks lead to dsb
below.312 This analysis detected hundreds of reproducible AID-
apposed gaps form dsb
dsb dsb dependent DNA break locations, some of which were syntenic
with DNA rearrangements found in human B-cell lymphomas.
Many of the AID-dependent dsbs occurred in nontranscribed
regions, unlike the CSR-related dsbs in IgH switch regions.
FIG 6.14. Mechanistic Model of Class Switch Recombination. This
figure incorporates features of the most widely accepted models,
Evidence indicates that AID-dependent off-target dsbs are
but some aspects are uncertain at present. Two strands of deoxy- normally repaired largely by homologous recombination: B
ribonucleic acid (DNA) near the 5′ boundary of a switch region are cells defective for the homologous recombination component
shown in black, with the S region repeats indicated by dots. DNA XRCC2 were found to harbor many more γ-H2AX foci than
with normal Watson-Crick basepairing is indicated by the gray shad- were found in Xrcc2+/+ cells or in Aicda− /− Xrcc2− /− cells.313
ing between the DNA strands. Ribonucleic acid (RNA) polymerase B cells apparently protect against these breaks by upregulating
moves to the right, transcribing the sequence into an RNA strand; Xrcc2 transcription when activated for CSR.
multiple other proteins accompany the RNA polymerase, but only
activation-induced deaminase (AID) (black circle) is shown. AID can Ubiquitously Expressed Components of Class Switch
deaminate C residues to U in single-stranded R-looped DNA dis-
placed by the RNA-DNA complex. The exosome complex (e in the Recombination Machinery
drawing) degrades RNA in the RNA-DNA hybrid, leaving an unpaired UNG and Mismatch Repair. The model of Figure 6.13 sug-
lower strand where AID can deaminate C residues. DNA cleavage gests that after AID-catalyzed deamination of a cytidine
triggered by processing of uracils can lead to double-strand breaks residue, the resulting uracil is removed by UNG, creating an
in pathways dependent or independent of mismatch repair compo- abasic site that is converted by APE1 to a single-strand DNA
nents as shown. break (nick), which can become double-stranded if there is
a nearby nick on the opposite strand. This model predicts
that both CSR and the creation of dsbs would be severely im-
oriented S regions such that CSR would occur by inversion, paired in the absence of UNG. Indeed, B cells from Ung− / −
preserving both recombination junctions on the same chro- mice showed almost complete inability to switch in vitro to
mosome. Several duplications at the ends of inverted DNA IgG1 or IgG3 secretion, and significant impairment in IgA
after CSR suggested that complementary overhangs at dsbs secretion.151 dsbs in Ung−/ − B cells, as detected by LM-PCR,
had been fi lled in before joining.305 Additional evidence for were also significantly reduced, but not abolished.308 Human
staggered breaks has come from several investigators using patients with homozygous UNG deficiency due to mutations
ligation-mediated–polymerase chain reaction (LM-PCR) to in both UNG alleles showed a hyper-IgM phenotype, with
detect double-strand DNA ends at switch regions in cells a more profound defect than in Ung−/ − mice: the patients
undergoing CSR.306–308 LM-PCR protocols involving liga- showed essentially no IgG, IgE, or IgA secretion by stimu-
tion of a blunt linker directly to blunt ends from genomic lated B cells, and no dsbs (by LM-PCR) in Sμ.314
DNA were successful in amplifying blunt ends from DNA of A role for APE1 or APE2 in CSR is supported by the ob-
B cells activated for CSR, but when the DNA was pretreated servation that in mice with engineered deficiencies these
with T4 polymerase, which would convert staggered end genes, B cells induced for CSR show decreased switching and
cuts to blunt, the yield of amplified LM-PCR products was decreased induction of dsbs.315
significantly increased, suggesting that most of the ends in The complementary pathways of UNG and MMR action in
the isolated genomic DNA (before T4 polymerase treatment) SHM were discussed previously (see Fig. 6.13); current evidence
were staggered. suggests a similar participation of MMR in CSR. Although

CSR was dramatically impaired in Ung− / − B cells (e.g., in vitro wild-type) in the number of chromosome aberrations in the
switching to IgG1 was reduced to about 6% of wild-type), the IgH locus detected by FISH assays330 ; these aberrations are
double knockout Ung− / − Msh2− / − caused significant further AID-dependent and only observed after activation of B cells
impairment (to 1.5% of wild-type IgG1 switching).316 Evidence for CSR. In humans, Artemis deficiency results in a SCID syn-
suggests that the C-terminal 10 amino acids of AID that are drome with severe defects in B- and T-cell development, but
required for CSR but not SHM may function by stabilizing the rare Sμ-Sα recombination junctions amplified from patient B
interaction of UNG2 and the MSH2-MSH6 dimer to DNA.317 cells were found to show a high index of microhomologies,331
If a single-strand break created on one strand by UNG suggesting that Artemis is required for normal classic-NHEJ
and APE1 is too far away from the the closest single-strand resolution of dsbs in human CSR.
break on the opposite strand to create a dsb, then Exo1 en- Studies on the role of DNA-PKcs in CSR have yielded
gaged by MMR can chew from a nick on one strand toward somewhat conflicting conclusions on whether this protein
a nick on the opposite strand, creating a dsb.310,318 Consistent is required for maximally efficient CSR, with the results of
with this idea, mice engineered with a knockout of the MMR individual studies perhaps depending on the knocked-in
component Exo1 show a significantly decreased efficiency of Ig genes used, the specific mutations of DNA-PKcs, or the
CSR, to roughly 15% to 30% of normal.319 genetic background of mice studied.332,333 However, DNA-
Thus, many of the same AID-triggered mechanisms that PKcs − / − B cells showed evidence of AID-dependent chro-
induce mutations in V H regions operate in switch regions to mosome aberrations in the IgH locus, similar to those seen
induce dsbs. Once created, these dsbs are resolved by mech- in Artemis−/− cells but more numerous.330 Thus it appears
anisms distinct from the mechanisms resolving mismatches that both Artemis and DNA-PKcs are required for efficient
in SHM; however, as discussed in the following, some of the repair of at least a subset of the dsbs associated with CSR.
components that resolve dsbs generated in V(D)J recombi-
nation play a similar role in CSR. DNA Damage Response Proteins in Class Switch
Recombination. NHEJ is one component of a larger mecha-
End Joining Proteins in Class Switch Recombination. If AID nism for detecting and repairing dsbs, collectively known
triggers DNA cleavage at switch regions, ubiquitous DNA as the DNA damage response (DDR). Because DNA dsbs
repair and ligation enzymes could participate in the subse- are potentially damaging to the cell, and occur in all cells
quent DNA repair steps of CSR, as in V(D)J recombination. through accidents of DNA metabolism, toxic chemicals,
This possibility has been tested by engineered gene knock- and radiation, DDR appeared early in evolution, and many
outs. However, as NHEJ knockouts would impair V(D)J re- components are conserved from yeast to mammals. Defects
combination, thus blocking B-cell development prior to the in DDR can cause developmental abnormalities, cancer pre-
stage of CSR, investigators have studied NHEJ in CSR using disposition, and sensitivity to radiation, as well as immuno-
mouse strains with knockins of productive recombined H deficiency resulting from impaired V(D)J recombination or
and L chain genes, which can undergo CSR. IgH/IgL knockin CSR. DDR factors participate in a baroque cascade of inter-
mice with intact Ku genes were able to switch to downstream actions to cluster at dsbs and initiate repair. The complexi-
isotypes, but the corresponding Ku-70– or Ku-80–deficient ties of the DDR are beyond the scope of this chapter, but are
mice were reported to be dramatically impaired in CSR, al- described in several recent reviews.78,334
though decreased cell proliferation could have contributed to Several DDR components have been documented to par-
this effect as more recent studies using different conditions ticipate in CSR and are listed below; mutations in some of
report 30% to 50% residual CSR activity.320 these cause specific genetic syndromes in humans, often as-
The other “core” factors of NHEJ—XLF/Cernunnos, sociated with immunodeficiency.
Ligase4, and XRCC4—catalyze the DNA ligation in NHEJ 1. Nbs1, product of the gene that is defective in the
and have also been investigated for participation in CSR.320–324 human disease Nijmegen breakage syndrome.309,335,336
B cells deficient in any one of these genes show variably de- Nbs1 functions as part of a complex that also includes
creased CSR efficiency, with increased junctional micro- Mre11 and Rad50 (MRN complex). Mre11 has a nu-
homologies compared with normal B cells. This observation clease activity, and may contribute to DNA cleavage in
suggests that one or more fairly robust backup pathways— CSR; mice and humans with defective Mre11 also show
commonly called alternative NHEJ—can repair dsbs using impaired CSR.337,338
microhomology-based ligation when “classic” NHEJ is inop- 2. ATM, product of the gene mutated in the human
erative. This conclusion is consistent with studies of alternative disease ataxia telengectasia and a member of the
NHEJ repair of dsbs unrelated to CSR.78 The proteins CtIP,325 phosphatidylinositol-3′-kinase family.339,340
PARP1,326,327 and XRCC1328 apparently contribute to alterna- 3. γ-H2AX—the phosphorylated form of histone H2AX—
tive NHEJ during CSR, as experimentally reduced expression which rapidly accumulates in foci at dsbs, and helps to
of each protein decreases microhomology at CSR junctions. assemble other proteins at dsbs to prevent the breaks
Apart from these core NHEJ components, Artemis from progressing to chromosome translocations.341
and DNA-PKcs are required for joining ends that require 4. The E3 ubiquitin ligase RNF8 (Ring Finger 8), which is
processing, like the hairpin coding ends produced in V(D)J known to monoubiquitinate histones at dsbs.342
recombination discussed previously in this chapter. Artemis- 5. The E3 ubiquitin ligase RNF168, found to be mutated in
deficient murine B cells have no obvious impairment in CSR the RIDDLE (radiosensitivity, immunodeficiency, dys-
efficiency329 but show several-fold increases (compared to morphic features, learning difficulties) syndrome.343

6. 53BP1 (p53 binding protein 1) originally discovered on Ser38-phosphorylated AID. Cellular concentrations of
as a protein binding to the tumor suppressor p53, but cAMP were found to rise rapidly after stimulation to CSR,
subsequently found to function in checkpoint control and a genetic inactivation of PKA activity that prevented its
and to localize rapidly to DNA breaks in vivo. 53BP1−/− activation by cAMP reduced both RPA binding and CSR.
murine B cells show dramatic impairment in CSR (to These data suggest a model in which CSR stimuli recruit
55 to 25% of wild-type), but increases in the frequency both PKA and nonphosphorylated AID to switch regions,
of AID-triggered deletions within Sμ,344 suggesting that and AID gains the ability to recruit RPA and trigger dsbs
formation of dsbs and repair of closely-spaced breaks efficiently only when cAMP-activated PKA phosphorylates
does not require 53BP1, but joining of more distant Ser38.
dsbs depends on this protein. Additional components of CSR may be discovered by
Although the we currently lack a detailed model explain- analysis of patients with a hyper-IgM phenotype unex-
ing the functions of all the DDR factors in dsb repair and plained by defects in known components.314
specifically in CSR, the emerging evidence suggests that
these proteins have mutual interactions and distinct but CONCLUSION
related roles, so that the elimination of any one protein re-
Recombinant DNA technology has revolutionized the study
duces CSR efficiency but permits residual CSR to occur by
of the antibody response. Initial investigations used power-
pathways that remain intact.
ful cloning and sequencing methods to define the structure
Other Proteins that may Target AID for Class Switch of the Ig genes as they exist in the germline and in actively se-
Recombination. As discussed in the context of SHM, ac- creting B-lymphocytes. Subsequent experiments have begun
curate targeting of AID is important since off-target activi- to shed light on the mechanisms of the processes unique to
ties of this protein can be deleterious, and off-target dsbs these genes: V(D)J recombination, CSR, and SHM.
triggered by AID could be particularly dangerous. Several The knowledge of Ig genes gained so far has answered
searches for proteins that might contribute to targeting of some of the most puzzling mysteries about antibody di-
AID to switch regions have uncovered candidates for this versity, as discussed previously, and has also led to many
function. In one study,345 biotinylated AID was used as bait practical ramifications involving these genes that are be-
to fish for in vitro binding proteins, which were identified yond the scope of this chapter. As one example, cloned Ig
by mass spectrometry. One protein identified by this screen genes have allowed the production of recombinant mono-
was PTBP2 (polypyrimidine tract-binding protein 2) which clonal antibodies and the bioengineering of Ig-fusion pro-
is considered to be a regulator of RNA splicing. Knockdown teins that exploit the exquisite specificity of antibody V
of PTBP2 by shRNA in CH12 B cells caused a substantial region binding (e.g., antibody-toxin fusions) or the abil-
decrease in binding of AID to Sμ (determined by ChIP) and ity of Ig C region domains to extend serum half-life or en-
significant impairment of CSR. The possible role of a splic- gage Fc receptors. Other engineered derivatives utilizing Ig
ing regulator in CSR could be related to the puzzling obser- genes include single-chain antibodies, bispecific antibod-
vation that germline transcripts must be spliced in order for ies, and “intrabodies” designed not to be secreted from a
efficient CSR, 280 as mentioned previously. cell but rather to bind to intracellular targets.348–350 Ig V
A screen for proteins that could bind selectively in vitro gene fragments cloned into bacteriophage so as to express
to AGCT sequences—which are found clustered in many single-chain V regions on the phage surface (phage dis-
switch regions—identified another protein that might target play libraries) can be used to obtain specific monoclonal
AID in CSR: the 14-3-3 proteins, a family of seven members antibodies without immunization or use of mammalian
widely expressed and known to bind to many signaling pro- cells, and in vitro mutation and selection protocols can
teins.346 ChIP assays on B cells stimulated for CSR showed mimic affi nity maturation to yield high-affi nity antibod-
that 14-3-3 proteins bind to switch regions in an isotype- ies.351 Even Ig gene regulatory regions have been exploited
specific way (depending on stimulus conditions) in both to achieve B-cell–specific expression of oncogenes352 and
normal and Aicda−/− cells, and reduction of 14-3-3 activ- of intracellular toxins that could be used to target B lym-
ity by either a peptide inhibitor or by genetic engineering phomas.353 Apart from these biotechnology applications,
decreased both CSR efficiency and AID binding to switch Ig gene probes have led to the identification of numerous
region DNA. Finally, a bimolecular fluorescence comple- proto-oncogenes that become activated by translocation
mentation assay revealed that in B-cell nuclei AID and 14- into Ig gene loci.354 For instance, Bcl2 was initially discov-
3-3 molecules form a complex that is dependent on the AID ered as the target of Ig H chain translocation in follicular
C-terminal amino acids required for CSR. These experi- lymphoma, and provided an entry into an entire family of
ments suggest that 14-3-3 proteins are additional candidates apoptosis-related genes. A fi nal example of medical benefit
for AID targeting molecules. from Ig gene technology has been the use of patient-spe-
PKA, the cAMP-regulated protein kinase responsible for cific Ig gene rearrangements of leukemias or lymphomas to
phosphorylation of AID Ser38 discussed previously, has also monitor disease status by PCR.355,356
been suggested as a candidate that targets AID-dependent Further practical applications of Ig genes can be antici-
DNA cleavage to switch regions.347 PKA was found by ChIP pated in the future, as well as a deeper scientific understand-
to bind to switch regions of stimulated B cells from normal ing of their molecular biology and their contribution to the
and Aicda−/−mice, whereas the binding of RPA depended immune system.

Antigen–Antibody Interactions and

CHAPTER
7 Monoclonal Antibodies
Jay A. Berzofsky • Ira J. Berkower
INTRODUCTION as well as to make use of antibodies as quantitative reagents.

The basic principles of antigen–antibody interaction are Furthermore, we discuss the derivation, use, and properties
those of any bimolecular reaction. Moreover, the binding of monoclonal antibodies.
of antigen by antibody can, in general, be described by the
same theories and studied by the same experimental ap- THERMODYNAMICS AND KINETICS
proaches as the binding of a hormone by its receptor, of a
substrate by enzyme, or of oxygen by hemoglobin. There The Thermodynamics of Affinity
are several major differences, however, between antigen– The basic thermodynamic principles of antigen–antibody
antibody interactions and these other situations. First, interactions, as we indicated previously, are the same as
unlike most enzymes and many hormone-binding sys- those for any reversible bimolecular binding reaction. We
tems, antibodies do not irreversibly alter the antigen they review these as they apply to this particular immunologic
bind. Thus, the reactions are, at least in principle, always reaction.
reversible. Second, antibodies can be raised, by design of
the investigator, with specificity for almost any substance Chemical Equilibrium in Solution
known. In each case, one can fi nd antibodies with affi ni- For this purpose, let S = antibody binding sites, L = ligand
ties as high as and specificities as great as those of enzymes (antigen) sites, and SL = the complex of the two. Then for
for their substrates and receptors for their hormones. The the reaction
interaction of antibody with antigen can thus be taken as a S + L D SL (1)
prototype for interactions of macromolecules with ligands
in general. In addition, these same features of reversibility the mass action law states
and availability of a wide variety of specificities have made [SL]
antibodies invaluable reagents for identifying, quantitating, K A = ______ (2)
[S][L]
and purifying a growing number of substances of biologic
and medical importance. where K A = association constant (or affinity) and square
One other feature of antibodies that in the past proved to brackets = molar concentration of the reactants enclosed. The
be a difficulty in studying and using them, compared to, say, import of this equation is that for any given set of conditions
enzymes, is their enormous heterogeneity. Even “purified” such as temperature, pH, and salt concentration, the ratio of
antibodies from an immune antiserum, all specific for the the concentration of the complex to the product of the con-
same substance and sharing the same overall immuno- centrations of the reactants at equilibrium is always constant.
globulin (Ig) structure, will be a heterogeneous mixture of Thus, changing the concentration of either antibody or li-
molecules of different subclass, different affinity, and differ- gand will invariably change the concentration of the complex,
ent fine specificity and ability to discriminate among cross- provided neither reactant is limiting, that is, neither has al-
reacting antigens. The advent of hybridoma monoclonal ready been saturated, and provided sufficient time is allowed
antibodies1–3 has made available a source of homogeneous to reach a new state of equilibrium. Moreover, because the
antibodies to almost anything to which antisera can be concentrations of antibody and ligand appear in this equa-
raised. Nevertheless, heterogeneous antisera are still in tion in a completely symmetrical fashion, doubling either the
widespread use and even have advantages for certain pur- antibody concentration or the antigen concentration results
poses, such as precipitation reactions. Therefore, it is critical in a doubling of the concentration of the antigen–antibody
to keep in mind throughout this chapter, and indeed much complex, provided the other reactant is in sufficient excess.
of the volume, that the principles derived for the interac- This proviso, an echo of the one just discussed, is inherent in
tion of one antibody with one antigen must be modified and the fact that [S] and [L] refer to the concentrations of free S
extended to cover the case of heterogeneous components in and free L, respectively, in solution, not the total concentra-
the reaction. tion, which would include that of the complex. Thus, if L is
In this chapter, we examine the theoretical principles not in great excess, doubling [S] results in a decrease in [L] as
necessary for analyzing, in a quantitative manner, the inter- some of it is consumed in the complex, so the net result is less
action of antibody with antigen and the experimental tech- than a doubling of [SL]. Similarly, halving the volume results
niques that have been developed to study these interactions in a doubling of the total concentration of both antibody and
183

The total concentration of ligand at which the antibody

begins to saturate is a function not only of the antibody
concentration but also of the association constant, K A, also
called the affinity. This constant has units of molar (M−1) or
L/mol, if all the concentrations in Equation 2 are molar. Thus
the product K A[L] is unitless. It is the value of this product
relative to 1 that determines how saturated the antibody is,
as can be seen from Equation 3′. For example, an antibody
with an affinity of 107 M−1 will not be saturated if the ligand
concentration is 10−8 M (product K A[L] = 0.1) even if the total
amount of ligand is in great excess over the total amount of
FIG. 7.1. Schematic Plot of Bound Ligand Concentration as a Function antibody. From Equation 3′ the fraction of antibody occupied
of Free Ligand Concentration at a Constant Total Concentration would be only 0.1/1.1, or about 9%, in this example. These
of Antibody-Combining Sites, [S]t. The curve asymptotically aspects of affinity and the methods for measuring affinity are
approaches a plateau at which [bound ligand] = [S]t. analyzed in greater detail in the next section.
ligand. If the fraction of both reactants tied up in the complex Free Energy
is negligibly small (as might be the case for low-affinity bind- Regarding thermodynamics, the affinity, K A, is also the
ing), the concentration of the complex quadruples. However, central quantity because it is directly related to the free
in most practical cases, the concentration of complex is a energy, ΔF, of the reaction by the equations
significant fraction of the total concentration of antigen or ΔF° = −RT ln K A (4)
antibody or both, so the net result is an increase in the con-
⫺ΔF°/RT
centration of complex but by a factor of <4. The other im- KA = e (4′)
portant, perhaps obvious, but often forgotten, principle to be where R = so-called gas constant (1.98717 cal/°K·mol),
gleaned from this example is that because it is concentration, T = absolute temperature (in degrees Kelvin), ln = natural
not amount, of each reactant that enters into the mass ac- logarithm, and e = base of the natural logarithms. The
tion law (Equation 2), putting the same amount of antigen minus sign is introduced because of the convention that a
and antibody in a smaller volume will increase the amount negative change in free energy corresponds to positive bind-
of complex formed, and diluting them in a larger volume will ing. The ΔF° is the standard free-energy change defined as
greatly decrease the amount of complex formed. Moreover, the ΔF for 1 mol antigen + 1 mol antibody sites combining
these changes go approximately as the square of the volume, to form 1 mol of complex at unit concentration.
so volumes are critical in the design of an experiment. It is also instructive to note an apparent discrepancy in
The effect of increasing free ligand concentration [L], at Equations 4 and 4′. As defined in Equation 2, K A has dimen-
constant total antibody concentration, on the concentration sions of M−1 (ie, L/mol), whereas in Equation 4′, it is dimen-
of complex, [SL], is illustrated in Figure 7.1. The mass action sionless. The reason is that for Equation 4′ to hold strictly, K A
law (Equation 2) can be rewritten as must be expressed in terms of mole fractions rather than con-
[SL] = K A[S][L] = K A([S]t − [SL])[L] (3) centrations. The mole fraction of a solute is the ratio of moles
of that solute to the total number of moles of all components
or in the solution. Because water (55 M) is by far the predomi-
K A[S]t[L] nant component of most aqueous solutions, for practical
[SL] = __________ (3′) purposes, one can convert K A into a unitless ratio of mole
(1 + K A[L])
fractions by dividing all concentrations in Equation 2 by 55 M.
where [S]t = total antibody site concentration; that is, This transformation makes Equation 4′ strictly correct, but it
[S] + [SL]. Initially, when the complex [SL] is a negligible frac- introduces an additional term, −RT ln 55 (corresponding to
tion of the total antibody [S]t, the concentration of complex the entropy of dilution), into Equation 4. This constant term
increases nearly linearly with increasing ligand. However, as cancels out when one is subtracting ΔF values but not when
a larger fraction of antibody is consumed, the slope tapers one discusses ratios of ΔF values.
off and the concentration of complex, [SL], asymptotically An important rule of thumb can be extracted from these
approaches a plateau value of [S]t as all the antibody becomes equations. Because ln 10 = 2.303, a 10-fold increase in affinity
saturated. Thus, the concentration of antibody-binding sites of binding corresponds to a free-energy change ΔF of only
can be determined from such a saturation binding curve (see 1.42 kcal/mol at 37°C (310.15°K). (The corresponding val-
Fig. 7.1), taking the concentration of (radioactively or other- ues for 25 and 4°C are 1.36 and 1.27 kcal/mol, respectively.)
wise labeled) ligand bound at saturation as a measure of the This is less than one-third the energy of a single hydrogen
concentration of antibody sites.* This measurement is some- bond (about 4.5 kcal/mol). Looked at another way, a very
times referred to as antigen-binding capacity. high affinity of 1010 M−1 corresponds to a ΔF of only 14.2 kcal/
mol, approximately the bonding energy of three hydrogen
*
This point is strictly true only for univalent ligands, but most multivalent ligands bonds. (Of course, because hydrogen bonds with water are
behave as effectively univalent at large antigen excess, where this plateau is measured. broken during the formation of hydrogen bonds between

CHAPTER 7 ANTIGEN–ANTIBODY INTERACTIONS AND MONOCLONAL ANTIBODIES | 185
antigen and antibody, the net energy per hydrogen bond is of the interacting groups. Most antigen–antibody reactions
closer to 1 kcal/mol.) It is apparent from this example that are studied near neutral pH and at physiologic salt concen-
of the many interactions (hydrophobic and ionic as well as trations (0.15 M NaCl). If the interaction is dominated by
hydrogen bonding) that occur between the contact residues ionic interactions, high salt concentration lowers the affinity.
in an antibody-combining site and the contacting residues of Conformational flexibility of an antigen can also affect the
an antigen (such as a protein), almost as many are repulsive affinity by affecting the entropy term in Equation 5. An out-
as attractive. It is this small difference of a few kilocalories standing example comes from the thermodynamics of bind-
between much larger numbers corresponding to the total of ing of a series of broadly neutralizing monoclonal antibodies
attractive and the total of repulsive interactions that leads to to the human immunodeficiency virus (HIV) envelope pro-
net “high-affinity” binding. If ΔF were any larger, binding re- tein gp120. The energetics of binding between HIV gp120
actions would be of such high affinity as to be essentially ir- and its cellular receptor cluster of differentiation (CD)45 or
reversible. Viewed in this way, it is not surprising that a small to a panel of monoclonal antibodies6 has been studied in de-
modification of the antigen can result in an enormous change tail. The results provide insight into the ways protein flex-
in affinity. A single hydrogen bond can change the affinity ibility can help the virus to evade antibody immunity. They
manyfold, and similar arguments apply to hydrophobic inter- may also explain part of the difficulty in eliciting antibodies
actions and other forms of bonding. This concept is impor- of this type when immunizing with the native protein.
tant when we discuss specificity and antigen structure later. GP120 is a glycoprotein that exhibits at least two confor-
mational states: one exists on the virus, and it changes to
Effects of Temperature, pH, Salt Concentration, and the other when it binds the CD4 receptor on the cell surface.
Conformational Flexibility Monoclonal antibodies F105 and b12 bind residues within
It was mentioned previously that K A is constant for any given the CD4 binding site on gp120 while monoclonal 2G12 binds
set of conditions such as temperature, pH, and salt concen- the opposite surface. The CD4 binding site defines a neutral-
tration. However, it varies with each of these conditions. izing surface that is conserved among a broad range of HIV
We have already seen that the conversion of free energy to isolates.
affinity depends on temperature. However, the free energy As shown in Table 7.1, CD4 and each antibody demon-
itself is also a function of temperature strated a strongly negative ΔF, corresponding to high affinity
ΔF° = ΔH° − T ΔS° (5) binding. However, as shown by microcalorimetry, they ar-
rived at the affinity in different ways. The CD4 and F105
where ΔH = change in enthalpy (the heat of the reaction),† have a strong negative ΔH of binding, but they also have a
ΔS = entropy (a term related to the change in disorder large negative entropy due to the conformational change
produced by the reaction),2 and T = absolute temperature required, as shown by strongly positive values for −TΔS. This
(in degrees Kelvin). entropy effect greatly reduces the overall ΔF of binding, so
It can be shown that the association constant K A will thus they depend on a very large negative ΔH to bind. In contrast,
vary with temperature as follows: monoclonal b12 has moderate levels of both ΔH and −TΔS,
ln K A ΔH°
d______ resulting in a nearly identical ΔF. Monoclonal 2G12 binds
= ____2 (6) gp120 with a less favorable ΔH but no entropy cost.
dT RT
The two discrete conformations of gp120 may be called the
or equivalently, open and closed forms. The excursion between these two con-
ln K A −ΔH°
d______ formations affects the entropy of binding. CD4 and F105 only
= ____ (6′) bind the open form, which reduces randomness and creates
d(1/T) R
an entropy barrier. Monoclonal b12 is less dependent on one
The derivation of these equations is beyond the scope of conformation, so it binds with less entropy cost, and 2G12 is
this book.4 However, the practical implications are as follows. indifferent to the two forms, so it has no entropy effect at all.
First, one can determine the standard ΔH° of the reaction The strong entropy effect observed for these monoclonals
from the slope of a plot of ln K A versus 1/T. Second, for an may illustrate one way for the virus to evade antibody bind-
interaction that is primarily exothermic (ie, driven by a large ing, through a mechanism called conformational masking.
negative ΔH, such as the formation of hydrogen bonds and
polar bonds), the affinity decreases with increasing tempera-
ture. Thus many antigen–antibody interactions have a higher
affinity at 4°C than at 25°C or 37°C, so maximum binding Energetics of Antibody Binding to
for a given set of concentrations can be achieved in the cold. TABLE 7.1
gp120 Core Structure
In contrast, apolar or hydrophobic interactions are driven
largely by the entropy term T ΔS, and ΔH° is near zero. In this Ligand ΔF ΔH −TΔS
case, there is little effect of temperature on the affinity. CD4 −10.4 −48.4 38.2
As for the effects of pH and salt concentration (or ionic F105 −11.4 −30.0 18.6
strength) on the affinity, these vary depending on the nature b12 −12.3 −18.0 5.7
2G12 −7.8 −6.2 −1.6
†
For a more complete description of these concepts, see a physical chemistry text
such as Moore.4 CD, cluster of differentiation. From Kwong, et al.6

The randomness of gp120, as found on the virus, means that to 4. Thus, for the interaction between two molecules of
it is rarely in the open conformation, making it difficult for equal size, the diffusive rate constant is the same regardless
most antibodies such as F105 to bind. Only the rare anti- of whether those molecules are large or small.8 However, if
body that is b12-like can bind multiple forms and eventually one molecule is large and the other small, the rate is greater
pull gp120 into its most favorable conformation. than if both molecules are large. This difference occurs be-
The same effect may explain the difficulty in eliciting cause reducing the radius r1 while keeping r2 constant (and
antibodies to this site using native gp120. If it is rarely in the larger than r1) has a greater effect on increasing the diffusion
open form, it will be unable to trigger B cells to make anti- constant term D, proportional to (1/r1 + 1/r2), in which the
bodies that require this form, and it will deliver only a weak smaller radius produces the larger term than it has on the
antigenic stimulus to those that bind partially to different term a, which is still dominated by the larger radius r2. For
conformations, such as b12. These considerations suggest example, if r2 = r as shown, but r1 = 0.1r, then the numerator
that a more favorable vaccine antigen could be made if gp120 in Equation 7b is only reduced from 4r2 to 1.21r2, whereas
could be anchored in the open conformation, so it could the denominator is reduced from 1r2 to 0.1r2. Thus, the ratio
stimulate B cells to make antibodies that require this form. is increased from 4 to 12.1. Viewed another way, the greater
The hallmark of this structure would be its ability to bind diffusive mobility of the small hapten outweighs its dimin-
F105 and b12 with a good ΔH and reduced values of −TΔS. ished target area relative to a large protein antigen because
the larger target area of the antibody is available to both.
The dissociation rate (or “off rate”) k −1 is determined by
Kinetics of Antigen–Antibody Reactions the strength of the bonds (as it affects the activation energy
A fundamental connection between the thermodynamics barriers for dissociation) and the thermal energy kT (where
and kinetics of antigen–antibody binding is expressed by k is Boltzmann constant), which provides the energy to sur-
the relationship mount this barrier. The activation energy for dissociation is
k1 the difference in energy between the starting state and the
K A = ___ (7) transition state of highest energy to which the system must
k −1
be raised before dissociation can occur.
where k1 and k −1 are the rate constants for the forward As pointed out by Eisen,9 if one compares a series of re-
(association) and backward (dissociation) reactions. lated antigens, of similar size and other physical properties,
The forward reaction is determined largely by diffu- for binding to an antibody, the association rates are all very
sion rates (theoretical upper limit 109 L/mol/sec) and by similar. The differences in affinity largely correspond to the
the probability that a collision will result in binding, that differences in dissociation rates.
is, largely the probability that both the antigen and the an- A good example is that of antibodies to the protein anti-
tibody will be oriented in the right way to produce a good gen staphylococcal nuclease.10 Antibodies to native nuclease
fit as well as the activation energy for binding. The diffu- were fractionated on affinity columns of peptide fragments
sive rate constant can be shown7 to be approximated by the to isolate a fraction specific for residues 99 through 126.
Smoluchowski equation The antibodies had an affinity of 8.3 × 108 M−1 for the native
antigen and an association rate constant, kon, of 4.1 × 105
kdl = 4πaD(6 × 1020) (7a)
M−1 ⋅ sec−1. This kon was several orders of magnitude lower
where a = sum of the radii in centimeters of the two reactants, than had been observed for small haptens,11 as discussed
D = sum of the diffusion constants in cm2/sec for the individ- previously. A value of koff of 4.9 × 10−4 sec−1 was calculated
ual reactants, and the constant 6 × 1020 is necessary to con- using these results in Equation 7. This is a first-order rate
vert the units to M−1 ⋅ sec−1. For example, if a = 10−6 cm and constant from which one can calculate a halftime for dis-
D = 10−7 cm2/sec, then kdl ≈ 7.5 × 108 M−1 ⋅ sec−1. Association sociation (based on t1/2 = ln 2/koff ) of 23 minutes. These rates
rates will generally be slower for large protein antigens than are probably typical for high-affi nity (K A ≈ 109 M−1) anti-
for small haptens. This observation may be due to the smaller bodies to small protein antigens such as nuclease (molecular
value of D, to the orientational effects in the collision, and weight [MW] ≈ 17,000). The dissociation rate is important
to other nondiffusional aspects of protein–protein interac- to know in designing experiments to measure binding be-
tions. Therefore, association rates for protein antigens are cause if the act of measurement perturbs the equilibrium,
more frequently on the order of 105 to 106 M−1 ⋅ sec−1 (see fol- the time one has to make the measurement (eg, to separate
lowing discussion). However, this observation can also be bound and free) is determined by this halftime for dissocia-
partly understood from diffusion-limited rates alone. If the tion. For instance, a 2-minute procedure that involves dilu-
radii of hypothetically spherical reactants are r1 and r2, then tion of the antigen–antibody mixture can be completed be-
in Equation 7a, a = r1 + r2, whereas D is proportional to 1/r1 + fore significant dissociation has occurred if the dissociation
1/r2. The diffusive rate constant is therefore proportional to halftime is 23 minute. However, if the on rate is the same,
but the affinity 10-fold lower, still a respectable 8 × 107 M−1,
1 __ 1 (r1 + r2)
2
(
(r1 + r2) __ )
_______
r1 + r2 = r1r2 (7b) then the complex could be 50% dissociated in the time re-
quired to complete the procedure. This caution is relevant
From this result, it can be seen that if r1 = r2 = r, then r can- when we discuss methods of measuring binding and affinity
cels out and the whole term in Equation 7b is simply equal in the following.

Because knowledge of the dissociation rate can be so vary the antibody concentration, but then, the analysis is
important in the design of experiments, a word should be slightly more complicated. Perhaps the theoretically most
said about techniques to measure it. Perhaps the most widely elegant experimental method to determine these quanti-
applicable one is the use of radiolabeled antigen. After equi- ties is equilibrium dialysis,12,13 depicted and explained in
librium is reached and the equilibrium concentration of Figure 7.2, in which ligand (antigen) is allowed to equili-
bound radioactivity determined, a large excess of unlabeled brate between two chambers, only one of which contains
antigen is added. Because any radioactive antigen molecule antibody, separated by a semipermeable membrane imper-
that dissociates is quickly replaced by an unlabeled one, meable to antibody. The important feature of this method,
the probability of a radioactive molecule associating again as opposed to most others, is that the concentrations of
is very small. Therefore, one can measure the decrease in ligand in each chamber can be determined without per-
radioactivity bound to antibody with time to determine the turbing the equilibrium. The disadvantage of this method is
dissociation rate.‡ that it is applicable only to antigens small enough to perme-
ate freely a membrane that will exclude antibody. Another
technical disadvantage is that bound antigen, determined
AFFINITY as the difference between bound plus free antigen in one
It is apparent from the previous discussion that a lot of chamber and free antigen in the other, is not measured in-
information about an antigen–antibody reaction is packed dependently of free antigen.
into a single value, its affinity. In this section, we examine Another category of method uses radiolabeled ligand
affinity more closely, including methods for measuring in equilibrium with antibody and then physically sepa-
affinity and the heterogeneity thereof, the effects of multiva- rates free antigen bound to antibody and quantitates each
lency of antibody and/or of antigen, and the special effects separately. The methods used to separate bound and free
seen when the antigen–antibody interaction occurs on a antigen are discussed in the section on radioimmunoassay
solid surface (two-phase systems). (RIA). These methods generally allow independent mea-
surement of bound and free antigen but may perturb the
Interaction in Solution with Monovalent Ligand equilibrium.
The simplest case is that of the interaction of antibody with
Scatchard Analysis
monovalent ligand. We may include in this category both
Once data are obtained, there are a number of methods
antihapten antibodies reacting with truly monovalent hap-
of computing the affinity, of which we shall discuss two.
tens and antimacromolecule antibodies, which have been
fractionated to obtain a population that reacts only with a
single, nonrepeating site on the antigen.** In the latter case,
the antigen behaves as if monovalent in its interaction with
the particular antibody population under study. The pro-
viso that the site recognized (antigenic determinant) be
nonrepeating, that is, occur only once per antigen molecule,
of course, is critical.
If the combining sites on the antibody are independent
(ie, display no positive or negative cooperativity for anti-
gen binding), then for many purposes one can treat these
combining sites, reacting with monovalent ligands, as if
they were separate molecules. Thus, many, but not all, of the
properties we discuss can be analyzed in terms of the con-
centration of antibody-combining sites, independent of the
number of such sites per antibody molecule (2 for IgG and
IgA, 10 for IgM).
To determine the affi nity of an antibody, one generally FIG. 7.2. Equilibrium Dialysis. Two chambers are separated by a
determines the equilibrium concentrations of bound and semipermeable membrane that is freely permeable to ligand but not
free ligand, at increasing total ligand concentrations, but at all to antibody. Antibody is placed in one chamber (B) and ligand
at constant antibody concentration. Alternatively, one can in one or both chambers. Regardless of how the ligand is distributed
initially, after sufficient time to reach equilibrium, it will be distrib-
uted as follows. The concentration of free ligand will be identical in
‡
This method assumes that all binding sites are independent, as is generally true for both chambers, but chamber B will have additional ligand bound to
antibodies and monovalent ligands. If there were either negative or positive coop- antibody. The concentration of bound ligand is thus the difference
erativity in binding, then the change in receptor occupancy that occurs when a large between the ligand concentration in the two chambers, whereas
excess of unlabeled antigen is added would probably perturb the dissociation rate of the free concentration is the concentration in chamber A. Because
radiolabeled antigen molecules already bound to other sites.
** these concentrations must obey the mass action law, Equation 2,
Such fractionated antibodies may contain mixtures of antibodies to overlapping
sites within a domain on the antigen, but as long as no two antibody molecules (or they can be used to determine the affinity KA, from Equation 3 or 3′,
combining sites) can bind to the same antigen molecule simultaneously, the antigen by any of several graphical procedures, such as Scatchard analysis
still behaves as effectively monovalent. (described in the text).

FIG. 7.3. Scatchard analysis of the binding

of [3H]-sperm whale myoglobin by a mono-
clonal antibody to myoglobin (A) and by the
serum antibodies from the same mouse
whose spleen cells were fused to prepare
the hybridoma (B). The monoclonal antibody
(clone HAL 43-201E11, clone 5) produces a
linear Scatchard plot, whose slope, −1.6 × 109
M−1, equals −KA, and whose intercept on the
abscissa gives the concentration of antibody-
binding sites. In contrast, the serum antibod-
ies produce a curved (concave up) Scatchard
plot, indicative of heterogeneity of affinity.
A B
From Berzofsky et al.,15 with permission.
Perhaps the most widely used is that described by Scatchard14 no assumptions about the number of sites per molecule)
(Fig. 7.315). The mass action equilibrium law is plotted in the to obtain
form of Equation 3
R = K (n − r)
__ (9)
[SL] = K A([S]t − [SL])[L] (3) c A
and B is substituted for [SL] and F for [L], referring to bound where r = the number of occupied sites per antibody mol-
and free ligand, respectively. Then the Scatchard equation is ecule, n = the total number of sites per antibody molecule,
and c = free ligand concentration, that is, c = F. Thus,
B = K ([S] − B)
__ (8)
A t
F B B
r = _____________ = ____
[total antibody] [A]t
Note that a very critical implicit assumption was made in
[total sites] [S]
this seemingly very simple conversion. The [SL] within the n = _____________ = ____t
[total antibody] [A]t
parentheses in Equation 3 was intended to be the concentra-
tion of bound antibody sites so that ([S]t − [SL]) = free [S]. where [A]t = total molar antibody concentration. In this
However, in Equation 8, we have substituted B, the concen- form of the Scatchard plot, r/c versus r, the slope is still −K A
tration of bound ligand. If the ligand behaves as monovalent, and the intercept on the r axis is n. Thus one can determine
then this substitution is legitimate, as every bound ligand the number of sites per molecule. Of course, if one deter-
molecule corresponds to an occupied antibody site. However, mines [S]t from the intercept of Equation 8, one can also cal-
if the ligand is multivalent and can bind more than one anti- culate the number of sites per molecule by dividing [S]t by
body site, then Equation 8 is valid only in ligand excess where any independent measure of antibody concentration. Thus,
the frequency of ligands with more than one antibody bound the only advantage of normalizing all the data points first to
is very low. In this section, we are discussing only monova- plot the r/c form arises when the data were obtained at vary-
lent ligands, but this proviso must be kept in mind when the ing antibody concentrations. If the antibody concentration
Scatchard analysis is applied in other circumstances. is unknown but held constant, then the B/F form is more
From Equation 8, we see that a plot of B/F versus B convenient and actually provides one measure of antibody
should yield a straight line (for a single affi nity), with a (site) concentration. Because today we know the value of
slope of −K A and an intercept on the abscissa correspond- n for each class of antibody (2 for IgG and serum IgA, 10
ing to antibody-binding site concentration (see Fig. 7.3). for IgM), the concentration of sites and that of antibody are
This is the so-called Scatchard plot. An alternative ver- easily converted in many cases.
sion that is normalized for antibody concentration is es-
pecially useful if the data were obtained at different values Heterogeneity of Affinity
of total antibody concentration, [A]t, instead of constant The next level of complexity arises when one is dealing with a
[A]t. However, for this version, one requires an indepen- mixture of antibodies of varying affinity for the ligand. This
dent measure of total antibody concentration, other than is the rule, rather than the exception, when one deals with
the intercept of the plot. Then one divides Equation 8 by antibodies from immune serum, even if they are fraction-
the total concentration of antibody molecules (making ated to be monospecific, that is, all specific for the same site

on the antigen. Contrast, for example, the linear Scatchard

plot for a homogeneous monoclonal antibody to myoglobin
(see Fig. 7.3A), with the curved Scatchard plot for the serum
antibodies from the same mouse used to prepare the hybrid-
oma monoclonal antibody (see Fig. 7.3B). This concave up
Scatchard plot is typical for heterogeneous antibodies. In
a system such as hormone receptor–hormone interaction,
in which negative cooperativity can occur between recep-
tor sites (ie, occupation of one site lowers the affi nity of its
neighbor), a concave up Scatchard plot can be produced by
negative cooperativity in the absence of any intrinsic hetero-
geneity in affinity. However, in the case of antibodies, where
no such allosteric effect has been demonstrated, a concave
up Scatchard plot indicates heterogeneity of affinity.
Ideally, one would like to imagine that the tangents all
along the curve correspond (in slope) to the affinities of the
many subpopulations of antibodies. Mathematically, this is
not strictly correct, but it is true that the steeper part of the
curve corresponds to the higher affinity antibodies and the
shallower part of the curve to the lower affinity antibodies.
Graphical methods have been developed to analyze more
quantitatively the components of such curves,16,17 and a very
general and versatile computer program (LIGAND) has been
developed by Munson and Rodbard18 that can fit such curves
using any number of subpopulations of different affinity. For FIG. 7.4. Analysis of a Curved Scatchard Plot Produced by a Mixture
of Two Antibodies with Different Affinities. The antibodies have
purposes of this chapter, we discuss only the case of two af- affinities K1 and K2 and have n1 and n2 binding sites per molecule,
finities and then examine the types of average affinities that respectively. The r is the concentration of bound antigen divided by
have been proposed when one is dealing with much greater the total antibody concentration (ie, bound sites per molecule), and
heterogeneity. We also examine mathematical estimates of c is the free antigen concentration. The curve is a hyperbola that
the degree of heterogeneity (analogous to a variance). can be decomposed into its two asymptotes, which correspond to
When an antibody population consists of only two sub- the linear Scatchard plots of the two components in the antibody
mixture. The tangents to the curve at its intercepts only approximate
populations of different affinities, K1 and K 2, then we can
these asymptotes so that the slopes of the tangents estimate but
add the component Equations 3′ to obtain do not accurately correspond to the affinities of the two antibodies.
n1K1c n 2K 2c However, the intercept on the r axis corresponds to n1 + n2. Note
r = r1 + r2 = ________ + ________ (10)
(1 + K1c) (1 + K 2c) that in this case n1 and n2 must be defined in terms of the total anti-
body concentration, not that of each component.
so that
n1K1
__r = ________ n 2K 2
+ ________ (10′)
c (1 + K1c) (1 + K 2c) Average Affinities
where the subscripts correspond to the two populations. In practice, of course, one rarely knows that one is deal-
Then the graph of r/c versus r can be shown to be a hyperbola ing with exactly two subpopulations, and most antisera are
whose asymptotes are, in fact, the linear Scatchard plots of significantly more heterogeneous than that. Therefore, the
the two components (Fig. 7.4). This situation has been ana- previously mentioned case is more illustrative of principles
lyzed graphically by Bright.19 Taking the limits as c → 0 and than of practical value. When faced with a curved Scatchard
as c → ∞, it can easily be shown that the intercept on the ab- plot, one usually asks what the average affinity is, and per-
scissa is just n1 + n2 (or, in the form B/F versus B, the intercept haps some measure of the variance of the affi nities, with-
is the total concentration of binding sites [S]t), and the inter- out being able to define exactly how many different affinity
cept on the ordinate is n1K1 + n2K2. Thus, one can still obtain populations exist.
the total value of n or [S]t from the intercept on the abscissa. Suppose one has m populations each with site concentra-
The problem is in obtaining the two affinities, K1 and K2, and tion [Si] and affinity K i, so that at free ligand concentration
the concentrations of the individual antibody subpopula- [L], the fraction of each antibody that has ligand bound will
tions (corresonding to n1 and n2). If K1 is greater than K2, one be given by an equation of the form of Equation 3′:
can approximate the affinities from the slopes of the tangents
K i[Si]t[L]
at the two intercepts (see Fig. 7.4), but these will not, in gen- Bi = _________ (11)
eral, be exactly parallel to the two asymptotes, which give the (1 + Ki[L])
true affinities, so some error is always introduced, depending Then the bound concentrations sum to give
on the relative values of n1 and n2 and K1 and K2. A graphical m m
method for solving for these exactly has been worked out by
Bright19 and computer methods by Munson and Rodbard.18
B= Σ B = Σ _________
i=1
i
K [S ] [L]
i
(1 + K [L])
i=1
i t
i
(11′)

Substituting F for [L] and dividing through by this quantity, From Equations 13 and 14, it is apparent that Kav is simply
one obtains the ratio of the two intercepts on the B/F and B axes, that is,
m the slope of the chord (see Fig. 7.5). This type of weighted
B =
__
F
Σ ________
K [S ]
(1 + K F)
i=1
i i t
i
(12) mean affinity, Kav, is therefore actually easier to obtain
graphically in some cases than K0, and we shall see that it is
or equivalently, useful in other types of plots as well.
m
__r =
c Σ ________
Kn
(1 + K c)
i=1
i i
(12′) Indices of Heterogeneity: The Sips Plot
i For a heterogeneous antiserum, one would also like to have
These can be seen to be generalizations of Equations 10 and some idea of the extent of heterogeneity of affinity. For
10′. Taking the limits as F → 0 and F → ∞, one again sees instance, if the affinities were distributed according to a
that the normal (Gaussian) distribution, one would like to know the
m variance.21,22 More complex analyses have been developed
intercept on ordinate = Σ K [S ]
i=1
i i t (13) that do not require as many assumptions about the shape
of the distribution,23–25 but the first and most widely used
and the index of heterogeneity arbitrarily assumes that the affinities
m fit a distribution, first described by Sips,26 which is similar in
intercept on abscissa = Σ [S ] = [S]
i=1
i t t (14) shape to a normal distribution. This was applied to the case
of antibody heterogeneity by Nisonoff and Pressman, 27 and
Therefore, one can still obtain the total antibody site con- is summarized by Karush and Karush.28 One fits the data to
centration from the intercept on the abscissa (Fig. 7.5).20 the assumed binding function
Two types of average affinity can be obtained graphically n(K0c) a
from the Scatchard plot.20 Perhaps the more widely used K0 r = __________ (16)
(1 + (K0c) a)
is actually more accurately a median affinity rather than a
mean affinity. It is defined as the slope of the tangent at the which is analogous to Equations 3′ and 11 (the Langmuir
point on the curve where half the sites are bound, that is, adsorption isotherm) except for the exponent a, which is
where B = [S]t/2 (see Fig. 7.5). A second type of average af- the index of heterogeneity. This index, a, is allowed to range
finity, which we call Kav, is a weighted mean of the affinities, from 0 to 1. For a = 1, Equation 16 is equivalent to Equation 3,
each affinity weighted by its proportional representation in and there is no heterogeneity. As a decreases toward 0, the
the antibody population. Thus, we take the ratio heterogeneity increases. To obtain a value for a graphically,
m one plots the algebraic rearrangement of Equation 16:
Kav = Σ K_____
[S ] i i t
(15) r
log ( _____
n − r ) = a log c + a log K0
i=1 [S] t (17)
so that the slope of log [r/(n − r)] versus log c is the hetero-
geneity index a.
C. DeLisi (personal communication) has derived the
variance (second moment) of the Sips distribution in terms
of the free energy RT ln K0, about the mean of free energy.
The result (normalized to RT) gives the dispersion or width
of the distribution as a function of a:
σ2Sips ________
____ π2(1 − a2)
= (18)
R 2T 2 3a2
This is useful for determining a quantity, σSips, which can
be thought of as analogous to a standard deviation, if one
keeps in mind that this is not a true Gaussian distribution.
In addition, as noted previously, the use of the Sips distribu-
tion requires the assumption that the affinities (really the
free energies) are continuously distributed symmetrically
about a mean, approximating a Gaussian distribution. This
assumption frequently is not valid.
FIG. 7.5. Types of Average Affinities for a Heterogeneous Population of The Plot of B/F versus F or T
Antibodies, as Defined on a Scatchard Plot. K0 is the slope of the tan- Another graphical method that is useful for estimating
gent to the curve at a point where B = [S]t/2, that is, where half the an-
tibody sites are bound. Thus, K0 corresponds to a median affinity. The
affinities is the plot of bound/free versus free or total ligand
Kav is the slope of the chord between the intercepts and corresponds concentration, denoted F and T, respectively20 (Fig. 7.6). To
to a weighted average of the affinities weighted by the concentrations simplify the discussion, let us define the bound/free ratio,
of the antibodies with each affinity. Adapted from Berzofsky.20 B/F, as R, and define R0 as the intercept, or limit, as free

seen from Equation 23, the assumption that the midpoint is

1/K will result in an error equal to half the antibody-binding
site concentration. Thus, in plots of B/F versus T, the mid-
point will be a good estimate of the affinity only if [S]t/2 <<
1/K, that is, if the antibody concentration is low compared
to the dissociation constant. In fact, if the affinity is so high
that 1/K << [S]t/2, then one will merely be measuring the
antibody concentration, not the affinity at all20 (see Fig. 7.6).
In the case of a heterogeneous antiserum, we have already
seen that
R0 = Σ K [S ]
i
i i t (13)
Therefore, at the midpoint, when B/F = R0 /2, it is easy to

see that
B ___ R0
2 = ___
FIG. 7.6. Schematic Plot of R, the Bound/Free Ratio, as a Function
of Free (F) or Total (T) Antigen Concentration. The curves have a
Kav = __( )( )
F [S]t [S]t
(24)
similar sigmoidal shape, but the midpoint (where R = R0/2) of the plot
of R versus T has a term dependent on antibody site concentration Thus, one can still obtain the average affinity, as defined
([S]t), whereas the midpoint of the plot of R versus F is exactly 1/K, previously.20
independent of antibody concentration. Adapted from Berzofsky.20 Regardless of average affinities, the effect of affinity het-
erogeneity is to broaden the curve or to make the slope shal-
lower. This can be seen by visualizing the curve of B/F ver-
ligand F → 0. First, for the case of a homogeneous antibody, sus F as a step function. Each antibody subpopulation of a
from Equation 3′, given affinity, K i, will be titrated to 50% of its microscopic
K[S]t
B = _______ B/F at a free ligand concentration F = 1/K i. The high-affinity
R = __ (19)
F (1 + KF) antibodies will be titrated at low F, but the low-affi nity anti-
bodies will require much higher F to be titrated. The result-
and ing step function is analogous to the successive transitions
B = K[S] corresponding to different pK values in a pH titration.
R0 = lim __ t (20)
F→0 F
Intrinsic Affinity
Let us define the midpoint of the plot (see Fig. 7.6) as the The affinity, K A, that we have been discussing so far is what
point at which R decreases to half its initial value, R0, that is, has been termed the intrinsic affinity, that is, the affinity of
at which R = K[S]t/2. For the case of homogeneous antibody each antibody-combining site treated in isolation. We have
(ie, a single affinity), simple algebraic manipulation,20 sub- been able to do this, regardless of the valence of the anti-
stituting K[S]t/2 (ie, R0 /2) for B/F in Equation 8, will show bodies, by using the concentration of combining sites, [S],
that at this midpoint†† in our equations rather than the concentration of antibody
molecules, [A], which may have more than one site. Even
1
F = __ (21) without any cooperativity between combining sites, there
K
is a statistical effect that makes the actual affinity different
and from the intrinsic affinity if the antibody is multivalent and
one uses whole antibody concentration rather than site con-
[S]
B = ___t (22) centration. The way this difference arises can best be seen
2
by examining the case of a bivalent antibody, such as IgG.
so that the total concentration, T, is We assume that the two sites are equivalent and neither is
affected by events at the other. The ligand, as in this whole
[S] 1
T = B + F = ___t + __ (23) section, is monovalent. Then there are two binding steps
2 K
K1 K2
Thus, if one plots B/F versus F, the midpoint directly A + L D AL, AL + L D AL2 (25)
yields 1/K. However, it is frequently more convenient
experimentally to plot B/F versus T. In this case, the mid- and the corresponding actual affinities are
point is no longer simply the reciprocal of the affinity. As [AL] [AL2]
K1 = ______ , K 2 = _______ (26)
[A][L] [AL][L]
††
It is important to note that R0 must be the limit of B/F as F truly approaches zero. If the intrinsic affinity of both equivalent sites is K, then
In an RIA in which the concentration of tracer is significant compared to 1/K, reduc- K1 will actually be twice K because the concentration of
ing the unlabeled ligand concentration all the way to zero will still not yield the true
limit R0. The tracer concentration must also be negligible. If not, R0 will be estimated available sites [S] will be twice the antibody concentration
falsely low, and the affi nity will also be underestimated. when the first ligand is about to bind in step 1. However,

once one site is bound, the reverse (dissociation) reaction more than one identical determinant of a multivalent an-
of step 1 can occur from only one site, namely, that which is tigen molecule. Karush 30 has termed this phenomenon
occupied. Conversely, for the second step, the forward reac- “monogamous bivalency.” Such monogamous binding
tion has only one remaining available site; however, in the can occur between two molecules in solution, or between
reverse reaction, AL2 → AL + L, either site can dissociate a molecule in solution and one on a solid surface, such as
to go back to the AL state. The second site bound need not a cell membrane or microtiter plate. We fi rst discuss the
be the first to dissociate, and because the sites are identical, situation in solution and then discuss the additional con-
one cannot tell the difference. Thus, for step 2, the apparent siderations that apply when one of the reactants is bound
concentration of sites for the reverse reaction is twice that to a solid surface.
available for the forward reaction, so the affinity K 2 for the
second step will be only half the intrinsic affinity, K. Monogamous Bivalency
It is easy to see how this statistical effect can be extrapo- Suppose a divalent antibody molecule reacts with antigen
lated to an antibody with n sites29 : that has two identical determinants. This situation has been
treated in detail by Crothers and Metzger,31 and by Karush.30
1
K1 = nK and ( )
K n = __
n K (27) Let us call the two antibody sites S and S⬘, and the antigenic
determinants D and D⬘, with the understanding that, in
For the steps in between, two derivations are available,9,29 actuality, we cannot distinguish S from S⬘ or D from D⬘.
which yield The interaction can be broken up into two steps, a bimo-
lecular reaction
(n − i + 1)
Ki = _________ K (28) S D SM D
i K
⏐+ ⏐ D⏐
1
⏐ (29)
The actual affinity, rather than the intrinsic affinity, S⬘ D⬘ S⬘ D⬘
becomes important with monovalent ligands when one is
interested in the effective affinity (based on a molar anti- followed by an intramolecular reaction
body concentration) under conditions where [L] is so low SM D SM D
K
that only one site can bind antigen. Then for IgG or IgM ⏐ ⏐ D⏐ ⏐
2
(30)
(with 2 or 10 sites per molecule, respectively), the apparent S⬘ D⬘ S⬘ M D⬘
affinity will be theoretically 2 or 10 times the intrinsic affin-
ity. For most purposes, it is easier to use site concentrations The association constant for the first step, K1, is related to the
and intrinsic affinities. The analyses given previously, such intrinsic affinity, K, simply by a statistical factor of 4 due to
as B/F versus F or the Scatchard plot, whether B/F versus the degeneracy (equivalence) between S and S⬘ and between
B or r/c versus r, will all yield intrinsic affinities. It is the D and D⬘. This is a typical second-order reaction between
intrinsic affinity that tells us something about the nature of antigen and antibody. However, the second step (Equation
the antibody–ligand interaction. 30) is a first-order reaction because it is effectively an inter-
Once one enters the realm of multivalent ligands, the conversion between two states of a single molecular com-
actual affinity or effective affinity involving multipoint plex, the reactants S⬘ and D⬘ being linked chemically (albeit
binding between multivalent antibody molecule and multi- noncovalently) through the S−D bond formed in the first
valent ligand molecule can be much greater than the intrin- step. Thus, the first-order equilibrium constant, K 2, is not a
sic affinity for binding at each site. This case is the subject of function of the concentrations of S–S and D–D in solution,
the next section. as K1 would be. Rather, the forward reaction depends on the
geometry of the complex and the flexibility of the arms; in
other words, the probability that S⬘ and D⬘ will encounter
Interaction with Multivalent Ligands each other and be in the right orientation to react if they do
So far, we have discussed only situations in which the li- come in contact depends on the distances and freedom of
gand is monovalent or effectively monovalent with respect motion along the chain S⬘−S−D−D⬘ rather than on the den-
to the particular antibody under study. However, in many sity of molecules in solution (ie, concentration).
situations, the ligand molecule has multiple repeating The reverse reaction for step 2, on the other hand, will
identical determinants, each of which can bind indepen- have a rate constant similar to that for the simple mon-
dently to the several identical combining sites on a di- ovalent S−D → S + D reaction, as the dissociation reaction
valent or multivalent antibody.‡‡ Although the intrinsic depends on the strength of the S⬘−D⬘ (or S−D) bond and is
affi nity for the interaction of any single antibody–com- not influenced by the other S−D interaction unless there is
bining site with any single antigenic determinant may strain introduced by the angles required for simultaneous
be the same as that discussed in the preceding section, bonds between S and D and S⬘ and D⬘. Note that K 2 will
the apparent or effective affi nity may be much higher inherently have a statistical factor of 1/2 compared to the
due to the ability of a single antibody molecule to bind intrinsic K⬘2 for the analogous reaction if the S⬘−S−D−D⬘
link were all covalent because in the forward reaction of
‡‡
Equation 30 only one pair can react, whereas in the reverse
If only the antigen is multivalent, and the antibody monovalent, such as an Fab
fragment, the situation can be analyzed using the same statistical considerations reaction either S⬘–D⬘ or S−D could dissociate to produce the
discussed previously. equivalent result.

We would like to know the apparent or observed affinity cell surface or an artificial surface (such as Sepharose or the
for the overall reaction plastic walls of a microtiter plate), the reaction of a multi-
valent ligand with antibodies on the surface of a B cell, a
S D SM D
K Sepharose bead, or a plastic plate, and the reaction of either
⏐+ ⏐ D ⏐
obs
⏐ (31) component with an antigen–antibody precipitate. For the
S D SM D reasons outlined previously, “monogamous” binding can
Because the free energies, ΔF1 and ΔF2, for the two steps are make the apparent affinity of a multivalent antibody or an-
additive, the observed affinity will be the product of K1 and K2 tigen for multiple sites on a solid surface be quite large to the
point of effective irreversibility.
Kobs = K1K 2 (32) However, another effect also increases the effective af-
where we have defi ned K1 and K 2 to include the statistical fi nity in a two-phase system. This effect applies even for
degeneracy factors.*** The equilibrium constants K1 and K 2 monovalent antibodies (Fab fragments) or monovalent
are each the ratios of forward and reverse rate constants, ligands. The effect arises from the enormously high ef-
as in Equation 7. Of these four rate constants, all are di- fective local concentration of binding sites at the surface,
rectly related to the corresponding terms for the intrinsic compared to the concentration if the same number of sites
affi nity between S and D except for the intramolecular for- were distributed in bulk solution.33 Looked at another
ward reaction of step 2, as noted previously. Thus, the diffi- way, the effect is due to the violation, at the liquid–solid
culty in predicting Kobs is largely a problem of analyzing the interface, of the basic assumption in the association con-
geometric (steric) aspects of K 2, assuming one already knows stants, K A, discussed previously, that the reactants are all
the intrinsic affi nity, Crothers and Metzger31 have analyzed distributed randomly in the solution. (To some extent, the
this problem for particular situations. Qualitatively, we can latter is involved in the enhanced affi nity of multivalency
say that whether K 2 will be larger or smaller than K will as well.) This situation has been analyzed by DeLisi34 and
depend on factors such as the enforced proximity of S⬘ and DeLisi and Wiegel, 35 who break the reaction down into two
D⬘ in step 2 and the distance between D and D⬘ compared steps: the diffusive process necessary to bring the antigen
to the possible distances accessible between S and S⬘, which and antibody into the right proximity and orientation to
in turn depends on the length of the antibody arms and react, and the reactive process itself. The complex between
the flexibility of the hinge between them. Thus, because K1 antigen and antibody, when positioned but not yet reacted,
can be approximated by K, except for statistical factors, the is called the encounter complex. The reaction can then be
apparent affi nity for this “monogamous bivalent” binding written
interaction, Kobs, may range from significantly less than to k+ k1
significantly greater than K 2. If K 2 is of the same order of S + D D S ⋅⋅⋅ D D SD (33)
magnitude as K, then Kobs will be of the order of K 2, which k− k −1
can be huge (eg, if K ≈ 109 M−1, Kobs could be ≈1018 M−1). The
halftime for dissociation would be thousands of years. It where S = antibody site, D = antigenic determinant, k+ and
is easy to see how such monogamous bivalent interactions k − = forward and reverse diffusive rate constants, and k1 and
can appear to be irreversible, even though in practice the k −1 = forward and reverse reactive rate constants once the
observed affi nity is rarely more than a few orders of mag- encounter complex is formed. If the encounter complex is
nitude larger than the K for a single site, possibly due to in a steady state, the overall rate constants will be given by
structural constraints.32 k1k +
If apparent affi nities this high can be reached by monog- k f = _______ (34)
(k1 + k −)
amous bivalency, even greater ones should be possible for
k −1k −
the multipoint binding of an IgM molecule to a multivalent k r = _______ (35)
ligand. Although IgM is decavalent for small monovalent (k1 + k −)
ligands, steric restrictions often make it behave as if pen- where subscripts f and r = forward and reverse.34 The asso-
tavalent for binding to large multivalent ligands. However, ciation constant, according to Equation 7, is the ratio of
even five-point binding can lead to enormously tight in- these two, or
teractions. Therefore, even though the intrinsic affi nity of
IgM molecules tends to be lower than that of IgG molecules k1k +
K A = _____ (36)
for the same antigen,30 the apparent affi nity of IgM can be k −1k −
quite high.
The relative magnitudes of k1 and k – determine the probable
fate of the encounter complex. Is it more likely to react to
Two-Phase Systems form SD or to break up as the reactants diffuse apart?
The same enhanced affinity seen for multipoint binding Now suppose that k- is slow compared to k1. Then the
applies to two-phase systems. Examples include the reac- SD bound complex and the encounter complex, S ⋅⋅⋅ D,
tion of multivalent antibodies with antigen attached to a may interconvert many times before the encounter com-
plex breaks up and one of the reactants diffuses off into
***
In some treatments where these statistical factors are not included in K1 and K 2, the bulk solution. If the surface has multiple antigenic sites,
equivalent equation may be given as Kobs = 2K1K 2. D, then even a monovalent antibody (Fab) may be much

more likely, when SD dissociates to S ⋅⋅⋅ D, to rereact with RADIOIMMUNOASSAY AND RELATED METHODS
the same or nearby sites than to diffuse away into bulk so- Since it was first suggested in 1960 by Yalow and Berson,37
lution, again depending on the relative magnitudes of these RIA has rapidly become one of the most widespread, widely
rate constants. This greater probability to rereact with the applicable, and most sensitive techniques for assessing the
surface rather than diffuse away is the essence of the effect concentration of a whole host of biologic molecules. Most of
we are describing. A more extensive mathematical treat- the basic principles necessary to understand and apply RIA
ment of reactions with cells is given in DeLisi34 and DeLisi have been covered previously in this chapter. In this section,
and Wiegel.35 we examine the concepts and methodologic approaches used
A somewhat different, and very useful, analysis of the in RIA. For a detailed methods book, we refer the reader to
same or a very similar effect was given by Silhavy et al.36 Chard,38 Rodbard,39 and Yalow.40
These authors studied the case of a ligand diffusing out of a The central concept of RIA is that the binding of an in-
dialysis bag containing a protein for which the ligand had a finitesimal concentration of highly radioactive tracer antigen
significant affinity. Once the ligand concentration became to low concentrations of a high-affinity–specific antibody is
low enough that there was an excess of free protein sites, very sensitive to competition by unlabeled antigen and is
then the rate of exit of ligand from the dialysis bag was no also very specific for that antigen. Thus, concentrations of
longer simply its diffusion rate nor was it simply the rate antigen in unknown samples can be determined by their
of dissociation of protein–ligand complex. These authors ability to compete with tracer for binding to antibody. The
showed that under these conditions the exit of ligand fol- method can be used to measure very low concentrations of a
lowed quasi–first-order kinetics, but with a half-life longer molecule, even in the presence of the many impurities in bio-
than the half-life in the absence of protein by a factor of logic fluids. Accomplishment of this requires an appropriate
(1 + [P]K A) high-affinity antibody and radiolabeled antigen, a method to
t + = t − (1 + [P]K A) (37) distinguish bound from free-labeled antigen, optimization
of concentrations of antibody and tracer-labeled antigen to
where [P] = protein site concentration, K A = affinity, and t + maximize sensitivity, and generation of a standard curve,
and t − = half-lives in the presence and absence of protein in using known concentrations of competing unlabeled antigen,
the bag. from which to read off the concentrations in unknown sam-
In this case, the protein was in solution, so the authors ples as well as the best method for representing the data. We
could use the actual protein concentration and the ac- review all these steps and pitfalls in this procedure except the
tual intrinsic affi nity, K A. In the case of protein on a two- preparation of antibodies and labeled antigens.
dimensional surface, it is harder to know what to use as the
effective concentration. However, the high local concentra-
tion of protein compartmentalized in the dialysis bag can Separation of Bound and Free Antigen
be seen to be analogous to the high local concentration at- Whatever parameter one uses to assess the amount of com-
tached to the solid surface. The underlying mechanism of petition by the unlabeled antigen in the unknown sample to
the two effects is essentially the same and so are the im- be tested, it will always be a function of bound versus free,
plications. For instance, in the case of dialysis, a modest radiolabeled antigen. Therefore, one of the most critical tech-
10 μM concentration of antibody sites with an affinity of nical requirements is the ability to distinguish clearly between
108 M−1 can reduce the rate of exit of a ligand 1000-fold. antibody-bound radioactive tracer and free radioactive tracer.
A dialysis that would otherwise take 3 hours would take This distinction usually requires physical separation of bound
4 months. It is easy to see how this “retention effect” can and free ligand. If the bound fraction is contaminated by free
make even modest affi nities appear infi nite (ie, the reac- ligand, or vice versa, enormous errors can result, depending
tions appear irreversible). This retention effect applies not on the part of the binding curve on which the data fall.
only to immunologic systems but also to other interactions
at a cell surface or between cell compartments where the Solution Methods
local concentration of a protein may be high. In particular, Solution RIA methods have the advantage that binding can be
these principles of two-phase systems should also govern related to the intrinsic affinity of the antibody. However, bound
the interaction between antigen specific receptors on the and free antigen must be separated by a method that does not
surface of T cells and antigen–major histocompatibility perturb the equilibrium. Three basic types of approaches have
complex (MHC) molecule complexes on the surface of an- been used: precipitate the antibody with bound antigen, leav-
tigen-presenting cells, B cells, or target cells. ing free antigen in solution; precipitate the free antigen, leaving
One final point is useful to note. Because these retention antibody and bound antigen in solution; or separate free from
effects depend on a localized abundance of unoccupied sites, antibody-bound antigen molecules in solution on the basis of
addition of a large excess of unlabeled ligand to saturate size by gel filtration. This last method is too cumbersome to
these sites will diminish or abolish the retention effect and use for large numbers of samples and is too slow, in general,
greatly accelerate the dissociation or exit of labeled ligand. to be sure the equilibrium is not perturbed in the process.
This effect of unlabeled ligand can be used as a test for the Therefore, gel filtration columns are not widely used for RIA.
retention effect, although one must be aware that in certain Methods that precipitate antibody are perhaps the most
cases the same result can be an indication of negative coop- widely used. If the antigen is sufficiently smaller (<30,000
erativity among receptor sites. MW) than the antibody that it will remain in solution at

concentrations of either ammonium sulfate41 or polyethyl- antibody coating the surface is unknown and because the
ene glycol, 6000 MW (10% W:W),42 which will precipitate affinity is not the intrinsic affinity, one cannot use these
essentially all the antibody, then these two reagents are methods for studying the chemistry of the antigen–antibody
frequently the most useful. Precipitation with polyethyl- reaction itself. A detailed analysis of the optimum param-
ene glycol and centrifugation can be accomplished before eters in this method is given by Zollinger et al.43
any significant dissociation has occurred due to dilutional A variation that does allow determination of affinity,
effects.43 However, if the antigen is much larger than about based on the enzyme-linked immunosorbant assay (ELISA)
30,000 to 40,000 MW, these methods will produce unac- described in the following, but equally applicable to RIA,
ceptably high background control values in the absence of was described by Friguet et al.45 This uses antigen-coated
specific antibody. If the antibody is primarily of a subclass microtiter wells and free antibody but measures competi-
of IgG that binds to staphylococcal protein A or G, one can tion by free antigen to prevent the antibody in solution from
take advantage of the high affinity of protein A or G for IgG binding to the antibody on the plate (see Fig. 7.9B). Thus,
by using either protein A (or G)-Sepharose or formalin- the antibody bound to the plastic is antibody that was free
killed staphylococcal organisms (Cowan I strain) to precipi- in the solution equilibrium. The affi nity measured is that
tate the antibody.44 Finally, one can precipitate the antibody between the antibody and antigen in solution, not that on
using a specific second antibody, an anti-Ig raised in another the plastic, so it is not directly influenced by the multiva-
species. Maximal precipitation occurs not at antibody excess lency of the surface. However, as pointed out by Stevens,46
but at the “point of equivalence” in the middle of the titra- the determination of affinity is strictly accurate only for
tion curve where antigen (in this case, the first antibody) monovalent Fab fragments because a bivalent antibody
and the (second) antibody are approximately equal in con- with only one arm bound to the plastic and one bound by
centration. Thus, one must add carrier Ig to keep the Ig con- antigen in solution will still be counted as free. Therefore,
centration constant and determine the point of equivalence there will be an underestimate of the ligand occupancy of
by titrating with the second antibody. Even worse, the pre- the antibody combining sites and thus an underestimate of
cipitin reaction is much slower than the antigen–antibody affinity. Stevens also points out a method to correct for this
reaction itself, allowing reequilibration of the antigen–anti- error based on binomial analysis. Subsequently, Seligman47
body interaction after dilution by the second antibody. Some showed that the nature and density of the antigen on the
of these problems can be reduced by enhancing precipitation solid surface can also influence the estimate of affinity.
with low concentrations of polyethylene glycol.
The other type of separation method is adsorption of
free antigen to an agent, such as activated charcoal or talc, Optimization of Antibody and Tracer Concentrations
which leaves antigen bound to antibody in solution. Binding for Sensitivity
of antigen by these agents depends on size and hydropho- The primary limitations on the sensitivity of the assay are
bicity. Although these methods are inexpensive and rapid, the antibody affi nity and concentration, the tracer concen-
they require careful adjustment and monitoring of pH, ionic tration, and the precision (reproducibility) of the data. In
strength, and temperature to obtain reproducible results general, the higher the affinity of the antibody, the more
and to avoid adsorption of the antigen–antibody complex. sensitive the assay can be made. Once one prepares the
Furthermore, because these agents have a high affinity for highest affinity antibody available, this parameter limits the
antigen, they can compete with a low-affinity antibody and extent to which the other parameters can be manipulated.
alter the equilibrium. Also, as charcoal quenches beta scin- For instance, because the unlabeled antigen in the unknown
tillation counting, it can be used only with gamma-emitting sample is going to compete against labeled tracer antigen,
isotopes such as 125I. the lower the tracer concentration, the lower the concentra-
tion of the unknown sample, which can be measured up to
Solid-Phase Methods a point. That point is determined by the affinity, K A, as can
Solid-phase RIA methods have the advantages of high be seen from the theoretical considerations discussed previ-
throughput and increased apparent affinity due to the effects ously.38 The steepest part of the titration curve will occur
at the solid–liquid interface noted previously. However, they in the range of concentrations around 1/K A. Concentrations
have the concomitant disadvantage that one is not measur- of ligand much below 1/K A will leave most of the antibody
ing the true intrinsic affi nity because of these same effects. sites unoccupied so that competition will be less effective.
The method itself is fairly simple. One binds the antibody Thus, there is no value in reducing the tracer concentration
in advance to a solid surface such as a Sepharose bead or more than a few-fold lower than 1/K A. Therefore, although
the walls of a microtiter plate well. To avoid competition it is generally useful to increase the specific radioactivity of
from other serum proteins for the solid phase, one must use the tracer and reduce its concentration, it is important to be
purified antibody in this coating step. Once the wells (or aware of this limit of 1/K A. Increasing the specific activity
Sepharose beads) are coated, one can incubate them with more than necessary can result in denaturation of antigen.
labeled tracer antigen with or without unlabeled competitor, Similarly, lowering the antibody concentration will also
wash and count directly the radioactivity bound to the plas- increase sensitivity, up to a point. This limit also depends
tic wells or to the Sepharose. The microtiter plate method is on 1/K A and on the background “nonspecific binding.”
particularly useful for processing large numbers of samples. Decreasing the antibody concentration to the point that
However, because the concentration, or even the amount, of binding of tracer is too close to background will result in

loss of sensitivity due to loss of precision. In general, the

fraction of tracer bound in the absence of competitor should
be kept greater than 0.2, and in general closer to 0.5.48
A convenient procedure to follow to optimize tracer and
antibody concentrations is first to choose the lowest tracer
concentration that results in convenient counting times and
counting precision for bound values of only one-half to one-
tenth the total tracer. Then, keeping this tracer concentration
constant, one dilutes out the antibody until the bound/free
antigen ratio is close to 1.0 (bound/total = 0.5) in the absence
of competitor. This antibody concentration in conjunction
with this tracer concentration will generally give near-optimal
sensitivities, within the limits noted previously. It is important FIG. 7.7. Schematic Plot of B/F or B/T (the Bound Over Free or Total
to be aware that changing the tracer concentration will require Antigen Concentration) as a Function of Free (F) or Total (T) Antigen
readjusting the antibody concentration to optimize sensitivity. Concentration, when Plotted on a Linear Scale. Contrast with simi-
lar plot on a log scale in Figure 7.6.
Analysis of Data: Graphic and
Numerical Representation
antibody molecules simultaneously and independently of
We have already examined the Scatchard plot (bound/free
one another, then the more such determinants capable of
versus bound) and the plot of bound/free versus free or total
being recognized by the antibodies in use, the steeper will
antigen concentration as methods of determining affinity. The
be the slope.49 This effect of multideterminant binding
latter lends itself particularly to the type of competition curves
on steepness arises because, in RIA, an antigen molecule
that constitute an RIA. In fact, the independent variable must
is scored as bound whether it has one antibody molecule
always be antigen concentration, as that is the known quantity
attached or several. It is scored as free only if no antibody
one varies to generate the standard curve. Let us use B, F, and
molecules are attached. Thus, the probability that an antigen
T to represent the concentrations of bound, free, and total an-
molecule is scored as free is the product of the probabilities
tigen, respectively. We have seen that the plot of B/F versus F
that each of its determinants is free. The effect can lead to
is more useful for determining the affinity, KA, than the plot
quite steep slopes and has been confirmed experimentally.49
of B/F versus T. However, in RIA, the quantity one wants to
A transform that allows linearization of the data in
determine is T, and correspondingly, the known independent
most cases is the logit transform.50,51 To use this, one first
variable in generating the standard curve is T. Another differ-
expresses the data as B/B0, where B0 is the concentration of
ence between the situation in RIA and that discussed previously
bound tracer in the absence of competitor. One then takes
is that, in RIA, one has both labeled and unlabeled antigen. The
the logit transform of this ratio, defi ned as
dependent variable, such as B/F, is the ratio of bound tracer
over free tracer, as only radioactive antigen is counted. The B/F Y
for the unlabeled antigen will be the same at equilibrium, as- [
logit (Y) = ln ______
(1 − Y) ] (40)
suming that labeled and unlabeled antigen bind the antibody
where ln = the natural log (log to the base e). The plot of
equivalently, that is, with the same KA. This assumption is not
logit (B/B0) versus ln T is usually a straight line (Fig. 7.8).
always valid and requires experimental testing.
The sigmoidal shape of B/F versus F or T, when F or T (the
“dose”) is plotted on a log scale, has been seen in Figure 7.6.
The shape for B/T versus F or T would be similar. Note that
because B + F = T,
B = _______
__ B/T
B = ________ (38)
F (T − B) (1 − B/T)
and
__ B/F
B = ________ (39)
T (1 + B/F)
These transformations can be useful. If one plots B/F or
B/T versus F or T on a linear scale, then the shape is approxi-
mately hyperbolic, as in Figure 7.7. The plot of B/F versus T
(log scale) was one of the first methods used to plot RIA data
FIG. 7.8. Schematic Logit. Log plot used to linearize radioimmunoas-
and is still among the most useful. The most sensitive part of say data. B and T are bound tracer and total antigen concentration,
the curve is the part with the steepest slope. respectively, and B0 is the value of B when no unlabeled antigen is
It has been shown by probability analysis that if the added to tracer. The logit function is defined by Equation 40, logit
antigen has multiple determinants, each capable of binding (Y) = In [Y / (1 − Y)].

The slope is usually −1 for the simplest case of a monoclo- Nonequilibrium Radioimmunoassay
nal antibody binding a monovalent antigen. The linearity So far, we have assumed that tracer and unlabeled competi-
of this plot obviously makes it very useful for graphical tor are added simultaneously, and sufficient incubation time
interpolation, which one would like to do to read antigen is allowed to achieve equilibrium. To measure the affinity, of
concentration off a standard curve. One additional advan- course, equilibrium must be assured. However, suppose one’s
tage is that linearity facilitates tests of parallelism. If the sole purpose is to measure the concentration of competitor by
unknown understudy is identical to the antigen used to RIA. Then one can actually increase the sensitivity of the assay
generate the standard curve, then a dilution curve of the by adding the competitor first, allowing it to react with the an-
unknown should be parallel to the standard curve in this tibody, and then intentionally adding the tracer for too short
logit–log coordinate system. If not, the assay is not valid. a time to reach a new equilibrium. One is essentially giving
These and other methods of analyzing the data are discussed the competitor a competitive advantage. It can be shown that
further by Feldman and Rodbard52 and Rodbard,39 including the slope of the dose-response curve, B/T versus total antigen
statistical treatment of data. Although a number of computer added, is increased in the low-dose range—a mathematical
programs have become available for rapid analysis of RIA data measure of increased sensitivity. A detailed mathematical
without using manual plots of standard curves, they are all analysis of this procedure may be found in Rodbard et al.53
based on these and similar methods, and their accurate inter- Note, however, that use of such nonequilibrium conditions
pretation depends on an understanding of these concepts. requires very careful control of time and temperature.
Corrections for B, F, and T
Before we leave this section on analysis of RIA data, we must Enzyme-Linked Immunosorbent Assay
point out a few controls and corrections to the data without An alternative solid-phase readout system for the detection
which the results may be fallacious. of antigen–antibody reactions is the ELISA.54 In principle,
First, in any method that precipitates antibody and bound the only difference from RIAs is that antibodies or antigen
antigen (or uses a solid-phase antibody), there may always be are covalently coupled to an enzyme instead of a radioisotope
a fraction of antigen that precipitates or binds nonspecifically so that bound enzyme activity is measured instead of bound
in the absence of specific antibody. Thus, one must always cpm. In practice, the safety and convenience of nonradioac-
run controls with normal serum or Ig to determine this back- tive materials and the commercial availability of plate readers
ground. The nonspecific binding usually increases linearly that can measure the absorbance of 96 wells in a few seconds
with antigen dose, that is, it does not saturate. This control account for ELISA’s widespread use. Because both ELISA and
value should be subtracted from B but does not affect F when RIA are governed by the same thermodynamic constraints,
measured independently, only F determined as T minus B. The and the enzyme can be detected in the same concentration
total antigen that is meaningful is the sum of that which is spe- range as commonly used radioisotopes, the sensitivity and
cifically bound and that which is free. Nonspecifically bound specificity are comparable. We consider three basic strategies
antigen should be deleted from any term in which it appears. for using ELISA assays to detect specific antibody or antigen.
A second correction is that for immunologically inac- As shown in Figure 7.9A, the indirect antibody method is
tive radiolabel, that is, either free radioisotope, or isotope the simplest way to detect and measure specific antibody in
coupled to an impurity or to denatured antigen. The frac- an unknown antiserum. Antigen is noncovalently attached
tion of radioactive material that is immunologically reac- to each well of a plastic microtiter dish. For this purpose, it is
tive with the antibodies in the assay can be determined by fortunate that most proteins bind nonspecifically to plastic.
using a constant, low concentration of labeled antigen and Excess free antigen is washed off, and the wells are incubated
adding increasing concentrations of antibody. If there is no with an albumin solution to block the remaining nonspe-
contamination with inactive material, all the radioactivity cific protein binding sites. The test antiserum is then added,
should be able to be bound by sufficient antibody. If the frac- and any specific antibody binds to the solid-phase antigen.
tion of tracer bound reaches a plateau at <100% bound, then Washing removes unbound antibodies. Enzyme-labeled anti-
only this fraction is active in the assay. The importance of Ig is added. This binds to specific antibody already bound
this correction can be seen from the example in which the to antigen on the solid phase, bringing along covalently at-
tracer is only 80% active. Then, when the true B/F is 3 (B/T tached enzyme. Unbound antiglobulin-enzyme conjugate
= 0.75), applying only to the active 80% of the tracer, the re- is washed off; substrate is then added. The action of bound
maining 20%, which can never be bound, will mistakenly enzyme on substrate produces a colored product, which is
be included in the free tracer, doubling the amount that is detected as increased absorbance in a spectrophotometer.
measured as free. Thus, the measured B/F will be only 1.5 (ie, Although this method is quick and very sensitive, it is
0.6/0.4) instead of the true value of 3 (ie, 0.6/0.2). This factor often difficult to quantitate. Within a defined range, the
of 2 will make a serious difference in the calculation of affi n- increase in optical density is proportional to the amount
ity, for instance, from a Scatchard plot. Also, it will result in of specific antibody added in the first step. However, the
a plateau in the Scatchard plot at high values of B/F, as with amount of antibody bound is not measured directly.
20% of the tracer obligatorily free, B/F can never exceed 4 (ie, Instead, the antibody concentration of the sample is esti-
0.8/0.2). To correct for this potentially serious problem, the mated by comparing it with a standard curve for a known
inactive fraction must always be determined and subtracted amount of antibody. It is also difficult to determine affin-
from both F and T. ity by this method because the solid-phase antigen tends to

FIG. 7.9. Three Strategies for the Detection of Specific Antibody–Antigen Reactions Using the Enzyme-Linked Immunosorbant Assay
Technique. A: Direct binding. B: Hapten inhibition. C: Antigen sandwich.
increase the apparent affinity. The sensitivity of this assay The absorbance produced is a function of the antigen con-
for detecting minute amounts of antibody is quite good, centration of the test solution, which can be determined
especially when affinity-purified antiglobulins are used as from a standard curve. Specificity of the assay depends on
the enzyme-linked reagent. A single preparation of enzyme- the specificity of the antibodies used to coat the plate and
linked antiglobulin can be used to detect antibodies to many detect antigen. Sensitivity depends on the affinity as well as
different antigens. Alternatively, class-specific antiglobu- amount of the first antibody coating the well, which can be
lins can be used to detect how much of a specific antibody increased by using affinity-purified antibodies or monoclo-
response is due to each Ig class. Obviously, reproducibility of nal antibodies in this step. The sandwich method depends
the assay depends on uniform antigen coating of each well, on divalency of the antigen or else the two antibodies must
and the specificity depends on using purified antigen to coat be specific for different antigenic determinants on the same
the wells. antigen molecule. When comparing two monoclonal anti-
Figure 7.9B shows the competition technique for detect- bodies to the same antigen, this technique can be used to
ing antigen. Soluble antigen is mixed with limiting amounts ascertain whether they can bind simultaneously to the same
of specific antibody in the first step. Then the mixture is molecule or whether they compete for the same site or sites
added to antigen-coated wells and treated as described in close enough to cause steric hindrance.55
Figure 7.9A. Any antigen–antibody complexes formed in When antibodies are serially diluted across a plate, the
the first step will reduce the amount of antibody bound to last colored well indicates the titer. Specificity of binding can
the plate, and hence will reduce the absorbance measured be demonstrated by coating wells with albumin and mea-
in the fi nal step. This method permits the estimate of affi n- suring antibody binding in parallel with the antigen-coated
ity for free antigen, which is related to the half-inhibitory wells. Because it can be used to test many samples in a short
concentration of antigen. Mathematical analysis of affi nity time, ELISA is often used to screen culture supernatants in
by this approach was described by Friguet et al.45 with mod- the production of hybridoma antibodies. The sensitivity of
ification by Stevens,46 as discussed under RIA solid-phase the method allows detection of clones producing specific
methods. In addition, some estimate of cross-reactivity be- antibodies at an early stage in cell growth.
tween the antigen in solution and that on the plate can be An important caution when using native protein
obtained. antigens to coat solid-phase surfaces (see Fig. 7.9A) is that
Figure 7.9C shows the sandwich technique for detect- binding to a surface can alter the conformation of the pro-
ing antigen. Microtiter plates are coated with specific anti- tein. For instance, using conformation specific monoclonal
body. Antigen is then captured by the solid-phase antibody. antibodies to myoglobin, Darst et al.56 found that binding
A second antibody specific for the antigen, and coupled to of myoglobin to a surface altered the apparent affinity of
enzyme, is added. This binds to the solid-phase antigen– some antibodies more than others. This problem may be
antibody complex, carrying enzyme along with it. Excess avoided by using the solution phase methods of Figure 7.9B
second antibody is washed off, and substrate is added. or 7.9C.

ELIspot Assay Karush30 has defined a related term, selectivity, as the ability
The normal ELISA assay can be modified to measure anti- of an antibody to discriminate, in an all-or-none fashion,
body production at the single cell level. In this method, tis- between two related ligands. Thus, selectivity depends not
sue culture plates are coated with antigen, and various cell only on the relative affinity of the antibody for the two li-
populations are cultured on the plate for 4 hours. During gands but also on the experimental lower limit for detection
that time, B cells settle to the bottom and secrete antibod- of reactivity. For instance, an anticarbohydrate antibody
ies, which bind antigen nearby and produce a footprint of with an affinity of 105 M−1 for the immunogen may appear
the antibody-secreting cell. The cells are then washed off, to be highly selective, as reaction with a related carbohydrate
and a second antibody, such as enzyme-labeled goat antihu- with a 100-fold lower affinity, 103 M−1, may be undetectable.
man IgG, is added. Finally, unbound antibody is washed off, On the other hand, an antibody with an affi nity of 109 M−1
and enzyme substrate is added in soft agar. Over the next for the homologous ligand may appear to be less selective
10 minutes, each footprint of enzyme activity converts the because any reaction with a related ligand with a 100-fold
substrate to a dark spot of insoluble dye, corresponding to lower affinity would still be quite easily detectable.
the localized zone where the B cell originally secreted its Conversely, cross-reactivity is defined as the ability to
antibody. react with related ligands other than the immunogen. More
Using this method, it is possible to detect as few as 10 to usually, this is examined from the point of view of the
20 antibody-producing B cells in the presence of 106 spleen ligand. Thus, one might say that antigen Y cross-reacts with
cells, and typical results for immunized mice range from antigen X because it binds to anti-X antibodies. Note that in
200 to 500 spot-forming cells per 106 spleen cells.57,58 Clearly, this sense, it is the two antigens that are cross-reactive, not
to work at all, this assay must be capable of detecting the the antibody. However, the cross-reactivity of two antigens,
amount of antibody secreted by a single immune B cell and X and Y, can be defined only with respect to a particular an-
specific enough to exclude nonspecific antibodies produced tibody or antiserum. For instance, a different group of anti-
by most nonimmune B cells. Sensitivity depends on the af- X antibodies may not react at all with Y so that with respect
finity and amount of antibodies secreted and may be opti- to these antibodies, Y would not be cross-reactive with X.
mized by titering the amount of antigen on the plate. One can also use the term in a different sense, saying that
This type of assay is useful in analyzing the cellular re- some anti-X antibodies cross-react with antigen Y.
quirements for antibody production in vitro, as the number In most cases, cross-reactive ligands have lower affinity
of responding B cells is measured directly. It can also be than the immunogen for a particular antibody. However,
used to detect antibodies made in the presence of excess an- exceptions can occur in which a cross-reactive antigen
tigen. For example, during acute infections59 and in auto- binds with a higher affinity than the homologous antigen
immunity,60 when antigen may be in excess over antibody, itself. This phenomenon is called heterocliticity, and the
this assay makes it possible to measure antibody-producing antigen that has a higher affinity for the antibody than
B cells, even though free antibody may not be detectable in does the immunogen is said to be heteroclitic. Antibodies
circulation. It can also be used to measure local produc- that manifest this behavior are also described as hetero-
tion of self-reactive antibodies in a specific tissue, such as clitic antibodies. A good example is the case of antibodies
synovium. By using two detecting antibodies, each specific raised in C57BL/10 mice against the hapten nitrophenyl
for a different Ig class and coupled to a different enzyme, acetyl. These antibodies have been shown by Mäkelä and
and two substrates producing different colored dyes, cells Karjalainen64 to bind with higher affinity to the cross-reac-
secreting IgA and IgG simultaneously can be detected.61 tive hapten, nitroiodophenyl acetyl, than to the immuno-
Recently, ELIspot was used to show that bacterial deoxyri- gen itself. Another example is the case of retro-inverso or
bonucleic acid (DNA)-containing CpG sequences is a poly- retro-D peptides.65–69 By reversing the chirality from L to
clonal B-cell mitogen.62 D amino acids, and simultaneously reversing the sequence
ELIspot can also detect secreted cytokines, as opposed to of amino acids, one can produce a peptide that is resistant
antibodies, by coating the plate with a capture antibody and to proteolysis and has its side chains approximately in the
detecting antigen with an enzyme-coupled second antibody same position as the original L amino acid peptide, with the
(as in a sandwich ELISA; see Fig. 7.9C). For example, using exception of some amino acids with secondary chiral cen-
plates coated with monoclonal antibody to interleukin (IL)-4, ters such as Thr and Ile. However, the backbone NH2 and
T cells secreting IL-4 could be detected,63 providing one mea- COOH moieties are reversed. Antibodies that interact with
sure of T helper 2 cells. only the side chains might not distinguish these peptides,
whereas antibodies that interact with the main chain as well
as side chains might distinguish them and have potentially
SPECIFICITY AND CROSS-REACTIVITY higher or lower affinity. In a study of monoclonal antibodies
The specificity of an antibody or antiserum is defi ned by its to a hexapeptide from histone H3, some bound the retro-D
ability to discriminate between the antigen against which it form with higher affinity than the native sequence and some
was made (called the homologous antigen or immunogen) did not.67,68 The former are examples of heterocliticity. In ad-
and any other antigen one might test. In practice, one cannot dition to greater binding affinity, the retro-D peptides may
test the whole universe of antigens but only selected anti- have even greater activity in vivo because of their resistance
gens. In this sense, specificity can only be defined experi- to proteolysis.65–69 This stability makes them more useful as
mentally within that set of antigens one chooses to compare. drugs as well.65,66,70

FIG. 7.10. Schematic Radioimmunoassay Binding Curves for Homo-

logous Ligand L and Cross-Reacting Ligands. Cross-reacting ligand B
CA manifests type 1 or true cross-reactivity demonstrated by com-
plete inhibition of tracer ligand binding and a lower affinity. Ligand CB
displays type 2 cross-reactivity or determinant sharing, as recognized
from the plateau at less than 100% inhibition, but not necessarily a
lower affinity. The ordinate R is the ratio of bound/free radiolabeled
tracer ligand, and R0 is the limit of R as the concentration of all ligands,
including tracer, approaches zero. From Berzofsky and Schechter,71
with permission.
Cross-reactivity has often been detected by methods such

as the Ouchterlony test, or hemagglutination (see the fol-
lowing for descriptions of both of these) or similar methods,
which have in common the fact that they do not distinguish C
well between differences in affinity and differences in con-
centration. This practical aspect, coupled with the hetero-
geneity of immune antisera, has led to ambiguities in the
usage of the terms “cross-reactivity” and “specificity.” With
the advent of RIA and ELISA techniques, this ambiguity in
the terminology, as well as in the interpretation of data, has
become apparent.
For these reasons, Berzofsky and Schechter71 have defined
two forms of cross-reactivity and, correspondingly, two
forms of specificity. These two forms of cross-reactivity are
illustrated by the two prototype competition RIA curves in FIG. 7.11. An Artist’s Drawing of the Amino Terminal Region of the
Figure 7.10. In reality, most antisera display both phenom- a Chain of Hemoglobin. A: The first 11 residues of the βA chain.
ena simultaneously. B: The comparable regions of the βS chain. The substitution of valine
Type 1 cross-reactivity, or true cross-reactivity, is de- for the normal glutamic acid at position 6 makes a distinct antigenic
fi ned as the ability of two ligands to react with the same site determinant to which a subpopulation of antibodies may be iso-
lated.72,73 C: A schematic diagram of the sequence in (A) unfolded as
on the same antibody molecule, possibly with different af- occurs when the protein is denatured. This region may be cleaved
fi nities. For example, the related haptens dinitrophenyl and from the protein, or the peptide synthesized,74 resulting in changed
trinitrophenyl may react with different affi nity for antibod- antigenic reactivity. An antiserum prepared to hemoglobin (or the
ies raised to dinitrophenyl hapten. In protein antigens, such β chain thereof) might exhibit cross-reactivity with the structures
differences could occur with small changes in primary se- shown in (B) and (C), but the molecular mechanisms would be dif-
quence (eg, the conservative substitution of threonine for ferent. Polypeptide backbone atoms are in white in the side chains,
oxygen atoms are hatched, nitrogen atoms are black, and carbon
serine), or with changes in conformation, such as the cleav-
atoms are lightly stippled. Adapted from Berxofsky and Schechter71
age of the protein into fragments (Fig. 7.11).71–75 If a peptide and Dean and Schecter.75
fragment contained all the contact residues in an antigenic
determinant (ie, those that contact the antibody-combining
site), it might cross-react with the native determinant for CA, but higher concentrations of CA than of the homolo-
antibodies against the native form but with lower affi nity gous ligand, L, are required to produce any given degree of
because the peptide would not retain the native conforma- inhibition.
tion. This type of affi nity difference is illustrated by com- A separate issue from affinity differences is the issue of
petitor CA in Figure 7.10, in which complete displacement whether the cross-reactive ligand reacts with all or only a
of tracer can be achieved at high enough concentrations of subpopulation of the antibodies in a heterogeneous serum.

FIG. 7.12. Schematic Radioimmunoassay Binding Curves Showing the Effect of Affinity on
the Midpoint and the Slope at the Midpoint and the Value of Using Free (Ligand) Rather
than Total (Ligand). Ordinate R is the ratio of bound/free radiolabeled tracer ligand, and R0
the limit of R as all ligand concentrations approach zero. If x and y are the concentrations of
ligands X and Y that reduce R to exactly R0/2, then if the abscissa is total ligand concentra-
tion, x = 1/KX + [S]t/2 and y = 1/KY + [S]t/2, where [S]t is the concentration of antibody binding
sites and KX and KY the affinities of the antibody for the respective ligands. However, if the
abscissa is free ligand concentration, x = 1/KX and y = 1/KY so that the ratio x/y (or the dif-
ference log x − log y on a log plot) corresponds to the ratio of affinities KY/KX. Note that the
slopes at the midpoints are the same on a log scale, but that for Y would be only KY/KX that
for X on a linear scale. From Berxofsky and Schechter,71 with permission.
This second type of cross-reactivity, which we call type 2 competition curve would be shifted to the right and would
cross-reactivity or shared reactivity, therefore can occur plateau before reaching complete inhibition.†††
only when the antibody population is heterogeneous, as In the case of a homogeneous (eg, monoclonal) antibody
in most conventional antisera. In this case, the affinity of in which only type 1 or true cross-reactivity can occur, one
the cross-reactive ligand may be greater than, less than, or can quantitate the differences in affi nity for different cross-
equal to that of the homologous ligand for those antibod- reactive ligands by a method analogous to the B/F versus F
ies with which it interacts. Therefore, the competition curve method described previously. Suppose that ligands X and
is not necessarily displaced to the right, but the inhibition Y cross-react with homologous ligand L for a monoclonal
will reach a plateau at less than complete inhibition, as il- antibody. If one plots the bound/free (B/F = R) ratio for
lustrated by competitor CB in Figure 7.10. As an example, let radiolabeled tracer ligand L as a function of the log of the
us consider the case of a protein with determinants X and Y, concentration of competitors X and Y, one obtains two par-
and an antiserum against this protein containing both anti- allel competition curves (Fig. 7.12),71 under the appropriate
X and anti-Y antibodies. Then a mutant protein in which de- conditions (see subsequent discussion). The first condition
terminant Y was so altered as to be unrecognizable by anti-Y, is that the concentration of free tracer be less than 1/KL, the
but determinant X was intact, would manifest type 2 cross- affinity for tracer. In this case, it can be shown71 that
reactivity. It would compete with the wild-type protein only
for anti-X antibodies (possibly even with equal affi nity), but 1
K X ≈ _____ (41)
[X]free
not for anti-Y antibodies.
Of course, both types of cross-reactivity could occur at the midpoint where R = R0 /2, where K X = affinity for
simultaneously. A classic example would be the pep- X. This is analogous to Equation 21 for the case in which
tide fragment previously discussed in the case of type I unlabeled homologous ligand is the competitor. Also, in
cross-reactivity. Suppose the fragment contained the resi-
dues of determinant X, albeit not in the native conforma-
tion, but did not contain the residues of a second determi- †††
An ambiguous case could occur experimentally in which the distinction between
nant, Y, which was also expressed on the native protein. If the two types of cross-reactivity would be blurred. For example, in the case of anti-
the antiserum to the native protein consisted of anti-X and bodies that all react with determinant X but have a very wide range of affi nities for
anti-Y, the peptide would compete only for anti-X antibod- X, some such antibodies may have such a low affi nity for cross-reactive determinant
X⬘ that they would appear not to bind X⬘ at all. Then a competition curve using X⬘
ies (type 2 cross-reactivity) but would have a lower affinity might appear to reach a plateau at incomplete inhibition, even though all the anti-
than the native protein even for these antibodies. Thus, the bodies were specific for X, and the only difference between X and X⬘ was affi nity.

analogy with Equation 23, it can be shown that if the total be relatively specific for the immunogen. Note that it does
concentration of competitor, [X]t, is used instead of the free not matter whether the affinity of a subpopulation that
concentration, [X]free, an error term will arise, giving reacts with a cross-reactive antigen is high or low (type 1
cross-reactivity). As long as that subpopulation is a small
[S]t
1 + ___
[X]t (at R = R0 /2) = ___ (42) fraction of the antibodies, the mixture is specific. Thus,
KX 2
type 2 specificity depends on the relative concentrations
Thus, with competitor on a linear scale, the difference in of antibodies in the heterogeneous antiserum, not just on
midpoint for competitors X and Y will correspond to the dif- their affinities. Also note that one can use these relative
ference 1/K X − 1/K Y regardless of whether free or total com- concentrations of antibody subpopulations to compare the
petitor is plotted, but the ratio of midpoint concentrations specificity of a single antiserum for two cross-reactive li-
will equal K X /K Y only if the free concentrations are used. gands. However, it would not be meaningful to compare the
This last point is important if one plots the log of competitor specificity of two different antisera for the same ligand by
concentration, as is usually done, as the horizontal displace- comparing the fraction of antibodies in each serum which
ment between the two curves on a log scale corresponds to reacted with that ligand. Although type 2 specificity may ap-
the ratio [X] / [Y], not the difference.71 pear to some a less classic concept of specificity than type 1,
If a second condition also holds, namely, that the concen- it is type 2 specificity that one primarily measures in such
tration of bound tracer is small compared to the antibody assays as the Ouchterlony double immunodiffusion test,
site concentration [S]t, then the slopes (on a linear scale) of and it carries equal weight with type 1 specificity in such
the curves at their respective midpoints (where R = R0 /2) assays as hemagglutination, discussed in the following sec-
will be proportional to the affinity for that competitor, K X tion. Type 2 specificity also leads naturally to the concept of
or K Y (71). (Both conditions can be met by keeping tracer L “multispecificity,” described in the following section.
small relative to both K L and [S]t.) When [X]free and [Y]free
are plotted on a log scale, the slopes will appear to be equal
(ie, the curves will appear parallel) because a parallel line Multispecificity
shifted m-fold to the right on a log scale will actually be 1/m The theory of multispecificity, introduced and analyzed by
as steep, at any point, in terms of the antilog as abscissa. Talmadge77 and Inman,78,79 and discussed on a structural
When the antibodies are heterogeneous in affinity, the level by Richards et al.,80 suggests a mechanism by which
curves will be broadened and in general will not be parallel. the diversity and specificity of antisera can be expanded and
When heterogeneity of specificity is present, and type 2 also understood. The idea is that each antibody may actually
cross-reactivity occurs, it should be pointed out that the bind, with high affinity, a variety of diverse antigens. When
fractional inhibition achieved at the plateau in a B/F versus one immunizes with immunogen A, one selects for many
free competitor plot will not be proportional to the fraction distinct antibodies, which have in common only that they
of antibodies reacting with that competitor but will be pro- all react with A. In fact, each antibody may react with other
portional to a weighted fraction, where the antibody con- compounds, but if fewer than 1% of the antibodies bind B,
centrations are weighted by their affinity for the tracer.71 and fewer than 1% bind C, and so on, then by type 2 speci-
These two types of cross-reactivity lead naturally to two ficity, the whole antiserum will appear to be highly specific
definitions of specificity.71 The overall specificity of a het- for A. Note that the subpopulation that binds B may react
erogeneous antiserum is a composite of both of these facets with an affinity for B as high as or higher than that for A so
of specificity. Type 1 specificity is based on the relative af- that the population would not be type 1 specific for A. This
finities of the antibody for the homologous ligand and any same population would presumably be selected if one im-
cross-reactive ligands. If the affinity is much higher for the munized with B. The net result would be that the diversity of
homologous ligand than for any cross-reactive ligand tested, highly (type 2) specific antisera an organism could generate
then the antibody is said to be highly specific for the homol- would be even greater than the diversity of B-cell clones (or
ogous ligand (ie, it discriminates very well between this li- antibody structures). This principle can explain how poly-
gand and the others). If the affinity for cross-reactive ligands clonal antisera can sometimes appear paradoxically more
is below the threshold for detection in an experimental situ- specific than a monoclonal antibody.
ation, then type 1 specificity gives rise to selectivity as was
discussed previously.30 The specificity can even be quanti-
tated in terms of the ratio of affi nities for the homologous li- OTHER METHODS
gand and a cross-reactive ligand.76 It is this type 1 specificity We mention only a few of the other methods for meas-
that most immunochemists would call true specificity, just uring antigen–antibody interactions. Other useful tech-
as we have called type 1 cross-reactivity true cross-reactivity. niques include quenching of the tryptophan fluorescence
The common use of the term “cross-reactivity” to include of the antibody by certain antigens on binding81 (a sensitive
type 2 or partial reactivity leads to a second definition of method useful for such experiments as fast kinetic studies),
specificity, which applies only to heterogeneous populations antibody-dependent cellular cytotoxicity, immunofluores-
of antibodies such as antisera. We call this type 2 specific- cence including flow cytometry, immunohistochemistry,
ity. If all the antibodies in the mixture react with the im- and inhibition by antibody of plaque formation by antigen-
munogen, but only a small proportion react with any single conjugated bacteriophage82 (a method as sensitive as RIA as
cross-reactive antigen, then the antiserum would be said to inhibition of even a few phage virions can be detected).

Quantitative Precipitin and is called the equivalence point. The amount of antibody
Among the earliest known properties of antibodies were protein in the precipitate at equivalence is considered to equal
their ability to neutralize pathogenic bacteria and their abil- the total amount of specific antibody in that volume of antise-
ity to form precipitates with bacterial culture supernatants. rum. The rising part of the curve is called the antibody excess
Both activities of each antiserum were highly specific for zone (antigen limiting), and the part of the curve beyond the
the bacterial strain against which the antiserum was made. equivalence point is called the antigen excess zone.
The precipitates contained antibody protein and bacterial Supernatants and precipitates were carefully analyzed
products. The supernatants contained decreased amounts for each zone of antibody or antigen excess, as shown in
of antibody protein and, under the right conditions, had lost Figure 7.13B. When antigen was limiting, the precipitate
the ability to neutralize bacteria. However, quantitation of contained high ratios of antibody to antigen. The superna-
the antibody precipitated was difficult because the precipi- tant in this zone contained free antibody with no detectable
tate contained antigen protein as well as antibody protein. antigen. As more antigen was added, the amount of antibody
Heidelberger and Kendall83,84 solved this problem when they in the precipitate rose, but the ratio of antibody to antigen
found that purified pneumococcal cell wall polysaccharide fell. At equivalence, no free antibody or antigen could be de-
could precipitate with antipneumococcal antibodies. In tected in the supernatant. As more antigen was added, the
this case, the amount of protein nitrogen measured in the precipitate contained less antibody, but the ratio of antibody
precipitate was entirely due to antibody protein, and the to antigen remained constant. The supernatant now con-
amount of reducing sugar was mostly due to the antigen. tained antigen–antibody complexes because the complexes
Plotting the amount of antibody protein precipitated from at antigen excess were small enough to remain in solution.
a constant volume of antiserum by increasing amounts of No unbound antibody was detected.
carbohydrate antigen gives the curve shown in Figure 7.13. The lattice theory83,84 is a model of the precipitation reac-
As shown in Figure 7.13A, the amount of antibody precipi- tion that explains these observations. It assumes that antibod-
tated rises initially, reaches a plateau, and then declines. The ies are multivalent and antigens are bivalent or polyvalent.
point of maximum precipitation was found to coincide with Thus, long chains can form consisting of alternating antibody
the point of complete depletion of neutralizing antibodies linked to antigen linked to antibody, and so on. The larger
the size of the aggregate, the less soluble the product. In the
antibody excess zone, branch points can form whenever three
antibodies bind to a single antigen, resulting in a three-di-
A mensional lattice structure, which precipitates. In the equiva-
lence zone, when equimolar amounts of antibody and antigen
are mixed, the likelihood of more than two antibodies bind-
ing the same antigen molecule decreases. With fewer branch
points, the product is more likely to consist of long chains of
alternating antibody and antigen molecules in the molar ratio
of 1:1. At even higher antigen ratios, each antigen molecule
will have 0 or 1 antibody bound, so shorter chain lengths are
found, until the product is small enough to remain in solu-
tion. Such soluble antigen–antibody “immune complexes”
are detectable in the antigen excess zone, where no free anti-
body is found.
B Besides explaining the observed precipitation phenomena
on a statistical basis, the lattice theory made the important
prediction that antibodies are bivalent or multivalent. The
subsequent structural characterization of antibodies re-
vealed their molecular weight and valency. Antibodies are
indeed bivalent, except for IgM, which is functionally pen-
tavalent and forms precipitates even more efficiently.
Antigens can be polyvalent either by having multiple cop-
ies of the same determinant or by having many different
determinants, each of which reacts with different antibod-
ies in a polyclonal antiserum. The predominant antigenic
determinants of polysaccharides are often the nonreduc-
ing end of the chain. Branched chain polysaccharides have
more than one end and are polyvalent. Nonbranched chains
FIG. 7.13. Quantitative Immunoprecipitation. To a fixed amount of such as dextran (polymer of glucose) are monovalent for
specific antibody are added increasing amounts of nonprotein an-
tigen. The figure shows the amount of antibody protein (A) and the
end specific antidextran antibodies and do not precipitate
ratio of antibody to antigen (B) found in the precipitate. At antigen them.85 However, a second group of antidextran antibodies
excess, soluble immune complexes are found in the supernatant, is specific for internal glucose moieties. Because each dextran
and the precipitate is decreased. polymer consists of many internal units, it is polyvalent for

internal α(1 Æ 6)-linked glucose-specific antibodies. Thus, at independent rates, depending on antigen concentration in
unbranched dextran polymer can be used to distinguish be- the sample, diffusion coefficient (size), and antibody con-
tween end-specific and internal-specific antibodies, as it will centration in the agar. Similarly, in the two-dimensional
precipitate with the latter antibodies but not the former.85,86 method, at a given radius of diffusion, antigen concentration
Monomeric protein antigens, such as myoglobin or lysozyme, will be equivalent to the antibody in the gel, and a precipitin
behave as if they were polyvalent for heterogeneous antisera ring will form. The higher the initial antigen concentration,
but monovalent for monoclonal antibodies. This results from the farther the antigen will diffuse before precipitating and
the fact that each antigen molecule has multiple antigenic de- the wider the area of the ring will be. The area of the ring
terminants but only one copy of each determinant. Thus, a is directly proportional to the initial antigen concentration.
polyspecific antiserum can bind more than one antibody to This method provides a convenient quantitative assay that
different determinants on the same molecule and form a lat- can be used to measure Ig classes by placing test serum in
tice. However, using antibodies directed against a single de- the well and antiserum to each class of human Ig in the agar.
terminant (such as a monoclonal antibody), no precipitate Sensitivity can be increased by lowering the concentration of
will form. In this case, antigen–antibody reactions must be antiserum in the gel, giving wider rings, as the antigen must
measured by some other binding assay, such as RIA or ELISA. reach a lower concentration to be at equivalence. However,
the antiserum cannot be diluted too far, or no precipitate
will form.
Immunodiffusion and the Ouchterlony Method The double diffusion methods use the same principles,
One of the most useful applications of immunoprecipitation except that instead of having one reactant incorporated
is in combination with a diffusion system.87 Diffusion could in the gel at a constant concentration, both antigen and
be observed by gently adding a drop of protein solution to antibody are loaded some distance apart in a gel of pure
a dish of water without disturbing the liquid. The rate of agarose alone and are allowed to diffuse toward each other.
migration of protein into the liquid is proportional to the At some point in the gel, antigen diffusion and antibody
concentration gradient times the diffusion coefficient of the diffusion will provide sufficient concentrations of both
protein according to Fick’s law, reactants for immunoprecipitation to occur. The line of
precipitation becomes a barrier for the further diffusion
dQ
___ dc
= −DA ___ (43) of the reactants, so the precipitin band is stable. If the an-
dt dx
tigen preparation is heterogeneous, and the antiserum is
where Q = amount of substance that diffuses across an area a heterogeneous mixture of antibodies, different bands
A per unit time t; D = diffusion coefficient, which depends will form for each pair of antigen and antibody reacting
on the size of the molecule; and dc/dx = concentration at positions dependent on concentration and molecular
gradient. Because antibody molecules are so large, their dif- weight of each. The number of lines indicates the number
fusion coefficients are quite low, and diffusion often takes of antigen–antibody systems reacting in the gel. The abil-
1 day or more to cover the 5 to 20 mm required in most ity of immunodiffusion to separate different antigen–anti-
systems. In order to stabilize the liquid phase for such long body systems gives a convenient estimate of antigen purity
periods of time, a gel matrix is added to provide support or antibody specificity.
without hindering protein migration. In practice, 0.3% to In the most widely used Ouchterlony method of double
1.5% agar or agarose will permit migration of proteins up diffusion in two dimensions, 87 three or more wells are
to the size of antibodies while preventing mechanical and cut in an agarose gel in a dish in the pattern shown in
thermal currents. By carefully adjusting the concentration Figure 7.14. Antigen a or b is placed in the upper wells,
of antibody and antigen, these systems can provide a simple whereas antiserum containing anti-a or anti-b is placed
analysis of the number of antigenic components and the in the lower well. Each antigen–antibody reaction sys-
concentration of a given component. By adjusting the geom- tem will form its own precipitin line between the wells.
etry of the reactants entering the gel, immunodiffusion can As shown in Figure 7.14A, this should extend an equal
provide useful information concerning antigenic identity or distance to the left and right of the wells. When different
difference, or partial cross-reaction, as well as the purity of antigens are present in different wells (see Fig. 7.14C), the
antigens and the specificity of antibodies. precipitating systems do not interact immunochemically,
In single diffusion methods,88–91 antibody is incorporated so the precipitin lines cross. However, when the same an-
in the gel, and antigen is allowed to diffuse from one end of tigen is present in both wells (see Fig. 7.14B), each line of
a tube gel in one dimension or from a hole in a gel in a petri precipitation becomes a barrier preventing the other anti-
dish in two dimensions. Over time, the antigen concentra- gen and antibody from diffusing past the precipitin line.
tion reaches equivalence with the antibody in the gel, and This shortens the precipitin line on that side of the well.
a precipitin band forms. As more antigen diffuses, antigen In addition, antigen diffusion from the neighboring wells
excess is achieved at this position, so the precipitate dissolves shifts the zone of antigen excess, causing the equivalence
and the boundary of equivalence moves farther. By integrat- line to deviate downward and meet between the two wells.
ing Fick’s law, we find that the distance moved is propor- Complete fusion of precipitin lines with no spurs is called
tional to the square root of time. If two species of antigen a a line of identity, indicating that the antigen in each well
and b are diffusing, and the antiserum contains antibodies reacts with all the antibody capable of reacting with anti-
to both, two independent bands will form. These will move gen in the other well.

A B
C D
FIG. 7.14. Immunodiffusion of Two Compo-

nents in Two Dimensions. Cross-reactions
produce inhibition (shortened bands) or de-
viation (curved bands). Lines of identity are
shown in (B) and (D).87 E F
The analytical power of this method is shown in gin of diffusion of different antigens. This is equivalent to
Figure 7.14D. When a mixed antigen sample is placed in one having each antigen start in a different well, as shown in
well and pure antigen b is placed in the other well, antise- the right-hand panel. A horizontal trough is then cut into
rum to a plus b gives the pattern shown. Two precipitin lines the agar and fi lled with antiserum to all the components.
form with the left well and one precipitin line with the right Immunodiffusion occurs between the separated antigens
well. The line of complete fusion allows us to identify the and the linear source of antibody. The results for a mixture
second band as antigen b; the first band is antigen a. From of three antigens approximate those shown for three anti-
their relative distance of migration, we can conclude that an- gens in separate wells.87 Fusion, deviation, and inhibition
tigen a is in excess over antigen b, assuming their diffusion between precipitin lines can be analyzed as described previ-
coefficients are comparable and both antibodies are present ously. The resolution of each band is somewhat decreased
in equal amounts. Finally, because the precipitin line of an- due to widening of the origin of diffusion during electro-
tigen a-anti-a is not shortened at all, there is no contamina- phoresis. However, the immunodiffusion of unseparated
tion of the right sample with antigen a, and the two antigens human serum proteins, for example, is greatly facilitated by
do not cross-react. prior electrophoresis. Starting from a single well, only the
The type of cross-reactivity detected by this Ouchterlony heavier bands would be visible. However, prior electropho-
double immunodiffusion in agar is what we have defi ned resis makes it possible for each electrophoretic species to
previously as type 2 cross-reactivity. The method is not suit- make its own precipitin line. Monospecific antiserum can
able for measuring affinity differences required for quanti- be placed in a parallel horizontal trough so that each band
tating type 1 cross-reactivity. Sensitivity can be increased by of precipitation can be identified. Immunoelectrophoresis
use of radioactive antigen and detection of the precipitate by is commonly used to diagnose myeloma proteins in human
autoradiography. serum. The unknown serum is electrophoresed, followed by
immunodiffusion against antibodies to human Ig heavy or
light chains. A widening arc of IgG suggests the presence of
Immunoelectrophoresis an abnormal Ig species. At this same electrophoretic mobil-
Some antigen–antibody systems are too complex for double ity, a precipitin line with anti-κ, but not anti-λ, reactivity
immunodiffusion analysis, either because there are too many strongly suggests the diagnosis of myeloma or monoclonal
bands or they are too close together. Immunoelectrophoresis gammopathy, as these proteins arise from a single clone that
combines electrophoresis in one dimension (Fig. 7.15) with synthesizes only one light chain. All normal electrophoretic
immunodiffusion in the perpendicular direction. In the species of human Igs contain both light chain isotypes, al-
first step, electrophoresis separates the test antigens ac- though κ exceeds λ by the ratio of 2:1 in humans. As shown
cording to charge and size, in effect, separating the ori- in Figure 7.15C, the abnormal arc with γ mobility reacts

A)
B)
FIG. 7.15. Immunoelectrophoresis. A sample

containing multiple components is electro-
phoresed in an agarose gel, separating the
antigens in the horizontal dimension. Then a
horizontal trough is cut into the gel and an-
C)
tiserum is added. Immunodiffusion between
the separated antigens and the trough is
equivalent to having separate wells, each
with a different antigen.87 This technique is
used to identify a myeloma protein in human
serum. Sera from the patient or normal indi-
vidual were placed in the circular wells and
electrophoresed. Antisera were then placed
in the rectangular troughs and immunodif-
fusion proceeded perpendicular to the di-
rection of electrophoresis. The abnormally
strong reaction with anti-immunoglobulin (Ig)
G and anti-κ, but no reaction with anti-λ anti-
bodies, indicate a monoclonal protein (IgG,κ),
as polyclonal Ig should react with both anti–
light-chain antisera. Failure to form a band
with anti-IgM and a reduced band with
anti-IgA show typical reduction of normal
Igs in this disease. (Photographs courtesy
of Theresa Wilson, NIH Clinical Chemistry
Section.)
with anti-IgG and anti-κ but not anti-λ antiserum. Thus, it coating red blood cells has been the limiting factor in the
is identified as an IgG-κ monoclonal protein. usefulness of this method for certain antigens. Apparently,
slightly aggregated or partially denatured protein antigens
are adsorbed preferentially.92
Hemagglutination and Hemagglutination Inhibition The test for specific antibodies is done by serially diluting
Hemagglutination the antiserum in the U-shaped wells of a microtiter plate and
A highly sensitive technique yielding semiquantitative val- adding antigen-coated red cells. When cross-linked by spe-
ues for the interaction of antibody with antigen involves the cific antibodies, agglutinated cells settle into an even carpet
agglutination by antibodies of red blood cells coated with spread over the round bottom of the well. Unagglutinated
the antigen.92 For exogenous antigens that are adsorbed red cells slide down the sides and form a much smaller but-
to the red blood cell surface, the reaction is called passive ton at the very bottom of the well. The titer of a sample is
hemagglutination. Untreated red blood cells are negatively the highest dilution at which definite agglutination occurs.
charged, and electrostatic forces keep them apart. Following With hyperimmune antisera, inhibition of agglutination is
treatment with tannic acid (0.02 mg/mL for 10 minutes at often observed at high doses of antibody, termed a prozone
37°C), however, they clump readily. effect. Two interpretations have been offered: One is that, at
Untreated red blood cells are easily coated with polysac- great antibody excess, each cell is coated with antibody, so
charide antigens, which they adsorb readily. After tanning, cross-linking by the same antibody molecule becomes im-
the uptake of some protein antigens is good, giving a sensi- probable. The second interpretation is the existence of some
tive reagent, whereas for others it tends to be quite variable; species of inefficient or “blocking” antibodies that occupy

A B
FIG. 7.16. A: Western blot technique. The antigen preparation is run through a polyacrylamide gel, which separates its components into dif-
ferent bands. These bands are then transferred to paper by electrophoresis in the horizontal dimension. The paper is cut into strips. Each
strip is incubated with test antibodies, followed by further incubation with enzyme-labeled second antibodies. If the test antibodies bind to the
component antigens, they will produce discrete dark bands at the corresponding positions on the strip. B: Clinical specimens from patients
with acquired immunodeficiency syndrome tested on strips bearing human immunodeficiency virus-1 viral antigens, showing antibodies to
viral gag (p15/17, p24, and p55 precursor), pol (p66 and possibly p32), and env (gp41 and gp120) proteins. Lane 1 is the negative control, and
lanes 2 to 4 are sera from three different patients.
antigen sites without causing aggregation of cells.9 To as- The immunoblot is often used to detect viral proteins
sure antigen specificity, the antiserum should be absorbed with specific antibodies that bind these proteins on the ni-
against uncoated red cells prior to the assay, and an un- trocellulose blot. Then a second antibody, which is either
coated red cell control should be included with each assay. enzyme conjugated or radiolabeled, is used to detect the an-
IgM is up to 750 times more efficient than IgG at causing tigen–antibody band. Crude viral antigen preparations can
agglutination, which may affect interpretation of data based be used, as only those bands that correspond to viral anti-
on titration. The titer may vary by a factor of 2 simply due to gens will be detected.
subjective estimates of the endpoint. Typical results are shown in Figure 7.16. HIV-1 was cul-
Once the hemagglutination titer of an antiserum is deter- tured in susceptible H9 cells. The viral proteins were sepa-
mined, its interaction with antigen-coated red blood cells can rated by polyacrylamide gel electrophoresis and detected
be used as a sensitive assay for antigen. To constant amounts by immunoblot, using the serum of infected patients. Each
of antibody (diluted to a concentration twofold higher than antigen band recognized by the antiserum has been identi-
the limiting concentration producing agglutination) are fied as a viral component or precursor protein. The gp160
added varying amounts of free antigen. Agglutination will precursor is processed to mature gp120 and gp41 envelope
be inhibited when half or more of the antibody sites are oc- proteins, whereas a p66 precursor is processed to the p51
cupied by free antigen. In a similar fashion, the assay can mature form of reverse transcriptase, and a p55 precursor
be used for the detection and quantitation of anti-idiotype becomes the p24 and p17 gag and matrix proteins of the
antibodies that react with the variable region of antibodies virus.94 The practical uses of the HIV western blot include
and sterically block antigen binding. diagnosing infection, screening blood units to prevent HIV
transmission, and testing new vaccines.
Immunoblot (Western Blot)

A most useful technique in the analysis of proteins is poly- Surface Plasmon Resonance
acrylamide gel electrophoresis, in which charged proteins Surface plasmon resonance (SPR) uses the electromagnetic
migrate through a gel in response to an electric field. When properties of light to measure the binding affinity of a vari-
ionic detergents such as sodium dodecyl sulfate are used, the ety of biologic molecules, including antigen–antibody pairs.
distance traveled is inversely proportional to the logarithm In this method, polarized light passes through a glass plate
of molecular weight. The protein components of complex coated on the back surface with a thin metal fi lm, usually
structures, such as viruses, appear as distinct bands, each at gold. Biologic materials binding to the metal fi lm behind
its characteristic molecular weight. Because antibodies may the plate can alter its refractive index in ways that affect the
be unable to diffuse into the gel, it is necessary to transfer angle and intensity of reflected light.
the protein bands onto a nitrocellulose membrane support At angles close to perpendicular, light will pass through
first, where they are exposed for antibody binding.93 the glass, although it will bend at the interface due to

FIG. 7.17. Monoclonal Antibody to gp120 was Introduced into the Flow Cell at Time 0, and Antibody Binding was Followed Over Time as
a Change in the Critical Angle, Measured in Response Units. After 1000 seconds, free antibody was washed out, and the release of bound
antibody was measured as a decrease in refractive index. Lower affinity binding to gp120 from the IIIB strain (left) was shown as a faster “off
rate,” as compared to the very slow rate of antibody release from the MN strain (right). These results, obtained under nonequilibrium condi-
tions, provide direct measurement of the forward and reverse rate constants for antibody binding, and the ratio of these two gives the affinity
constant. Modified from VanCott et al.,97 with permission.
differences in the refractive index of glass and what is be- to the gold layer, and various concentrations of monoclo-
hind it. Above a certain angle, called the critical angle, bend- nal antibody to gp120 were added to the flow cell.97 Over
ing will be so great that total internal reflection will occur. the first 1000 seconds, antibody binding was measured as
Small changes in refractive index behind the glass can be a change in reflected light (in response units), allowing a
detected as significant changes in the critical angle, where calculation of the rate constant for the forward reaction of
light reflection occurs, and in the intensity of reflected light antibody binding. Once the signal reached a plateau, anti-
at this angle. By reading the reflected light intensity in a body was washed out of the flow cell, and the decrease in
diode array detector, the critical angle and intensity can be SPR signal over time indicated the rate at which antibody
determined simultaneously. Due to the wave nature of light, came off the antigen. The “on rate” for antibody binding to
the effect of refractive index in the gold fi lm extends about IIIB gp120 (left) was about twice as fast as for MN (right)
one wavelength beyond the glass or about 300 to 700 nm.95 at each antibody concentration. However, the “off rate” was
Within this layer, if an antigen is covalently attached to the about 50-fold less for MN than for IIIB. Combining these
gold, then antibody binding can be detected as a change in kinetic results indicates much greater binding affinity for
refractive index, resulting in a different critical angle and gp120 of MN type, which may explain the observation that
intensity of the reflected light. MN type virus was 10-fold more sensitive to neutralization
SPR systems have three essential features96 : an optical by this antibody than was the IIIB strain.
system that allows determination of the critical angle and
light intensity at the same time, a coupling chemistry that
links antigen or antibody to the gold surface, and a flow MONOCLONAL ANTIBODIES
system that rapidly delivers the binding molecule in the mo- Homogeneous Igs have long played important roles in
bile phase, so SPR can measure the rate of binding, rather immunologic research. Starting in the 1950s, Slater et al.98
than the rate of diffusion. Because binding causes a physi- studied sera from human patients with multiple myeloma
cal change in the gold film, there is no need for radioactive and recognized the relationship between abnormal my-
labels or enzyme conjugates. Molecular binding interactions eloma proteins and normal serum globulins. Potter 99 char-
can be followed in real time. acterized numerous mouse myeloma tumors and identified
A typical SPR experiment is shown in Figure 7.17. HIV the antigenic specificities of some of them. Human and
gp120 of type IIIB (left panel) or MN (right panel) were fi xed mouse myeloma proteins were studied as representative Igs

and recognized for the advantages they had with proteins

as diverse as antibodies for studies of Ig structure, func-
tion, and genetics. It was not yet possible, however, to in-
duce monoclonal Igs of desired specificity.
This goal was achieved by the introduction of hybridoma
technology by Köhler and Milstein1,100 and by Margulies et
al.101 in the 1970s. Since that time, monoclonal antibodies
have come to play an enormous role in biologic research and
applications. They offer as advantages the relative ease of the
production and purification of large quantities of antibody,
the uniformity of antibody batches, and the ready availabil-
ity of Ig messenger ribonucleic acid and DNA from the hy-
brid cell lines.
Derivation of Hybridomas
Hybridomas producing monoclonal antibodies are gener-
ated by the somatic cell fusion of two cell types: antibody-
producing cells from an immunized animal, which by
themselves die in tissue culture in a relatively short time, and
myeloma cells, which contribute their immortality in tissue
culture to the hybrid cell. The myeloma cells are variants
carrying drug selection markers so that only those myeloma
cells that have fused with spleen cells providing the missing
enzyme will survive under selective conditions. Initial work
used myeloma cells that secreted their own Ig products, but
later such fusion partners were replaced by myeloma vari-
ants that fail to express Ig102,103 so that the fused cell secretes FIG. 7.18. Production of Hybridomas. Steps in the derivation of
exclusively antibody of the desired specificity. Successful hybridomas can be outlined as shown. Spleen cells from immunized
hybridoma production is influenced by the characteristics donors are fused with myeloma cells bearing a selection marker.
of the cell populations (immune lymphocytes and myeloma The fused cells are then cultured in selective medium until vis-
fusion partner), the fusion conditions, and the subsequent ible colonies grow, and their supernatants are then screened for
selection and screening of the hybrids. A diagrammatic antibody production. HAT, hypoxanthine, aminopterin, thymidine;
HGPRT, Hypoxanthine-guanine phosphoribosyltransferase; PEG,
version of the overall process of hybridoma derivation is polyethylene glycol.
presented in Figure 7.18.
This section will not attempt to provide a detailed, step-
by-step protocol for laboratory use. For that purpose, the antibody to the mouse CD3 equivalent.106 Rabbit–mouse hy-
reader is referred to monographs and reviews on the subject, bridomas have also been described.107
including a detailed laboratory protocol with many hints A second approach is the use of fusion partner cells from
and mention of problems to avoid.104 the desired species. Myeloma variants carrying drug selec-
tion markers are available in a number of species. A rat my-
eloma line adapted for this purpose, IR983F, was described
Hybridomas Derived from Species Other than Mice by Bazin.108 Production of human hybridomas is of special
Laboratory mice are the most common species immunized importance because their use in therapies would avoid the
for hybridoma production, but for a variety of reasons, other problem of human immune responses to Ig derived from
animal species often have advantages. If an antigen of inter- other animal species, as discussed in detail in the later sec-
est is nonpolymorphic in the mouse, the mouse component tion on applications.
might be immunogenic in other species while mice would
be tolerant to it. In the case of hybridomas for clinical use,
mouse antibodies have the drawback of inducing antimouse Use of Gene Libraries to Derive
Ig immune responses with possible deleterious effects. Monoclonal Antibodies
Several approaches have been taken to the derivation of Monoclonal antibodies produced by hybridoma technology
hybridomas in species other than mouse. First, interspecies are derived from B cells of immunized animals. A recent alter-
hybridization can be performed using mouse myeloma fu- native technology uses gene libraries and expression systems
sion partners. The resulting hybrids are often unstable and instead. This approach has the advantages of avoiding labor-
throw off chromosomes, but clones can sometimes be se- intensive immunizations of animals and the screening of
lected that produce antibody in a stable fashion. Examples of antibody-containing supernatants. Another advantage of the
this would be rat–mouse fusion to produce antibody to the approach is circumventing tolerance. One can derive mono-
mouse Fc receptor105 and hamster–mouse fusion to produce clonal antibodies to antigens expressed in the animal species

that donated the gene library, including highly conserved an- activity. The efficiency of antibody production has been as
tigens for which there may be no available responder that does high as 30% to 60% per well in different experiments.
not express the antigen. Advantages include the ability to focus on a particular
The first version of such an approach involved prepara- B-cell compartment, such as memory B cells, during the
tion of V H and Vκ libraries and expression of the libraries sorting step, the large numbers of monoclonals produced
in bacteria. Further development of the system led to use in a single experiment (typically more than 100), and the
of VH and VL libraries (see Chapters 5 & 6) made sepa- absence of antigen boosting prior to cloning. Due to the
rately, and then preparation of a combinatorial library by high efficiency of cloning, these monoclonals are thought
cleaving, mixing, and religating the libraries at a restriction to represent the expressed B-cell repertoire at a given time,
site.109,110 A linker can be used so that V H and V L can both location, and stage of differentiation. For example, human
be expressed on one covalent polypeptide; the flexibility of monoclonals had the same ratio of kappa to lambda light
the linker allows association of the V H and V L in a normal chains (60% to 40%) as found in circulating B cells, and
three-dimensional configuration and thus formation of an they showed allelic exclusion in 95% of wells. Key features
antigen-binding site.110 of the method include the choice of fluorescent antigen for
Another innovation involves expression of V H and Vκ B-cell sorting, called the “bait,” and the choice of primers for
genes on the surface of bacteriophage as fusion proteins nested PCR. The bait should capture as many antigen spe-
with a phage protein to permit rapid screening of large cific B cells as possible; different bait will capture different
numbers of sequences.110–112 Adsorption of antibody-bearing monoclonals from similar populations of B cells (compare
phage on antigen-coated surfaces allows positive selection Pietzsch et al.120 and Scheid et al.121). A complex bait with
of phage containing DNA encoding the desired variable re- multiple epitopes, such as HIV gp140, will capture antibod-
gion fragment (Fv) from combinatorial variable region gene ies to different parts of the antigen.119,122,123 In order to am-
libraries.111,112 plify highly variable antibodies, the forward (and reverse)
Human antibody gene sequences can be recovered by PCR primers should be chosen upstream (or downstream)
polymerase chain reaction (PCR) from peripheral blood of the V H region so they will capture sequence variation near
cells,113 bone marrow,114 or human cells reimmunized in se- both ends of the coding sequence.
vere combine immunodeficiency–hu mice.115 The phage dis- In one study, 1.5 × 105 memory B cells from a chroni-
play technique can then be used to select antigen-binding cally HIV-infected subject were sorted based on binding to
clones and derive human reagents of desired specificity, such a stabilized form of gp120.121 Individual antigen-specific B
as antibody to hepatitis surface antigen113 or HIV envelope.114 cells were distributed in microtiter wells, and they produced
One limitation in the phage library technique initially 576 monoclonal antibodies that bound gp120 by ELISA.
was low affi nity of the monoclonal antibodies derived Sequencing the V H and V L chains showed that each mono-
because they were generated by a random process and not clonal was often clonally related to several others, which dif-
subject to further somatic mutation. Several approaches fered only by somatic mutation. Overall, these monoclonals
have now been used to improve affi nities. Hypermutation represented 200 distinct B-cell clones. The monoclonals
and selection has now been achieved in vitro by a strategy could be divided into groups, based on binding four distinct
using a bacterial mutator strain.116 The process involves epitopes on gp120. Within the same clonally related group,
multiple rounds of mutation followed by growth in nonmu- some monoclonals had potent HIV-neutralizing activity
tator bacteria and then selection for high-affi nity binding while other closely related monoclonals were inactive. The
led to an overall 100-fold increase in affinity.116 Improved B-cell donor was a long-term nonprogressor. The analysis
affi nity has also been achieved by use of site-directed mu- of a significant sampling of the antibody repertoire may
tagenesis to alter residues in hypervariable regions affecting provide insight into how these patients control their circu-
dissociation rates.117 lating virus, and these findings may lead to improved vac-
cine antigens.
Recombinant Monoclonal Antibodies Derived from
Single B Cells After Cell Sorting
A new and very productive approach to making monoclonal Applications of Monoclonal Antibodies
antibodies is based on amplification of both the V H and V L Because monoclonal antibodies can be made easily and re-
chains of single B cells by reverse transcription–PCR.118,119 In producibly in large quantities, they allow many experiments
the first step, B cells are sorted by fluorescence-activated cell that were not possible or practical previously. Affinity chro-
sorting, based on antigen specificity or other characteristics, matography based on monoclonal antibodies can be used as
and then distributed in 96 well plates at a dilution of one a step in purification of molecular species that are difficult
B cell per well. The cells are lysed, followed by reverse tran- to purify chemically. Homogeneous antibody can be crys-
scription of the messenger ribonucleic acid of a single cell tallized and can also be crystallized together with antigen
and amplification of the V H and V L chains by nested PCR to permit the study of the structure of antibody and of anti-
of each chain separately. The two chains are cloned into gen–antibody complexes by x-ray diffraction. Homogeneous
expression vectors in frame with the appropriate constant antibodies are also valuable in the study of antibody diver-
regions. Both chains are coexpressed to reconstitute the an- sity. Such analyses have revealed much about the roles of so-
tibody. The resulting monoclonal antibodies are then ana- matic mutation, changes in affinity, and changes in clonal
lyzed for sequence, antigen specificity, affinity, and biologic dominance in antibody responses.

Catalytic Antibodies One of the most powerful uses of hybrid antibodies is

One area of recent interest is the use of antibody molecules in redirecting cytolytic cells to targets of a defined specific-
to catalyze chemical reactions.124 In this role, antibodies ity. In one early demonstration of this use,134 a monoclonal
serve as an alternative to enzymes, an alternative that can antibody specific for the Fcγ receptor and one specific for
be customized and manipulated more easily in some cases. the hapten dinitrophenyl were chemically cross-linked. In
The concept of antibodies as catalysts had been proposed the presence of this hybrid antibody, FcγR-bearing cells
a long time ago by Woolley.124 Use of homogeneous antibod- were able to lyse haptenated target cells specifically. The
ies permitted identification of some with significant catalytic FcγR played a critical role; antibody to MHC class I anti-
effects; MOPC167 accelerates the hydrolysis of nitrophenyl gens on the cell could not be substituted. Antibody to the
phosphorylcholine by 770-fold.125 Polyclonal antibodies T-cell receptor complex has also been used extensively to
have also been reported to possess detectable enzymatic ac- redirect T-cell lysis to desired targets. For example, anti-
tivity.126 With the advent of hybridoma technology, purpose- CD3 was cross-linked to antitumor antibodies and mixed
ful selection of antibodies with potent enzymatic function with effector cells. These “targeted T cells” were able to
became possible. Antibodies have been characterized that inhibit the growth of human tumor cells in vivo in nude
catalyze numerous chemical reactions, with rates nearing mice.135 Bispecific antibodies have also been used recently
108-fold above the spontaneous rate.124 One common strat- to alter the tropism of a viral gene therapy vector to target
egy for elicitation of such antibodies is immunization with specific cells.136
transition state analogs,127 although there are other strate- Cumbersome cross-linking chemistry can now be re-
gies.128 Antibodies function as catalysts in a stereospecific placed by genetic engineering for creation of designer an-
manner,129 a valuable property. tibodies.133 Bifunctional and bispecific antibodies can be
Molecular mechanisms of antibody-mediated catalysis engineered as single chain variable fragment constructs or
vary, as do enzymatic reactions.128,130 To accelerate a reaction, by specialized strategies using two chains. A wide variety of
an antibody has to lower the activation energy barrier to the configurations are possible and can be used to make multi-
reaction, which means lowering the energy of the transition valent reagents as well as reagents with one site of each speci-
state by stabilizing it. For this reason, an antibody that rec- ficity. Tags can be built in by fusion of additional sequence
ognizes the transition state is favorable, and immunizations such as streptavidin, or, as mentioned previously, antibody
with analogs of the transition state have advantages. domains can be combined with other functional domains
Antibodies can serve as what has been termed an entropy such as toxins, enzymes, or cytokines.
trap124 ; binding to the antibody “freezes out” the rotational
and translational degrees of freedom of the substrate and
Clinical Applications
thus makes a chemical reaction far more favorable energeti-
cally. Interactions with chemical groups on the antibody can The possible clinical uses of monoclonal antibodies are
neutralize charges or bury hydrophobic groups, thereby sta- many. In vitro, they are widely used in RIA and ELISA mea-
bilizing a constrained transition state. surements of substances in biologic fluids, from hormones to
Discovery of such catalytic antibodies opens practical toxins. They are also extremely valuable in flow cytometric
opportunities: Antibodies can be customized for an applica- assays of cell populations using antibodies specific for dif-
tion by appropriate selection and can be produced relatively ferentiation antigens expressed on cell surfaces. Monoclonal
cheaply and purified easily. Catalytic antibodies can be de- antibodies plus complement or toxin-conjugated monoclo-
veloped to perform chemical reactions for which no enzyme nal antibodies have also been used to remove T cells from
is available. They can shield intermediates from solvent, for bone marrow prior to transplantation.137
example, allowing reactions that do not occur in aqueous In vivo, although it took more than two decades for the
solution.131 They can form peptide bonds,132 suggesting a original promise of hybridoma technology to be translated
new approach to polypeptide synthesis. Thus, catalytic anti- into widespread clinical applications, a number of mono-
bodies will likely have many practical applications. clonal antibodies are now in use or in trials for a variety
of purposes.138–140 Monoclonal antibody OKT3 directed
to a marker on human T lymphocytes is used as a treat-
Bispecific and Bifunctional Antibodies ment for rejection reactions in patients undergoing kidney
Antibodies produced naturally by a single B cell have only transplant.141 Other monoclonal antibodies, for example,
111
one binding site specificity, and their effector functions are In-labeled CYT-103 referred to as Oncoscint (Cytogen
determined by the structure of the Fc domain. The avail- Corp, Princeton, NJ),142 are used as diagnostic tumor im-
ability of monoclonal antibodies made possible the gen- aging reagents. Monoclonal antibodies have now been ap-
eration in quantity of artificial antibodies as cross-linking proved for various therapeutic uses.140,143 Cancer therapies
reagents by linking binding sites of two specificities to use either unconjugated monoclonal antibody138,144–148
form bispecific antibodies. A variety of techniques have or toxin-coupled149,150 or radiolabeled monoclonal anti-
been used to prepare such hybrid or bispecific antibodies, body.144,148,151 Molecules targeted in cancer therapies include
and they have been put to a variety of uses. In addition, CD25 (IL-2 receptor alpha chain) in adult T-cell leuke-
antibody-binding sites can be linked to other functional mia,138,140,148,152–154 CD20 in non-Hodgkin lymphoma using
domains such as toxins, enzymes, or cytokines to create either unlabeled147,155 or more recently radionuclide-labeled
“bifunctional antibodies.”133 anti-CD20,156,157 the HER-2/neu oncoprotein in breast and

ovarian cancer,145,146,158,159 CD22 in hairy cell leukemia,150 Another approach to production of monoclonal anti-
vascular endothelial growth factor to limit angiogenesis in bodies with human characteristics involves application
diverse tumors (especially colorectal, renal, and non–small- of genetic engineering. The part of the antibody structure
cell lung cancer),160–165 antiepidermal growth factor receptor recognized as foreign by humans can be minimized by
in colorectal carcinoma and others,166 anti-CD52 in chronic combining human constant regions with mouse variable
lymphocytic leukemia,167 and anti-CD33 in acute myelog- regions206,207 or even just mouse hypervariable segments208
enous leukemia.168 Other therapies studied include antili- by molecular genetic techniques. Antigen-binding specific-
popolysaccharide for treatment of sepsis, anti–IL-6 receptor ity is retained in some cases, and the “humanized” chi-
for treatment of multiple myeloma and for rheumatoid ar- meric molecules have many of the advantages of human
thritis, anti-IgE for treatment of allergy,139 anti–tumor ne- hybridomas.
crosis factor for treatment of arthritis,169,170 anti–respiratory Production of fully human monoclonal antibodies in
syncytial virus for prevention of respiratory syncytial virus transgenic mice has now been achieved by multiple labora-
morbidity and mortality in infants,171,172 and anti–IL-2 re- tories. The strategy has involved insertion into the mouse
ceptor (CD25) for prevention of graft rejection148,173 as well germ line of constructs containing clusters of human Ig
as for treatment of autoimmune diseases such as uveitis174 V, D, J, and C genes (see Chapter 6) to generate one trans-
and multiple sclerosis.175 Other monoclonal antibodies have genic line and targeted disruption of the mouse heavy chain
been developed to modulate immune responses. On the and κ chain loci to generate another transgenic line. From
one hand, antibodies to block checkpoint inhibitors in the these two lines, mice are then bred that express only human
immune response, especially against cancer, have been de- antibodies.
veloped, such as the first licensed one, ipilimumab, against To show feasibility of this approach, cosmids carrying
cytotoxic T-lymphocyte antigen-4 (CD152), an inhibitory parts of the human heavy chain locus were used to make
receptor expressed on activated T cells,176–181 and anti-PD-1, transgenic mice.209 The next step was to produce mice car-
another inhibitory receptor on activated T cells that is cur- rying human genes for both heavy and light chains to gen-
rently in trials.182–186 An antibody to transforming growth erate a functional human repertoire. Several groups using
factor–beta, a soluble inhibitor of immune responses, is different technologies constructed heavy chain miniloci
also in clinical trials in both cancer and fibrosis.187–190 On containing functional V segments representing several
the other hand, monoclonal antibodies are being developed major V region families, D and J segments, constant and
to amplify immune responses and synergize with vaccines, switch regions, and enhancers. The κ chain constructs were
such as agonist anti-CD40 antibodies.191–193 made that contained multiple functional Vκ segments, the J
In the specialized case of B-cell lymphoma, mono- segments, Cκ, and enhancers.210,211 Mice were bred that were
clonal anti-idiotypes against the idiotype expressed by homozygous both for the transgene loci and for disruption
the patient’s tumor have been tested as a “magic bullet” of the mouse heavy chain and κ light chain loci; note that
therapy.194 Active immunization of the patient with idiot- the mouse λ locus was left intact. The human Ig genes could
ype195–199 has the advantage that escape mutants200 are less rearrange in the mouse genome, and expression of human Ig
likely to emerge because multiple idiotopes are recognized. resulted. If these mice were immunized with a fragment of
Another approach under study is immunization using not tetanus toxin, resulting antibodies included some that were
idiotype as protein but plasmid DNA-encoding patient fully human.211 In one of the studies,210 serum contained
idiotype.201 This approach would have additional advan- human μ, γ1, and κ as well as mouse λ and γ. Immunization
tages, such as ease of preparing customized reagents for of such mice with various antigens led to class switching,
each patient. somatic mutation, and production of human antibodies of
affinities of almost 108.
Ig expression in these mice demonstrates cross-species
Production of Human or Humanized compatibility of the components involved in antibody gene
Monoclonal Antibodies rearrangement and diversification. The mice also provide a
Many of the side effects of monoclonal antibodies in clinical responder able to provide fully human antibodies to clini-
use are due to the foreign Ig constant regions. Recognition of cally important antigens, and they have the advantage that
foreign Ig epitopes can lead to sensitization and so preclude they are not tolerant to human antigens, such as the human
subsequent use in the same individual of different monoclo- IgE and human CD4 used by Lonberg et al.210
nal antibodies. Thus, monoclonal antibodies with some or
all structure derived from human Ig have advantages. Several
approaches have been taken employing fusion of human cells Nucleotide Aptamers: An Alternative to
with animal myelomas or with human tumor cells of vari- Monoclonal Antibodies
ous kinds202,203 and use of Epstein-Barr virus to immortalize Antibodies are not the only biologic macromolecules that
antibody-producing cells.204 Production of populations of have evolved to permit an enormous range of specific struc-
sensitized human cells to be fused presents another special tures. Oligonucleotides selected for ability to bind a ligand
problem, as the donors cannot be immunized at will. In one with high affinity and specificity are termed “aptamers”
example, in vitro stimulation of lymphocytes with antigen and can be used in many of the ways antibodies have been
followed by fusion with mouse myeloma cells has been used used. Selection, properties, and uses of aptamers have
to generate a series of antibodies to varicella zoster.205 been reviewed.212 Aptamers have the advantage that their

production does not require animals or cell culture. These functions as the receptor for HIV. Monoclonal antibod-
well-defined reagents may be used increasingly in diagnostic ies to CD4 can be divided into several groups based on
testing and are also being tested in clinical trials for use as competitive inhibition.216 The cluster containing the site
imaging agents or therapeutics. recognized by OKT4A is closely related to virus infection,
as antibodies to this site block syncytium formation. The
cluster recognized by OKT4, however, is not related to in-
Specificity and Cross-Reactivity fection because antibodies to it do not block syncytium
Specificity of Monoclonal Antibodies formation, 216 and cells expressing variant forms of the CD4
Because all the molecules in a sample of monoclonal anti- molecule lacking the OKT4 epitope can still be infected by
body have the same variable region structure, barring vari- HIV.217 This information about the sites on the molecule is
ants arising after cloning, they all have the same specificity. important in understanding the molecular interactions of
This uniformity has the advantage that batches of mono- virus with its receptor and may be useful in designing vac-
clonal antibody do not vary in specificity as polyclonal sera cine candidates.
often do. The most obvious fact about cross-reactions of While most antibodies are not MHC-restricted in their
monoclonal antibodies is that they are characteristic of all recognition of antigens, distinguishing them from T-cell
molecules and cannot be removed by absorption without receptors, antibodies can be selected that recognize pep-
removing all activity. An exception would be an apparent tide–MHC complexes218–220 (MHC-restricted antipeptide
cross-reaction due to a subset of denatured antibody mol- antibodies or peptide-dependent anti-MHC class I anti-
ecules, which could be removed on the basis of that binding. bodies). Several monoclonal antibodies have been selected
The homogeneity of monoclonal antibodies allows refine- that require both a certain MHC class I antigen and a par-
ment of specificity analysis that was not possible with poly- ticular peptide for reactivity. Such monoclonal antibodies
clonal sera. A few examples follow. are useful reagents capable of detecting cells presenting the
First, one can use monoclonal antibodies to distinguish appropriate peptide-MHC complexes on their surfaces.218
closely related ligands in cases where most antibodies in a Such monoclonal antibodies may also be useful in dissec-
polyclonal serum would cross-react, and so absorption of a tion of T-cell responses. In one study, the monoclonal an-
serum would not leave sufficient activity to define additional tibodies could inhibit IL-2 secretion by a T-cell hybridoma
specificities. This ability is useful in designing clinical assays of corresponding specificity and could also block induc-
for related hormones, for example. Such fine discrimination tion of cytotoxic T lymphocyte recognizing that epitope
also allows the definition of new specificities on complex when given in vivo during priming.220 Such monoclonal
antigens. When large numbers of monoclonal antibodies antibodies have been used to address structural questions
specific for class I and class II MHC antigens were analyzed, about antigen recognition by T and B cells.219 Such antibod-
some defined specificities that could not be defined with ex- ies also appeared to skew the repertoire of T cells for this
isting polyclonal antisera.213–215 particular HIV peptide MHC complex to specific T-cell re-
On the other hand, monoclonal antibodies are also a ceptor Vβ types and T-cell avidities.221 However, only very
powerful tool for demonstrating similarities rather than dis- rare monoclonal antibodies have this type of specificity
tinctions between two antigens. In some cases, only a minor and they were purposely selected in the fusions, so they do
portion of an antibody response detects a cross-reaction, not provide a general comparison of T-cell receptor and Ab
and so it is not detected by polyclonal reagents. For example, characteristics.
determinants shared by the I-A and I-E class II MHC anti-
gens in the mouse were demonstrated using monoclonal an- Cross-Reactions of Monoclonal Antibodies
tibodies215 while they had been suspected but were difficult Monoclonal antibodies display many type 1 cross-reactions,
to demonstrate using polyclonal sera. emphasizing that antibody cross-reactions represent real
Another type of fine specificity analysis possible only similarities among the antigens, not just an effect of het-
with monoclonal antibodies is the discrimination of spa- erogeneity of serum antibodies. Even antigens that dif-
tial sites (epitope clusters) by competitive binding. In some fer for most of their structure can share one determinant,
cases, such epitope clusters correspond to specificities that and a monoclonal antibody recognizing this site would
are readily distinguished by other means. However, in other then give a 100% cross-reaction. An example is the re-
cases, the epitope clusters may not be distinguishable by any activity of autoantibodies in lupus with both DNA and
serologic or genetic means. An example is the splitting of cardiolipin.222
the classical specificity Ia.7 into three epitope clusters by It should be emphasized that sharing a “determinant”
competitive binding with monoclonal antibodies.215 The does not mean that the antigens contain identical chemi-
epitopes cannot be distinguished genetically, as all three are cal structures, but rather that they bear a chemical resem-
expressed on cells of all Ia.7-positive mouse strains. Thus, blance that may not be well understood, for example, a dis-
polyclonal sera cannot be absorbed to reveal the different tribution of surface charges. Antibodies to the whole range
specificities. Only with the use of monoclonal antibodies of antigens can react with Igs in idiotype anti-idiotype re-
were the epitopes resolved from each other. actions, showing a cross-reactivity of the same antibodies
The importance of this type of analysis is shown by an- with proteins (the anti-idiotypes) and with the carbohy-
other example, the defi nition of epitope clusters on CD4, drates, nucleic acids, lipids, or haptens against which they
a surface molecule on a subset of human T cells that also were raised.

Polyclonal versus Monoclonal Antibodies These can be naturally divalent, for example in the case of
When monoclonal antibodies first became available, some IgG or multivalent, for example in the case of IgM or can be
people expected that they would be exquisitely specific and made as monovalent molecules such as Fab or recombinant
would be superior to polyclonal sera for essentially all pur- Fv fragments. They serve not only as a major arm of host de-
poses. Further thought about the issues discussed previ- fense playing a major role in the protective efficacy of most
ously, however, suggests that this is not always the case and existing antiviral and all antibacterial vaccines but also as
depends on the intended use of the antibodies. Not only do very versatile tools for research and clinical use. RIAs and
monoclonal antibodies cross-react, but when they do, the ELISAs have revolutionized the detection of minute quanti-
cross-reaction is not minor and cannot be removed by ab- ties of biologic molecules, such as hormones and cytokines,
sorption. A large panel of monoclonal antibodies may be and thus have become indispensable for clinical diagnosis
needed before one is identified with the precise range of re- and monitoring of patients as well as for basic and applied
activity desired for a study. research. Current solid-phase versions of these take advan-
In polyclonal sera, on the other hand, each different tage not only of the intrinsic affinity and specificity of the
antibody has a distinct range of reactivity, and the only antibodies but also of the implicit multivalency and local
common feature would be detectable reactivity with the high concentration on a solid surface. Cross-reactivity of
antigen used for immunization or testing. Thus, the serum antibodies often provides the first clue to relationships be-
as a whole may show only a low-titered cross-reaction with tween molecules that might not otherwise have been com-
any particular other antigen, and that cross-reaction can be pared. Conversely, methods that use antigens to detect the
removed by absorption, leaving substantial activity against presence of antibodies in serum have become widespread in
the immunizing antigen. For the purposes of an experi- testing for exposure to a variety of pathogens, such as HIV.
ment, a polyclonal serum may be “more specific” than any Antibodies also provide specific reagents invaluable in the
one of its clonal parts and may be more useful. This concept rapid purification of many other molecules by affinity chro-
is the basis of the theory of “multispecificity” (see previous matography. They have also become indispensable reagents
discussion). for other branches of biology, for example, in histocompat-
Polyclonal sera also have advantages in certain technical ibility typing and phenotyping of cells using a myriad of
situations such as immunoprecipitation in which multiva- cell–surface markers that were themselves discovered with
lency is important. Many antigens are univalent with respect monoclonal antibodies and for separating these cells by
to monoclonal antibody binding but display multiple dis- fluorescence-activated cell sorting, panning, or chromato-
tinct sites that can be recognized by different components of graphic techniques. Monoclonal antibodies have also finally
polyclonal sera. Thus, a greater degree of cross-linking can emerged as clinically important therapeutics in cancer, ar-
be achieved. thritis, organ graft rejection, and infectious diseases. Thus,
The ultimate serologic reagent in many cases may well antibodies are among the most versatile and widely used
be a mixture of monoclonal antibodies that have been cho- types of reagents today, and their use is constantly growing.
sen according to their cross-reactions. The mixture would Understanding the fundamental concepts in antigen–anti-
be better defined and more reproducible than a polyclonal body interactions thus has become essential not only to an
antiserum and would have the same advantage of overlap- understanding of immunology but also to the effective use
ping specificities. of these valuable molecules in many other fields.
CONCLUSION ACKNOWLEDGMENTS
In conclusion, antibodies, whether monoclonal or poly- We thank Drs. Charles DeLisi, Elvin A. Kabat, Henry
clonal, provide a unique type of reagent that can be made Metzger, and Fred Karush for helpful suggestions and dis-
with high specificity for almost any desired organic or cussion, and Dr. Suzanne L. Epstein for her valuable contri-
biochemical structure, often with extremely high affinity. bution to the monoclonal antibody section.

B-Lymphocyte Development
CHAPTER
8 and Biology
Richard R. Hardy
INTRODUCTION mutations. It can also be extended by analysis of gene or

B lymphocytes constitute one of the major arms of the protein expression at distinct intermediate stages. Critical
immune system, being responsible for humoral immunity. processes, such as D-J rearrangement and immunoglobulin
B cells in humans and mice are produced throughout life, (Ig) heavy chain expression, can also be mapped onto this
primarily in the fetal liver before birth and in the bone mar- framework. Progress in this work facilitates experiments
row after birth. Their development from hematopoietic stem that address additional issues, such as identification of key
cells (HSCs) has been extensively characterized in mice, and regulatory interactions, developmental checkpoints, and the
the generation of numerous gene-targeted and transgenic mechanism of B-lineage commitment.
lines in many cases has provided crucial information on the The following sections will cover the sites of B-lineage
role of transcription factors, cellular receptors, and interac- development at different stages of ontogeny, then focus on
tions that are critical in their generation. Recently, the role what is known about their development in the bone marrow
of microribonucleic acids (miRNAs) in regulating hemato- of adult mice, highlighting the function of the pre–B-cell
poietic development has also emerged. The complexity of receptor and the crucial role of Ig heavy and light chains in
this process is now apparent, and B-cell progenitor differ- guiding development. Later sections will consider their dif-
entiation into multiple peripheral subsets with distinctive ferentiation into various specialized peripheral populations
functions is also widely appreciated. This chapter will focus and emphasize insights into B-cell selection gained from
on B-cell development and function in the mouse, touch- various transgenic models of tolerance.
ing more briefly on aspects of human B cells that are similar
or distinctive, with a focus on immunodeficiency and B-cell Early Development
neoplasias. It will conclude with a brief description of novel Sites of B Lymphopoiesis during Ontogeny
aspects of B-lymphocyte development in other species, high- In the mouse, hematopoiesis occurs predominantly in the
lighting differences from development in mouse and human. fetal liver prior to birth, in the spleen just prior to and shortly
after birth, and in the bone marrow thereafter. Prior to liver
B-CELL DEVELOPMENT IN MICE hematopoiesis, the blood islands of the yolk sac (YS) contain
In mice, B cells are produced from HSCs through a complex the first identifiable hematopoietic cells, nucleated erythro-
process of differentiation that has been uncovered over the cytes with embryonic forms of hemoglobin.1 However, these
past 30 years or so. One of the goals of classical hematology early YS precursors appear incapable of generating other
has been the delineation of differentiation pathways for dif- blood cell lineages, and generation of all blood cell types,
ferent lineages of blood cells. There has been considerable including lymphocytes,2,3 initiates at around 9 to 10 dpc in
progress in utilizing the ordered expression of a diverse set an embryonic region referred to as the splanchnopleura/AGM
of cell surface and internal proteins, some with known func- (or simply Sp/AGM). Cells from this site are capable of long-
tions, others whose roles are only suspected, to construct a term repopulation of lethally irradiated adult recipients with
description of the intermediate stages that cells transit as all blood lineages.4,5 These cells colonize the fetal liver at about
they develop into B lymphocytes. A simplified example of 11 dpc, initiating hematopoiesis there. Thus, there are two
such a description is presented in Figure 8.1. Thus, HSCs with sites of very early hematopoietic precursors, with one in the
the capacity to generate all the cell types in blood generate YS largely limited to erythropoiesis and the other in the Sp/
progeny with a more restricted capacity, recognizable in AGM capable of complete (referred to as “definitive”) hema-
this example by expression of the receptor for interleukin topoiesis. However, it may be that precursors in the YS have
(IL)-7. These in turn produce yet more restricted progeni- a broader lineage potential in the fetal microenvironment, as
tor cells identified by expression of cluster of differentiation when they are injected directly into the newborn liver.6
(CD)45R/B220 (and, importantly, by absence of CD19). HSCs capable of developing into all the blood cell types
This kind of pathway can be constructed by isolation are produced in the Sp/AGM and migrate to the fetal liver
and short-term culture of intermediate stages, allowing at about d10. Thereafter, B-lineage cells develop largely in
progression to occur, which helps to define the order. This a wave, with earlier stages present at earlier times and later
framework for development serves as a starting point for stage predominating at later times, close to (and shortly
analysis of the effects of transcription factors, microenviron- after) birth.7,8 This progression with gestation day is easily
mental interactions, cytokines, and natural or engineered visualized by staining with antibodies that delineate B-cell
215

FIG. 8.1. Differentiation Diagram for Development of B Cells from Hematopoietic Stem Cells. Expression of the each surface protein is
indicated by a line. Changes in level of expression is indicated by line thickness. CLP, common lymphoid progenitor; ELP, early lymphoid
progenitor; Fo B, follicular B cell; HSC, hematopoietic stem cell; MLP, multilineage progenitor; NF B, newly formed B cell.
development, as shown in Figure 8.2. Early precursors can B-cell development shifts to the bone marrow and there-
also be found in the fetal omentum.9 In contrast with the after it continues for the life of the animal. B lymphopoi-
bone marrow, cells at most differentiation stages in the fetal esis decreases in aged mice. This may be due to diminished
liver appear to be rapidly proliferating so that larger and responsiveness of precursors to IL-7.24,25
larger numbers of B-lineage cells are detected at progressive
days of gestation. Another distinction of fetal liver from bone Stem Cells, Commitment, and Early B-Cell Progenitors
marrow development in the adult is the absence of termi- in Bone Marrow
nal deoxynucleotidyl transferase (TdT),10,11 an enzyme that B cells are continually generated from HSCs in the bone
mediates nontemplated addition of nucleotides at the D-J marrow of adult mice. Considerable effort has focused on
and V-D junctions of Ig heavy chain.12–14 Therefore, heavy evaluating the functional capacity of fractions of bone mar-
chains produced during fetal development have little or no row cells to repopulate different lineages of blood cells;
N-region addition, and CDR3 diversity is constrained even this work has progressed to the stage of defining a phe-
further by favoring of short stretches of homology at the V-D notype for such cells, with expression of c-KIT constitut-
and D-J junctions.15,16 Rearrangement of certain V or D ele- ing an important marker in the so-called lineage negative
ments may also differ between fetal and adult development, subset.26,27 This is the small fraction of bone marrow cells
as for example, the reported high utilization of the DFL16.1 (< 5%) that lacks expression of a panel of “differentiation
segment in fetal liver.17 Differential expression of genes other markers,” cell surface molecules that are expressed on later
than TdT also distinguish B-cell development during fetal stages of various hematopoietic cell lineages. Careful analy-
life from that in the adult including precursor lymphocyte sis of this HSC fraction using additional markers has shown
regulated myosin light chain like PLRLC transcripts11,18 that it represent perhaps 1/30,000 of nucleated bone marrow
and major histocompatibility complex (MHC) class II.19,20 cells with as few as 10 mediating multilineage repopulation
Interestingly, although absence of the cytokine IL-7 com- in cell transfer assays.28–30 An important capacity of “true”
pletely eliminates bone marrow B-lineage development,21 or “long-term repopulating” HSCs is their ability to give
it nevertheless spares some fetal development,22 suggesting rise to cells in a recipient mouse that can also repopulate all
a difference in growth requirements. The B-cell progeny of the blood cell lineages upon retransfer into a second host,
this early fetal wave may largely consist of B cells quite dis- indicating a capacity for extensive self-renewal without dif-
tinct from adult-derived cells, populating the B-1 subset.23 ferentiation into more restricted progenitors.
At birth, B-cell development can also be detected in A major focus in research on hematopoiesis has been
spleen, but development at this site gradually decreases to defining and characterizing lineage-restricted progeni-
very low levels by 2 to 4 weeks of age. Over this same period, tors, such as the common myeloid and common lymphoid

CHAPTER 8
B-LYMPHOCYTE DEVELOPMENT
FIG. 8.2. Phenotypic Progression of Developing B-Lineage Cells in Mouse Fetal Liver Analyzed at Different Days of Gestation. Note that B220+cluster of differentiation (CD)43+ cells
precede B220+CD43− cells and that within the B220+CD43+ fraction, heat stable antigen (HSA)− cells precede HSA+ cells, and BP-1− cells precede BP-1+ cells. Within the B220+CD43−
|
fraction, the immunoglobulin M+ percentage increases until birth (at about day 20).
217
9/17/12 5:28 AM
progenitors (CLPs).31–33 The CLP cell fraction was identified deoxyribonucleic acid (DNA) segments lost upon D-J rear-
by lack of a panel of “lineage markers,” expression of the rangement, and the formation of such D-J segments in indi-
IL-7Rα chain, and distinctive, intermediate levels of c-KIT, vidual cells isolated by electronic cell sorting showed that 30%
compared to higher levels on HSCs. Initial characterization to 50% of cells in CLPs and more than 80% of cells in Fr. A
of these cells in various functional assays suggested that contained a D-J rearrangement on at least one chromosome.44
these cells could generate B, T, natural killer (NK), and a This is consistent with high-level expression of genes important
subset of dendritic cells but no other blood cell lineages. The in Ig rearrangement, including TdT, Rag-1, and Rag-2, in CLP46
reason for this restriction has been intensively studied, and and Fr. A stage cells.
downregulation of the receptor for granulocyte–myeloid An emerging view of CLPs (and possibly even the earlier
colony stimulating factor has been suggested to be a key multilineage progenitor [MLP]) stage cells considers them
event in this process.34 Cells with the phenotype of CLP con- to be early B-lineage precursors rather than branch points
stitute about 1/3000 of bone marrow cells. Prior to the CLP in the production of other hematopoietic cell lineages. Thus,
stage, multipotent progenitors exhibit low-level expression analysis of CD4/CD8/CD3 “triple-negative” cells in thymus
of genes characteristic of diverse cell lineages, leading to the failed to identify cells with a surface phenotype comparable
idea that such promiscuous expression indicates chromatin to CLPs, and mutant mice lacking CLPs in bone marrow
accessibility that facilitates flexibility in cell fate decisions.35 nevertheless have relatively intact thymic development, lead-
CLP cells can give rise in short-term cultures to cells ing these authors to suggest a distinct “early T progenitor”
of the B lineage, naturally raising the issue of when cells different from CLPs.47 Furthermore, cells with T/myeloid
become restricted to the B lineage. Most cells growing in potential, but lacking B-lineage capacity, have been de-
stromal cultures give rise only to B cells upon transfer into scribed.48 It seems reasonable to hypothesize that MLP, CLP,
mice, and the phenotype of these cells has been well char- and Fr. A stage cells occupy a distinctive microenvironmen-
acterized.36 Most have at least some heavy chain rearrange- tal niche in bone marrow where they receive signals that
ment and bear the B-lineage marker CD19.37,38 There is less guide them along the early stages of B-lineage development
certainty concerning the cells isolated directly from primary culminating in CD19 + pro-B cells that are irreversibly com-
lymphoid tissues, as such cells are quite rare similar to CLPs mitted to becoming B cells due to expression of PAX539 (see
and HSCs. Most of the CD45R/B220 + cells in bone marrow following section). Interestingly, cells considered to be pro-
are also CD19 + ; such cells are committed to the production gressing down a lymphoid/B-lineage path can be redirected
of B cells.39 However, a subset of B220 + cells lacks detect- to become dendritic cells by signals through toll-like recep-
able CD19 expression; cells within this fraction can generate tors (TLRs), suggesting that infection can profoundly alter
CD19 + cells in short-term stromal culture with IL-7. Such early stages in hematopoietic development.49
cells are included in the CD43 + CD24low fraction (Fr. A, 1% Additional issues remain regarding the lineage restric-
of bone marrow) of B220 + cells in bone marrow, but this tion of cells at these early stages in B-cell development. For
fraction also contains other cell types including NK-lineage example, there is evidence that cells restricted to generating
precursors.38,40 Thus, it is necessary to exclude cells lacking B and myeloid/macrophage (but not T) lineage may exist
AA4.1 (about half38) and expressing Ly6c.41 Many of these in the fetal liver50 and even in bone marrow.51 There is also
Ly6c + cells also express CD4; recent work suggests that apparently a different dependence of fetal liver B lympho-
these are plasmacytoid dendritic cells.42,43 A phenotypic ap- poiesis on expression of the transcription factor BSAP com-
proach for enriching and fractionating very early B-lineage pared to bone marrow, as determined by analysis of PAX5
subsets is shown in Figure 8.3A. null mice.52 Further comparison of B-cell development in
Careful analysis of the LIN− (including CD19) IL7Rα + cKIT+ fetal liver with that in bone marrow is needed to clarify this
CD45/B220 − (CLP) and CD45/B220 + (Fr. A), as delineated in point. Finally, the precise delineation and characterization
Figure 8.3, modified our understanding of the earliest stages of of B-cell precursors earlier than CLPs, prior to IL-7 expres-
B-cell development in bone marrow.44 First, while CLP stage sion, remains imprecise. It seems likely that at least some
cells fail to efficiently generate myeloid cells upon transfer into of the MLP stage cells mentioned previously are initiating a
irradiated hosts, they nevertheless retain significant capacity to B-lineage program based on their expression of E2A, Rag-1,
produce such cells in short-term cultures, likely due to contin- Rag-2, and TdT.44 However, potential heterogeneity in this
ued (albeit reduced) expression of receptors for myeloid growth fraction needs to be assessed. Determination of Rag-1 tran-
factors. In contrast, this myeloid capacity is greatly reduced as scriptional activity at the single cell level by a green fluo-
cells begin to express CD45R/B220 (ie, become “Fr. A”), con- rescent protein reporter, used for identification of the early
comitant with reduced expression of receptors for myeloid lymphoid progenitor fraction,53 may provide a key approach
growth factor receptors. Yet these Fr. A cells, while poorly re- for such studies.
constituting T cells in cell transfer assays, nevertheless retain
the capacity to generate T-lineage cells in culture, mediated by Transcription Factors Regulating
engagement of Notch by its ligand DL1.45 Thus, the potential for B-Lineage Development
alternate hematopoietic lineages appears to be lost somewhat The GATA-2 and Runx1/AML1 transcription factors are
later in progression down the B-lineage pathway in mouse bone required for the development of HSCs that are the precursors
marrow than previously thought. On the other hand, it appears of all the blood cell lineages, including B cells54–58 (Fig. 8.4).
that initiation of Ig rearrangement is initiated earlier than some Experiments with the core-binding factor–associated leu-
studies have indicated. Determination of the extent of germline kemia fusion protein CBFbeta-SMMHC, whose expression

CHAPTER 8
FIG. 8.3. A: An approach for purifying the earliest stage of B-lineage cells in mouse bone marrow. Bone marrow cells expressing cell surface proteins characteristic of differentiated stages of T,
myeloid, erythroid, and B lineages are depleted sequentially by electronic gating in the first three panels. Cells with low-level expression of cluster of differentiation (CD)24/heat stable antigen
(HSA) and intermediate levels of CD43(S7) are selected in the fourth and the distribution of CD45R/B220 versus CD93/AA4 is shown in the fifth. AA4+B220− cells contain multilineage progenitors
(MLPs) and common lymphoid progenitors (CLPs), resolved by analysis for c-Kit versus interleukin (IL)-7Rα in panel six. AA4+B220+ cells, shown in the final panel, are enriched for c-Kit+IL-
7Rα+ cells, termed Fr. A. CLP stage cells resemble Fr. A, but lack detectable expression of CD45R/B220. In contrast with CLP and Fr. A, MLP stage cells have higher levels of c-Kit and lack IL-
7Rα expression. B: Functional analysis of early B-lineage cells by in vivo competition assay, showing absence of myeloid or T-lineage generation, but production of B-lineage cells from Fr. A. In
|
contrast, CLP stage cells generate B and T cells, whereas MLPs repopulate B, T, and myeloid lineage cells. MLP, CLP, and Fr. A as identified in A. Fr. B stage cells are DJ/DJ rearranged pro-B
cells, identified as CD19+CD43+CD24(HSA)+. C: Functional analysis of early B-lineage cells by in vitro S17 stromal cell assay, showing predominant B-lineage colony formation from Fr. A, but
219
some myeloid generation from CLP stage cells. D: Functional analysis of early B-lineage cells by in vitro DL1-OP9 stromal cell culture, revealing significant T-lineage potential in Fr. A stage cells.
9/17/12 5:28 AM
FIG. 8.4. Transcription Factors Important at Different Stages in B-Cell Development in Mouse Bone Marrow. Some regulatory networks are
also shown. Positive/activating activity is indicated by arrows, whereas negative or blocking activity is indicated by bars. The rapidly cycling
stage, early pre-B cell, is also indicated. Predominant stages of expression are indicated below the diagram.
inhibits RUNX function, have revealed that its expression concert with PAX5 to maintain B-lineage commitment, but
negatively impacts pre–pro–B-cell through pre–B-cell pop- also altering chromatin compaction around the IgH locus,
ulations in bone marrow.59 Although the frequency of CLP which together with Rag expression results in heavy chain
stage cells was unaffected, the expression of B-lineage asso- rearrangement. Aiolos is detected somewhat later in devel-
ciated genes (such as CD79a and λ5) was decreased, demon- opment at about the stage of B-lineage commitment, and its
strating the key importance of early RUNX activity. expression increases further at later stages. It is induced by
Somewhat later acting, but still very early in develop- pre–B-cell receptor (BCR) signaling (see section on the role
ment, is the Ikaros transcription factor.60–62 Ikaros and the of Ig heavy chain and the pre-BCR), and it acts to down-
related transcription factor Aiolos63 play important roles in regulate transcription of λ5 (initially induced by action of
lymphocyte development. Ikaros is expressed early in hema- Ikaros), a part of the pre-BCR complex. In this way, Ikaros
topoietic precursors. Ikaros null mice lack B-lineage cells62 ; family members play key roles both initiating and termi-
a different Ikaros mutant that acts as a dominant negative nating pre-BCR signaling, a critical checkpoint in B-cell
completely blocks lymphoid development.60 Ikaros activates development.
numerous early B-lineage genes, including TdT, Rag-1, λ5, The PU.1, an Ets family transcription factor, is critical for
and VpreB. Expression of EBF1 in Ikaros − / − hematopoi- progression to the earliest stage of lymphoid development,
etic progenitor cells restored generation of CD19 + cells, but as demonstrated by the inability of PU.1 null precursors to
these cells were not committed to the B-cell fate and failed generate lymphocytes.65,66 An important target of PU.1 for
to rearrange IgH genes.64 Thus, Ikaros acts in a transcrip- B-lineage development is the gene for Igβ, known as MB-1.
tion factor pathway, inducing EBF1 expression, and acting in The level of PU.1 appears to be critical for development

CHAPTER 8 B-LYMPHOCYTE DEVELOPMENT | 221
along the B lineage, as, while low-level expression induced marrow development at the pro-B stage, likely due to the
in PU.1 null mice allowed B-lineage development, high- lack of complete heavy chain rearrangements and also due
level expression blocked this and fostered myeloid lineage to the absence of the critical B-cell adaptor protein BLNK
development,67 likely due to differential induction of the that serves to link the pre-BCR to the intracellular signal-
IL-7Rα and macrophage colony-stimulating factor recep- ing pathway via the tyrosine kinase Syk.93 BSAP/PAX5 also
tor chains.68 In fact, retroviral mediated expression of the acts to repress alternate cell fates, as pro-B phenotype cells
IL-7Rα chain complements defective B lymphopoiesis in isolated from PAX5 null bone marrow can generate diverse
PU.1 null bone marrow HPCs.69 Surprisingly, recent work hematopoietic cell lineages, in contrast with such cells from
from several groups indicates that some B-cell development wild-type mice that are B-lineage restricted.39,94 This occurs
can occur in the absence of PU.1 expression.70 Furthermore, by repression of the myeloid growth factor receptor gene
analysis of conditional PU.1 knockout mice showed that c-fms95 and by repression of the Notch1 signaling pathway 96
expression of this transcription factor was not required after critical for T-cell fate specification.45,97 Conditional targeting
the pre–B-cell stage.71 of PAX5 in more mature B-cell stages shows that its contin-
E2A codes for two proteins, E12 and E47, members of ued expression is necessary for maintenance and function
the basic helix-loop-helix family of transcription factors; of mature B cells.98 Finally, as mentioned previously, in con-
its induction is crucial from the earliest stages of B-lineage trast with bone marrow, the absence of BSAP/PAX5 arrests
development, as all stages after CD19 expression are absent B-cell development prior to the B220 + stage in fetal liver,
from E2A null mice.72,73 These mice lack detectable D-J suggesting a crucial difference in the early dependence on
rearrangements, and, interestingly, such rearrangements this transcription factor.52
can be induced in nonlymphoid cells by introduction of the The Forkhead family transcription factor FoxO1 plays
Rag genes and ectopic expression of E2A,74 implicating this important roles at several stages of B-cell development.99
transcription factor in the process of chromatin remodel- Early in development, it acts to induce expression of the re-
ing of the Ig heavy chain locus that permits accessibility by ceptor for IL-7 at the CLP stage. It also functions to regulate
the recombinase machinery.75 The regulation of E2A is cru- Rag-1 and Rag-2 expression during heavy and light chain re-
cial for B-lineage development, as negative regulators such arranging stages of development.100 Finally, in mature B cells,
as Notch1 and ID2 have been show to block this lineage it is needed for normal expression of L-selectin, a homing
and induce alternate cell fates, the T and NK lineages.76–79 receptor important for normal recirculation of peripheral B
Consistent with this picture, ectopic expression of genes that cells through the lymphatics. GA binding protein, a ubiqui-
negatively regulate Notch1, lunatic fringe, and Deltex-1 in- tously expressed Ets family transcription factor, is another
duce the B-cell fate.80–82 player regulating expression of the IL-7R.101 Importantly,
Expression of the early B-cell factor, EBF1, a member of through interaction with PAX5, it acts in concert to induce
the O/E protein transcription factor family, is requisite for expression of critical PAX5 target genes such as CD79a.
progression of early B-lineage progenitors to the D-J rear- There is recent evidence that FoxO1 regulates Ikaros ac-
ranged pro-B stage (Fr. B), as shown in EBF1 null mice.83 tivity by altering splicing of its messenger ribonucleic acid
The expression of EBF1 is induced by action of the epigenetic (mRNA), rather than altering Ikaros transcription.102 FoxO1
histone H2A deubiquitinse MYSM1, as revealed by targeting activation of Ikaros was sufficient for induction of rear-
this gene.84 EBF1 and E2A act at a similar stage in early in rangement of proximal VH genes, but expression of PAX5
B-lineage development; these two transcription factors can was required for rearrangement of distal VH genes. Thus,
act together to upregulate a family of early B-lineage–specific FoxO1, Ikaros, and PAX5 appear to function in a network
genes, including Ig-α / β, VpreB/λ5, and Rag-1/2.85,86 There is to coordinate the ordered rearrangement of Ig genes during
evidence that E2A upregulates expression of EBF1, found B-cell development.
by transfection of E2A in a macrophage cell line,87 suggest- Lymphoid enhancer binding factor (LEF-1) shows a pat-
ing an ordering of these two in development. Furthermore, tern of expression restricted to the pro-B and pre-B stages
recent studies showed that there are two distinct promoters of B-cell development.103 Targeted inactivation of the LEF-1
for EBF1 that are regulated differently.88 A distal promoter is gene allows B-cell development but with reduced num-
activated by IL-7 signaling, E2A and EBF1, whereas a proxi- bers.104 This is because LEF-1 regulates transcription of
mal promoter is regulated by PAX5, Ets1, and PU.1. Such the Wnt/ β -catenin signaling pathway whose activation
complex regulation indicates that B-cell development occurs increases proliferation and decreases apoptosis of early
through the action of several feedback loops in a regulatory B-lineage cells. In fact, exposure of normal pro-B cells to
network that is becoming understood.89,90 Wnt protein induces their proliferation.104 Interestingly,
BSAP, the product of the PAX5 gene, is expressed through- there is a counterproliferative signal that can act at the pre-B
out B-cell development until the plasma cell stage.91 PAX5/ proliferative stage, mediated by transforming growth factor-
BSAP transcriptional targets include CD19 and BLNK; β1.105 It appears that this occurs due to induction of the ID3
expression of this transcription factor acts to upregulate V inhibitor that negatively regulates the activity of E2A.106
to D-J heavy chain rearrangement.52 Analysis of chromatin Another transcription factor whose expression is similar to
structure around the Ig heavy chain locus revealed that PAX5 LEF-1 is SOX-4; its inactivation also results in the inability
induces V to D-J locus contraction, thereby promoting rear- of normal early B-lineage cell expansion and a block at the
rangement.92 PAX5 null mutant mice show an arrest in bone pro-B stage.107

Several forms of NF-κb subunits are expressed through- expression; such a study with miR-150 reveal its role in regu-
out B-cell development; this transcription factor can regulating c-Myb, a transcription factor that regulates pro-B to
late kappa light chain expression and also growth factor pre-B progression and also the survival of mature B cells.120
signaling.108 Mice lacking the p65 subunit die before birth, so Transgenic overexpression of miR-17-92, a miRNA often
development must be analyzed by transfer of fetal liver pre- found to be amplified in lymphoma, resulted in a lympho-
cursors into wild-type recipients. Such experiments showed proliferative syndrome and autoimmune disease, resulting
diminished B-lineage cell numbers, but the major defect was in premature death.121 One target of miR-17-92 is Bim; its
in mature B-cell mitogenic responses.109 Mice lacking the decrease in the transgenic animals may have resulted in
p50 subunit showed relatively normal B-cell development excessive cell survival and a loss of normal tolerance to self-
but again poor response to mitogen by mature B cells.109 antigens. The potential relevance to normal growth regu-
However, mice lacking both the p50 and p65 subunits failed lation is quite interesting, considering the amplification of
to generate any B220 + B-lineage cells. Curiously, when this miRNA in some lymphomas. Another miRNA with a
mixed with wild-type fetal liver cells, normal numbers of cancer association, miR-21, has been studied as a transgene
mature B cells could be generated from the double-defective in mice, where it results in tumors with a pre-B malignant
precursors, suggesting that the defect could be overcome by lymphoid-like phenotype.122 The miRNAs that are amplified
secreted or membrane-bound signals provided by the wild- in cancers and likely contribute to the neoplastic process are
type precursors. Another double mutant, p50p52, showed a now termed “oncomirs.”
late stage defect in B-cell development, with a failure to gen- miRNAs can also influence lineage choice or act at key
erate mature B cells in spleen.110 checkpoints during hematopoietic development. Retroviral
Inactivation of the Oct-2 transcription factor results provision of miR-34a in bone marrow hematopoietic pro-
in neonatal lethality, but transfers of fetal precursors can genitors blocked B-cell development at the pro-B to pre-B
reconstitute lymphoid cells in wild-type recipients, allow- checkpoint, resulting in reduced numbers of mature B cells
ing assessment of effects on the B lineage. Such studies in mice repopulated with such precursors. A possible expla-
have shown that fewer mature follicular B cells are gener- nation for the developmental block is action of miR-34a on
ated in these mice and B-1 (CD5 +) B cells are completely Foxp1, a transcription factor that appears to be important
eliminated.111–113 Similarly, the Oct binding factor, OBF-1, at this stage, as cotransfection of FoxO1 lacking its nor-
also known as OCA-B and BOB-1, appears to function in mal 3′ UTR target of miR-34a restored B-cell development.
the maturation of newly formed B cells in the bone mar- A novel regulator of lineage choice appears to be Let-7, a
row to become follicular B cells in the periphery, as inactiva- family of miRNAs regulated by the highly conserved ribo-
tion of this gene resulted in a significant deficit in mature nucleic acid–binding protein Lin28.123 While Lin28 has been
B cells.114–116 Both of these transcription factors have been studied for its role in pluripotency, developmental timing,
shown to regulate the follicular B-cell chemokine recep- and oncogenesis, a recent study indicates that it may regu-
tor CXCR5, which may explain at least part of the defect.117 late a developmental switch in both B and T cells, such that
Curiously, unlike Oct-2 null mice, there was reportedly no its expression in adult hematopoietic progenitors results in
deficit in B-1 B cells in OBF-1 null mice. Interestingly, when reprograming development toward a more “innate-like”
the OBF-1 mutant mouse is crossed with btk-deficient mice, pattern, normally only seen during fetal/neonatal timing.124
there is a complete lack of peripheral B-cell generation,118
suggesting that this transcription factor may function in the
Bone Marrow Developmental Stages
BCR-mediated selection of mature B cells.
Functional Definition
Impact of Microribonucleic Acids on Distinct stages of developing B-lineage cells can be delineated
B-Cell Development based on their capacity for growth under different culture
The miRNAs are small noncoding ribonucleic acids that conditions. That is, the earliest precursors require cell con-
facilitate the degradation of mRNAs and thereby act at a tact with the stromal microenvironment in addition to spe-
posttranscriptional level to regulate gene expression. The cific cytokines, notably IL-7.105,125 This stromal cell/precursor
generation of mature functional miRNAs requires action adhesive interaction is mediated, at least in part, by binding
of Dicer, a protein that cleaves pre-miRNAs, so the target- of VLA-4 to intercellular adhesion molecule-1.126,127 Later
ing of Dicer allows assessment of the global effect of miRNA stage cells do not require cell contact but maintain a need
on B-cell development. Ablation of Dicer in early B-lineage for cytokines.128,129 Both cell types can undergo considerable
progenitors results in a block at the pro- to pre-B cell stage, cell proliferation in culture. Interestingly, the difference be-
likely due to upregulation of the proapoptotic molecule tween cell contact requirement and independence is linked
Bim.119 Counteraction of Bim function by a Bcl-2 transgene to the expression of heavy chain protein.128,130 A population
reveal further dysregulation of normal development, includ- of cytoplasmic heavy chain–expressing B-lineage cells later
ing nontemplated nucleotide addition (N-regions) at the Ig than either of these, the so-called late or small pre-B cells,
light chain V-J junction, due to aberrant expression of the does not proliferate in culture. These cells likely require dif-
terminal deoxynucleotidyl transferase gene that is normally ferent culture conditions for survival, as they usually do not
extinguished at the pre–B-cell stage. persist for extended periods, but rather die with a half-life of
Another approach for assessing the importance of spe- less than 24 hours unless protected from apoptosis by a Bcl-2
cific miRNAs is direct overexpression or knockdown of transgene.131

FIG. 8.5. Flow Cytometry Approach for Analyzing Different Stages of Cluster of Differentiation (CD)19+ B-Cell Development in Mouse Bone
Marrow. Note that the antibody used for CD24/heat stable antigen (HSA) staining, 30F1, is important, as other monoclonal antibodies that rec-
ognize HSA do not resolve high from low-level expression as well. The cells expressing low levels of CD24/HSA, labeled “A”, are CD19− and
is enriched for very early B-lineage precursors but also is contaminated by other cell types that can be detected by staining for AA4, NK1.1,
DX5, and Ly6c; early B-lineage precursors are AA4+NK1.1−DX5−Ly6c−.
Phenotypic Definition in Figure 8.5. Again, these cell populations can be isolated
Further clarification of the heterogeneity in bone marrow and short-term culture used to determine their order in
can be achieved by analysis using fluorescent staining re- the pathway. Alternative approaches based on other devel-
agents and either microscopic or flow cytometric analysis. opmental markers can be correlated with this framework
For example, the earliest determination that there were of cell stages, notably the system developed by Melchers’
both heavy chain surface-positive B cell and cytoplasmic- group137 using expression of CD45R/B220, CD19, c-KIT, and
positive pre-B cells was through microscopic examination the IL-2Rα chain. A diagram summarizing this type of phe-
using anti-Ig staining.132,133 Later studies in mouse showed notypic subdivision and relating different nomenclatures is
that there were specific surface proteins or “markers” that shown in Figure 8.6.
could be useful in identifying these populations, notably
a restricted isoform (CD45Ra) of the common leukocyte Culture Systems and Critical
antigen, CD45.134 This largely B-lineage–restricted 200 kDa Microenvironmental Interactions
molecular mass isoform is often referred to as “B220.” Some The combination of phenotypic characterization coupled with
highly B-lineage–restricted monoclonal antibodies, such as analysis of growth and differentiation in culture has provided
RA3-6B2, recognize a specific glycosylation of the CD45Ra a powerful approach for the further understanding of B-cell
isoform.135 However, as described previously, even highly development, as employed by many different investigators.
specific antibodies such as 6B2 may also recognize other cell Bone marrow cultures developed by Whitlock and Witte,138–140
types, such as particular differentiation stages or subsets of and fetal liver cultures developed by Melchers’ group141,142 have
NK or dendritic cells. allowed determination of the critical cytokines and some of
The application of multiparameter/multicolor flow cy- the cell adhesion molecules important in the in vivo develop-
tometry and additional monoclonal antibodies specific for ment of these cells. Many of these are summarized in Table 8.1.
other cell surface proteins differentially expressed during A typical B-lineage colony proliferating on S17 stromal cells in
B-lineage development has facilitated delineation of mul- the presence of IL-7 is shown in Figure 8.7.
tiple additional intermediate stages in this pathway.129 For Survival and growth of the earliest stages of develop-
example, the B220 + population in bone marrow can be ing B-lineage cells require cell contact with nonlymphoid
further fractionated into an earlier subset expressing CD43 adherent cells that can be isolated from bone marrow, cells
(about 3% to 5% of marrow cells) and a later fraction with referred to generically as “stromal cells.” A number of lines
much lower CD43 expression (20% to 30% of marrow). The have been derived from primary cultures of bone marrow
precursor/progeny relationship of cells in these two frac- adherent cells and characterized in terms of their capacity
tions can be readily demonstrated by short-term culture, to support B lymphopoiesis in vitro.143 This work has led
with CD43 + cells giving rise to CD43− cells. These two pop- to the discovery of adhesion molecules that play important
ulations can be further subfractionated based on additional roles in mediating the organization of clusters of developing
developmentally regulated surface proteins, such as CD24/ B-lineage cells on stromal layers, including CD44 interact-
heat stable antigen (HSA), BP-1 (a zinc-dependent cell suring with hyaluronate and VLA-4 interacting with vascular
face metallopeptidase also known as aminopeptidase A136), cell adhesion molecule 1.126,127,144,145 Both of these interac-
and the surface Ig molecules IgM and IgD.129 This is shown tions could be disrupted by addition of blocking antibodies

FIG. 8.6. Diagram of Distinct Phenotypic Stages and Characterization of Terminal Deoxynucleotidyl Transferase, Biphasic Rag Expression,
Immunoglobulin-`/a, and Surrogate Light Chain Expression. Genes characteristic of myeloid and T-cell lineage are also shown. The cell
type descriptions are cross-referenced to the alphabetic phenotypic fraction nomenclature and also to the Basel nomenclature. The early
lymphoid progenitor population is identified by activation of the Rag-1 locus in a green fluorescent protein reporter mouse.
to CD44 and VLA-4 on B-cell precursors, resulting in a dis- and differentiation; the most important of these for mouse
ruption of normal pre-B proliferation in vitro.146 Such ad- B-cell development is IL-7.105,125,148,149 IL-7 promotes the sur-
hesion interactions may serve to transmit signals directly vival and proliferation of pro-B and pre-B stage cells, both
to the stromal cells or B precursors, or both. There is some in vivo and in vitro.150,151 Neutralizing antibody to IL-7 can
evidence that stromal cells are induced to elaborate specific block B-cell development in vitro,129 and IL-7 expressed as a
growth mediators after interaction with B-cell precursors or transgene can deregulate normal B-cell development, lead-
soluble regulators.147 ing to B-cell lymphadenopathy.152 The IL-7 receptor con-
Another function of the stromal cells is to produce sists of a unique IL-7Rα chain153 paired with the common
growth factors critical to B-lineage survival, proliferation, gamma chain (γc) that is also found in the receptors for
TABLE 8.1 Regulators of Growth of Early B-Lineage Cells

Mediator Effect Reference
105,148,149,590
L-7 Stimulates CLP and B-precursor proliferation
156,157
TSLP Alternate IL-7–like cytokine
185
IGF-1 Stimulates accumulation of Cμ+ cells in culture
168,172,591,592
FLT-3/FLK2-L Critical for earliest stages of B-lineage development
26,166
c-KIT-L Synergizes in IL-7–induced proliferation
176
IL-3 Substitute for IL-7 in proliferation of pre-B clones
194,196,198,199
CXCL12/CXCR4 Crucial chemokine interaction for early B-lineage precursors
187
Hemokinin Novel regulator of B lymphopoiesis
126,127,593
VLA-4/VCAM-1 Adhesive interaction; antibodies to either block B lymphopoiesis
144,146
CD44/hyaluronate Adhesive interaction; mediates association of B-lineage/stromal cells
105
TGF-β Inhibits proliferation stimulated by IL-7
594,595
Sex steroids Decrease B-lineage precursors in bone marrow
188,189,596
Growth hormones Required for normal B lymphopoiesis
190
TLRs Innate immune system regulation
CD, cluster of differentiation; CLP, common lymphoid progenitor; IL, interleukin; TGF, transforming growth factor; TLR, toll-like receptor; TSLP, thymic stromal lymphopoietin; VCAM,
vascular cell adhesion molecule.

marrow.162 When TSLP is overexpressed from an inducible

transgene, B1 B cells expand, apparently at the expense of
marginal zone (MZ) B cells.163
The earliest precursors in the B-lineage pathway, prob-
ably including cells that are not B-lineage committed but
that can efficiently give rise to B cells in a short time in
vitro, have receptors for SCF/c-KIT-ligand26,27,164–167 and
FLK2/FLT3-ligand.168–172 Thus, the most permissive cultures
for expanding precursors of B-lineage cells will include
these cytokines, in addition to IL-7 and a stromal adher-
ent cell layer, such as S17.173–175 IL-3 has also occasionally
been suggested as playing a support role for pre-B cells in
vitro,176 although its role in vivo may be at a much earlier
stage. While culture conditions have been reported that can
support B-lineage development in the absence of stromal
cells,177,178 the clear-cut alteration of contact-dependence
prior and post heavy chain expression128–130,179,180 argues that
the most physiologic model for early B-lineage growth will
include stromal cells. Besides providing important cell–cell
contacts that may signal survival, proliferation, and differ-
entiation, it is also likely that stromal cells bind at least some
cytokines to their surface, providing higher local concentra-
tions to the clusters of B-lineage precursors that adhere.181,182
B-lineage development may be modified by exposure to
hormones; considerable interest has focused on sex steroids
released during pregnancy that serve to depress B lympho-
poiesis, particularly the pre–B-cell pool.183 This may be
important to avoid autoimmune responses by the mother
but could have negative consequences due to possible tran-
sient immunodeficiency. Interestingly, fetal B lymphopoiesis
FIG. 8.7. Photomicrographs of B-Lineage Colony Proliferating on is not similarly depressed due to the absence of hormone re-
S17 Stromal Layer (in the Well of a 96-Well Microplate) in the ceptors on fetal B-lineage cells.184 Insulin-like growth fac-
Presence of Interleukin-7. Day 10 colony derived from a single Fr. tor has been reported to potentiate progression in vitro to
A phenotype (see Fig. 8.5) cell. Low power and high power views. the C μ + stage,185,186 and, more recently, there is a report of
All of these cells now express CD19 and many have progressed to a bioactive peptide, a type of tachykinin, that synergizes
BP-1+. with IL-7 to enhance the growth of IL-7–dependent cul-
tures.187 Besides insulin-like growth factor, other pituitary
hormones, thyroxine, and growth hormone have effects on
IL-2, IL-4, IL-9, IL-15, and IL-21.154 IL-7Rα null mice have a B lymphopoiesis.188 For example, thyroxine treatment can
severe deficit of both B and T cells in the periphery and lack restore normal B-cell development in dwarf Pit-1 mutant
most B-lineage cells in bone marrow.155 Mice with targeted mice with deficient pituitary function.189 Recent work has
inactivation of the γc or IL-7 do have some B-lineage de- highlighted the effect that activation of TLRs during infec-
velopment, suggesting an alternate cytokine; this appears to tion may have on altering development.190 Thus, it is likely
be thymic stromal lymphopoietin (TSLP). This protein was that more detail remains to be fi lled in to complete our pic-
first identified as a pre–B-cell growth factor produced by a ture of the growth requirements and modulating influences
thymic stromal line156 and shows some of the same effects of B-lineage cells in mouse bone marrow.
in culture as IL-7, although possibly inducing less prolif- Another function of cell–cell interaction is cell fate
eration and more differentiation.157 Its receptor has been determination during the lineage commitment stage, very
cloned; it consists of two chains, the TSLP receptor and IL- early in development of B-lineage cells. The Notch sig-
7Rα .158 The TSLP receptor shares both sequence homology naling pathway is implicated in cell fate determination in
and genomic exon organization with the common gamma invertebrates and more recently has been shown to func-
chain.159 Signaling through the IL-7 receptor requires JAK3 tion in lymphoid lineage specification.45,76,191 Notch family
and activates the transcription factor STAT5, whereas sig- transmembrane receptors regulate transcription by being
naling through TSLP is JAK3 independent but also activates cleaved upon ligand binding to release an intracellular
STAT5.157,160 Unexpectedly, the growth response to TSLP cytoplasmic domain that translocates to the nucleus where
requires synergy with the pre-BCR in bone marrow but it interacts with the transcriptional repressor CSL.192 Recent
not in fetal liver,161 leading some to propose that this might studies have shown that Notch1 can play a pivotal role in
be a marker for distinctive B1 B-cell development in bone commitment of common lymphoid progenitors to the T-cell

lineage.76 That is, expression of Notch1 by retroviral trans- The extent of heavy chain or light chain rearrangement
duction has been shown to redirect B-lineage differentiation can be quantitated either in bulk isolated populations129
in bone marrow along the T lineage. Furthermore, a recip- or in individual cells.137,209,210 At the heavy chain locus, D-J
rocal result was found in conditional Notch1 null mice, rearrangement occurs prior to V-DJ rearrangement, and
blocking T-cell development in the thymus to be replaced by cells with extensive D-J but little V-DJ rearrangement can
B-cell development.97 Finally, altering the Notch1 modifier be detected at the B220 + CD43 + CD19 − (Fr. A) stage, where
lunatic fringe by overexpressing this molecule under regula- Rag-1/2 and TdT are strongly expressed.44 V-DJ rearrange-
tion of an lck promoter resulted in B-cell development in ments are readily detected in the abundant B220 + CD43 − (Fr.
the thymus.80 Differentiation of lymphoid precursors to D) stage small pre-B cells, although productive rearrange-
NK or dendritic cell lineages was unaffected in Notch1 null ment has already completed by the large pre-B (Fr. C-prime)
CLP cells, so Notch apparently affects only the B/T-lineage stage (see following discussion). Single-cell sequence analysis
decision. of rearrangements in Fr. C stage cells shows a large proportion
with nonproductive rearrangements on both chromosomes,
Role of Chemokines in Migration of B-Cell Precursors suggesting that this may represent a dead-end fraction.209
One of the most distinctive features of B-cell development Some light chain rearrangement is detectable in early stage
in bone marrow is the migration of developing precursors B220 + CD43 + (Fr. B) cells, and this is consistent with the ob-
from early stages nearest the bone endosteum layer to lat- servation that low-level kappa light chain rearrangement is
ter stages progressively closer to the central arteriole, where detectable in bone marrow of mice where μ heavy chain has
they will eventually exit.193 This migration is likely due to been crippled by deletion of the membrane exon.211 However,
differential expression of specific adhesion molecules and much higher levels of kappa rearrangement can be detected
also to expression of chemokine receptors. Analysis of B-cell in B220 + CD43 − (Fr. D) stage cells consistent with the finding
migration has identified a critical chemokine that is impor- of sterile kappa mRNA increase just prior to this stage likely
tant in this process, SDF-1, now known as CXCL12,194,195 and induced by pre-BCR signaling (see following discussion).
its receptor CXCR4.196 CXCL12 is expressed by fetal liver and
bone marrow stromal cells, whereas CXCR4 is found on he- Role of Immunoglobulin Heavy Chain and
matopoietic precursors and B-cell progenitors.197 Deletion the Pre–B-Cell Receptor
of either the receptor or ligand results in severely impaired Many years ago, the analysis of severe combined immunode-
B lymphopoiesis.198–200 Interestingly, the critical defect ap- ficiency (SCID) mouse212 bone marrow revealed the presence
pears to be failure to retain precursors in the primary lym- of a population of B220 + cells, all with a very early CD43 +
phoid organ, as progenitors and precursors can be found in phenotype, suggesting a block in B-cell development at this
the blood of mutant mice.201 stage.213 SCID mice have a defect in the catalytic subunit of
the DNA-dependent protein kinase DNA-PKcs,214,215 and, as
Gene Expression and Immunoglobulin Rearrangement a result, B-lineage cells in these mice are very ineffective at
In addition to delineation of developmental stages based on completing productive Ig heavy chain rearrangements. This
changes in protein surface expression, B-lineage cells can also block could be overcome by introduction of an Ig heavy
be characterized for expression of internal proteins related chain transgene, indicating a critical role for μ protein in
to critical processes in their progression along this pathway, progressing past an early developmental checkpoint.216
specifically those related to rearrangement and expression Furthermore, a gene targeting experiment that eliminated
of the B-cell antigen receptor (Fig. 8.8). Thus, expression of the membrane exon of μ heavy chain (μ-mt) also generated
μ heavy chain constant region, prior to Ig rearrangement, a block at this stage.217,218
from a cryptic promoter generates a “sterile transcript” The μ heavy chain is associated with a set of B-cell–spe-
that reflects an open chromatin structure important for cific peptides at the early pre–B-cell stage,219 and this com-
the onset of rearrangement,202–204 and so analysis of sterile plex is referred to as the pre-BCR. It seems clear that pre-
μ expression can be used to investigate very early stages of BCR mediates a type of signaling function analogous to
B-cell development. Classical northern analysis can be done the BCR in mature B cells. Prior to light chain expression,
with transformed lines, but much work analyzing RNA levels two peptides known as λ5 and VpreB, originally isolated
in B-lineage cell fractions, whether directly isolated or cul- as B-lineage–specific complementary DNAs,220,221 associate
tured, has depended on polymerase chain reaction ampli- with heavy chain. The λ5 shows homology to a lambda con-
fication of complementary DNA.38 For example, using this stant region and VpreB is so-called because it has homol-
approach, sterile μ can be detected in a very early fraction ogy to a variable region domain. Together, these peptides
of B220 + CD43 + CD19 − (Fr. A) cells. Expression of the reconstitute a pseudo- or surrogate light chain (SLC). The
combinase activating genes Rag-1 and Rag-2, which together critical role of λ5 was demonstrated unambiguously in gene-
make the double-strand breaks in DNA required for Ig rear- targeted mice, where B-cell development was blocked at the
rangement,205–207 also occurs in Fr. A stage cells, which also B220 + CD43 + stage.222 The production of some mature cells
have high levels of TdT, the enzyme responsible for adding that accumulate in this mutant is likely due to early kappa
nontemplated nucleotides at the D-J and V-D junctions of rearrangement, with light chain substituting for SLC, as
the heavy chain.12,13 Rag-1 binds in a highly specific fashion demonstrated in light chain transgenic experiments.223
to discrete sites within the IgH locus, recombination signal The μ heavy chain has a very short cytoplasmic region
sequences, defining “recombination centers.”208 consisting of only three amino acids; signal transduction

CHAPTER 8
FIG. 8.8. A: Profiles of immunoglobulin (Ig) rearrangement–related gene expression. Cells isolated using fractionation scheme in Figure 8.3 for multilineage progenitors and Fr. A; in Figure 8.5
for Fr. B through Fr. F. Relative messenger ribonucleic acid levels assessed by performing semiquantitative reverse transcription-polymerase chain reaction using limited numbers of cycles,
blotting, then probing, and quantitating the probe signal. Note the biphasic expression of Rag genes and the early expression of sterile μ (labeled m0) during the first wave, where heavy chain
rearranges, and the upregulation of sterile kappa transcripts (k0) during the second Rag wave, when most light chain rearrangement takes place. B: Single-cell polymerase chain reaction
analysis of Ig heavy chain rearrangement in four early stages of developing B-lineage cells. Deoxyribonucleic acid prepared from individual cells isolated following the scheme shown in
Figure 8.3 was divided into two aliquots and analyzed for retention of a deoxyribonucleic acid segment lost upon any D-J rearrangement (labeled GL, germline) and also for D-J rearrange-
ment. Note that D-J rearrangement initiates at the common lymphoid progenitor stage, where 30% to 50% of cells show a rearrangement.
|
227
9/17/12 5:28 AM
through the BCR is mediated by accessory peptides, similar show a block in development at the CD43 + stage that can
to the CD3 components of the T-cell receptor, known as Ig- be complemented by introduction of a functionally rear-
α and Ig-β.224–227 Inactivation of Ig-β228 results in a block at ranged μ heavy chain as a transgene180 (Fig. 8.9C). Analysis
the B220 + CD43 + stage in mouse bone marrow, similar to of several types of pre-BCR defective mutant mice shows
that seen in μ-mt and λ5 null mice. Finally, the Syc tyrosine a complete absence of Fr. C-prime stage cells, suggest-
kinase plays a critical role in transducing BCR cross-linking ing that pre-BCR signaling results in the upregulation of
signals in mature B cells and inactivation of this gene results CD24/HSA and also entry into rapid cell proliferation.
in a “leaky” block at this same stage.229,230 Thus, any muta- Thus, a model of pre-BCR function is that it signals the
tion that affects this pre-BCR complex (see following sec- clonal expansion phase of pre–B-cell development, ampli-
tion; see Fig. 8.10A) precludes efficient progression past the fying cells with in-frame VDJ rearrangements capable of
earliest stages of B-cell development. making heavy chain protein.
Careful examination of B-cell development in normal The precise nature of pre-BCR signaling remains to be
mice shows that heavy chain is fi rst expressed at a late completely understood. An early model suggested that
fraction of the B220 + CD43 + stage, termed Fr. C-prime cross-linking of heavy chain was mediated through interac-
(Fig. 8.9A). This fraction is also interesting because it tion of SLC with a bone marrow expressed ligand. However,
shows a much higher proportion of cells in cycle (revealed subsequent experiments showed that normal light chain
by a high frequency of cells with greater than 2N DNA could substitute for SLC, and that even a V H truncated μ
content; Fig. 8.9B), compared with any other B220 + stage heavy chain could mediate progression past this stage.
in bone marrow.129 Mice unable to assemble a pre-BCR, Furthermore, intensive searches for the putative ligand over
due to inability to rearrange heavy chain (Rag-1 null), a 10-year period have been fruitless, leading to the model
FIG 8.9. A: Western blot of immunoglobulin μ heavy chain expression showing high-level expression in Fr. C-prime. B: Cell cycle analysis of
individual fractions shows most cells in Fr. C-prime are cycling. Propidium iodide staining of permeabilized sorted cells allows determination
of deoxyribonucleic acid content per cell using flow cytometry. C: Block in B-cell development in Rag-1–deficient mice can be overcome by
introduction of an immunoglobulin μ heavy chain transgene.

thought to modulate BCR signaling strength,243 and this is

probably also the case for pre-BCR signaling, allowing only
strongly signaling pre-BCRs to progress in the mutant mice.
BLNK/SLP65 serves to link the pre-BCR to the Syk kinase
critical in BCR signaling.244,245 Mutant mice lacking BLNK
show a partial block in B-cell development at the pro-B to
pre-B transition.246 Curiously, whereas pre-BCR signaling is
thought to mediate allelic exclusion (expression of a single
heavy chain allele), this remains intact in BLNK-deficient
mice.247 Animals deficient in both BLNK and Btk develop
an extensive pre-B expansion that progresses to lymphoma,
leading to study of such mice as a model for human pre-B
acute lymphoblastic leukemia (ALL).248,249
Syk-deficient mice show a more severe block at the pro-B
to pre-B transition and a lack of allelic exclusion.229,230,247
There is also evidence that pre-BCR signals through Erk to
induce proliferation.250
Outcomes of pre-BCR signaling, in addition to pre-B
proliferation, are downregulation of the Rag genes,251 down-
regulation of TdT,235 and transcriptional activation of the
kappa locus, detected as upregulation of sterile kappa tran-
scripts.252 A control element for regulating Rag expression
has been identified.253,254 Extinction of recombinase activity
FIG 8.10. A: Diagram of the pre–B-cell receptor (BCR) μ heavy chain is probably important for chromosomal stability during the
with surrogate light chain (λ5 and VpreB) in place of conventional clonal burst period of B-cell development255–257 and is also at
light chain. As in the BCR, immunoglobulin (Ig)-α and Ig-β serve to least a part of the mechanism that assures allelic exclusion,
couple signals between the receptor and cytoplasmic components, the expression of a single heavy chain by any given B cell.258
such as BLNK and Syk. Starred m transmembrane residues are im- There is evidence that pre-BCR selection requires low levels
portant in mediating interaction with Ig-α/Ig-β, as mutation of these
diminishes BCR function. B: Clonal expansion mediated by pre-BCR of IL-7259 and probably occurs naturally as the developing
assembly. Association of newly generated μ heavy chain with pre- precursors migrate through different stromal cell microen-
existing surrogate light chain leads to a burst of proliferation at the vironments in bone marrow.
pre-B stage. The function of the pre-BCR may be more complex than
simply to sense whether an in-frame VDJ rearrangement has
occurred. This possibility is suggested by the observation
that pre-BCR signaling is more akin to “tonic” signaling in that heavy chains with different VDJ segments vary in their
mature B cells.231–233 That is, simple assembly of the complex capacity to assembly with SLC components.260–263 V regions
(or possibly some degree of multimerization fostered by the are classified into families based on sequence homology, and
self-aggregating nature of SLC234) probably is sufficient for many members of two of these families, the 7183 and Q52,
the cell to pass this developmental checkpoint (Fig. 8.10B). appear to frequently generate heavy chains that assemble
One clear-cut finding is that pre-BCR signaling in a trans- poorly with SLC.262 A consequence of this will likely be
formed pro–B-cell model system can occur in the absence poor pre-BCR signaling and little clonal expansion; such
of any additional cell type, suggesting that if a ligand exists, cells will become underrepresented at later stages of B-cell
it must be expressed on B-lineage cells rather than stromal development relative to cells containing heavy chains that
cells.235 Possibly, pre-BCR homodimerization is mediated signal effectively. One explanation for this SLC assembly-
through glycosylation at a conserved asparagine residue in mediated clonal expansion is that it serves a quality control
the first μ constant region domain.236 Studies of the T-cell function to test heavy chain V regions for their potential to
analog of the pre-BCR, pre-Tα , provide strong evidence fold with real Ig light chain, a critical requirement if the cell
that it signals through spontaneous dimerization, without is to express a complete BCR. An alternative (not necessarily
requirement for an external ligand.237–239 mutually exclusive) explanation is that making pre–B-cell
Mutations in other molecules in the pre-BCR signaling proliferation dependent on pre-BCR expression provides a
pathway have been shown to affect B-cell development and simple mechanism for regulating the extent of clonal expan-
pre–B-cell clonal expansion. Although the Btk mutation sion, as an immediate consequence of pre-BCR signaling is
is less severe in mouse than in human, there nevertheless to terminate SLC expression. Thereafter, SLC protein levels
is an alteration in pre–B-cell expansion in Btk-deficient decay and are diluted by cell division so that after several
mice.240 Also, X-linked immunodeficiency (xid) B cells (de- rounds of proliferation, pre-BCR levels will decrease to below
ficient in Btk) have been reported to proliferate more in the threshold required to provide the signal to maintain the
stromal cell cultures, possibly due to decreased differentia- cell in cycle. Figure 8.11 illustrates a model for bone marrow
tion to later nonproliferative stages.241,242 The role of Btk is B-cell development, showing the pre-BCR checkpoint.

FIG. 8.11. Model of Mouse Bone Marrow B-Cell Development Showing Relationship of Immunoglobulin Rearrangement with
Progression and Proliferation.
One of the most striking examples of pre-BCR selection productive/nonproductive V H81X does not decrease during
is seen with the D-proximal V H gene, V H81X, where early cultures of fetal precursors.272,273 Furthermore, the prolifera-
precursors show biased overutilization due to preferential tive burst that pre-BCR assembly provides to bone marrow
rearrangement of this V H gene.11,264,265 In fact, regulatory pre-B cells may instead result in exit from cell cycle in fetal
sequences interspersed throughout the distal VH region, precursors,263 leading to selection of very different BCR rep-
termed PAX5-activated intergenic repeat elements, appear ertoires during fetal and adult B lymphopoiesis. The possible
to enhance utilization of distal J558 family VH genes.266 significance of this is discussed subsequently in the section
Furthermore, a recombination regulatory region, the inter- on B1 B cells. Leaving aside such developmental difference,
genic control region 1, which lies between the VH and D the role of the pre-BCR in B-cell selection remains subject to
clusters, has been shown to inhibit proximal and promote debate, with evidence suggesting it functions to eliminate274
distal VH gene utilization.267 or enrich275 self-reactivity.
Curiously, although it is abundant in early B-cell pre-
cursors, V H81X is rarely seen in the mature B-cell compart- Light Chain Rearrangement and Generation of
ment. The demonstration of the decrease in representation Immature B Cells
of cells with V H81X rearrangements at the pre-B clonal Besides termination of TdT and SLC gene expression, pre-
expansion phase in bone marrow,268,269 together with the BCR signaling also results in the downregulation of Rag-1
fi nding that heavy chains utilizing VH81X frequently fail and Rag-2 expression, mediated by activity of Gfi1b on Erag,
to assemble functional pre-BCRs, 260,261,270 explained this the B cell-specific Rag regulatory element,254 and the repres-
paradox. Models of pre-BCR three-dimensional structures sion of FoxO1.276 It appears that induction of Bcl-6 by pre-
based on x-ray crystallography may help to explain varia- BCR signaling functions in this progression by repressing
tion in the ability of different heavy chains to assemble Myc and Ccnd2, thereby promoting exit from cell cycle.277,278
with SLC.271 As mentioned previously, there is evidence that the pre-BCR
However, it is still puzzling that the most frequently rear- signals through ERK and this Ras–MEK–ERK signaling acts
ranged V H gene is so strongly selected against at the clonal to silence transcription of Ccd3, encoding the cell cycle pro-
expansion stage. A possible explanation may lie in compari- tein cyclin D3, thus promoting cell quiescence.279 Opposing
sons of V H utilization during fetal development. That is, in cell cycle exit, IL-7R signaling activates STAT5, maintain-
contrast with bone marrow precursor cultures, the ratio of ing Ccd3 expression. Eventually, the pre-B cell escapes from

IL-7 signaling, exits the cell cycle, and the Rag genes are formed cells in these cultures, and possibly also in vivo, is
reexpressed at high levels. Sterile kappa transcripts become due to their low level of antiapoptotic mediators. Both Bcl-2
detectable during the cycling stage, likely reflecting chro- and Bcl-X L mRNA are present at only very low levels in these
matin remodeling to make the kappa light chain locus ac- cells, in contrast with other B-lineage stages where either one
cessible,252 so induction of the recombinase machinery can or the other predominates. Overexpression of Bcl-X L from a
initiate kappa light chain V to J rearrangement. transgene results in accumulation of a population of pro-B
An interesting feature of the Vκ locus is that the ap- phenotype cells with nonfunctional rearrangements,291 im-
proximately 100 genes are in both transcriptional orienta- plicating this protein in the process of pre-BCR selection of
tions, and so these genes can rearrange either by deletion cells with functional rearrangements.
(generating an extrachromosomal excision circle) or by
inversion.280 The absence of intervening D segments also
means that it is possible for upstream V kappa genes to Peripheral Maturation Stages and Functional Subsets
rearrange to downstream J kappa segments, “leapfrogging” Transitional B Cells
the initial rearrangement, assuming it was to any Jκ other Newly formed immature B cells migrate to the spleen where
than Jκ5. The successive association of different kappa they either die or undergo further maturation to a mature
chains with the same heavy chain in a B cell is referred to as B cell. These maturing B cells can be subdivided based on
BCR “editing” and was originally observed in the context of differential expression of several surface proteins, including
autoreactivity, which maintains Rag expression even at the CD21, CD23, CD24/HSA, and AA4.1. These subdivisions
B-cell stage281,282 (see following section on B-cell tolerance). have been referred to as “transitional B cells.”292 One recent
Because assembly and expression of a complete BCR (that is subdivision based on CD21, CD23, AA4.1, and IgM levels has
not self-reactive) terminates Rag expression, an additional shown progression from an AA4.1+ CD21− CD23 − T1 stage
reason for light chain editing in the bone marrow may be to an AA4.1+ CD21− CD23 + T2 stage, followed by down-
to replace an initial light chain that fails to assemble effec- regulation of IgM as a T3 stage, and finally loss of AA4.1
tively with the particular heavy chain present in that pre-B with upregulation of CD21 to yield the mature follicular
cell. This is probably the explanation for multiple light chain phenotype.289 As shown in Figure 8.12, this approach also
rearrangements detected in single early B-lineage cells.283 resolves two AA4.1− subsets that lack CD23, the B-1 subset
Complete failure of kappa rearrangement, possibly after with low CD21 and the MZ subset with very high CD21
receptor editing to avoid self-reactivity, leads the pre-B cell (see following). The transitional stage cells are all short-lived
to a second phase of light chain rearrangement, dependent as shown by bromodeoxyuridine incorporation.289,293 They
on IKK-mediated NF-κ B signaling, where the λ light chain are also not functionally competent, as shown by inability
locus rearranges.284 to proliferate after BCR cross-linking.287,294 Another well-
Newly formed B cells can be distinguished from mature characterized functional distinction is that B-cell tolerance,
B cells on the basis of their inability to proliferate in re- rather than an immune response, is induced by BCR cross-
sponse to BCR cross-linking (ie, they are functionally linking of immature B cells.295–297 More recent studies with
immature). This is also the stage where negative selection transgenic models of self-reactivity have shown that these
is reported in transgenic models of autoreactivity.285,286 Cells B cells can be deleted, undergo receptor editing, or rendered
at this stage have a short half-life of only a few days, com- functionally unresponsive (anergic) by BCR signaling at the
pared to mature follicular B cells with a half-life measured immature stage.281,285,298–302 One group has suggested that the
in months. They can be distinguished by surface phenotype “T3” stage is not an intermediate in production of follicu-
from other B cells based on expression of certain combina- lar B cells, but rather a population of autoreactive anergic
tions of markers, such as IgM + IgD −, absence of CD23, and cells,303 so additional work needs to be done to establish the
high-level expression of CD24/HSA.287 Recently, there have identities of all the B-cell subpopulations present in spleen.
been reports of single markers that are useful in distinguish- For example, careful comparison of T1 and T2 B-cell popu-
ing newly formed cells from any mature subset, such as the lations shows that a fraction of T2 cells do respond to cer-
molecules recognized by monoclonal antibodies 493288 and tain microenvironmental stimuli and include a portion
AA4.1.289 The AA4.1 target molecule has been cloned and of cycling cells that appear to constitute a branchpoint in
identified as the mouse ortholog of a component of the generation of MZ B cells.304
human C1q receptor.290 It is not simply the inability to receive T-cell help due
Cells similar to newly formed B cells can be generated to differences in microenvironment or receptor expression
to varying extent during stromal cell culture of B-cell pre- that makes immature B cells incapable of responding as
cursors, although the more primitive cycling pre-B or pro-B mature B cells. Different from the result with mature B cells,
cells are usually more abundant and tend to increase in fre- cross-linking the BCR on immature B cells has been shown
quency with prolonged culture. It is possible to induce dif- to induce apoptosis, suggesting distinctions in the signal-
ferentiation of B cells in these cultures by withdrawing IL-7, ing pathways between these two stages.305,306 Studies with
which induces a wave of small pre-B and then newly formed transgenic mice suggest that this apoptosis is not mediated
B-cell generation. Such cells do not persist for more than through the Fas/Fas-ligand pathway, as central deletion
a day, unless the cultures are established from Bcl-2 trans- is intact in Fas-mutant mice.307,308 Prior to induction of
genic mice,131 suggesting that the short half-life of newly apoptosis, immature B cells have been shown to complete

FIG 8.12. Flow Cytometry Approach for Resolution of Transitional (T1–T3) and Mature (Follicular, B1, Marginal Zone) Populations of B Cells
in Mouse Spleen. Left panels show the distribution of AA4.1 and cluster of differentiation (CD)23 on B cells (defined as CD19+immunoglobulin
[Ig] M+). Cells in the boxed regions are then analyzed for correlated expression of CD21 and IgM, facilitating resolution of three AA4.1+
fractions (T1–T3), the follicular subset (AA4.1−CD23+CD21+), the marginal zone subset (AA4.1−CD23−CD21++), and the B-1 subset (AA4.1−
CD23−CD21low).
some of the early events associated with entry into cell cycle likely plays a role at several stages of B-cell development and
while failing to complete this program.309 Distinct stages in activation, which complicates the analysis, but it appears
maturation appear more or less capable of responding to clear that one major consequence is altered BCR signaling
BCR cross-linking by reinduction (or maintenance) of Rags that has a profound effect on progression through the vari-
to facilitate receptor editing.286 Furthermore, it appears that ous transitional stages in spleen. A likely consequence of
immature B cells are more sensitive to smaller changes in diminished strength of BCR signaling is a compensatory
intracellular free calcium, compared to mature B cells.310 requirement for higher surface BCR expression that even-
It is possible that the capacity to upregulate antiapoptotic tually produces at decreased frequency a type of “mature”
molecules, such as A1, may play a critical role in the inabil- B cell that is still functionally handicapped.313,314 Several
ity of immature B cells to survive and complete a normal groups have produced xid mice on a nu/nu T-cell less back-
response.311,312 The characterization of signaling pathways ground that results in a more profound absence of mature
in different immature stages of developing B cells is ongo- B cells, suggesting a requirement for T cells or T cell–pro-
ing and should eventually provide insights into the detailed duced factors in the maturation of xid B cells.315,316 A more
mechanism for immature B-cell tolerance. recent variation of this type of investigation is the produc-
Analyses of various normally occurring or engineered tion of xid/CD40 -deficient mice that show a similar deficit
mutant mice have provided approaches for investigation of in mature B cells, suggesting a role for the CD40/CD40L
the process of progressing from a newly formed B cell to a interaction in the generation of mature B cells from transi-
mature follicular B cell. B-cell populations and B-cell functional B cells, particularly when the BCR is handicapped by
tion has been studied for many years in CBA/N mice bear- defective Btk.317
ing the xid mutation. This mouse has a mutation in the Btk Lyn is a Src family protein kinase that is associated with
gene that produces a milder phenotype than the complete the BCR and functions in signaling in mature B cells.318 Lyn-
absence of peripheral B cells seen in humans. The Btk gene deficient mice exhibit defects in maturation of immature cells,

suggesting a positive role for BCR/Lyn signaling at this stage, selection.” It is interesting to note that ablation of heavy chain
but these mice also develop a severe autoimmune condition, from immature B cell, thereby eliminating any possibility of
suggesting an additional negative regulatory role for Lyn in pre-BCR or BCR tonic signaling, results in a “reversion” of
maintaining tolerance in mature B cells.319,320 Cross-linking cell phenotype to an earlier developmental stage.334 Finally,
of the BCR on T2 and mature B cells results in c-Rel activa- the maintenance of follicular B cells has been found to de-
tion that upregulates antiapoptotic genes and the prosurvival pend on the function of the c-Myb transcription factor due
BAFF receptor.321 Furthermore, migration of early bone mar- to its role in BAFF signaling335 (see following section).
row emigrants to areas where T1 and T2 cell mature has been The repertoire of the follicular B-cell pool appears to dif-
shown to depend on chemokine signaling that is disrupted fer from the earlier immature splenic B-cell population, as
in Rac1/Rac2 double-deficient mice.322 Finally, progression assessed by sequence analysis of the light chain repertoire in
of transitional cells to the mature B-cell stage is dependent heavy chain transgenic mice.336 The approach of fi xing the
on “tonic” BCR signaling, as mice engineered to express low- heavy chain and then examining the light chain repertoire
surface BCR levels, and, thereby lower tonic BCR signaling simplifies the analysis, and the results of this study were
showed decreased generation of mature B cells.323 Analysis of interpreted to indicate that BCR-mediated antigen selection
these mice suggested that maturation was dependent on acti- is indeed operating. However, the resolution of the analy-
vation of ERK via a signaling pathway requiring Ras. sis probably could not have rigorously excluded populations
CD72 is a predominantly B-lineage restricted C-type known to show V gene biases, such as B-1 or MZ B cells
lectin, and ligating this molecule was recognized for many (see following sections), so further work will be required
years (when it was known as Lyb2) as having functional to provide convincing evidence of antigenic selection in the
consequences.324 Recent analyses of a CD72 null mouse follicular B-cell pool.
has clarified its function in B-cell development and activa-
tion.325 CD72 has been shown to recruit SHP-1 to the BCR, B-Cell Migration and Maintenance
supporting a negative regulatory role in BCR activation.326 Newly formed B cells migrate from the bone marrow to the
Consistent with this model for CD72 function, null gene- spleen, undergo further maturation in the red pulp, and
targeted mice have been shown to produce B cells that are eventually enter the follicle where they constitute the mature
hyperresponsive.325 Interestingly, late stages of B-cell devel- B-cell pool that recirculates. Their migration is dependent on
opment are affected, with fewer mature B cells and relatively chemokines/receptor interactions, notably the SLC(CCL21)/
normal numbers of immature B cells in spleen.325 Thus, too CCR7 interaction, as demonstrated by the inability of ma-
intense signaling may also delay maturation of immature ture B cells to be retained normally in spleens of CCR7 null
B cells. mice.337 The role of the CXCR5 receptor on B cells in homing
A number of studies have demonstrated the important to the lymphoid follicle due a gradient of the B-lymphocyte
role that the phosphoinositide 3-kinase (PI3K) signaling chemoattractant CXCL13 is also well known, and CXCL13
pathway plays in the maturation of transitional cells and can directly induce Ltα1β2 on the recruited cells.338 Finally,
generation of mature B cells. Under conditions of normal it is also possible that the SDF1 (CXCL12)/CXCR4 interac-
tonic signaling, PI3K signaling inhibitors block the nor- tion, critical for normal B lymphopoiesis, may also be im-
mal downregulation of Rags in immature B cells and also portant at this later stage, although investigation of this issue
decrease progression past this stage.327 This BCR signal is de- is complicated by the early defect. This is an ongoing area
pendent on CD19 and BCAP, a BCR complex adapter mole- of investigation and may eventually be clarified by develop-
cule,328 and acts through AKT to activate PI3K. The survival mentally regulated gene targeting studies.
of B cells at the mature stage has been shown to depend on Recently, considerable interest has focused on the role
an intact BCR. More recent work revealed that the survival of a tumor necrosis factor family member cytokine known
of BCR-deficient B cells can be rescued by activation of the variously as BAFF, BLyS, TALL-1, zTNF4, or THANK in
PI3K signaling pathway alone.329 the process of peripheral B-cell maturation.339 BAFF is a
tumor necrosis factor family member found to enhance
Follicular B Cells survival of B cells or even produce autoimmunity in trans-
The major population of mature recirculating B cells in the genic mice constitutively expressing it.340,341 Initially, two
spleen is located in the B-cell follicle region, hence the term receptors defined for BAFF, BCMA and TACI, provided a
follicular B cells. Entry into this anatomic site appears to complex picture, as targeted inactivation of BCMA yielded
constitute a final stage in maturation for developing bone no B-cell defect and deletion of TACI had increased B cell
marrow B cells, as competition for this site is compromised numbers (suggesting that TACI might be a negative regu-
in several transgenic models of B-cell tolerance.330–333 Cells lator). This puzzle was resolved by identification of a third
in this compartment do not proliferate but persist in the receptor, BAFF-R/BR3,342,343 which was mutated in a strain
resting state for several months. A conditional knockout of mice known to lack most mature B cells, A/WySnJ.344
study, eliminating expression of the BCR (by deleting the A second ligand, APRIL, can bind to BCMA and TACI, but
V region), revealed that expression of the BCR is required for not to BAFF/BR3, and this binding is proliferative rather
cell survival.232,233 It is not yet established whether this is due than survival promoting.345 Thus, the critical interaction
to “tonic signaling” (simple assembly of the BCR signaling for maintenance of follicular B cells is BAFF/BLyS with its
complex) or instead reflects signaling by low-affinity bind- receptor. B-1 cells are not deficient in A/WySnJ,344 suggesting
ing to cross-reactive self-determinants, a kind of “positive that their maintenance does not depend on this pathway

but instead is more BCR dependent. Interestingly, exces- Germinal Center B Cells
sive levels of BAFF promote survival of highly autoreactive T cell–dependent immune responses usually give rise to
BCR transgenic mouse B cells, consistent with the fi nding anatomically distinctive structures in spleen and lymph
of elevated levels in patients with autoimmune diseases such nodes that are referred to as germinal centers (GCs) and
as systemic lupus erythematosus and Sjögren syndrome.346 contain large numbers of rapidly cycling B cells.354,355 These
cells can be recognized in stained sections of spleen by bind-
Mechanism of B-Cell Receptor Signaling ing of high levels of peanut agglutinin and by the absence
Considering that the structure of the Ig molecule was elu- of IgD.356,357 Many of the B cells with this phenotype have
cidated more than 40 years ago, it is surprising that the downregulated-Bcl-2/upregulated-Fas expression and, in
details of initiation of B-cell activation following antigen the absence of strong BCR signaling, will likely die by
binding are still not firmly established. Recent work using apoptosis.358–361 The termination of IgD expression means
single-molecule imaging demonstrated that antigen binding that surface BCR expression decreases at least 10-fold, and
altered the mobility of BCRs within microclusters on the B so limiting amounts of antigen will favor the cells with in-
cell, resulting in rigid oligomeric structures that generated creased affinity for antigen, generated by a process termed
a signal.347 Interestingly, this oligomerization required the somatic hypermutation. Recently, a miRNA, miR-155, has
membrane proximal domain of IgH-μ (Cμ), and such do- been found to be critical for the generation of GCs by regu-
mains in isolation spontaneously formed aggregates in the lating cytokine production.362
membrane, leading to the hypothesis that antigen ligation A major advance in understating the GC response
induces a conformational change in Cμ4 that then pro- has been the discovery that activation-induced cytidine
motes assembly of signaling clusters. Other work visualiz- deaminase (AID), an RNA-editing enzyme that can induce
ing simple Ig molecules has demonstrated the importance class switch recombination in fibroblasts,363–365 is also a key
of the membrane cytoskeleton in limiting diffusion of the player in the process of hypermutation.366,367 In fact, a third
BCR and revealed the importance of Igb/CD79b in medi- means for Ig gene diversification used in nonmammalian
ating such restriction, showing that alterations in BCR dis- species, V gene conversion, is also dependent on AID.368,369
tribution rapidly result in strong intracellular signaling.348 AID is now thought to be responsible for genetic muta-
Finally, an alternative model has been proposed, based on tions outside the Ig loci in lymphomas that arise from GC
analysis using quantitative bifluorescence complementation, B cells.370
suggesting that preexisting BCR aggregates are less active in The precise mechanism of selection for higher affinity
signaling, and that the disruption of such complexes upon B cells generated by hypermutation of the BCR V regions
antigen binding is responsible for generating a signal.349 remains to be fully understood, but the regulation of pro-
Additional work using such advanced imaging techniques apoptotic and antiapoptotic genes likely plays a major role.
will undoubtedly resolve this apparent contradiction and B cells able to bind antigen with high affinity can present
lead to a more complete understanding of BCR signaling. antigen to specialized CD4 + T cells (follicular helper T cells)
that then signal the B cell through a CD40/CD40L interac-
B-Cell Turnover tion, resulting in the upregulation of Bcl-X L .358–360 Most GC
It is estimated that 10 to 20 million B cells are produced in B cells have sharply downregulated levels of Bcl-2 and up-
bone marrow of the mouse each day,350 yet it appears that regulated levels of Fas,360,361 and so in the absence of rescue
only about 10% of this number reach the periphery.293 Thus, by expression of the alternative antiapoptotic mediator Bcl-
there is considerable loss at this bone marrow emigration X L, cell death by apoptosis will be the fate of most B cells in
stage/spleen entry stage, possibly due to elimination of auto- the GC. Careful regulation of selection is critical to affinity
reactive cells (B-cell tolerance) or to homeostatic regulation. maturation, and elimination of self-reactive cells that poten-
The latter possibility is supported by the observation that tially could be generated during this process must also occur
depletion of the mature B-cell population results in a rela- efficiently to avoid the potential of autoimmune disease.
tively rapid recovery of this pool, suggesting that most of the Recently, the role of the immunoinhibitory receptor
immature B-cell population can enter the mature follicular PD-1, present at high levels on follicular helper T cells, in
subset in this situation.351 GC B-cell selection has been described.371 Furthermore,
Once functionally mature B cells are generated, it has high-resolution imaging of fluorescent reporters is revealing
been difficult to unambiguously determine their half-life, details of cell dynamics within the GC, indicating that com-
although accumulating data from several laboratories using petition for T-cell help is a key limiting factor in the selec-
bromodeoxyuridine labeling has led to the idea that fol- tion of higher affinity B cells.372
licular B cells have a relatively long half-life on the order
of months.293,352 A recent elegant study provided definitive Memory B Cells
confirmation of this by conditional elimination of Rag-2 Memory B cells were initially defined functionally as cells that
expression, allowing termination of B-cell development in could respond rapidly by production of high-affinity antibody
adult mice.353 This study showed that follicular B cells have when challenged in a host reconstituted with B cells and T
a half-life of about 4.5 months. This same analysis showed cells from a primed animal.373–378 Subsequently, such cells have
that two other subsets of B cells, B-1 and MZ B cells, did not been purified based on their antigen-binding properties379 and
diminish over time, consistent with their well-known capac- shown to consist primarily of isotype-switched (IgG +) B cells
ity for self-renewal and lifelong persistence. that continue to express CD45R/B220 and have distinctively

lower levels of cell surface BCR.380 They arise during the encompasses both CD5 + and CD5 − B cells that are IgMhigh,
T cell–dependent immune response, probably only from IgDlow/−, CD23 −, and CD43 +. They constitute a large propor-
follicular B cells, in the GC. They are very long-lived or self- tion of the B cells found in the peritoneal and pleural cavities
regenerating, as cell transfer assays have shown that memory (30% to 50% of B cells, around 106 cells), but are also found
responses can be detected for long periods after the primary in spleen where they are present at numerically similar levels,
immunization.380,381 More recent work has focused on exam- but constitute a much lower proportion of the total B-cell pool
ining the heterogeneity of secondary B-cell subsets, examining (2% of B cells, around 106 cells). The B1 B cells in the peri-
cell surface proteins that may aid in distinguishing them.382 In toneal cavity are also CD11b/Mac-1+ unlike those in spleen.
terms of genes that distinguish memory B cells, it appears that They appear early in ontogeny, representing 30% or more of
phospholipase C gamma 2 is important both for the genera- the B cells in spleen of 1-week-old animals. They also have a
tion and maintenance of memory B cells.383 distinctively higher frequency of λ light chain usage compared
It is not clear whether all B cells are capable of giving to follicular B cells (20% vs 5%). Also unlike follicular B cells,
rise to memory B cells. Memory or “secondary” B cells they maintain their population in adult animals largely by self-
were originally described as expressing distinctively low renewal (possibly dependent on periodic stimulation by self-
levels of CD24/HSA, recognized by the monoclonal anti- antigen; see following discussion), rather than by input from
body J11d.384 Some years later, fractionation of naïve spleen precursor cells, as shown in cell transfer studies.392 There is
B-cell precursors into J11d low and J11d high subsets in a spleen also a recently described population of IgM-secreting cells in
focus assay system showed that while rapid antibody secre- bone marrow with a B1 cell surface phenotype.393
tion derived from J11d high cells, memory came largely from Perhaps the most distinctive feature of CD5 +/B1 B cells is
the J11d low subset.385 Subsequent experiments demonstrated their enrichment of certain self-reactive specificities, notably
that GCs (the site where most memory B cells are generated) for branched carbohydrates, glycolipids, and glycoproteins,
were only produced in cell transfers of J11d low B cells and including phosphorylcholine, phosphatidylcholine (PtC), the
not with J11d high or CD5 + B cells.386 Considering the rapid Thy-1 glycoprotein, and bacterial cell wall constituents.394–397
Ig secretory response of B1 B cells and MZ B cells, both These antibodies, although autoantibodies, are not pathogenic,
contained in the CD24/HSA high fraction, and the fact that but rather referred to as “natural autoantibodies” whose exis-
most other CD24/HSA high cells are immature (transitional) tence has been recognized in serum for several decades.398–400
B cells, likely to be highly susceptible to apoptosis, it seems Their function is still under active investigation, but at least
quite reasonable that memory B cells would not be a major some natural autoantibodies are thought to function in clear-
product of this fraction. Rather, the most likely candidate ance of senescent cells or proteins and to provide an initial im-
for the memory B-cell precursor is the follicular B-cell sub- munity to common bacterial or viral pathogens, serving as a
set (Fo), which has variable but lower expression of CD24/ kind of “hardwired” memory B-cell population.401–404
HSA. Whether there is heterogeneity for GC formation or Under physiologic conditions in normal mice, most of the
memory B-cell generation within the Fo population remains CD5 + B cells in the B1 population (so called “B1a B cells”)
to be determined. arise from precursors in fetal liver, as cell transfer studies
Whether the maintenance of memory requires periodic re- showed many years ago that B cells with this phenotype
stimulation by antigen has been a longstanding controversial were inefficiently generated from bone marrow precursors in
issue. On the one hand, transfer of B cells and T cells into adult mice, compared to fetal or neonatal precursors.405 This
irradiated recipients usually required simultaneous challenge is shown by repopulation of SCID mice by pro-B stage cells
with antigen in order to elicit the full response and maintain isolated from fetal liver and adult bone marrow (Fig. 8.13).
B-cell memory in recipients.387 Antigenic fragments can per- Part of the reason for this difference may be that novel BCRs
sist for very extended periods on follicular dendritic cells, are enriched by distinctive mechanisms of fetal B lymphopoi-
which are very potent antigen-presenting cells, and, so in this esis, including recombination in the absence of TdT (thereby
model, the memory B cells are periodically triggered to self- favoring rearrangement of certain D-J and V-D junctions
renew by interaction with antigen on follicular dendritic cells. possessing short regions of homology16) and distinctive pre-
However, on the other hand, memory B cells can be main- BCR selection.263,273,406 Forced expression of transgenic BCRs
tained in the apparent absence of T cells or follicular dendritic cloned from CD5 + B cells can give rise to CD5 + B cells from
cells.388,389 Furthermore, analysis by bromodeoxyuridine la- bone marrow of adult animals, but the physiologic relevance
beling of memory B-cell populations showed that they were of this remains to be carefully assessed. B1 B cells that lack
nondividing.381 This issue has been addressed in elegant ex- CD5, termed B1b B cells, can be generated from bone marrow
periments using inducible Cre recombinase to switch the BCR and a distinctive phenotype for a precursor capable of pro-
on memory cells away from the immunizing antigen.390 Such ducing such cells has been described,162 although the devel-
antigen-negative memory cells still persisted for extended pe- opmental relationships between conventional (“B2”) B-cell
riods, clearly demonstrating that this was a physiologic prop- precursors and these cells are complex.407 Embryonic day 9
erty of the cell type, independent of the presence of antigen. YS hematopoietic progenitors have been sown to give rise to
B1 and MZ B cells but not conventional follicular B cells.
B1 B Cells The development of B1/CD5 + B cells is thought to be
B1 B cells, initially described as Ly-1/CD5 + B cells, are dis- more dependent on antigenic selection compared to fol-
tinguished from follicular B cells by phenotype, anatomic licular B cells. This idea was first suggested by the finding
distribution, and function.23,391 The B1 B-cell phenotype of particular specificities enriched in this population and

FIG 8.13. Generation of B Cells in Severe

Combined Immunodeficiency (SCID)
Mice Shows the Distinctive Phenotypes
Produced from Fetal and Adult Pro-B
Cells. Similar numbers of pro-B cells,
isolated as in Figure 8.5, Fr. B/C, were
injected intravenously into sublethally
irradiated adult SCID mice. Recipients
were analyzed 3 weeks later for spleen
cell lymphocytes by staining as shown
in the figure.
strengthened by the observation of repeated occurrences signaling components or positive mediators in this pathway
of particular V H/V L pairs.408 Thus, for example, the anti- such as an Ig-α tail mutant,416 Btk deficiency (in xid or Btk
PtC specificity is predominantly encoded by V H11Vκ9 and null mice),417,418 CD19 null mice,419,420 CD21 null mice,421 CD45
V H12Vκ4 utilizing two V H genes rarely found in conventional null mice,410 and vav null mice422 all negatively impact this
T-dependent immune responses.395,409 These cells appear to B-cell population. A further indication of a signal-dependent
participate in T-independent responses, but in normal phys- selection model is the accumulation of B1-type B cells with
iology may in fact provide an initial low-affinity “first wave” high-level expression of a BCR surrogate, LMP2A, but only
response to many pathogens that eventually will also elicit a B2 B cells and MZ B cells with low-level expression.423
T-dependent response.404,410 The functional role of B1 B cells is still incompletely
The observation of their self-reactive bias, their capacity understood. Whereas it is clear that they produce natural
for self-renewal, and their restricted repertoire of distinctive autoantibody, there also is evidence for their participation in
BCRs all are consistent with an important role for BCR– immune responses. B1 B cells have also been shown to accu-
antigen interaction in generation and maintenance of this mulate in draining lymph nodes of the respiratory tract dur-
population. This has been formally confirmed recently by ing influenza infection of mice.393 Treatment of mice with
studying mice bearing a BCR transgene specific for a glycosyl- Francisella tularensis –derived lipopolysaccharide protects
ation present only on the Thy-1 membrane protein.411 In these them from subsequent exposure to the live bacteria (the
mice, transgene-encoded serum anti–Thy-1 autoantibody396 etiologic agent of tularemia) by activating a small fractions
was readily detected, and there was a corresponding accumu- of B1 B cells with this antigen specificity.424
lation of a population of CD5 + transgene BCR+ B cells in the IL-10 is a cytokine with anti-inflammatory properties
peritoneal cavity. Importantly, in Thy-1 null mice generated that also enhances B-cell survival and antibody production.
by gene targeting, neither serum autoantibody nor the B-cell Thus, it can function to damping down cellular immune re-
population was found, demonstrating the critical role for an- sponses and promote humoral immunity. Many years ago,
tigen in selection of this B-cell population.412 B1 B cells were identified as the B-cell source of this cytokine
Consistent with the importance of antigen selection in the that is also secreted by monocytes, mast cells, and regula-
generation and/or maintenance of these cells, this population tory T cells.425 Recently, based on observations of alterations
is often severely affected in mice bearing mutations that alter in the level of inflammation in mice lacking or overexpress-
BCR signaling intensity. The loss of negative mediators such ing CD19, a CD1d(high) subset of B-1 cells expressing IL-10
as PTP1C/SHP-1 in “moth-eaten” mice,413 of CD72 that re- has been identified.426 Considering the anti-inflammatory
cruits this phosphatase,413 and of the CD22 coreceptor414,415 all regulatory effects of IL-10, this subset has been referred to
result in an increased frequency of B1/CD5 + B cells relative to as “B Regs” or “B10” B cells. Investigation of this subset in
follicular B cells. On the other hand, the loss of critical BCR the experimental autoimmune encephalomyelitis mouse

model of multiple sclerosis showed that initiation and pro- a significant portion of MZ B cells are produced during fetal
gression of disease was influenced reciprocally by depletion development. Consistent with this fetal origin, many MZ
of B cells. Because experimental autoimmune encephalomy- B cells have Ig heavy chains with little or no TdT-mediated
elitis is considered a T cell–mediated disease, this provides N-regions at their V-D and D-J junctions.443 Interestingly,
strong evidence for the capacity of B1 cells in modulating irradiated mice repopulated with bone marrow precursors
T cells during the initiation of disease. Curiously, depleting still accumulate significant numbers of MZ B cells with
B cells after experimental autoimmune encephalomyeli- low levels on N-addition, indicating that the strong BCR-
tis was already established actually diminished symptoms, mediated selection guiding entry into this subset can select
indicting that B cells were required for persistence of auto- such cells from adult bone marrow–generated B cells.
reactive T cells. This suggests that simple depletion of all In recent years, there has been considerable progress in
B cells in certain autoreactive pathologies may have different understating some of the non–BCR-related pathways that
consequences, depending on the progression of the disease. regulate the generation of MZ B cells. For example, the can-
Finally, examination of B10 function in the autoimmune nabinoid receptor 2 appears important in localization of MZ
lupus-prone NZB/W mouse F1 indicated that these cells B-cell precursors to the anatomic region where Notch sig-
were responsible for inducing regulatory T cells that could nals are available.444 Notch ligand expression for generation
ameliorate the course of disease in these animals.427 of MZ B cells is regulated by the E3 ubiquitin ligase Mind
bomb 1, which also is critical for T-cell development.445
Marginal Zone B Cells Furthermore, Notch2 activation in MZ B cells is potenti-
Another B-cell subset distinct from follicular B cells is the ated by expression of the fringe glycosyltransferases lunatic
MZ B-cell population. MZ B cells are localized in a distinct fringe and manic fringe.446 Notch activation requires its pro-
anatomic region of the spleen that represents the major teolysis, and this is now known to occur through activity of
antigen fi ltering and scavenging area (by specialized macro- a disintegrin and metalloproteinase 10 (ADAM10).447
phages resident there). It appears that they are preselected to
express a BCR repertoire similar to B1 B cells biased toward B-Cell Tolerance and Receptor Editing
bacterial cell wall constituents428 and senescent self-compo- A number of transgenic mouse models have been devel-
nents (such as oxidized low-density lipoprotein [LDL]).429,430 oped for the study of B-cell tolerance (Table 8.2). In two
Similar to B1 B cells, they respond very rapidly to antigenic of these systems, high-affinity BCRs are expressed as IgM–
challenge, likely independently of T cells, but participating IgD transgenes, one specific for the antigen hen egg white
in the early phase of T-dependent responses.431–433 Uniquely, lysozyme (HEL), the other for a specific polymorphic de-
among all populations of B cells, MZ B cells are dependent terminant on MHC class I.281,298–300,448–450 The advantage of
on Notch2 signaling for their development.434,435 these systems is that they utilize antigens that can be regu-
There are similarities and differences in the cell sur- lated: Transgenic B cells can develop in either the presence
face phenotype of MZ and B-1 B cells. Thus, they both are or absence of antigen, and cell transfer experiments can be
IgM +++ IgD−/+, CD23 −, and CD9 +.436 However while B1/ employed to alter the B cell’s antigenic milieu. These have
CD5 + B cells express CD5 and CD43, MZ B cells do not, and been used initially to confirm ideas on B-cell tolerance that
MZ B cells express distinctively high levels of CD21, while originated in work with nontransgenic B cells, namely that
B1 cells have distinctively low levels (see Fig. 8.12). Also, MZ exposure to antigen is generally deleterious to developing
B cells have high levels of CD1d, while most B1 B cells do B cells, resulting in their elimination or failure to mature
not,437 the exception being the “B10” subset mentioned pre- (instead entering an “anergic” state). However, the resolu-
viously. Certain mutant mice show similar effects on MZ tion of these systems, coupled with advances in gene tar-
and B1 B cells, distinct from follicular B cells. For example, geting and other molecular technologies, have uncovered
most of the mutations described for B1 B cells that alter important new details regarding the way that self-reactive
BCR signaling have similar consequences for MZ B cells,438 B cells develop (or fail to develop). For example, studies in
although cells that resemble MZ B cells are present in xid/ the anti–H-2 model uncovered an alternative to deletion in
Btk-deficient mice, leading to some controversy in their or- response to immature B-cell encounter with antigen, BCR
igins.439 Both are decreased by a mutation in the Ig-α tail editing to escape autoreactivity.281
that weakens overall BCR signaling.440 They are also both In the HEL system, differences have been discovered in
decreased in the Aiolos transcription factor null mouse.63 immature B-cell responses to soluble versus membrane-
Interestingly, deletion of the Pyk-2 tyrosine kinase results in bound antigen, suggesting that the extent of BCR cross-
elimination of MZ B cells while B1 cells are still found.441 linking can influence cell fate.299 Furthermore, studies with
MZ B-cell development can be studied in a heavy chain this system on different mutant backgrounds that shift
transgenic mouse model system where large numbers of BCR signaling thresholds up or down has shown that such
such cells are produced.428,442 In this V H81X heavy chain alterations can result in striking alterations in selection out-
mouse, B cells with a specific light chain accumulate with an comes.451,452 Work in this system has shown that one conse-
MZ phenotype. This MZ population is eliminated by dele- quence of B-cell tolerance may be arrest of B-cell migration
tion of CD19, Btk, or CD45, all genetic changes that weaken so that follicular entry is inefficient.330,331 Presumably, failure
BCR signaling.438 MZ B cells are generated in mice even to reach such follicular niches contributes to handicapping
when B-cell development is blocked shortly after birth by the autoreactive B cells, resulting in their relatively speedy
conditional deletion of Rag-2,353 showing that like B1 B cells, elimination.

TABLE 8.2 Transgenic Models of B-Cell Tolerance
Ig Transgene Antigen Background Effect Reference

3-83 μk MHC class I H-2Kk Deletion, receptor editing 281,300,450
H-2Kk,b H2-Kd Normal development

3-83 μk MHC class I H-2Kk lpr autoimmune-prone Deletion unaffected 308
H-2Kk,b
anti-HEL μδ HEL sHEL-Tg Anergy 298,448,449
Wild-type Normal development

anti-HEL μδ HEL mHEL-TG Deletion 299
anti-HEL μδ HEL HEL-Tg lpr autoimmune prone Deletion unaffected 307
anti-HEL μδ HEL sHEL-Tg Self-antigen promoted 452
CD45 null development

anti-HEL μδ HEL sHEL-Tg motheaten Deletion by lower valency 451
autoantigen
3H9 μ-only ssDNA with many light BALB/c No anti-DNA autoantibodies 301
chains
3H9 μ-only ssDNA with many light lpr Anti-DNA 597
chains Autoimmune-prone autoantibodies

3H9 μκ dsDNA BALB/c Deletion, editing 302
3H9 μκ dsDNA JH−/Jκ− Deletion 285
3H9 μλ dsDNA Wild-type and lpr Anergy 332,333
Autoimmune prone
598
3H9-R/Vκ4-R dsDNA BALB/c Deletion, editing
3H9-R/Vκ8-R ssDNA anergy
3H9-R/Vκ4-R dsDNA, ssDNA Rag-2− Deletion, activation 599
3H9-R/Vκ8-R
600
3H9/56R dsDNA, ssDNA BALB/c Deletion, editing
3H9/56R76R
Antierythrocyte Red blood cell Wild-type Tg+ cells only in peritoneal cavity 457–459
601,602
VH11μ, VH12μ PtC, BrMRBC Wild-type Increased number of B1 B cells
412
6C10μ ATA determinant Wild-type ATA B cells ATA serum
(glycosylation of Thy-1) Thy-1 null Normal development; no ATA
ATA μκ (6C10μVκ21c) ATA determinant Wild-type Receptor editing or development 464,465
block in bone marrow; ATA

in serum
Thy-1 null Normal development
ATA, antithymocyte autoantibody; DNA, deoxyribonucleic acid; HEL, hen egg white lysozyme; Ig, immunoglobulin; MHC, major histocompatibility complex.
Another major line of investigation has focused on a more will result in autoreactivity by some of these cells. Studies
“physiologic” example of pathogenic autoreactivity, the with a NF-κ B–dependent Iκ B alpha gene reporter mouse
anti-DNA antibodies produced in lpr (Fas-deficient) mice revealed that autoreactive BCRs resulted in activation of the
that are generally considered to model the human disease of reporter and that reporter + cells have elevated levels of IRF4,
systemic lupus erythematosus. Analysis of transgenic mice indicating that BCR cross-linking by self-antigen signals
bearing a heavy chain transgene known to be capable of gen- through a NF-κ B pathway and acts through IRF4 to regu-
erating anti-dsDNA reactivity with numerous light chains late receptor editing.456
showed that only light chains with ssDNA activity were tol- Later work analyzing transgenic B cells expressing λ light
erated in the periphery, and even these did not contribute chain (or in kappa null mice) where B cells have dsDNA
to the serum antibody pool.301 Follow-up work uncovered binding has shown the failure of follicular entry previously
receptor editing in this model302,453–455 and also showed that described in the HEL system.332,333 Interestingly, similar
when such editing was blocked, the B cells were eliminated analyses on an “autoreactive” (Fas-deficient) background
in the bone marrow at an immature stage.285 have shown that this follicular exclusion is lost and pro-
Receptor editing, due to failure of Rag downregulation, duction of pathogenic autoantibodies ensues, providing
in replacement of the original light chain by a different a powerful model for the further characterization of the
one, resulted in decrease or elimination of autoreactivity. development of autoimmunity due to breakdown of B-cell
Although discovered in several BCR transgenic models, it tolerance.
also occurs at some level in the newly formed B-cell stage in A different model of a pathogenic antierythrocyte auto-
bone marrow, as random pairing of heavy and light chains antibody has shown another possible mechanism whereby

self-reactive B cells may avoid deletion or receptor editing by The role of BCR signaling intensity in directing B cells to
sequestration from self-antigen.457,458 In this system, the trans- various specialized subpopulations has been studied using
genic B cells are largely absent from spleen but instead sur- the ATA μκ BCR transgenic model, by altering the levels
vive in the peritoneal cavity where exposure to the distinctive of Thy-1 self-antigen and then examining B-cell develop-
microenvironment may also contribute to the persistence of ment.465 This analysis observed follicular B-cell develop-
these cells. Eventual activation of the B cells by mitogen or an- ment in the absence of antigen, but MZ B-cell production
tigen can lead to an autoimmune condition in these mice.459 in the presence of very low levels of autoantigen. As shown
B cells may express BCR coreceptors or modulators of previously, normal physiologic levels of the Thy-1 autoan-
signaling pathways that change the response to self-antigen tigen results in a block in bone marrow development, but
cross-linking. For example, it appears that enzymatic acety- allowed B1 production, with concomitant generation of ATA
lation and deacetylation of a cell surface carbohydrate by detected in serum. This work led the authors to propose a
sialate: O-acetylesterase alters B-cell responsiveness, by reg- model of B-cell development where strong BCR signaling
ulating the function of CD22, a sialic acid binding Ig-like produces B1 B cells (particularly during fetal development),
lectin (siglec) that inhibits BCR signaling.460 As mentioned intermediate BCR signaling yields MZ B cells, and weak
previously, the level of CD19 also modulates BCR signaling (or negligible) BCR signaling results in follicular B-cell
and selection, such that altering CD19 expression can alter generation.
the frequency of B1 B cells. Recent work has revealed that
mutation of the Wiskott–Aldrich syndrome protein selec- Role of Complement, Serum Antibody, and Cluster of
tively in B cells results in severe autoimmunity due to BCR Differentiation 5 in B-Cell Tolerance and Response
and TLR hyperresponsiveness.461 The importance of complement in the immune system has
A common thread in all of the studies described here is the long been recognized based on classic experiments showing
negative impact that the B cell experiences upon interaction that cobra toxin decreased responses.466–468 Over the past 10
with self-antigen, an expected result for systems that model years, the importance of complement in immune responses,
the regulation of pathogenic autoantibodies. However, a explaining the function of adjuvants, has become clearer.469
class of autoantibodies is produced in healthy individu- In the context of B-cell development, components of the
als, and these “natural autoantibodies” may play a role in complement system also play a role in modulating BCR re-
early responses to certain classes of pathogens.401,402,404,462 sponses and negative selection of B cells. At least in the HEL–
Such a natural autoantibody has been used to construct a Ig/sHEL system, altering the strength of the BCR signal, by
transgenic model system where the self-antigen can be regu- CD45R or PTP1C/SHP1 inactivation, could either reduce or
lated. Most natural self-antigens are common glycosylations enhance B-cell deletion.452 Similar results could be obtained
or cell constituents such as PtC that cannot be eliminated, by altering the BCR-associated chain CD19.470 Considering
but a class of natural autoantibodies binds to thymocytes the role of complement as a coreceptor for modulating BCR
(antithymocyte autoantibody [ATA]) and many of these signaling thresholds, the finding that HEL double transgenic
recognize a glycosylation that is only present on the abun- mice that also are Cr2 (CD21/CD35) null develop peripheral
dant thymocyte cell surface glycoprotein CD90/Thy-1.412 B cells that are apparently not fully anergized, as they could
Thy-1–null mice have already been generated,463 so pro- still respond to antigen challenge.471 This could be due to
duction of ATA–BCR μ-only transgenic mice enabled the an inefficient retention of antigen (sHEL) on stromal/den-
study of the role of antigen in the generation of this natural dritic cells and, therefore, weaker signaling by a monomeric
autoantibody. Interestingly, both production of serum ATA soluble antigen to developing/transitional B cells in the bone
and accumulation of B cells with the appropriate light chain marrow or spleen. Potentially, a major role for complement
(by rearrangement of endogenous Ig light chain locus) for would be to localize self-antigens on bone marrow stromal
the ATA–BCR required the presence of Thy-1 self-antigen.412 and dendritic cells facilitating tolerance at the newly formed
Thus, at least some B cells are selected for binding to self- B-cell stage. An implication of these findings is that comple-
antigen, although these may belong exclusively to special- ment deficiency could result in accumulation of functional
ized B-cell compartments, such as the B1 B-cell subset. autoreactive cells, possibly leading to autoimmunity.
This ATA–BCR/Thy-1 tolerance model has been extended Natural autoantibody, a major component of the serum,
by the production of ATA μκ transgenic mice, where most de- may have a role (together with complement) in maintain-
veloping B cells express the natural autoreactive specificity.464 ing tolerance to highly conserved self-antigens.469 Some such
In such mice, most of B cells developing from bone mar- IgM antibodies, highly conserved phylogenetically, have
row become arrested in spleen and either cease maturation been shown to recognize conserved carbohydrates beta-
or else “edit” their BCR by replacing the transgene-encoded glucan and chitin structures present on fungi and so serve to
light chain by a different light chain as a consequence of re- enhance host defense.472 It also appears that it plays a role in
arranging the endogenous light chain locus. This resembles amplifying immune responses, as suggested by studies with
typical “negative selection,” as seen with other transgenic a mutant mouse lacking the μ heavy chain secretory exon.473
models.281,298–300,448,450 Nevertheless, some B1 B cells are de- The deficit due to a lack of serum IgM was particularly
tected, and high levels of serum autoantibody are produced, severe in an acute peritonitis model, where some restoration
suggesting that some B cells with this BCR are capable of ma- of responsiveness could be obtained by injection of a mono-
turing, perhaps in specialized microenvironments or at dis- clonal IgM antibody derived from CD5 + B cells.473 Finally, in
tinctive (fetal/neonatal) times. infection with influenza virus, it appears that early presence

of natural autoantibody was equally as important as anti- and, as in mouse, the bone marrow is the predominant site
body induced during the course of infection in mediating in adults, with production continuing over most of an indi-
viral clearance and survival.404 This work has led to the pro- vidual’s lifespan, with some decrease in the aged. The fetal
posal of a two-phase description of immune responses, with omentum has also been described as a site for B-cell develop-
an early T cell–independent phase dominated by natural ment similar to mouse.9,479 There is some indication that VH
autoantibody, likely critically dependent on innate immune gene usage is more restrictive in fetal development.480 TdT,
recognition and activation of complement, leading to a later absent from mouse fetal development, is similarly missing
T cell–dependent phase culminating in the GC reaction, very early in human fetal development but is expressed by
producing high-affinity antibody and memory B cells.474 the eighth or ninth week of gestation based on detection of
The presence of CD5 on many of the B1 B cells has led N-region addition.480 In general, there is more N-addition
to the question of whether it plays a direct role in main- (and consequently longer CDR3 regions) in human anti-
taining such self-reactive B cells, particularly in light of its bodies compared to mouse at all stages of development,
role in altering T-cell selection thresholds as demonstrated including the adult, although the significance of this is not
in the CD5-null mouse.475 Analysis of B-cell responses to known. Analysis of human T-cell development indicates
BCR cross-linking in CD5-null mice suggested that the that fetal and adult T cells arise from distinct developmen-
presence of CD5 makes the normal population less likely to tal pathways, with fetal cells allowing more self-reactivity to
respond by secreting IgM,476 effectively “raising the thresh- persist481 similar to mouse.482 Finally, as recently described
old” for their response, possibly by promoting interaction in mouse, B cells with the capacity to rapidly secrete IL-10
of the BCR with SHP-1.477 Analysis of the effect of CD5 have also been identified in humans, and such cells appear
expression on tolerance in the HEL-Ig/sHEL system also to be increased in patients with autoimmune disease, such
suggested a modulating role, with absence of CD5 leading to as rheumatoid arthritis.483
a loss of B-cell tolerance and production of serum anti-HEL
autoantibody.478
Bone Marrow Development
B-cell development can be detected in human bone mar-
B-CELL DEVELOPMENT IN HUMANS:
row from 20 weeks of gestation and continues throughout
SIMILARITIES AND DIFFERENCES FROM MOUSE life.484,485 Phenotypic subdivisions similar to those in the
Fetal Development mouse have been described for developing bone marrow
As described for mouse, human B-cell development can be cells486 (Fig. 8.14). Thus, CD19 identifies B-lineage cells at all
broadly divided into that taking place prior to birth and stages, with the earliest stages also expressing CD34, a mole-
that operating after birth and throughout life. The liver and cule found on MLPs.487 These CD19 + CD34 + cells have been
spleen are major sits of fetal B lymphopoiesis in humans shown to express TdT, Rag-1 and Rag-2, the SLC orthologs,
FIG. 8.14. Stages of Developing B-Lineage Cells in Human Bone Marrow can be Delineated, Based on Expression of Combinations of Cell
Surface Proteins, Similar to Mouse.

and Ig-α /Ig-β.486 They contain D-J, but not productive VDJ expression is high in GC B cells and is required for for-
rearrangements. The next stage can be defined by loss of mation of the GC.503 Strikingly, its 5′ regulatory region is
CD34, with a fraction that is still IgM− (pre-B cells) and mutated as a consequence of hypermutation in GC B cells,
another subset that is IgM +. The pre-B cells express heavy the first example of hypermutation targeted outside the Ig
chain in their cytoplasm and have downregulated expres- regions,501,502,504 and links this process to deregulated cell
sion of TdT.486 All of the IgM + cells express a marker lost growth and lymphomagenesis. These findings have focused
upon final maturation, CD10, so they are immature B cells. interest on Bcl-6, and recent work has shown that it binds
A recent study examined the B-lineage potential of cells to the promoter of the antiapoptotic proto-oncogene Bcl-2
expressing different levels of CD10 and found that increas- by interacting with the transcriptional activator Miz1, also
ing levels of this molecule were found on more differenti- suppressing Miz1 activation of Bcl-2.505 In lymphomas,
ated B-lineage cells, suggesting its potential for identifying Bcl-6–mediated suppression of Bcl-2 is lost, either by chro-
distinctive stages in B-cell development in humans.488 mosomal translocations to Bcl-2, mutation of Bcl-2, or Miz1
Human early B-lineage cells have proven much more deregulation.
difficult to grow in culture than mouse cells, although con-
ditions have now been defined that allow their expansion
Abnormalities of Development
in vitro. One of the most surprising differences between
mouse and human B precursor growth has been much less A key discovery in the past decade has been the finding
dependence on IL-7, suggesting that a different cytokine that a well-known immunodeficiency, xid, characterized
may substitute in humans.489–491 The finding that lack of by inability to respond to bacterial infections and a severe
the cytokine common gamma chain (γc) in mouse blocks deficit in peripheral B cells,506 is due to mutations in Btk.507
B-cell development but spares T-cell development, whereas Shortly after this finding, the mouse ortholog of Btk was
the reverse is the case in humans,154,492 demonstrates clear shown to be the cause of murine xid, an extensively studied
differences in cytokine dependence between mouse and mutation originally identified in CBA/N mice.508 Btk defi-
human lymphopoiesis. ciency in humans is more severe than xid, with little B-cell
Human B precursors also express orthologs of λ5 and development past the early B-cell stage, in contrast with an
VpreB, although their organization and number differ from absence of normal peripheral B-cell development in mouse
the mouse.493–495 Their expression occurs at a corresponding and inability to respond to certain types of T-independent
stage and analysis of the human EBF gene has shown that its antigens.509 This difference is not simply due to specific dif-
targets are similar to those of mouse EBF, including Ig-α , ference in the mutations, as a complete null mutation in
Ig-β, and 14.1 (the human ortholog of λ5).496 Importantly, mouse is indistinguishable from xid.418 Thus, human B-cell
mutations in components of the pre-BCR lead to immuno- development, likely at the pre-BCR signaling stage, is much
deficiency diseases in humans, pointing out the similar and more dependent on Btk.
crucial role that pre-BCR signaling plays in human B-cell While xid is by far the most common B-cell deficiency,
development.497–499 amounting to more than 80% of those identified, non–
X-linked mutations have also been observed. These cor-
respond to mutations in the pre-BCR signaling complex,
Germinal Center Differentiation and, in most cases, similar effects had been observed in the
As in mouse, GCs are anatomic structures in secondary mouse. For example, deletions or mutations in the μ con-
lymphoid organs where T-dependent immune responses stant regions accounted for another 5%.499 Examples have
occur, selecting high-affinity clones during the process of been found of mutations in the 14.1 gene, the human ortho-
somatic hypermutation and eventually generating memory log of the mouse surrogate light chain λ5 protein,497 and also
B cells. The GC consists of B cells at a variety of differen- in Ig-α.498 In both of these cases, early B-lineage cells, iden-
tiation states, from early activated B cells through the plas- tified as CD19 + CD34 + were present in normal numbers in
mablast stage. Therefore, study of this process has benefited bone marrow, but CD19 + CD34– Cμ + cells (pre-B cells) and
significantly from resolution of intermediate stages using all later stages were absent. Finally, a patient with a mutation
multiparameter flow cytometry and carefully chosen cell in BLNK, an adaptor protein that links pre-BCR signaling
surface markers. In one study, this allowed defi nition of two from syk to the rest of the signaling cascade, showed a simi-
IgD + stages, two GC stages, and a memory stage in tonsillar phenotype.510
lar B cells.500 This work focused on the levels of somatic Very few peripheral B cells are detected in any of these
mutations accumulating as cells passed through this path- disorders, which suggests that pre-BCR signaling is more
way, finding that it was first detected in the initial GC critical in human than in the mouse, where mutations in
stage (centroblast). Subsequent work has shown that IgD + λ5 and BLNK allow the generation of variable numbers of
somatically mutated cells can be detected in peripheral peripheral B cells. The reason for this difference is not yet
blood, based on expression of CD27, suggesting that they are understood. Interestingly, common variable immunode-
IgM + IgD + memory B cells.500 ficiency has been shown to result from mutations in the
Recent work has found an interesting link between so- common gamma chain154 or in the JAK3 gene511 consistent
matic hypermutation and certain Bcls. Bcl-6 is a transcrip- with IL-7 playing a much less critical role in human pre–B-
tional repressor that is linked to both GC B cells and to B cell growth than in mouse, as had already been determined
lymphomas that likely derive from GC B cells.501,502 Bcl-6 from culture studies.489,490 B-cell development is relatively

intact, whereas T cells are ablated. The reverse is true for undergoing somatic hypermutation of their IgV genes
mouse.512,513 mediated through activity of AID.366,367 As mentioned pre-
Analysis of B cells with mutations in genes required for viously, aberrant somatic hypermutation in DLBCL has
normal TLR signaling has revealed the importance of such been shown to induce mutations in the Bcl-6 gene.501,502,504
signaling for maintenance of B-cell tolerance.514 This study As mentioned previously, Bcl-6 is often found to be overex-
examined the binding specificity of BCRs isolated from pressed in DLBCL by either hypermutation of its promoter
B cells of patients with defects in IL-1 receptor-associated or chromosomal translocation, but investigation of DLBCLs
kinase 4, myeloid differentiation factor 88, and UNC-93B. lacking these genetic alterations has revealed that Bcl-6 is
As all TLRs but TLR3 depend on IL-1 receptor-associated targeted for ubiquitylation and proteasomal degradation by
kinase 4 and because TLR3 depends on UNC-93B, this work a SKP1-CUL1-F-box protein ubiquitin ligase complex that
assessed the contribution of all TLR signaling. Autoreactive contains the orphan F-box protein FBXO11.527 FBXO11 was
BCRs were found at high frequency in the newly formed and found to be mutated or deleted in DLBCL cell lines and also
mature B-cell pools in all patients, establishing a critical role in primary leukemias, thus revealing another mechanism
for TLR signaling in establishing and maintaining B-cell for Bcl-6 dysregulation, leading to lymphoma.
tolerance. Interestingly, none of these patients had serum AID in DLBCL likely is responsible for widespread ge-
autoantibody or autoimmune disease, suggesting that some nome instability,528 and using a mouse model system, AID
further signal, possibly TLR mediated, is needed for disease was clearly identified as responsible for c-myc to Ig V-J(H)
progression. translocations.529 Examination of known non-Ig targets of
AID also identified a high rate of mutation of PAX5 in both
classic and nodular lymphocyte-predominant Hodgkin lym-
Insights into B-Cell Malignancies phoma.530 Microarray gene profi ling showed that the NF-κ B
One of the principle reasons for studying the regulation of signaling pathway is constitutively activated in DLBCL and
B-cell development is that defects in this process may result as normal NF-κ B activation by BCR cross-linking requires
in lymphoma. Human B-lineage neoplasias can be viewed CARD11, a cytoplasmic scaffolding protein, this gene was
as transformed counterparts of normal B-cell developmen- examined for mutations. Ten percent of DLBCL tested had
tal stages, such as pro-B, pre-B, immature B, mature B, or missense mutations, providing a rationale for develop-
plasma cell, based on rearrangement status and surface phe- ment of new therapeutic agents targeting this pathway.531
notype.493 In some cases, transformed cells may even retain Other work has identified a negative regulator of NF-κ B
growth characteristics of the normal counterpart. This type signaling, A20, as a frequent target in B-cell lymphomas,
of classification scheme, correlating features of lymphomas by using genome-wide analysis of genetic lesions.532 Thus,
and leukemias with their normal counterparts, has been early “whole genome” analyses of leukemias and lympho-
useful in diagnosis and prognosis of B-lineage neoplasias. mas by microarray gene profi ling has been followed up by
For example, B-precursor ALL, the most common type of high-resolution determination of gene copy and targeted
ALL in children (ALL accounts for 25% of childhood can- sequence analysis. Now, the availability of whole exome or
cer), is a clonal expansion of a cell defi ned by surface pheno- even whole genome high-throughput sequence determina-
type and Ig rearrangement status as representing the pro-B tion will undoubtedly identify as yet unsuspected key driver
stage.515–517 Recent analyses of B-precursor ALL suggests mutations in these diseases, potentially leading to new and
that they can be subdivided into a pro-B type, predomi- more personalized therapies.
nant in pediatric patients, and a pre-B type, more frequent Another therapeutic approach makes use of knowledge of
in adults.518 A subtype of adult pre-B ALL is Philadelphia the surface phenotype of the transformed cell, as in the re-
chromosome-positive, being defi ned by the oncogenic cent development of anti-CD20 therapy for several types of B
BCR–ABL1 kinase and showing deletion of the Ikaros gene lymphomas.533 CD20 is a 33 kDa phosphoprotein expressed
in most cases. Pre-BCR signaling mediated cell cycle arrest highly on the surface of mature B cells.534–536 Antibodies to
has been shown to depend on normal Ikaros function,519 CD20, originally called B1, were initially characterized by
providing an explanation for the high incidence of Ikaros their stimulatory and inhibitory effects on human B cells,
deletion in this disease. indicating the importance of CD20 in regulating B-cell pro-
Whereas traditional chemotherapeutic agents typically liferation and differentiation.537–539 Therapeutic anti-CD20,
target proliferating cells without specificity for malignant called rituximab, is a chimeric antibody derived by fus-
cells, the identification of molecular abnormalities that result ing the V regions of a mouse antibody to the human IgG1
in the abnormal survival and proliferation of leukemic cells constant segment.540 The precise mechanism of depletion
may lead to the design of novel therapies that specifically is not yet fully understood but likely include contribu-
target these cells. For example, the successful treatment of tions from antibody-dependent cell-mediated cytotoxicity,
mice carrying human B-precursor leukemias with antisense complement-mediated cytotoxicity, and direct antibody
strategies and tyrosine kinase inhibitors holds promise for binding effects, including sensitization to apoptosis.533
efficacy in humans.520,521 The identification of novel translo- As most of the cells in many indolent B-cell neoplasms,
cations in B-precursor ALL also may identify mechanism(s) such as non-Hodgkin lymphoma or chronic lymphocytic
responsible for this disease at the molecular level.522–525 leukemia (CLL), are not predominantly in cycle, the problem
Another example is diffuse large B-cell lymphoma may be more a failure to die appropriately, rather than a fail-
(DLBCL) that appears to arise from GC B cells526 that are ure to regulate proliferation. A greater understanding of the

growth and sensitivity to apoptosis of different types of lym- This proliferation selectively expands cells that have BCRs.
phomas and leukemias may allow more specific targeting These receptors have limited diversity, as the heavy chain is
using this approach.541 For example, while CLL cells express formed by rearrangement of a single VH, several Ds, and a
relatively low levels of CD20, likely requiring higher doses single JH pairing with a light chain generated by rearrange-
of anti-CD20 for a response, a combination with another ment of a single VL with a single JL.551,552 As in mouse fetal
antibody recognizing a molecule that is highly expressed on development, there is no TdT-mediated N-region addition at
CLL, CD52 has shown promise.542 Alternatively, treatment the junctions.552
with anti-CD20 may render the cells generally more sensi- This “prebursal” receptor is diversified by gene conversion
tive to apoptosis so that use in combination with more con- by a set of V pseudogenes during this proliferative phase. At
ventional chemotherapeutic agents will be efficacious. These about hatching, these cells become exposed to the contents
combination therapies are already in clinical trials. Finally, of the bursal lumen that is connected to the gut lumen via
new insights into distinctive growth properties of trans- the bursal duct similar to the appendix. Thus, these prolif-
formed B cells, such as the role of Wnt signaling in CLL,543 erating B cells are exposed to the contents of the digestive
may suggest new approaches to treat such disorders. tract, and there is also reverse peristalsis at the end of the
gut that transports external antigens into the bursal duct.546
At about this time, the level of apoptosis increases dramati-
ALTERNATIVE STRATEGIES FOR cally, and it is possible that only 5% of the cells generated
B-CELL DEVELOPMENT in the bursa eventually emerge.553 This death may be due to
The broad outlines and even many of the details of B-cell generation of nonfunctional receptors during the course of
development are quite similar in mouse and man, but there gene conversion or it may reflect antigenic selection.
are striking differences in other species. There is a notable At hatching, emigration of B cells from the bursa increases,
common alternative approach that involves generation of but most of these cells constitute a population with a rela-
Ig + cells during fetal/neonatal development with a relatively short half-life measured in days.554 The long-lived pool
tively restricted repertoire that is then diversified by novel colonizes the peripheral lymphoid organs over several weeks
approaches (gene conversion or somatic hypermutation) in as the bursa atrophies. By 3 weeks after hatching, bursectomy
specialized lymphoid organs that are associated with the no longer results in agammaglobulinemia, indicating that
gut. In these species, most development from Ig − precursors the postbursal phase has become established.
appears to cease by birth, and the B-cell population is main-
tained by self-renewal of mature B cells. Here, we consider
the development and diversification of B cells in chicken Rabbit
and rabbit. B-cell development in rabbit is similar to chicken, in that
B cells initially are produced with a limited BCR diver-
sity555–557 during fetal life, with little new production after
Chicken birth.558,559 This repertoire is then expanded through gene
B-cell development in the avian occurs in the bursa of conversion555,560 and somatic hypermutation560,561 in a spe-
Fabricius.544,545 In fact, the term “B” cell refers to “bursa cialized gut-associated organ, the appendix.562 However,
derived,” reflecting the historic origins of research in lym- unlike chicken, this diversification process is dependent on
phocyte development. That is, removal of the bursa just antigen availability.563–565
after hatching eliminated the ability to mount an antibody Pre-B cells can be found in rabbit before birth in the liver,
response, demonstrating the importance of this organ in bone marrow, and omentum,558,566–568 but B lymphopoi-
generating cells capable of antibody formation.544 In con- esis decreases at birth and is negligible in adult animals.559
trast with the bone marrow, the bursa, being associated with Ig rearrangement during fetal and neonatal times is domi-
the gut,546 facilitates exposure of developing cells to external nated by usage of the most D-proximal V H gene paired with
antigens and bacterial flora. B-cell development in chicken multiple V L genes in the light chain.556,557,569–572 In contrast
is usually divided into three stages: prebursal, bursal, and with chicken and mouse, there is significant N-addition.557,571
postbursal.547 From 4 to 8 weeks after birth, there is a striking increase
During prebursal development, at day E5, early precur- in the diversity of this primary repertoire that occurs dur-
sors can be identified in the para-aortic foci,548 likely cor- ing proliferation of B cells in the gut-associated lymphoid
responding with similar precursor stages localized in this tissues (GALTs).559,560 V genes are diversified by both gene
anatomic site in mammals.2 B-lineage commitment, as conversion555 and also by somatic hypermutation.560,561
indicated by D-J rearrangements, is detected in the YS at Importantly, surgical removal of GALT organs, appendix,
day E5/6, and V gene rearrangement is found 3 days later.549 sacculus rotundus, and the Peyer patch, from neonatal rab-
Unlike mammalian ordered development, light chain rear- bits led to unresponsiveness to many antigens, suggesting
rangement is detected at about the same time as heavy chain, that this diversification was crucial for normal immune
and light chain can precede heavy chain.550 This means that function.573 This finding has been confirmed by sequence
there is no pre-B stage, per se, and also probably no require- analysis, demonstrating that removal of the GALT blocks
ment for SLC. Rearranging B-lineage precursors migrate diversification.562
into the bursal mesenchyme at about day E12, and, thereaf- Considering the relatively late diversification in rabbit
ter, these cells begin to proliferate in bursal epithelial buds. compared to the chicken, these gut areas will provide a milieu

of microbial antigens, and this appears to be a critical aspect repertoire during fetal/neonatal life, diversification at a later
of the diversification process.574 For example, surgery to pre- time, and maintenance of the B-cell pool in adult life by self-
vent access of intestinal flora to the appendix blocked diversi- renewal (rather than de novo generation from unrearranged
fication in this organ575,576 that could be restored by reversing precursors) is a major pattern in the design of the immune
the ligation.575 Furthermore, analysis of rabbits reared in system (Fig. 8.15). This pattern contrasts with that de-
germfree conditions revealed abnormal cellular development scribed for mouse and man, where ordered heavy and light
in the GALT564 and a lack of responsiveness to certain anti- chain rearrangement, pre-BCR selection, and replacement
gens.563 The dependence of V gene diversification on antigen of senescent B cells by newly generated B cells are major
has been directly demonstrated in animals where the sacculus aspects of development. This raises the interesting question
rotundus and Peyer patches were removed at birth and the of whether there is an analogous B-cell development path-
appendix ligated. Testing the peripheral blood B cell V gene way in humans or mouse, perhaps as a vestige.
repertoire showed an absence of diversification in contrast to The idea of two pathways in bone marrow develop-
controls.565 Although the mechanism for this stimulation re- ment was suggested previously based on the observation
mains to be established, possibilities include B-cell activation that heavy and light chain ordering is not absolute: Light
by a BCR superantigen577–579 or through a B-cell TLR.580 chain rearrangements are detected in mice incapable of
making heavy chain rearrangements and can be detected in
cells early in B-cell development.581 In one pathway, heavy
Two B-Cell Developmental Pathways chains rearrange first and the pre-BCR is assembled and sig-
The similarities of chicken and rabbit B-cell develop- nals downregulation of Rag-1/2 expression (ending heavy
ment with that in other species such as sheep, swine, and chain rearrangement), clonal expansion, and accessibility
cow suggests that the initial production of a limited BCR of the kappa locus. Cessation of proliferation is coincident
FIG. 8.15. Alternative Strategies of Repertoire Diversification and B-Cell Development. In both mouse and human bone marrow B-cell de-
velopment, ordered rearrangement predominates, with a pre–B-cell receptor (BCR_ selection phase dividing heavy and light chain rear-
rangement. The eventual outcome is a population of newly formed B cells with a diverse set of BCRs produced throughout life. In chicken
and rabbit, immunoglobulin rearrangement appears less efficient (usually the other allele is germline) and much less diverse, with a single
BCR for chicken. However, these cells proliferate and undergo gene conversion during fetal/neonatal life, followed by selection, eventually
generating a set of B cells with a more diverse repertoire that persist for the life of the animal.

with reinduction of Rag-1/2 and light chain rearrangement. BCRs with weak reactivity to self or environmental antigens
Expression of a complete BCR signals a final termination as an alternative to ordered rearrangement, pre-BCR selec-
of Rag expression. In this model, pre-BCR and BCR medi- tion, and BCR selection—an elaborate process fi ne tuned to
ates allelic exclusion by a feedback mechanism regulating generate more variation.
Rag expression. Thus, most B cells have D-J or even VDJ
rearrangements on both alleles. In the alternative pathway,
heavy and light chains rearrange stochastically, and allelic CONCLUSION
exclusion is mediated by relative inaccessibility of the loci There has been considerable progress over the past decade in
(and thereby low frequency of rearrangement). If this is the understanding the mechanisms that regulate B-cell develop-
major pathway in chicken and rabbit, it is consistent with ment, with important insights coming from application of
the observation that most chicken B cells have only one al- the novel genetic approaches of transgenesis and gene target-
lele rearranged.552,582,583 Furthermore, the careful ordering of ing. In a sense, this has allowed progression from research
TdT expression, being downregulated at the light chain rear- with simple model systems using cell lines to analysis of
ranging stage, is apparently not the case for rabbit, where a “normal B cells” in whole animals. The use of such mutant
large percentage of light chains have N-regions.572 mouse “reagents,” together with much higher resolution of
Most rabbit B cells express CD5.584 In mouse, much of normal development made possible by multiparameter flow
the CD5 + B-cell population is generated during fetal/ cytometry, has proven a powerful combination for unrav-
neonatal development and persists in the adult through eling much of the complexity of this process. It is daunt-
self-renewal.392 Furthermore, these cells express a relatively ing to attempt to predict where new advances in the field
restricted BCR repertoire that is dependent on antigen will come, but undoubtedly, the completion of the genome
selection.23 Finally, the pre-BCR selection phase of several sequence for both mouse and human will provide impetus
heavy chains abundant in CD5 B cells appears to follow dif- to large-scale gene profi les of B-cell development, as have
ferent rules than classical pre–BCR-mediated expansion.263 already begun.586,587 A complementary approach to gene tar-
In fact, the process of development in mouse fetal liver may geting based on such profi ling will involve characterizing
predominantly follow this alternative pathway.406 Thus, in new mutant mice generated by chemical mutagenesis.588,589
mouse, fetal development, culminating in production of Considering the recent unanticipated discoveries of AID in
B1 B cells, may represent a type of alternative or “primitive” isotype switching and BAFF in peripheral B-cell develop-
B-cell development, as has been proposed previously.585 It ment, it seems likely that the coming decade will provide
is less clear whether a similar distinction exists in human many surprises. A goal will be to eventually understand how
B-cell development, as data are lacking. Nevertheless, one the interplay of the innate and adaptive immune systems
can envision a primordial pathway, randomly combining generates protective responses while avoiding autoimmune
heavy and light chains and simply selecting cells that express pathologies at the organism level.

B-Lymphocyte Receptors, Signaling

CHAPTER
9 Mechanisms, and Activation
Akanksha Chaturvedi • Angel Davey • Wanli Liu • Hae Won Sohn • Susan K. Pierce
INTRODUCTION Antigen binding to the BCR triggers several signaling

The primary function of mature B lymphocytes in response cascades that lead to the transcriptional activation of a va-
to foreign antigens is to proliferate and differentiate into an- riety of genes associated with B-cell activation (see Fig. 9.1).
tibody-producing plasma cells that secrete large quantities The BCR is also an internalizing receptor; under the influ-
of antibodies. Effective antibody responses are highly spe- ence of the signaling cascades, the BCR and bound antigen
cific and of high affinity for the inducing antigen and of the are actively endocytosed into the cell and trafficked to spe-
appropriate isotype to allow the antibodies to carry out the cialized intracellular compartments in which the antigen is
effector functions, as described in Chapter 5, that best serve processed and presented on major histocompatibility com-
to eliminate an antigen-containing pathogen. For B cells plex (MHC) class II molecules, as described in Chapter 22.
to produce the most effective antibodies, they must have The MHC-peptide complexes then move to the B-cell sur-
mechanisms that allow them to discern their specificity and face where they activate antigen-specific helper T cells to
affinity for antigens and to produce plasma cells that secrete provide T-cell help, a complex process involving both direct
antibodies of the appropriate isotype. In this chapter, we interactions between B cells and T cells as well as the release
explore what we know and are learning about these mecha- of soluble cytokines by the T cells.
nisms. Understanding how B cells recognize and respond to A considerable amount of what we know about the
antigens will aid in our efforts to develop effective vaccines biochemical nature of the signaling pathways triggered by
and to target therapies to block hyper-B-cell responses as in antigen binding to the BCR comes from studies in which
systemic autoimmune diseases and in some B-cell tumors. multivalent antigens are provided to the B cell in solution
B cells are activated by antigen by a process of clonal se- (see Fig. 9.1). Although B cells can respond to antigens in
lection, a fundamental feature of the adaptive immune resolution, a variety of recent studies suggest that antigens
sponse. Each B cell expresses a membrane form of a single are presented to B cells on the surface of antigen-presenting
heavy (H) chain (VHCH) and a single light (L) chain (VLCL) cells (APCs) including macrophages and dendritic cells
that are assembled together with a noncovalently bound im- present in the local microenvironments of the lymph nodes
munoglobulin (Ig) α –Igβ heterodimer into an antigen re- and spleen in which B cells are activated in vivo (see Fig. 9.1).
ceptor, the B-cell receptor (BCR), which is expressed on the Antigens may be presented on macrophages and dendritic
B-cell surface (Fig. 9.1). It is conservatively estimated that cells as part of complement-fi xed complexes bound to the
during a lifetime, an adult human expresses more than 1013 complement receptors or immune complexes bound to FcRs
unique clonally distributed BCRs. This estimate is based, in expressed by these cells. We do not yet know if the molecular
part, on the diversity of H and L chain variable (V) regions mechanisms by which antigens in solution and antigens on
that can be generated given the germ line encoded V gene cell surfaces initiate signaling are identical, even though it
segments and the molecular combinational and mutational appears in both cases that the engagement of antigen by the
mechanisms that create the V genes described in Chapter 6. BCR results in the triggering of similar signaling cascades.
When an antigen enters the immune system, it selects from In this chapter, we describe what we are learning about the
among this extraordinarily large array those B cells whose events that occur following the binding of the BCR to anti-
receptors fit it best and signal these to proliferate and dif- gen that trigger signaling cascades that lead to B-cell activa-
ferentiate into antibody-secreting plasma cells. Even very tion. The question that will be addressed is how is the BCR’s
simple antigens are predicted to find hundreds of B cells that specificity and affinity for an antigen read by the BCR and
express BCRs with sufficient specificity and affinity to be ac- transduced across the B cell’s membrane to trigger intracel-
tivated. This process of antigen selecting best fits from among lular signaling cascades that induce the B cell’s response. The
the enormous array of preexisting B cells to proliferate and signaling cascades that are set off upon antigen binding have
differentiate into antibody-secreting cells is a complicated been described in considerable biochemical detail and will be
process that takes place in highly specialized microenviron- reviewed in this chapter. However, biochemical analyses do
ments in the secondary lymphoid organs and involves the not capture the spatial and temporal dynamics of the events
functions of both T cells and innate immune system cells, as that are triggered by antigen binding to the BCR. The recent
will be described in Chapter 10. Here, we focus on the events applications of live cell imaging technologies are providing the
that follow antigen binding to the BCR that initiate signal- first views of B cells as they encounter antigen with the resolu-
ing, an essential, critical first step in B-cell activation. tion to follow individual BCRs. These images are showing us
246

CHAPTER 9 B-LYMPHOCYTE RECEPTORS, SIGNALING MECHANISMS, AND ACTIVATION | 247
FIG. 9.1. Overview of the Activation of B

Cells by Antigen. B-cell activation is initi-
ated by the binding of antigen to the B-cell
receptor (BCR). B cells encounter anti-
gen either in solution or on the surface of
an antigen-presenting cell (APC), likely as
complement coupled antigens or as immune
complexes bound to APC complement
receptors or FcRs. Antigen binding triggers
signaling cascades that lead to the activa-
tion of a variety of genes associated with
B-cell activation. The BCR is also a traffick-
ing receptor; under the influence of BCR
signals, it transports antigens into intracel-
lular compartments where the antigens are
processed and presented on major histo-
compatibility complex class II molecules for
recognition by helper T cells.
that the activation of B cells by antigen is far more dynamic of BCR signaling is also influenced by the environmental
than originally thought. It has been possible to infer from the context in which the antigen is encountered. Antigens enter
behavior of the B cells and BCRs in these images the mecha- the immune system in various contexts as relatively simple
nisms underlying antigen-driven events that lead to B-cell ac- vaccines to complex microorganisms including viruses, bac-
tivation. Such images are changing our view of BCR-mediated teria, and parasites. Both vaccines and pathogens bring with
B-cell activation and are providing the tools to gain an un- them materials that can activate B cells through coreceptors
derstanding of the mechanisms underlying diseases that result other than the BCR as well as activate T cells and cells of
from inappropriate B-cell activation including systemic auto- the innate immune system to secrete products that bind to
immune diseases and certain B-cell tumors. B cell coreceptors that influence how the BCR signals in re-
As we consider the mechanisms by which antigens trig- sponse to the antigen. The molecular form of the antigen
ger B cells through BCRs, we need to be aware that B cells itself can influence the outcome of BCR antigen binding, as
encounter antigen in both developmental and environmen- described in Chapters 10 and 23. Bacteria and some viruses
tal contexts, and that these contexts greatly influence the display rigid arrays of antigens on their surface that induce
outcome of antigen engagement. Consider, for example, that antibody responses in the absence of helper T cells (coined
B cells express BCRs throughout their development from T-independent antigens), as do polysaccharides on bacteria
immature B cells to memory B cells. However, engagement in which the carbohydrate moieties are arrayed as multimers.
of the BCR by antigen at different developmental stages Thus, in predicting the outcome of BCR-antigen binding, we
has different outcomes. The binding of self-antigens to the need to consider not only the antigen’s interaction with the
BCRs expressed on immature B cells signals these cells, but BCR but also the developmental state of the B cells, the en-
the signals do not result in activation. This means that the vironment in which the B cell is activated, and the nature of
developmental state of the B cell dictates the outcome of the the antigen itself. In other words, BCR signaling always oc-
BCR’s engagement of antigen, either driving proliferation curs in a context; to predict the outcome of antigen binding
and differentiation in mature B cells or triggering mecha- to the BCR, we need to be aware of that context.
nisms that lead to elimination or silencing of self-reactive The BCR also signals during development to provide sur-
immature B cells, as described in Chapter 8. The basic vival signals to the B cell, often referred to as tonic signal-
structure of the BCR does not change from the immature ing, and to keep B cells in nonresponsive or tolerant states,
to the mature B-cell stage, but the BCR signals differently termed anergic signaling; however, we do not yet know how
in different developmental cellular contexts. The outcome these signals relate to those that activate mature B cells.

Concerning the impact of the environmental context in

which B cells are activated, this chapter will focus on the
effect of B-cell coreceptors that directly affect BCR signal-
ing. We will provide a comprehensive list of the coreceptors
and what they respond to and describe how two of the best
studied coreceptors, the enhancing cluster of differentiation
(CD)19/CD21 complex and the inhibitory Fc γRIIB receptor,
function to influence B-cell responses. We will also com-
ment on the interactions between the innate immune sys-
tem’s toll-like receptors (TLRs) and the BCR that appear to
serve to regulate each other’s signaling.
We hope that this chapter leaves the reader with a clear
view of the early events that follow the engagement of anti-
gen by the BCR and the signaling cascades that are triggered
that ultimately activate B cells to proliferate and differentiate
into antibody-secreting cells. We also hope that the reader
gains an appreciation that B-cell activation occurs in both a
developmental and environmental context that dictates the
outcome of antigen binding to the BCR. Lastly, we consider
the repercussions of uncontrolled BCR signaling that may re-
sult in B-cell tumors and in systemic autoimmune diseases.
THE STRUCTURE OF THE B-CELL RECEPTOR

The BCR belongs to the multichain immune recognition re-
ceptor (MIRR) family that includes the T-cell receptor for
antigen and the high-affinity receptor for IgE. MIRR family
members contain ligand-binding chains, which for the BCR is
a membrane form of Ig (mIg) (Fig. 9.2). B cells express BCRs
composed of Igs of all isotypes that are expressed in a develop-
mentally controlled fashion beginning with IgM-BCRs in im- FIG. 9.2. The Structure and Organization of the B-Cell Receptor
mature B cells, IgM- and IgD-BCRs in mature B cells, and then (BCR). The immunoglobulin (Ig)M BCR is composed of a membrane
isotype switched BCRs, containing IgGs, IgAs, and IgEs in form of IgM associated with a covalently linked heterodimer of Igα
memory B cells. The mIgs have short cytoplasmic tails of 3 to and Igβ that contain in their cytoplasmic domains immunoreceptor
28 amino acids that, with the exception of IgG- and IgE-BCRs, tyrosine-based activation motifs. The BCR is depicted compartmen-
do not connect directly with the cell’s signaling apparatus.1–3 talized on the plasma membrane by actin cytoskeleton “fence”9 as
Rather, the MIRRs ligand binding chains noncovalently asso- proposed by Batista et al.21 Two different views are provided: a side
view showing the BCR short cytoplasmic tail and the Igα-Igβ tails in
ciate with membrane proteins that contain immunoreceptor the cytosol and a view from above showing the BCRs compartmen-
tyrosine-based activation motifs (ITAMs) in their cytoplasmic talized by actin “fences.”
tails. For the BCR, the mIg associates with a disulfide linked
heterodimer, Igα and Igβ, each chain of which has a single
ITAM in its cytoplasmic tail (see Fig. 9.2).4 The stoichiometry heterodimer with the existing structure of the Cα4 domain.8
of the BCR complex is 1mIg:1Igα-Igβ determined by biochem- These results predicted an unexpectedly extensive contact
ical analysis as well as by live cell imaging (see Fig. 9.2).5,6 surface between the ectodomains of mIg and both Igα and
In considering how the BCR’s engagement of antigen Igβ through multiple charged residues that could potentially
triggers signaling, it is helpful to consider what we know be sensitive to antigen-induced changes in the BCR’s mIg.
about the structure of the BCR. From x-ray crystallographic In considering how the BCR’s engagement of antigen
studies, we know that the antibody Fab does not undergo triggers signaling, it is also useful to consider the organiza-
large conformational changes between the free and antigen- tion of the plasma membrane in which the BCR resides (see
bound states that could transduce the information that the Fig. 9.2). Our current understanding of the plasma mem-
BCR has bound antigen to the BCR’s cytoplasmic domains brane is that it is not simply a fluid mosaic structure of freely
to initiate signaling.7 However, the BCR is a complicated diffusing proteins in a phospholipid bilayer, but rather, the
multichain complex and to date, there is no structure of a plasma membrane appears to be partitioned into highly dy-
complete BCR containing mIg, Igα , and Igβ. Thus, a full un- namic compartments formed by actin cytoskeleton “fences”
derstanding of the molecular basis of the antigen-induced and actin-anchored protein “pickets.”9,10 These compart-
initiation of BCR signaling will await the determination of ments serve to organize the plasma membrane to control
the structure of the BCR. Recently, as an effort toward this the diffusion of membrane proteins and concentrate pro-
goal, the structure of the disulphide-linked homodimer of teins in one compartment or conversely segregate proteins
Igβ was solved, which allowed the modeling of an Igα –Igβ into separate compartments. The fences and pickets are also

dynamic structures that can disassemble and reassemble activated by antigens expressed on the surfaces of APCs in
during B-cell activation. vitro resulting in the formation of a polarized bull’s eye–like
structure in which the BCRs are concentrated in the center,
surrounded by the adhesion molecule lymphocyte function–
THE ANTIGEN-INDUCED CLUSTERING OF THE associated antigen, which engages intercellular adhesion
B-CELL RECEPTOR molecule on the APC surface.13 Recent intravital imaging
Like other MIRRs, the BCR has no intrinsic receptor kinase studies showed that B cells interacted with antigens on the
activity, but upon antigen binding, the ITAM tyrosines in the surfaces of APCs in lymph nodes in vivo.14–18 Antigens in lym-
Igα and Igβ chains are phosphorylated by the membrane- phatic fluid enter lymph nodes through efferent vessels and
tethered kinase Lyn.11,12 The phosphorylation of the BCR gain access to B cells through various mechanisms.19 Small
ITAMs leads to the recruitment of the Src homology 2 (SH2) soluble antigens move through follicular conduit networks
domain-containing kinase Syk and the initiation of a variety of and are presented to B cells within the follicles. Particle-like
downstream signaling pathways described later in this chap- antigens, including viruses and immune complexes, are cap-
ter. The focus of this section is on the early events following tured by macrophages lining the subcapsular sinuses and are
antigen binding that lead to the phosphorylation of the BCR. then transported into the cortex of the lymph node where
Until recently, most of what we learned about the re- they are presented to B cells. In addition, B cells are also able
sponses of B cells to antigens came from biochemical stud- to engage antigens on dendritic cell surfaces in the lymph
ies of B cells responding to antigens in solution. Such studies nodes.20 These findings provided a new view of the initia-
showed that BCR signaling can only be initiated by the bind- tion of antigen-driven BCR signaling in which BCR signal-
ing of multivalent antigens. As an example, only the bivalent ing is initiated at the contact interface between B cells and
F(ab′)2 fragments, but not the monovalent Fab fragments, of APCs. Live cell imaging technologies are providing the tools
anti-IgM antibodies trigger BCR signaling.12 By immunoflu- to observe B cells as they first engage antigen on membrane
orescence imaging, following multivalent antigen binding, surfaces. To facilitate these studies and gain high-resolution
BCRs were observed to form microscopic clusters or patches images, B cells are often activated by antigens incorporated
on the cell surface that then move to one pole of the B cell into fluid planar lipid bilayers as surrogate APCs. These
to form a cap. Based on these biochemical studies, a widely studies are revealing the B cells’ engagement of membrane-
accepted concept emerged that the physical aggregation of bound antigens to be a remarkably dynamic event 21,22 that
the BCR by multivalent antigens promoted the patching and contrasts with the view of the patching and capping of BCRs
capping of BCR that initiated signaling. that resulted from the simple physical cross-linking of BCRs
However, there is growing evidence both from studies in by multivalent antigens in solution.
vitro as well as from intravital imaging in live animals that B cells first touch the antigen-containing bilayer through
B cells are activated by membrane-bound antigens and not finger-like protrusions of their plasma membrane from which
by antigens in solution. B cells were shown to be efficiently the BCRs engage antigen and form microclusters (Fig. 9.3).
FIG. 9.3. The Activation of B Cells on Antigen-Presenting Cell (APC) Surfaces. The initial contact of the B cell with an antigen-containing
APC surface is through finger-like protrusions of the membrane. Antigen binding induces B-cell receptor (BCR) clustering and signaling
that triggers the B cell to spread over the APC surface, forming additional BCR clusters as antigens are encountered. After maximal
spreading, the B cell contracts, actively moving BCR clusters to the center of the contact area ultimately forming an immune synapse.

These BCR microclusters are enriched in tyrosine phosphor- and coreceptors.29 The actin cytoskeleton may also segre-
ylated proteins and are the site of Lyn and Syk recruitment gate BCRs from kinases and phosphatases at steady state.
and thus appear to be the elemental signaling units. B cells Current evidence suggests that the cytoskeleton fences and
subsequently spread over the bilayer allowing the engage- pickets confine the BCR and inhibitory phosphatases to the
ment of additional antigens (see Fig. 9.3). Following maximal same areas, and that BCR signaling serves to disrupt the cy-
spreading, the BCR-antigen microclusters are actively moved toskeleton and allow the BCRs and phosphatases to diffuse
toward the center of the contact area, and the B cell contracts away, promoting B-cell activation.
to form an immunologic synapse. This dynamic process of A third model, the dissociation activation model, is based
touching, spreading, and contraction occurs within minutes on biochemical studies.6,30 In this model, in resting B cells
of the BCR’s first contact with the bilayer.23 These observa- in the absence of antigen most BCRs exist on the B-cell
tions bring us back to the fundamental questions: How are surface as signaling-inactive, autoinhibited oligomers in
BCR microclusters formed, and how do BCR microclusters equilibrium with a small number of signaling-active BCR
trigger BCR signaling? Our understanding of these processes monomers.30 The binding of multivalent antigen disrupts
is still incomplete. However, based on current data, several the oligomers shifting the equilibrium toward BCR active
models have been proposed that address these questions. monomers, which are then clustered by the multivalent an-
One model, the conformation-induced oligomerization tigen in a manner that prevents the formation of inactive
model, is based on evidence from live cell imaging at the oligomers within the cluster. This model also accounts for
level of individual BCRs. Single BCR tracking provided evi- the ability of structurally diverse antigens to activate B cells
dence that BCRs are dispersed and freely diffusing in the by proposing that antigens keep BCRs apart rather than
plasma membrane of resting B cells. Upon binding antigens bringing them together into well-ordered oligomers.
presented on lipid bilayers, BCRs formed immobile oligo-
mers that grew into microclusters by trapping additional
antigen-bound BCRs.24 Remarkably, contrary to soluble an- THE ROLE OF B-CELL RECEPTOR AFFINITY AND
tigens, monovalent antigens presented on fluid lipid bilayers, ISOTYPE IN SIGNALING
which could not physically cross-link BCRs, activated B cells Affinity maturation and class switching are two hallmark
equivalently to multivalent antigens. This observation led to features of humoral immune responses.31 In a typical T-cell–
the conclusion that the physical cross-linking of BCRs by dependent antibody response, antigen-specific antibodies
multivalent antigens was not a requirement for BCR oligo- become increasingly higher in affinity and predominantly
merization or clustering. Early oligomerization and micro- of the IgG isotype through the linked molecular processes
cluster formation that occurred within 60 seconds of antigen of somatic hypermutation and class switching. Because the
binding were shown to be BCR-intrinsic events that did not B-cell immune response functions through clonal selection,
require the signaling apparatus of the BCR.25 Mutagenesis it is presumed that affinity maturation and class switching
studies showed that the membrane-proximal Cμ4 portion of reflect an advantage of B cells expressing high-affinity, class-
the ectodomain of mIgM (or the Cγ 3 membrane-proximal switched BCRs in the selection process. Indeed, adoptive
domain of mIgG) was both necessary and sufficient for BCR transfer studies provided clear evidence that high-affinity
oligomerization and signaling.25 Thus, the membrane proxi- B-cell outcompete low-affinity B-cell clones for survival in
mal domain of the mIg appeared to contain an oligomeriza- vivo.32–38 Similarly, when comparing IgM and class-switched
tion domain that was not exposed in the absence of antigen BCRs, several seminal studies showed that class-switched B
binding. It was hypothesized that the B cell’s binding antigen cells outcompete IgM B cells for survival in vivo, and that
presented on an APC surface exerted a force on the BCR that this survival advantage can be attributed to the 28 amino acid
revealed the oligomerization domain, allowing oligomeriza- cytoplasmic tail of mIgG,1,2 missing in mIgM and mIgD. The
tion of antigen-bound BCRs as they randomly bumped on question is at what point in the antigen-selection process does
the membrane. This model provides a mechanism by which a high-affinity, class-switched BCR gain an advantage?
the universe of foreign, structurally distinct antigens can Biochemical experiments suggested that the signals that
bring BCRs into a precise signaling-active oligomer by the are triggered through high-affi nity BCRs are qualitatively
monovalent binding of BCRs to epitopes on APC-presented different from signals through low-affinity BCRs.39 At an
antigens. This model also suggests how BCRs are able to so earlier point in B-cell activation, the affinity of the BCR for
exquisitely discriminate their affinity for antigen by mon- antigen was shown to determine the degree to which B cells
ovalent binding, avoiding the affinity obscuring effects of spread over antigen-containing fluid lipid bilayers, allowing
the avidity contributed by multivalent antigen binding to increased accumulation of antigen into the immune synapse
bivalent BCRs. and subsequent enhanced responses.23 The IgG-BCR cyto-
Another model for the initiation of BCR signaling based plasmic tail is responsible for the enhanced B-cell prolifera-
on live cell single molecule imaging focuses on the role of tive responses to antigen in vitro and signaling cascades that
the actin cytoskeleton in signaling.26–29 In this model, the have been shown to be qualitatively different from those
membrane cytoskeleton fences and pickets restrict BCR triggered by IgM-BCRs.3,40–42 Enhanced signaling through
mobility and interactions with signaling molecules and IgG-BCRs is dependent on the phosphorylation of a tyrosine
coreceptors. BCR signaling results in the disruption of in the IgG tail that serves to recruit growth factor-recep-
the cytoskeleton barriers and increases the likelihood that tor-bound protein 2 (Grb2), resulting in sustained kinase
the antigen-engaged BCR will encounter activated kinases activation and enhanced B-cell proliferation.3

Recent studies using high-resolution imaging to follow is essential to trigger and regulate downstream signaling. The
single BCRs showed that the earliest events that occur follow- importance of these PTKs in the B-cell signaling is under-
ing antigen binding to the BCR are also highly sensitive to scored by the fact that deficiencies in any one of the three
both the affinity of the BCR for antigen and the isotype of the families result in aberrant B cell development and function
BCR. Comparing BCRs that differed 50-fold in their affin- (Table 9.1). The first kinases that are activated following BCR
ity for antigen, it was observed that high-affinity BCRs more cross-linking are the Src family PTKs, primarily Lyn, but
readily formed BCR microclusters that grow more rapidly, also Blk and Fyn followed by the activation of Syk and the
thereby resulting in larger microclusters that recruit more Tec family kinase, Btk45 (Fig. 9.4). ITAM phosphorylation of
Syk and signal for more robust calcium responses.24 These Igα-Igβ by Lyn generates phosphotyrosine motifs that allow
imaging experiments also demonstrate that IgG-BCRs are the binding of the SH2 domains (Box 9.1) of the second ki-
dramatically enhanced in their ability to oligomerize and to nase, Syk, resulting in rapid Syk activation. Upon binding to
grow microclusters, ultimately leading to increased recruit- phosphorylated ITAMs, Syk undergoes autophosphorylation
ment of Syk and more robust calcium responses as compared at multiple tyrosines within its linker regions that not only
to IgM-BCRs of the same affinity.43 The enhanced function prolongs Syk’s activation but also creates SH2 binding sites
of IgG-BCRs was mapped to a novel membrane proximal 15 on Syk for the recruitment of downstream signaling mol-
amino acid region of the cytoplasmic tail. These studies place ecules, including PLC-γ 2,46 leading to a positive feedback of
the effects of both affinity maturation and class switching at BCR signaling and the concomitant influx of calcium.12 Btk
the earliest steps in the initiation of BCR signaling. is the third PTK that is activated upon BCR cross-linking.47
The importance of Btk in BCR signaling for the development,
HOW B-CELL RECEPTOR CLUSTERING activation, and differentiation of B cells is underscored by the
TRIGGERS SIGNALING fact that the loss-of-function mutations of the gene encoding
Btk lack circulating B lymphocytes, are unable to generate Igs,
Another fundamental yet still open question related to the ini- and cannot mount humoral immune responses.48 This pri-
tiation of BCR signaling is how BCR clustering in response to mary immunodeficiency is named X-linked agammaglobu-
antigen recruits Lyn to phosphorylate the antigen-bound BCRs linemia.49,50 Similarly, a spontaneous mutation in the mouse
and triggers signaling. One simple explanation is that Lyn is Btk gene leads to X-linked immunodeficiency.51 Btk consists
constitutively associated with the cytoplasmic domains of some of multiple protein domains including PH, SH2, SH3, and
monomeric BCRs, but only phosphorylates BCRs in trans when kinase domains (see Box 9.1), which define its subcellular
the BCRs are clustered. However, recent studies suggest that the location and regulate its activity. For Btk activation, plasma
mechanism for recruitment of Lyn may be more complicated. membrane localization is important, which is governed by
Lyn is lipidated and tethered to the inner leaflet of the plasma the interactions between the PH domain of Btk with phos-
membrane. BCR clustering has been shown to perturb the phatidylinositol (PI)(3,4,5)P3, the product of PI3K activity
local lipid environment leading to the transient coalescence of and between the SH2 domain of Btk with phosphorylated
lipid rafts around the BCR oligomers followed by a more stable BLNK, an adaptor protein. Mutation in the PH domain of Btk
association of the BCR microcluster with lipid tethered Lyn.44 (R28C) leads to classical X-linked agammaglobulinemia,52
Thus, the BCR’s perturbation of the membrane may serve to substantiating the importance of plasma membrane local-
recruit Lyn to the BCR cluster. BCR clustering has also been ization for Btk activation. Following BCR cross-linking, Btk
shown to alter the way the Igα and Igβ chains’ cytoplasmic translocates from the cytosol to the plasma membrane where
domains associate. In the absence of antigen, the domains are it is activated by phosphorylation at Y551 in its catalytic do-
in close proximity in a “closed” conformation and upon anti- main by Lyn53–55 followed by an autophosphorylation of its
gen binding, the domains “open”; the opening is simultaneous SH3 domain.54,56 Once activated, Btk triggers a cascade of sig-
with Lyn’s phosphorylation of the ITAMs.5 Thus, BCR cluster- naling events that culminate in calcium mobilization through
ing may serve to reveal the ITAMs for phosphorylation by Lyn. phosphorylation of PLC-γ 2, cytoskeletal rearrangements, Vav
activation, and transcriptional activation involving NF-κB.
B-CELL RECEPTOR–TRIGGERED
SIGNALING CASCADES
Amplification of B-Cell Receptor Signaling—
BCR signaling is a multistep process that involves the initia-
Recruitment of Adaptors
tion of signaling by the activation of protein tyrosine kinases
(PTKs), serine-threonine kinases, and lipid kinases; ampli- BCR-mediated signaling is a complex process in which each
fication of signaling by recruiting adaptors; generation of phosphorylation event is linked with another in a regulated
second messengers; and finally activation of the transcrip- manner, thereby generating a large number of protein–
tion of genes involved in B-cell responses. protein interaction networks ultimately resulting in the for-
mation of a large, well-ordered structure often referred to as
a signalosome. Interactions between signaling networks are
Initiation of B-Cell Receptor Signaling—Protein regulated by a number of scaffolding or adaptor proteins,
Tyrosine Kinase Activation which regulate BCR-mediated signaling cascades not only
Following BCR clustering, three different families of PTKs, by recruiting multiple signaling intermediates to the proper
Src, Syk, and Tec, are activated sequentially. This sequential location but also by controlling interactions between signal-
activation of the members of the three different PTK families ing components. Recent studies have identified the roles of

TABLE 9.1 The Phenotype of Mice Deficient in Key B-Cell Signaling Components
Target Protein Major Phenotype Reference
184
Igα cytoplasmic domain Normal pre-B-cell development; completely impaired mature B-cell development
185
Igβ Complete block at the pro-B-cell stage
186
Igμ Deletion of mature B cells; increased Fas expression
187
Lyn Normal B-cell development in the bone marrow; reduced number of peripheral B cells;
increased proportion of immature B cells; enhanced BCR induced ERK activation and
hyperproliferative responses
188,189
Syk Block in transition of the pro-B- to the pre-B-cell stage; intact Igα-Igβ ITAM
phosphorylation in remaining B cells; abolished BCR-induced calcium influx;
failure to transmit downstream signals
190–193
Btk Reduced number of peripheral B cells; increased immature B cells; complete loss of B1 B
cells; fail to respond to TI-II antigens
194–196
BLNK Block in transition of the pro-B- to the pre-B-cell stage; incomplete block in B-cell
development; fail to respond to both TD and TI antigens
197,198
PLC-γ2 Decreased mature B cells; block in pro-B-cell differentiation; B1 B–cell deficiency; block
in BCR-induced calcium influx and proliferation
82
BCAP Reduced number of B cells; B1 B–cell deficiency; reduced serum IgM and IgG3 levels;
abolished TI antibody response; reduced BCR induced calcium influx and proliferation
199,200
PI3K, p110 subunit Reduced numbers of B1 and marginal zone B cells; reduced serum Ig levels; defective
primary and secondary response to TD antigens; diminished response to TI-II antigens;
reduced BCR, CD40, and LPS induced proliferation
201,202
PI3K p85 subunit Reduced number of peripheral B cells and B1 cell; reduced serum Ig levels; reduced
BCR- and CD40-induced proliferative response; abolished TI antibody response
203,204
PKC-β Failure to activate IKK and degrade IκB; failure to upregulate NF-κB–dependent survival
signals; impaired humoral immune responses and reduced cellular responses
89
Bam32 Normal B-cell development; impaired TI-II antibody responses; reduced responses to BCR
cross-linking
205,206
Vav1/2 Reduced number of B cells; defects in formation of germinal centers and class switching;
defective immune responses against TD and TI antigens; impaired BCR-induced calcium
influx and proliferation
207
Rac Reduced number of mature and marginal zone B cells and B1-a cells and IgM secreting
plasma cells; increased number of peripheral B cells in blood; reduced serum IgM and
IgA concentration
191,208,209
CD19 Defective response to TD antigens; failure to form germinal centers and undergo affinity
maturation; lack of B-1, marginal zone B cells; reduced BCR- and CD40-induced
proliferative responses
210,211
CD22 Decreased surface IgM levels; augmented BCR-induced calcium; compromised marginal
zone B-cell compartment
212
PIRB Increased number of peritoneal B1 cells; constitutive activation of follicular B cells;
increased BCR-induced proliferation
102
Calcineurin Reduced number of B1 cells; reduced plasma cell differentiation; decreased TD antibody
response; BCR-induced proliferation defects
213,214
IKKα Reduced mature B-cell population; impaired basal and Ag-specific Ig production; disrupted
splenic architecture including germinal center formation; decreased expression of
NF-κB target genes
215
IKKβ Disappearance of mature B cells
215
NEMO Disappearance of mature B cells
BCR, B-cell receptor; CD, cluster of differentiation; Ig, immunoglobulin; ITAM, immunoreceptor tyrosine-based activation motif; LPS, lipopolysaccharide; TD, T-dependent; TI, T-independent.
adaptors in B-cell activation, leading to important insights macromolecular complexes are assembled, allowing both spa-
into how they integrate BCR signaling. Although adaptors tial and temporal regulation of signaling cascades. To illus-
generally lack any enzymatic activity, they consist of multiple trate the importance of adaptors, we describe the functions of
protein–protein or protein–lipid interaction domains, includ- three: B-cell linker protein (BLNK); B-cell adaptor for PI3K
ing SH2, SH3, PH and PX homology domains (see Box 9.1). (BCAP), and B-cell adaptor molecule of 32 kDa (Bam32).
Each of these domains has the potential to interact with a
number of proteins that are critical to amplify BCR signaling BLNK—Integrating Protein Tyrosine Kinases and PLC-g 2
by facilitating the coupling of multiple downstream signal- BCR clustering activates Lyn and Syk that regulate a variety
ing pathways. Essentially, the adaptors define where and when of effectors including PLC-γ 2, an essential phospholipase for

FIG. 9.4. Schematic Depiction of B-Cell Receptor (BCR) Signaling. Presented is a highly simplified version of the signaling pathways
triggered by antigen-induced BCR clustering. Antigen binding to the BCR leads to the formation of signalosome consisting of BCR,
protein tyrosine kinases, adaptors, and effector molecules. BCR clustering results in PI3K activation, leading to the production of PIP3
from PIP2 and PLC-γ2 activation resulting in the production of IP3 and diacyl glycerol (DAG) from PIP2. Binding of IP3 to the IP3 recep-
tor on the endoplasmic reticulum (ER) opens channels in the ER membrane leading to calcium release and depletion of ER calcium.
Calcium depletion is sensed by STIM, a calcium-binding protein, which results in its oligomerization and movement to regions of the
ER membrane in close proximity to the plasma membrane where it interacts with Orai opening the calcium release–activated channel,
resulting in calcium influx. Calcium, DAG, and PIP3 trigger signaling cascades leading to mitogen-activated protein kinase and tran-
scription factor activation and activation of the Akt pathways promoting cell survival and growth.
calcium responses. Although Syk can directly phosphorylate multiple effector molecules facilitating interactions among
PLC-γ 2 in vitro,57 expression of a functional BCR, Lyn, and them and allowing them to phosphorylate and activate their
Syk in nonlymphoid cells did not induce PLC-γ 2 phosphor- respective signaling pathways. One of the best studied ex-
ylation or calcium mobilization,58 leading to the discovery amples of how BLNK bridges multiple signaling components
of the B-cell–specific adaptor protein, BLNK. BLNK (also is provided by the activation of PLC-γ 2. PLC-γ 2 activation
known as SLP-65 or BASH) is a cytoplasmic protein con- is completely abolished in Syk-deficient B cells.60 In ad-
sisting of an N-terminal region containing a short leucine dition, BCR-induced PLC-γ 2 activation is diminished in
zipper motif, a proline-rich region within the middle third Btk-deficient B cells, suggesting that both Syk and Btk are
of the molecule, and a C-terminal SH2 domain. BLNK con- required for PLC-γ 2 activation.61 However, the mechanisms
tains 13 tyrosine residues, of which 6 are in putative SH2 underlying how these two families of PTKs regulate PLC-γ 2
binding motifs and are phosphorylated upon BCR cluster- activation became clear only after the discovery of BLNK.
ing. BLNK is recruited to the clustered BCR by binding After phosphorylation by Syk, BLNK creates binding sites
through its SH2 domain to the phosphorylated non-ITAM for the SH2 domains of both Btk and PLC-γ 2, bringing
tyrosine 204 in Igα . Mice with a Igα Y204 mutation exhibit them into close proximity with each other, and hence facili-
reduced BLNK phosphorylation and calcium fluxes.59 Once tating PLC-γ 2 phosphorylation at Tyr753 and Tyr759 by Btk,
translocated to the plasma membrane, BLNK is phosphory- which is required for PLC-γ 2 activation.62–68 Cells expressing
lated by Syk, which creates docking sites for SH2 domains of a mutant BLNK lacking either Btk or PLC-γ 2 binding sites

BOX 9.1. PROTEIN MODULES INVOLVED IN PIP(4)P to ensure that activated PLC-γ 2 does not run out
SIGNAL TRANSDUCTION of its substrate.70 Once activated, PLC-γ 2 hydrolyzes PI(4,5)
P2 to generate two important second messengers: inosi-
Src homology 2 (SH2) domain. The SH2 domain is a structur- tol trisphosphate (IP3) and diacyl glycerol (DAG). BLNK
ally conserved protein domain present in the Src oncopro- not only couples Btk and PLC-γ 2, but it also binds to the
tein from which its name is derived and in many signaling SH2 domains of Vav and Nck, resulting in activating the
molecules including those involved in B-cell signaling. SH2 mitogen-activated protein (MAP) kinase pathway, cytoskel-
etal rearrangements, and BCR internalization.71
domains bind with high affinity to peptide sequences within
target proteins that contain phosphorylated tyrosine resi- BCAP—Integrating Protein Tyrosine Kinases and PI3K
dues but have no affinity for the unphosphorylated sequence. PI3K is a heterodimeric enzyme consisting of a catalytic
This allows tyrosine phosphorylation to act as a molecular subunit (p110) and a regulatory subunit (p85) that phos-
switch, recruiting SH2 domain–containing proteins to acti- phorylates phosphatidylinositol lipids generating lipid de-
vated receptors and initiating signaling cascades. Important rivatives at the inner leaflet of the plasma membrane that
SH2 domain–containing proteins involved in B-cell signaling serve as second messengers for downstream signaling.72,73
include Lyn, Syk, Btk, PLC-γ2, BLNK, and Grb2. Of particular interest is the generation of PI(3,4,5)P3 from
Src homology 3 (SH3) domain. SH3 domains are protein PI(4,5)P2, which serves as a ligand for the proteins con-
modules that recognize proline-rich sequences, in par-
taining PH domains and tethers these to the membrane.
Although multiple isoforms of PI3K are known, p110δ and
ticular those containing a PxxP motif, which do not require
p85α are the predominant isoforms in B cells. Gene target-
phosphorylation. However, recent studies have suggested ing experiments have established the role of these subunits
that the SH3 domains also bind to nonproline-rich in the development and activation of B cells (Table 9.2).
sequences accounting for diverse functions they mediate. Deletion or inactivation of either of the two subunits of PI3K
Some examples in B cells are Lyn, Btk, and Grb2. demolish BCR-induced phosphorylation of Akt, FoxO, and
Pleckstrin homology (PH) domain. The PH domain is a lipid protein kinase D and results in reduced calcium flux, im-
binding module that recognizes phosphoinositides phos- paired cell cycle progression, and reduced glucose metabo-
phorylated at the 3 position of the inositol ring, the most lism.74–80 Despite the critical role of PI3K in BCR signaling,
important of which is PI(3,4,5)P3, the product of PI3K’s phos- the mechanisms by which the BCR-associated PTKs regulate
phorylation of PI(4,5)P2. This property allows PH domain– the PI3K pathway is unclear. It has been proposed that tyro-
containing proteins to bind to the inner leaflet of the plasma sines in the cytoplasmic tail of CD19, which are phosphory-
lated upon BCR clustering, provide binding sites for the
membrane following PI3K activation where they function.
SH2 domains of the p85 subunit of PI3K, recruiting it to the
PH domain–containing proteins in B cells include Btk, Vav,
plasma membrane and increasing the specific activity of the
Akt, Gab, and Bam32. enzyme. However, the observation that defects in p85α KO
Phox homology (PX) domain. The PX domain is another mice are more severe than those in CD19 KO mice suggested
phosphoinositide-binding domain that was originally identi- that additional adaptor molecules may participate in bridg-
fied in two cytosolic components of the nicotinamide ad- ing BCR-associated PTKs and PI3K, leading to the discovery
enine dinucleotide phosphate (NADPH) oxidase, p40phox of another B-cell–specific adaptor molecule, BCAP.81 BCAP
and p47phox. In a majority of cases, PX domains contain contains multiple tyrosines, which are phosphorylated after
a proline-rich motif, PxxP, in the middle that serves as SH3 BCR stimulation through the concerted actions of Syk and
binding motif, suggesting that PX domains might directly Btk, thereby creating binding sites for SH2 domains of the
interact with SH3 domains as well. PX domains have been p85 subunit of PI3K. BCAP was shown to regulate PI3K
shown to play diverse cellular functions including vesicle
signaling in CD19-deficient B cells, but in BCAP KO mice,
PI3K signaling was largely unaffected in B cells,81,82 suggest-
docking and fusion, cell signaling, and protein sorting. An
ing that in the absence of BCAP, CD19 functions as a major
example in B cells is phospholipase D. adaptor. Indeed, CD19/BCAP double knockout mice show
an almost complete block of BCR-mediated Akt activation
and severe defects in the generation of immature and mature
show significant reduction in calcium activation, further B cells.83 Recently a GTPase, TC21, was also identified as an
suggesting that BLNK acts as a scaffold for bridging Btk adaptor for PI3K activation by binding directly to Igα-Igβ.84
to PLC-γ 2.69 A single BLNK molecule can bind three PLC-
γ 2 molecules, thus BLNK also functions as an amplifier of Bam32—Linking B-Cell Receptors to Mitogen-
BCR signaling.62 Following BCR clustering, both BLNK and Activated Protein Kinase Activation and Cytoskeletal
Btk are also translocated to the plasma membrane bringing Rearrangements
along PLC-γ 2 and allowing it to gain access to its substrate Bam32 bridges clustered BCRs to two important pathways,
PI(4,5)P2 in the inner leaflet of the plasma membrane.66–68 namely, Rac and cdc42, which trigger cytoskeletal rearrange-
In addition to phosphorylating PLC-γ 2 and recruiting it to ment leading to cell adhesion, polarization, and motility
the plasma membrane, Btk also recruits PIP5 kinase to the and to activation of the MAP kinase c-Jun N-terminal ki-
plasma membrane, which catalyzes PI(4,5)P2 synthesis from nase (Jnk) and extracellular signal-regulated kinases (ERK)

TABLE 9.2 B-Cell Coreceptors that Influence B-Cell Receptor Signaling

Signaling Molecules Expressed in
Activating (+) Phosphorylated Recruited upon Humans (H) or
Coreceptor or Inhibitory (-) Motif(s) Ligand(s) Motif Phosphorylation Mice (M) References
CD19 + Tyr residues C3-coated antigen PI3K + Vav + Lyn + H+M 126,127,216
PLC-γ2
FcμR (FAIM3/TOSO) + Ser + Tyr residues IgM PI3K + PLCγ-1 H+M 131,217
FcRL1 (FcRH1) + ITAM Unknown Unknown Tyr kinase? H+M 132,218,219
PIRA + ITAM on associated MHC-1 Unknown M 133,134,220,221
FcRγ
CD45R (B220) + PTP domain CD22? C-terminal Tyr of SFKs H+M 135,136,222,223
CD148 (HPTPh) + PTP domain Unknown C-terminal Tyr of SFKs H+M 137,138
FcγRIIB (CD32B) − ITIM IgG-IC SHIP H+M 139–141,224
CD22 (BL-CAM) − ITIM Sialic acid SHP-1 H+M 223–227
PIRB − ITIM MHC-1 SHP-1 (+ SHP-2?) M 133,220,221,224
ILT-2 − ITIM HLA-A, -B + -G SHP-1 H 224
(MHC-1)
CD72 (Lyb-2) − ITIM CD100, CD5 SHP-1 H+M 224,227,228
CD5 (Ly1) − ITIM CD72 (+ others?) SHP-1 H+M 224,229
PD-1 − ITIM IgSF protein SHP-2 H+M 224
CD66a (BGP-1) − ITIM CD66a SHP-1 + SHP-2 H+M 224
FcRL4 − ITIM Unknown SHP-1 + SHP-2 H 132,219
FcRL2 (FcRH2) − ITAM/ITIM Unknown SHP-1 H 132,153
FcRL3 (FcRH3) − ITAM/ITIM Unknown SHIP + SHP-1 + SHP-2 H 132,154
FcRL5 (FcRH5) − ITAM/ITIM Unknown SHP-1 H+M 132,155
FcRL6 (FcRH6)a − ITIM HLA-DR (MHC-2) SHP-2 M 132,230,231
CD, cluster of differentiation; HLA, human leukocyte antigen; Ig, immunoglobulin; IC, immune-complex; IgSF, immunoglobulin superfamily; ITAM, immunoreceptor tyrosine-based
activation motif; ITIM, immunoreceptor tyrosine-based inhibitory motif; MHC, major histocompatibility complex; PTP, protein tyrosine phosphatase; SFK, Src-family kinase;
SHIP, SH2-containing inositol 5′-phosphatase; SHP, SH2-domain-containing protein tyrosine phosphatase.
a
FcRL6 motif, ligand, and signaling molecule information was obtained from studies of human T and natural killer cells.
that play central roles in the activation of transcription fac- The binding of IP3 to the IP3R activates the receptor and
tors that regulate the expression of genes involved in B-cell opens the channel resulting in the release of calcium from
proliferation, differentiation, and survival.85–88 Following BCR the ER into the cytosol. Depletion of intracellular ER calcium
clustering, Bam32 translocates to the plasma membrane in a stores results in the oligomerization of the ER calcium sen-
PI3K-dependent mechanism through its PH domain88 where sor, STIM1, an ER membrane protein that contains in its cy-
it binds to Rac and cdc42 leading to their activation, trigger- toplasmic domain an EF-hand calcium-binding domain.92,93
ing downstream signaling to Jnk and ERK. Mice deficient in STIM1 oligomers rapidly redistribute into clusters and move
Bam32 show diminished Jnk and ERK phosphorylation.89 Rac to junctions of the ER and plasma membrane where STIM1
activation is diminished in Bam32-deficient B cells, whereas interacts with Orai, a calcium release–activated channel,
overexpression of Bam32 results in enhanced Rac activation.90 and activates Orai to open, triggering the entry of extracel-
lular calcium into the cytosol.94–96 A single missense mutation
Generation of Second Messengers (R91W) in human Orai or a truncation mutation in STIM1
causes severe combined immune deficiency supporting the
As described previously, binding of the PTKs and various
functional roles of STIM1 and Orai in calcium signaling in
signaling intermediates such as PLC-γ 2 and PI3K on various
the immune system.97,98 The increase in cytosolic calcium is
adaptors results in the activation of these enzymes resulting
sensed by various calcium-binding proteins, ultimately lead-
in the generation of second messengers including calcium,
ing to activation of several downstream pathways including
DAG, and phosphoinositides. These second messengers play
activation of protein kinase C (PKC) and transcription factors
a crucial role in regulating BCR signaling by virtue of their
nuclear factor of activated T cells (NFAT) and NF-κB. One
ability to affect downstream signaling initiated by the BCR.
of the major cytosolic calcium-binding proteins, calmodulin,
Calcium as a Second Messenger functions as a calcium sensor by binding to calcium with its
Calcium mobilization is a multistep process that is initi- two EF hands. Calcium-bound calmodulin activates several
ated with the activation of PLC-γ 2 that hydrolyzes PI(4,5)P2 molecules including calcineurin, which is a phosphatase that
present in the inner leaflet of the plasma membrane to gener- dephosphorylates NFAT resulting in NFAT’s translocation
ate IP3 and DAG. A receptor for IP3 (IP3R) is present as a tet- into the nucleus where it activates the transcription of tar-
ramer on the ER membrane that forms a calcium channel.91 get genes. One of the targets of NFAT in B cells is interferon

regulatory factor 4, a critical transcription factor for the dif- position generating PI(4,5)P2, and SH2-domain-containing
ferentiation of plasma cells.99–101 Indeed, calcineurin-deficient inositol 5′ phosphatase (SHIP), which generates PI(3,4)P2 by
B cells show reduced plasma cell differentiation and decreased removing a phosphate from the five position and essentially
antigen-specific antibody responses to antigens.102 shuts down signaling. One of the pathways that activates SHIP
in B cells is the ligation of the inhibitory FcR FcγRIIB,122 as
Diacylglycerol as Second Messenger described in the next section of this chapter.
The second product of PI(4,5)P2 hydrolysis by PLC-γ 2 is DAG
that together with calcium activates PKC-β that regulates the
NF-κ B pathway. NF-κ B in resting B cells is retained in the The Spatial and Temporal Dynamics of B-Cell
cytoplasm by binding to its inhibitor, Iκ B. BCR clustering in- Receptor Signaling Cascades
duces Iκ B phosphorylation, through a PKC-β –initiated path- Advanced fluorescence-based imaging techniques are just
way, ubiquitination, and subsequent degradation, allowing now providing the first view of the spatial and temporal dy-
translocation of NF-κ B into the nucleus.103–105 Aberrant acti- namics of the recruitment of individual signaling molecules
vation of NF-κ B has been linked to defective B-cell activation to BCR microclusters following BCR oligomerization in re-
leading to multiple immune disorders. Ablation of PKC-β, sponse to membrane-bound antigens. Each newly formed
which leads to the defective activation of the NF-κ B path- BCR cluster recruits the earliest PTKs, Lyn and Syk24,123,124 ; as
way, results in defective B-cell activation and maturation.106 the BCR moves to form an immune synapse, Syk is lost and
In humans, mutations in components of the NF-κ B pathway PI3K is recruited. It has also recently been possible to use im-
are associated with the formation of lymphomas.107–113 In ad- aging to follow signaling as the BCR is internalized into the B
dition to activating PKC-β, DAG also recruits the DAG bind- cell. Such studies showed that BCR signaling is initiated at the
ing Ras guanine nucleotide releasing protein 3 (RasGRP3) plasma membrane with the recruitment and phosphorylation
to the plasma membrane, thus initiating the Ras to Raf to of Lyn and Syk and continues after the BCR is endocytosed
MEK to ERK pathway.114 Deletion of the DAG-binding do- and trafficked through early endosomes then to late endo-
main from RasGRP blocks the movement of RasGRP3 to the somes and multivesicular antigen processing compartments
plasma membrane and the activation of Ras. The Ras–Raf– with the recruitment and phosphorylation of downstream
MEK–ERK signaling pathway plays a critical role in antigen- kinases, initially cRaf and subsequently the MAP kinases,
induced proliferation of mature B cells.115 Ras is also involved ERK, p38, and Jnk.125 Internalization of the BCR is neces-
in developing B cells, as transgenic expression of dominant sary for proper signaling as when internalization is blocked,
negative forms of Ras blocks developmental progression to phosphorylation of kinases become dysregulated as does gene
the pre–B cell and immature B-cell stages.116,117 Major tar- transcription. The continued application of imaging technol-
gets of ERK in B cells are Ets-family transcription factors, ogies to describe B-cell activation should lead to an increas-
resulting in the expression of early response genes including ingly detailed spatial and temporal view of BCR signaling.
Egr1 that has been shown to promote expression of adhesion
molecules.118
CORECEPTOR REGULATION OF B-CELL
Phosphoinositides as Signaling Mediators—A Diverse RECEPTOR SIGNALING
Group of Molecules Regulating Diverse Pathways The outcome of signaling through the BCR is carefully regu-
PI is unique among membrane lipids as it can undergo related not only by the developmental stage of the B cells but
versible phosphorylation at multiple sites on its inositol head also by a variety of B-cell coreceptors that both promote and
group by lipid kinases and phosphatases to generate a vari- attenuate BCR signaling in response to antigen. Ideally, co-
ety of phosphorylated PI lipids called phosphoinositides. receptors ensure that sufficient amounts of antigen-specific
Phosphoinositides play crucial roles in cell signaling and antibodies are produced to control an infection, and that once
membrane trafficking. As described previously, BCR cluster- controlled, antibody production is turned off. We have pro-
ing leads to the accumulation of the phosphoinositide, PI(4,5) vided a list of B-cell coreceptors and have indicated whether
P2, the substrate for PLC-γ 2, which generates two crucial they function to enhance or inhibit BCR signaling and, when
second messengers, IP3 and DAG. Another important phos- it is known, what they recognize and the molecular basis of
phoinositide is the PI3K-generated PI(3,4,5)P3 that provides a their effect on BCR signaling (see Table 9.2). Presumably, all
ligand for recruiting important PH domain–containing pro- B-cell coreceptors respond to clues from the environment as
teins to the plasma membrane, including Btk, PLC-γ 2, and to the course of the immune response and relate this informa-
Akt. Akt is activated at the membrane by phosphorylation tion to the B cells by enhancing or inhibiting BCR signaling.
and in turn phosphorylates several proteins and transcription The best understood examples of the role of coreceptors in
factors that regulate protein synthesis, cell survival, and pro- B-cell responses are the activating CD19/CD21 complex and
liferation. Some of the targets of Akt in B cells are mamma- the inhibitory FcγRIIB. Both coreceptors become physically
lian target of rapamycin, glycogen synthase kinase 3, forkhead cross-linked to the BCR through the recognition of antigen-
family transcription factors, Caspase-9, and proapoptotic pro- containing complexes. The CD19/CD21 complex binds com-
tein, BAD.119–121 Given the crucial role of PI(3,4,5)P3 in B-cell plement fixed antigens (C3d-modified antigen, as described in
activation, its levels are tightly regulated by two different lipid Chapter 36) through CD21, a complement receptor that binds
phosphatases that oppose the activity of PI3K, phosphatase to C3d. Complement modification of antigens can be viewed
and tensin homolog, which removes a phosphate from the D3 as a sign that the antigen is dangerous, having the ability to

FIG. 9.5. Inhibitory and Activating B-Cell Receptor (BCR) Coreceptors. Antigen-containing immune complexes engage both the BCR
and the inhibitory coreceptor FcγRIIB (Left). Lyn associated with clustered BCRs phosphorylates the immunoreceptor tyrosine-based
inhibitory motif of FcγRIIB, recruiting SH2-domain-containing inositol 5′ phosphatase that dephosphorylates PI(3,4,5)P3, the product of
PI3K activation, to PI(3,4)P2, blocking downstream signaling. Antigen binding to BCRs induces their clustering and signaling that results
in the phosphorylation of CD19 by BCR-associated Lyn, recruiting PI3K and Vav (Right) and enhancing BCR signaling. Enhancement of
BCR signaling is stronger when complement conjugated antigens engage both the BCR and the complement receptor, CD21, recruiting
CD19 to the signaling complex where it is phosphorylated by Lyn recruiting PI3K and Vav (Right).
activate the complement pathway, and thus, the CD19/CD21 enzymatic activity and phosphorylation of Lyn during BCR
complex serves to enhance BCR signaling. The FcγRIIB signaling.128 CD81 may aid in the stable association of the
binds to antigen–antibody immune complexes (described in BCR with saturated lipids that coalesce around the antigen-
Chapter 24) through the Fc portion of the IgG antibody com- bound BCRs to sustain antigen-induced signaling when the
plexed with the antigen. Immune complexes form when an- CD19-CD21 complex is coengaged with the BCR.129 CD81
tigen-specific antibodies reach sufficient levels relative to the and Leu-13 may also be involved in mediating cell–cell in-
antigen, an indication that further B-cell activation should be teractions during the B-cell response to antigen.130
curbed, and thus, the FcγRIIB inhibits BCR signaling. In addition to the CD19-CD21 complex, B cells express
CD19/CD21, a transmembrane glycoprotein complex other activating coreceptors whose ligands and mechanisms
expressed on the surface of all B cells, is the best understood by which they affect BCR signaling are less well characterized
activating coreceptor for BCR signaling. On mature B cells, (see Table 9.2). These activating coreceptors include the re-
the CD19 component can exist alone or as part of a tetra- cently identified Fc receptor for IgM (FcμR), which contains
meric complex with CD21 (complement receptor type 2), the several conserved tyrosine and serine residues in its cytoplas-
tetraspanin family protein, CD81, and the small interferon- mic tail that are targets for phosphorylation after receptor li-
induced transmembrane protein 1, Leu-13126 (Fig. 9.5). CD21 gation with IgM-containing immune complexes.131 Although
links this complex to the BCR in response to complement- the physiologic role of FcμR in B-cell responses is not known,
coupled antigens and boosts BCR signaling and ultimately it has been hypothesized that FcμR may become coligated to
antibody production. Upon binding of complement-coupled the BCR and CD19/CD21 complex through the recognition
antigen, the BCR and CD19/CD21 complex are coligated, of complement-coupled IgM-antigen immune complexes
and several tyrosine residues in the cytoplasmic tail of and drive the B cell response toward isotype-switched an-
CD19 are phosphorylated by Lyn, providing binding sites tibodies. Other activating coreceptors are Fc receptor-like
for the SH2 domains of the p85 subunit of PI3K, Vav and 1 (FcRL1)132 and paired Ig-like receptor A (PIRA)133,134 that
Lyn, thus enhancing the signaling triggered by BCR anti- respectively contain or associate with cytoplasmic ITAMs.
gen binding.127 It is hypothesized that the association of Lyn The tyrosine residues within these ITAMs are phosphory-
with CD19 may counteract the autoinhibition of Lyn, which lated by one of the Src-family tyrosine kinases of B cells (Lyn,
is maintained by interaction between its SH2 domain and Fyn, or Blk) upon receptor engagement by their ligands,
its carboxy-terminal phosphotyrosine, thus stimulating the mediating costimulatory signals by creating docking sites

for other proteins involved in the B-cell signaling pathway. ITIM-mediated mechanism, such as those shown in
Instead of ITAMs, the cytoplasmic domains of the activating Table 9.2. In general, the Lyn-phosphorylated ITIMs recruit
coreceptors CD45R and CD148 contain an active protein lipid phosphatases (eg, SHIP) or protein phosphatases (eg,
tyrosine phosphatase domain that dephosphorylates a key SH2-domain-containing protein tyrosine phosphatase-1
regulatory tyrosine at the C-terminal end of Src kinases dur- [SHP-1]) that effectively block BCR signaling.151 SHP-1
ing BCR signaling.135–138 likely downmodulates BCR signaling by dephosphory-
While various coreceptors promote BCR signaling, there lating early signaling components that include BLNK.152
are also numerous coreceptors that are poised to counter Coreceptors such as FcRL2, 3, and 5 have both ITAM-like
B-cell activation. Fc γRIIB, the best characterized inhibitory and ITIM sequences in their cytoplasmic tails, but the in-
coreceptor for BCR signaling, is expressed on the surface of hibitory effects of their ITIMs appear to dominate their
all B cells and is composed of two extracellular Ig-like do- regulatory function.153–155 Negative modulators of BCR sig-
mains: a transmembrane domain and a cytoplasmic domain naling other than Fc γRIIB do not require coligation to the
that contains a single immunoreceptor tyrosine-based in- BCR by immune complexes to attenuate signaling but are
hibitory motif (ITIM)139 (see Fig. 9.5). When a B cell binds instead regulated functionally by binding various ligands on
IgG-antigen immune complexes, Fc γRIIB and the BCR are the B-cell surface.
coengaged. Serum contains a relatively high concentration A recurrent theme throughout the discussion of the
of soluble IgG, but the affinity of the Fc γRIIB for soluble IgG mechanism by which the CD19/CD21 complex and
is low and such that it is only able to stably engage multi- Fc γ RIIB function is the requirement for coligation to the
meric IgG-antigen immune complexes. The presence of IgG BCR by complement-coupled antigens or immune com-
immune complexes indicates that sufficient antigen-specific plexes. The requirement for joint binding of the BCR and
antibody has been made to counter any foreign threat, and coreceptors to antigen-containing complexes ensures that
that antibody production can be attenuated or stopped, de- only the antigen-specific B-cell response is regulated.
pending on the amount of IgG-immune complex present. However, some coreceptors that function to regulate BCR
Coengagement of FcγRIIB and the BCR promotes phos- signaling do not appear to require physical cross-linking
phorylation of the ITIM of Fc γRIIB by BCR-associated Lyn, with the BCR to affect BCR signaling. In fact, for several
recruiting the lipid phosphatase SH2-containing inositol such coreceptors their ligands, if they exist, are not known.
5′-phosphatase (SHIP).140 SHIP dephosphorylates PI(3,4,5) These coreceptors raise a number of yet unanswered ques-
P3 to PI(3,4)P2, preventing the recruitment of PH domain– tions concerning how they sense that a BCR is activated
containing kinases (eg, BTK and PLC-γ2) to the cell mem- and the mechanism by which they regulate BCR signaling.
brane, thereby downregulating downstream signaling and For example, CD19 regulates BCR signaling independently
proliferation.141 SHIP also recruits RasGAP via the adaptor of CD21. In response to antigen presented on fluid lipid
Dok-1, which inactivates Ras and thus further downregu- bilayers as surrogate APCs, CD19 was shown by live cell
lates proliferation.142,143 imaging to transiently associate with signaling-active BCR
Besides affecting downstream BCR signaling, it has microclusters, mediating the recruitment and activation
also been shown that coengagement of Fc γ RIIB with the of associated signaling molecules and, therefore, ampli-
BCR greatly destabilizes newly formed BCR oligomers fying signaling within the individual microclusters.28,124
that are essential for the initiation of signaling.144 This CD19-deficient B cells are unable to form BCR microclu-
early point of inhibition is not influenced by interactions sters and undergo a spreading response or flux calcium in
between the cytoplasmic tails of the two receptors but in- response to membrane-bound antigens, suggesting that
stead by the perturbation of the local lipid environment the BCR-CD19 complex may be a basic signaling unit, es-
by Fc γ RIIB, which destabilizes BCR association with sential for BCR activation. Notably, B-cell spreading is not
saturated lipids and Lyn.144,145 It is of interest that a loss- abrogated in the absence of CD21, suggesting that CD19
of-function mutation in the transmembrane domain of can function independently of the CD19/CD21 complex.
Fc γ RIIB that is associated with the autoimmune disease As evidence toward an independently functioning CD19,
systemic lupus erythematosus (SLE) prevents the associa- mice lacking CD19 exhibited a more severe phenotype
tion of ligated Fc γ RIIB with saturated lipids146,147 and thus than those lacking CD21.156 It is interesting that a CD19
inhibits the ability of Fc γ RIIB to block BCR oligomeriza- deficiency in mouse B cells does not impair the BCR re-
tion and signaling. Fc γ RIIB may also function when en- sponse to soluble antigens,127 suggesting that the process
gaged by immune complexes independently of the BCR. by which BCRs cluster in response to membrane-associ-
Cross-linking Fc γ RIIB by immune complexes that do not ated antigens provides a mechanism for recruiting CD19
contain the antigen for which the BCR is specific leads to to the BCR, resulting in enhanced signaling. In addition
apoptosis through a c-Abl– dependent pathway.148,149 Recent to affecting early BCR signaling, a CD19 deficiency is also
evidence indicates that this pathway may be important in implicated in impaired pre–BCR-dependent development
eliminating preexisting Fc γ RIIB-expressing plasma cells in and responses to antigen in vivo.157,158 Another potential
the bone marrow after infection or immunization, creating example of ligand-independent coreceptor regulation of
space in the limited niches in the bone marrow for newly BCR signaling is FcRL4, a member of the FcR-like family of
formed plasma cells.150 proteins, which has no known ligand.132 FcRL4 is expressed
In addition to Fc γRIIB, B cells express a variety of on a subpopulation of atypical memory B cells that are ex-
other coreceptors that inhibit BCR signaling via a similar panded in individuals with chronic infections, including

acquired immunodeficiency syndrome and malaria, and mutations affecting the activity of coreceptors or the kinases
are hyporesponsive to BCR triggering. Antigen binding in the BCR signaling pathway. Hyperactivation of the BCR is
leads to association of FcRL4 with the BCR and a block in associated with both B-cell tumors and autoimmune disease.
BCR signaling at the point of Syk phosphorylation159 and In the future, it may be possible to develop new therapeutic
eliminating FcRL4 expression in atypical memory B cells strategies that target spontaneous BCR oligomerization clus-
from individuals infected with human immunodeficiency tering and chronic active signaling to treat B-cell tumors and
virus restores BCR signaling.160 An important unanswered systemic autoimmune disease.
question is in the absence of coligation, by what mechanism The BCR is required for the survival of the activated B
do these coreceptors associate with the BCRs? Possible sug- cell-like (ABC) subtype of diffuse large B-cell lymphomas
gested mechanisms are through intracellular adaptor mol- (DLBCLs).164 Using live cell imaging, the BCRs on these
ecules or through membrane perturbations that facilitate activated DLBCLs were found to form immobile oligomers
BCR-coreceptor interactions. In the absence of a mecha- within microclusters similar to those observed on nor-
nism that triggers the association of these coreceptors with mal antigen-stimulated B cells. In contrast, the BCRs on
the BCR, it is also not easy to guess what environmental Burkitt’s lymphoma, mantle cell lymphoma, and germinal
clues these receptors are responding to. center B cell-like DLBCL tumors, which are not dependent
Another family of innate immune system receptors that on the BCR for survival, show no clustering. Somatic mu-
receive information from the B cell’s environment and can tations affecting the cytoplasmic domains of Igα and Igβ
affect signaling through the BCR is the TLR family. TLRs are detected frequently in ABC DLBCL biopsy samples but
are distinctive in that they recognize highly conserved mo- rarely for other DLBCLs and never for Burkitt lymphoma or
tifs present in microorganisms, including bacteria, viruses, mucosa-associated lymphoid tissue (MALT) lymphoma, al-
fungi, and protozoans, referred to as pathogen-associated though the contributions of these mutations to chronic BCR
molecular patterns, as described in Chapter 15.161 As for clustering is not known.
the BCR, TLR-initiated signaling results in the activation Interestingly, the spontaneous BCR clustering on ABC
of MAP kinases and NF-κ B162 ; thus, TLR and BCR signal- DLBCLs is similar to that observed for human H chain dis-
ing have the potential to synergize. Of particular interest to ease in which ligand-independent BCR self-aggregation and
BCR signaling are TLR7 and TLR9, which reside in intra- constitutive activation are a consequence of BCR H chain
cellular compartments of B cells and respectively recognize gene mutations that cause misfolding and disrupt antigen
pathogen-associated molecular patterns in single-stranded binding.165 Impaired glycosylation and folding of the μ-H
ribonucleic acid (ssRNA) derived from RNA viruses and un- chain are implicated in chronic lymphocytic leukemia,
methylated CpG-containing deoxyribonucleic acid (DNA) of which has symptoms similar to those of μ-H chain dis-
bacterial and viral origin. The ssRNA-induced TLR7 signal- ease; however, it has not yet been determined whether this
ing and CpG-containing DNA–induced TLR9 signaling have misfolding leads to enhanced BCR ligand-independent, or
been shown to synergize with antigen-induced BCR signal- chronic active, signaling.
ing in NF-κ B activation.162 It was recently shown that syn- In addition to the BCR itself, other BCR signaling mol-
ergistic signaling in response to CpG-containing DNA–con- ecules that are key regulators of the NF-κ B pathway are also
taining antigens is mediated by a novel mechanism in which, known targets of mutations found in B-cell lymphomas.
following antigen binding, the BCR signals from the plasma Specifically, MALT1 and Bcl10 are independently associated
membrane to recruit TLR9 from small endocytic intracellu- with chromosomal translocations in MALT lymphoma,
lar vesicles to autophagosomal compartments into which the whereas CARD11 has activating point mutations in a per-
BCR traffics the CpG-containing antigen and from which centage of ABC DLBCLs that result in constitutive NF-κ B
synergistic signaling occurs.163 By this mechanism, TLR9 can activation.113,165–167 The ABC DLBCLs are also dependent for
alert the BCR to the presence of pathogenic DNA in the en- their survival on MyD88, the adaptor that mediates TLR sig-
vironment and heighten BCR signaling. Similar mechanisms naling.168 Remarkably, nearly 30% of ABC DLBCLs have a
may be at play for BCR and TLR7 signaling in response to single gain-of-function mutation in MyD88 that promotes
ssRNA-containing antigens. cell survival by activating the NF-κ B pathway. The dual de-
pendence on both the BCR and TLR pathways suggest that
these work in concert to drive B-cell tumorigenesis.
ABNORMAL B-CELL RECEPTOR SIGNALING IN B cells from patients with systemic autoimmune disease
HUMAN DISEASE exhibit B-cell hyperactivation that may be due to alterations
In this chapter, we have provided an overview of our current in the BCR signaling pathway or alterations in corecep-
understanding of the molecular mechanisms underlying the tor function.169 In SLE, stimulating peripheral blood B cells
early BCR-intrinsic events that lead to normal BCR oligomer- with BCR ligand leads to increased intracellular calcium flux
ization and subsequent signaling and have emphasized that and increased phosphorylation of various cytosolic proteins
early events in the initiation of BCR signaling are tightly reg- as compared to healthy individuals and patients with other
ulated. It might be predicted that alterations in any step of the rheumatic diseases.170 In most of patients with SLE analyzed,
initiation of signaling could drive the BCR to hyperactivation. expression of Lyn is also significantly decreased in resting,
For example, the outcome of antigen engagement could be af- and BCR stimulated peripheral blood B cells171; the abil-
fected by mutations in the BCR itself, changes in the compo- ity of Lyn to associate with saturated membrane lipids is
sition of the membrane lipids that stabilize BCR oligomers, or reduced in SLE B cells as compared with healthy donors.172

Decreased Lyn expression negatively impacts its ability to messenger RNA is readily detected in the peripheral blood
inhibit BCR signaling via phosphorylation of FcγRIIB and B cells of healthy individuals, and LMP2A is often found in
other inhibitory coreceptors that contain ITIMs. In addition, tumors from EBV-associated malignancies.177 To subvert im-
a loss-of-function mutation in the transmembrane domain mune responses and consequently maintain viral latency in
of FcγRIIB that is associated with SLE susceptibility pre- B cells, EBV blocks signaling and antigen-trafficking func-
vents BCR-ligated FcγRIIB from associating with saturated tions of the BCR through the activity of the cytoplasmic tail
membrane lipids146 and consequently inhibits the ability of of LMP2A178–180 that, when phosphorylated, binds to Lyn
FcγRIIB to block BCR oligomerization, clustering, and sub- (Y112) and Syk (Y74, Y85),181,182 thus blocking BCR signaling.
sequent signaling. The synergistic engagement of the BCR LMP2A also activates the PI3K/Akt pathway, which normally
and TLR7 and TLR9 in response to antigens containing provides a survival signal in response to BCR signaling.183
ssRNA or CpG-containing DNA has also been implicated in Overall, LMP2A maintains viral latency by preventing nor-
the activation of autoimmune B cells. Many of the antigens mal BCR activation, which initiates viral replication, while
targeted in SLE contain DNA and RNA that are thought to sustaining survival pathways in latently infected B cells.
be released from apoptotic cells. Indeed, genetic variations in
TLR9 are linked to SLE susceptibility173; in a mouse model of
SLE, multiple copies of a normal TLR7 gene are sufficient to CONCLUSION
drive SLE.174 Overall, it seems that enhanced B-cell signaling B-cell activation is initiated by the binding of antigen to the
as a result of a multiple genetic abnormalities is the defining BCR resulting in the triggering of a number of signaling path-
pathogenic event of SLE and provides a therapeutic target for ways that ultimately drive B-cell proliferation and differentia-
treating autoimmune disease. tion to antibody-secreting cells. At present, we understand the
BCR signaling can also be altered in infectious diseases biochemical nature of the signaling pathway in some detail.
by viruses that commandeer B cells for their own replica- What remains less well understood is the nature of the events
tion. Epstein-Barr virus (EBV) is the causative agent of following antigen binding to the BCR that trigger these sig-
infectious mononucleosis and is implicated in Burkitt’s lym- naling cascades. New tools of live cell imaging both in vitro
phoma, Hodgkin disease, nasopharyngeal carcinoma, and and in vivo are anticipated to provide an increasingly detailed
various lymphoproliferative disorders that arise in immuno- spatial and temporal picture of events that initiate signaling in
compromised individuals.175 EBV infects B cells of the oral both time and space. Our increased understanding of B-cell
epithelium and establishes a lifelong latent infection in a sub- activation is likely to lead to new approaches for developing
set of those cells.176 Latent membrane protein 2A (LMP2A) therapies for diseases caused by hyper–B-cell activation.

CHAPTER
10 B-Lymphocyte Responses
Michael McHeyzer-Williams
INTRODUCTION with cognate TFH cells that determines subsequent B-cell

B-lymphocyte responses provide the effective and long- fate. Naïve B cells recognize foreign antigen and then pres-
lasting immune protection induced by most vaccines in ent antigenic peptides to specific TFH cells to progress in de-
use today. We measure antibody molecules as the circulat- velopment across two major pathways (pre–germinal center
ing agent of immune protection but now understand much [GC] development). Extrafollicular development permits
more about the underlying B-lymphocyte response that pro- antibody class switch and plasma cell differentiation while
gressively matures in response to foreign antigen exposure. entry into the GC reaction is the major pathway to high-
This chapter will focus on the highly-regulated cellular and affinity B-cell memory (GC cycle). GC B cells in this path-
molecular development of antigen-specific B-lymphocyte way can switch antibody class and affinity mature their ex-
responses. pressed B-cell receptor (BCR) following access to antigen
The three cardinal features of effective B-cell immunity and presentation to GC TFH cells. Upon antigen reexposure,
are antigen specificity, antibody class, and antigen binding memory B cells recognize, uptake, and then present anti-
affinity. To be effective, antibodies must bind vulnerable an- gen to memory TFH cells to promote rapid memory B-cell
tigens on the targeted pathogen. During infection, this is a responses and boost circulating high-affinity antibody
struggle between the evasion mechanisms of the pathogen (memory B-cell response). Antigen recall is the least studied
and preexisting diversity within the adaptive immune sys- facet of B-lymphocyte responses but provides an important
tem. Different classes of antibodies engage distinct mecha- developmental juncture for vaccine-based prophylactic or
nisms of antigen clearance. These various classes of anti- therapeutic intervention.
body provide either more or less protection depending on This chapter mainly focuses on what is known of antigen-
the pathogen’s portal of entry at infection. Finally, the right specific B-lymphocyte responses in mouse models with ref-
antigen-specific antibody still requires induction of suffi- erence to work conducted in humans. Further, there is an
ciently high binding affinity to provide adequate sensitivity emphasis on the response to model antigens that provides
for long-term immune protection. Many promising antigens a greater understanding of how to manipulate adaptive im-
fail to achieve adequate immunogenicity in contemporary munity for preventative vaccination rather than a focus on
vaccine strategies due to poor affinity maturation. Here, we the immune response to infection.
focus on the sequential mechanisms that program these cen-
tral attributes of effective antigen-specific B-cell memory. PRE–GERMINAL CENTER DEVELOPMENT
In the past few years, experimental access to immune The initial antibody response to many infectious agents is
response biology has dramatically shifted with the advent largely based on the rapid T-cell independent (TI) expansion
of multiphoton laser-based intravital imaging techniques. of B cells and their subsequent differentiation into plasma
These studies provide direct access to the mechanics and cells. B1 and marginal zone B cells are largely responsible
cell dynamics of antigen-specific cognate regulation in vivo. for these rapid TI humoral responses suggesting that some
This information serves to integrate existing knowledge in level of “natural memory” function resides in these B-cell
the field using a real-time scaffold for developmental pro- subsets and may be predetermined in an evolutionarily con-
gression in vivo. Importantly, follicular helper T (TFH) cells served manner. TI antigens can be separated into two broad
have recently emerged as a new class of antigen-specific im- categories based on their ability to polyclonally activate B
mune regulator that controls multiple stages of high-affi nity cells (TI-1) or require BCR recognition of multivalent epit-
B-cell immunity. Understanding antigen-specific TFH-cell opes to induce B-cell differentiation in the absence of T-cell
development and function remains an active research chal- help (TI-2). TI-2 antigens can activate BCR signaling but
lenge with great potential for the rational design of future require accessory signals to promote the development of an-
vaccines. Current information regarding the role of antigen- tigen-specific plasma cells. In contrast, monovalent protein
specific TFH cells will be integrated into this chapter to pro- antigens require antigen-specific helper T-cell regulation
vide a regulatory dimension to B-lymphocyte responses. to promote high titer antibody responses and the develop-
Following initial exposure to antigen, TFH-cell–regu- ment of B-cell memory. These TFH-cell–dependent antibody
lated B-cell immunity progresses in three separable stages responses take longer to emerge and display a spectrum of
of antigen-specific development. Each stage is characterized BCR affinities and antibody isotypes as multiple strategies
by a B-cell antigen-recognition event followed by contact for antigen-specific clearance in vivo. Immune responses to
261

c SCS
MO
Secondary
Primary follicle formation
B-cell zone
follicle
Naive B B
B
B B B Commitment
B B to Memory Tfh
B B
B Ag-
B B primed B B
B
B B B g B B
B B B
B
B
Tfh B
e f
Th Th
B
T-cell zone
Th PC
Th Th Th
d B B
B
B B PC
Ag- B
primed Naive Th
eTh PC
DC Th Th
a b
Time (days) 3 5 7
FIG. 10.1. Pre–Germinal Center (GC) Development. A: Local protein vaccination induces dendritic
cells (DC) maturation and migration to the T-cell zones of draining lymph nodes (LNs). B: Peptide
major histocompatibility class II expressing DCs will engage naive antigen-specific helper T (TH)
cells to induce proliferation and effector TH-cell differentiation. C: Whole antigen will be trapped
by subcapsular sinus macrophage and presented to naive follicular B cells. Antigen-specific B cells
will become activated and uptake, process, and present antigenic peptides and migrate toward the
T-B borders of the draining LN. D: Effector TH cells emerge in multiple forms with emigrant TH cells
exiting the LN to function at distal tissue sites and TFH cells relocating to T-B borders and interfol-
licular regions. E: Cognate contact between pre-GC TFH cells and antigen-primed B cells is required
for multiple programming events in the pathway to B-cell immunity. F: Clonal expansion, antibody
class switch, and non-GC plasma cell development proceeds within extrafollicular regions of LNs.
G: Secondary follicle formation and antibody class switch precede formation of the GC reaction as
the dominant pathway to memory B-cell formation.
model protein antigens provide experimental access to this However, populations of lymph node (LN) subcapsular
complex cascade of cellular and molecular events that un- sinus (SCS) macrophages appear most effective at present-
derpin long-term protective immunity. ing cell–associated antigen to follicular B cells.3 B cells
B-cell immunity is initiated on two fronts. Initially, naïve take up noncognate antigen presented by SCS macrophages
B cells are activated through cell-associated antigen to uptake through complement receptors and transfer it into follicu-
antigen, process and present peptide major histocompatibil- lar regions and onto follicular DCs (FDCs), which can serve
ity complex [MHC] class II (pMHCII) complexes to enable as a source of antigen to prime naïve B cells.4 In contrast,
cognate contact with pMHCII-specific TFH cells. These anti- priming with the cognate antigen at first contact with SCS
gen-primed B cells relocate to T-B–cell borders in lymphoid macrophages results in movement of antigen-specific B cells
tissues to increase the likelihood of contact with antigen- to the T-B–cell borders and antigen-specific B-cell responses
primed TFH cells. On the second front, populations of den- to captured antigens.4–6 Hence, the SCS macrophages fi lter-
dritic cells (DCs) also take up antigen, process and present ing the lymphatic fluid not only protect from systemic in-
pMHCII, and migrate to draining lymphoid tissues to initiate fection,7 but also effectively initiate T helper cell–regulated
TFH-cell responses. One outcome of contact with pMHCII- antigen-specific B-cell immunity.
expressing DCs is differentiation of the TFH-cell lineage and DCs can also effectively prime B-cell immunity. Injection
the productive contact between pMHCII-specific TFH cells of DCs pulsed with protein antigen can induce isotype switch
and antigen-primed B cells. This pre-GC cell contact and bi- and promote efficient B-cell responses. B cells can form
directional exchange of molecular programming is central to synapse-like interactions with antigen-pulsed DCs.8,9 More
the development of effective B-cell immunity. Antibody class recently, two-photon imaging revealed that naïve B cells en-
switch, extrafollicular plasma cell differentiation, and initia- tering local LNs surveyed protein antigen-pulsed DCs before
tion of the GC reaction are major B-cell fates associated with entering the follicular areas.10 In this model, engagement of
this initial phase in B-cell immunity (Fig. 10.1). BCR led to calcium flux, migration arrest, and the local ac-
cumulation of the antigen-specific B cells. Furthermore,
there is a reticular network of collagen fibers that physically
Antigen Presentation to B Cells connects the subcapsular and paracortical sinuses of LNs to
B cells can acquire soluble antigen by free diffusion into lym- blood vessels and separates these regions from T- and B-cell
phoid follicles1 or through the lymphoid system of conduits.2 areas.11 This organization facilitates the efficient delivery of

CHAPTER 10 B-LYMPHOCYTE RESPONSES | 263
soluble antigen toward the lumen of high endothelial ve- lysosomes for presentation with MHCII8 as the cognate
nules without entering the LN parenchyma. While resident point of TFH contact. Curiously, antigen presentation ap-
DCs can access this conduit transport system, follicular B pears asymmetrically segregated with one daughter re-
cells may have difficulty accessing soluble antigen. taining larger antigen stores thereby more able to contact
Nevertheless, in the context of protein antigens, B cells pMHCII-specific TFH cells.27 Whether these developmental
must recognize their cognate antigen and internalize, pro- outcomes are directed by initial context of antigen presen-
cess, and present peptides from this antigen in the context tation or stochastically assorted 28 remains an interesting
of MHCII in order to receive pMHCII-specific T-cell help. If fundamental issue with an early impact on antigen-specific
the antigen is cell associated, in the presence of adequate co- B-cell development.
stimulation, some aspects of the B-cell response can proceed
in a TI manner. B-cell proliferation and plasma cell devel-
opment can occur in the absence of cluster of differentia- Dendritic Cell Maturation
tion (CD)40-CD40L interactions with some residual isotype DCs are essential antigen-presenting cells (APCs) for initi-
switch induced by the action of TACI and BAFF-R.12 Thus, B ating adaptive immunity. Multiple DC subsets exist prior
cells can respond in a TI manner to protein antigens but do to antigen challenge. Different phenotypic schemes can be
not express the characteristic range of outcomes and the full used to characterize DC subsets with the origins and devel-
extent of protective immunity that is found with cognate opmental relatedness of different DC subsets still subject to
regulation and T-cell help. debate.29,30 In the murine system, three main bone marrow–
derived DC subsets enter all secondary lymphoid organs
via the blood and reside at different levels. These blood-
Antigen-Specific B-Cell Activation derived DCs all express CD11c and are distinguishable as
Initial activation of naïve B cells through the BCR triggers CD11bhiCD8aneg DCs, CD8a hi DCs, and 6B2hi plasmacytoid
multiple gene expression programs that enable effective DCs. In LNs draining the skin, there are at least two fur-
contact with cognate T helper cells. Dynamic contacts with ther CD11c + DC subsets, Langerhans cells (LCs) and der-
membrane-associated antigens determines the amount of mal DCs, that emigrate from the skin, even at homeostasis.31
antigen naïve B cells accumulate following antigen expo- Thus, before antigen challenge, multiple subsets of DCs are
sure.13 Effective cell contacts require expression of the signal- available to differentially process and present antigen to the
ing adaptor dedicator of cytokinesis 8 (DOCK8).14 Mutations adaptive immune compartment.
in DOCK8 disrupted integrin ligand accumulation in the Protein antigen administration in the absence of inflam-
immune synapse without altering BCR signaling events. mation induces immune tolerance. In contrast, coadmin-
B-cell–specific conditional ablation of the calcineurin regu- istration of an immune adjuvant activates facets of innate
latory subunit 1 (CNB1),15 myocyte enhancer factor 2c,16,17 immunity, induces inflammation, and primes antigen-
and stromal interaction molecule 1 and stromal interac- specific adaptive immunity. Sensing pathogens involves
tion molecule 218 have shown that calcium responsiveness pattern recognition receptors such as the evolutionarily
is necessary for cell cycle progression in these early pre-GC conserved toll-like receptors32,33 and the more recently
stages of B-cell responses. Hydrogen voltage-gated proton described nucleotide oligomerization domain–like recep-
channels 1, which are internalized with the BCR, have been tors.34 Most forms of antigen and innate stimulators require
recently implicated in early B-cell programming events.19 DC priming at some level for B-cell immunity, whereas
Single-pulsed BCR signaling 20 that only partially activated particulate antigen in virus-like particles are more reliant
NF-κ B increased CC chemokine receptor 7 and MHC class on B-cell innate receptor activation.35 Nevertheless, both
II expression, and responsiveness to CD40, indicating some families of innate receptors recognize different types of mi-
of the early facets of B-cell activation. Severe defects in early crobial components initiating programs of DC maturation
B-cell proliferation have also implicated integrin-binding that promote immediate local inflammation and innate ef-
CD98hc21 and extracellular signal-regulated kinase activa- fector clearance mechanisms.
tion22 in preparing the antigen-primed B cells to receive cog- Temporal and spatial constraints on DC maturation
nate T-cell help in vivo. Hence, initial antigen recognition, provides another layer of regulation for the innate system
uptake, processing, and presentation critically impact the that can impact adaptive immunity. In the steady state, DCs
early B-cell developmental fate. form dense networks at the T-B borders of LNs. Interestingly,
High-resolution dynamic imaging has provided sub- motile lipopolysaccharide-activated DC immigrants will
stantial insight into the earliest events associated with rapidly coalesce with this preexisting network in vivo.36 In
initial BCR engagement on naïve B cells. The membrane separate studies using genetically tagged LCs, the emigrants
cytoskeleton controls BCR diffusion; disruption of this of LCs and dermal DCs were shown to emerge separately
organization initiates signaling.23,24 Discrete microclusters in time and colonize separate regions of the T-B border.37
form upon antigen binding 25 and recruit multiple compo- A similar temporal regulation was seen using antibodies to
nents of the intracellular BCR signaling network to initiate specific pMHCII complexes with resident DCs presenting an
signal transduction.26 One rapid response involves B-cell early wave of pMHCII and dermal DCs emerging later in a
spreading to increase surface contact with antigen on the second wave.38 In each of these studies, the immune stimu-
presenting cells.13 Central clustered antigen at the cellu- lus was varied and the coordinated response of the innate
lar interface is then internalized into antigen-processing system also qualitatively and quantitatively different.

Clonal Selection in Helper T Cells TFH cells and other known TH-cell subsets. CXCL13 was
The initial outcome for pMHCII+ DC interactions with naive highlighted early 62 with evidence for ICOS, IL-21,63,64 IL-
TH cells is T-cell receptor (TCR)-driven clonal selection. TH- 21R,65 and the differential expression of Bcl-663 being used
cell responses that focus the specific TH-cell response to a set as the most reliable attributes of TFH function in vivo. Thus,
of dominant TCRs provide access to the mechanisms that acquisition of special pre-GC TFH-cell functions was associ-
underpin TH clonal selection.39–41 Earlier studies indicate the ated with the programming of a separate TH-cell lineage.
importance of TCR-pMHCII affinity in determining TH cell Molecules important in the development of normal B-cell
fate.39,42–44 In this model, selective expansion of clonotypes ex- immunity were implicated in early studies of TFH develop-
pressing higher-affinity TCRs would rely on competition for ment. CD28 deficiency or treatment with blocking CD28 an-
pMHCII complexes on APCs. Alternatively, in vitro studies tibodies led to profound defects in B-cell immunity.66 CD28
using altered peptide ligands suggested that the duration of was required early to initiate naive TH-cell responsiveness to
TCR-pMHCII contact was critical to cell fate and therefore pMHCII + CD80- and CD86-expressing DCs. In contrast,
defined “best fit” in vivo.45–48 In support of this model, TH CD40-CD40L interactions were central to the delivery of
cells expressing TCR with fast off rates were lost over time in T-cell help to B cells.67–69 The CD28 family member ICOS70
vivo.49 Further, there is evidence for different peptides stabiliz- was implicated in TFH-cell function in pre-GC interactions
ing the pMHCII to create a hierarchy of dominant peptides.50 with pMHCII-expressing B cells.71–73 ICOS-deficient humans
Each and all of these variables would impact what is consid- and mice74–76 and ICOS-L–deficient mice had marked defi-
ered “best fit” and influence the outcome of clonal selection. cits in all aspects of B-cell immunity. ICOS deficiency was
More recently, we provide evidence for a model of TH clonal associated with decreased TFH-cell development and consid-
selection that is based on a TCR affinity threshold.51 Multiple ered an important molecule in the delivery of effector TFH
affinity-based thresholds underpin cell fate and TH clonal ex- function.77,78 Conversely, overexpression of ICOS in mice
pansion52 with evidence that the highest affinity TCR assort with a regulatory defect in ICOS expression79 resulted in an
into the TFH compartment.53 Hence, selection thresholds gen- overproduction of CXCR5 + TFH cells and breakthrough au-
erate antigen-specific clonal diversity in ways that impact the toimmune disease.64 Recent studies indicated that ICOS can
development of effector TH cell function in vivo. substitute for CD28 and rescue the TFH defects and B-cell de-
fects in CD28-deficient mice.80 Furthermore, the abundant
CXCR5 + TFH cells in this model act in a T-cell autonomous
Follicular Helper T Cells manner to promote autoantibody production.81 PD-1, an-
There were early reports of effector TH cells specialized to other CD28 family member that has been implicated in the
regulate B-cell responsiveness. CXCR5 expression was first negative regulation of chronically activated T cells, was also
reported on CD4 + CD45RO + cells in the peripheral blood found on GC TFH cells in human tonsil and mouse.82 Positive
and secondary lymphoid tissue in humans.54 Gene ablation and negative influences of antigen-specific TFH-cell costimu-
studies emphasized the role of CXCR5 and CC chemokine lation are balanced in ways that remain poorly understood in
receptor 7 in the correct positioning of T and B cells in sec- vivo and are an active avenue of current research in this field.
ondary lymphoid tissue that was also needed to support BCL6 is required for development of the TFH program83–85
effective TH cell–dependent B-cell immunity.55,56 Blocking and is reinforced within TFH cells upon pre-GC B-cell con-
CD28 and OX40 interactions in vivo blocked the develop- tact.86,87 IL-21 also plays a major role in TFH function with
ment of CXCR5 + TH cells and the GC reaction.57 Further, substantial loss of B-cell immunity in its absence.88,89 Recent
CXCR5 expression was induced in an antigen-specific man- studies have identified BCL6 in antigen-specific TFH cells and
ner on TH cells in vivo, and these cells relocated to follicular BLIMP1, which has an opposing function, in non-TFH cells.53
areas and the GC of responding lymphoid tissue.58 CXCR5 + BCL6 and BLIMP1 expression was mutually exclusive across
TH cells were sorted from human tonsil and shown to sup- these two TH-cell subtypes, already evident by the second cell
port antibody production in vitro.59,60 The tonsillar CXCR5 + division in vivo.29 More recently, transcription factors c-Maf
TH cells expressed high levels of CD40L and ICOS and were and BATF were shown to act with BCL6 to program TFH
found in both the follicular mantle and GC. Adoptive trans- development.90,91 Hence, distinct transcriptional program-
fers distinguished CXCR5 + B cell helper activity (TFH) from ming of unique cellular functions directs the early TFH-cell
P-selectin ligand hi tissue homing inflammatory mediation development that is central to subsequent B-cell immunity
(DTH-promoting TH) that emerged together from the same (Fig. 10.2).
set of precursors in vivo.61 The term TFH cells was coined to
categorize this functionally and phenotypically distinct ef-
fector TH cell compartment.59,60 Pre-Germinal Center Follicular Helper
As the name implies, the cardinal characteristic of all T-B–Cell Contact
TFH cells is their repositioning into the follicular regions of First contact between antigen-specific TFH and antigen-
secondary lymphoid tissues. Early assessments of cytokine primed B cells has also been captured through dynamic
production by in vitro restimulated CXCR5 + TFH cells indi- imaging.92 Stable “monogamous” interactions between one
cated interleukin (IL)-2, interferon (IFN)-γ, and IL-10 from antigen-specific TFH cell and one B cell can last for a duration
human peripheral blood60 with evidence for IL-4 and IFN- of 10 to 60 minutes in the follicular regions of the LNs. These
γ from TCR transgenic mouse TFH cells.61 Early microarray interactions were accompanied by highly dynamic move-
analyses suggested separable gene expression programs for ments, with the B cells migrating extensively and leading

a b
Multiple Antigen
IgG1
TFH subsets primed B cell BCL6
CXCL13
IgG2a
IL4
PD1 BCR BCL6
TFH1 CXCR5
OX40 OX40L Nγ
IF
Cytokine Cytokine
Receptor IgA
TGFβ
Pre-GC CD40L CD40 Ag-primed BCL6
TFH2 GC
TFH TCR pMCHII B pathway
ICOS ICOSL e
BCL6
IL21 IL21R d Non GC
TFH21 SLAM SLAM pathway
IgG1
Blimp1
c
Cognate IgG2a
TFH10 Blimp1
contact
IgA
TFH17 Blimp1
FIG. 10.2. Pre–Germinal Center (GC) Follicular Helper T (TFH)-B-Cell Contact. A: Multiple subsets
of antigen-specific pre-GC TFH are produced to regulate B-cell immunity. The organization of func-
tion within TFH subsets remains speculative. There is evidence for different cytokine-secreting TFH
cells that regulate commitment to separate antibody class as well as different types of TFH cells that
regulate non-GC plasma cell differentiation. Bcl-6 and CXCR5 expression is thought to be a common
feature of all TFH cell subsets. B: Antigen-primed B cells must process and present peptide major
histocompatibility class II to receive cognate help from pre-GC TFH cells. Upregulation of molecules
involved in helper T cell contact is a poorly resolved component of early antigen-driven B-cell matu-
ration. C: Cognate contact between antigen-specific T-cell receptor and peptide major histocompat-
ibility class II complexes focus pre-GC TFH-B cell intercellular exchange of molecular information.
Modifying interactions at first contact are known to involve costimulatory molecules (eg, cluster of
differentiation [CD]40L-CD40, ICOS-ICOSL), accessory interactions (eg, signaling lymphocytic activa-
tion molecule family interactions, OX40-OX40L), cytokine and cytokine receptors (eg, IL4-IL4R, inter-
feron [IFN]γ-IFNγ receptor, IL21-IL21R). Distribution of these functional attributes within pre-GC TFH
compartments is not yet well resolved in vivo. D: The non-GC pathway to plasma cell development
permits antibody class switch recombination (CSR) without somatic hypermutation (SHM) depending
largely on the cytokine stimulus provide by pre-GC TFH cells. Blimp-1 expression is required for plasma
cell commitment across all antibody classes. E: The GC pathway to memory B-cell development be-
gins with extensive B-cell proliferation within secondary follicles that will polarize into dark and light
zone to initiate the GC reaction. The GC pathway is associated with Bcl-6 upregulation and activation-
induced cytidine deaminase expression to supports both antibody CSR and SHM. These GC features
emerge across all antibody classes and require productive and long duration pre-GC TFH contact.
the TFH cells.92 A recent study demonstrated that expression B cells are capable of forming GCs99 but fail to do so in the
of the adaptor molecule signaling lymphocytic activation presence of high-affinity competition.100 In contrast, there is
molecule–associated protein (SAP) was needed to form long evidence for the highest-affinity B cells preferentially enter-
duration contacts with antigen-primed B cells.93 In these ing the non-GC plasma cell pathway, leaving lower-affinity
studies, there was no role for SAP in early TFH-DC contacts, B cells to mature within the GC cycle.101 This issue has been
and the SAP-deficient TFH cells reached the follicular regions addressed more recently using intravital imaging to examine
and expressed all the hallmarks of pre-GC TFH cells (CXCR5hi, the early pre-GC selection events.102 In this model, access to
CD40L+, ICOShi, and OX40hi).93 Furthermore, in the absence antigen was not impacted by BCR affinity, but the differen-
of SAP, TFH cells were still capable of cytokine production94 tial capacity to present antigen to pre-GC TFH cells assorted
but unable to promote GC formation.95,96 More recently, with BCR affinity. Increased T-cell help promoted greater ac-
dynamic imaging places ongoing critical pre-GC contacts cess to both the plasma cell pathway and the GC reaction.
in the interfollicular zones of LNs97 with a requirement for Thus, BCR affinity thresholds regulate B-cell fate at the earli-
persistent BCL6 expression in B cells to maintain effective est pre-GC junctures of antigen-specific TFH-B interactions.
cognate contact.98 Therefore, early TFH-cell developmental Overall, the precise function of different costimula-
programs establish the capacity for cognate contact needed tory molecules and their combinatorial impact on antigen-
to promote ongoing antigen-specific B-cell immunity. specific B cell fate remains an exciting area of current interest.
It has been unclear how differential BCR affinity can im- Quantitative differences in cell surface molecules in combi-
pact the early fate of antigen-primed B cells. Very-low-affinity nation with mixtures of cytokines will likely synergize in

predetermined ways to skew pMHCII-expressing B cells into switch while CD27-CD70 interactions promote plasma cell
separate pathways of B-cell immunity. Unraveling these mo- (PC) production.118 Thus, the range of molecules expressed
lecular combinations will help to define the rules of molecu- at the pre-GC TFH cell surface influences the developmental
lar control for antigen-specific B-cell immunity. impact of TCR-pMHCII contact on antigen-primed B cells.
Considering the range of antibodies that can be pro-
duced, there must be multiple subtypes of pre-GC TFH cells
Controlling Antibody Class that control antibody class. Different classes of pre-GC TFH
Pre-GC cognate TFH-cell contact controls multiple antigen- would vary in production of TFH cell–derived cytokines to
specific B-cell differentiation options. Beyond this early pre- control antibody isotype. IL-4 and IFN-γ are reciprocal
GC developmental juncture, antibody class switch appears regulators of IgG1 and IgG2a production.119 Animals lacking
in both in the extrafollicular pathway to plasma cell differ- IL-4 or Stat 6 have decreased IgG1 levels and no IgE.120,121
entiation and the GC pathway to memory development.103 IL-4 also acts together with IL-21 to control IgG subtypes
Class switch also requires multiple rounds of cell division and IgE levels.88 In contrast, transforming growth factor
in the target cell before expression of non-immunoglobulin (TGF) β is implicated in the induction IgA, while IL-2 and
(Ig)M antibody.104 Hence, it is likely that initial commitment IL-5 augment IgA production.122 Similarly, IL-6 may selec-
to antibody class occurs at the first pre-GC point of cognate tively support IgG2a and IgG2b expressing B cells in vivo.123
contact and is propagated in a lineal manner across each de- Each of these factors can exert their effects in vitro or in a
velopmental option (see Fig. 10.2). bystander manner in vivo. However, it is thought that the
Class switch recombination (CSR) is an intrachromo- directed delivery of these soluble molecules toward points of
sonal deletional process between the switch (S) regions TCR-pMHCII contact allows soluble signals to focus locally
that reside 5′ of each constant region gene in B cells (except in an antigen-specific cognate manner.
Cδ).105 Signaling through CD40 and cytokine receptors in- Within antigen-responsive B cells, the molecular machin-
duces germline transcription through the targeted S regions ery that regulates CSR is deployed in an antibody class–spe-
providing activation-induced cytidine deaminase (AID; cific manner. The global CSR machinery is targeted by tran-
also known as Aicda) access to deaminate cytosines in the scription factors downstream of the cytokine receptors that
single stranded template. AID is required and sufficient for control specific antibody classes. For example, IFNγ acti-
the initiation of the CSR reaction in the activated locus.106 vates signal transducer and activator of transcription down-
AID-deficient animals107 and humans108 display no CSR or stream of the IFNγ receptor to induce T-bet and promote
somatic hypermutation (SHM) of the Ig genes. Recent evi- IgG2a class switching.124,125 Similarly, TGFβ signals through
dence indicates that following antigen stimulation, AID ex- TGFβ -receptor to activate SMAD and RUNX transcription
pression is regulated in B cells by paired box gene/protein 5, factors to promote IgA class switch.126 Furthermore, the
E-box proteins,109 homeobox C4,110 and forkhead box O1.111 transcriptional regulator BATF required for TFH develop-
Removal of the resulting uracils and deoxyribonucleic acid ment is also required in B cells to generate germline switch
(DNA) cleavage generates double strand breaks that trigger transcripts and promote AID expression.91 Finally, Ikaros
the recruitment of DNA damage machinery, mismatch re- regulates antibody class decisions by differentially control-
pair, and nonhomologous end joining to complete the CSR ling transcriptional accessibility of constant region genes.127
event.112 The adapter protein 14.3.3 is recruited with AID to How the initial commitment to antibody class is maintained
switch regions113 and polymerase ζ has been implicated in and propagated during clonal expansion, BCR diversifica-
the repair process associated with CSR.114 Peripheral B cells tion, and affinity-based selection within the GC reaction
undergoing CSR in the absence of the x-ray–repair cross- remains an important but unresolved issue. Therefore, it
complementing protein 4 component of the double strand remains plausible that functional reprogramming accompa-
break repair machinery are also highly susceptible to trans- nies CSR creating separable lineages of class-specific mem-
location events and oncogenic transformation.115 Antibody ory B cells in vivo.
class switch is a destabilizing and potentially dangerous cel-
lular event that can proceed without SHM in the extrafol-
licular pathway. Extrafollicular Plasma Cell Development
Induced CD40L on pre-GC TFH cells and the receipt of Under the cognate regulation of pre-GC TFH cells, a cohort
this signal through CD40 on B cells is required for antibody of antigen-primed B cells clonally expand within the T-cell
class switch.67,116 Animals and humans lacking CD40 dis- zones of secondary lymphoid organs and rapidly give rise to
play a hyper-IgM syndrome with profound defects in class PCs. Within the first few days after antigen exposure, small
switch, GC formation, and the development of affinity ma- foci of B-cell blasts can be seen within the T-cell zones.103 This
tured B-cell memory.116 ICOS expression on activated TFH plasmablast stage appears transitional and defines pre-PCs
cells is thought to act upstream of CD40L in this temporally that may secrete antibody but also retain the capacity to pro-
orchestrated set of events.117 ICOS-deficient animals also liferate. In contrast, PCs are typically considered terminally
have clear defects in antibody class switch, GC formation, differentiated and in a postmitotic state.128,129 PCs display a
and the development of B-cell memory.74–76 Some residual marked increase in IgH and IgL messenger ribonucleic acid
class switch in the absence of CD40/CD40L interactions and prominent amounts of rough endoplasmic reticulum
may be explained by the action of TACI and BAFF-R.12 to accommodate translation and secretion of abundant Ig.
OX40/OX40L interactions also quantitatively impact class They have reduced or lost numerous cell surface molecules

including B220, CD19, CD21, and CD22 with an increase in blocking the induction of Blimp-1 and PC development.
the proteoglycan syndecan-1 (CD138),130,131 often used as a In contrast, ablation of the transcription factor MitF leads
distinguishing marker for PCs. We recently demonstrated to the spontaneous development of PC that appears inde-
that class-switched plasma cells retained robust antigen- pendent of antigen stimuli.144 Interestingly, reexpression of
presentation capacity as a negative feedback loop for regu- bcl-6 and its cofactor MTA3 re-activates the B-cell program
lation of TFH cells.132 These studies further highlight the and increased CD19 and MHC II, and decreased CD138 in
bidirectional programming between antigen-primed B cells plasma cell lines.145 This remarkable study suggests that PC
and pMHCII-specific TFH cells during developing immune fate is not as terminal and passive as it has been thought to
responses. be, and that the PC fate remains subject to dynamic gene ex-
In mice deficient for SAP, the extrafollicular pathway to pression programs. Regulation of the unfolded-protein re-
PC development, remains largely intact but the GC pathway sponse by X-box-binding protein 1 is not needed for plasma
is blocked.95 Recent studies indicate that SAP-deficient TFH cell development but is necessary for antibody secretion.146,147
cells exhibit heightened CD40L expression but decreased Epstein-Barr virus-induced molecule 2 also appears es-
ICOS induction that alone can account for this defect.94 More sential for B-cell movement to extrafollicular sites and the
recent studies indicate that SAP-deficient TFH cells can only non-GC plasma cell response.148,149 In addition, Epstein-Barr
form short duration cognate contact with antigen-primed B virus-induced molecule 2 guides recently activated B cells to
cells, which are insufficient for GC entry but adequate for interfollicular LN regions and then to outer follicular areas
PC formation. It will be important to dissect the separable as a prelude to GC formation. Furthermore, there appears
signals and cellular subsets of pre-GC TFH cells that regulate to be an early pre-GC proliferative phase at the perimeter
the rapid but short-lived effector B-cell response in vivo. of follicles that also precedes GC formation and BCR diver-
Extrafollicular PCs have half-lives of 3 to 5 days133 and sification.150 Interestingly, recent dynamic imaging studies
express germline-encoded antigen-specific antibodies.131,134 indicate migration of TFH cells to the follicle interior, even
Newly formed PCs migrate via the marginal zone bridging before accumulation of GC B cells.97
channels into the red pulp of the spleen and into the medul-
lary cords of LNs. Migration patterns in vivo appear con-
trolled by increased responsiveness to the CXCR4 ligand THE GERMINAL CENTER CYCLE
CXCL12 and decreased expression of CXCR5 and CC chemo- Movement of antigen-primed B cells into the follicular re-
kine receptor 7.135 There is evidence for BCR affinity-based gions of lymphoid organs after effector TFH-cell contact
selection even at this early stage in the response.131 In some initiates the GC pathway to memory B-cell development.
studies, B cells expressing low affinity BCR remain non-GC These pre-GC B cells expand rapidly within the follicular
and high-affinity B cells preferentially enter the GC path- region to form areas of B220 + IgDlow antigen-specific B cells
way.100 The converse can also be seen with high-affi nity B referred to as secondary follicles. Secondary follicles polar-
cells preferentially entering the non-GC short-lived PC ize into T-cell proximal dark zones of cycling centroblasts
pathway.101 While seemingly contradictory, these different and opposing light zones of largely noncycling centrocytes
outcomes in vivo may reflect the plasticity of intercellular among the dense processes of FDC and sparse presence of
control at the TFH-B point of contact. GC TFH cells.151,152 This broad anatomical distribution of the
The transcriptional control of PC development also ex- GC reaction orients the activities of clonal expansion, BCR
ists in multiple layers controlling a cascade of developmen- diversification, and clonal variant selection that underpin
tal change. The transcriptional repressor prdm-1 (encoding the evolution of high-affinity memory B cells and post-GC
Blimp-1) plays a central role in the regulation of PC devel- plasma cells in this pathway (Fig. 10.3).
opment.136 B cells lacking Blimp-1 do not differentiate into Under normal physiologic conditions, the GC reaction
extrafollicular PC or post-GC PC, with TI and TD antibody emerges as the most efficient means to control affinity mat-
responses profoundly diminished in vivo.137 These Blimp-1– uration and memory B-cell development. However, in the
deficient animals also display defective levels of serum anti- disorganized or absent preimmune lymphoid subcompart-
body, suggesting that spontaneous production of antibody ments of mice lacking LTa. , LTb, tumor necrosis factor re-
by B-1-B cells also requires Blimp-1 expression. Structure- ceptor I, and LTbR, the activities of the GC reaction remain
function analysis indicates the modular action of Blimp-1 disorganized but largely intact.153–156 Vestiges of the GC or
integrating a variety of environmental cues that lead to PC small cell aggregates manage the expansion, diversification,
development.138,139 Blimp-1 represses proliferation through and selection steps required for affinity maturation and
c-myc as one direct target among many others involved in memory B-cell development.
cell cycle control. Blimp-1 also induces antibody secretion
by repressing the transcription factor Pax-5, thereby dere-
pressing Xbp-1. The transcription factor Xbp-1 controls the Germinal Center Cellular Dynamics
unfolded protein response, and many facets of the cellular Dynamic imaging of the GC reaction in situ has provided
secretory mechanism that are critical to PC function and outstanding clarity to the kinetics of GC B cell157,158 and GC
survival.140–142 TFH-cell movements and interactions.159 There was evidence
The transcriptional coactivator OBF-1 also appears im- for interzonal movement of GC B cells indicating bidirec-
portant for B cells to complete the PC program.143 In the tional migration between light and dark zones,157–160 with
absence of OBF-1, Bcl-6, Pax-5, and AID are not repressed, one study emphasizing the majority of movement to be

Germinal Germinal center reaction

center reaction
GC
FDC c Tfh
Tfh
B-cell zone
Tfh
B-cell zone
Tb Mem B B
MO B a B GC
B d B B Light
B b Apop
B
B zone
Dark
B Memory zone B
B
B B Mem B
PC B B
Affinity B B
Maturation
T-cell zone
FIG. 10.4. The Germinal Center (GC) Cycle. The GC cycle is initiated
through pre-GC contact with cognate follicular helper T (TFH) cells that
promotes extensive proliferation of antigen-primed B cells. The GC
cycle is thought to begin when immunoglobulin D– secondary follicles
polarize into micro-anatomically distinguishable dark zones (DZ) (con-
Time (days) 14 taining centroblasts) and light zones (LZ) (containing centrocytes, an-
tigen-laden follicular dendritic cell [FDC] networks, and antigen-spe-
FIG. 10.3. The Germinal Center (GC) Reaction. A: Polarization of cific GC TFH cells). Antigen-specific GC B cells’ clonal expansion in the
the secondary follicle anatomically signifies initiation of the GC DZ is accompanied by B-cell receptor (BCR) diversification through
cycle with the dark zone supporting GC centroblast expansion, SHM and antibody class switch. Both SHM and CSR are associated
class switch recombination and B-cell receptor (BCR) diversifica- with transcriptionally active gene loci, are associated with deoxyribo-
tion through somatic hypermutation. B: Noncycling GC centrocytes nucleic acid replicative and repair machinery, and occur during cell
move to the light zone and continually scan follicular dendritic cell cycle. Hence, we associated these activities with the DZ phase of the
networks. Centrocytes that lose antigen binding will undergo apop- GC cycle. Exit from cell cycle coincides with relocation of noncycling
tosis while those that express variant BCR with higher affinity can GC B cells to the LZ. Continual scanning of immune complex–coated
compete for binding to antigen-specific GC TFH cells. C: Cognate FDC is observed in the LZ and has been associated with the potential
contact with GC TFH cells requires peptide major histocompatibility for GC B cells to test variant BCR for antigen binding. Loss of antigen
class II expression by GC centrocytes that can promote reentry into binding can lead to death by apoptosis and clearance of dead cells by
the dark zones and the GC cycle or exit from the GC. D: Entry into tingible body macrophages in the LZ. Positive signals through the BCR
the affinity-matured memory B cell compartments of nonsecreting during FDC scanning program GC B cells to compete for contact with
memory B cells and post-GC plasma cells. cognate GC TFH cells. Productive contact with GC TFH cells can induce
reentry into the GC cycle through movement back into the DZ and in-
duction of cell cycle and BCR rediversification. Alternatively, affinity-
matured GC B cells can exit the GC as either a nonsecreting memory
B cell precursors for the memory response, or secreting long-lived
intrazonal.157,160 Surprisingly, cell division appeared in both “memory” plasma cells that contribute to serologic memory.
zones of the GC in contrast to classical models.157–159 There
was also evidence for the ability of naive B cells to traverse
the GC environment with evidence that high-affi nity B cells to the light zone. Importantly, movement back into the dark
could also enter and dominate existing GCs.158 All studies zone and reinitiation of proliferation was controlled by an-
identified highly motile GC B-cell movements with evidence tigen presentation to GC TFH cells.162 These studies provide
for continuous uninterrupted movement over FDC pro- experimental evidence for the reiterative cycles of BCR di-
cesses. This movement contrasted with the capacity of GC versification and positive selection as central events during
B cells to form frequent short-term contacts but infrequent affinity maturation driving clonal evolution in the antigen-
stable contacts with GC TFH cells.159 The latter group sug- specific memory B cell compartment (Fig. 10.4).
gested that these infrequent cognate contacts play a domi-
nant role in the selection of high-affinity BCR expressing
GC B cells.161 Hence, a unique model for affinity-based se- Memory B-Cell Evolution
lection emerges with uninterrupted access of GC B cells to The GC reaction requires T-cell help as it is absent in athymic
antigen-coated FDCs followed by the ability of “reprimed” nude mice, CD40- and CD40L-deficient mice, and is dimin-
B cells to express pMHCII complexes to amounts that gain ished using reagents that block or deplete TH-cell function
competitive access to pMHCII-specific GC TFH cells. such as anti-CD4, anti-CD40, and anti-CD28.163–165 Mice de-
More recently, labeling of B cells based on GC zonal ficient in ICOS also display profound defects in all aspects
location with a photo-activatable green fluorescence protein of T-cell–dependent B-cell responses.74–76 Conversely, mice
tag provided more conclusive evidence for these activities in deficient in Roquin, an inhibitor of ICOS expression, pro-
vivo.162 These elegant studies indicated that proliferation was duce excessive numbers of TFH cells and GCs and are prone to
largely restricted to the dark zone followed by a net movement autoimmunity.64 However, TI antigens can also promote GC

reactions, but they collapse within the first week after prim- frequencies. Thus, approximately one mutation is intro-
ing with no evidence of BCR diversification.166 In contrast, duced with each cell division. Analysis of mutation in “pas-
marginal zone B cells can be recruited into T-cell– dependent senger” Ig transgenes that are not under selection pressure
responses and form GCs that diversify and affinity mature.167 indicate intrinsic sequence hot spots for the mutator mecha-
Interestingly, in these studies, the marginal zone B cells re- nism.175 Hence, BCR diversification accompanies extensive
sponding to antigen challenge expressed a preimmune BCR clonal expansion within the GC reaction generating progeny
repertoire distinct from that of the follicular B cells respond- that express variant antigen-binding BCR (Fig. 10.5).
ing in the same animals. Hence, there is some redundancy to AID is the central component of the SHM mechanism.
the control of GC formation and flexibility to the origins of Originally discovered through complementary DNA sub-
the B cells recruited into the GC reaction. traction focused on novel genes in GC B cells,176 it was then
The transcriptional repressor Bcl6 is highly expressed in found to be defective in an autosomal-recessive form of
the GC and is necessary for GC formation.168 However, IgM hyper-IgM syndrome.108 Mice deficient in AID were able
and IgG1 antigen-specific memory B cells can develop in the to form the GC reaction but were unable to undergo CSR
absence of Bcl6.169 These studies were based on direct labeling or SHM.107 AID deaminates cytosine to uracil in single-
with antigen and their capacity to respond to soluble low-dose stranded DNA that can be processed by a mutagenic repair
antigen recall. However, there were no GC formed, and the pathway.105,177 The initial changes target sequence-dependent
memory B cells expressed no SHM or evidence for affinity hotspots within rearranged variable region genes of antibod-
maturation. Bcl6 has been shown to repress p53, the tumor ies. Uracil excision by uracil DNA glycosylase is then pro-
suppressor gene that controls DNA-damage–induced apopto- cessed by error-prone DNA replication to introduce point
sis.170 Regulation of p53 in this manner may protect GC B cells mutations within the actively transcribed Ig locus. Error-
to allow for the DNA breaks that are necessary intermediates prone processing using mismatch repair and base excision
in SHM and CSR. Bcl6 also directly represses Blimp-1171 and repair factors is selectively offset with high-fidelity process-
hence must be lost at some point during the GC reaction to ing to protect genome stability.178 The targeting spectrum of
allow subsequent development of memory B cells and post- AID associated with its active site can be altered to modify
GC PCs in vivo.137 variable region gene SHM179 and rate of antibody diversifica-
The GC cycle of activity regulates clonal evolution within tion.180 AID stability within the cytoplasm of Ramos B-cell
antigen-primed B-cell responders. All pre-GC B cells ex- lines can be regulated by heat shock protein 90 with specific
press some measure of antigen specificity and appear pre- inhibition leading to destabilized AID181 providing a means
selected into the memory B-cell pathway based on germline to modify the rate of antibody diversification. The details of
BCR expression. Massive and rapid clonal expansion with this mutating complex, as well as its action and regulation
doubling times of 6 to 8 hours underpins secondary follicle within the GC reaction, are active areas of research that have
formation and drives clonal expansion in the GC reaction. It been reviewed in detail elsewhere.182
is now clear that GC B cells in the light zone also have the ca-
pacity to proliferate.172 Intravital studies emphasize the more
Cell Cycle Arrest and Apoptosis
open nature of the GC itself demonstrating the capacity of
naive B cells to traverse the follicular region occupied by the The control of cell cycle in the GC environment is of funda-
antigen-responsive GC.173 Importantly, antigen-specific B mental importance to the evolution of high-affinity B-cell
cells could also be recruited into ongoing GCs if they ex- memory. Variant GC B cells exit the cell cycle and move to-
pressed sufficiently high-affinity BCR. While these studies ward the light zone to undergo selection. Dysregulated cell
use high frequencies of BCR transgenic B cells pulsed into cycle arrest promotes enlarged GCs in situ. This phenotype
the ongoing response, it is intriguing to consider the ramifi- is found in multiple genetic knockout models that also influ-
cations of such an open network of affi nity-based selection ence the composition of the memory B-cell compartment.
and its impact on the composition of the memory B-cell Large GCs form in the absence of AID and are thought to
compartment. indicate a negative feedback mechanism that follows SHM
and/or CSR.107 The absence of the CDK inhibitor p18INK4c
more directly impacts GC size through blocking cell cycle
Somatic Hypermutation arrest and also decreases the formation of long-lived PCs.183
BCR diversification is dependent on DNA replication and The large GCs in the absence of Blimp-1137 may also be re-
largely restricted to GC B cells in the pathway to memory. lated to the lack of cell cycle control that may contribute to
Upon expansion in the GC reaction, antigen-specific B cells the post-GC PC defect in these mice. The absence of the
downregulate their germline BCR and diversify their vari- main regulatory subunit of calcineurin, CnB1, results in the
able region genes through SHM. Single base substitutions, large GC phenotype, the loss of late stage antibody produc-
rare insertions, and deletions are introduced into a region tion, and the diminution of the memory B-cell response
spanning 1.5 to 2.0 kb downstream of the transcription to antigen recall.184 It will be important to assess whether
initiation site; however, activity peaks within the V(D)J these putative cell cycle defects directly impact high-affinity
region and decreases within the J-C intronic region of IgH memory B-cell evolution or indirectly modify GC B-cell fate
and IgL variable genes.174 The mutation rate approaches and the onset of apoptosis.
10 −3 per base pair per generation, which is approximately It is generally believed that the majority of GC B cells ex-
six orders of magnitude higher than spontaneous mutation pressing variant BCR will die in situ. Extensive local apop-

Dark zone
re-entry
Apoptotic
B
Cell cycle
cell proteins
c d
AID
Expansion
Dark zone Transcription
Factor
Dark zone
B B
IgG1
BCL6
Error-prone polymerase DNA repair
AID AID
SMH CSR
a IgG2a
BCL6 Transcription Transcription
Factor Factor
Commitment
to Ab class b
IgA Cell cycle,
BCL6
CSR & SHM
FIG. 10.5. The Germinal Center (GC) Dark Zone (DZ). A: Pre-GC cognate follicular helper T cues instruct antigen-
primed B cells to initiate the GC reaction. It is likely that commitment to antibody class is preprogrammed at this initial
juncture and all classes of B cells can seed the primary GC response. B: Molecular control of the cell cycle is an inte-
gral component of DZ B-cell dynamics and Bcl6 expression involved in ways that remain poorly resolved. Activation-
induced cytidine deaminase (AID) expression, cytosine deaminase activity, and uracil deoxyribonucleic acid (DNA)
glycosylase initiate the somatic hypermutation (SHM) machinery targeted to single-stranded DNA. Uracil excision is
then processed by error-prone DNA replication to introduce point mutations into the rearranged antibody variable
genes. Class switch recombination (CSR) can also occur during this DZ phase using AID to target DNA cleavage to
antibody switch regions, generate double strand DNA breaks that trigger the DNA damage machinery to complete the
CSR event. The interrelatedness of cell cycle control, SHM, and CSR is not clearly resolved in vivo. C: DZ GC B cells
will reexpress diversified B-cell receptors and enter the light zone (LZ) of the GC reaction to undergo antigen-based
selection. D: Following antigen-based selection and cognate contact with GC follicular helper T cells, LZ GC B cells
can reenter the DZ regions and rediversify the BCR to initiate secondary rounds of affinity maturation.
tosis is characteristic of the GC reaction. Furthermore, B-cell development that occurs after the fi rst TH-B–cell
overexpression of antiapoptotic molecules such as Bcl-2 or checkpoint in development.
Bcl-xL prolongs survival of GC B cells without improving
selection efficiency.185 In the absence of CD95 (Fas), there
is also a dilution of high-affi nity clonotypes in the memory Affinity Maturation
B-cell compartment.186 Cell death is a prevalent outcome Affinity maturation requires the positive selection of GC B
of the GC cycle.187 Fas receptor expression on B cells regu- cells expressing high-affinity variant BCRs. The details of
lates negative selection in the GC.188 Moreover, myeloid cell this process are still poorly understood. Receipt and integra-
leukemia sequence 1 has emerged as a major antiapoptotic tion of signals through the BCR are clearly involved in posi-
factor controlling GC B-cell formation and survival.189 tive selection and must be based on the affinity for antigen.
Positive selection of variant GC B cells must be a major Ablation of BCR signals such as calcineurin-dependent sig-
driving force within the GC and is based on the increased nals, mutations in the CD19 signaling pathway, and loss of
capacity of the mutated BCR to bind antigen. Mutations in CD45 interferes with positive selection and memory B-cell
the comodifiers of BCR signal also impact GC B-cell dy- development.172,184,190,191 However, as discussed previously, it is
namics. Unlike the CnB1 mutation, defects in CD45 and difficult to dissect the GC BCR interactions with antigen from
CD19 reduce proliferation and survival of GC B cells argu- the initial BCR triggers that recruited naive B cells into the pri-
ing for a more global impact of these molecules on BCR mary response. Most models suggest immune complex (IC)
signal integration.172,190,191 Initiation of the GC development trapping on FDCs as the most likely means for variant BCRs to
program may also be altered to differing degrees in these receive a rescuing signal from native antigen. Antigen appears
animal models in ways that are difficult to dissect. Driving rapidly in this location and can persist for extended periods of
the expression of Cre recombinase at a stage in develop- time focused to FDC in lymphoid tissue draining the site of
ment when germline Cγ1 has been transcribed has helped antigen administration.193 In support of this notion, comple-
to overcome these problems.192 This model allows the con- ment receptors CD21/CD35 on FDCs are needed to generate
ditional deletion of alleles at a late stage in antigen-driven long-term serum antibody responses.194 However, animals that

do not secrete antibody and therefore cannot form ICs can still nuclear factor of activated T cells, cytoplasmic 1 also com-
support affinity maturation.195 In the absence of inhibitor of promises B-cell responses in vivo197 but the level of the defect
NF-κB kinase 2–dependent activation of the NF-κB pathway, remains unclear. Nevertheless, as BCR signal strength must
FDCs can still capture ICs but are not stimulated to upregulate drive affinity maturation at some level, it remains important
vascular cellular adhesion molecule-1 and intercellular adhe- to resolve the B-cell intrinsic mechanisms that also help to
sion molecule-1.196 In the absence of these adhesion molecules, shape the affinity of the memory B-cell compartment.
GC B cells appear more susceptible to apoptosis with evidence
for altered gene expression and decreased affinity maturation.
Evidence connecting BCR signal strength within the GC Antigen Scanning on Follicular Dendritic
B-cell compartment and affinity maturation has been lack- Cell Networks
ing. BCR signaling and antigen presentation are required to Polarity in the GC microenvironment is partly controlled by
initiate the GC reaction and hence are difficult to manipulate the differential expression of chemokines and their receptors.198
specifically within the GC. In the absence of DOCK8, which Higher expression of CXCL12 in the dark zone assorts CXCR4-
causes compromised early immune synapse formation, early expressing centroblasts, while higher CXCL13 in the light zone
GCs still develop.14 However, without DOCK8, these GCs do attracts CXCR5-expressing centrocytes. This study used flow
not persist and GC B cells do not undergo affinity matura- cytometry and cell cycle status to identify GC B-cell subsets
tion. Calcium influx as a consequence of BCR signaling also and demonstrate the aberrant behavior of GC B cells from
appears dispensable for affinity maturation under multiple various genetically modified host animals. Recent two-photon
T cell–dependent priming conditions in vivo. Although B analysis revealed the movement of GC B cells along FDC pro-
cells deficient for stromal interaction molecule 1 and stromal cesses within the GC.159 This study emphasized the lack of dwell
interaction molecule 2 or for CNB1 exhibit profound defects time for GC B cells on FDC processes with little change in GC
in proliferation in vitro,15,18 these signaling molecules are dis- B-cell velocity upon FDC contact. Therefore, if antigen bind-
pensable for the maturation of antibody responses in vivo. ing is associated with GC B cell–FDC contact, there appears to
Downstream of BCR signals, the transcription factor myocyte be no evidence for interclonal competition between different
enhancer factor 2c is necessary for early B-cell proliferation GC B cells. These studies describe when variant GC B cells are
and GC formation,16,17 but dissecting the pre-GC versus most likely to contact antigen with the opportunity to test the
GC function remains unresolved. B cell–specific deletion of binding properties of their expressed mutated BCR (Fig. 10.6).
FDC
CR2 CD40L
FcγR
C3
IgG
FIG. 10.6. Antigen Scanning on Follicular

Dendritic Cell (FDC) Networks. FDC scan- Chemokine
ning appears as continuous movement BCR CD40 IL21R Adhesion
of germinal center B cells along mature Molecule
immune complex-laden FDC processes. Chemokine
These contacts are more similar to stromal pMCHII
Receptor
cell-associated trafficking behavior than SLAM
stable immune synapse-like interactions. FDC
Differential B-cell receptor (BCR) affinity for scanning Light zone
antigen may influence antigen-uptake and
peptide major histocompatibility class II pre- B
sentation at this juncture of development.
Programs of gene expression for molecules BCL6
able to modify cognate contact may also
be differentially induced due to BCR signal
strength through FDC scanning.

As with all in vivo studies, there are substantial influences expression also correlates with GC localization of the TFH com-
associated with the choice of models, affinities of BCR, type partment82 and the absence of programmed cell death ligand
of antigen, and method of immunization that will impact 2 on B cells impacts PC production and affinity maturation.204
the observations presented in the studies. The issues of an- Most interestingly, cytokine production and class-specific GC
tigen receptor monoclonality are self-evident in BCR trans- B cells appear to be associated in the GC well beyond the origi-
genic models in the context of repertoire studies, but are nal CSR event.205 This surprising functional pairing between
less clearly understood in association with the expression of GC TFH cells and switched GC B cells (eg, IL-4 + TFH cells with
diverse effector functions in vivo. More importantly, in the IgG1+ GC B cells; IFNγ TFH cells with IgG2a + GC B cells) hints
context of T-cell responses, it is now apparent that elevated at the extended level of heterogeneity that exits within the GC
precursor frequency deviates the dynamics of clonal expan- cycle of memory B-cell development. Hence, it is likely that
sion199,200 and can alter the development of T-cell memory in each separable class-specific GC B-cell compartment requires
vivo.201 Changing the balance or affinity of regulator popu- cognate contact with separate class-specific GC TFH cells.
lations will impact cell fate of the target populations in ways During antiprotein immune responses, the TFH cells
that remain difficult to assess. Nevertheless, new insights within the light zone of the GC express pMHCII-specific
offered by dynamic imaging continue to buttress our appre- TCRs. There appears to be sequential movement of antigen-
ciation for the workings of the immune system and serve to specific TFH cells from the T-cell zones into the GC micro-
challenge existing dogma in powerful and productive ways. environment40,206 with evidence that suggests GC TFH cells
can also move between different GCs.207 Interestingly, all
antigen-specific TFH cells responding to a protein antigen
Germinal Center Follicular Helper T Cells are not represented within the GC reaction, indicating pre-
In contrast to pre-GC effector TFH cells discussed previously, GC functional differentiation for responders TFH cells.208
CXCR5 + TH in GC will be referred to as GC TFH cells. Many Furthermore, non-GC TH cells can reemerge in a memory
CXCR5 + TFH cells were found very early in the immune re- response, indicating that the GC is not required for memory
sponse (day 3), at the T-B borders and within follicular re- TH-cell development.40 Downregulation of CD90 (Thy-1) has
gions of LNs.58 These early studies also provided evidence for been used as a marker of GC TFH cells in the mouse, although
a circulating compartment of CXCR5 + TH cells. Hence, it is when this occurs and how well it discriminates only GC TFH
clear that CXCR5 expression is not a reliable marker of GC cells is not clear.207 Interfering with CD40L-CD40 interac-
TFH cells only. Further, the expected function of GC TFH cells tions disrupts the GC reaction, while blocking B7-2 inter-
would be broadly different from those described for pre-GC action impairs memory B-cell development.209 Hence, the
effector TFH cells. The presence of GC TFH cells among the cognate control of GC B-cell fate appears to be modified by
dense FDC networks of the GC light zone predict a role in the cellular and molecular context of pMHCII presentation.
the propagation of high-affinity variant GC B cells into the The work on human TFH cells use expression of CD57, as
long-lived memory B-cell compartment198 (Fig. 10.7). well as CXCR5, to define TFH cells that are contained within
Subsequent to FDC scanning, some GC B cells were shown the GC of tonsils.63,210 Microarray analysis highlights the
to make more stable, immune synapse-like contacts with GC distantly related functional programs of CD57+ GC TFH
TFH cells, as determined by two-photon imaging.159 These early cells as compared to naive TH cells, central memory TH cells,
images gave rise to the notion that competition for GC TFH and effector memory TH cells from the peripheral blood.210
cells may be the limiting factor in GC B cell selection of vari- Differences in adhesion molecules, chemokine receptors, cy-
ant high-affinity BCRs.161 More recently, antigen presentation tokines, and transcription factors appear as distant as dif-
by GC B cells without engaging the BCR was shown to domi- ferent lineages. Interestingly, the CXCR5 ligand, CXCL13, is
nate the selection apparatus within GCs.162 GC B cells capable highly expressed in GC TFH cells. In vitro–derived TH1 and
of presenting more antigen rapidly exited the GC reaction and TH2 effector TH cells also appear distant in gene expression
produced more post-GC plasma cells. These studies indicated program to GC TFH cells.63 This analysis associates CD84,
similar mechanisms to the pre-GC selection event102 and ar- CD200, IL-21, and BCL6 with CD57+ GC TFH cells. A separate
gued strongly that antigen presentation to GC TFH cells was the analysis of TFH cells by function and gene expression demon-
rate-limiting event during affinity maturation in the GC cycle. strates that TFH activity in human tonsil is independent of
It remains technically difficult to manipulate cellular and CD57 expression.65 These studies highlight ICOShiCXCR5hi
molecular activities within the GC cycle without interfering TFH as CXCL13 secretors and the most potent inducers of an-
with the developmental programs that initiate the GC reac- tibody production in vitro. Therefore, it is clear in situ that all
tion in the first place. Many of the molecules associated with GC TFH cells express CXCR5 and many express CD57; how-
pre-GC TFH function may also function within the GC. BCL6 ever, these molecules are not exclusively expressed on GC TFH
expression itself is reinforced within TFH cells upon contact cells nor expressed on all TFH cells in the tonsil. Nevertheless,
with pre-GC B cells.98 ICOS–ICOSL interactions are impor- it remains important to clarify GC TFH phenotype so that the
tant throughout this pathway at early DC contact,202 pre-GC details of their function can be pursued in vivo.
contact,78 and likely during the GC reaction itself. IL-21 and its
receptor appear of continued importance at the pre-GC stage
and during the GC reaction.86,87 Sphingosine-1 phosphate MEMORY B-CELL RESPONSE
receptor 2 has an important role in confining GC B cells to It is now clear that centrocytes can undergo proliferation in
the GC niche in vivo.203 Elevated programmed cell death 1 the light zone,191 and there is direct evidence for movement

a
Cognate
control Mem
Chemokine
GC Receptor IgG1
PD1 Mem
TFH IgG2a
GC BCL6 Adhesion
TFH1 Co-stim
BCR Mem
TCR
IgA
Accessory
Adhesion
GC Co-stimulatory
SLAM
TFH2 pMCHII
Accessory
Cytokine
GC exit c
GC Receptor Light zone
SLAM
TFH17 B
BCL6
Mem
IgG1
Blimp1
Mem
IgG2a
b Blimp1
Mem
Dark zone IgA
Blimp1
re-entry
FIG. 10.7. Germinal Center (GC) Follicular Helper T (TFH)-B-Cell Contact. A: Longer duration cognate contact with GC TFH cells in the light
zone can be visualized directly in vivo. Similar to earlier pre-GC events these contacts must focus around T-cell receptor–peptide major
histocompatibility class II interactions and can be modified by a multitude of intercellular exchanges of molecular information. There is still
little detailed analysis of these interactions in vivo. We depict the class of molecules that can be associated with this critical programming
event but do not appreciate the organization of these interactions or their precise developmental imprint as yet. B: Antigen presentation by
B cells can influence dark zone reentry and reinitiation of B-cell receptor diversification with cell proliferation, somatic hypermutation, and
class switch recombination. C: GC cognate contact can also initiate GC exit into separable nonsecreting memory B cell and post-GC long-
lived memory plasma cell compartments.
of centrocytes back into the dark zone of GCs.159,173 Thus, it cells.187 The most typical memory B cells are the precursors
appears that one outcome of GC B cell-GC TFH interactions for a memory response to antigen recall. These memory B
is reentry into the GC cycle permitting reiterative rounds of cells are easily distinguished functionally from the second
expansion, diversification, and selection. In this model, each cellular compartment of memory, the post-GC PCs.211 The
subsequent round of clonal expansion and BCR diversifica- post-GC PCs are terminally differentiated cells that contin-
tion is applied to GC B-cell variants that have been posi- ually produce high-affinity antibody and will not be drawn
tively selected based on increased BCR affinity. Hence, the into a secondary response. In both categories of memory B
introduction of a few mutations in each clonal progeny of cells, the expression of antibody isotype distinguishes sepa-
selected variants is less likely to destroy BCR specificity and rable memory B-cell compartments. By definition, these
leaves room for further increases in an already high-affinity subsets differ in the antibody they can produce and hence
variant BCR. their developmental program and cellular function is dis-
The second major developmental outcome of cognate tinct. However, these “class-specific” memory B-cell subsets
contact in the GC is exit from the GC cycle and entry into may also differ in migration patterns, survival needs, and
the memory B-cell compartment. The GC reaction produces the requirements for reactivation at the time of antigen re-
at least two broad categories of affinity-matured memory B call. These issues define a level of heterogeneity that has re-

Memory Secondary
a response Germinal center
FDC
B-cell zone
Mem mem
f
Tfh
Memory
B Tb mem mem
MO B B
mem
mem
mem B
B B
mem
B mem
mem B mem mem
Mem B B mem
B
e mem
B B
B mem mem B
B B mem
B
c
Mem mem
B
b Tfh d mem
B
T-cell zone
Mem
PC
Mem
PC Mem
Mem PC
PC
Days (after recall) 1 5
FIG. 10.8. Memory Response to Antigen Recall. A: Memory B-cell responses can emerge in
the absence of innate inflammatory stimuli. In this case, the central antigen presenting cells
are the affinity-matured memory B cells themselves. B: Memory B-cell response to T-cell–
dependent antigens still require helper T (TH)-cell regulation upon antigen recall. When
priming and recall antigens are identical, memory TH cells are the rapid responders and are
thought to emerge preferentially over their low-frequency naïve counterparts. Regarding the
regulation of memory B-cell responses, antigen-specific memory TFH cells are the most likely
candidate for rapid cognate regulation. C: Cognate contact at this developmental juncture
occurs across sets of memory B cells and memory TFH cells with an organization and kinet-
ics that remain poorly resolved in vivo. There is rapid and vigorous local clonal expansion
within the first 2 to 3 days in both memory B and memory TFH compartments. D: Expansion
of affinity-matured memory B response plasma cells occurs very quickly with evidence that
most memory-response plasma cells already appear affinity-matured. E: There is evidence
for memory B-cell subsets that express a germinal center (GC) phenotype and create GC-
like structures upon antigen recall. Whether these structures are residual from the primary
response GC or reemerge with GC activities upon recall has not been resolved. F: Increased
numbers of memory B cells and memory-response plasma cells persist after antigen recall.
It remains unclear whether these cells are the product of memory GC reactions or extrafol-
licular non-GC memory response outcomes.
ceived very little attention at the current time and remain 8 months after priming) depending on the form of antigen
important features in the organization of B-cell memory used to immunize. In these studies, class-switched memory
in vivo (Fig. 10.8). B cells rapidly promoted plasma cell differentiation, whereas
their IgM + counterparts promoted secondary GC reactions.
Depending on the form of antigen delivery and the com-
Memory Response Precursors bination of innate stimuli, B-cell responses can be skewed
There are reports of non-GC early memory B-cell toward memory formation with extended GC reactions,
development,86,169 although how well these germline BCR- which can last over 1.5 years.214 Hence, it is possible that per-
expressing memory B cells compete with post-GC memory sistent GCs can continuously produce nonsecreting memory
B cells in the response to antigen recall remains to be evalu- B cells well after the initial priming event.
ated. Affinity-matured IgM + memory B cells can emerge Memory response precursors are more typically affi n-
from the GC reaction and persist for long periods in vivo.212 ity-matured post-GC B cells that have switched to non-
These nonswitched memory cells appear more active in IgM antibody isotypes and can respond, in a T H-cell–de-
secondary responses in the absence of circulating antibod- pendent manner, to low-dose soluble antigen recall. The
ies. Genetic labeling of AID-expressing cells with yellow long-term persistence in vivo of antigen-binding B cells,
fluorescent protein allowed memory B cells to be monitored after the decline of the primary response, provides the
over long periods.213 Surprisingly, primary response GC single best indicator for all memory response precursors.
reactions were persistent for extended periods of time (over Beyond this particular attribute, antibody isotype and

stage of development within the memory B-cell compart- mal.133,237–241 In the bone marrow, long-lived PC need signals
ment introduces a level of cellular heterogeneity that is through the tumor necrosis factor receptor family member,
only recently being appreciated. BCMA, for survival.242 It has also been proposed that a pre-
As soluble antigen is the ligand for the BCR, most experi- PC precursor243 or memory cells themselves233 produce PCs
mental approaches use variations of labeled antigen to iden- in a non–antigen-dependent manner as a means of main-
tify antigen-specific B cells. While panning techniques used taining serum antibody levels for extended periods.
gel-associated antigen to detect antigen-specific B cells,215,216 The extended longevity of the post-GC PC can be dem-
flow cytometry provided the most reliable access to antigen- onstrated using BrdU incorporation and adoptive trans-
specific B cells.217,218 These earliest studies217,218 coupled cell fer.238–241 The extinguished gene programs associated with
sorting technology and direct labeling to enrich antigen-spe- PC development142,244,245 support a terminally differentiated
cific B cells for adoptive transfer. These early studies helped end-stage cell that needs to arrest cell cycle progression183
to demonstrate that B cells with receptors for antigen were and will not be reactivated on antigen recall. Nevertheless,
the precursors for antibody-forming cells. This approach based on the evidence of a GC phase in development, the
has been adopted by many groups,219–222 with the subsequent extended longevity of these PCs and the continued produc-
evaluation of specificity and purity demonstrated by the tion of high-affinity antibody, it is reasonable to consider
frequency for antibody forming cells or the enrichment for that these end-stage B cells belong to the memory B-cell
production of antigen-specific antibodies in vitro. compartment. We have recently demonstrated that post-
Early studies indicated antigen-binding B-cell popula- GC antibody-secreting B cells not only express BCR, but
tions of long-lived cells with slow or no turnover in vivo.223 also present antigen and can modulate cognate TFH-cell
These memory B cells can survive independent of their ex- responses.132 These surprising studies further demonstrate
pressed BCR specificity and hence do not require persistent that PCs negatively regulate BCL6 and IL-21 expression in
antigen depots in vivo.224 Loss of surface IgM/IgD and ex- antigen-specific TFH cells.132 Thus, PCs are not only the pro-
pression of downstream isotype are also indicative of anti- ducers of antibody, but can also engage in antigen-specific
gen experience but are not required for memory-cell devel- immune regulation. Signals through the BCR or MHC class
opment.218,220,225 Further, the expression of mutated BCRs II on post-GC PCs may serve to regulate the ongoing pro-
with evidence for affinity increasing changes is the most duction of high-affinity antibodies in the serum. The long-
useful molecular marker for the memory B-cell compart- term antigen presenting or regulatory function of post-GC
ment.131,226–228 However, there are also abundant examples PCs has not yet been elucidated.
of germline-encoded BCRs expressed by memory response Based on variability in memory B-cell formation across
precursors. Location has also been a reliable means for iso- different antigen models, it appears likely that different sets
lating memory B cells from the blood very soon after inten- of initiating stimuli modify the balance of cells within the
tional priming.226 The combination of location, phenotype, different memory B-cell compartments. The memory cell
genotype, and time after intentional priming has many ele- balance may be a quantitative “interpretation” of the qual-
ments of a comprehensive definition for memory B cells. ity of the initial immunizing signals. Ultimately, memory
B-cell fate is likely to be controlled by interactions in the GC
itself. The strength of BCR signal and costimulatory context
Post–Germinal Center Plasma Cells serve to select variant BCR and initiate changes in memory
High-affinity antibody-producing PCs that emerge from the B-cell development. Cognate interactions with GC TFH cells
GC reaction can also be considered an integral part of anti- may consolidate these functional outcomes. Understanding
gen-specific B-cell memory. High-affinity GC B cells prefer- the role of the different memory B-cell subsets in long-term
entially assort into the PC compartment giving rise to high- protection and the rules that govern their development in
affinity circulating antibodies.229 In LNs, affinity-matured vivo remain important unresolved areas in this field.
PCs dwell in paracortical areas to mature230 then migrate
toward the medullary regions prior to export.231 CD93 is ex-
pressed at this early stage and is required for PC survival in Memory Response to Antigen Recall
the bone marrow.232 Clearly, the circulating antibody that is Persistent high-affinity serum antibody provides the first
produced by post-GC PCs contributes to ongoing serologic layer of protection against antigen recall. While binding to
immune protection.233 antigen is a clearance mechanism, it also serves to increase
Post-GC PCs do not self-replenish through turnover, the efficiency of antigen presentation to memory B cells
secrete isotype-switched antibody, and display evidence of through rapid IC formation and binding to FcR or comple-
SHM with affinity increasing mutation patterns.211,234,235 ment receptors on cells of the innate system. In this manner,
This post-GC antigen-specific B-cell compartment appears antibody may amplify the sensitivity of the memory B-cell
during the second week after initial antigen exposure131 and response to antigen recall. However, the boost with antigen
preferentially homes to the bone marrow for growth factor rechallenge also seems necessary to establish adequate long-
support of stromal cells.236 Based on gene ablation studies, term protection following protein vaccination. The cognate
these cells use a variety of redistribution mechanisms such cellular dynamics and molecular regulation of the memory
as upregulation of CXCR4, α4β1 integrin binding to its li- response boost are important factors in the consolidation
gand vascular cellular adhesion molecule-1135 to get to the of B-cell memory that remain poorly understood and inad-
bone marrow, where they can persist for the life of the ani- equately optimized in most vaccine regimes.

Primary response GCs persist for about 3 weeks after and circulating antibody levels that provide long-term im-
initial priming, but this timing may vary substantially de- mune protection.
pending on the immune stimuli.151,152 There is evidence for
continued selection in the memory B-cell compartment even
after the demise of the GC reaction.235 This selection pro- Antigen-Specific Memory Follicular Helper T Cells
cess represents more typical interclonal competition with- Memory B-cell response to protein antigens require antigen-
out further BCR diversification.246 Where this secondary specific T-cell help.187 Under normal physiologic circum-
selection occurs and how it relates to the selection mecha- stances, this secondary T-cell help must be antigen-specific
nisms in the GC remains to be determined. Nevertheless, and is most likely delivered by memory TH cells that are spe-
secondary selection events appear capable of reshaping the cific to the immunizing antigen. Under typical low-dose an-
memory compartment toward higher-affinity clonotypes tigen rechallenge regime in the absence of immune adjuvant,
that can substantially influence the quality of the secondary antigen-specific B cells are likely to be the predominant APC
response and consolidation of the memory compartment.233 for the memory response. Hence, the cognate interactions of
Recent evidence indicates that memory B cells can reinsti- pMHCII + memory B cells and antigen-specific memory TH
tute a GC reaction upon antigen recall. The form of antigen cells define a major developmental checkpoint in the evolu-
appears to have a role in primary response GC persistence tion of long-term immune protection. Alternately, circulat-
with particulate antigen more likely to promote GC longev- ing antibody and innate immune activation induce an ac-
ity.213 Moreover, innate system stimuli differentially impact celerated memory response due to increased frequencies and
persistent GC structures with combinations of toll-like re- heightened responsiveness of memory response precursors.
ceptor-4 and toll-like receptor-7 more effective than single In this scenario, the memory TH cell–memory B-cell check-
stimuli.214 Whether the secondary GC is a continuation and point may be preceded by an independent series of antigen
reexpansion of a primary GC activity remains unclear. More presentation events prior to cognate memory TFH-memory
importantly, whether these secondary or persistent GC-like B-cell contact.
structures support the rediversification of affinity-matured The major cellular outcome of the secondary boost is
BCR and the selection of even higher-affinity clonotypes re- rapid and exaggerated memory B-cell expansion and the
mains to be determined. These issues are central to the fu- production of large numbers of high-affinity PCs. While
ture management of prime-boost vaccination protocols with high-affinity B cells can be drawn into GC reactions,158 it
substantial practical impact in this field. is not clear that the GC pathway is operative at the time of
Using priming doses of antigen and adjuvant, antigen- the secondary boost. Regardless, there is clear evidence for
specific memory TH-cell responses40,41,247 and memory B-cell secondary selection events upon antigen rechallenge that
response248 emerge more rapidly than their naive response are more likely to be driven by cellular selection without
counterparts. Memory TH-cell responses reach peak levels somatic BCR diversification.187 The nature of the selection
more rapidly but to similar levels as the primary response. mechanism for B cells at this stage of the response is also
In contrast, memory B cells display accelerated kinetics and poorly understood. In contrast, the accelerated local reex-
reach substantially higher maximal levels compared to the pansion of antigen-specific memory TH cells occurs with
primary response. Even in the presence of priming doses of little change in the expressed TCR repertoire256 or cytokine-
antigen and adjuvant, the memory responders dominate the secreting potential.247 Hence, the antigen-specific memory
recall response,40,130 outcompeting naive lymphocytes that TH-cell compartment appears to conserve and reexpress
may express specific Ag-R. many of the functions associated with the initial primary
Antigen recall promotes accelerated clonal expansion and immune response.
rapid differentiation to high-affinity PCs. Initiation and mi- There is evidence for a division of labor in the memory
croclustering of IgG1 BCRs is enhanced at the single cell level T-cell compartment that involves the migratory capacity of
due to membrane proximal regions in the cytoplasmic tail memory TH-cell subsets.257 Recirculating central memory
of IgG1.249 The cytoplasmic tails of class-switched memory T cells and tissue-homing effector memory T cells further
B cells can contribute substantially to the expanded clonal divide on their readiness to reexpress effector function
burst associated with retriggering by antigen.250 There is evi- upon reexposure to antigen. It is reasonable to propose that
dence for distinct changes in BCR signaling pathways.251–254 each antigen-specific effector TH-cell subset will produce a
Increased affinity of the BCR on memory cells must also memory TH-cell counterpart.247 However, the range of TH
contribute to memory B-cell sensitivity to low-dose soluble memory effector attributes required to control antigen-
antigen that does not induce a primary immune response. specific memory B-cell responses is not well understood.
In addition to these intrinsic attributes, circulating high- Compartments of tissue-localized memory TH cells provide
affinity antibody contributes to differential management of rapid “reactive” immune protection that is antigen-spe-
antigen in vivo. Rapid presentation of immune complexes to cific.257 For example, the focal placement of effector memory
the memory B cells maybe enhanced. Furthermore, memory T-cell subsets at sites of original antigen entry preempts the
B cells require antigen-specific TH-cell regulation to initiate behavior of the pathogen. In this manner, location provides
secondary immune responses.255 These issues have not been an opportunity to accelerate the recall response enhancing
well studied but remain central to the capacity of memory B the capacity of the immune system to block overt infection.
cells to expand, self-replenish, and boost high-affinity PCs As the precursors of the memory response to antigen recall,

high-affinity memory B cells belong to a similar “reactive” restricted to the LNs draining initial vaccination site and
set of memory cells. However, in contrast to local hypersen- persistent antigen presentation induced naïve T H-cell pro-
sitivity responses by TH cells, antigen reexposure promotes liferation.261 We proposed that pMHCII complexes on im-
rapid and exaggerated memory B-cell responses in the lym- munocompetent APCs had a role in confi ning the antigen-
phoid tissue that drains the site of rechallenge. In this con- specific memory TFH-cell compartment to LNs draining
text, we can refer to the memory TH-cell compartment that the site of initial priming.262 Although it has been known
controls memory B cell responses as memory TFH cells. for some time that FDC networks are capable of trapping
We have provided evidence for the local persistence of an whole antigen as immune complexes for extended periods
antigen-specific memory TFH cell compartment. CXCR5 + of time,161 the nature of the long-lived local APC remains
TFH cells bind pMHCII tetramers, express lower levels of unresolved.
ICOS, and have lost the capacity to express messenger ribo- Unlike CD8 T cells, CD4 T H cells appear to require
nucleic acid for a series of cytokines.53 These putative mem- continued presence of antigen to reach maximal clonal
ory TFH cells rapidly upregulated cytokine signals following expansion in vivo. The conditional induction and abroga-
specific peptide restimulation in vivo. We propose that these tion of pMHCII molecules demonstrate that as soon as a
locally confined memory TFH cells are the cognate regulators lower threshold level of antigen is breached, T H cells cease
of the memory B-cell response. Many of the antigen-specific to divide.263 Surprisingly, this model induces no form of
memory TFH cells retain expression of CD69, which suggests inflammation and would otherwise be considered a model
recent contact with pMHCII complexes and provides a plau- of tolerance induction. Recent intravital imaging studies
sible mechanism for local retention. It is likely that memory also indicate that antigen-responsive T H cells engage mul-
B cells with high-affinity BCR will rapidly capture minute tiple DCs at successive early stages after priming. 264 These
levels of secondary antigen and present this antigen to the multiple DC–T H-cell contacts also impact the develop-
cognate memory TFH cells. Hence, memory B cells may be ment of effector T H-cell function in vivo. Most surpris-
the primary APC in the memory response, further acceler- ingly, there is evidence for persistent depots of pMHCII
ating the memory B-cell response to recall. complexes up to 3 weeks after viral infection.265 These
It is not clear whether memory TFH cells would neces- pMHCII depots are present at times after viral clearance
sarily express CXCR5 + like their primary response TFH in vivo. The same persistent depots can be demonstrated
compartment. Furthermore, it is not clear whether both ef- for pMHCI with the capacity for local activation of viral-
fector TFH and GC TFH produce memory TFH counterparts. specific naive CD8 cells.266 While persistent antigen is not
Nevertheless, it will be important to unravel the programs required for the maintenance of antigen-specific T-cell
of cognate control used by antigen-specific memory TFH to memory, 267,268 there may be a role for persistent depots
regulate the memory B-cell response to antigen rechallenge. of pMHC complexes as local guidance cues for antigen-
It is plausible that manipulating cellular and molecular in- specific memory T cells.
teractions at this developmental juncture can alter the shape
of antigen-specific B-cell memory.
CONCLUSION
Understanding the cellular and molecular control of anti-
Antigen Persistence In Vivo gen-specific memory B-cell development remains a high pri-
Tonic signaling through the BCR and the downstream acti- ority for basic research with important application to future
vation of phosphoinositide 3-kinase, together with B cell-ac- vaccine design. The largest growing class of pharmaceuti-
tivating factor signaling through the B cell-activating factor cal agents today is the antigen-targeting antibodies with an
receptor are required for survival of naïve B cells in the pe- endless spectrum of biologic impact. Producing potent anti-
riphery.224,258 Similarly, inducible deletion of phospholipase bodies with high affi nity is an important facet of this grow-
Cγ 2 after the formation of antigen-specific B-cell memory ing new industry.
substantially depleted the memory B-cell compartment There has been substantial progress in our understand-
and suggested a BCR signaling requirement in memory.259 ing of this complex developmental cascade in vivo with
Nevertheless, earlier genetic studies indicated that cognate many of the cellular processes being more carefully defined
BCR specificity was not required after formation of B-cell in their most relevant in vivo context. The sophisticated ma-
memory to provide the tonic survival signal.224 Thus, per- nipulation of mouse genetics provides a powerful set of tools
sistent antigen appears not to be required for the survival that begin to unravel the regulatory mechanisms operative
of antigen-specific memory B cells, although memory B-cell across each developmental checkpoint controlling cell fate
function has not been addressed in this model. in vivo. Real-time intravital imaging provides a major ad-
More recently, there has been evidence of persistent vance to our appreciation of cell dynamics and intercellu-
pMHCII complexes in the context of antiviral responses lar communication as it proceeds in the crowded confines
in vivo, 260 leading to local activation of naïve T H cells of secondary lymphoid tissue. Most importantly, a surge
even after clearance of virus. We recently demonstrated a in research activity has revealed many of the mechanisms
similar persistence of pMHCII complexes for longer than controlling TFH-cell development and function in ways that
100 days following protein antigen vaccination in a non- have changed the conceptual landscape surrounding our ap-
depot adjuvant.261 The depots of pMHCII complexes were preciation of cognate regulation in B-cell memory.

Beyond antigen recognition, antibody class determines strategy that can directly target antigen-specific adaptive re-
immune function and binding affinity controls sensitivity sponses. Unraveling the molecules and programs that control
within memory B cells. Antigen contact initiates presentation each phase of memory B-cell development provides a plethora
by B cells and a program of events, which is consolidated fol- of new targets for vaccine-based modification in vivo.
lowing B-cell contact with antigen-specific TFH cells. We have
outlined these events as a progressive developmental program
across three related but distinct phases of antigen encounter. ACKNOWLEDGMENTS
We remain cautious of manipulating critical adaptive immune I would like to acknowledge Louise McHeyzer-Williams
functions but quietly optimistic. The molecular regulation of in the collaborative unfolding of the ideas that structure
antigen-specific cellular events initiates a complex but finite this chapter and the images that illustrate the flow of ideas
set of regulatory programs that can be modified both indi- presented here. There has been immeasurable input from
rectly, following vaccination with innate stimuli, and perhaps the members of our laboratory and collaborators over the
directly, during the acquisition of high-affinity B-cell memory. years that help to shape and reshape the nature of our under-
The vaccine boost is the most readily accessible phase of this standing of these fascinating processes in vivo.

SECTION
IV T-Lymphocytes
CHAPTER
11 T-Cell Antigen Receptors
Mark M. Davis • Yueh-Hsiu Chien
T-lymphocytes expressing αβ or γδ T-cell antigen recep- the mucosal compartments.10,11 Although αβ T cells perform
tors (TCRs) are found together with B-lymphocytes in all most of the functions classically attributed to T cells, mice
but the most primitive vertebrate animals. These three cell lacking γδ T cells are clearly have a compromised immune
types are the only ones that use random variable (V), diver- defense indicating that γδ and αβ T cells contribute to host
sity (D) in the case of TCR-β and -δ, joining (J), gene rear- immune defense differently.11,12 γδ TCRs also recognize
rangement to generate diverse antigen receptors. During the antigens directly, like antibodies, with no apparent need for
last two decades, there has been a great deal of progress in antigen processing,13 at least in the most thoroughly studies
identifying the molecules and genes of TCRs, and there is cases. During the past few years, there have been consider-
considerable information on their biochemistry and struc- able advances in our understanding of antigen recognition
ture. While TCRs share structural and genetic similarities by γδ T cells. This should lead to a better understanding of
with B-cell antigen receptors (immunoglobulins [Igs]), they how γδ T cells contribute to immune competence.
also possess a number of unique features pertinent to their
specific functions. The first major difference was suggested
by the experiments of Zinkernagel and Doherty, who found T-CELL ANTIGEN RECEPTOR POLYPEPTIDES
that cytotoxic cells specific for a viral antigen could only lyse The search for the molecules responsible for T-cell recogni-
infected cells that expressed a particular major histocom- tion first focused on deriving antisera or monoclonal anti-
patibility complex (MHC) molecule.1,2 This phenomenon bodies specific for molecules on T-cell surfaces. Ultimately,
of “MHC-restricted recognition” is in marked contrast to a number of groups identified “clonotypic” sera14 or mono-
the recognition of intact antigens by Igs.3,4 Later work dem- clonal antibodies.15–19 Several of these antibodies were
onstrated that what was being recognized by these T cells, able to block antigen-specific responses by the T cells they
which were of the αβ TCR type, were fragments of antigens were raised against or, when coated on a surface, could ac-
or peptides bound to a characteristic groove in MHC mol- tivate the T cells they are specific for. They were also able
ecules.5 These αβ TCRs are expressed on classical helper to immunoprecipitate 85,000 to 90,000 molecular weight
and cytotoxic T cells, which predominate in most lymphoid (MW) disulfide-bonded heterodimers from different T-cell
compartments (90% to 95%) of humans and mice.6 They clones or hybridomas consisting of two 40,000 to 50,000
are also expressed on natural killer (NK)T cells,7 regulatory MW glycosylated subunits referred to as α and β. Peptide
T cells,8 and T cells in the mucosal sites such as the intes- mapping studies showed that there was a striking degree of
tinal epithelial compartment (IEL).9 In most cases, the αβ polymorphism between heterodimers isolated from T cells
TCR ligand is a peptide antigen bound to a class I or class II of differing specificity, thus suggesting that these antigen
MHC molecule; but in the case of NKT cells, the antigen is recognition molecules might be akin to Igs.20,21
a glycolipid bound to a nonclassical class I MHC molecule, Work in parallel to these serologic studies exploited the
cluster of differentiation (CD)1d.7 small differences (approximately 2%) observed between B-
T cells bearing γδ TCRs are less numerous than the αβ and T-cell gene expression,22 and isolated both a mouse23,24
type in most cellular compartments of humans and mice and a human25 T cell–specific gene that had antibody-
(<5%). However, they make up a substantial fraction of like V, J, and C region sequences and could rearrange in
T-lymphocytes in cows, sheep, and chickens.10 γδ T cells T-lymphocytes.24 This molecule was identified as TCR- β by
coexist with αβ T cells but seem to be better represented in partial sequence analysis of immunoprecipitated materials.26
279

280 | SECTION IV T-LYMPHOCYTES
Subsequent subtractive cloning work rapidly identified two a single C region domain (versus up to four for Igs) followed
other candidate TCR complementary deoxyribonucleic by a connecting peptide. These usually contain the cysteine
acids (DNAs) identified as TCR-α27,28 and TCR- γ.29 It was for the disulfide linkage that joins the two chains of the het-
quickly established that all antigen-specific helper or cyto- erodimer (some human TCR-γδ isoforms lack this cysteine
toxic T cells expressed TCR- αβ heterodimers. Where TCR- γ and consequently are not disulfide-linked36). N-linked gly-
fit in remained a puzzle until work by Brenner et al.30 cosylation sites vary from two to four for each polypeptide
showed that it was expressed on a small (5% to 10%) sub- with no indications of O-linked sugar addition. C-terminal
set of peripheral T cells together with another polypeptide, to the connecting peptide sequences are the hydrophobic
TCR-δ. The nature of TCR-δ remained unknown until it transmembrane regions. These have no similarity to those
was discovered within the TCR-α locus, between the Vα and of the IgH genes but instead have one (TCR-β and -γ) or two
Jα regions.31 Formal proof that the TCR-α and -β subunits (TCR-α and -δ) positively charged residues. As discussed
were sufficient to transfer antigen/MHC recognition from later, these charged residues are critical for the association of
one T cell to another came from gene transfection experi- the ligand-binding TCR polypeptides with the CD3 signal-
ments,32,33 and equivalent experiments have also been done ing polypeptides. This is important because the TCR poly-
with γδ TCRs.34 peptides have very short cytoplasmic regions with no known
As shown in Figure 11.1, all TCR polypeptides have a role in signaling.
similar primary structure, with distinct V, D in the case of A more recent member of the TCR polypeptide family is
TCR-β and -δ, J, and constant (C) regions exactly analogous the pre-Tα chain, which serves as a chaperone for TCR-β in
to their Ig counterparts. They also share many of the amino early thymocytes, similar to the role of λ5 in pre-B cells.37
acid residues thought to be important for the characteris- It was first identified and cloned by von Boehmer and col-
tic variable and constant domains of Igs.35 The C β region leagues.38 It has an interesting structure that consists of a
is particularly homologous, sharing 40% of its amino acid single Ig constant region-like domain followed by a cysteine-
sequences with CK and Cλ . The TCR polypeptides all contain containing connecting peptide, a transmembrane region
FIG. 11.1. Structural Features of T-Cell Receptors and Pre-T a Polypeptides. Leader (L), variable (V), diversity (D), joining (J), and con-
stant region (C) gene segments are indicated. Transmembrane and bold horizontal lines delineate the putative transmembrane regions;
CHO indicates potential carbohydrate addition sites; C and S refer to cysteine residues that form interchain and intrachain disulfide bonds;
R and K indicate the positively charged amino acids (arginine and lysine, respectively) that are found in the transmembrane regions.

CHAPTER 11 T-CELL ANTIGEN RECEPTORS | 281
containing two charged residues: an arginine and a lysine the γ-chain associated with the FcεRI (FcεRI γ) and FcγRIII
spaced identically to the TCR-α transmembrane region. (CD16) receptors.49,50 These CD3 subunits, in their various
The cysteine in the connecting peptide is presumably what forms, are an integral part of TCR-mediated T-cell recog-
allows heterodimer formation with TCR-β and the similar- nition because only they possess the immunoreceptor tyro-
ity to TCR-α in the transmembrane region is most likely to sine-based activation motifs (ITAMs) that are necessary for
accommodate the CD3 polypeptides. Recently, Rossjohn cellular activation when the TCR engages ligand.
and colleagues have presented a structure of the pre-Tα
chain that is comprised of a single Ig-like domain that is dis-
Characterization and Structural Features of the Cluster
tinct from the C domain of the TCR-α chain; nevertheless,
the mode of association between pre-Tα and TCR-β mir-
of Differentiation 3 Polypeptides
rored that mediated by the Cα-Cβ domains of the αβTCR.39 Figure 11.2 illustrates the principal structural features of
The pre-TCR has a propensity to dimerize in solution, and the CD3 γ, δ, ε , and ζ polypeptides as derived from gene
the molecular envelope of the pre-TCR dimer correlated well cloning and sequencing,42,51 and more recently by pro-
with the observed head-to-tail pre-TCR dimer. This mode tein crystal structures of the extracellular domains of γ,
of pre-TCR dimerization enables the pre-Tα domain to in- δ, and ε .52–55 The extracellular domains of the γ, δ, and
teract with the Vβ domain through residues that are highly ε chains show a significant degree of similarity to one an-
conserved across the Vβ and Jβ gene families, thus mim- other. These domains retain the cysteines that have been
icking the interactions at the core of the αβ TCR’s Vα-Vβ shown to form intrachain disulfide bonds and each con-
interface. Disruption of this pre-Tα-Vβ dimer interface ab- sists of a single Ig superfamily domain. The spacing of the
rogates pre-TCR dimerization in solution and impaired pre- cysteines in these domains produces a compact Ig-fold,
TCR expression on the cell surface. Their work suggests a similar to a constant region domain. The γ and δ subunits
mechanism of pre-TCR self-association that allows the pre- form distinct heterodimers with ε via highly conserved
Tα chain to simultaneously “sample” the correct folding of residues at the dimerizing interface.52–55 The connecting
both the V and C domains of any TCR β -chain, regardless peptides of the CD3 γ, δ, and ε chains all contain highly
of its ultimate specificity,39 which likely represents a critical conserved, closely spaced cysteines just before the mem-
checkpoint in T-cell development. brane-spanning regions. These residues are likely candi-
In both the mouse and humans, the cytoplasmic tail dates for the formation of interchain disulfide bonds and
of pre-Tα is much longer than any of the TCR chains appear to play a role in the assembly of the CD3 and TCR
(37 and 120 amino acids, respectively), and the murine polypeptides.52,56 The extracellular domain of the ζ chain
sequence contains two likely phosphorylation sites and consists of only nine amino acids and contains the only
sequences homologous to an SH3 domain binding region. cysteine, which is responsible for the disulfide linkage of
These are not present in the human sequence, however, the ζζ homodimer or the ζ Fc εRIγ heterodimer. Each of
and so their functional significance is unclear. 38 Thus, the γ, δ, ε , and ζ polypeptides contain a conserved, nega-
at least in principal, the murine pre-Tα molecule could tively charged amino acids in their transmembrane region
function as signaling intermediate independent of the complementary to the positive charges seen in the TCR
CD3 polypeptides, and it has recently been shown that transmembrane regions.57–60
at least one CD3 component (δ ε , see the following) is not The cytoplasmic regions of the γ, δ, ε , and ζ chains
required for it to function normally in early thymocyte are the intracellular signaling “domains” of the TCR
differentiation.40 heterodimer. Each of these molecules contains one or
more amino acid sequence motifs that can mediate cel-
lular activation.61 The intracellular sequences responsible
CLUSTER OF DIFFERENTIATION 3 POLYPEPTIDES for this activation are contained within an 18 amino acid
Immunoprecipitation of the human TCR with anti-idiotypic conserved ITAM62 with the sequence X 2YX 2L/IX7 YX 2L/I.
antibodies after solubilization with the nonionic detergent, Both of the tyrosines in this motif are absolutely required
noniodet P-40 (NP-40), initially revealed only the α- and to mediate signal transduction as mutation of either
β -chain heterodimer. However, the use of other deter- completely prevents the mobilization of free calcium or
gents, such as digitonin or Triton-X100, revealed four other cytolytic activity.63 This sequence occurs three times in
proteins.41–44 These are known as the CD3γ, δ, ε, and ζ. γ the ζ chain and once in each of the CD3 γ, δ, ε , and Fcε RI
and δ form distinct heterodimers with ε within the TCR/ γ chains. There are also pairs of tyrosines present in the
CD3 complex (γε and δε), and ζ usually occurs as a disul- cytoplasmic domains of the γ, δ, ε , and ζ chains. This
fide-linked homodimer. In mouse T cells, NP-40 does not sequence motif is also present in the mβ -1 and B29 chains
dissociate TCR heterodimers from CD3 molecules.44,45 In associated with the Ig β -cell receptor and in the Fc ε RI
some cases, the ζ-chain can be part of a heterodimer in at β -chain but there are many more10 in TCR/CD3 than in
least two forms. In mouse T cells, the ζ-chain can disulfide any other receptors which use ITAMs. The tyrosines in
bond with a minor variant called the η (eta) chain.46,47 This these cytoplasmic sequences are substrates for the tyro-
latter chain is an alternate splicing variant of the ζ-chain sine phosphorylation that is one of earliest steps in T-cell
gene.48 This alternatively spliced species of the ζ-chain is signaling61 and is thought to occur aberrantly in nonpro-
not found in significant quantities in human T cells.48 The ductive T-cell responses (eg, antagonism; see following).
second type of ζ-chain containing heterodimer contains Serine phosphorylation of the CD3 γ also occurs upon

FIG. 11.2. Structural Features of the CD3 Molecules. As in Figure 11.1, transmembrane regions, carbohydrate
addition sites, and cysteine residues are indicated. In addition, negatively charged transmembrane residues (D
for aspartic acid and E for glutamic acid) and putative phosphorylation sites are shown.
antigen or mitogenic stimulation of T cells 64 and may play still mysterious is how ligand binding by the TCR disengages
a role in T-cell activation as well. these signaling modules from the plasma membrane and trig-
Recently, live-cell imaging studies have shown a close gers the kinase cascade needed to activate T cells.
interaction of the CD3epsilon cytoplasmic domain of the TCR
with the plasma membrane, using fluorescence resonance
energy transfer (FRET) between a C-terminal fluorescent pro- Assembly and Organization of the T-Cell Receptor/
tein and a membrane fluorophore.65 Electrostatic interactions Cluster of Differentiation 3 Complex
between basic CD3epsilon residues and acidic phospholipids The assembly of newly formed TCR- α and -β chains with
enriched in the inner leaflet of the plasma membrane were the CD3 γ, δ, ε, and ζ chains and their intracellular fate have
required for binding. Nuclear magnetic resonance studies of been studied in detail.39–41,61,63 Early studies have focused
the lipid-bound state of this cytoplasmic domain revealed a on mutant hybridoma lines, which fail to express TCR on
deep insertion of the two key tyrosines of the ITAM into the their cell surface, and on transfection studies using comple-
hydrophobic core of the lipid bilayer, likely preventing their mentary DNA for the different chains in the receptor; but
phosphorylation.65 Similar studies of CD3 zeta have con- recently, Wucherpfennig and colleagues have developed an
firmed that this is also the case for that component of the elegant in vitro translation and assembly system that has
TCR/CD3 complex as well.66 This then likely defines the “off” clarified a number of important issues.54,55
state of the TCR/CD3 complex and is a novel explanation for Experiments in a nonlymphoid cell system67 have shown
how it can prevent tyrosine phosphorylation by lck. What is that TCR-α can assemble with CD3 δ and ε but not CD3 γ

and ζ. In contrast, the TCR-β chain can assemble with any suggest a highly ordered assembly process as they found
of the CD3 chains except the ζ chain. When the ζ chain that the TCR/CD3δε association facilitates the assembly of
was transfected with either α or β chain genes, or any of CD3γ ε into the complex and that the association of the TCR
the three CD3 chains, no pairwise interaction occurred. and CD3 heterodimers was a prerequisite for incorporation
Only when all six complementary DNAs were cotransfected of CD3ζζ into the complex. Importantly, the data show that
was it shown that the ζ chain could be coprecipitated with there is only one TCR heterodimer per nascent TCR/CD3
the other chains.67 Based on these data, a model has been complex in their in vitro expression system.
proposed that suggests that TCR-α pairs with CD3 δ and In contrast, a number of groups have found evidence
ε chains and that TCR-β pairs with the CD3 γ and ε chains that there can be two TCR heterodimers in a given TCR/
in the completed molecule. The ζ chain is thought to join CD3 cluster on T-cell surfaces. In particular, Terhorst and
the TCR and other CD3 polypeptides in that last stage colleagues42 showed that in a T-T hybridoma, a monoclonal
of assembly. antibody against one TCR-αβ pair could comodulate a sec-
Pulse-chase experiments have shown that all six chains ond αβ heterodimer. In addition, sucrose gradient centrifu-
are assembled in the endoplasmic reticulum (ER), trans- gation of TCR/CD3 showed a predicted molecular weight
ported to the Golgi apparatus, and then transferred to the of 300 kDa, more than 100 kDa larger than expected from
plasma membrane. It also appears that the amount of ζ a minimal δ subunit complex (α , β, γ, δ, ε2, ζ2).70 Another
chain is rate limiting, as it is synthesized at only 10% the study suggesting that there are least two TCRs in a given CD3
level of the other chains. This results in the vast major- complex is the Scatchard analysis indicating that the number
ity of newly synthesized α , β, or CD3 components being of CD3ε molecules on a T-cell surface equals the number of
degraded within 4 hours of their synthesis. The remaining αβ TCRs.71–73 Finally, there is the work of Fernandez-Miguel
nondegraded chains are long lived due to the formation of et al.,74 who showed that in T cells which have two transgenic
complete TCR/CD3 complexes with the limiting ζ chain.68 TCR-β chains, antibodies to one Vβ can immunoprecipitate
TCR/CD3 lacking CD3 ζ chains migrate through the ER and the other. It was also found that they are often close enough
Golgi intact but then are transported to and degraded in the to allow fluorescence energy transfer, meaning that the two
lysosomes. The immunologic significance of this pre-Golgi TCR-βs in a cluster are within 50 Angstroms of each other.74
degradation pathway is most evident in CD4 + CD8 + thymo- Interestingly, it appears that the TCR complexes with CD3
cytes where, despite high levels of synthesis of both messen- either have CD3 γ or CD3 δ, but not both, and these two
ger ribonucleic acid and protein for all the TCR, CD3, and ζ
chains, surface expression is relatively low. The TCR chains
in immature thymocytes seem to be selectively degraded.68
Thus posttranslation regulation appears to be an impor-
tant means of controlling the cell surface expression of
TCR heterodimers.
The TCR and CD3 γ, δ, and ε chains contain ER retention
signals.68,69 If the γ and δ signals are removed, then the chains
are transported through the Golgi and rapidly degraded in
the lysosomes. In contrast, removal of the CD3 ε ER reten-
tion signal allows this chain, and any associated chains, to
be transported to the cell surface. Thus, association of the
TCR and other CD3 chains with ε renders their ER reten-
tion signals inoperative. However, the ε ER retention signal
remains functional. This prevents the surface expression
of partial complex intermediaries until CD3 ζ is incorpo-
rated into the complex, which then masks the ε ER retention
signal and allows the transport of mature complexes to the
cell surface.
The overall stoichiometry of the αβ TCR/CD3 com-
plex is controversial. The work of Call and colleagues57 has
shown the relationships between different CD3 dimers (γε,
δε, and ζ ζ) and a single TCR αβ heterodimer (as shown
in Fig. 11.3). Using mutagenesis, they found very specific
interactions based on the positive charges in each TCR
transmembrane domain with complementary negative
charges in the transmembrane domains of the different CD3 FIG. 11.3. A Model of the ab T-Cell Receptor (TCR)-Cluster of
components. The two positively charged residues of TCR-α Differentiation (CD)3 Complex. This shows the approximate
positions of the ab TCR chains and the CD3a, d, e, and z chains
mediate interactions with the negatively charged residues of based on the membrane reconstitution and mutagenesis experi-
the of CD3δε and CD3ζζ dimers, while the single positively ments of Call and Wucherpfennig.55 More recent experiments sug-
charged residue of TCR-β mediates interactions with the gest that the CD3 polypeptides may be somewhat more clustered on
negatively charged residues of CD3γε. In addition, their data the side of the TCR heterodimer.299

receptor types are expressed in different ratios in different engineered to lack recombinase activating genes 180 or 281 are
cells. These data may not be irreconcilable, because while the unable to rearrange either TCR or Ig gene segments properly.
initial TCR/CD3 assembly may involve only one TCR, these Many of the other molecules involved in Ig gene rearrange-
may dimerize or multimerize later on the cell surface.50,75 ment serve the same function in TCRs as well.82 As with
The composition of TCR/CD3 complexes on γδ T cells is Igs, if the V region and J region gene segments are in the
distinct from that of αβ T cells and changes with the activa- same transcriptional orientation, the intervening DNA is
tion state of the cell. Biochemical analysis showed that most deleted during recombination. DNA circles of such material
murine γδ TCRs contain only CD3 γε dimers. Interestingly, can be observed in the thymus,83,84 the principal site of TCR
a differentially glycosylated form of CD3 γ was found to recombination (see the following). In the case of TCR-β and
associated with γδ TCRs dependent on the activation state TCR- δ, there is a single V region 3′ to the C in the opposite
of the cells.76 In addition, while C3ζζ is incorporated into the transcriptional orientation to J and C. Thus, rearrangement
complexes of naïve cells, activation results in the expression to these gene segments occurs via an inversion. Variable
and incorporation of FcεRI γ into the γδTCR complex.76 Using points of joining are seen along the V, D, and J gene segments
quantitative immunofluorescence, Hayes and Love have de- as well as random nucleotide addition (N regions) in postna-
rived data and proposed a model of murine γδTCR stoichi- tal TCRs. The addition of several nucleotides in an inverted
ometry in which there are two CD3γε dimers, as well as one repeat pattern, referred to a P element insertion, at the V-J
CD3ζ dimer in each TCR complex.77 Taken together, these junction of the TCR-γ chains has also been observed.85
findings strongly suggest that signal transduction through
the TCR will occur differently in γδ versus αβ T cells.
Organization of the T-Cell Receptor a/d Locus
In humans and in mice, there is a single α-chain C-region
T-CELL RECEPTOR GENES gene that is composed of four exons encoding: 1) the constant
As shown in Figure 11.4, TCR gene segments are orga- region domain, 2) 16 amino acids including the cysteine that
nized similarly to those of Igs and the same recombination forms the interchain disulfide bond, 3) the transmembrane
machinery is responsible for joining separate V and D seg- and intracytoplasmic domains, and 4) the 3′ untranslated
ments to a particular J and C. This was initially indicated region (see Fig. 11.4). The entire α /δ locus spans about 1.1
by the fact that the characteristic seven and nine nucleotide MB in both mice and humans. There are 50 different J-region
conserved sequences adjacent to the V, D, and J regions with gene segments upstream of the C-region in the murine locus.
the 12 or 23 nucleotide spacing between them, first described At least eight of these J-regions are nonfunctional because of
for Ig genes, are also present in TCRs.78 The most conclusive in-frame stop codons or rearrangement and splicing signals
evidence of this common rearrangement mechanism is that that are likely to be defective. A similar number of α-chain
both a naturally occurring recombination-deficient mouse J-regions are present in the human locus. This very large
strain (severe combined immune deficiency79) and mice number of J-regions compared to the Ig loci may indicate that
TCR
TCR β PD β1 Eβ
Vβ Vβ Vβ Dβ1 Jβ1 Cβ1 Dβ2 Jβ2 Cβ 2 Vβ14
Murine TCR α/δ Eδ TEA J α49 Eα
Vα Vδ Dδ1 Dδ2Jδ1 Jδ2 Cδ Vδ5 Jα Cα
Human TCR α/δ
Vα Vδ1 Vα Vδ2 D1D2 D3 J1 J2 J3 Cδ Vδ3 ψJα ~80 Jα Cα
D1 D2 D3 J1 J2 J3 Cδ Vδ3
Murine TCR γ
Eγ1 Eγ3 Eγ2 Eγ4
V7 V4 V6 V5 J1 C1 V3 J3 C3 C2 J2 V2 V1 J4 C4
FIG. 11.4. T-Cell Receptor Gene Organization in Mice

Human TCR γ and Humans. Schematic of V, D, J, and C elements
of the T-cell receptor genes. Transcriptional ori-
V1 V2 V3 V4 V5 V5P V6 V7 V8 VA V9 V10 VB V11 JP1 JP J1 C1 JP2 J2 C2 entation is from left to right, except where noted.
E designates enhancer elements, and S are silencer
Vγ I Vγ II Vγ III Vγ IV elements.

the functional diversity contributed by the J segment of the Organization of the T-Cell Receptor b Locus
TCR (which constitutes a major portion of the complemen- The entire human 685 kb β chain gene locus was originally
tarity-determining region [CDR]3 loop) makes an important sequenced by Hood and coworkers91 (Fig. 11.5). One interest-
contribution to antigen recognition (see the following). ing feature is the tandem nature of Jβ-C β in the TCR-β locus.
In both the murine and human loci, the Cδ, Jδ, and two This arrangement is preserved in all higher vertebrate spe-
Dδ gene segments are located between the Vα and Jα gene cies that have been characterized thus far (mouse, human,
segments. In the murine system, there are two Jδ and two chicken, frog). The two C β coding sequences are identical in
Dδ gene segments on the 5′ side of Cδ and the Cδ exons are the mouse and nearly so in humans and other species. Thus
approximately 75 kb upstream of the Cα gene and approxi- it is unlikely that they represent two functionally distinct
mately 8 kb upstream of the most 5′ known Jα gene segments. forms of C β. However, the Jβ clusters have relatively unique
The human organization is similar, with three Dα gene seg- sequences, and thus this may be a mechanism for increas-
ments and three Jδ s. Surprisingly, in both species all of the ing the number of Jβ gene segments. Together with the large
D elements can be used in one rearranged gene rather than number of Jα gene segments, there is far more combinato-
alternating as is the case with TCR-β or IgH. That is, in mice rial diversity (Jα × Jβ = 50 × 12 = 600) provided by J regions
one frequently finds Vδ D1, D2, and Jδ rearrangements,86 and in αβ TCRs than in Igs.
in humans Vδ, D1, D2, D3, and Jδ.87 This greatly increases the Most of the V-regions are located upstream of the join-
junctional or CDR3 diversity that is available, especially being and constant regions and in the same transcriptional
cause of the potential for N-region addition in between each orientation as the D and J gene element, and rearrange to
gene segment. This property makes TCR-δ the most diverse DβJβ gene via deletion. Similar to the case of Vδ5, a single
of any of the antigens receptors known, with approximately Vβ gene, Vβ14 is located 3′ to C-regions and in the opposite
1012 to 1013 different amino acid sequences in a relatively transcriptional orientation, thus rearrangements involving
small (10 to 15 amino acid) region.86 The implications for Vβ14 occur via inversion.
this and comparisons with other antigen receptor genes are In the NZW strain of mouse, there is a deletion in the
discussed subsequently. β chain locus that spans from Cβ1, up to and including the
The location of Dδ, Jδ, and Cδ genes between Vα and Jα Jβ 2 cluster.92 In SJL, C57BR and C57L mice, there is a large
gene segments suggests that TCR-δ and -α could share the deletion93 in the V-region locus from Vδ5-Vβ9. These mice
same pool of V gene segments. While there is some overlap
in V gene usage, in the murine system, four of the com-
monly used Vδ genes (Vδ1, Vδ2, Vδ 4, Vδ5) are very different
than known Vα sequences and they have not been found to
associate with Cα .88 The other four Vδ gene families overlap
with or are identical to Vα subfamilies (Vδ3, Vδ6, Vδ7, and
Vδ8, with Vα 6, Vα7, Vα4, and Vα11, respectively).
The mechanisms that account for the preferential usage
of certain gene segments to produce δ versus α chain are not
known. While some Vδ genes are located closer to the Dδ and
Jδ fragments than Vα genes (such as Vδ1), other Vδ s (such
as Vδ6) are rarely deleted by Vα Jα rearrangements and thus
seem likely to be located 5′ of many Vα gene segments.
One of the Vδ gene segments, Vδ5, is located approxi-
mately 2.5 kb to the 3′ of Cδ in the opposite transcriptional
orientation and rearranges by inversion. Despite its close
proximity to Dδ Jδ gene segments, Vδ5 is not often found
in fetal γδ T cells. Instead, the Vδ5 → DJδ rearrangement
predominates in adult γδ T cells.
An implicit characteristic of the α /δ gene locus is that a
rearrangement of Vα to Jα deletes the entire D-J-C core of
the δ -chain locus. In many αβ T cells, the α-chain locus
is rearranged on both chromosomes and thus no TCR δ
could be made. In most cases, this is due to Vα → Jα re- FIG. 11.5. Complementarity-Determining Region (CDR)3 Length Anal-
arrangement, but evidence suggesting an intermediate step ysis of T-Cell Receptor (TCR) Polypeptides versus Immunoglobulin
in the deletion of TCR-δ has been reported.89 This involves (Ig) Heavy and Light Chains. These data, modified from Rock
rearrangements of an element termed T early alpha (TEA) et al.,129 show that whereas TCR-a and -b CDR3 regions are rela-
to a pseudo-Jα 3′ of Cδ. The rearrangement of TEA to this tively uniform with respect to each other, the other antigen receptor
psuedo-Jα would eliminate the δ -chain locus in αβ T cells. pairs show a marked asymmetry. Specifically, both Ig light chains
(k and l) show very short CDR3s, as do TCR-g chains. In contrast,
Gene targeting of the TEA element resulted in normal levels
both IgH and TCR S TCRs are quite heterogeneous and tend to be
of αβ and γδ T cells, but usage of the most Jα s was severely longer. These data suggests that g d TCRs have a more antibody-like
restricted,90 suggesting that its function is to govern the ac- structure and binding properties. This has been borne out by subse-
cessibility of the most proximal 5′ Jα s for recombination. quent analysis (see text).

also express a V gene, Vβ17, that is not expressed in other There are also reports suggesting that some intestinal epi-
strains of mice. Deletion of about half of the V genes (in SJL, thelial γδ T cells develop extrathymically.107 Regardless,
C57BR and C57L mice) does not seem to have any partic- it has been suggested that the Vγ gene rearrangement is a
ular effect on the ability of these mice to mount immune programmed process.108,109
responses whereas mice that have deleted the Jβ2 cluster
show impaired responses.94
Control of Transcription and Rearrangement
It has become increasingly apparent that transcriptional
Organization of the T-Cell Receptor g Locus accessibility and rearrangement of TCR and Ig loci are closely
The organization of the mouse and human γ-chain loci linked, following the early work of Alt and Yancopoulous.110
are shown in Figure 11.4. The human γ genes span about Factors governing accessibility and rearrangement include
150 kb85 and are organized in a fashion similar to that of histone methylation,111–114 DNA methlyation, and the pres-
the β chain locus with two Jγ Cγ regions. There is more than ence of enhancer and specific promoter elements.115 Even
one nomenclature commonly used to describe the γ chain specific variations in the recombination signal sequences
genes.85,96–98 Here, we use that of Lefranc and Rabbitts99 have been shown to elicit specific biases in V(D)J joining.86
and Tonegawa and colleagues85 for the human and mouse γ With respect to enhancer elements in the TCR loci, these
chains, respectively. The organization of the human γ chain were first identified in the TCR-β locus, 3′ of Cβ2,116,117 and
genes consists an array of Vγ s in which at least six of the subsequently for the other TCR loci as well,115 as indicated in
V-regions are pseudogenes is located 5′ to these Jγ Cγ clus- Figure 11.4. These TCR enhancers all share sequence simi-
ters, and each of the V genes are potentially capable of rear- larities with each other. Some of the transcriptional factors
ranging to any of the five J-regions. The sequences of the that bind to the TCR genes are also found to regulate Ig gene
two human Cγ regions are very similar overall and only expressions. It has been shown that TCR-α enhancer (Eα) is
differ significantly in the second exon. In Cγ 2, this exon is not only important for normal rearrangement and expres-
duplicated two or three times and the cysteine that forms in sion for the α chain locus but also is required for the normal
the interchain disulfide bond is absent. Thus, Cγ 2-bearing expression level of mature TCR-δ transcripts.118 Also inter-
human T cells have an extra large γ-chain (55,000 MW) that esting is the work of Lauzurica and Krangel,119,120 who have
is not disulfide bonded to its δ -chain partner. shown that a human TCR-δ enhancer-containing minilocus
The organization of the murine γ chain genes is very dif- in transgenic mice is able to rearrange equally well in αβ
ferent than that of the human genes in that there are three T cells as in αδ T cells but that an Eα-containing construct
separate rearranging loci that span about 205 kb.100,101 Of was only active in αβ lineage T cells. Similar to Ig genes,
four murine Cγ genes, Cγ 3 is apparently a pseudogene in promoter sequences are located 5′ to the V gene segments.
BALB/c mice, and the Jγ 3 Cγ 3 region is deleted in several Although D → Jβ rearrangement and transcription occur
mouse strains including C57 Bl/10. Cγ1 and Cγ 2 are very fairly often in B cells and in B-cell tumors,121 Vβ rearrange-
similar in coding sequences. The major differences between ment and/or transcription appears highly specific to T cells.
these two genes is in the five amino acid deletion in the In addition to enhancers, there also appear to be “silencer”
Cγ 2 gene that is located in the C II exon at the amino acid sequences 3′ of Cα122,123 and in the Cγ1 locus.124 It has been
terminal of the cysteine residue used for the disulfide forma- suggested that these “repressor sites” could turn off the
tion with the δ chain. The Cγ4 gene differs significantly in expression of either of these genes, influencing T-cell dif-
sequences from the other Cγ genes (in 66% overall amino ferentiation toward either the αβ or the γδ T-cell lineage.
acid identity). In addition, the Cγ4 sequences contains a The murine TCR Cγ1 gene cluster comprises four closely
17 amino acid insertion (compared to Cγ1) in the C II exon linked Vγ gene segments, in the order Vγ 7, 4, 6, and 5, which
located at similar position to the five amino acid deletions in rearrange to a single common downstream J gene segment,
the Cγ 2 gene.101a Each Cγ gene is associated with a single Jγ Jγ1 (see Fig. 11.4). In early fetal thymocytes, rearrangements
gene segment. The sequences of Jγ1 and Jγ 2 are identical at of Vγ 5 and Vγ 6 genes predominate, and the resulting Vγ 3 +
the amino acid level, whereas Jγ4 differs from Jγ1 and Jγ 2 at and Vg4 + cells migrate to the skin or reproductive tissue, re-
9 out of 19 amino acid residues. spectively. Later in ontogeny, Vγ4 and Vγ 7 rearrangements
The murine Vγ genes usually rearrange to the Jγ Cγ gene predominate, and cells expressing these V regions migrate
that is most proximal and in the same transcriptional ori- from the adult thymus to the secondary lymphoid organs
entation. Thus Vγ1 rearranges to Jγ4; Vγ 2 to Jγ 2; and Vγ4, and the intestinal epithelium.100,101 At least two cis-acting,
Vγ 5, Vγ 6, and Vγ 7 to Jγ1. Interestingly, some Vγ genes are enhancer/locus control region (LCR) elements are present in
rearranged and expressed preferentially during γδ T-cell the Cγ1 cluster. One is a T cell–specific transcriptional en-
ontogeny and in different adult tissues as well.101 In particu- hancer, 3γEγ, located 3 kb downstream of the Cγ1 gene seg-
lar, Vγ 5 + and Vγ 6 + T cells are generated in the fetal thymus ment.125 A second element, “has,” was found between the Vγ 7
with very limited/no junctional diversity. Instead, the adult and Vγ4 genes, based on DNAse I hypersensitivity.126 Similar
thymus produces γδ T cells expressing Vγ1, Vγ 2, Vγ4, and enhancers have also been found to be associated with the
Vγ 7 gene segments with highly diverse junctional sequences. Cγ 2 and Cγ 3 genes.95 Experiments suggest that simultaneous
Moreover, γδ T cells that localize to the secondary lymphoid deletion of both enhancer elements in Cγ1 cluster severely
organs tend to express Vγ4, Vγ1, and Vγ 2, whereas those diminishes TCR-γ transcription, but only modestly reduces
that localize to the intestinal epithelium express Vγ 7.88,102–106 TCR-γ gene rearrangement, while deletion of each element

separately has little effect.127 In contrast to these results in a confined, largely two-dimensional volume (eg, between
thymocytes, deletion of “has” alone reduces transcription of two cells). Cells employing such receptors require weak,
one Vγ gene specifically in peripheral γδ T cells. Thus, the but highly specific, interactions so that they can disengage
two elements not only exhibit functional redundancy in thy- quickly.137,138 The rapid off-rates seen with TCRs (see later
mocytes but also have unique functions in other settings. section) may even amplify the effects of small numbers
of ligands.139,140
There has also been no enduring evidence for a naturally
Allelic Exclusion secreted form of either an αβ or γδ TCR. Here again, it can
In Igs, normally only one allele of the heavy chain locus and be argued that such a molecule would have no obvious use as
one of the light chain alleles is productively rearranged and the affinities are too low to be very useful in solution. Thus,
expressed, a phenomenon termed “allelic exclusion.” With for most TCRs, the concentration of protein would have to
respect to αβ TCR expression, while TCR-β exhibits allelic be extremely high in order to achieve an effect similar to
exclusion,128 TCR-α seems much less constrained,129,130 and soluble antibodies (in the milligram/milliliter range).
many mature T cells express two functional TCR-α chains. A third mechanism seen in antibodies but not TCRs is
As the chances of forming an in-frame joint with any CH switching, which allows different Ig isotypes to maintain
antigen receptor is only one in three, the probability that a given V region specificity and associate it with different
a T cell would have two productively rearranged TCRα s is constant regions that have different properties in solution
only ¹⁄³ × ¹⁄³ = ¹⁄9, or 11%. However, even when this happens, (such as complement fi xation, basophil binding, etc.). As
the two TCR-α chains may not form heterodimers equally there is no secreted form of the TCR, this feature would also
well with the single TCR-β that is expressed; thus, only one lack any obvious utility.
heterodimer may be expressed. But this simple calculation Where TCRs are equal—and in fact generally superior—
is complicated by the likelihood that only thymocytes that to Igs is in the sheer number of possible receptors that can be
make at least one productive TCR rearrangement of each generated through recombination alone. Table 11.1 summa-
type will have a chance at maturation, which would eliminate rizes the potential V region diversity that TCRs are capable
almost half of the T cells (four-ninths, which is the product of when the number of V region gene segments is multi-
of a two-thirds chance of failure on one chromosome, fol- plied by D, J, and N region diversity. It can be seen from this
lowed by the same failure rate on the second). Secondly, it table that while the V region number is generally lower in
has been found that there is a mechanism called receptor murine TCRs, particularly TCR-δ and TCR-γ, this is more
editing, which means that a given rearrangement that is not than compensated for by the degree of junctional diversity
productive, either because of an out-of-frame joint or for (where V and J or V, D, and J come together) and chain com-
reasons of self reactivity, can induce a V region further 5′ binations, such that overall TCRs have orders of magnitude
or “upstream” of the initial VJ joint to rearrange to one of greater potential diversity than Igs. This junctional region
the remaining Jas.131 In any event, it has been reported that corresponds to CDR3 as originally defined by Kabat and
one-third of human T cells express two TCRs,132 which is Wu for Igs.141 With respect to αβ TCRs, the concentration of
slightly higher than what is expected from the probabilities diversity in this region (in both chains) can be explained by
discussed previously (approximately 20%), perhaps reflect- the key role that these sequences play in recognizing diverse
ing the effect of receptor editing. peptides in MHC molecules (see later section), as supported
There also appears to be an important role for the pre- by mutagenesis and structural studies. For γδ TCRs and Igs,
TCR heterodimer (eg, pre-Tα :TCR-β) in blocking further however, the diversity is almost all in just one chain (TCRδ
TCR-β rearrangement and thus ensuring allelic exclusion and IgH, respectively), and the implications of this are
at that locus.133,134 In particular, pre-Tα –deficient mice discussed subsequently. Recent work using high throughput
had a significant increase in the number of cells with two sequencing techniques are in remarkable agreement with
productive TCR = β rearrangements, compared with wild- these crude early estimates of TCR-β diversity.142
type mice.133
The Complementarity-Determining Region 3
T-Cell Receptor Diversity Length Distributions of g d T-Cell Receptors
Although the basic organization and V(D)J recombination are More Similar to Those of Immunoglobulin
machinery are shared between TCR loci and Igs, there are than to Those of a b T-Cell Receptors
a number of striking differences. One of these is somatic Because CDR regions are loops between different β strands
hypermutation. In antibodies, this form of mutation typi- of an Ig or TCR V region (see later section), the configu-
cally raises the affinities of antigen specific Igs several order rations they adopt are generally very sensitive to their
of magnitude, typically from the micromolar (10−6 M) to the length, such that a difference of even one amino acid may
nanomolar (10−9 M) range.135,136 We now know that most cell- produce a significant change in the overall structure.3,143 A
surface receptors that bind ligands on other cell surfaces, comparison of CDR3 length distributions between the αβ
including TCRs, typically have affinities in the micromolar TCRs, γδ TCRs, and Igs (see Fig. 11.5)144,145 showed that
range (see later section) but that they compensate for this those of TCR- α and -β have a very constrained distribution
relatively low affi nity by engaging multiple receptors simul- of lengths and that these are nearly identical in size. These
taneously (eg, increasing the valency) and by functioning in length constraints may reflect a requirement for both the α

TABLE 11.1 Sequence Diversity in T-Cell Receptor and Immunoglobin Genes

Immunoglobulin TCR-a /b TCR-g/d
H k a b g d
Variable segments 250–1,000 250 100 25 7 10
Diversity segments 10 0 0 2 0 2
Ds read in all frames Rarely — — Often — Often
N-region addition V-D, V-J none V-J V-D, V-J V-J V-D1, D1-D2, D1-J
Joining segments 4 4 50 12 2 2
Variable region combinations 62,500–250,000 2,500 70
Junctional combinations ∼1011 ∼1015 ∼1018
Calculated potential amino acid sequence diversity in TCR and immunoglobulin genes without allowance for somatic mutation. The approximate number of V gene segments
are listed for the four TCR polypeptides and contrasted with immunoglobulin heavy and light chains. CDR1 and CDR2 are encoded within the V gene segments. The pairing
of random V regions generates the combinatorial diversity listed as “variable region combinations.” Because there are fewer TCR V gene segments than immunoglobulin V
gene segments, the combinatorial diversity is lower in TCRs than in immunoglobulins. Estimates for the number of unique sequences possible within the junctional region
are contrasted for TCRs and immunoglobulins. Amino acids within CDR3 are encoded almost entirely within the D and/or J region gene segments. (The last few amino acids
encoded by a TCR V gene segment can contribute to diversity within the TCR CDR3-equivalent region, but the effects of these residues on junctional diversity are not included
in these calculations.) The mechanisms for generation of diversity within the junctional region that are used for this calculation include usage of different D and J gene seg-
ments. N region addition up to six nucleotides at each junction, variability in the 3′ joining position in V and J gene segments, and translation of D region in different reading
frames. Numbers are corrected for out-of-frame joining codon redundancy and N-region mimicry of germ-line sequences. Modified from Elliott et al.86
CDR, complementarity-determining region; TCR, T-cell receptor.
and β chains of TCRs to contact both the MHC molecules rearrangements (by inversion) have been observed in some
and bound peptides on the same plane, as borne out by human tumor material.148,149 The functional significance of
structural studies (see later section). In contrast, the CDR3s this is not known.
of Ig heavy chains are long and variable, whereas those of Ig Particularly frequent is the chromosome 8–14 trans-
light chains are short and constrained. This may reflect the location [t(8;14) (q24;q11)] that joins the α /δ locus to the
fact that Igs recognize both very small molecules (eg, hap- c-myc gene, analogous to the c-myc → IgH translocation
tens) as well as very large ones (eg, proteins). Surprisingly, in many mouse myeloma tumors and in Burkitt lympho-
γδ TCR CDR3 length distributions are similar to those of mas in humans. In one cell line, a rearrangement occurred
Igs in that the CDR3 lengths of TCR-δ chains are long and between the Jα-region coding sequences, and a region 3′ of
variable, whereas those of the TCR γ chains are short and c-myc.150 In both B- and T-cell malignancies, the transloca-
constrained. Thus, on the basis of this measure of ligand tion of c-myc into IgH or TCR-α / β appears to increase the
recognition, one might expect γδ TCRs to be more similar to expression of c-myc and may be a major factor in the unreg-
Igs than to αβ TCRs. This has been validated in subsequent ulated cell growth that characterizes cancerous cells. Other
biochemical and structural studies (see later sections).
Chromosomal Translocations and Disease Chromosomal Locations of T-Cell

The chromosomal locations of the different TCR loci have TABLE 11.2 Receptor, Immunoglobulin, and
been delineated in both mouse and humans, and the results Related Loci in Mouse and Human
are summarized in Table 11.2. One significant factor in can-
cers of hematopoietic cells are chromosomal translocations Mouse Chromosome Human Chromosome
that result in the activation of genes normally turned off or
TCR-a 14 14(q11–q12)
the inactivation of genes that are normally turned on. Thus, TCR-d 14 14(q11–q12)
B- or T-lymphocyte neoplasia is frequently associated with IgH 12 14(qter)
inter- or intrachromosomal rearrangements of Ig or TCR TCR-b 6 7(q35)
loci or in some cases both.146,147 CD4 6 12
These translocations seem to mediated by the V(D)J CD8 6 2(p11)
recombinase machinery, indicating the inherent danger Igk 6 62(p12)
and need for tight regulation of this pathway. Such rear- TCR-g 13 7(p14)
CD3-g 9 11(q23)
rangements are particularly common in the α /δ locus,
CD3-d 9 11(q23)
perhaps because this locus spans the longest developmen- CD3-e 9 11(q23)
tal window in terms of gene expression, with TCR-δ being CD3-z 1 1
the first and TCR-α the last gene to rearrange during T-cell Thy-1 9 11(q23)
ontogeny (as discussed in more detail in the following). Ig-l 16 22(q11.2)
In addition, the α /δ locus is in excess of 1 mb in size, and MHC 17 6(p21)
this provides a larger target for rearrangement than either Pre-Ta 17 6
TCR-β or TCR- γ. Interestingly, in humans, TCR-α /δ is CDR, complementarity-determining region; Ig, immunoglobulin; MHC, major
on the same chromosome as the IgH locus and V H → Jα histocompatibility complex; TCR, T-cell receptor.

putative proto-oncogenes that have been found translocated or all of the carbohydrates on each chain to achieve high-
into the TCR-α / β locus are the LIM domain– containing quality crystals. An alternative is to express soluble TCRs
transcription factors Ttg-1151 and Ttg-2,152,153 which are in insect cells, where they have compact N-linked sugars,
involved in neural development; the helix-loop-helix pro- or in Escherichia coli, where they are unglycosylated. The
teins Lyl-1154 and Scl,155 which are involved in early hema- first successes in TCR crystallization come from the labora-
topoietic development; and the homeobox gene Hox 11,156 tory of Mariuzza and collaborators who solved the structure
which is normally active in the liver. How these particular of first a Vβ Cβ polypeptide161 and then a Vα fragment.162
translocations contribute to malignancy is unknown, but In the following year, the first complete αβ TCR structures
they presumably causes aberrations in gene expression that were solved.163,164 The structure of the 2C TCR, by Garcia
contribute to cell growth or escape from normal regulation. and colleagues, is shown in Figure 11.6.163 In general, as pre-
In patients with T-cell leukemia infected with the human dicted from sequence homologies, these domains are all Ig-
T-cell lymphotrophic-I virus, there are large numbers of like, with the classical β -barrel structure in evidence in all
similar translocations; it is thought that human T-cell three domains. At each end of the barrel in each V-region
lymphotrophic-I itself is not directly leukemogenic but acts domain there are four loops between the β sheets, three
by causing aberrant rearrangements in the T cell that it in- of which form the CDRs of Igs, which are numbered in
fects, some of which become malignant. Figure 11.6. The fourth loop, between the D and E strands,
Another disorder, which frequently associates with TCR has been implicated in superantigen binding. The six CDR
and Ig locus translocation, is ataxia telangiectasia, an auto- loops from the two variable domains form the antigen-
somal recessive disorder characterized by ataxia, vascular binding surface in both Igs and TCRs. The major anomaly
telangiectasis, immunodeficiency, increased incidence of in terms of similarity of TCRs to Igs is the structure of the
neoplasia, and an increased sensitivity to ionizing radiation. Cα .165 Cα consists of one-half of the classical β -barrel, that
Peripheral blood lymphocytes from patients with ataxia is, one set (or “sheet”) of β strands while the rest of the par-
telangiectasia have an especially high frequency of trans- tially truncated domain exhibits random coils. This type of
locations involving chromosomes 7 and 14.157 These sites structure is unprecedented in the Ig gene family. The func-
correspond to the TCR-γ, -β, and -α loci, and the Ig heavy tional significance of such a variant structure in unknown,
chain locus. Thus, it appears as though one of the charac- but it has been suggested that this incompletely formed Ig-
teristics of patients with ataxia telangiectasia is a relatively like domain may be responsible for the observed lability of
error-prone rearrangement process that indiscriminately TCR- α , and this may allow greater flexibility in the regula-
recombines genes that have the TCR and Ig rearrangement tion of its expression. Another possible explanation is that
signals.158 this configuration is designed to accommodate one or more
of the CD3 molecules.
The Structure of a b and g d T-Cell Receptors
As discussed previously, the sequences of TCR polypeptides
show many similarities to Igs, and thus it has long been sug-
gested that both αβ and γδ heterodimers would be antibody-
like in structure.24,25,159 The similarities between TCRs and
Igs include the number and spacing of specific cysteine resi-
dues within domains, which in antibodies form intrachain
disulfide bonds. Also conserved are many of the inter- and
intradomain contact residues and, in addition, secondary
structure predictions are largely consistent with an Ig-like
“β barrel” structure. This consists of three to four antipar-
allel β strands on one side of the “barrel” facing a similar
number on the other side, with a disulfide bridge (usually)
connecting the two β “sheets” (sets of β strands in the same
plane). All Ig variable and constant region domains have this
structure, with slight variations in the number of β strands
in variable region domains (by convention including V, D,
and J sequences) compared with constant domains. FIG. 11.6. Ribbon Diagrams of the T-Cell Receptor (TCR) Structures.
This shows the structures of the 2Cab TCR150 versus the G8g d
TCR.153 The TCR-b and the -g chains are in cyan, and the TCR-a
a b T-Cell Receptor Structure and -g chains are in vermillion. The complementarity-determining
Efforts to derive x-ray crystal structures of TCR heterodi- regions (CDRs) of both are in yellow. The very long TCR-d CDR3 in
mers and fragments of heterodimers presented many G8, which binds the T10/T22 ligands, is very apparent here but is
shorter in most other g d TCRs. Note the different C-region interac-
technical hurdles.160 One difficulty is that structure deter- tions with these TCRs and the deviations from the classic “b barrel”
mination required engineering the molecules into a solu- structure in both Ca and Cd. The prominent Cb loop to the left is also
ble form. A second problem is that many of the TCRs are unusual and may mediate interactions with CD3 or other molecules
heavily glycosylated, and it was necessary to eliminate most on the T cell surface. (Figure courtesy of Dr. K.C. Garcia)

The now large number of solved αβ TCR structures can In terms of ligand binding surfaces, we note that the Vδ
be compared to the three γδ heterodimers (discussed in CDR3 of G8 protrudes significantly away from the other
more detail in the following), and while these also resemble CDRs, as shown in Figure 11.6. This has significance in that
the Fab fragment of an antibody, there are several features this is the major region of contact with the T22 ligand (see
that are unique to the αβ molecules, which may be signifi- later section). In the case of G115, both Vδ and Vγ CDR3
cant. These include the following: loops protrude from the rest of the putative binding surface
and create a cleft between them. Portions of the CDR1γ and
1) In one structure,165 four out of seven N-linked sug-
δ and CDR2γ combine with the clefts between the CDR3
ars diffracted to high resolution, indicating that they
loops to form a pocket, which is surrounded by positively
are not free to move very much and thus are likely to
charged amino acid residues contributed by CDR2γ and δ,
play a structural role, particularly in Cα:Cβ interac-
and CDR3γ. The jagged surface of this TCR resembles the
tions. This correlates with mutagenesis data indicating
surface of an antibody that binds a small-molecule antigen.
that certain Cα sugars cannot be eliminated without
Although this would be consistent with the supposition
abolishing protein expression166 and the disordered
that this TCR binds the negatively charged phosphate com-
state of a Cα domain in the structure of a TCR lacking
pounds,170 direct binding between the TCR and phosphoan-
glycosylation.164
tigen including crystal-soaking and cocrystallization exper-
2) There is significantly more contact between Vβ and Cβ
iments have not been successful. Instead, a soluble G115 was
and between Vα and Cα than in the equivalent regions
found to bind a soluble form of adenotriphosphate (ATP)
of antibodies.
synthase F1 and apolipoprotein A-1.171
3) The geometry of the interaction of Vα and Vβ more
While the δ1A/B-3 TCR maintains an overall fold similar
closely resembles that of the CH3 domains of antibodies
to the other γδ TCR structures, it was noted that unlike the
than VHVL.
G115 and G8 CDR3 regions, which are protruding out, the
4) Between the CDR3 loops of Vα and Vβ, there is a pocket
δ1A/B-3 CDR loops together generate a nearly flat surface
that can (and does in at least one case165) accommodate
on the combining site. This difference is anticipated, as the
a large side chain from the peptide bound to an MHC.
CDR3 length distribution of the TCRδ chains is quite vari-
able as discussed previously, and like antibodies should have
a broad range of binding site shapes.
g d T-Cell Receptor Structure
There are now three γδ heterodimer structures: a γδ TCR ab T-CELL RECEPTOR–LIGAND RECOGNITION
from a human T-cell clone G115,167 which can be activated
by natural or synthetic pyrophosphomonoesters, a γδ TCR Binding Characteristics
from the murine T cell clone G8 together with its ligand, Although it has long been established that this type of T cell
the nonclassical MHC class I molecule T22,168 and a human generally recognizes a peptide bound to an MHC molecule,
MHC class I chain–related–reactive γδ TCR (δ1A/B-3),169a a formal biochemical demonstration that this was due to
which was determined as a single-chain Fv construct. The TCR binding to a peptide/MHC complex took many years to
G8 structure is shown in Figure 11.6, alongside the 2C αβ establish. Part of the difficulty in obtaining measurements of
TCR. The structure of a single human Vδ domain also has this type has been the intrinsically membrane-bound nature
been determined.169 The Vδ domain of the G115 structure is of MHC and TCR molecules. Another major problem is that
similar to the isolated Vδ domain and the quaternary struc- the affinities are relatively low, in the micromolar range,
ture of G8 is similar to that of G115.168 which is too unstable to measure by conventional means.
The most distinctive feature of both the G115 and the To some extent, the problem of measuring the interac-
G8 TCR, when compared with αβ TCRs and Igs, is that the tions of membrane-bound molecules can be circumvented
C domains “swing out” from under the V domains. This by expressing soluble forms of TCR and MHC, which is also
unusual shape is highlighted by both a small elbow angle of essential for structural studies (see previous discussion).
110 degrees, defined as the angle between the pseudo two- For TCRs, many successful strategies have been described,
fold symmetry axes that relate V to V and C to C, and a including replacing the transmembrane regions with signal
small V-C interdomain angle. This contrasts with an aver- sequences for glycolipid linkage,172 expressing chains with-
age of 149 degrees for αβ TCR structures. The small angle out transmembrane regions in either insect or mammalian
between the Vγ and Cγ domains shifts both Cδ and Cγ to cells,173 or a combination of cysteine mutagenesis and E. coli
one side. Moreover, the molecular surfaces of the constant expression.164 Unfortunately, no one method seems to work
domains are different than those of αβ TCRs with no clear for all TCR heterodimers, although the combination of
similarities either in the shape or the nature of the CαCβ insect cell expression and leucine zippers at the c-terminus
and Cγ Cδ surfaces; there are only a few solvent-exposed res- to stabilize heterodimer expression has been successful in
idues that are conserved in both Cβ and Cγ domains as well. many cases.174 The production of soluble forms of MHC
Thus, it is unclear where or how the extracellular domains molecule has a much longer history, starting with the enzy-
of the CD3 subunits interact with the extracellular portions matic cleavage of detergent solubilized native molecules175 as
of γδ TCRs compared with αβ TCRs. This may explain why well as some of the same methods employed for TCR such as
the CD3 components of αβ TCRs are so different from those glycophosphatidylinositol (GPI) linkage,176 E. coli expression
of γδ TCRs. and refolding,177,178 and insect cell expression of truncated

(or leucine zippered) molecules.179 One interesting variant radioactive labels. Recently, microcalorimetry has also been
that seems necessary for the stable expression of some class used to measure some TCR ligand affi nities; and these anal-
II MHC molecules in insect cells has been the addition of a yses have confirmed the surface plasmon resonance (SPR)
covalent peptide to the N-terminus of the β chain.180 values,184 but do not allow kinetic measurements. These and
The first measurements of TCR affinities binding to other data138,160 showed definitively that TCR and peptide-
peptide/MHC complexes were performed by Matsui et al.181 loaded MHC molecules alone are able to interact and also
and Weber et al.182 Matsui and colleagues used a high con- that expression in a soluble form has not altered their ability
centration of soluble peptide/MHC complexes to block the to bind to each other. As shown in Table 11.3, SPR measure-
binding of a labeled anti-TCR Fab fragment to T cells spe- ments show that while the on-rates of TCRs binding to pep-
cific for those complexes, obtaining an equilibrium binding tide/MHC molecules vary from very slow (1,000 M sec) to
affinity (Kd) value of approximately 50 μM for several differ- moderately fast (200,000 M sec), their off-rates fall in a rela-
ent T cells and two different cytochrome peptide/I- Ek com- tively narrow range (0.5 to 0.01 sec−1) or a t ½ of 12 to 30 sec-
plexes (as shown in Table 11.3). Weber and colleagues used onds at 25°C. This is in the general range of other membrane
a soluble TCR to inhibit the recognition of a flu peptide/I-Ed bound receptors that recognize membrane molecules on
complex by a T cell and obtained a K D value of approximately other cells,137 but it has been noted that most TCRs have very
10 μM. While these measurements were an important start slow on-rates,138 which reflects a flexibility in the binding site
in TCR biochemistry, they gave no direct information about that might help to foster cross-reactivity (see the following).
the kinetics of TCR-ligand interactions. Fortunately, the In the case of a class I MHC–restricted TCR, 2C, this rela-
development of surface plasmon resonance instruments, tively fast off-rate may be stabilized (10-fold) if soluble CD8
particularly the BIAcore™ (Pharmacia Biosensor, Uppsala, is introduced,165 but this result is controversial.185 CD8 stabi-
Sweden) with its remarkable sensitivity to weak macromo- lization of TCR binding has been seen by Luescher et al. in
lecular interactions,183 has allowed rapid progress in this cell-based TCR labeling assay186 ; however, no enhancement
area. In this technique, one component is covalently cross- of TCR binding has been seen using soluble CD4187 (see the
linked to a surface and then buffer containing the ligand is following for more discussion of CD4 and CD8).
passed in solution over it. The binding of even approximately More recently, new methodologies have been utilized to
5% of the surface-bound material is sufficient to cause a measure TCR binding in its native state, that is, on a T-cell
detectable change in the resonance state of gold electrons on surface and interacting with peptide-MHC ligands on either
the surface. This method allows the direct measurement of an artificial bilayer188 or on the surface of a red blood cell189
association and dissociation rates, that is, kinetic parame- “reporter.” These types of measurements are important
ters, and also has the advantage of requiring neither cells nor because studies of TCRs and peptide/MHCs binding in
TABLE 11.3 T-Cell Receptor-Ligand Binding

T Cell Ligand KD (mM) k on (M−1s−1) Koff (s−1) Method Reference
TH Cells
5C.C7 MCC/Ek 50 — — Anti-TCR comp. 166
2B4 MCC/Ek 50 — — Anti-TCR comp. 166
2B4 MCC/Ek 30 — — Anti-P/MHC comp. 166
2B4 MCC/Ek 90 600 0.057 BlA1 166
228.5 MCC 99E/Ek 50 — — Anti-TCR comp. 166
14.3d Flu H1N1/Ed ∼10 — — Sol. TCR 167
14.3d SEC 1,2,3 5.4–18.2 >100,000 >0.1 BlA1 251
HA1.7 HA/DR1 >25 — — BlA1 300

300
HA1.7 SEB 0.82 13,000 0.001 BlA1
Tc cells
2C p2Ca/Ld 0.5 11,000 0.0055 Anti-TCR comp. 176
2C p2/Ca/Ld 0.1 21,000 0.026 BlA1 301
2C OL9/Ld 0.065 53,000 0.003 Labeled MHC 176
4G3 pOV/Ld 0.65 22,000 0.02 Labeled MHC 176
42.12 OVA/Kb 6.5 3,135 0.02 BlA4 179
2C p2Ca/Ld 3.3 8,300 0.027 BlA1 208
HY M80/Db 23.4 6,200 0.145 BlA1 208
HY CD8 a /b + M8/D/b 2.0 5,100 0.01 BlA1 208
2/C CD8 a /b + p2Ca/Ld 0.32 1,200 0.0038 BlA1 208
Tgd cells
274
G8 T10/T22 0.13 65,000 0.0081 BlA1
BlA1, TCR amine coupled; BlA2, TCR cysteine coupled; BlA3, MHC-peptide amine coupled in competition experiment; BlA4, TCR coupled by using H57 antibody and MHC coupled
via amine chemistry; MHC, major histocompatibility complex; P, peptide; sol., soluble; TCR, T-cell receptor.

solution only partially reflect what is happening between two a ligand? One of the most intriguing discoveries concerning
membranes. This is because of the very severe constraints T-cell reactivity has been the phenomenon of altered pep-
on binding that are present in this specialized environment. tide ligands. These are single amino acid variants of anti-
For one, it is likely that TCRs on T cells and peptide/MHCs genic peptides that either change the nature or degree of the
on antigen-presenting cells are fi xed into a fairly rigid align- T-cell response (partial agonists) or prevent a response to a
ment that maximizes their ability to bind to each other, as normally stimulating ligand (antagonists).193,194 Discussions
opposed to the same molecules in solution that are diffusing concerning the mechanism of these “altered peptide”
in all directions and for which only a fraction of collisions responses have centered on whether they are due to some
will result in a productive binding event. Secondly, binding conformational phenomenon involving TCRs and/or CD3
is happening in a very small space/volume of liquid, making molecules or to affinity or kinetic characteristics. With the
the effective concentration very high. While this environ- data now available, we can now say that most, but not all,
ment is not strictly two-dimensional, Dustin and others T-cell responses correlate well with the binding characteris-
have used this as a simplifying assumption with relative suc- tics of their TCRs. In particular, Sykulev et al.195 first noted
cess. But a direct measurement of TCR affinity and kinetics that higher-affinity peptide variants elicited more robust
has been made by Huppa and colleagues,188 who used a fluo- T-cell responses. Subsequently, Matsui et al.196 found that in
rescent probe on an anti-TCR antibody to create a system in a series of three agonist peptides increasing dissociation rates
which this fluorophore could form a FRET pair with a com- correlated with decreasing agonist activity. Lyons et al.197
plementary label on the c-terminus of the peptide that could found that this correlation extended to antagonist peptides
emit photons of the appropriate wavelength only when the in the same antigen system (moth cytochrome c/I-Ek). They
TCR bound its peptide-MHC ligand. This enabled precise also showed that while an antagonist peptide might differ
measurements of the dissociation rate (koff) and affi nity only slightly in affinity compared with the weakest agonist,
(KD) within an immunologic synapse, and these measure- its dissociation rate differed by 10-fold or more. This data in
ments could also be used to estimate an average association a class II MHC–restricted system is largely supported by the
rate (kon). Another advantage of these measurements is that studies of Alam et al. in a class I MHC system,198 who also
they can be done at 37°C, whereas most of the SPR mea- saw a drop-off in affinities and an increase in off-rates (with
surements cited previously were performed at 25°C due to one exception as noted in Table 11.2) with antagonist ver-
instrument limitations. Dissociation rates at higher temper- sus agonist ligands. In the cell-based TCR labeling system of
atures are significantly faster. As expected from the previ- Luescher and colleagues, a survey of related peptide ligands
ous discussion, the association rate was greatly accelerated of varying potency also found a general, but not absolute,
approximately 100 × over the solution rate, presumably be- correlation between receptor occupancy and stimulatory
cause of the ideal orientation of the TCR and peptide/MHC ability.199 Thus, while there is a general trend toward weaker
with respect to each other. Not expected was an accelerated T-cell responses and faster off-rates and lower affi nities, this
dissociation rate of 4 to 10 times faster. This turns out to does not seem to be an absolute rule, and thus other factors
be a function of actin polymerization and depolymerization may be important in some cases. Alternatively, Holler200 has
activity within the T cell, as shown dramatically in the actin suggested that some or all of the discrepancies may derive
dynamics studies of Vale and colleagues.190 Once this was in- from differences in peptide stability (in the MHC) between
hibited, the dissociation rates within the synapse were quite the relatively short (minutes) time scale of BIAcore analy-
close to the solution values.191 Still, it is worth noting that sis at 25°C compared with the much longer (days) cellular
TCR binding to peptide/MHC is even more unstable within assays at 37°C. But this explanation probably only applies to
an active synapse than the solution measurements have sug- a fraction of the anomalous cases in which TCR ligands fail
gested, with half lives of less than 1 second in many cases. to adhere to the “t ½ rule.” In particular, Krogsgaard et al.201
A second experimental system that has been applied performed a very comprehensive survey of both known and
to the question of TCR binding in the context of cell-cell newly derived cytochromic peptide antigens and found that
contact has been that of Chung and colleagues,189 in which almost half of the peptide-MHC ligands analyzed failed to
red blood cells coated with peptide-MHC complexes are exhibit a linear relationship between t ½ and T-cell stimula-
brought into repeated contact with a T cell of the appropri- tory ability (Fig. 11.7). Extensive thermodynamic analyses
ate specificity. By visualizing the response of the peptide/ of these ligands showed that one particular parameter, the
MHC–coated red blood cell to different frequencies and du- change in heat capacity (−ΔCp), which reflects changes in
rations of T-cell contact, one can infer parameters of TCR conformation or flexibility upon binding, seems to be syn-
affinity, at least in the two-dimensional aspect.192 While ergistic with t ½ in enhancing a ligand’s stimulatory capacity.
this system has worked well in describing integrin binding, In fact, as shown in Figure 11.7, when ΔCp values are com-
it is indirect and it some ways contradicts results obtained bined with t ½, the x axis values for the range of ligands cor-
with the FRET system of Huppa et al.188 and even some of relate much better with T-cell stimulation. This suggests
the basic properties determined in solution measurements. that ΔCp may be the “missing” variable in correlating ligand
Nonetheless, it may be telling us important properties of binding to stimulation.
cell-cell contact that are more difficult to discern with more How could a negative change in heat capacity synergize
directly molecular methods. with the stability of binding? One possibility is that large
To what extent are we now able to predict a T-cell conformational changes at the binding surface of a TCR
response based on the binding characteristic of its TCR to that can occur when it engages peptide/MHC ligands (as

the pathway, in this case gene transcription in the nucleus.

Thus antagonism may occur at one threshold and an ago-
nist response at another. Alternatively, an antagonist ligand
may traverse the activation pathway just far enough to use
up some critical substrate, as proposed by Lyons et al.197 Yet
another possibility that has also been suggested is that some
antagonists may act even earlier by blocking TCR clustering
at the cell surface.206 Lastly, Germain and colleagues have
found evidence that a feedback loop involving the phospha-
A tase SHP-1 may act as an alternative pathway to inactivate
TCR signaling of insufficient strength.207,208
Another controversy that relates to TCR binding charac-
teristics is the “serial engagement” model of Valittutti and
Lanzavecchia and colleagues,139,209 which proposed that one
way in which a small number of peptide/MHC complexes
can initiate T-cell activation is by transiently binding many
TCRs in a sequential fashion. While the dissociation rates
reviewed here show that TCR binding is likely to be very
transient, they do not in fact support the statement that
more interactions are better. This is because, in most cases,
improvements in TCR-peptide/MHC stability within any
one system result in a more robust T-cell response. This has
B been shown by the work of Kranz and colleagues,210 who
selected a nanomolar affinity TCR from a mutagenized
FIG. 11.7. Dissociation Rate and Heat Capacity with Peptide–Major library expressed in yeast. With an approximately 100-fold
Histocompatibility Complex (MHC) Ligands Both Influence T-Cell slower off-rate than the original, this TCR should have been
Activation. Data from Krogsgaard et al. show that where deviations
from the general dependence of T-cell activation on dissociation only poorly stimulatory based on the serial engagement
rate occur, they may be compensated for by another factor, namely model. Instead, T cells bearing it are considerably more
the heat capacity (ΔCP) of the T-cell receptor (TCR)-peptide/MHC sensitive to antigen, which casts considerable doubt on this
interaction. This is a measure of changes in mobility or conformation aspect of the model. Similar work of Allen and colleagues on
during the binding interaction and suggests that other binding pa- another TCR-ligand interaction produced similar results.211
rameters can influence TCR-mediated activation besides half-life.184 Furthermore, recently Xie et al. cross-linked peptide-MHCs
A: The relationship between the half-maximal peptide concentration
to a cognate TCR and found that when aggregated they can
needed for T-cell activation (EC50) and the half-life (t½) of the TCR
binding to that particular ligand. In this series, three of the seven still signal efficiently.212 But the concept that the rapid disso-
peptides tested (K2, K3, and PPC) do not seem as dependent on t½ as ciation rates of TCRs for peptide-MHC ligands could serve
the others (PCC* represents a correction for a lack of stability when to amplify signaling under limiting conditions is still very
bound to the MHC at 37°C). B: When ΔCP is factored in, all of the likely and has been cited in the context of the “pseudodimer”
peptides can be plotted on a line. model,140,213 as discussed in the following.
shown by structural studies; see the following) can trans-

locate the TCR deeper into the membrane—as suggested Topology and Cross-Reactivity
by the “piston”52,202 and “twist cap”184 models—and trig- As discussed previously, TCR sequence diversity resides
ger conformational changes in the CD3 signaling domains. largely in the region between the V and J region gene
Chakraborty and colleagues recently suggested another segments, which corresponds to the CDR3 regions of anti-
possibility for how conformational changes at the binding bodies.214 This has led to models in which the CDR3 loops of
surface might exert their effect.203 They found that in the Vα and Vβ make the principal contacts with the antigenic
context of membrane-membrane interactions, conforma- peptide bound to the MHC.214 Support for this model has
tional changes in the surfaces of relatively rigid proteins come from many studies in which it has been shown that
(such as all but the CDR3 regions of TCRs appear to be) the CDR3 sequences of TCRs are important predictors of
would act to increase the effective half-life of TCR-peptide/ specificity,214 as well as elegant mutagenesis studies which
MHC interactions. Thus t ½ and ΔCp may be equivalent in showed that a single CDR3 point mutation could alter the
the unique environment between two cell surfaces. specificity of a TCR 215 and also a CDR3 “transplant” could
How might the relatively small differences in the bind- confer the specificity of the donor TCR onto the recipient.216
ing characteristics of the ligands cause such different In addition, a novel approach to TCR-ligand interactions
T-cell signaling outcomes as agonism or antagonism? As was developed by Jorgensen et al.,217 who made single amino
McKeithan204 and Rabinowitz205 have noted, a multistep acid changes in an antigenic peptide at positions that af-
system such as T-cell recognition has an inherent abil- fect T-cell recognition but not MHC binding. These vari-
ity to amplify small differences in signals that are received ant peptides are then used to immunize mice that express
on the cell surface to much larger differences at the end of either α or β chain of a TCR that recognizes the original

recognition constitutes a major departure from antibody-

antigen interactions and may reflect a need to accommo-
date other molecules into a particular configuration that
is optimal for signaling, such as CD4, CD8, and/or CD3
components. Particularly relevant to this possibility are the
recent observations by Garcia and colleagues222 suggesting
that TCR signaling can be limited by the docking geometry
of the TCR to the peptide/MHC complex. They used a yeast
library in which millions of different peptides bound to a
class I MHC can be interrogated by a single TCR. Structural
FIG. 11.8. Sensitivity of T-Cell Receptor (TCR) Complementarity- analysis showed that three of the peptides isolated bound
Determining Region 3 (CDR3) Sequences and Va / Vb Usage to to the TCR with the typical orientation and induced TCR
Changes in the Antigen Peptide. This figure summarizes the data signaling. But a fourth peptide bound the TCR not in the
of Jorgensen et al.196,198 and Sant’Angelo et al.,197 who immunized typical diagonal mode but in a parallel fashion, and failed
single-chain transgenic mice (TCR-a or TCR-b) with antigenic pep- to induce signaling.222 This suggests that the typical TCR-
tides (MCC or CVA) altered at residues that influence T-cell recogni- MHC docking mode is selected for in the thymus and may
tion but not major histocompatibility complex binding. These data
show that such changes invariably affect the CDR3 sequences of
be critical for signaling.
Va or Vb or both, and that there appears to be a definite topology As αβ TCR heterodimers are first selected in the thymus
in which Va governs the N-terminal region and Vb seems more for reactivity to self-peptides bound to MHC molecules, all
responsible for the C-terminal portion of the peptide. foreign-peptide reactive TCRs could be considered to be
inherently cross-reactive. Indeed, a number of T cells have
reactivity to very different peptide sequences, as shown by
peptide and the responding T cells are analyzed. Using these Nanda and Sercarz.223 It has also been argued by Mason224
“ hemitransgenic” mice allows the resulting T cells to keep that the universe of peptides is so large that each T cell must
one-half of the receptor constant, while allowing consid- on average be cross-reactive to approximately 106 differ-
erable variation in the chain that pairs with it. The results ent peptides (although many of the differences in peptide
from this study and work in another system by Sant’Angelo sequence in this calculation would not be accessible to the
et al.218 are very similar in that every mutation at a TCR- TCR, being buried in the MHC binding groove). Several
sensitive residue triggered a change in the CDR3 sequence large-scale screens of a random 9-mer peptide library with
of Vα , Vβ, or both, and in some cases, changed the Vα or different T cells did turn up a great many stimulatory
Vβ gene segment as well (as summarized in Fig. 11.8). One
of the more striking examples of a CDR3-peptide interac-
tion occurred in the cytochrome c system where a Lys →
Glu change in the central TCR determinant on the peptide
triggered a Glu → Lys charge reversal in the Vα CDR3 loop,
suggesting a direct Lys → Glu contact between the two mol-
ecules.217 This prediction has recently been confirmed in the
structural work of Newell et al.219 and extended to a closely
related TCR, 2B4, which contacts the central lysine on the
peptide with an oppositely charged residue in CDR3β.
Another interesting finding was the order of Vα → Vβ
preference going from the N-terminal to the C-terminal
residues of the peptides. This led Jorgensen to a proposed
“ linear” topology of TCR-peptide/MHC interaction in
which the CDR3 loops of Vα and Vβ line up directly over
the peptide.217,220 Sant’Angelo et al.218 proposed an orienta-
tion of the TCR in which the CDR3 loops are perpendicular
to the peptide. This was partially based on intriguing data
they found suggesting an interaction between the CDR1 of
Vα and an N-terminal residue of the peptide.
This controversy regarding orientation has been largely
resolved by the numerous crystal structures of TCR-peptide/
MHC complexes.160,221 These studies show that TCRs bind
in roughly diagonal to 90-degree configurations, ranging
over 30 degrees. In these structures, one of which is shown
in Figure 11.9,165 the CDR3 loops are centrally located over FIG. 11.9. T-Cell Receptor–Peptide/Major Histocompatibility
the peptide, but the Vα CDR1 and the Vβ CDR1 are also Complex Crystal Structure of a T-Cell Receptor–Peptide/Major
in a position to contact the N-terminal and C-terminal Histocompatibility Complex Complex. Peptide and complementary-
peptide residues, respectively. The confined nature of TCR determining regions are portrayed in different colors.150

Direct evidence of TCR plasticity was first obtained by Garcia

et al.,229 who in comparing the x-ray crystal structures of the
same TCR bound to two different peptide-MHC ligands
found a large conformational change in the CDR3 loop and
a smaller one in the CDR1α loop. An even larger conforma-
tional change (15 Å) has been found in the CDR3β residue
of another TCR as it binds to a peptide/MHC complex,230
as shown in Figure 11.10. That each TCR may have many
different conformations of its CDR3 loops is suggested by
the two-dimensional nuclear magnetic resonance studies of
Reinherz and colleagues, who found that the CDR3 regions
of a TCR in solution were significantly more mobile than
the rest of the structure.231 That this may be a general fea-
FIG. 11.10. Complementarity-Determining Region (CDR)3 Movement ture of most TCRs is supported by thermodynamic analy-
in ab T-Cell Receptor (TCR) Binding. In most cases where there
is a structure for an ab TCR as well as for one (or more) for that ses of various TCRs binding to their peptide/MHC ligands,
TCR in complex with a peptide–major histocompatibility complex, both class I and class II. Here, the binding is often accom-
there is a marked movement of one or the other of the Va or Vb panied by a substantial loss of entropy (Table 11.4) and, in
CDR3s. This example from Mallissen et al.209 shows a particularly most cases, an “induced fit” mechanism.184,192,232 This seems
large (14 A) movement of CDR3b with binding. This meshes well with to be a situation where an inherently flexible binding site
thermodynamic data showing an “induced fit” binding mechanism achieves greater order upon binding. This is a mechanism
for most TCRs.173,211
which is also employed by DNA recognition proteins, and
Boniface et al.192 have suggested that it might represent a
peptides, but very few have changes in the two to four key common mechanism of “scanning” an array of very similar
TCR-sensitive residues.225 This led Garcia and colleagues to molecular structures (MHCs or DNA) rapidly for those few
conclude that cross-reactivity to different agonist peptides that “fit” properly. We have seen previously that the associa-
may not be a general feature of T-cell specificity.226 Instead, tion rates are often remarkably slow, with Kas ranging from
the major requirement for cross-reactivity may be in thymic 1,000 to 10,000 M−1s −1 (see Table 11.3). This indicates that
selection227 and/or in the use of endogenous peptide-MHCs either a multistep process is occurring before stable bind-
to augment T-cell sensitivity to agonist ligands.207,213,228 ing can be achieved or that only a fraction of the TCRs in
Thermodynamic and Structural Parameters for T-Cell Receptor Peptide–Major Histocompatibility

TABLE 11.4
Complex Interactions
DG° DH° DCp TCR Conformational pMHC Conformational
TCR pMHC (kcal/mol) (kcal/mol) (kcal/mol) Change Change Reference
2C dEV8-H2-K b
−6.3 −22.7 −1.1 CDR3a (6 Å) None a 169,208
2C p2Ca-H2-Kb −6.1 −29 −1.5 ND ND 169
2C SIYR-H2-Kb −7.2 −8.4 −1.1 CDR3a (3.9 Å) None 169,208
2C dEV8-H2-Kbm3 −5.8 ND ND CDR3a (6 Å) None 305,308
JM22z MP(58–66)-HLA-A2 −7.1 −23 ND — Q155-A2(2.4Å) 211,307
KB5-C20 PKB1-H2-Kb ND ND ND CDR3b (15 Å) None 209
LC13 EBNA3A-HLA-B8 −6.8 ND ND CDR3a (2.5 Å); (Q155)-HLA-B8 304
CDR1a (1.9 Å):Cα

A6 Tax-HLA-A2 −8.2 ND ND CDR3b (4.4 Å) A2-a 2(1.4Å) 149,209
BM3.3 VSV8-H2-Kb −5.4 ND ND CDR3a (5.3 Å) None 303
D10 CA-I-Ak −7.0 ND ND CDR3b None 172,210
F5z AM9-H2-Kb −6.7 −19 ND ND ND 211
2B4 MCC-I-Ek −6.9 −13 −0.6 ND ND 169,173
2B4 K2-I-Ek −6.9 −9.4 −2.1 ND ND 169
2B4 K3-I-Ek −6.1 −30.5 −4.0 ND ND 169
2B4 K5-I-Ek −7.5 −8.0 −1.2 ND ND 169
2B4 102S-I-Ek −5.6 −13.2 −0.3 ND ND 169
2B4 PCC-I-Ek −6.1 13.3 −1.8 ND ND 169
2B4 PCC-103K-I-Ek −7.0 −8.4 −1.0 ND ND 169
172.10 MBP(1–11)-I-Au −6.9 −21.2 −0.16 ND None 305
1934.4 MBP(1–11)-I-Au −6.0 −15.7 −1.2 ND ND 305
D3 SL9-HLA-A2 −7.5 −10.4 −0.4 ND ND 309
The ΔG values were derived from the equation ΔG = −RT ln(KA) where R = 0.001987 kcal/mol/K. CDR, complementarity-determining region; ND, not determined; p/MHC, peptide/
major histocompatibility complex; TCR, T-cell receptor.
a
No major conformational change observed after ligand recognition.

solution have the correct conformation. Just how such a colleagues,236 have documented such specific footprints that
scanning mechanism might work for TCRs is seen in the correlate to particular Vβ usage, although it does seem that
analysis of Wu et al.,233 who found that a cytochrome c/ there are multiple ways way in which a particular V region
MHC II–specific TCR derived most of its stability of bind- can contact a particular MHC.
ing from antigenic peptide residues, but very little of its ini- The rules for the conserved reaction of αβ TCRs with
tial activation energy from these residues. In contrast, MHC MHC proteins plus peptides are addressed by evaluating
residues contributed by far the most of the initial binding, the contact residues between TCR and MHC in cocrystal
but had relatively modest effects on stability. This indicates structures of TCR/peptide/MHC complexes. Thus it has
that “scanning” may be a process (as shown in Figure 11.11) been suggested that each TCR variable-region gene prod-
that first involves contact with (and orientation by) the uct engages each type of MHC through a “menu” of struc-
α-helices of the MHC and then a “fitting” process with and turally coded recognition motifs that have arisen through
stabilization by peptide residues that involves a substantial coevolution.235,236
loss of entropy. This model of TCR binding might help to
explain the striking efficiency and sensitivity of T-cell rec-
ognition with the MHC helices guiding the TCR into the Role of Cluster of Differentiation 4 and
correct orientation. It might also be the structural basis for Cluster of Differentiation 8
cross-reactivity in which structurally very different peptides What is the role of CD4 and CD8 with respect to the T-cell
bind to the same TCR, as the CDR3 regions of TCR could response to agonist and antagonist peptides? Clearly, their
“fold” into the peptide in many possible configurations. expression greatly augments activation, and in some cases,
While attractive, there remain caveats about this “two-step determines whether there is a response at all.237 In addition,
binding” model; one is the existence of some important the results of Irvine et al.238 and Purbhoo et al.,239 using a
human class I MHC antigens that “bulge” out of the bind- single-peptide labeling technique, found an appreciable
ing groove, creating a barrier to “scanning” as a mechanism T-cell response to even one agonist peptide in all four
of binding to those antigens.234 A second issue is that one T cells analyzed, resulting in a “stop” signal for the T cell
would expect a reproducible “footprint” for MHC binding and a small, but detectable rise in intracellular calcium.
of particular TCR V regions, which until recently has been In CD4 + T cells, these effects are attenuated by antibody
elusive. But Garcia and colleagues,235 as well as Kappler and blockade of CD4, such that many more (25 to 30) peptides
FIG. 11.11. As Shown by Wu etal.,212

Mutational Analysis of T-Cell Recep-
tor (TCR)-Peptide/Major Histocom-
patibility Complex (MHC) Binding
Indicates that the TCR First Contacts
MHC Residues (in the Transition
State), and the Peptide has Very
Little Influence. Subsequently, how-
ever, the peptide residues contribute
greatly to the stability of the complex.
Thus, we have proposed that the tran-
sition state largely involves TCR-MHC
contact followed by stabilization of
mobile complementarity-determining
region 3 residues into a stable state,
usually involving significant conforma-
tional change and loss of entropy.

are required in order to elicit a stop signal and a calcium Superantigens

flux.238 Much of this effect likely comes from the recruit- One of the most interesting and unexpected areas to emerge
ment of lck to the TCR/CD3 complexes via the cytoplasmic from the study of αβ T-cell reactivities is the discovery of
tails of either molecule. But in addition, there is also a sig- “superantigens.” Whereas a particular antigenic peptide
nificant positive effect even with CD4 molecules that are might only be recognized by 1 in 100,000 or fewer T cells in a
unable to bind lck and thus there appears to be an effect naive organism, a given superantigen might stimulate 1% to
on TCR-ligand interaction as well. Nonetheless, while a 20% of the T cells.247,248 As will be discussed in more detailed
weak binding of CD4 to class II MHC has been observed,187 in the following, the physical basis for this is that the supe-
most recently in a complete crystal structure of CD4-TCR- rantigen binds to a Vβ domain of the TCR on T cells while
peptide/MHC,240 there was no apparent stabilizing effect on simultaneously binding to an MHC class II molecule on an
TCR binding to peptide-MHC in measurements of bind- antigen-presenting cell (although not in the peptide-binding
ing within a synapse by Huppa et al.188 This is in contrast groove). This allows a single superantigen, such as staphylo-
to the earlier data of Luescher and colleagues,241 who found coccal enterotoxin A (SEA) in Table 11.5, to stimulate vir-
that CD8 had a measureable stabilizing effect on TCR bind- tually every murine T cell bearing Vβ 1, 3, 10, 11, 12, or 17
ing to ligand on T cells expressing that molecule. This later (about 15% of all αβ T cells), in most cases regardless of what
result is consistent with measurements of CD8 affinity for Vα it is paired with or what CDR3 sequence is expressed.
class I MHCs that are very similar, or even superior, to that Clearly, this is a unique class of T-cell stimulatory molecules.
of many TCRs for peptide/MHC.242 So, are CD4 and CD8 The first indication of a superantigen effect was the dis-
playing very different roles with respect to TCR binding to covery of minor lymphocyte stimulating determinants by
peptide/MHC? Chakraborty and colleagues suggest not in Festenstein in the early 1970s.249 Many years later, Kappler
a model they have developed that indicates that the differ- and colleagues characterized a mouse strain–specific
ences in CD4 and CD8 affinity for MHCs contribute at best deletion of T cells expressing a specific TCR Vβs that were
marginally to activation and that delivering lck to produc- attributable to these loci.250 It emerged that these effects
tive TCR-peptide/MHC complexes is their most important were due to endogenous retroviruses of the mouse mam-
property.243 In this view, the specificity of these molecules mary tumor virus family.251–255 Different family members
for their respective MHC molecules is more of a targeting bind different TCR Vβ domains (as shown in Table 11.5)
mechanism than the stabilization function implicit in the and stimulate T cells expressing them. Meanwhile, Janeway
ubiquitous illustrations originally proposed by Janeway116 and colleagues256 had shown earlier that staphylococcus
and featured in many subsequent models. enterotoxins could polyclonally active naive T cells in a Vβ -
How could CD4 be facilitating the recognition of small specific manner without a requirement for antigen process-
numbers of peptides? Irvine et al.238 proposed a “psuedodimer” ing. Many of these enterotoxins have been characterized
model, which suggests that a CD4 molecule associated with a extensively.247,248,257,258 Unlike the MMTV proteins, which are
TCR binding to an agonist peptide-MHC could bind laterally type II membrane proteins, the enterotoxins are secreted.
to an endogenous peptide-MHC complex that is also being Subsequently, proteins having similar properties have been
bound by an adjacent TCR. This takes advantage of the ap- isolated from other bacteria (Yersinia pseudotuberculosis,259,260
parent abundance of endogenous peptide/MHCs that can streptococcus,261 and from mycoplasma262,263). There is also
be bound by a given TCR 244 and uses two weak interactions evidence of superantigen-like activities in other mammalian
(CD4 → MHC II and TCR → endogenous peptide-MHC) viruses such as rabies,264 cytomegalovirus,265 herpes virus,266
to help create a dimeric “trigger” for activation. This is sup- Epstein-Barr virus,267 and also in Toxoplasma gondii.268 As so
ported by the work of Krogsgaard et al.,213 who also showed many pathogenic or parasitic organisms possess these mol-
that disabling CD4 binding to the endogenous peptide/MHC ecules, apparently by convergent evolution, there must be
in a heterodimer had a much a greater effect on activation some selective advantage, but in most cases there is no con-
than the same mutation on the agonist peptide/MHC. clusive evidence as to what this might be. The one exception
What about the timing of CD4 or CD8 binding with is the case of the MMTV superantigens, where it has been
respect to TCR-peptide/MHC? The first hints of this came shown that polyclonal T-cell stimulation allows the virus to
from the work of Hampl et al.,245 who found that antago- more efficiently infect the B-lymphocytes that are activated
nist peptide/MHCs showed no apparent influence of CD4 by the T cells.269,270 This may be a special case, however, and
versus agonists, and because these interactions are much most authors have suggested that superantigens primarily
briefer, suggested that CD4 binding was subsequent to TCR serve to confuse and occupy the immune system while the
engagement with a ligand. That this is likely to be a general pathogen escapes specific targeting and elimination. Large
effect was shown recently by Jiang et al.,246 who found that doses of superantigens have also been implicated in various
whereas very brief T-cell contacts with peptide/MHC loaded “shock” syndromes, such as food poisoning or “toxic
red blood cells involved TCR but not CD8, longer contact shock,”247 but this is probably not their everyday purpose,
periods involved both molecules. Thus it seems that the as it would violate the general rule that the host and parasite
TCR must first bind to a high quality (eg, agonist) ligand should coexist.
and then CD4 or CD8 comes in to deliver lck and dissoci- It has also been suggested by Stauffer et al.271 that supe-
ates (based on the lack of stabilization, at least in the case of rantigens might be involved in triggering autoimmune dis-
CD4). But more direct measurements are needed, as well as eases. Here the hypothesis is that a large number of some
further definition of the role of endogenous ligands. Vβ-bearing T cells are activated by a pathogenic superanti-

TABLE 11.5 Vb Specificity of Exogenous and Endogenous Superantigens

Bacterial Superantigen Human Vb Specificity Reference Murine Vb Specificity Reference
(as referenced in 235,314)
310,311
SEA ND 1,3,10,11,12,17
315,316 310,311,312
SEB 3,12,14,15,17,20 (3), 7,8.1,8.3, (11), (17)
316 310
SEC1 12 7,8.2,8.3,11
315,316 310
SEC2 12,13,14,15,17,20 8.2,10
316 310
SEC3 5,12 (3), 7,8.2
316 310
SED 5,12 3,7,(8.2),8.3,11,17
315,316 310
SEE 5.1,6.1–6.3,8,18 11,15,17
316 310
TSST-1 2 15,16
315 310
ExFT 2 10,11,15
239
Strep M 2,4,8 ND
Endogenous Proviruses Vb Specificity Mls Typea Chromosome Reference
(as referenced in 235,236)
Mtv-1 3 c, 4a 7
Mtv-2 14 NA 18
Mtv-3 3,17 c 11
Mtv-6 3,17 c, 3a 16
Mtv-7 6,7,8.1,9 a, 1a 1
Mtv-8 11,12 f, Dvbll.1 6
Mtv-9 5,11,12 f, Etc-1 12
Mtv-11 11,12 f, Dvbll.3 14
Mtv-13 3 c, 2a 4
Mtv-43 6,7,8.9,9 Mls-like ND
Exogenous Viruses Vb Specificity Mls Type Chromosome Reference
231
MMTV-C3H 14,15 NA
247
MMTV-SW 6,7,8.1,9 Mls-like
242
Rabies ND
245
EBV ND HERV-K18
243
CMV ND
244
Herpesvirus
Other Pathogens Vb Specificity Name Chromosome Reference
240,241
Mycoplasma arthritidis h17,6,8.1,8.3 MAM
246
Toxoplasma gondii 5
237
Yersinia enterocolitica ND
238
Yersinia pseudotuberculosis ND
Vb in parentheses are reactive with commercial but not recombinant enterotoxins.
CMV, cytomegalovirus; EBV, Epstein-Barr virus; MMTV, mouse mammary tumor virus; NA, not applicable; ND, not determined.
a
The nomenclature in use before the discovery that the phenotype resulted from endogenous retroviruses.
gen and that subsequently self-reactive T cells within those from separate structures) would displace the TCR somewhat
activated cells are more easily stimulated by a particular tis- (but not entirely) away from the MHC binding groove,275
sue antigen. That this may occur in some cases is supported thus making the interaction largely insensitive to the TCR/
by their work on a human endogenous retrovirus which spe- peptide specificity. Other TCR-SAg-MHC complexes have
cifically stimulates Vβ7 T cells and is implicated in the initia- very different geometries.276,277
tion of type I diabetes.271 Another report implicates a supe- Why do all the many independently derived superanti-
rantigen in Crohn disease, another autoimmune disorder.272 gens interact only with the TCR β -chain? One possibility
While the biochemistry of superantigen binding to TCR is that the β -chain offers the only accessible “face” of the
and MHC is similar to that of TCR peptide/MHC interac- TCR, perhaps because the CD4 molecules hinders access to
tions,273 mutagenesis, and particularly x-ray structural data, the Vα side, as suggested by the antibody blocking studies of
has shown that the topology is both quite different and vari- Janeway and colleagues.278,279
able.248 In particular, it has been found that Mls-la presenta-
tion to T cells is most affected by mutations on the “outside”
surface of the Vβ domain, which do not affect peptide/MHC ANTIGEN RECOGNITION BY g d T CELLS
recognition.162 In contrast, CDR1 and CDR2 of regions of γδ T cells together with αβ T cells are present together in all
Vβs are involved in bacterial superantigen reactivity.274 but the most primitive vertebrates. This fi nding argues that
An example of the structural data is shown in Figure 11.12, γδ T cells functions are different from those of αβ T cells.
which shows how a model TCR-SAg-MHC complex (derived Yet, most γδ T cells produce cytokines that are similar to

by MHC and related molecules, it was assumed that γδ T-cell

recognition would also follow the same rules. Even in cases
where classical MHCs are clearly not involved, it has been
suggested that nonclassical MHC molecules or some as
yet to be identified surface molecules might play a similar
role. These possibilities are difficult to test experimentally.
Thus, analysis of γδ T cells specific for MHC molecules were
carried out to determine the antigen recognition require-
ment of γδ TCRs. This approach asks the following ques-
tions: When γδ T cells recognize MHC molecules, what kind
of antigen processing is required, is any of the specificity
conferred by bound peptide, and which part of the MHC
molecule is recognized? This approach took advantage of
the detailed knowledge of the molecular structure of MHC
class I280,281 and MHC class II molecules,282 thus making
T-cell epitope mapping feasible and interpretable. In ad-
dition, and more importantly, the biosynthetic pathways
and antigen-processing requirements for both MHC class I
and MHC class II molecules had been extensively studied.
Mutant cell lines defective in either pathway were readily
available and could be transfected with various MHC class
I and class II genes. Therefore, potential antigen processing
and presentation requirements for alloreactive γδ T cells
could be studied with precision and compared with those
of αβ T cells.
The γ δ T cell LBK5283,284 recognizes I-Eb,k,s but not I-Ed.
An analysis of the fine specificities of LBK5 showed that the
peptide bound to the I-E molecules does not confer speci-
ficity and that no known antigen-processing pathways were
required in the recognition of I-E by LBK5. All variations
in the ability of different stimulator cells to activate LBK5
can be attributed solely to their level of surface I-E expres-
FIG. 11.12. Crystal Structure of a T-Cell Receptor (TCR) b/Super- sion and are independent of their species origin (mouse,
antigen (SAg) Complex. Fields et al.253 crystallized TCR-SAg com-
hamster, human) and cell type (B cells, T cells, fibroblasts).
plexes and from the structure of the same superantigens with a
class II major histocompatibility complex (MHC) molecule and were Modifications of the repertoire of peptides loaded onto MHC
able to deduce the relative spatial arrangement of the three mole- molecules also showed no effect because native I-Ek, GPI-
cules. This model suggests that TCR does not contact the MHC very linked I-Ek, I-Ek expressed with or without invariant chains,
strongly, which is consistent with the relative peptide insensitivity and the presence or absence of functional MHC class I or
of SAg activation. class II antigen-processing pathways all stimulated LBK5
similarly.285 Thus, LBK5 recognizes native I-Ek molecules
those produced by αβ T cells. γδ T cells also can mount cyto- with a variety of different bound peptides or in the case of
toxic responses upon activation like CD8 + αβ T cells. These GPI-linked I-Ek, which most likely does not complex with a
results, together with the fact that αβ T cells are almost peptide. In addition, LBK5 recognizes E. coli –produced I-Ek
always found alongside and usually in excess of γδ T cells, a and b chains folded with a single peptide.285
suggest that the difference in how γδ T cells and αβ T cells The functional epitope on I-E for LBK5 recognition maps
contribute to host immune competence is less likely because to the β67 and β70 residues.285 This explains why LBK5 rec-
of differences in effector functions or in tissue distribution ognizes I-Eb,k,s but not I-Ed and why peptide bound to I-Ek
but rather because of differences in how these two types of does not confer the specificity. Surprisingly, I-Ek mutants
cells are triggered. Thus, to understand how γδ T cells func- with an altered carbohydrate structure on the α84 position
tion, it is essential to know the antigens and the target cells are not recognized by LBK5.286 Although the binding inter-
recognized by γδ T cells. face of the LBK5 γδ TCR and I-E has yet to be determined,
it is reasonable to assume that it will not deviate much from
those reported for antibodies, αβ TCRs and the G8 γδ TCR.
Antigen Recognition Requirements From the coordinates of a published crystal structure of
and g d T-Cell Ligands I-Ek,287 the distance between β67–70 (the LBK5 epitope) and
Because the study of γδ T cells is relatively recent and does the carbohydrate attachment site at position α82 was esti-
not stem from any knowledge of their biologic function, mated to be around 28–33Å. Hence, it is likely that the car-
experiments designed to characterize their specificity and bohydrate structure is peripheral to the core of the LBK5/I-E
function had drawn heavily on our knowledge of αβ T cells. interaction. This type of interaction may be similar to that
Because αβ T-cell recognition requires antigen presentation described for human growth hormone and the extracellular

domain of its receptor288 where a central hydrophobic region T10/β2m and T22/β2m molecules can stimulate the G8 γδ
at the contact site, which is dominated by two tryptophan T cell,295,296 which provides unequivocal evidence that these
residues, accounts for more than three-quarters of the bind- peptide-free molecules retain their immunologic function.
ing free energy and where peripheral electrostatic contacts Direct binding between soluble G8 γδ TCRs and T10/ β2m
contribute substantially to the specificity of binding but and T22/ β2m complexes has been measured.297 Surface
not to the net binding energy. This type of protein-protein plasmon resonance showed that the dissociation rates for
interaction has been postulated to ensure the specificity of the interaction between G8 and T10b and T22b were similar
an interaction without requiring a high affinity.289 (kd = 8.1 ± 2.3 × 103s −1) and slower than those that had been
Two independently derived γδ T-cell clones, KN6290 and observed for most interactions between αβ TCRs and pep-
G8,291 are found to recognize T10 and T22, two closely tide/MHC complexes. The association rates (k a = 6.53 ± 1.73
related, nonclassical MHC class I molecules that have 94% × 104 M−1s −1) are among the fastest that had been reported
amino acid identity. T22 appears to be expressed constitu- for αβ TCRs and their ligands. Therefore, compared to αβ
tively on a variety different cell types, whereas the expression TCRs, the affinity between G8 γδ TCRs and their ligand is
of T10 is inducible on cells of the immune system. Among rather high (K D = 0.13 ± 0.05 μM). A Scatchard analysis of
strains of mice tested so far, all express T22. However, mice equilibrium binding generated a similar affinity of 0.11 ±
of the H-2d or H-2k MHC haplotypes (eg, BALB/c and C3H, 0.07 μM for this interaction.297
respectively) lack functional T10 molecules.291,291a T10 and There are only two other pairs of ligand-TCR interaction
T22 have also been identified as natural ligands for murine have been reported. The binding between the soluble form
γδ T cells. Approximately 0.2% to 1% of the γδT cells in nor- of G115 and its ligand F1-ATPase and the apolipoprotein A-I
mal, unimmunized mice are T10/T22-specific (Fig. 11.13).292 complex showed a K D of 1.5 μM and 0.8 μM, respectively.171
The primary sequences of T10 and T22 suggest that the These affinities are within the range of soluble αβ TCRs
necessary structural features that enable classical MHC and their peptide/MHC ligands. Surprisingly, a K D of 110 to
class I molecules to bind peptides are absent. Indeed, x-ray 900 μM was reported for δ1A/B-3:MICA interactions.299
crystallography has shown that T10 and T22 adopt a severely This affinity is one to two orders of magnitude weaker than
modified MHC-like fold that lacks a classical peptide- the NKG2D:MICA interaction, more on a par with MHC
binding groove and exposes part of the β sheet “floor” of the class I:CD8 interactions (from 11 to ≥ 1,000 μM).
α1/α2 platform.293,294 Consistent with the structural data, no It should be noted that the same human γδ T-cell clones
endogenous peptides can be eluted from chimeric T10/Ld mol- that recognize AS/ApoA-I complexes are activated by a set
ecules that are expressed by transfected cells.295,295a This indi- of nonpeptidic pyrophosphomonoesters that are collectively
cates that these molecules can reach the cell surface devoid referred to as phosphoantigens (phosphoAgs).170 It has been
of peptide. Importantly, the E. coli–produced, in vitro–folded observed for well over a decade that some human periph-
eral blood Vγ 9Vδ2 cells can be stimulated by phosphoAgs
such as isopentenyl pyrophosphate and dimethylallyl pyro-
phosphate. These cells show in vitro responses to tumors,
phosphoAgs produced by eukaryotes and prokaryotes, natu-
ral and synthetic alkylamines, and aminobisphosphonates.
However, repeated attempts to show interactions between
phosphoAgs and Vγ 9Vδ2 TCRs have failed. The identifica-
tion of AS/ApoA-I complexes as antigens of some of these T
cells further challenges this notion.
As indicated in Table 11.6,145 other than toward phos-
phoAgs, the reactivities of Vγ 9Vδ2-expressing T cells to other
challenges such as tumors and AS/ApoA-I are far from homo-
geneous. Further, these cells mount a much more robust re-
sponse to phosphoAgs than to CD3 cross-linking. Thus, it is
possible that isopentenyl pyrophosphate and other isoprenoid
intermediates enhance antigen-specific responses of Vγ 9Vδ2
cells without being TCR antigens themselves. Isopentenyl py-
rophosphate and dimethylallyl pyrophosphate are metabolites
of the mevalonate pathway that regulates the biosynthesis of
FIG. 11.13. T10/T22-Specific g d T Cells can be Detected in Normal cholesterol as well as of isoprenoids that mediate the mem-
Mice Through Use of a Tetrameric T22 Staining Reagent. As shown brane association of certain GTPases. The addition of isopren-
by Crowley et al.,274 a T22 reagent that was generated by similar oid intermediates has been shown to augment antigen-specific
methods as tetrameric peptide/major histocompatibility complex αβ T-cell responses and to alter their cytokine profiles.300
reagents stained approximately 0.6% of splenic g d T cells in normal These experimental observations together with the CDR3
animals. More than 90% of these cells are cluster of differentiation
(CD)4–CD8–; the rest are either CD4 or CD8 single positive (about
length distribution analysis described previously make it
3% to 4% each). A similar frequency of tetramer-positive g d T cells clear that the molecular nature of γδ T-cell antigen recog-
was also found in the intestinal intraepithelial lymphocyte popula- nition is fundamentally different than that of αβ T cells.
tion (data not shown). While MHC and MHC-related molecules are recognized by

TABLE 11.6 Partial List of γ δ T-Cell Reactivitiesa

Name (referred to as) Source Reported Reactivities Comments Reference
Murine
DGT3 Lymph node cells from DBA2 mouse Qa-1/(Glu50 Tyr50) 317
primed with poly (Glu50 Tyr50)

KN6 Double-negative thymocytes from T10/T22b,k not d 318
C57Bl/6
319,320
TgI4.4 Lymph node of HSV-infected C3H HSV-gl Can be stimulated by gI protein
mouse and restimulated with alone
L cells transfected with HSV-gI
G8 BALB/c nu/nu immunized with B10. T10/T22b,k not d Direct binding and cocrystal 153,274,323
BR APCs structure have been shown

LBK5 C57Bl/10 nu/nu immunized with B10. I-Eb,k,s not d Can be stimulated by I-E 322,323
BR splenocytes proteins alone; reactivity is

not peptide-specific
LKD1 B10.BR immunized with B10.D2 I-Ad 322
splenocytes
324,325
69BAS-122 C57Bl/10, adult splenocyte HSP-60 peptide, Transferring TCR transfers
cardiolipin, reactivity
β2-glycoprotein 1
325,326
BNT-19.8.12 C57Bl/10, newborn thymocyte Mycobacterium PPD, Transferring TCR transfers
cardiolipin, reactivity
β2-glycoprotein 1
7–17 and other den- γ δ T cells from murine epidermis Keratinocytes Transferring TCR transfers 33
dritic epidermal reactivity

T cells
Human
327,328
Panels of T cell PBMC stimulated with irradiated Tumor cells (eg, 20% of Vγ9Vδ2 clones do not
clones expressing PBMCs and PHA Molt-4), MT, react to MT; only slightly more
Vγ 9Vδ 2 metabolites in than 50% of MT- or Molt-4–
the mevalonate specific clones recognize the
pathway other specificities
Panels of T-cell γ δ T cells from PBM cultured phosphoAg, tumor Close correlation between Daudi 156,329,330
clones expressing with irradiated PBLs and cells (eg, Daudi), and mycobacterial reactivity;
Vγ 9Vδ2 including lymphoblastoid cells AS/ApoA- direct binding between
G42 and G115 I(G115TCR) AS/ApoA-I; heterogenous
reactivity to ApoA-I
DG.SF13 (Vγ 9Vδ 2) γ δ T cells isolated from RA synovial Daudi cells, Transferring TCR transfers 331
fluid stimulated with sonicate of Mycobacterium reactivity

Mycobacterium tuberculosis tuberculosis, MEP
Panels of T-cell PBMC γ δ T cells stimulated with phosphoAg, Reactivities require cell-cell 332,333
clones expressing Mycobacterium tuberculosis alkylamines, contact

Vγ 9Vδ 2 including extract aminobisphos-
CP1.15 and phonates
DG.SF68
334,335
Clones 1,2,3,4, 5 (Vδ 1) Lymphocytes extracted from human MICA, MICB Transferring TCR transfers
intestinal epithelial tumors reactivity; Vδ 1-Jδ 1 with
cultured with irradiated CIR- diverse CDR3 are used;
MICA and CIR-MICB cells not all Vδ 1-Jδ 1 cells are
MIC-specific
335
JR.2 and XV.1 (Vδ 1) PBL stimulated with autologous CD1c expressing Transferring TCR transfers re-
CD1+ DC and M. tuberculosis cells activity; many Vδ1 positive γδ
extract T cells are not CD1c-reactive
Vγ 1.3Vδ 2 expressing γδ TCR chains from muscle- Muscle cell extract 336,337
BW5147 infiltrating T cells of a poly- and Escherichia

myositis patient and transfected coli extract
into TCR-deficient BW5147
338
4–29 and 5–3 PBMC from CMV-infected trans- CMV-infected
(Vδ 2 negative) plant recipients stimulated with fibroblasts, Hela,
irradiated PBMC and PHA HT-29, Caco-2
a
The names of the T cells, the way they are generated, and their reported reactivities.
CMV, cytomegalovirus; HSA, herpes simplex virus; MEP, monoethylphosphate; PBMC, peripheral blood mononuclear cell; PHA, phytohaemagglutinin; PPD, microbacterium purified
protein derivative; RA, rheumatoid arthritis.

γδ T cells, γδ T-cell antigens need not be MHC or MHC- While ligand expression does little to constrain antigen
related molecules. Moreover, in all cases, the antigens are specificities of the γδ T cell repertoire, it does play a role in en-
recognized directly. This suggests that pathogens and dam- dowing γδ T cells with different functional programs, in that
aged tissues can be recognized directly and cellular immune TCR signaling strength experienced by γδ thymocytes seem
responses can be initiated by γδ T cells without a requirement to determines the functional specification of γδ T cells: γδ
for antigen degradation and specialized antigen-presenting thymocytes that have not encountered cognate ligands make
cells, such as B cells, macrophages, and dendritic cells. This IL-17, γδ thymocytes that have encountered thymic ligands
would allow for greater flexibility than is present in classi- make IFN-γ,306a and those that are strongly self-reactive make
cal αβ T-cell responses. T10/T22, I-E, as well as MICA and IL-4. Importantly, regardless of ligand experience, some γδ T
MICB expression can be induced under certain physiologic/ cells in normal mice are able to make cytokines immediately
pathologic conditions.301 ATP synthase F1 is normally local- upon TCR engagement. IL-17 is a cytokine, which regulates
ized in the membranes of mitochondria, but it has also been the expansion and recruitment of neutrophils and monocytes
found on the surfaces of tumors, hepatocytes, and endothe- to initiate the inflammatory response.306b In acute inflamma-
lial cells.302–303 Thus, γδ T-cell activation may be regulated tion, a swift IL-17 response must be elicited without prior an-
through the level of protein expression. In addition, the rec- tigen exposure. Therefore, γδ T cells may be uniquely suited
ognition of I-E by LBK5 is acutely sensitive to changes in to produce IL-17 at the onset of the inflammatory response to
the glycosylation of the I-E molecule. This suggests a novel initiate an acute inflammatory response to pathogens and to
way by which antigen recognition by γδ T cells can be regu- host antigens revealed by injury. In addition, by acting early
lated. Changes in the posttranslational modifications of sur- in the inflammatory response, γδ T cells may modulate and
face glycoproteins often indicate that tissues have become shape the subsequent αβ T cell and B cell responses that de-
infected, have undergone neoplastic transformations, or velop during the inflammatory process and thus may play a
have experienced other types of cellular stress. For example, much larger role in the adaptive immune response than previ-
it has been shown that the infection of mice with Listeria ously recognized.
monocytogenes impairs the addition of sialic acid to host cell
glycoproteins that include MHC molecules.304 Additionally,
Antigen Recognition Determinants of g d T-Cell
whereas the surface glycoprotein mucin is heavily glyco-
sylated in normal cells, it is underglycosylated in breast,
Receptors: V Genes Versus Complementarity-
ovarian, and pancreatic carcinomas such that the peptide Determining Region 3 Regions in g d T-Cell Receptor
backbone is unmasked.305 Thus, both the quantity and the Ligand Recognition
quality of the ligand could contribute significantly to the In part because γδ T cells from different anatomic sites show
specificity of γδ TCR recognition. preferential V gene expression, γδ T cells are commonly been
divided into subsets based on V gene usage.309,310 Numerous
studies have also reported that γδ T-cell functions segre-
Thymic Ligand Recognition and g d T-Cell Repertoire gate with Vγ or VγVδ usage.311–314 These observations have
and Function Development led to suggestions that the bias in V gene usage enables γδ
γδ T cells, like αβ T cells, develop in the thymus before en- T cells to respond to antigens that are specific to their resi-
tering the periphery. In the case of αβ T cells, thymic de- dent tissues.11,312,315 If this were the case, then γδ TCRs would
velopment entails endogenous ligand–driven positive and function like innate immune receptors even though VDJ
negative selection, which determine what αβ T cells can recombination of the TCR δ chain leads to orders of mag-
recognize and whether these T cells will develop into CD4 + nitude higher potential junctional (CDR3 region) diversity
helper or CD8 + cytolytic T cells. However, the role of ligand- than is found in Ig and αβ TCRs (as discussed previously).
mediated selection in γδ T-cell development and function To resolve this issue, it is essential to understand the basis
has been less clear. Recently, this issue has been reexamined of γδ TCR antigen recognition. Based on the analysis of T22-
and the emerging picture is fundamentally different from specific γδ T cells isolated from normal mice, the majority
what we know about αβ T-cell development and differen- of T22-specific γδ TCRs use Vγ1 and Vγ4 in the spleen and
tiation. These work suggest that thymic development does a sizeable number use Vγ 7 in the IEL compartment.316 Thus,
little to constrain γδ T cell antigen specificities, but deter- at least for T22 specificity, Vγ usage reflects tissue origin
mines γδ T cell effector fate.306 In particular, γδ thymocyte and not antigen specificity. Consistent with this, it has been
development and exit into the periphery is not contingent demonstrated that the preferential usage of Vγ 7 by γδ T cells
on encountering cognate antigen in the thymus and TCR that can migrate into the IEL compartment primarily results
dimerization may be sufficient to induce signaling for γδ T from interleukin-15–driven control of Vγ 7 accessibility dur-
cells to develop in the thymus.306a Thus, once the antigen ing thymic VJ rearrangement.317
specificity repertoire is generated by V(D)J rearrangement, While different Vγs and Vδ s were associated with T22-
it is only marginally modified by thymic selection. In fact, a specific TCR sequences, there is one defining feature that is
large fraction of peripheral γδ T cells have not encountered common among them.316 This is a prominent CDR3δ motif
ligand either in the thymus or in the periphery, and this shown in Figure 11.14 that consists of a Vδ or Dδ1-encoded
antigen-naïve population is actively maintained by rapidly Trp (W); a Dδ2-encoded sequence of Ser, Glu, Gly, Tyr, and
turning over, and cells that have encountered self-ligands do Glu (SEGYE); and a P-nucleotide-encoded Leu (L). Gene
not accumulate.306a transfer experiments established that TCRs with the W-(S)

A B C
FIG. 11.14. The T22 Structure and the Interaction with the G8 g d T-Cell Receptor (TCR). A: The structure of a T22 molecule, a nonclassical
class I MHC that is missing half of an alpha helix and does not bind peptides.270 B,C: The intersection with the G8 TCR, where the long TCR
d CDR3 accounts for almost all of the contact residues.153
EGYEL motif bound T22 while those lacking the motif did αβ TCR and the peptide/MHC when in complex (Fig. 11.15).
not316 (see Table 11.6). The majority of the contact residues are contributed by the
Proof that the W- (S)EGYEL motif is the antigen contact β-sheet floor and the α1 helix of T22, and the fully extended
site came from the crystal structure of the G8 γδ TCR bound to CDR3d loop with Trp anchoring at the N-terminal end and
T22.168 As discussed in the previous section, T22 has an MHC residues Gly, Tyr, Glu, and Leu together with a Thr residue
class I–like fold, but one side of what would normally be a encoded in the J region anchoring the loop at its C-terminal
peptide-binding groove is severely truncated and exposes the end. This is similar to Ig, where antigen specificity in non-
β-sheet “floor.”296 G8 binds T22 at a tilted angle that contrasts somatically mutated antibodies resides predominantly in the
with the essentially parallel alignments of the long axes of the CDR3 of the heavy chain.318 This mode of antigen recognition
FIG. 11.15. Conserved T-Cell Receptor (TCR) d Complementarity-Determining Region (CDR)3 Sequences Correlate with T10/T22 Specificity
in g d T Cells. Sequence data from Shin et al.290 showing that both established cell lines (G8 and KNG) and γ δ T cells from clonal cultures that
are specific for the T10/T22 share highly conserved TCRδ CDR3 sequences. These are largely derived from the Dδ 2 gene segment and from
the structural data.Adams et al., color plate 4, 153 These conserved sequences constitute the main interaction between these γ d TCRs and their ligands.

also fits well with analyses of CDR3 length distributions of T cells and could allow for a significant response without an
all immune receptor chains that first suggested that γδ TCRs initial need for clonal expansion as is required for most αβ
bind antigens more similarly to Igs than to αβ TCRs.144 T-cell responses. It was shown that rearrangements at the
There are two G8/T22 complexes in the asymmetric unit.168 TCR δ locus are biased towards full-length Dδ2 sequences
While the contact residues are similar at the interface of the rather than extensive D region nucleotide deletion, as is the
CDR3/T22 β sheet, the two complexes differ by a relative ro- case for the TCR-β locus.313 Thus, different reading frames of
tation between the G8 TCRs. This shift alters the contacts Dδ2 may contribute to the recognition of other ligands by γδ
formed between the CDR1, CDR2, HV4, and CDR3 loops TCRs in a manner similar to that of T22-specific γδ TCRs and
and T22 in each TCR, suggesting that the CDR3 loop acts as a would lead to a repertoire that is biased toward a relatively
pivot point for G8 binding, with some flexibility in the inter- small number of ligands, more on the order of hundreds to
action between the other CDR loops and T22. The hinge-like thousands versus millions as estimated for αβ T cells, but
flexibility around this pivot point stands in stark contrast to with highly variable antigen-binding affinities. A repertoire of
interactions seen in antibody/antigen and αβ TCR-peptide/ this type would allow more flexible and efficient responses to
MHC complexes. In those cases, the relatively straight-on changes in ligand expression. These are testable hypotheses,
docking mode results in multipoint (ie, multi-CDR) attach- especially once more γδ TCR ligands have been identified.
ment of the receptor to the ligand, essentially rigidifying
the intermolecular orientations between the two binding GENERAL FEATURES OF T-CELL RECEPTOR
partners. In most cases, the αβ TCR CDR1 and CDR2 loops AND IMMUNOGLOBULIN DIVERSITY
provide a perimeter of contacts with the MHC helices sur-
rounding the CDR3 loops, and so far, no variation has been A Dominant Role for Diverse
seen in the docking angle of TCR to MHC in cases where Complementarity-Determining Region 3
multiple complexes exist in the asymmetric unit. Thus, the Regions in Antigen Specificity
CDR3 motif of G8 and other γδ TCRs may be thought of as One interesting observation that emerges from a detailed
a somewhat autonomous binding entity that is presented by a analysis of the gene rearrangements that create both TCR
variety of germline-encoded variable domain scaffolds with- and Igs is how the diversity of the CDR3 loop region in one
out strong preference for particular CDR1 and 2 sequences. or both of the chains in a given TCR is so much greater than
It would be interesting to see other examples of γδ TCR/li- that available to the other CDRs. A schematic of this skewing
gand binding. CDR3 regions were found to be important for of diversity is shown in Figure 11.16 for human Igs and for
γδ T-cell recognition of MICA/MICB. In this case, the reac- αβ and γδ TCR heterodimers. In the case of αβ TCRs, this
tivity correlates only with a junction between Vδ1 and Jδ1. concentration of diversity occurs in both Vα and Vβ CDR3
Aside from the W-(S)EGYEL motif, the T22-specific CDR3γδ loops, and numerous TCR-peptide/MHC structures160 have
sequences were diverse and were encoded by various Vδs, N- confirmed that these loops sit largely over the center of the
and P-nucleotides, and Dδ1s of different lengths and reading antigenic peptide (see previous section). While this concen-
frames. Importantly, it was shown that sequence variations in tration of diversity in αβ TCRs in the regions of principal
the CDR3 regions around this motif modulated the affinity contact with the many possible antigenic peptides seems
and the kinetics of T22 binding.313 In fact, the T22-specific reasonable, it is much harder to explain for Ig or γδ TCRs.
repertoire in normal mice covers a range of affinities, as is Clearly, there must be some chemical or structural “logic”
evident by the large range of T22 tetramer staining intensi- behind this phenomenon. A clue as to what this might be
ties.300,316 This allows for the selection of T cells with the “most comes from the elegant work of Shin et al.313 in the demon-
optimal” antigen-binding capabilities during an immune re- stration that for at least one γδ T-cell specificity, the antigen
sponse, which is a hallmark of the adaptive immune response. recognition determinants are encoded by germline V or D
Nonetheless, such a repertoire that is created mainly by region residues, with remaining sequence diversity modulat-
V, D, and J region–derived germline-encoded nucleotides, ing the affinity. While this may be a feature of some or many
despite requiring VDJ recombination, would be “innate” γδ TCRs, it could not explain the much broader repertoire of
in character, because the antigen specificities would be pre- Igs. Instead, one possible explanation comes from the stud-
determined and the repertoire would be much less variable ies of Wells and colleagues,288 who systematically mutated
among individuals of the same species. all of the amino acids (to alanine) at the interface of human
Analysis of the formation of T10/T22-reactive repertoire growth hormone and its receptor as determined by x-ray
indicates that biases linked to the recombination machinery crystallography. Interestingly, only a quarter of the 30 or so
influence the generation of a γδT-cell repertoire toward certain mutations on either side had any effect on the binding affin-
specificities. A repertoire that is generated by recombination ity, even in cases where the x-ray structural analysis showed
but conferred by a limited set of germline or germline-like that the amino acid side chains of most of the residues were
residues at the CDR3 region will be created at a much higher “buried” in the other. This study illustrates an important ca-
frequency than one whose specificity is conferred primarily veat to the interpretation of protein crystal structures, which
by N-nucleotide additions, as is the case with that of αβ TCRs. is that while they are invaluable for identifying which amino
Indeed, the 1 in 100 frequency of T22-specific γδ T cells in acids could be important in a given interaction, they do not
normal mice is much higher than the estimated 1 in 105 to indicate which ones are the most important. This is presum-
106 frequency of naïve peptide/MHC-specific αβ T cells.315–317 ably because the “fit” at that many positions is not “exact”
This could provide a solution to the apparent paucity of γδ enough to add significant binding energy to the interaction.

FIG. 11.16. Diversity “Map” of Immunoglobulins and T-Cell Receptors, Calculated Potential for Sequence Diversity in Human
Antigen Receptor Molecules.Adams et al., 153 The N region addition is assumed to contribute zero to six nucleotides to the junction of
each gene segment, except for immunoglobulin K chains, in which this form of diversity is seldom used.
In this context, we have proposed a new model319,320 in which contacts. For γδ TCRs, it is not yet clear whether the T10/
the principle antigen specificity of an Ig or TCR is derived T22 specificity313 is an isolated case or the general rule. From
from its most diverse CDR3 loops. In the case of antibodies, the hypothesis discussed here, if there are γδ TCRs that use
we imagine that most of the specific contacts (and hence the the very large inherent diversity in the Vδ CDR3 directly for
free energy) with antigen are made by the VH CDR3 and that antigen recognition, it may be that the lack of somatic muta-
the other CDRs provide “opportunistic” contacts that make tion forces it to provide more diversity in the initial reper-
generally only minor contributions to the energy of bind- toire (versus Igs).
ing and specificity. Once antigen has been encountered and
clonal selection activates a particular cell, somatic mutation
would then “improve” the binding of the CDR1s and 2s to CONCLUSION
convert the typically low-affinity antibodies to the higher- Because TCR genes were first identified in the early 1980s,
affinity models as observed by Berek and Milstein,135 and also information about their genetics, biochemistry, structure,
by Patten et al.136 As a test of this model, Xu et al.315 ana- and function has accumulated to become almost a field
lyzed mice that have a severely limited Ig V region repertoire, unto itself. Despite this very real progress, many issues still
consisting of one VH and effectively two VLs (Vλ1 and Vλ2). remain unsolved such as: What do γδ T cells normally “see,”
These mice are able to respond to a wide variety of protein and what function do they serve? What do superantigens
and haptenic antigens, even with this very limited comple- actually do during the course of a normal response and how
ment of V regions. In several cases, hybridomas specific for is this of benefit to the pathogen/parasite? What is the struc-
very different antigens (ovalbumin versus 2,4-dinitrophenol tural/chemical basis of TCR specificity? What sort of rear-
[DNP], for example) differ only in the VH CDR3. A limited V rangements or conformational charges occur in the TCR/
region repertoire also seemed no barrier to deriving high-af- CD3 molecular ensemble upon ligand engagement? These
finity antibodies with somatic mutation, as repeated immu- and other questions will require many more years of effort.
nizations produced IgG monoclonals with very high affini-
ties (109 to 10 −10 M). The major immune deficit in these mice
was in their inability to produce antibodies to carbohydrates, ACKNOWLEDGMENTS
which may require a special type of binding site or specific V We are grateful to M. Call, K. Wucherpfennig, M. Krangel,
region. Thus, while these experiments only involved one V H, M. Krogsgaard, and K.C. Garcia for providing some of
the results are highly suggestive about the inherent malle- the figures and tables used here. We also thank C. Tato,
ability of VHVL in general, at least with respect to protein and M. Kuhns, M. Krogsgaard, Qi-jing Li, and K.C. Garcia for
haptenic epitopes. With respect to αβ TCRs, we expect that invaluable help with the manuscript, and R. Cuevas for
most of the energy of the interaction with a typical ligand expert assistance. This work was supported by grants from
will reside in the CDR3-peptide contacts and here again the the National Institutes of Health (to M.M.D. and Y.C.) and
CDR1 and 2 regions will make less energetically important from the Howard Hughes Medical Institute (to M.M.D.).

Mechanisms of T-Lymphocyte
CHAPTER
12 Signaling and Activation
Takashi Saito
INTRODUCTION tial finding that CD3 chains are responsible for transducing
The immune system protects us against attack by pathogens activation signals into T cells came from the observation that
through two separate but interacting systems: innate immu- treatment of T cells with a crosslinking anti-CD3ε antibody
nity and acquired (or adaptive) immunity. Whereas innate (OKT3) induces strong T-cell polyclonal activation7 simi-
immunity exhibits rapid and transient responses by recogni- lar to that induced upon stimulation with mitogens such as
tion of conserved molecular patterns of pathogens, adaptive concanavalin A. It was also shown that there is a physical but
immunity distinguishes any small difference/heterogeneity noncovalent association between TCR dimers and the CD3
of a huge number of pathogens and antigens and mediates complex. It is now known that the association of the TCR
a rather slow but long-lasting response. For this purpose, as dimer and the CD3 complex is required first for cell surface
is discussed elsewhere (see Chapter 6) in this volume, our expression and then to transduce antigen-recognition signals
body prepares a vast variety of receptors on T cells and B cells into T cells.8,9 The CD3 complex is composed of four different
(immunoglobulins [Igs]) to be able to recognize virtually all chains. Three of them, γ, δ, and e, are closely related in their
potential antigens. Because each lymphocyte possesses single protein structure, gene structure, biosynthesis, and chromo-
antigen specificity, antigen-specific responses begin with the some localization. While the γ, δ, and ε are genes are located
recognition by and activation of a single cell and consequently within a neighboring region of the chromosome (chromosome
the cells proliferate, a process called clonal expansion and 11 in human and 9 in mouse), ζ is located on chromosome 1
selection many years ago by Macfarlane Burnet.1 To under- in both species (Fig. 12.1). The CD3 complex is composed of
stand the mechanism of antigen-specific activation of lym- three distinct dimers, γε, δε, and ζζ, which associate with the
phocytes, it is necessary to study how a single cell of relevant TCRαβ dimer. Whereas γε and δε are noncovalently assem-
specificity recognizes the antigen and induces its activation bled, ζ– ζ is a covalently linked dimmer with a disulfide bond.
to mediate various functions and protection against patho- The ζ chain has an isoform termed η, which is derived from
gens. Recognition of antigen only on the cell surface by the alternative splicing of the ζ transcript and is found only in
T-cell receptor (TCR) of a T cell is unique and totally different rodents but not in humans. The cell surface expression of the
from that of Ig on a B cell. Such differences in T-cell anti- TCRαβ dimer requires its assembly with the CD3 complex
gen recognition force the establishment of extensive cell-cell due to a stringent quality control checkpoints in the endoplas-
interactions/communications in the body to achieve protec- mic reticulum (ER) that prevent the release of incompletely
tive immunity and to maintain immune homeostasis. In this assembled TCR-CD3 complexes. Studies of the biosynthesis
chapter, we describe the unique feature of antigen recognition of these components of the this complex showed that CD3γ,
of T cells through their unique antigen receptor complex and δ, and ε as well as TCRα and β chains are synthesized in great
the mechanisms that trigger T-cell activation. excess and readily degraded in the ER in the absence of as-
sembly with the other chains. The assembly with the partner
chain protects the complex from ER degradation. CD3ζ is
T-CELL RECEPTOR COMPLEX synthesized at limiting levels, and even the TCRαβ-CD3γδε
Antigen specific recognition is mediated by the TCR, which is complex cannot be transferred to the cell surface without as-
composed of a heterodimer of either αβ or γδ polypeptides. sembly with CD3ζζ and is degraded in lysosome.10 Therefore,
This chapter will focus on signaling mediated through the αβ the assembly with ζ– ζ defines the fully functional TCR-CD3
TCR. As described in Chapter 11, both TCRα and β chains complex, and the entire TCRαβ-CD3γδεζζ complex is now
contain variable regions, which together are responsible for allowed to transport to the plasma membrane where it is
binding to the complex of antigen peptide-major histocom- expressed as the complete and functional TCR-CD3 complex.
patibility complex (MHC). After a decades-long search for the
biochemical nature of TCR, the TCR genes were finally cloned
and found to be quite similar to those encoding Ig genes.2–6 THE IMMUNORECEPTOR TYROSINE-BASED
TCRα and β chains both possess very short cytoplasmic do- ACTIVATION MOTIF AND INITIAL SIGNALING
mains without any particular protein-binding motif, and thus The cytoplasmic tails of all CD3 chains contain a common
cannot transduce antigen-recognition signals into the cell. It signaling motif, immunoreceptor tyrosine-based activation
is necessary, therefore, for the TCR to assemble with signal- motif (ITAM), which has the consensus sequence YxxL/I
transducing components, the cluster of differentiation (CD)3 x6-8 YxxL; composed of two repeats of YxxL/I motif (Y; ty-
complex, to transduce antigen-recognition signals. The ini- rosine, L/I; leucine/isoleucine) with a spacer of six to eight
306

CHAPTER 12 MECHANISMS OF T-LYMPHOCYTE SIGNALING AND ACTIVATION | 307
β α
S S
S S
ε γ S S
δ ε
S S
S S S S
S S S S
ζ−ζ
-S-S-
-S-S-
FIG. 12.1. Structure of the T-Cell Receptor R +
– D E – – D D – + K
K + – D D –
(TCR)-Cluster of Differentiation (CD)3
Complex. The TCRαβ dimer is associated with
three dimers of CD3 chains: CD3εγ, CD3εδ,
and CD3ζζ. Whereas TCRα and β possess
positively charged amino acids within the
transmembrane region, all CD3 chains contain
negatively charged amino acids, which are
necessary to be assembled as the complex.
CD3 chains possess immunoreceptor tyrosine-
based activation motif (ITAM) within their cy-
toplasmic tails as shown in red boxes (γ, δ, and
ε have one; ζ has three).
amino acids. CD3δ, γ, and ε have a single and CD3ζ has receptors on natural killer (NK) cells (DAP12 and DAP10)
three tandem ITAMs within the intracellular region. The (Fig. 12.2) and further in various pattern-recognizing recep-
ITAM was first identified as a signature sequence present tors of innate cells such as macrophages and dendritic cells
in several cell surface receptors on immune cells.11 ITAM- (DCs).12 The ITAM turns out to be specific motif capable of
containing molecules have since been found to be widely transducing receptor–mediated recognition signals into cel-
distributed among various important immune receptors, not lular activation.13 The function of the ITAM in transducing
only TCR (CD3s) but also in B-cell antigen receptors (Igα T-cell activation signals has been shown first by analyzing the
and Igβ), various types of Fc receptors (FcRγ), adaptor pro- capacity of a chimeric protein composed of the extracellu-
teins associated with MHC-recognizing receptors, or paired lar domain of CD8 and the intracellular domain of CD3ζ to
Amino acid sequence
TCR CD3γ E Q L Y Q P LK D – R E Y D Q Y S H L
CD3δ E Q L Y Q P LR D – R E D T Q Y S R L
CD3ε NP D Y E P I R K – G Q RD L Y S G L
CD3ζa NQ L Y N E LN L – GR R E E Y D V L
CD3ζb E G V Y N A LQ KD K M A E A Y S E I
CD3ζc DG L Y Q GL S T – A T K D T Y D A L
BCR Igα E N L Y E GL N L – D D C S M Y E D I
FIG. 12.2. Structure and Signaling Function of
Immunoreceptor Tyrosine-based Activation Motif Igβ DH T Y E GL N I – D Q T A T Y E D I
(ITAM). Cluster of differentiation 3s (T cells), Ig-α/β
(B cells), FcR (myeloid cells; macrophages, dendritic FcR FcRγ DA V Y T GL N T – R S Q E T Y E T L
cells, neutrophils, etc.), NK cells, as well as some
viruses possess ITAM within the cytoplasmic do- FcεRIβ DR L Y E E L N H – V Y S P I Y S E L
main. There are increasing numbers of ITAM-bearing
or ITAM+ adaptor–associated receptors in innate NKR DAP12 E S P Y Q E L Q G– Q R P E V Y S D L
systems. The consensus sequence of the ITAM is
composed of YxxL/I- (seven or eight spacers)- YxxL/I Virus BLV gp30 DS D Y Q A L L P – S A P E I Y S H L
where x is any amino acid. The tyrosine residues
within ITAM are phosphorylated by src family kinase EBV LMP2A HS D Y Q P L G T – Q D Q S L Y L G L
Lck upon TCR engagement, followed by the binding of
the tandem two SH2 domains of ZAP-70 to both phos- consensus Dx x Y x x L x x – x x x x x Y x x L
phorylated tyrosines in T cells. E I I

APC
CD4 MHC CD4 MHC

TCR
CD3 CD3 CD3
T cell T cell
Lck
P
Lck P
P
P
ZAP70
SH2 SH2 kinase

ZAP70
FIG. 12.3. Initial Activation of T Cells by Sequential Two Tyrosine Kinases. Left: The T-cell receptor (TCR)-cluster of differen-
tiation (CD)3 complex and the complex of CD4 and the Src kinase Lck are located separately in resting T cells. Middle: Upon
antigen peptide-major histocompatibility complex (MHC) recognition by TCR, the TCR complex and the CD4-Lck complex get
near through the CD4-MHC binding, which induces Lck close to the CD3 chains. The immunoreceptor tyrosine-based activation
motif (ITAMs) of CD3 chains are tyrosine-phosphorylated by Lck. Right: The phosphorylated ITAM of the CD3 chains became
the binding sites for the tandem two Src-homology 2 domains of the second tyrosine kinase ZAP-70. ZAP-70 is phosphorylated
by Lck and by itself and activated, and then transduces activation signals by phosphorylating downstream signaling molecules.
induce T-cell activation.14 Antibody crosslinking of the CD8 complex, making it competent for signal transduction. ZAP-
domain showed that the cytoplasmic tail of CD3ζ can induce 70 assembled with the CD3 phospho-ITAMs must then be
almost all the events observed during normal T-cell activa- enzymatically activated by phosphorylation of otherwise in-
tion, including tyrosine phosphorylation of various proteins, hibitory tyrosines in its juxtamembrane region by Lck as well
induction of intracellular calcium mobilization and inositol as by itself. The binding and activation of ZAP-70 may be as-
phosphate metabolism, induction of activation markers, cy- sociated with the structural changes of CD3 chains induced
tokine production, and cell proliferation. Mutational analyses by phosphorylation, a topic that will be described under the
of the ITAM sequences within the CD3ζ cytoplasmic region section Triggering Mechanism. Activated ZAP-70 induces
revealed that both tyrosine and leucine (isoleucine) residues phosphorylation of tyrosine residues of various downstream
are critical for transducing a T-cell activation signal.15–16 target molecules.21 These include adaptor molecules LAT
Biochemical analyses to define the downstream signaling and SLP-76. These phosphorylation events lead to further
molecules recruited upon TCR stimulation resulted in the activation of several adaptor and effector molecules and sub-
discovery of several important proteins that are involved in cellular assembly as multimolecular signaling complex.
tyrosine phosphorylation and activation of several adaptor
molecules including a Src family tyrosine kinase, Lck, and
a Syk family kinase, Zeta-associated protein-70 (ZAP-70).17
REGULATION OF PROXIMAL SIGNALING
The ITAM is phosphorylated by Src kinase Lck upon TCR Regulation of Tyrosine Kinases
engagement by antigen. Once tyrosines are phosphorylated, Src family tyrosine kinases (Lck, Fyn in T cells) are com-
they become susceptible to binding in general by proteins posed of an N-terminus unique region, SH2, SH3, and a ki-
possessing Src-homology 2 (SH2) domains. The phosphory- nase domain at the C-terminus.22 The N-terminus unique
lated ITAMs recruit the second tyrosine kinase ZAP-70, one region contains palmitoylation sites by which Lck is localized
of Syk family kinases. ZAP-70 is contains two tandem SH2 in lipid rafts on the inner leaflet of the plasma membrane.
domains in its N terminus and a tyrosine kinase region in Because Lck specifically associates with the cytoplasmic re-
the C-terminus.18,19 The tandem SH2 domains with exactly gions of CD4 and CD8 coreceptors, Lck accumulates at the
the right spacing can bind the two phosphorylated tyrosines site of TCR engagement because CD4 or CD8 are also pres-
within an ITAM20 (Fig. 12.3). Because TCR and CD3 chains ent within the engaged site through binding to MHC.23,24 As
do not have any intrinsic effector function, unlike tyrosine a result, Lck is recruited to the vicinity of the CD3 chains
kinase-containing receptors such as the epidermal growth where it is able to phosphorylate them.
factor receptor or the insulin receptor, the binding of ZAP- The activity of Lck is regulated allosterically by phos-
70 to the phosphorylated CD3ζ ITAM changes the TCR-CD3 phorylation of a tyrosine in the carboxyl terminus by the

CD4 CD4 CD3

Inactivated Activated ζ−ζ
CD8 CD8
Lck Lck
-S-S-
FIG. 12.4. Regulation of the Src Kinase Lck Activation by C C C C

the Tyrosine Kinase Csk and the Phosphatase Cluster C C C C
CD45
of Differentiation (CD)45. In resting T cells, Lck has a
closed configuration through the intra-association be- SH3 SH3 P
tween the phosphorylated tyrosine at the C terminus
P
and the Src-homology 2 domain. C-terminus src kinase SH2
Y Y P
Csk SH2
(Csk) phosphorylates this regulatory tyrosine residue.
The association between CD4 and Lck is mediated by P
the interaction of the cysteine-containing regions of
both molecules. CD45 dephosphorylates the C-terminus Y
tyrosine, which alters the configuration from the closed
L
Lck Lck
structure as the “inactivated Lck” to an open shape as Y
the “activated Lck” with active kinase function.
C-terminal kinase (Csk).25 This phosphotyrosine inter- LAT is a transmembrane protein with a short extracellular
acts with the SH2 domain of Lck, and this induces a con- region and resides in lipid rafts, membrane microdomains
formational change to maintain Lck in a catalytically enriched in cholesterol and sphingolipids, via two palmi-
inactive status. Csk normally and continuously acts to toylation sites within the juxtamembrane region. Mutation
reduce Lck activity and thereby attenuate TCR signaling. of these palmitoylation sites results in the failure of LAT to
Indeed, in the absence of Csk, peripheral T cells become be expressed on the plasma membrane or to transduce T-cell
autonomously activated. Dephosphorylation of the C- activation, indicating that the association within lipid rafts
terminus tyrosine releases Lck from its inactive status. In regulates trafficking to the membrane. Upon TCR stimu-
addition, in order to fully activate Lck, the tyrosine phos- lation, LAT becomes tyrosine-phosphorylated on multiple
phorylation in the catalytic domain is also required. Because tyrosine residues (five tyrosines among nine conserved
the tyrosine phosphatase CD45 induces the dephosphoryla- residues), and these serve as the docking sites for several
tion of both of these regulatory tyrosines, CD45 functions as SH2-containing molecules. Important effector molecules
both a positive and negative regulator of T-cell activation26 containing SH2 domains that bind to phosphorylated LAT
through the C-terminal and kinase-domain tyrosines, re- include PLCγ, growth factor receptor binding protein 2 fam-
spectively. The balance between Csk and CD45 regulates the ily members, Grb2 and Gads, and a Tec family kinase, in-
status of T-cell activation27 (Fig. 12.4). A considerable frac- ducible T-cell kinase Itk.31 Because Grb2 constitutively binds
tion of Lck in naïve T cells is catalytically active because of to son of sevenless (Sos), which is a guanine exchange factor
this dynamic balance. Such active Lck may be responsible (GEF) that mediates guanine triphosphate (GTP) binding
for constitutive activation signals including phosphoryla- of Ras, this LAT-Grb2-Sos complex induces Ras activation;
tion of CD3ζ and help in initiating T-cell activation.28 A re- however, RasGRP plays a more critical role for Ras activa-
lated src-family kinase Fyn is also expressed in T cells and tion than the Grb2-Sos pathway in T cells. Gads binds to
is weakly associated with the TCR complex. Fyn appears to phosphor-LAT upon T-cell activation, and Gads associates
have some signaling role because T cells deficient in both with SLP-76. Although all tyrosine mutations of LAT lost
Lck and Fyn have a more complete block in T-cell develop- the function for T-cell activation, the LAT136 mutant mouse
ment than Lck-deficient mice. exhibits a lymphoproliferative disorder and induction of
strong Th2-type cytokine production, suggesting the pos-
sibility of a LAT-independent pathway to activate T cells.32,33
Adaptor-Mediated Signaling SLP-76 is a cytosolic adaptor with three domains: an
Among the targets of ZAP-70–mediated phosphorylation, a amino terminal region containing three major tyrosines
transmembrane adaptor protein, linker for the activation of that are phosphorylated upon TCR ligation and become
T cells (LAT),29 and an intracellular adaptor protein, SH2 binding sites for various SH2-containing proteins, a central
domain–containing leukocyte phophoprotein of 76 kDa proline-rich domain that binds to SH3-containing proteins,
(SLP-76),30 are the most important substrates (Fig. 12.5). and a carboxy-terminal SH2 domain by which SLP-76 can
The discovery of these two adaptor proteins revealed the bind to phosphorylated tyrosines of other proteins.34 SLP-76
connection between the ZAP-70 tyrosine kinase and PLCγ constitutively associates with Gads through its SH3 region,
activation. These adaptor proteins create an important sig- and the SLP-76/Gads complex is recruited to the phosphor-
nal assembly to induce downstream activation signals. The ylated LAT upon TCR engagement (Fig. 12.6). Both LAT
critical roles of these two adaptors are evidenced by the and SLP-76/Gads bind to PLCγ1 independently, and for-
observation that T cells deficient in either LAT or SLP-76 mation of this tetramolecular assembly stabilizes the com-
exhibit complete defects in TCR activation signaling. plex. Because LAT also binds to the Itk Tec family kinase,

Adaptor Structure Associated protein FIG. 12.5. Adapter Molecules Functioning

Downstream of T-Cell Receptor Signals. The
structure of several adapters described in the
text are depicted. These adapter molecules are
LAT PLCγ-1, Grb2, Gads composed of several modules critical for sig-
nal transduction such as Src-homology 2 (SH2)
Gads,Nck, Vav1, ADAP, and Src-homology 3 (SH3). Whereas tyrosine
SLP-76 P SH2 residues (black dot ) functions as the dock-
Itk, PLCγ-1, HPK1
ing sites for SH2 upon phosphorylation, the
prolin-rich regions (P) interact with SH3. Some
Gads SH3 SH2 P/Q SH3 SLP-76, LAT, Gab2 of these adaptors with the transmembrane re-
gions are localized on the plasma membrane
as “membrane adaptors” such as LAT and
Grb2 SH3 SH2 SH3 Sos, LAT, Shc, Gab2
PAG/Cbp, whereas others within the cytosol
as “intracellular adaptors.” ADAP, adhesion-
Fyn, SLP-76, VASP and degranulation-promoting adapter protein;
ADAP P EVH1 SH3
Skap55 Gads, Grb2-related adapter downstream of
Shc; LAT, linker for activated T cells; PAG/Cbp,
SAP SH2 SLAM, Fyn phosphoprotein associated with glycosphin-
golipid-enriched domains/csk-binding protein;
SAP, signal lymphocyte activation molecule–
PAG/Cbp P P L Csk, Fyn, EBP50 associated protein; SLP-76, SH2 domain–con-
taining leukocyte-specific phosphoprotein of
76 kDa.
P : Prolin-rich : Tyrosine
TCR-CD3
T cell ZAP70 P p85
LAT
ITK P Grb2
P P SOS
Nck P PLCγ
SL
P
P-
Vav P
76
Gads
P
ADAP
Cytoskeletal Integrin Ca2+ response Ras/MAPK

reorganization activation PI metabolism activation
FIG. 12.6. Formation of Multimolecular Signaling Complex Critical for Initial T-Cell Receptor (TCR)
Activation Signaling Pathways. Sequential activation of two tyrosine kinases Lck and ZAP-70 upon an-
tigen recognition by TCR induces phosphorylation of two critical adapter proteins LAT and SLP76. Gads
constitutively associates with SLP76 and is phosphorylated by TCR activation and recruited to LAT as
the Gads-SLP76 complex. Thereafter, several adapter proteins form a physical complex centralized by
LAT and SLP-76. Further recruitment of various adapters and effector molecules to the complex leads
to the activation of differential signaling pathways. Phospholipase γ1 (PLC-γ1) and Gads link between
LAT and SLP76 and a Tec-kinase family ITK, which binds to SLP76, is responsible to phosphorylate/ac-
tivate PLCγ1. Association of these molecules generates a stable complex of LAT-Gads-SLP76-PLCγ1-
ITK critical for further calcium and phosphatidylinositol (PI) responses (as in Fig. 12.7). SLP76 also binds
Nck, Vav1, and ADAP, which links to cytoskeletal reorganization (as in Fig. 12.11). LAT binds Grb2/Sos,
which induces Ras/MAPK activation (as in Fig. 12.8).

which is responsible for phosphorylating PLCγ1, the stable Calcium–Nuclear Factor of Activated T Cells
complex of Itk-LAT-SLP-76-PLCγ1 induces PLCγ1 activa- Upon TCR stimulation, the levels of intracellular free cal-
tion. Besides PLCγ1 activation, SLP-76 induces other func- cium are regulated in two phases: there is an initial and tran-
tions through its association with several other important sient induction of calcium release from storage in the ER, and
molecules. At its N-terminus, SLP-76 binds to Vav, a GEF, this is followed by the induction of a strong influx of high
and Nck, an adaptor protein, both of which are critical for levels of calcium from outside of the cells by opening a cal-
TCR-mediated cytoskeletal changes, and at its C-terminus cium channel called the calcium release–activated calcium
it binds adhesion- and degranulation-promoting adapter channel (CRAC) in the plasma membrane. IP3 generated by
protein (ADAP),35,36 an adapter protein coordinating TCR PLCγ1 catalysis diffuses into the cytoplasm and binds to IP3
signals with integrin activation (see Fig. 12.12). receptors on the ER membrane. IP3 receptors are calcium
channels and, after IP3 binding, they open and allow the re-
lease of the calcium stored within ER into the cytosol. The
DOWNSTREAM SIGNALING PATHWAYS released low level of calcium induces cluster formation by the
The next step of the T-cell activation pathway after ZAP- calcium-binding transmembrane protein, stromal interac-
70–induced phosphorylation of the adaptor LAT and SLP- tion molecule-1 (STIM1), within the ER.39 STIM1 is an ER-
76 upon TCR ligation is the activation of the key signaling residential protein that contains an N-terminal sterile motif
enzyme PLCγ1. PLCγ1 is recruited to the plasma membrane and paired EF hands that function as a calcium-binding
through the binding of its PH domain to PIP3, which is gen- motif and a coiled-coil motif at the C terminus. A critical
erated within the membrane by the phosphorylation of PIP2 CRAC called Orai1 has been identified from the analysis of a
by PI3-kinase. PLCγ1 is activated upon phosphorylation by patient with severe combined immunodeficiency who had a
Itk,37,38 which is composed of PH, SH2, SH3, and kinase do- major defect in lymphocyte activation.40 Clustered STIM1 on
mains, and is also recruited to the plasma membrane through the ER membrane is colocalized and assembled with Orai1
its PH domain by interacting with PIP3. Phosphorylated on the nearby plasma membrane to open the channel and
and activated PLCγ1 then cleaves the membrane bound introduce extracellular calcium into the cytosol (Fig. 12.7).
PIP2 to generate two critical products: the membrane- Recent analysis has shown that oligomerization of STIM1
bound lipid diacylglycerol (DAG) and the diffusible inositol is sufficient to induce CRAC activation independent of
1,4,5-triphosphate (IP3), both of which function as second ER calcium store depletion, but the mechanism by which
messengers for further inducing downstream signaling. STIM1 induces Orai1 oligomerization remains unknown.
TCR-CD3
Orai1
T cell Stim1/2
PLCγ Ca2+ Ca2+
DAG IP3 Ca2+
Ca2+
IP3R
ER
Stim1/2
Ca2+ Ca2+
FIG. 12.7. The Mechanism of Calcium Response in T-Cell Activation. T-cell receptor engagement in-
duces activation of PLCγ1 and generates inositol trisphosphate (IP3), which binds to the IP3 receptors
(IP3R) present on the endoplasmic reticulum (ER) and stimulates the calcium release from intracellular
calcium pool within the ER. Increase of intracellular calcium level activates a calcium-binding paired EF
hands-bearing protein Stim1/2, which then translocates to the vicinity of the plasma membrane. Stim1
localizes with and induces dimerization of ORAI calcium release–activated calcium modulator 1 (Orai1),
resulting in opening functional calcium release–activated calcium channel. The assembly between
Orai1 and Stim1/2 induces the activation pathway dependent on intracellular calcium concentration.

Surprisingly, STIM1, STIM2 (another family member), and plex with activator protein (AP)-1, which is a heterodimer
Orai1 are dispensable for T-cell development as deficiency of composed of members of the Jun and Fos family, to induce
these molecules does not affect this process. Because calcium IL-2 transcription. NFAT and AP-1 represent two signaling
flux is also induced and is important in thymocytes, it is as- pathways of calcium and Ras/MAPK, respectively. It has been
sumed that there must be other functionally redundant and shown that T-cell activation in the absence of the induction of
critical calcium channels in thymocytes.41 AP-1 results in anergic or unresponsive T cells.44
Upon T-cell engagement, intracellular calcium binds to a
protein, calmodulin, and induces a conformational change
that allows the protein to bind to various target proteins. An Ras–Mitogen-Activated Protein Kinase
increase of cytosolic calcium induces activation of signaling In parallel with IP3, DAG is generated by PLCγ and remains
molecules including calcium calmodulin-dependent kinase localized within the plasma membrane where it recruits vari-
and a phosphatase calcineurin. Activated calcineurin acts on ous signaling molecules that contain a specific DAG-binding
a critical transcription factor, nuclear factor of activated T motif (called the C1 domain), including members of the
cells (NFAT), to dephosphorylate the protein, which induces protein kinase C (PKC) family45 and RasGRP,46 a guanine-
its translocation into the nucleus.42 There are five NFAT fam- nucleotide exchange factor for Ras. DAG binding activates
ily members (NFATc1, c2, c3, c4, and NFAT5), and they are PKC and promotes RasGRP to activate Ras by exchange
expressed in many different tissues.43 NFAT is present in the from the guanosine diphosphate (GDP) to GTP-binding
cytoplasm in the resting state in T cells. Cytosolic serine/ form.47 Ras is active in the GTP-bound form, and this acti-
threonine kinases such as glycogen synthase kinase 3 and ca- vation step is mediated by GEFs and suppressed by GTPase-
sein kinase 2 phosphorylate the nuclear localization signal on activating proteins (GAPs). RasGRP and Sos are two major
NFAT to prevent its translocation into nucleus. Upon antigen GEFs in the TCR activation pathway (Fig. 12.8). In T cells,
stimulation, activated calcineurin dephosphorylates this crit- RasGRP functions dominantly for early activation of Ras.
ical phosphorylation site, which allows NFAT to enter into RasGRP is inducibly associated with the plasma membrane
the nucleus, where it triggers various gene expression pro- through binding of its C1 domain to DAG and is phosphory-
grams including various cytokine genes by cooperating with lated by PKCθ. On the other hand, Sos constitutively associ-
other transcription factors. The most well-characterized gene ates with an adaptor protein Grb2 and the Grb2 SH2 domain
among NFAT targets is the interleukin (IL)-2 gene. NFAT binds to phosphorylated LAT upon T-cell activation; Sos-
binds to the IL-2 promoter by forming a cooperative com- Grb2 binds to LAT-SLP76, which results in its recruitment to
TCR-CD3
KSR
Ras GTP P p85

ZAP70
Ras
T cell
MAPKKK Raf LAT
P Grb2
GTP
MAPKK Mek PLCγ
SOS
MAPK Erk RasGRP
DAG IP3
Ras
P
GTP
PKC
Elk
P
FIG. 12.8. Ras Activation Pathways by T-Cell Receptor (TCR) Stimulation. Activation of Ras in T cells fol-
lowing antigen engagement involves activation of two different guanine triphosphate (GTP) exchanging
factors: one is Ras-GRP, and the other is son of sevenless (Sos). TCR activation induces diacylglycerol
generation by PLCγ1, which recruits Ras-GRP to the membrane where Ras-GRP is activated by PKCθ.
Activated Ras-GRP induces Ras activation by exchanging guanosine diphosphate (GDP) to GTP, which
then binds to Sos, because Sos constitutively assembles with Grb2 and the Grb2-SOS complex is recruited
to LAT upon TCR activation. Activated Ras then induces the activation cascade of mitogen-activated pro-
tein kinase which sequentially contains MAPK kinase kinase (Raf), MAPK kinase (Mek), and MAPK (Erk).

the vicinity of the TCR. The relationship between RasGRP express the PKCθ isoform of protein kinase C. PKCθ has a
and Sos appears to be cooperative rather than competitive as DAG-binding domain, is recruited to the plasma membrane
RasGRP-mediated Ras activation enhances Sos activity as a through binding to DAG, and is activated upon generation of
positive feedback to induce strong Ras activation.48 DAG by PLCγ1. One of the main pathways activated by PKCθ
Ras activation triggers a cascade of kinase activation, is NF-κ B. PKCθ plays a critical role in initiating several cas-
which finishes by activating a serine/threonine kinase known cades of the classical pathway of NF-κ B activation down-
as a mitogen-activated protein kinase (MAPK). The MAPK stream of the TCR. NF-κ B is a family of transcription factors
cascade is composed of three kinases: MAPK kinase kinase composed of homo- and heterodimers of five members of
(MAP3K), MAPK kinase (MAP2K), and the MAPK itself. the Rel family including NF-κ B1(p50), NF-κB2 (p52), RelA
In the case of the TCR signaling pathway, the first MAP3K (p65), RelB, and c-Rel (Rel).50 In resting T cells, NF-κ B is
is Raf, which is a serine/threonine kinase that phosphory- found in the cytosol associated with one of a family of inhibi-
lates the next MAP2K, which in T cells is MEK1. MEK1 is tors of NF-κ B (Iκ B) that prevents NF-κ B from translocating
called a dual-specificity kinase because it can phosphorylate into the nucleus. Upon T-cell activation, Iκ B is phosphory-
both a tyrosine and a threonine of the last member of the lated by the Iκ B kinase (IKK) complex, ubiquitinylated and
cascade, the MAPK extracellular signal-regulated kinase 1 degraded, which allows NF-κ B to translocate into nucleus,
(Erk1) and Erk2 in T cells and B cells, respectively. Erk1/2 where it activates various genes involved in survival, homeo-
induces the activation of Elk1, which in turn activates the stasis, and activation of T cells and inflammatory responses.
AP-1 transcription factor complex composed of Fos and Jun. Whereas this signaling pathway to activate NF-κ B is
common among many cell types and has been known for
quite some time, the specific pathway by which PKCθ in-
PKC-CARMA1/Bcl-10/Malt1-Nuclear Factor of KappaB duces NF-κ B activation has only been elucidated in the past
Historically, it has long been known that phorbol ester (phor- decade. The critical molecule responsible for mediating the
bol 12-myristate 13-acetate) plus a calcium ionophore, a activation signal that connects PKCθ and NF-κ B is now
molecule that allows calcium ions to cross cell membranes, known as the CBM complex (Fig. 12.9). It consists of three
induce strong activation signals similar to TCR-induced sig- proteins; a scaffold protein, CARMA1 (caspase recruitment
nals.49 Phorbol ester is known to activate PKC; thus, PKC is a domain [CARD] and membrane-associated guanylate ki-
critical component in T-cell activation. T cells predominantly nase–containing scaffold protein), a CARD-containing
TCR-CD3
PKCθ
CARMA1
T cell
CBM
Bcl10
complex
MALT1
TRAF6
TAK1 IKK I B
IκB
NFκB degradation
NFκB
FIG. 12.9. Signaling Pathway for T-Cell Receptor (TCR)-Mediated NF-jB Activation. Upon TCR stimula-
tion, PKCθ is activated and induces phosphorylation and aggregation of caspase-recruitment domain-
containing membrane-associated guanylate kinase protein-1 (CARMA1). Aggregated CARMA1 forms
a complex (as the CBM [CARMA1-Bcl-10-Malt1] complex) with two other proteins: B cell lymphoma
10 (Bcl-10) and mucosal-associated lymphoid-tissue lymphoma-translocation gene 1 (Malt1). Malt1
induces the degradation of the regulatory component of IκB kinase (IKK) complex (IKKγ) by ubiquitina-
tion via activation of tumor necrosis factor receptor–associated factor-6. IKKγ degradation induces
phosphorylation and subsequently degradation of the inhibitor of NF-κB (IκB), which results in activa-
tion and nuclear translocation of NF-κB.

adaptor protein, B-cell lymphoma 10 (Bcl10), and mucosa- CBM complex–mediated NF-κ B activation for appropriate
associated lymphoid tissue lymphoma translocation gene immune responses without inducing excess inflammation
1 (Malt1).51–53 Upon TCR stimulation, this trimolecular and diseases.
complex is induced by PKCθ-mediated phosphorylation of
CARMA1, which is required for CARMA1 oligomerization
and association with Bcl10. Malt1 binds to Bcl10 and medi- THE IMMUNOLOGIC SYNAPSE
ates the degradation of the IKKγ subunit (also called NEMO) T-cell activation is induced upon recognition of antigen pep-
by inducing polyubiquitination through activation of an E3 tide/MHC on antigen-presenting cells (APCs) such as DCs
ubiquitin-ligase, tumor necrosis factor receptor–associated or B cells by the TCR on T cells upon the antigen-specific
factor 6 (TRAF6).55,56 TAK1 recruited by TRAF6 phosphory- interaction between these two cell types. The interaction
lates IKK, which in turn induces Iκ B phosphorylation. This induces a specific structure with molecular segregation at
event, together with degradation of NEMO, induces conse- the interface between the two cells, which is called the im-
quently Iκ B degradation and release and translocation of munologic synapse (IS) or supramolecular activation cluster
NF-κ B to the nucleus, and thereby induces gene activation. (SMAC).58–59 In the IS, several surface receptors and intracel-
Overexpression of the CBM complex induces spontane- lular signaling molecules are accumulated and segregated to
ous NF-κ B activation without T-cell (or B-cell) activation. form the concentric “bulls-eye”–like structure composed of
Recently, it has been shown that a high proportion of diffuse three discrete regions (Fig. 12.10); there is enriched accumu-
large B-cell lymphomas are induced by specific oncogenic lation of molecules such as TCR, CD28, CD2, and PKCθ in
mutations of CARMA1, which induce its spontaneous oligo- the central region (cSMAC), adhesion molecules such as inte-
merization of CARMA1, formation of the CBM complex, grin LFA1 in the peripheral region (pSMAC), and large mole-
and subsequent continuous chronic activation of NF-κ B, cules such as CD43 and CD44 in the distal region (d-SMAC).
ultimately resulting in tumor induction.57 This illustrates The IS, particularly the cSMAC, was initially thought to be
the importance of maintaining appropriate control of the the structure responsible for transducing signals that lead to
T cell APC
ICAM-1
ICAM-1 (CD54)
(CD54) CD4
CD48 CD80 CD45
CD43 pMHC CD86
APC
TCR
CD2 CD3 CD28

LFA-1
LFA-1
(CD11a
(CD11a
CD18)
CD18)
Lck
TCR PKCθ
c-SMAC C
PKCθ MA
c-S
p-SMAC LFA1
CD2 C
MA
p-S
CD45 C
d-SMAC MA
CD43 T cell d-S
Immunological synapse
FIG. 12.10. Schematic Structure of the Immunologic Synapse. Antigen recognition and activation of T cells starts by the specific
interaction between T cells and antigen-presenting cells (APCs) (left). In the interface of two cells, special structure with segregated
arrangement of receptors and signaling molecules is induced where T-cell receptros (TCRs) and major histocompatibility complexes
(MHCs) are accumulated in the central region termed as the central supramolecular activation cluster (SMAC), the adhesion mol-
ecule leukocyte function-associated antigen (LFA)-1–intercellular adhesion molecule 1 is localized at the surrounding area around
the TCR-MHC (as the peripheral SMAC), and further large molecules such as cluster of differentiation (CD)45 are located outside
peripheral region SMAC as distal SMAC (right). This “bulls eye–like” structure is dynamically induced after T cell-APC contact; LFA-1
moves outside and the TCR-CD3 moves into the center within 10 minutes. Activation signals are induced not at the synapse but at TCR
microclusters, which are generated at the interface between T cell and APCs prior to central SMAC formation, which is composed
of TCR, kinases, and adaptors.

T-cell activation. However, the formation of IS takes as long triggering, and the proposed models can be divided into two
as 10 minutes after cellular conjugate formation, whereas the main conceptual categories. One is through the receptor
initial activation events such as intracellular calcium mobili- clustering by ligand binding, and the other is through the
zation and protein phosphorylation occur within 1 minute, induction of conformational changes within the TCR-CD3
indicating that the activation signal is initiated much ear- complex. None of these models or their variations (see the
lier than the IS is formed.60 Recent studies revealed that the following discussion) can satisfactorily account for the di-
real signal transducing structure, called the TCR microclu- verse experimental observations regarding TCR triggering.
ster, is generated immediately after the T cell–APC contact 1. Kinetic proofreading model. This is an early quantita-
and prior to IS formation.61–63 TCR microclusters contain tive model that attempted to explain a T-cell response
the TCR, kinases such as Lck and ZAP70, adaptor proteins by the half-life of the TCR-MHC/peptide interaction.68
such as LAT and SLP76, and several effector molecules such The length of time between the initial ligand binding
as PLCγ and PI3K. TCR microclusters function as the sig- and receptor signaling was proposed to induce qualita-
nalsome to induce T-cell activation because all TCR micro- tively differential signaling. Longer and shorter half-life
clusters induce tyrosine phosphorylation of various proteins would induce a stronger and weaker response by either
including CD3ζ, ZAP-70, and PLCγ, and their generation oc- strong or weak agonist, respectively.69
curs in parallel to the initiation of the intracellular calcium 2. Serial triggering model. T-cell activation requires a sus-
flux. Therefore, the TCR microcluster is thought to be the tained signal that lasts for several hours. However, TCR
minimal functional unit for TCR signaling. TCR microclu- affinity for MHC/peptide is very low, and activation of
sters are generated initially in the periphery of the interface TCR induces only a brief spike of intracellular signals.
between T cells and APCs, and are then translocated into the This model attempts to resolve this paradoxical re-
central region of the interface to make the cSMAC.64 quirement for T-cell activation. The model states that
Contrary to the initial idea that the cSMAC is the site for sustained signaling is accomplished by the concerted
signal transduction, it is now thought that the cSMAC is the action of multiple TCRs that are sequentially engaged
place where the accumulated TCR complex is internalized with and triggered by the MHC/peptide complex.70 A
and degraded. In contrast to the negative function of the single complex could serially engage and trigger up to
cSMAC for TCR degradation, costimulation receptors such approximately 200 TCRs.
as CD28 and cytotoxic T-lymphocyte antigen (CTLA)-4 ac- 3. Kinetic segregation model. Signal initiation takes place
cumulate in the cSMAC, suggesting that the cSMAC also by excluding inhibitory molecules such as the phospha-
serves a signaling function for costimulation.65 In addition, tase CD45 from the tight contact area (cSMAC of the
T-cell activation results in the polarization of the T cells and IS) between T cell and the APC, thereby shifting the en-
movement of the microtubule organizing center toward the zymatic steady state toward an activating status. Such
TCR-MHC/peptide contact sites. The microtubule organiz- segregation is induced by sizes of the ectodomains of
ing center movement appears to be driven by localized ac- the excluded proteins.71 This model is supported by the
cumulation of DAG66 and also involves ADAP, both of which structure and kinetics of IS, as shown by the segregated
recruit the microtubule motor protein dynein. cSMAC for- regions.
mation is closely regulated by cytoskeletal arrangement and 4. Permissive geometry model. Binding between TCR
leads to the polarization of T cells for polarized secretion dimers and MHC/peptide dimers induces rotational
of secretory vesicles including cytokines, chemokines, and scissor-like conformational changes in the CD3 chains
lytic factors such as granzyme B. Various cytokine-contain- that reveal previously hidden intracellular activation
ing vesicles are accumulated at the TCR engaged site and are motifs. This model integrates receptor clustering and
released directionally and locally. During the T cell–APC conformational change models, together with the ex-
interaction, whereas some of cytokines such as interferon γ, istence of preformed oligomeric receptors, providing a
IL-2, and IL-4 are secreted directionally toward the inter- mechanism to explain TCR signal.72
face, some as tumor necrosis factor (TNF) α are randomly 5. Multimerization model. This model proposes that the
secreted.67 minimum unit of T-cell activation is a dimer and is
based on antibody-induced dimerization of B-cell re-
ceptors and MHC structure studies. A single MHC/
TRIGGERING MECHANISM peptide could induce initial and transient calcium sig-
Signaling mechanisms for T-cell activation have been exten- naling, but monomeric MHC/peptide in solution did
sively analyzed and are now well understood, as described not induce stimulation of T cells. Obviously, more
previously. It has been known for a long time that the anti- multioligomeric ligands induce more intense T-cell
gen recognition event by the TCRαβ dimer is transmitted to activation.73
the ITAM phosphorylation of the CD3 chains, which induce 6. Conformational change model. Extensive analyses of
signal transduction for T-cell activation. However, the actual the crystal structures of TCR-MHC/peptide have not
mechanism by which antigen recognition by the TCR initi- revealed any significant structural changes upon MHC/
ates the initial activation event of the signal transduction peptide binding by the ectodomains of TCRαβ dimers.
cascade is still widely debated and controversial. There are However, biochemical analyses demonstrated struc-
several different models to explain the first events of T-cell tural changes of TCR complex, particularly in CD3ε. A

proline-rich region of CD3ε is a hidden determinant in molecular complex, including transmembrane regions and
the resting state but is exposed and recruits Nck upon cytoplasmic tails, will provide a definite answer to this still
stimulation to induce downstream activation signals.74 confusing issue.
7. Safety model. This model is based on recent findings
that the cytoplasmic domains of CD3ε and CD3ζ have CYTOSKELETAL REGULATION
affinity for the acidic lipids present on the plasma
T-cell stimulation also induces a program for alteration of
membrane, which may result in prevention of ITAM
the cytoskeleton: actin polymerization, a microtubule re-
phosphorylation. Conversely, these chains in free
arrangement, which induces polarization and activation of
aqueous solution may be readily phosphorylated by
T cells, and consequently changes the shape and regulates
Lck. These results have led to the following model:
adhesion and movement of T cells. T cell–APC interaction
whereas CD3εζ tails are tightly associated with the
results in morphologic changes: the cell becomes round
lipid-rich inner membrane of plasma membrane in
and accumulates actin fi laments at the stimulatory inter-
resting T cells, they are released from this membrane-
face. These changes reflect the increase of fluidity of plasma
bound configuration and can be phosphorylated upon
membrane and a decrease of cellular motility. Accumulation
TCR ligation.75
of fi lamentous actin at the T cell–APC interface is the results
8. Pseudodimer model. Only a few MHCs carry foreign
of TCR-induced localized activation of multiple actin regu-
cognate peptides for the specific TCRs, whereas a great
latory and polymerizing pathways. T-cell activation induces
majority of MHCs carry endogenous self-peptides.
SLP76 phosphorylation, which recruits actin cytoskeleton
This model proposes that endogenous MHC/peptide
regulatory proteins, Itk, Vav-1, Nck, and Cdc42 as the reg-
amplify signals produced by agonist MHC/peptide by
ulatory proteins. Localized activation of Cdc42 stimulates
promoting TCR aggregation. Agonist MHC/peptide-
the actin-related protein 2/3 (Arp2/3)-regulating protein
TCR recruits a second TCR in a CD4-dependent man-
WASp, Wiskott-Aldrich syndrome protein, which interacts
ner, which binds endogenous MHC/peptide, stably
with Arp2/3 and activates actin polymerization.80 WASp is
forming a pseudodimer that triggers T-cell activation.76
recruited to the TCR activation complex by association with
The clustering models have been claimed to be incompatible Nck and is activated by assembly with Vav-1, a GEF for the
with the presence of preformed oligomeric receptors on the Rho family of GTP binding proteins and Vav-1, a central reg-
surface of resting cells. However, the recent finding that TCR ulator of cytoskeleton, migration, and adhesion (Fig. 12.11).
microclusters are the functional unit to transduce activation WASp functions as a regulator of the Arp2/3 complex that
signals shows that aggregation of preformed TCR “nanoclu- is critical for actin polymerization. Itk recruits Vav-1 to the
sters” may form microclusters to induce signals.77–79 On the immune synapse. Cdc42 regulated by the Itk-Vav-1 complex
other hand, models based on CD3 conformational changes directly stimulates WASp actively, and the Nck-WASP com-
that induced as a direct effect of ligand binding are not plex regulates actin dynamics. Vav-1-mediated activation of
consistent with the requirement for multivalent ligand to a second Rho family, GTPase Rac1, results in the activation
initiate TCR signaling. Structural analyses of the complete of WAVE2.
TCR-CD3
T cell
P ZAP70
LAT
P
Nck Gads
SLP-76
Vav ITK
FIG. 12.11. Cytoskeletal Regulation by T-Cell Receptor (TCR)

Signaling. TCR stimulation induces actin reorganization,
WIP
which is required for the alteration of cell shapes, cell move-
Dock2
Cdc42 WASP
ment, and receptor clustering. Actin-related proteins such
as Arp2/3, WASp, and WAVE2 regulate accumulation of
F-actin. The LAT-SLP-76 signaling cluster induced upon TCR
Rac1
engagement recruits Nck, Vav1, and Itk. Whereas WASp
WAVE2
binds to the adapter protein Nck and is activated by the Rho-
Arp2/3 family GTPase Cdc42, WAVE2 binds to Vav1 and is activated
by another Rho-family GTPase Rac1 through Dock2.

CELL ADHESION REGULATION—INSIDE-OUT Ras proximity 1 (Rap1), which belongs to the Ras super-
SIGNALING family of GTPases, plays a critical role in the inside-out signal-
Antigen stimulation of T cells induces their strong adhesion ing for integrin activation. Active Rap1 induces increased cell
to APCs. This adhesion is mediated mainly by activation of adhesion through integrin activation without changing their
integrin on T cells such as leukocyte function-associated expression levels on T cells.83 TCR stimulation and chemo-
antigen (LFA)-1, increasing its affinity and avidity for the kines induce Rap1 activation, which then induces clustering
ligand on APCs. Integrin activation is induced by a signal- of high-affinity LFA-1 on the leading edge of the membrane.
ing cascade initiated by the TCR engagement and is a pro- Following TCR ligation, SLP-76 inducibly interacts with
cess termed inside-out signaling (Fig. 12.12). The inside-out ADAP, which then forms a complex with SKAP55. The SLP-
signaling by TCR engagement results in integrin clustering 76-ADAP-SKAP55 pathway leads to regulation of T-cell ad-
and an increase in affinity and avidity for the ligands, which hesion.84 Both ADAP- and SKAP55-deficient T cells showed
then induces high-affinity ligand binding.81 It has been defective LFA-1 clustering and adhesion. ADAP-induced
shown that the integrin LFA-1 goes through three conforma- T-cell adhesion involves Rap1-GTP-interacting adaptor
tional changes to achieve high affinity ligand: resting state, molecule (RIAM), which is required for Rap1 localization
transitional state, and active conformation.82 The integrins at the membrane.85 Because RIAM associates with integrin,
predominantly expressed on T cells are LFA-1 and very late the complex with ADAP-SKAP55-RIAM makes a complex
antigen-4, which bind to their ligands, intercellular adhesion with LFA-1, resulting in adhesion to ICAM-1 upon TCR li-
molecule (ICAM)-1 and 2, and vascular cell adhesion mol- gation. RIAM also binds to a cytoskeletal binding protein,
ecule and fibronectin, respectively. The process of inside-out Talin, and the Talin-RIAM-Rap1 complex may induce Talin
signaling from TCR to integrin activation has been shown to to bind LFA-1 with high affi nity. RIAM is also the ligand of
include downstream signaling and actin-cytoskeletal rear- Ena/VASP and profi ling, which bind to actin cytoskeleton.
rangement. Several critical adaptor molecules, particularly In addition, similar to Talin, other cytoskeletal proteins
Rap1, ADAP, SKAP55, and RIAM, are necessary to translate such as Vinculin, WAVE2, and the Arp2/3 complex are also
TCR engagement to integrin activation. involved in the TCR-induced integrin activation.
Affinity change Clustering LFA-1

LFA-1
TCR-CD3
RapL
Ta
lin
T cell Rap1
LAT
P Vinculin
RI SKAP
P AM
EVL/ ADAP
P VASP
Vav Profilin
Gads
P
SLP-76
P P
ITK ADAP SKAP
RIAM
FIG. 12.12. Inside-out Signaling for T-Cell Receptor (TCR)-Mediated Integrin Activation as the Mechanism of TCR-Induced Cell Adhesion.
TCR engagement induces signaling complex of linker for the activation of T cells and SLP-76. SLP-76 recruits Vav1 and Itk important for actin
reorganization, as in Figure 12.11, and also adhesion- and degranulation-promoting adapter protein (ADAP), which constitutively associates
with Src kinase–associated phosphoprotein of 55 kDa (SKAP55) and Rap1-GTP–interacting adapter molecule (RIAM). The complex of ADAP-
SKAP55-RIAM associates with and activates Rap1 upon TCR activation and then translocates Rap1 to the membrane. The Rap1-RIAM/SKAP/
ADAP complex induces the association of talin with the tail of integrin β. The local accumulation of talin is required for high-affinity binding
by integrin. Integrin changes the status by clustering and then affinity maturation through structural changes. Integrin clustering involves
Rap1 and Rap1-binding molecule RAPL that also associates with RIAM. RIAM associates with F-actin through the association with EVL/VASP
proteins and the actin-binding protein profilin. PKD binds to the β1 integrin tails and recruits and activates Rap1 by Rap1-GEF C3G.

As to the effector mechanism of Rap1 signaling, Rap-1 and mediate cytokine secretion in the absence of costimu-
binds to RAPL, the effector regulator of cell adhesion and lation signals through these costimulation receptors. TCR
polarization enriched in lymphoid tissues.86 The interaction engagement induces various early activation signals such
induces the membrane localization of RAPL and the bind- as an increase in the levels of intracellular calcium, tyro-
ing to the αL subunit of LFA-1, which is critical for LFA-1 sine phosphorylation, and inositol metabolism, but these
clustering and increasing its affinity. RAPL associates with are not sufficient to induce full activation including cyto-
a kinase Mst-1 as an effector molecule. Inside-out signaling kine production, cell proliferation, and effector functions.
induced by TCR engagement results in integrin clustering Instead, in the absence of costimulation, T cells fall into a
and increased affinity and avidity for the ligand, which then state of unresponsiveness or anergy where they do not re-
induces high-affinity binding to the ligand. Further analy- spond to antigen stimulation. Therefore, two signals are re-
sis of the precise signaling cascade for inside-out signaling quired for efficient T-cell activation: “signal 1” is defi ned as
downstream of TCR is required. the antigen-specific TCR signal by the binding of peptide/
MHC complex, and “signal 2” refers to the additional co-
stimulation signal through a costimulation receptor. This
COSTIMULATION
“two signal model” for T-cell activation, which was pro-
Cluster of Differentiation 28–Mediated posed in the early study of T cells,87,88 has been proved by
Costimulation various observations, particularly of the peripheral toler-
T-cell activation is influenced by signals through several ance induction in the absence of “signal 2.”89,90 Although
surface receptors including costimulation receptors (see there are a number of receptors that can induce a costimu-
following discussion), cytokine receptors, and adhesion latory signal, the most physiologic, critical and well charac-
molecules. T cells cannot be fully activated to proliferate terized is CD2891 (Fig. 12.13A).
A B
Signal 1 Signal 1
+
Signal 2 Y YMNM PI3K PDK1 AKT
PLCγ1 PKCθ
B7 P PRRP Grb2
mTOR GSK3β
CD28 Vav CARMA1
Bcl10
Malt1
Cdc42
Grb2 Rac1
P PYAP
FLNA
JNK NFκB NFAT
Unresponsive Full activation Lck
to antigen Clonal expansion
(Anergy) Effector function
AP1/
CD28RE Bcl-xL IL-2
Cell Cell
Cytoskeleton survival proliferation
FIG. 12.13. Costimulation and Cluster of Differentiation (CD)28-Mediated Signals in T Cells. A: Full T-cell activation such as cell prolifera-
tion and cytokine secretion requires two signals: one (signal 1) through the T-cell receptor–CD3 complex as antigen-recognition signal, and
the other (signal 2) through costimulation receptors such as CD28 as costimulatory signals. In the absence of costimulation signal, T cells
become unresponsive to antigen stimulation (called anergy). B: Detailed signal pathway of CD28-mediated costimulation. The cytoplasmic
tail of CD28 has major three motifs: N-terminus tyrosine motif YMNM associates with PI3K leading to the activation of Akt, which induces
cell survival through mammalian target of rapamycin–mediated NFκB activation and cell proliferation, the upper Prolin-rich motif binds to
Grb2 leading to Vav1 activation, which mediates cytoskeletal regulation and NFκB activation, and the last proline region associates with
Lck and Filamin and Lck, which leads to actin-cytoskeletal organization.

CD28 is constitutively expressed on the surface of all T Enhancement of IL-2 production by CD28 engagement
cells as a homodimer. The ligands for CD28 are the B7 fam- is regulated at both transcriptional and posttranscriptional
ily molecules, CD80 (or B7-1) and CD86 (B7-2), which are levels.99 A specific region responsible for CD28-mediated
expressed on APCs. While CD80 is constitutively expressed, transcriptional activation (termed a CD28-response ele-
CD86 is expressed at low level on APCs in the resting state ment, CD28RE) was found in the IL-2 promoter.100 The
but is highly induced upon activation by innate signals, such binding of transcription factors to CD28RE is dependent on
as those derived from pathogens, in parallel with the induc- CD28 signal.
tion of high level expression of MHC class II.92 The induction In addition to enhancement of the transcription of the
of MHC class II and B7 molecules on activated APCs such as IL-2 gene, CD28-mediated costimulation augments IL-2
mature DCs are critical for T-cell stimulation because they production by increasing the stability of IL-2 mRNA.101
can deliver a strong signal 1 and signal 2, respectively. Many cytokine transcripts including IL-2 contain adenyl-
CD28 induces a profound increase in IL-2 production ate-uridylate (AU)-rich elements within the 3′ untrans-
by both transcriptional and posttranscriptional regula- lated region that stabilizes the message. Upon TCR stimu-
tory mechanisms, as well as increased proliferation and lation, an AU-binding protein, TTP, binds to the AU-rich
cell survival, which is partly due to the induction of the elements within the 3′ untranslated region and induces
antiapoptotic gene Bcl-xL (Fig. 12.13B). Whether CD28 degradation of the mRNA. By contrast, CD28 costimula-
induces costimulation totally independently of TCR en- tion induces other proteins, such as NF90, a transcription
gagement has been a controversial question for a long time. factor associated with NFAT, which may compete with
CD28-specific signals responsible for the costimulation TTP for binding at the AU-rich elements and enhance IL-2
signal have been extensively analyzed. CD28 has a short mRNA stability.
cytoplasmic tail (41 amino acids in human, 38 amino CD28 signaling activates I-κ B and consequently NF-κ B.
acids in mouse) and no intrinsic enzymatic function. It CD28 regulates the IKK activation step, and IKK activation
contains several tyrosines including an YxxM motif and results in I-κ B activation/degradation and induces NF-κ B
two PxxP motifs. The YxxM motif, which is conserved activation. Extensive analysis to define the connection be-
in CD28, CTLA-4, and inducible costimulator (ICOS), is tween CD28 signaling and IKK activation resulted in the
a consensus motif for the binding of the p85 subunit of discovery of CARMA1 as caspase recruitment domain–
the lipid kinase phosphatidyl-inositol 3-kinase (PI3K). containing membrane-associated guanylate kinase pro-
Functional relevance of the YxxM motif is suggested by tein-1, followed by identification of its associated proteins,
the fact that PI3K activation is enhanced by CD28 signal- Bcl-10 and Malt-1, which together form the CBM com-
ing.93,94 PI3K activation generates PIP2, which recruits var- plex.102,103 CARMA1 is phosphorylated by PKCθ and binds to
ious molecules containing pleckstrin homology domains Bcl10 through a CARD-CARD domain interaction.104 The
such as Tec family kinases, a serine/threonine kinase CBM complex is responsible for activation of NF-κ B, thus
Akt, phospoinositide-dependent kinase 1, Vav, and WASP T cells defective in any of the CBM components show im-
to the plasma membrane. Akt can be phosphorylated by paired NF-κ B activation and impaired T-cell proliferation.
phospoinositide-dependent kinase 1 and plays an impor- CD28-mediated activation of the CBM complex and NF-κ B
tant role in cell survival, which is one of the benefits of dramatically enhances IL-2 production.
CD28-mediated costimulation.95 The PxxP motif, which Given its powerful activity as a second signal for T-cell
ICOS and CTLA-4 do not have, also has binding specificity activation, CD28 agonists and antagonists have been se-
for Grb2/Gads. Grb2 recruits Sos and Vav, which in turn riously considered for clinical application in humans.
phosphorylates and activates Rac1 and Cdc42, which then CD28 superagonist monoclonal antibodies were shown
activates the MEKK1 cascade to ultimately activate the Jun to polyclonally activate T cells in vivo, but such treatment
kinase JNK.96 Whereas the TCR signal predominantly ac- ultimately led to the expansion of regulatory T cells, and
tivates ERK as an MAPK family member, CD28 engage- several studies showed therapeutic benefit in autoimmune
ment induces activation of the other member of the MAPK and inflammatory disease models in mice and rats. After
family, JNK, and the balance between the activity of ERK preclinical safety testing in nonhuman primates, six healthy
and JNK as a consequence of CD28-mediated costimula- human volunteers were injected with the TGN1412 CD28
tion may regulate cell survival. superagonist monoclonal antibodies, and the results were
Early studies showed that the distal PxxP (PYAP) motif disastrous.105 All six individuals suffered an immediate and
in CD28 may recruit Lck, and it was recently shown that life-threatening release of systemic proinflammatory cy-
fi lamin A also binds to this motif to connect to the cyto- tokines, a “cytokine storm” that is now termed cytokine
skeleton. Although in vitro analysis and an overexpression release syndrome. Recent studies suggest that the source of
system indicated a critical function for these motifs in co- these cytokines was the effector memory population of T
stimulation, in vivo analysis with specific knock-in muta- cells and have clarified why no such adverse effects were
tions revealed that mice with mutated YMNM motifs that seen in rodents and nonhuman primates. Rodents do not
failed to bind PI3K and activate Akt had no overt phenotype. accumulate a large population of effector memory popu-
By contrast, mice with the mutant distal motif had evidence lation of T cells because the generation of these cells re-
of impaired CD28-mediated costimulation, such as reduced quires repeated exposure to infections; these experimental
proliferation and IL-2 production.97,98 animals are kept in extremely clean conditions. The CD4 T

cells in macaques, the nonhuman primate used in the pre- of PI3K. ICOS-mediated PI3K activation is stronger than
clinical studies, unlike human T cells, lose CD28 expres- that mediated by CD28, whereas other signals through
sion when they differentiate into effector memory popula- CD28 that are not triggered through ICOS include Grb2
tion of T cells. binding.108 Such differences in signaling between ICOS
and CD28 lead to differential function. Most notably, un-
like CD28, ICOS does not induce IL-2 gene transcription.
COSTIMULATION THROUGH RECEPTORS OTHER Instead, ICOS-mediated costimulation is critical for B-cell
THAN CLUSTER OF DIFFERENTIATION 28 help for Ig production and germinal center formation.
CD28 is most prominent costimulation receptor and thus ICOS deficiency causes one of the multiple forms of com-
CD28-deficient mice have a general impairment in immune mon variable immunodeficiency, and these patients suf-
responses. However, not all immune response are dampened fered from impaired B-cell function and germinal center
by CD28 deficiency, which indicates that other costimula- formation and from hypogammaglobulinemia including
tion molecules can compensate for some of the remaining IgA deficiency.109
functions. These costimulation receptors include CD2, CD5, Among TNF receptor family molecules, the mole-
CD30, the Ig gene superfamily member ICOS, the TNF re- cules with costimulatory function represent CD27, OX40
ceptor family members CD137 (4-1BB) and CD134 (OX40), (CD134) and 4-1BB (CD137), CD30, herpes virus entry
and LFA-1. These receptors are expressed on the T-cell sur- mediator A (HVEM), and glucocorticoid-induced TNF-
face to induce costimulation signals upon crosslinking with receptor (GITR). Among them, OX40 and 4-1BB mediate
their ligands (Fig. 12.14). the most prominent functions. These molecules augment
In contrast to CD28, which is constitutively expressed T-cell activation upon engagement of their ligands, OX40L
on T cells, ICOS is expressed at a very low level on resting and 4-1BBL, respectively. Engagement of OX40 and 4-1BB
T cells and is inducibly expressed on activated T cells upon with their ligand induces activation of signaling pathways
TCR stimulation, which is the major difference in its func- similar to CD28: PI3K and Akt, and NF-κ B and MAPKs
tion from CD28 on resting T cells.106 As ICOS-deficient (JNK and p38), but this is through activation of the TNF
mice have impaired immune responses similar to that seen receptor–associated factor (TRAF) family of adaptor mol-
with CD28 deficiency, ICOS must play an important role ecules, which is different from CD28.110
in immune regulation.107 ICOS binds to a novel B7 family One of the most important functional differences be-
member, ICOS ligand, which is expressed rather broadly as tween CD28 and others such as ICOS, OX40, or 4-1BB is that
compared to CD80/86 including nonhematopoietic tissues CD28 is constitutively expressed on naïve T cells and plays
upon stimulation with inflammatory cytokines. ICOS is a critical role to stimulate naïve T cells for proliferation, ef-
expressed on the cell surface as a dimer similar to CD28 fector function, and functional differentiation, whereas all
and shares some structural features with CD28, including the others are inducibly expressed upon T-cell activation
the YMXM motif in the cytoplasmic tail that binds p85 and therefore play roles to induce costimulation on activated
APC (DC) T cell
PD-L1 PD-1
PD-L2 FIG. 12.14. Various Costimulation Receptors
ICOS and Their Ligands. Various adhesion
B7 family ICOS-L (B7h2) CD28 family molecules between T cells and antigen-
B7-1 (CD80) CTLA-4 presenting cells (APCs) function to mediate
B7-2 (CD86) CD28 costimulation signals for T-cell activation.
CD40 These molecules are divided into three
CD40L (CD154)
groups: cluster of differentiation (CD)28
OX40L (CD134L) OX40 (CD134) family, tumor necrosis factor receptor fam-
TNF family TNFR family
4-1BBL (CD137L) 4-1BB (CD137) ily, and others including integrin leukocyte
function-associated antigen-1. CD28 family
CD27L (CD70) CD27 molecules consist of two positive (CD28,
MHC TCR-CD3 inducible costimulator) and two negative
CD4 /CD8 (cytotoxic T-lymphocyte antigen [CTLA]-
ICAM-1 (CD54) 4, programmed death-1) receptors and
LFA-1(CD11a/CD18) the individually corresponding ligands.
LFA-3 (CD58) CD2 Whereas some receptors are constitu-
Others ? CD5 tively expressed such as CD28, others are
Others
expressed upon T-cell stimulation such as
? CD9 CTLA-4 and OX-40. Whereas some of the
hyaluronic acid CD44 ligands are expressed exclusively on APCs,
others are widely expressed on variety
cell types.

effector T cells and memory T cells. Thus, there is a big dif- also induces negative regulatory pathways.
ference in the timing of their function during immune re-
sponses. In addition, the distribution of the ligands for these
Inhibitory Effector Enzymes
two groups is different: stimulation through CD28 has to
be induced by professional APCs such as DCs and macro- T-cell activation is initiated by phosphorylation by the src
phages, whereas costimulation through ICOS, OX40, and family kinase Lck, and this most proximal signaling event is
4-1BB can even be induced by ICOS-L, OX40L, and 4-1BB-L actively regulated by the balance between the kinase Csk and
expressed on nonhematopoietic cells in the periphery. The the phosphatase CD45, as described previously (Fig. 12.4).
latter is critical for regulation of autoimmunity and infec- Positive and negative regulation is actively and in parallel
tious diseases. induced downstream of TCR signaling. Therefore, Csk is
the dominant negative regulator of TCR proximal signal by
phosphorylating the inhibitory tyrosine on Lck, thus main-
Signal Lymphocyte Activation Molecule Family for
T-Cell Activation
Signal lymphocyte activation molecule (SLAM; CD150) is a
receptor of a family expressed on various cell types includ-
ing T cells, B cells, NK cells, macrophages, and other cells,
which mediate homotypic cell adhesion. This homotypic in-
teraction induces costimulation in T cells. However, the sig-
naling mechanism for SLAM-mediated costimulation has
been unclear. Studies of the human immunodeficiency dis-
ease X-linked lymphoproliferative syndrome were quite in- SLAM
formative when mutations in the SLAM-associated protein
(SAP; SH2D1A) gene were identified as the cause of X-linked
lymphoproliferative syndrome.111 SAP was found to associ- SLAM
ate with SLAM through homotypic interaction upon activa-
tion. SAP is a small intracellular adaptor protein consisting SAP
of only single SH2 domain. SAP was found to associate with
the src kinase Fyn and appears to function as a bridge be-
tween SLAM and Fyn (Fig. 12.15). This binding mode is
quite unique in that the phosphotyrosine of SLAM binds T cell
SH3
P
SAP through its SH2 domain, whereas the SH3 domain of
Fyn binds through an atypical proline-rich domain in the
SH2
P
SH2 domain of SAP.112 This unusual binding configuration
has been confirmed by crystal structure analysis of the com- Fyn
P
plex. The SAP-Fyn interaction is critical for optimal activa-
tion of PKCθ and NF-κ B upon T-cell activation. In addition,
SAP-induced signaling is important to communicate with B
cells for delivering T-cell help for Ig production. Follicular
T cells express ICOS and SLAM family members and may
provide help for B-cell function in germinal center and hu-
moral responses.113,114
The SLAM family is composed of six different members,
SLAM (CD150), 2B4 (CD244), NTBA (Ly108), Ly9 (CD229), SHIP PKCθ
CD84, and CRACC (CD319), all of which associate with Dok-1/2
SAP. Mice deficient in each of these molecules exhibit a weak
Ras-GAP NF-κB
phenotype, probably indicating that the family members
have overlapping and redundant functions.
Cytokine production
NEGATIVE REGULATION OF ACTIVATION FIG. 12.15. Signal Transduction by Signal Lymphocyte Activation
T-cell activation is induced to trigger appropriate immune Molecule (SLAM) Family Members Through the Association with
responses and must be regulated to have the right timing, the Adaptor Protein SLAM-Associated Protein (SAP). The arginine
to be at the right location, and to have the proper strength. 78 of SAP binds the SH3 domain of Fyn, which recruits Fyn to the
To appropriately regulate these responses, it is necessary to SLAM/SAP complex. Fyn subsequently phosphorylates tyrosine res-
idues in the cytoplasmic domain of SLAM, which serve as the dock-
impose negative regulation in order to promote a balanced ing sites for SH2 domain–containing inositol phosphatase, which
response, to prevent an excessive response, and to terminate then phosphorylates the adaptor proteins Dok1/2 and Ras-GAP. SAP
the response properly. As described previously, TCR stimula- also binds to PKCθ and activates NF-κB through CARMA1/Bcl-10/
tion induces several activation pathways, but simultaneously Malt1 complex, which cooperates with T-cell receptor signals.

taining it in the inactive state. CD45 has an opposing func- On the other hand, strong agonistic stimuli induce Erk,
tion by dephosphorylating this negative regulatory tyrosine which is rapidly activated and phosphorylates Lck. This
and activating Lck. prevents the phosphorylation and recruitment of SHP-1, re-
Another set of inhibitory effector enzymes is a family sulting in T-cell activation. Whether such a dynamic regu-
of E3 ubiquitin ligases, which are responsible for ubiqui- lation of the kinase-phosphatase regulatory loop is actually
tination of proteins with certain specificity to lead to deg- operating in vivo remains to be determined. Despite the
radation. These are represented by Cbl and Cbl-b,115 and similar structure and phosphatase activity, SHP-2 appears
also include others such as GRAIL, Itch, and Nedd4. Cbl to have different specificity for SH2 binding receptors than
proteins mediate ubiquitination and targets for degrada- SHP-1, such as programmed death (PD)-1 (see following
tion of early signaling proteins include ZAP-70, effectively discussion).
inducing downregulation of T-cell activation. Evidence of Another phosphatase critical for negative regulation of T
the important negative regulatory role of these proteins in cells is the lymphoid tyrosine phosphatase (Lyp), encoded
vivo comes from the analysis of Cbl- and Cbl-b–deficient by the PTPN22 gene. Lyp is a cytosolic protein phosphatase
mice, which display hypercellularity and spontaneous belonging to the prolin-, glutamic acid–, serine- and threo-
T-cell activation.116,117 It has been shown that the anergic/ nine domain (PEST)-rich family of protein tyrosine phos-
unresponsive status of T cells is induced by these E3 ligases, phatase, and its mouse homolog is the PEST domain-rich
particularly Cbl and Itch, by degradation of various signal- phosphatase (Pep).121 Lyp dephosphorylates various TCR
ing molecules. upstream signaling molecules including CD3, Lck, ZAP-70,
and Vav. The negative regulation of Lyp/Pep is mediated by
its association with Csk. The Lyp-Csk complex is formed
Immunoreceptor Tyrosine-Based Inhibitory
upon T-cell activation and exhibits synergistic negative
Motif–Mediated Negative Regulation by regulation. Deletion of Pep in mice caused expansion of the
Phosphatases memory T cells and increased TCR signaling in effector T
Whereas the ITAM recruits the SH2-bearing tyrosine kinase cells as well as increased positive selection in the thymus, in-
ZAP-70/Syk to induce activation signals in T cells, the immu- dicating that Lyp/Pep is a key negative regulator of TCR sig-
noreceptor tyrosine-based inhibitory motif (ITIM) recruits naling.122 Another critical finding is that a single-nucleotide
SH2-bearing phosphatases to induce negative signals.118 polymorphism in the PTPN22 gene correlates strongly with
The ITIM consensus amino acid sequence is very similar to the incidence of type 1 diabetes, rheumatoid arthritis, ju-
the ITAM: S/I/V/Lx Y xx I/V/L. Receptors containing ITIM venile rheumatoid arthritis, systemic lupus erythematosus,
include killer inhibitory receptors, which mainly function Graves disease, generalized vitiligo, and various other au-
in NK cells to inhibit activation upon recognition of MHC toimmune disease.123 How such single-nucleotide polymor-
and also inhibit killing by cytotoxic T cells, the inhibitory phism results in increased susceptibility to various autoim-
receptors of various paired receptors expressed on NK cells, mune diseases remains to be determined.
DCs and macrophages such as PIR-B and PILR, and the
inhibitory IgG Fc receptor Fc γRIIB on B cells, macrophages,
and mast cells. The tyrosine residue of an ITIM in T cells Inhibitory Adaptors
is phosphorylated by the src family kinase Lck, similar to There are several adaptor molecules that function to mediate
an ITAM, which then recruit phosphatases such as the SH2 negative regulation of T-cell activation by recruiting phos-
domain-containing phosphatase (SHP)-1 and SHP-2 and phatases.124 These include SLAP, TSAd, Dok, and Gab2.
the SH2 domain–containing inositol phosphatase (SHIP). Dok1,2 and Gab2 recruit rasGAP and SHP-2, respectively.
SHP-1/2 is cytosolic protein phosphatase containing two- Although the Gab and Dok family are known to be positive
tandem SH2 domains at the N terminus and a phosphatase regulators of cytokine or growth factor signaling in nonhe-
domain at the C terminus. The phosphatase is in an inac- matopoietic cells, they show opposite inhibitory function in
tive closed configuration in the resting stage and becomes antigen receptor signaling. Gab2- and Dok-1/2–deficiency
active upon tyrosine phosphorylation of the ITIM. These results in enhanced phosphorylation of upstream signaling
enzymes are recruited to the vicinity of the TCR complex molecules and increased T-cell activation including cytokine
by association with Lck and then dephosphorylate kinases secretion and proliferation. Dok-1/2 expressed in T cells as-
themselves (Lck and ZAP-70), and upstream signaling mol- sociates with the phosphorylated ITAM of the CD3ζ chain
ecules close to the TCR complex including Vav, PLCγ1, and and therefore inhibits CD3ζ binding with ZAP-70. Dok-1/2
SLP-76. The functional importance of SHP-1 is evident by then functions to bridge the TCR complex and rasGAP to
the phenotype of SHP-1–deficient mice, which have severe inhibit T-cell activation.125 Gab2 is an adaptor protein and
autoimmunity.119 also negatively regulates TCR signaling by recruiting the
Positive versus negative regulation of Lck kinase activ- phosphatase SHP-2. Gab2 associates with LAT through
ity appears to correlate with the capacity of T cells to dis- binding with Gads upon T-cell activation. Because Gads
criminate between strong and weak stimuli.120 It has been constitutively associate with SLP-76 and the Gads-SLP-76
shown that weak antigens or antagonists induce rapid phos- complex binds to phosphorylated LAT, Gab2 is competing
phorylation of SHP-1 by Lck prior to activation of a posi- with the Gads binding with SLP-76. Upon TCR stimula-
tive signaling cascade. SHP-1 then dephosphorylates the action, ZAP-70 phosphorylates Gab2, which serves as a bridge
tive site of Lck, resulting in inhibition of T-cell activation. between LAT and SHP-2 to mediate inhibitory signals, or

phosphorylates SLP-76, which serves to induce downstream phosphorylated upon stimulation and several phosphatases
activation signals.126 It is important to note that expression including SHP-1/2 and PP2A have been suggested to be in-
of these inhibitory adaptor proteins is induced upon T-cell volved in CTLA-4–mediated inhibitory function,131 further
activation, indicating that these inhibitors function in feed- analysis is required to clarify this point.
back regulation of T-cell activation. PD-1 is also inducibly expressed on T cells upon TCR
Whereas Dok targets the ITAM of CD3ζ and Gab2-SHP-2 stimulation; its ligands are PD-L1 and PD-L2 (Fig. 12.14).
targets LAT, the serine/threonine kinase HPK1 inhibits SLP- Whereas PD-L1 is mainly expressed by professional APCs
76 function by inducible association with its SH2 domain such as DCs and macrophages, PD-L2 is expressed by a rela-
and recruitment of 14-3-3.127 Deficiency of HPK1 in T cells tively wide range of cells including in nonlymphoid tissues.
results in enhanced phosphorylation of early TCR signaling PD-1 possesses two tyrosine motifs within its cytoplasmic
molecules, indicating that HPK1 serves an inhibitory func- domain: ITIM and immunoreceptor tyrosine-based switch
tion for T-cell activation in vivo. motif, and the immunoreceptor tyrosine-based switch motif
(ITSM) is the site for recruitment of the phosphatase SHP-2
upon PD-1 ligand engagement. PD-L1/2 ligation induces
Inhibitory Costimulation SHP-2 phosphorylation and dephosphorylation of various
In contrast to the positive costimulation receptors leading TCR upstream signaling molecules including CD3ζ, ZAP-
to activation signals, there also exist sets of costimulation 70, Vav, and Erk, which results in inhibition of T-cell activa-
receptors that mediate negative signals for T-cell activation. tion.134,135 There is an interesting contrast between the mech-
These include CTLA-4 and PD-1. Neither of these inhibi- anisms of PD-1 and CTLA-4 inhibition of T-cell activation.
tory costimulatory receptors is expressed in resting T cells, Whereas PD-1 inhibits the very proximal signaling just
but both are inducibly expressed on activated and special- downstream of TCR complex, CTLA-4 mainly suppresses
ized T cells, indicating that these inhibitory molecules are CD28-mediated costimulation signals targeting NF-κ B ac-
involved in the feedback regulation of T-cell activation to tivation. Recent studies indicate that T cells from chronic
prevent excess activation during immune responses. The viral infection and immune senescence are in a nonfunc-
physiologic importance of both receptors for inhibition of tional or unresponsive/suppressive state that is mediated
T-cell activation is evident by the strong phenotype of the by PD-1.136 Because the suppressive state can be reversed by
respective gene targeted mice. CTLA-4–deficient mice have interruption of the interaction between PD-1 and PD-ligand
an excess of activated peripheral T cells, splenomegaly, and by anti-PD-1 or PD-L1 antibody, PD-1 appears to induce an
infi ltration of lymphocytes in nonlymphoid tissues, and ongoing active suppression of T-cell activation during the
exhibit a lymphoproliferative disease in all immune tissues course of chronic infection.
and early death.128,129 On the other hand, deletion of PD-1
results in less dramatic and slow-onset autoimmune dis-
eases, including the development of a lupus-like disease and CONCLUSION
cardiomyopathy in old mice.130 We describe here the current knowledge of the mechanisms
CTLA-4 and CD28 have common ligands, CD80 and of T-cell recognition and activation as the central regulator
CD86, on APCs (Fig. 12.14). However, as CTLA-4 has much of immune responses. The unique features of the antigen
higher affinity for these ligands than CD28, even low ex- recognition by TCR are twofold. One is gene rearrangement
pression of CTLA-4 on the T-cell surface upon stimulation to generate enormous diversity together with the unique
can effectively compete for ligand binding with CD28. This T-cell selection processes as positive and negative selec-
is the main mechanism by which CTLA-4 inhibits T-cell tion during thymic development. The second is the result
activation.131 Except for regulatory T cells, naïve resting T of MHC restriction, whereby T-cell recognition of antigen
cells do not express CTLA-4. CTLA-4 is transcribed upon can only occur on the cell surface of APCs in the complex
T-cell activation and usually accumulates in the lysosome with MHC. That is why T cells communicate with many cell
and is not expressed on the cell surface, probably because types of APCs.
such strong inhibitory molecules should be transiently ex- The TCR genes and the TCR-CD3 complex were cloned
pressed only when needed upon T-cell activation. Upon and characterized nearly three decades ago, and the
TCR stimulation, the CTLA-4–containing lysosomes are peptide-MHC binding structure and TCR recognition of
translocated and fuse with the plasma membrane, and a single MHC/peptide complex were clarified two decades
CTLA-4 is then expressed on the cell surface.132 Once on the ago. During the ensuing two decades, new insight regarding
cell surface, CTLA-4 can compete for ligand binding with T-cell recognition and activation was obtained by the dis-
CD28 and thereby push CD28 and the associated PKC θ covery of the IS, and more recently TCR microclusters, as
away from the central region of the interface between T cells the machinery for T-cell activation. The discovery of various
and APCs, where the CD28-mediated costimulation signal T-cell subsets such as Th17 and regulatory T cells during the
is mediated.133 The competition for ligand binding with following decade brings closer to reality the possibility of
CD28 by the ectodomain of CTLA-4 is a widely accepted regulating T-cell function by analyzing the detailed mecha-
phenomenon, but it has been controversial whether CTLA-4 nism of their differentiation and function. Further analysis
can also exhibit inhibitory function by recruiting a phos- on T-cell recognition, differentiation, and activation can be
phatase to its cytoplasmic tail in a manner independent of addressed by using the combination of current technolo-
ligand binding. Although it has been shown that CTLA-4 is gies of conditional knockout mice with tissue or temporal

specificity, of imaging analysis of the dynamic features of Foxp3 and its regulatory function in regulatory T cells, and
various molecules within a cell at single molecule level, and SAP and SLAM family receptors. All of these insights came
of in vivo dynamics of lymphocytes, and also system biologic from the analysis of primary immunodeficiency patients.
approaches including quantitative analysis of the expression For the cure of such diseases, the development of suitable
of genes and proteins by proteome analysis, together with drugs is also critical in the near future. During the past de-
the pathway analyses of signaling molecules and modeling cade, several immunologic reagents including monoclonal
of regulatory networks. antibodies and inhibitors have been successfully developed
Further elucidation of the mechanism of T-cell activation and applied, such as FK506 as a potent inhibitor of T-cell
and function will provide benefits for overcoming immuno- activation and critical for organ transplantation, monoclo-
logic diseases. The immunologically important questions, nal antibodies against TNF, IL-6 receptors for various auto-
such as repertoire selection, induction, and maintenance of immune diseases, and against JAK3 for therapy of tumors
self-tolerance and long-term memory, are still elusive and and immune disorders. These therapeutic reagents should
all are related to the regulation of autoimmune diseases. be further developed on the basis of our understanding of
Therefore, the future development of research in this area the fundamental mechanisms of immune regulation. Such
should be focused on overcoming these immunologic dis- new developments should be targeted not only for simple
orders by clarifying the principles of this regulation. In this inhibitors of kinases and phosphatases, but also to target
regard, analysis should move more toward human diseases. the dynamic movement of lymphocytes, transcriptional and
Analysis of patients with immune disorders have revealed the epigenetic regulation of gene expression, and the develop-
identity of critical genes and regulatory mechanisms such as ment of antibodies against functional surface molecules of
the calcium channel Orai1 and its activation and regulation, several critical T-cell subsets.

T-Lymphocyte
CHAPTER
13 Developmental Biology
Ellen V. Rothenberg • Ameya Champhekar
OVERVIEW OF T-CELL DEVELOPMENT or useless TCR specificities. These “repertoire selection”

Introduction checkpoints are major landmarks in the T-cell development
process, and they often provide choice points in develop-
T-cell development is a distinctive branch of hematopoiesis mental fate as well.
as well as a crucial input to immune system function. T-cell Both TCR recognition specificity and effector function
precursors, like other blood cells, are continuously gener- subdivide T cells, and one of the most notable features of
ated from hematopoietic stem cells through a process that T-cell development is that particular functional subtypes
involves loss of access to alternative hematopoietic fates as become preferentially associated with particular specificities
well as gain of lineage-specific characteristics. In addition, of antigen recognition. For one major divergence, the clus-
like B cells, T cells must undergo programmed rearrange- ter of differentiation (CD) 4+ “helper” versus CD8 + “killer”
ment of gene segments in order to assemble the genes that decision, a gene network explanation has emerged, but ex-
will encode their receptors for antigen, the α , β, γ, and δ planations for differentiation of TCRγδ versus TCRαβ and
T-cell receptor (TCR) chains. T-cell development is more “innate-type” versus “conventional” T cells of both receptor
distinctive in other ways. Unlike almost all other hematopoi- types remain a challenge.
etic cell types, T-cell precursors must leave the bone marrow This chapter covers T-cell differentiation both as a
in order to adopt their fate. Most of them must migrate to hematopoietic developmental program and as an immu-
the thymus for their differentiation: a specialized epithelial nological process that trains new cohorts of cells for appro-
organ that provides a framework for intensive signaling that priate discrimination between recognition of those targets
guides the cells to become T cells at the expense of other de- that should elicit attack and those that should elicit protec-
velopmental fates. The T-cell program itself is also remark- tion. It begins with an introduction to the molecules that
able because of the variety of different types of T cells that define T-cell identity and distinguish stages in the T-cell
it produces. Despite indelibly establishing common T-cell development process; a global synopsis of intrathymic T-cell
features, the cells that pass through T-cell development re- development is then presented as a framework for the rest of
tain great functional plasticity and eventually fan out into the chapter. Subsequent sections zero in on particular steps
a spectrum of different effector subtypes. Some of their in the process, developmental branch points, and check-
diversification begins in the thymus, even while their uni- points, in which crucial aspects of mechanism have been
versal T-cell properties are being acquired. Remarkably, as illuminated. The chapter then closes with a survey of some
described in the following, some of the same transcription open challenges that are faced in the next few years.
factors simultaneously play roles both in universal T-cell
gene expression and in competitive oppositional mecha-
nisms that determine particular effector subtype choices. Key Molecules for T-Cell Identity: Receptors,
Thus, the T-cell program is more than a simple progression Stage Markers, and Effector Molecules
toward maturity: it is a mosaic of developmental processes One goal of T-cell development is to arm mature T cells with
that on the one hand forge a durable T-lineage commitment a full set of the molecular criteria for T-cell identity. The ma-
and on the other hand open up many different routes to use- ture T cell requires not only recognition apparatus, centered
ful T-cell functionality. around the TCR complex, but also signal transduction ap-
T-cell function depends on recognition of antigen, and paratus to convey quantitative recognition information to
so a crucial part of the arming of the T-cell precursor for the nucleus, and a poised transcriptional response system
function is assembly of the TCR complex. To generate co- that can induce an appropriate combination of effector
horts of T cells with various different recognition speci- genes in response.
ficities, this process depends on individual cells’ random
choice of TCR gene segments for rearrangement, and on ad- T-Cell Receptor Complex Components
ditional chance incorporation of somatic mutations in the In all jawed vertebrates, there are two forms of TCR complex:
TCR genes, potentially different for every cell. The process one with antigen recognition mediated by an αβ heterodimer
is capable of generating autoreactive receptors or defective and one using a γδ heterodimer. Individual T cells use either
receptors as well as useful ones. Thus, the T-cell develop- αβ or γδ receptors, but not both. All four chains, α , β, γ,
ment program also incorporates several discrete fi ltering and δ, are multidomain immunoglobulin superfamily pro-
steps to eliminate newly made cells if they have dangerous teins, type I transmembrane proteins (single-pass proteins
325

326 | SECTION IV T LYMPHOCYTES
with N termini on the exterior of the cell). The highly vari- CD25 is most often a marker for Foxp3 + Treg cells. However,
able N-terminal domain sequences of TCR chains mediate all three markers have different roles in T-cell development,
recognition of antigen, usually a peptide bound to a major where their expression patterns usefully distinguish vari-
histocompatibility complex (MHC) molecule on the surface ous stages.
of an antigen-presenting cell (APC). The variable domains Thy-1, a glycolipid-linked membrane receptor, is highly
are encoded by combinations of gene segments that as- expressed in all murine T-lineage cells from an early stage
semble variably during a specific developmental phase when on, though not in human. CD24 or “heat-stable antigen,”
somatic mutation/recombination occurs. TCRα and TCRγ another cell surface molecule expressed in many hemato-
coding genes are assembled by joining segments of two types poietic precursors, is highly expressed in all TCR-negative
(V to J), while TCRβ and TCRδ genes are assembled by join- stages, gradually downregulated in a stepwise fashion
ing V, D, and J segments. Once the rearrangements have throughout T-cell development, and finally turned off when
finished, the receptor sequences and recognition specifici- intrathymic maturation is complete.
ties of the T cells are fi xed forever afterwards. As a rule, each
T cell expresses only a single TCRα (or γ) and only a single Growth Factor Receptors
TCRβ (or δ) chain. TCR heterodimer presentation at the cell Despite the importance of interleukin (IL)-2 and IL-4 in
surface then depends on assembly with a complex of signal- mature T cells, complete IL-2 and IL-4 receptors are not
ing chains that do not themselves mediate antigen recogni- required for most aspects of T-cell development. However,
tion, called CD3γ, δ, and ε and TCRζ (or CD3ζ). Like the receptors for IL-7 are crucial for the initial expansion of
TCR genes themselves, expression of the CD3 and ζ chains is T-cell precursors in early stages of T-cell development, in-
T-cell specific and a result of gene activation induced by the dispensable for TCRγδ cell development, and probably
T-cell developmental program. also important for divergence of CD4 + (helper) and CD8 +
(killer) branches of TCRαβ cell development. At the earliest
T-Cell Receptor Coreceptors and Signal Modulators stages of T-cell development, IL-7R function is anticipated
TCR engagement with MHC-peptide ligands on APCs is and supplemented by Kit, which is the receptor for “stem
greatly enhanced by coreceptors called CD4 and CD8 that cell factor” (steel factor, Kit ligand), that is also essential for
interact with the most conserved parts of the MHC mol- early T-cell development. Kit and IL-7R are dynamically
ecules. CD4, a linear, monomeric immunoglobulin super- regulated with distinct patterns of expression across early
family molecule, interacts with class II MHC, while CD8, stages of T-cell development.
a dimeric molecule (either CD8αβ or CD8αα), interacts
with class I MHC. The cytoplasmic domains of both mol-
ecules provide docking sites for a Src-family protein tyrosine Effector Response Molecules
kinase, Lck, that plays a key role in T-cell signaling. Thus, Mature T cells are armed for particular stereotyped re-
CD4 and CD8 not only stabilize TCR-MHC complex inter- sponses to antigen recognition. Most CD4 helper cells make
actions but also ferry a major signaling mediator to the site their first antigen response by turning on expression of the
of TCR-MHC interaction. growth factor IL-2. Killer T cells respond to activation by ex-
Lck activation upon TCR engagement triggers a complex pressing the membrane-perforating molecule perforin, gran-
cascade of signaling pathways. While some of the pathway zymes A and B, and relatively toxic cytokines like interferon
components are universal mediators of calcium, PI3-Kinase, (IFN)γ. Th1 cells do so by expressing IFNγ and TNFα , but
Ras/MAP kinase, and NF-κ B activation, T-cell responses much less perforin. Th2 cells respond by expressing IL-4 and
also use specialized components such as the adaptor LAT, in some cases IL-5 and/or IL-13. Natural killer T (NKT) cells
protein kinase Cθ and the T-cell-specific kinases Zap70 and rapidly activate IL-4 expression together with IFNγ. Th17
Itk. In addition, other cell surface receptors also feed into cells, often found defending epithelia, express inflammatory
these pathways to modulate the signals. One signal-ampli- cytokines such as IL-17A, IL-17F, and IL-22. Additional sub-
fying receptor is CD28, an immunoglobulin superfamily sets, follicular helper T cells and Th9 cells, also have particu-
disulfide-linked dimer. Another transmembrane signaling lar canonical effector programs that they use to respond to
molecule, CD5, is induced by TCR signaling and appears TCR signals. Whereas all these subtypes promote immune
to provide feedback signal-damping functions. These mol- activation upon stimulation, regulatory T cells (Treg cells) re-
ecules are not strictly T-cell specific but are tightly develop- spond to stimulation by blocking the effector responses of
mentally regulated within the T-cell pathway. neighboring T cells, at least in part through expression of
IL-10, TGFβ, and specific cell-surface interaction molecules.
Activation Markers/Developmental Stage Markers In mature T cells, the poised state of distinct sets of cyto-
When T cells are activated, three additional cell-surface kine genes that distinguishes each subset is maintained, in
molecules are usually upregulated with different kinetics. between rounds of stimulation, by transcription factor ex-
CD69 is usually fi rst, expressed transiently, followed by pression patterns and site-specific chromatin modification.
CD25 and then by CD44, which is the most persistently ex- Particular transcription factors provide distinctive, crucial
pressed. CD25 can be used as a subunit of the receptor for regulatory inputs in particular cell types. Thus, Runx3,
the growth factor IL-2 (as IL2Rα) but can also be expressed T-bet, and eomesodermin are important to maintain killer
alone. In mature T cells, persistent CD44 expression marks function, T-bet to maintain Th1 function, GATA-3 to pro-
memory and antigen-experienced effector cells, whereas mote Th2 function, PLZF to promote NKT-cell function,

CHAPTER 13 T-LYMPHOCYTE DEVELOPMENTAL BIOLOGY | 327
RORγ t to promote Th17 function, Bcl6 for follicular helper fraction of thymocytes (∼80%), and many of them come
T-cell function, and Foxp3 to promote Treg function. Of par- to express complete TCRαβ receptors as well. However, the
ticular interest is the fact that most of these factors also play overwhelming majority of these cells are fated never to go
pivotal earlier roles in the generation of T cells in general. further in their differentiation. Only a small fraction are
The differentiation of particular effector subsets is usu- allowed to progress further through positive selection, to
ally considered to start with antigen-inexperienced (naïve) become either CD4 + CD8 − TCRαβ + cells (“CD4 cells”) or
mature T cells, after they have finished with their intra- CD4 − CD8αβ + TCRαβ + cells (“CD8 cells”). It is estimated
thymic development. The specialization itself is part of the that the cells have only about 3 days within the DP stage in
mature T cell’s portfolio of possible responses to antigen, es- which to achieve this success before their window of op-
pecially among CD4 T cells. Conventionally, one may con- portunity closes. Left in the cortex, the other DP cells of
sider these antigen-driven events to be a completely different their cohort die “of neglect,” to be replaced by progeny of
type of developmental process from the intrathymic events the next cohort of precursors, at turnover rates up to 50 ×
that mold hematopoietic precursors into T cells. However, 106 per day.
we will see in the following that effector programming can
also originate in the thymus for subsets like NKT cells, cer- T-Cell Receptor-à Rearrangement and
tain classes of TCRγδ cells, and some Tregs: the distinction is Selection Checkpoints
not absolute. The regulated process of TCR rearrangement and selection
helps to explain why so many DP cells must be produced.
TCR coding genes are assembled through a highly regulated
Narrative Summary: Major Stages of Intrathymic but highly imprecise recombination process mediated by the
T-Cell Development recombinase complex RAG1/RAG2, the same recombinase
The major known site for T-cell development in all jawed that rearranges immunoglobulin VDJ gene segments in B
vertebrates is the thymus.1 The thymus does not harbor its cells. The imprecision is useful to add to the diversity of the
own long-term source of precursors but is continuously pop- eventual T-cell recognition pool, but it also results in many
ulated throughout life by new precursors that are replaced as gene rearrangements that are out of reading frame or oth-
their descendants complete maturation and selection. These erwise defective. To winnow out the useful cells, TCR gene
precursors migrate to the thymus from bone marrow or fetal rearrangement occurs in two sequential bursts, each fol-
liver when they are still multipotent. The thymus not only lowed by a quality control checkpoint at which all the cells
enforces their commitment to a T-cell fate and T-lineage with defective receptors can be eliminated.
differentiation but also provides a proliferation-inducing The timing of different phases of recombination compe-
environment. It is estimated that the numbers entering the tence is determined by the specific activation and deactiva-
thymus per day are quite modest, on the order of < 100 per tion of the RAG1 and RAG2 gene products; the genetic loci
day for a typical young adult mouse.2,3 The descendants of on which their effects can be focused at any given time are
that small cohort of input cells expand over a 2-week period determined by developmentally regulated unmasking via
to a cohort of cells that have undergone T-lineage commit- localized chromatin opening. For TCRαβ cells, the most
ment, TCR gene rearrangement, and TCR expression, even- numerous T cells in mice and humans, the TCRβ gene must
tually yielding ∼50 × 106 new thymocytes a day before their rearrange first, while the cells are still in the DN stage. The
proliferation stops. cells carrying out this rearrangement need to pause prolifer-
ation in order to allow the RAG complex to work, and most
Major Subdivisions Based on CD4, CD8, and T-Cell often they are never allowed to divide again unless they suc-
Receptor-à Status ceed in generating a good TCRβ coding sequence through
The steps of T-cell differentiation in the mouse thymus are rearrangement at one allele. This strict condition is the
summarized in Figure 13.1.3–6 Important landmarks for the “β -selection checkpoint.” Only cells passing this checkpoint
process are provided by the expression patterns of the TCR are allowed to become DP cells, receiving a bonus of multi-
coreceptors, CD4 and CD8, and the timing of rearrange- ple rounds of cell division to expand the winners and dilute
ment of the TCR-coding genes themselves. the stalled, dying losers. Through the process of β -selection,
Cells in the early stages of development in the thymus then, the cells also shift the gene loci that are accessible for
are CD4− CD8− (“double negative” [DN]) and cannot rearrangement, losing the ability to rearrange their TCRβ
yet express TCR. While the DN stages can be subdivided genes (and TCRγ and TCRδ genes) any further and acquir-
(described subsequently), the cells in all the DN stages are ing the ability to rearrange their TCRα genes.
mostly determining their commitment to a T-cell fate, pro- Once they become DP cells, TCRα rearrangement gets
liferating, and preparing for their fi rst expression of TCR underway. As a successful in-frame TCRα chain rearrange-
complexes. With successful expression of a TCRβ chain, ment occurs, the heterodimer TCR complexes of the new
they turn on CD4, CD8α , and CD8β genes together. The TCRα chains with the previously generated TCRβ chains
resulting CD4 + CD8αβ + “double positive” (DP) cells are become the next subject of testing at a second checkpoint.
highly distinctive to the thymus, not found among mature This second checkpoint is more draconian than the first,
T cells in peripheral lymphoid organs, and are diagnos- as the criterion for success here is not simply successful
tic of a pivotal stage in T-cell development and survival. expression of a TCRαβ heterodimeric protein, but also the
The DP cells in steady state constitute by far the major quantitative details of the recognition specificity of the new

THYMUS
~ 2 weeks ~ 3 days ~3 days ~5 days
DC TCRγδ
NK TCRβ+
TCRαβ+
Other
hematopoietic
cell types Positive CD8
β−Selection Selection TCRhi
CD4
CD4
CD8 CD8 TCRhi
(Prethymic) DN1 (ETP) DN2 DN3 DN4 ISP DP iNKT
(death) (death)
c-Kit
CD44
CD25
CD24
CD69
Thy-1
Notch/Delta
dependence
IL-7R dependence
RAG-1,2
recombinase
TCRγ, δ rearrangement
TCR Dβ-Jβ rearrangement
TCR Vβ-DJβ rearrangement
TCRα rearrangement
pTα expression
T genes ON (Specification)
T lineage commitment
FIG. 13.1. Schematic Summary of T-Cell Developmental Stages. Top: Cartoons depict the approximate transit times
and population expansion through the different stages. Middle: Recognized major stages of mouse T-cell develop-
ment are shown, indicating CD4, CD8, and T-cell receptor (TCR) phenotype, and phases that are TCRβ-dependent
or TCRαβ-dependent. Development through positive selection of CD4, CD8, and natural killer T cells is shown; not
shown are pathways to regulatory T cells, CD8αα innate-type cells, alternative γδ pathways, or late-stage negative
selection. White circles: nondividing or slowly dividing stages. Yellow circles: stages of extensive proliferation. Blue
circles: prethymic cells and non-T developmental alternatives. Green arrows: enabled by TCRγδ expression. Magenta
arrows: enabled by TCRβ expression. Purple arrows: enabled by TCRαβ expression and selection. Lower: timing of
expression of additional markers, phases of TCR gene rearrangement, and specification and commitment events (see
text for details).
receptor, as measured by its interaction with ligands in the and low affinity for self-ligands. These are the ones that
thymic environment. The newly expressed TCR could fail are allowed to become CD4 or CD8 cells through a process
to recognize anything, making it useless, or it could confer termed “positive selection,” with the CD4 cells generally
on the cell a dangerous autoreactivity that could lead to au- becoming cytokine-producing “helpers” and the CD8 cells
toimmune disease. All of these failures must be eliminated generally becoming “killers.” The positively selected cells
before they appear in the mature T-cell population. The only permanently silence their RAG gene expression and un-
successful cells are the ones that happen to have heterodi- dergo further maturation steps but little if any additional
meric receptors with the right combination of functionality cell division before they are sent out to the periphery.

The effects of the ordered TCR gene rearrangements acid [DNA]-dependent protein kinase needed for rejoin-
and quality control checkpoints are seen clearly in the ing of the gene segments during rearrangement [a very
effects of different mutations on the progression of T-cell instructive mutant, Prkdcscid, causes severe combined im-
development. Figure 13.2A,B depicts the CD4/CD8 pro- mune deficiency]). Because TCRαβ -lineage cells are so
fi les and DN subsets in a normal young mouse thymus of predominant in the mouse thymus, the effect of mutations
about 2 to 3 × 108 cells. Figure 13.2C,D shows the effects of Tcrb alone appears very similar (Fig. 13.2E). In contrast,
on thymus populations in mutant mice in which TCR gene mutations of the Tcra locus alone allow production of a full
rearrangement is impossible (eg, via mutation of Rag1 or complement of DP cells, although CD4 and CD8 SP cells are
Rag2 or Prkdc, the gene that encodes a deoxyribonucleic completely missing (Fig. 13.2F).
T-Cell Receptor–based Lineage Diversification for T-Cell

A B Receptor-à Cells
CD4+ The first functional distinctions between TCRαβ cell
SP DP DN1 DN2
subsets apparently arise only during emergence from the
DP stage. Positive selection itself probably triggers these
CD44
CD4
divergences through distinctive modes of positive selec-

tion signaling, as we discuss in detail in a later section. The
CD8+ three main outputs of TCRαβ cell positive selection are
SP
CD4 cells, CD8 cells, and NKT cells, so called because they
DN DN4 DN3
CD8 CD25
share some functional qualities and surface receptors with
Wildtype ~2.5 x 108 natural killer (NK) cells. DP cells fi rst experiencing positive
selection signals immediately turn on the activation marker
C D CD69, start increasing the surface density of their newly
validated TCRαβ complexes, upregulate CD5, and then
begin the process that results in shutting off either CD8αβ
CD44
CD4
or CD4 to generate TCRαβ -high “CD4 single positive” and

“CD8 single positive” cells, respectively. As the cells arrive
at one of these “single positive” (SP) phenotypes, they still
retain expression of another marker of immaturity, the
CD24 “heat-stable antigen” that is expressed from the early
CD8 CD25 DN stages onward. However, they turn off both CD69 and
RAG knockout ~2 x 106
CD24 as they reach completion of their functional matu-
rity, in a process that takes 3 to 7 more days.7 NKT cells
E F undergo a distinctive maturation sequence that is discussed
separately in the following, but they too lose CD69 and
(pseudo DP
DP)
CD24 as they mature.
The selection of these different subsets is based on TCR
CD4
CD4
and coreceptor interactions with different ligands in the

DN
DN thymic environment. CD4 cells are positively selected by
TCR and CD4 joint interaction with MHC class II mole-
cules. CD8 cells are positively selected by TCR and CD8 αβ
CD8 CD8 joint interaction with MHC class I molecules. NKT cells
TCRβ knockout TCRα knockout are positively selected by nonclassical class I MHC mole-
~ 107 ~ 2 x 108 cules, using non-CD4, non-CD8 accessory molecules (sig-
naling lymphocytic activation molecule [SLAM] family
FIG. 13.2. T-Cell Developmental Progression and Thymus Cellularity molecules) to assist in signal triggering. Although the pri-
Controlled by T-Cell Receptor (TCR) Assembly. A,B: Normal wild-type mary trigger in each case is TCRαβ -ligand recognition, the
mouse thymocytes: cluster of differentiation (CD)4/CD8 phenotype of
effector programs of these different major lineages diverge
all thymocytes (A), CD44/CD25 phenotype of double negative (DN) thy-
mocytes only (B). C,D: Recombination-deficient mouse thymocytes: significantly, through the consequences of using these dif-
CD4/CD8 phenotype of all thymocytes (C), CD44/CD25 phenotype of the ferent types of coreceptors to assist in “interpreting” the
cells, which are all DN (D). Data for Rag2 knockout cells are shown, TCR signals. How this signal-dependent feature of T-cell
but Rag1 knockout phenotype is the same and similar phenotypes are maturation works is a major focus of later sections of this
seen for Prkdcscid or Tcrb −/− Tcrd−/− thymocytes. E: Cartoon depict- chapter.
ing CD4/CD8 phenotype of TCRβ knockout (γδ still available). A few
pseudo–double positive cells are generated through transient use of Negative Selection of Autoreactive
a γδ TCR, but the great majority of TCR+ cells in these populations are
TCRγδ+ DN. F: Cartoon depicting CD4/CD8 phenotype of TCRα knock-
T-Cell Receptor-à Cells
out: development through β-selection is unimpaired, but no single Many cells with TCR that might be dangerously autoreac-
positive populations are formed. Note the effects of each genotype on tive can be eliminated at the DP stage, but another crucial
the cell numbers per thymus. period for this aspect of quality control is after positive

selection. As newly generated single-positive cells display Distinctive Paths for T-Cell Receptor-à Cells
increased surface density of their TCRαβ, raising their T cells that will use TCRγδ instead of TCRαβ actually follow
avidity for MHC-bound ligands, those that have strongly a special program or set of program options from the DN2
autoreactive receptors are generally killed off by interac- stage on. Whereas TCRβ rearranges strictly before TCRα
tion with specific subsets of thymic stromal cells in a pro- and mostly in DN3 stage, the TCRγ and TCRδ genes can
cess called “negative selection.” This continual purging of rearrange in parallel within the DN2 or DN3 stage. Thus,
the newly made T-cell recognition repertoire is crucial to a RAG1/RAG2 + cell can in some cases acquire successful in-
avoid autoimmune disease, as important mouse mutations frame rearrangements of both TCRγ and TCRδ genes before
and human disease models reveal. An alternative pathway it is successful with TCRβ alone. As a result, newly TCRγδ +
to defeat autoreactivity, at least among CD4+ cells, is to cells arising within the DN2 or DN3 population leave the
alter the functional subtype of the autoreactive cells, to TCRαβ lineage mainstream, undergoing a γδ -selection pro-
program them for function as tolerogenic regulatory T cess that is different from β -selection. The γδ -selected cells
cells before they even have a chance to leave the thymus. generally keep CD4 and CD8 silent, shut off RAG1/RAG2
Generation of such “natural Tregs” (nTregs) is but one case permanently without promoting TCRα rearrangement, and
where TCR interaction properties direct the cells to a par- limit their proliferation to a much smaller number of cell
ticular T-cell functional class. cycles than β -selected cells.
It is not clear yet whether all TCRγδ cells must go through
Generation of T-Cell Precursors in the two sequential checkpoints analogous to β -selection and
Double Negative Compartment positive selection. Like TCRαβ cells, TCRγδ cells are now
For years, the least well understood set of thymocytes was understood to emerge in multiple functional subtypes, and
the DN cells. The DN-stage cells comprise only a small frac- we will discuss in a separate section possible ways that these
tion of thymocytes in steady state, generally less than 5%, subtypes become differentially programmed.
but in fact represent the generative compartment of the thy-
mus in which the largest number of total cell cycles takes
place. Profound developmental transformations also occur CONTEXT OF T-CELL DEVELOPMENT
during transit through the DN stages. Anatomical Organization of T-Cell Development
Understanding the basis of T-cell specification became Major Thymic Domains
possible only through the discovery of markers that could The structure of the thymus plays a substantial role in the
subdivide the DN compartment into successive stages ordering and tempo of the steps in T-cell development.3
(reviewed in refs. 2,3). The earliest stages of precursor The epithelial framework of the thymus is a derivative of
differentiation within the thymus are represented by DN the fetal pharyngeal endoderm, emerging from the third
(CD4 − CD8 − TCR− or sometimes CD4low) cells that ex- branchial pouch in midgestation and only later descending
press high levels of the growth factor receptor Kit and the from the neck region to its later position lying just over the
activation/adhesion molecule CD44 but low levels of CD25, heart. At its peak of function, the thymic epithelium looks
termed Kit-high DN1 cells or early T-cell precursors (ETPs). very different from most endodermal epithelia, as it has be-
Then, these cells turn on expression of CD25, signaling their come a highly porous, “lacy” three-dimensional lattice with
definitive entry into the T-cell pathway, and the resulting few tight junctions, no obvious apical or basal polarity, and
Kit + CD44 + CD25 + cells are called “DN2” cells. Both the most of the spaces between epithelial cells crammed with
DN2 and ETP stage cells proliferate, and IL7R is increas- developing lymphocytes. The structure is not uniform, as
ingly expressed in the DN2 stage. The DN2 cells then prog- each lobe of the thymus is organized into cortical (outer)
ress into the “DN3” stage (CD25high but now CD44low and and medullary (inner) regions. The epithelial lattice appears
Kit low), and here proliferation slows or stalls, RAG protein quite open in the cortex but more compact in the medulla,
levels rise, and efficient TCRβ rearrangement can proceed. where the lymphocyte:epithelial cell ratio is much lower.
Development terminates at the DN3 stage unless the cells These different regions contain distinct epithelial cell types
can pass β -selection or γδ -selection. and different nonlymphoid hematopoietic cell types in addi-
Successful TCRβ expression and passage of β -selection tion to the developing T cells (Fig. 13.3).
enable the cells to restart proliferation, shut off expression
of CD25, and pass through a transitional “DN4” or “preDP” Ordered and Checkpoint-Controlled Migration Pathways
stage (CD25low, CD44low, Kit low). They then proceed to ac- for Developing T Cells
quire CD8 and CD4 on the cell surface, sometimes CD8 Multipotent precursors follow a distinctive traffic pattern
slightly before CD4 (“immature SP”), and fi nally culminate through the thymus8 (see Fig. 13.3). They enter through
their differentiation into DP cells by finishing their prolif- postcapillary venules at the cortical/medullary border, and
eration (large DP to small DP transition). This extended they spend the bulk of the ETP stage proliferating quite
β -selection-dependent transition from DN3 to small DP close to this boundary. Then, as they enter the DN2 stage,
cells creates a unique regulatory state in the resulting DP they begin to migrate centrifugally, toward the outer cor-
cells that prepares them for selection. Notably, although tex. They are thought to reach DN3 stage in the outer sector
some of these cells may be positively selected as described of the cortex and undergo the burst of proliferation trig-
previously, only a minority may ever divide again until after gered by β -selection at the extreme periphery of the cortex,
they have left the thymus. just under the capsule and far away from the initial site of

SUBCAPSULAR DN3
ZONE β-selection
T-lineage DN4/
commitment ISP
proliferation
γδ? DN2
PCV Death DP
ETP
CORTEX
CORT/MED Negative Positive

JUNCTION selection selection Immigrant CD69+
DP
Maturation
MEDULLA
cTEC
CD4
mTEC CD8
Export
A DC B
FIG. 13.3. Anatomical Compartments of the Thymus and Migration Pathways of Developing T-Cell Precursors.
A: Major compartments and migration pathways are depicted with key checkpoints in T-cell development
indicated. Cort/med junction, cortical-medullary junction; PCV: postcapillary venule (immigration portal).
B: Phenotypes of T-cell precursors (pink, red, violet) and stromal cell types in different compartments. cTEC,
cortical thymic epithelial cells; DC, dendritic; mTEC, medullary thymic epithelial cells. For clarity, the “out-
bound” path from ETP to DN3 is separated from the “inbound” path from DN4 to CD4 or CD8 single positive
maturation, but in vivo the “outbound” transit is surrounded with the “inbound” cells.
precursor entry. Then, the DP cells generated during this Thymic Anatomy Orders the Presentation of Inductive
proliferation start to fall back down through the cortex, and Selective Signals
covering the same ground as the DN cells that were their The pathway of T-cell precursors through the thymus helps
precursors, but now traveling in the opposite direction. The to order their exposure to key differentiation-inducing
sheer number of these DP cells dominates the population in stimuli.2,3,9 The cortical epithelium expresses high levels
the cortex: in a healthy young adult mouse thymus, for the of integral membrane cell surface ligands for the Notch
∼2 × 104 DN2 cells migrating toward the outer cortex, there signaling receptor, mostly Delta-like 4 (DLL4, or DL4).
are over 1.5 × 108 DP cells falling back toward the medulla. This is important because Notch-Delta interactions play a
Thus, the DN cells must work their way up through the vast central role in specification of T-cell progenitors from un-
excess of DP cells as they acquire T-cell character, and every committed precursor cells, as discussed in detail in the
other cell type in the cortex is embedded in a slow-moving following section. In addition, the epithelium closest to the
sea of DP cells. As discussed in the next major section, this cortical-medullary junction makes the highest concentra-
countercurrent migration may play an important role in tions of the cytokine IL-7, which supports both early and
helping the earliest precursors to receive inductive signals late T-cell development. Epithelial cells expressing Kit ligand
efficiently from the epithelium without competition. appear to be present in multiple domains of the thymus. As
Despite their huge numbers, the DP cells are normally the precursors enter the thymus, therefore, they can rapidly
denied entry into the medulla. Positive selection signals engage in Notch-DLL4 signaling interactions in the presence
are needed to induce new chemokine receptor expression of maximal concentrations of IL-7 and Kit ligand. These are
(CCR7) on the fortunate minority of DP cells, and these are ideal conditions for proliferation in the ETP and early DN2
the only ones that can plunge into the medulla to continue stages. As the cells move outward, they encounter decreasing
their differentiation.2,9 After further maturation, once all IL-7, which enables them to slow proliferation and begin to
signs of TCR signaling subside, the cells are fi nally allowed carry out efficient TCR gene rearrangement.
to upregulate emigration receptors that will allow them to Thymic epithelium is one of the rare epithelia in the body
leave the thymus entirely. It is also in the medulla, however, that can express MHC class II molecules, which are normally
that even more stringent testing for TCR autoreactivity and limited to hematopoietic APCs, as well as the more broadly
purging of autoreactive cells takes place. Ultimately, less expressed MHC class I cell-surface molecules. Because TCR
than 5% of the cells from each cohort of DP cells emerge interactions with class II MHC are crucial for positive selec-
from the thymus alive. tion to CD4 fates and TCR interactions with class I MHC

usually direct selection to CD8 fates, this expression pattern that will populate the thymus in later waves. Descendants
makes it possible for the cortical epithelium to support posi- of early intraembryonic hematopoietic precursors, they
tive selection to both CD4 and CD8 fates. At the cortical/ have a unique capability to give rise to certain specialized
medullary junction, there is a high concentration of thymic types of TCRγδ cells as well as more conventional kinds
macrophages, which can rapidly dispose of the dead cells of TCRγδ cells and TCRαβ cells.12,13 Second, the thymus
that are generated in large numbers by TCR-dependent epithelium itself is naïve in the sense that it has not yet
selection every day. been affected by signals from lymphocyte gene products,
Medullary thymic epithelium is distinctive in an ad- nor dilated into an open mesh by lymphoid proliferation.
ditional way.9a Not only does it express class I and class II The medulla is not evident yet as a distinctive domain, and
MHC, but also it is specialized for the detection and elimi- the spatial distribution of cortical epithelial products like
nation of highly autoreactive cells. Medullary epithelial IL-7 and Notch ligands is not necessarily ordered as it will
cells mature to an end stage in which they lift the repres- later become.14,15
sion of multiple silent loci in their genomes, genes that have One result is that differentiation occurs much faster in
no function in the thymus per se and would normally be the fetal mouse thymus than in the adult. Passage from DN1
expressed only in other organs in the body. These diverse to DN3 stage occurs within ≤ 2 days from embryonic gesta-
tissue-specific self-antigens, presented on class I or class II tion days E12.5 to E14.5 instead of 7 to 10 days in adults.16,17
MHC, are offered as “bait” to any newly positively selected Passage to DP occurs just 2 days later, requiring about
cells whose receptors might recognize them. The presen- 4 days instead of ∼2 weeks overall.3 The fetal-specific types
tation of these molecules is further enhanced by special of TCRγδ cells are in fact produced as the first wave of ma-
populations of dendritic cells (DCs) in the thymic medulla, ture T cells, by E15.5, even before DP cells emerge. Finally,
and these help to stimulate any autoreactive T cells to com- the gene expression requirements for T-cell development in
mit suicide. As we will see in the following sections, these the fetal thymus are subtly different than in the adult (eg, a
medullary cell types can also suppress autoreactive T-cell more stringent requirement for transcription factors Ikaros
precursors through an alternative mechanism, by directing and E2A, and a reduced requirement for transcription factor
their differentiation into nTregs. TCF-1 and the IL-7R).
It is interesting that there is generally reduced TCR
sequence diversity in the fetal T cells as compared to those
Variations of T-Cell Development: Ontogeny and made after birth. Normally, TCR sequences have two sources
Species Differences of variation: the combinatorial diversity based on the par-
T-cell development in the thymus is common to all jawed ticular V, D, and J segments that are rearranged, and the
vertebrates.10 A thymus, originating embryologically in sloppiness of cleavage and addition of untemplated nucleo-
the neck region and presenting Notch ligands to RAG1/ tides at the recombination breakpoints (ie, junctional diver-
RAG2 + lymphocyte precursors that turn on TCR gene ex- sity). One enzyme, terminal deoxynucleotidyl transferase
pression, is found from cartilaginous fish to bony fish and (encoded by Dntt), is responsible for much of the junctional
amphibia through birds and all mammalian species. There diversity. It is normally turned on in the earliest stages of
is even an apparent equivalent structure newly discovered prethymic precursor development and expressed through-
in the lamprey, a jawless vertebrate that probably shared out T-cell development until positive selection. However,
its last common ancestor with mammals shortly after the the waves of lymphoid precursors developing in the fetus
Cambrian explosion (∼500 million years ago).11 Despite do not express it, and this situation only changes around
important commonalities, there are significant differences the time of birth. As a result, some TCR types expressed in
between T-cell development in mice and in other verte- fetal thymocytes including the “first wave” TCRγδ recep-
brates, and even between mice and other mammals. In addi- tors are essentially invariant, with highly predictable read-
tion, even within the mouse, T-cell development undergoes ing frames, even though the genes that code for them are
significant changes between the first wave of production that assembled by RAG-mediated recombination. It seems likely
occurs in the fetus and later waves of T-cell development in that these TCR specificities have some evolutionary selective
fetal and postnatal life, described subsequently. Thus aspects advantage for the young mouse.
of T-cell development can be flexible even though they pro- The production of TCRγδ cells as the initial wave in the
duce similar T-cell populations as outputs. fetal thymus is remarkably shared between different classes of
vertebrates: one of the initial reports of first-wave TCRγδ cells
Ontogeny: Distinctive Features of the Mouse (then called “TCR1 cells”) was in chickens.18 In the mouse,
Embryonic Thymus there is considerable evidence that the first-wave cells are dis-
When the thymus rudiment is first formed in midgestation tinctive in more ways than their use of a particular TCRγδ
in the developing neck region of the embryo, it is not vascu- receptor complex, as discussed in the section on γδ cells.
larized. Thus, the first wave of hematopoietic precursors that
enter the thymus do so by migrating across the mesenchyme Evolutionary Conservation of Diverse T-Cell Lineages but
and penetrating the thymic rudiment from the outside. Flexibility of Roles
There are two distinctive things about this fi rst wave Focus on the mouse system may in fact handicap our abil-
of thymic lymphopoiesis. First, the precursors of the fi rst ity to perceive the roles of TCRγδ cells, because mice and
wave are intrinsically somewhat different from the cells humans are the two known mammalian species with the

most limited systemic use of TCRγδ cells. In fact, an αβ /γδ on TCRαβ as compared to TCRγδ cell production.23 While
distinction between T-cell subclasses is an ancient feature of most of the major features of development are similar, this
vertebrate lymphopoiesis. Even cartilaginous fish like skates kind of variation is a caveat against overgeneralizing data
and sharks have well-defined γδ and αβ TCR complexes that from the mouse.
are used by T cells in different anatomical domains.19 This
suggests that two TCR-defined branches of T-cell develop-
ment have been coselected for persistence over > 400 million
Creating an In Vitro Context for T-Cell Development
years of vertebrate evolution. Major divisions within the The three-dimensional structure of the thymus was for
TCRαβ lineages, between CD4 + helpers and CD8 + killers, many years a challenge to the full defi nition of T-cell
are similarly conserved among all bony vertebrates. Thus developmental mechanisms. It was clear after many at-
although the genes for immunological receptors themselves tempts that dissociated thymic epithelial cells could not
undergo rapid evolutionary change and reorganization, the reconstitute thymic function if plated in monolayer cul-
underlying programs for lymphocyte development appear ture, and thus it seemed that the migration of precursors
to be ancient and conserved. through organized three-dimensional domains would be a
crucial aspect of T lymphopoiesis. Redox conditions also
Evolutionary Flexibility at Close Range: seemed crucial for thymic function. There were two early
Mouse versus Human breakthroughs in experimental dissection of T-cell de-
Most comparative information is only available for mouse velopment. First was the generation of fetal thymic organ
and human T-cell development. The outlines of T-cell de- cultures, which were found to require a high oxygen con-
velopment in the thymus are strikingly similar in these centration to work, either by plating a thymic lobe at an
closely related mammalian species.20–22 However, many air-medium interface, or by pumping up to 70% oxygen
details of T-cell development have proven to be different into the incubator in which a submerged fetal thymic lobe
between man and mouse. The cell surface markers that de- was cultured.24,25 Second was the discovery that in these
fine stages of human T-cell development are different from excellent culture conditions, even dissociated, purified
their mouse counterparts, and even some of the same mol- thymic stromal elements could regenerate a competent
ecules are expressed in different developmental patterns, as thymic microenvironment if mixed together in a reaggre-
shown in Table 13.1. One surprising difference concerns the gate fetal thymic organ culture. The development of reag-
quantitative influences of different doses of Notch signaling gregate fetal thymic organ culture has allowed the separate
roles of distinct epithelial and hematopoietically derived
elements of the cortical and medullary environments in
positive and negative selection to be rigorously demon-
strated, as discussed in later sections.26 Both fetal thymic
organ culture and reaggregate fetal thymic organ culture
13.1 Comparison of Markers in Human
TABLE systems promote T-cell development from prethymic pre-
and Murine T-Cell Development
cursor stages all the way through positive and negative
Stage Mouse Phenotype Human Phenotype selection to generate functional mature T cells.
The fetal and reaggregate thymic organ cultures both
ETP-like: CD44+ c-Kit+ CD34+ CD38lo CD7− make use of natural thymic stromal elements, which makes
Intrathymic CD25− CD1a−
them close to physiological conditions, but also a logistical
T/DC/NK
precursors challenge. Furthermore, the organ cultures still re-form into
Specification stage CD44+ c-Kit+ CD34+ CD38+ CD7+ closed structures, making the thymocytes developing within
(DN2-like) CD25+ CD1a− them hard to count, track phenotypically, or experimentally
Committed CD44+/low c-Kitint CD34+ CD38+ CD7+ manipulate. Finally, these organ cultures remain very small,
T precursors CD25+ CD1a+ hampering use for molecular biology.
TCRδ, γ, β CD44+/− c-Kit+/− CD34+ CD38+ Against this background, the development of stromal
rearrangement: CD25+ (DN2, CD1a+ CD4−/+
monolayer coculture systems for T-cell differentiation has
αβ/γδ DN3)
precursors revolutionized the field. Two lines of research converged
β-selection CD44− c-Kit− CD34+ CD38+ to make this possible. First, Nakano, Kodama, and Honjo
CD25+ CD4- CD1a+ CD4+to developed OP9 bone marrow stromal lines from myeloid
CD8− (DN3) to CD4+ CD8α+ colony stimulating factor-deficient mice (Op, osteopetrotic
CD25− CD4− (CD4 ISP, mice),27 which proved to be ideal for supporting lymphoid
CD8− early DP) development. However, only B and NK lymphocytes, not
TCRα CD25− CD4+ CD4+ CD8α+ CD8β+
T cells, develop on these original OP9 stromal cell lines.
rearrangement CD8αβ+ (DP) (early DP to DP)
Mature cells TCR/CD3hi and CD1a− TCR/CD3hi Second, crucial experiments by the groups of Freddy Radtke
CD4+ or and CD4+ or and Warren Pear showed that environmental activation of
CD8αβ+ CD8αβ+ Notch signaling is crucial to induce hematopoietic precur-
sors to become T cells.28,29 Engineering OP9 cells to express a
CD, cluster of differentiation; DC, dendritic cell; DN, double negative; DP, double posi-
tive; ETP, early T-cell precursor; NK, natural killer cell; T, T cell.
potent Notch ligand, Schmitt and Zuniga-Pflucker developed
Data from reviews by Blom & Spits283 and Taghon & Rothenberg.22 the OP9-DL1 stromal line to support T-cell development.30

This created robust, highly efficient, high-yield conditions Notch signaling accomplishes this large-scale transcrip-
for driving and supporting T-lineage development in vitro tional mobilization by first activating the expression of sev-
from a wide range of uncommitted hematopoietic progeni- eral T-lineage transcription factors that can help the cells
tor types, while keeping the cells accessible, easily moni- to open up previously silent genetic loci.32–34 The most im-
tored, and easily transferable throughout. portant T-cell–specific members of this group are GATA-3
The OP9-DL1 system and variations of it have made (encoded by the Gata3 gene) and TCF-1 (encoded by the
possible enormous advances in understanding of the early, Tcf7 gene). These T-cell factors collaborate with the Notch-
Notch-dependent stages of T-cell development. This sys- activated transcription factor RBPJκ (also called CBF1 –
tem exposes to experimental dissection all the stages from suppressor of hairless – Lag-1 = CSL), and together they
initiation of the T-cell developmental program through interact with a set of transcription factors that is already
commitment and up to the DP stage (ie, all the stages present and active in multilineage hematopoietic progeni-
that DN precursors would normally undergo during their tors. These preexisting factors include “E protein” basic
“outward bound” migration through the thymic cortex). helix-loop-helix factors E2A (encoded by Tcf3 = Tcfe2a)
It has also opened up the ability to track and experimen- and HEB (Tcf12), Ikaros family members like Ikaros (Ikzf1)
tally perturb early stages of human T-cell development, and Helios (Ikzf2), Runx family factors like Runx1 and
starting from cord blood precursors, in a direct parallel to Runx3, Ets family factors including PU.1 (SPI1 in human,
early mouse T-cell development.23 However, the “inward Sfpi1 in mouse) and Ets1, and other key hematopoietic fac-
bound” phases of T-cell development after β -selection or tors like Myb and Gfi1.35 The knockout of any one of these
γδ -selection appear to depend on special properties of thy- factors or factor families is sufficient to abort early T-cell
mic epithelial cells that remain to be duplicated by OP9-DL1 development.
or any stromal coculture. This complexity is important to acknowledge because de-
spite the essential role of Notch, no one transcription factor
alone turns on the whole T-cell program in a single stroke.
EARLY LINEAGE CHOICES AND COMMITMENT In fact, different factors collaborate with each other in a
Notch and the Specification of T-Cell Identity balanced way throughout the DN stages, to coordinate dif-
Notch signaling is crucial for induction of the T-cell pro- ferentiation with the right degree of proliferation before the
gram, for progression from stage to stage of the specification cells reach the DN3 stage. Furthermore, the E proteins E2A
process, and for maintenance of viability of the developing and HEB not only regulate T-cell identity genes, but also
T cells throughout the DN stages, all the way to β -selection. serve throughout T-cell development to enforce TCR signal-
DN thymocytes gain privileged access to Notch ligands in ing checkpoints, undergoing cycles of inhibition and reacti-
the environment despite the large excess of DP cells. This is vation in response to TCR stimulation. We will see others of
because the DP cells “deny themselves” the chance to bind these factors that also have different, dose-dependent roles
efficiently with Delta-class Notch ligands in the environ- later on, working with or against each other to cause one
ment by downregulating key Notch modifier genes, Lfng T-cell lineage to diverge from another.
and Mfng, during β -selection,31 which are needed to com-
pete for stromal DLL4 binding. Thus, in a healthy normal
adult thymus, the masses of DP cells actually insulate indi- Options, Precursors, and Stages
vidual DN cells from competition with each other, allow- The cells that populate the thymus are not yet committed
ing Notch triggering to be renewed repeatedly from ETP to to a T-cell fate when they arrive. Elegant single-cell culture
DN3 stage. experiments using either adapted organ culture systems or
Notch-DLL4 signaling activates the regulatory cascade stromal coculture systems show that the same individual
that turns on T-lineage genes from ETP stage to DN2 stage cells that can give rise to T cells would still be able to gener-
to DN3 stage. T-lineage–specific signaling mediators and ate NK cells, myeloid cells, and DCs under other conditions,
invariant TCR components are upregulated starting by the when they arrive in the thymus.36–39 Several lines of research
DN2 stage or in the DN2-DN3 transition. The IL-7R (Il7r also imply that many of them are initially able to give rise to
gene) is fully turned on by DN2 stage to support prolif- B cells.40,41 The commitment process is the process through
eration of the precursors as they are acquiring their T-cell which the developing T cells decide permanently against
identity. Between DN2 and DN3 stage, RAG1 and RAG2 these other options.42
recombinases and multiple TCR signaling complex compo- Given how similarly T and B cells rearrange their immu-
nents and mediators of TCR signal transduction, including noreceptor genes and observe developmental checkpoints,
the kinases Zap70 and Itk, are all induced or upregulated. it is not surprising that B-cell potential should be present
The specialized surrogate T-cell receptor α-like chain, pre- in T-cell precursors. It is notable how early B-cell potential
TCRα (pTα , encoded by the Ptcra gene), is also turned on is lost—soon after the thymus-settling precursors enter the
to provide a partner for complex formation with any suc- thymus; many experiments even imply that access to the
cessfully-encoded TCRβ chain. Thus, the cells that reach the B-cell fate can be lost even before thymic entry.43–45 However,
DN3 stage are already fully armed with substantial levels of after losing the B-cell option, DN thymocytes maintain a
the T-cell specific components that will be needed to trans- clearly demonstrable ability to shift to a myeloid fate, either
duce signals through the TCR, as soon as they have made a macrophage or granulocyte, and this persists into the DN2
productive TCR gene rearrangement. stage.37–39 As long as the cells remain in an intact thymus,

these possibilities remain latent and development progresses potential up to this point. Under normal conditions, how-
overwhelmingly toward a T-cell fate, but all that is needed to ever, the thymus is not a good source of myeloid-cell growth
reveal these options is to remove the cells from the thymus factors or differentiation factors, and so myeloid develop-
and offer them cytokines that support myeloid cell survival. ment is a low-efficiency path in the thymus regardless of the
Even into the DN2 stage, some T-cell precursors also appear programming of the cells. In addition, active Notch signal-
to be able to develop into mast cells, yet another nonlym- ing appears to add a further conditional obstacle to myeloid
phoid cell type.46 The cells that start the T-cell program thus development through interference with the action of my-
begin with many alternative possibilities. eloid transcription factors.52,55 This mechanism still allows
A great deal of work has been carried out to link the de- cells to keep myeloid potential as a latent option through
velopmental repertoire of newly arrived thymic immigrants several cell divisions, accessible if they are removed from the
with uncommitted cells in the bone marrow that are likely thymus. But eventually the T-cell developmental program
to be their precursors. This lively field of research has shown reaches a stage when the myeloid-enabling transcription fac-
that thymic immigrants can be derived not only from cells tors themselves are repressed. This occurs in the later DN2
that were mostly committed to some kind of lymphoid fate stage, and it marks the point when the myeloid paths too are
prethymically (ie, common lymphoid precursors), but also permanently cut off.
from cells that were broader-spectrum lymphoid-myeloid The most persistent fate alternative is the NK fate, which
multilineage precursors (lymphoid-primed multipotent falls into a somewhat different category. The choice against
precursors).43,47–50 Furthermore, even cells that have already becoming an NK cell appears to be made about the same
begun to be restricted to a nonlymphoid, myeloid pathway time as the choice to forgo the DC fate option,36,56,57 but in
(granulocyte-macrophage precursors) can also develop into fact one could argue that many CD8 + T cells never fully give
T cells, if transferred from the bone marrow into a thymic up the NK possibility. There are many differences between
environment or OP9-DL1 coculture.51 Many T-cell precur- NK cells and T cells: most obviously, the NK cells do not
sors in the thymus in fact show evidence of having earlier rearrange their TCR genes and utilize a substantially differ-
activated a myeloid lineage differentiation gene, lysozyme ent antigen recognition system than T cells. Thus, any TCR +
M.52 Thus, the dominant developmental program imposed cell would be classified as T rather than NK no matter how
by the thymic environment is a kind of boot camp that can NK-like it was in other respects. In an apparent dichotomy,
force cells of different developmental origins to converge on too, developing T cells require a class of transcription fac-
a common T-cell fate, erasing signs of their previous distinct tors, the E proteins (E2A, HEB) that NK cells must inhibit:
identities. the E protein antagonist Id2 is an NK developmental re-
quirement. However, NK cells mobilize effector programs
that are almost completely congruent with those of effec-
Mechanisms of Commitment tor CD8 + killer T cells. On persistent antigen stimulation,
The thymus uses different mechanisms to eliminate vari- CD8 killer T cells turn on expression of NK-cell activation
ous fate alternatives for the cells. One of these is the envi- receptors as well as TCR and behave even more like NK cells
ronmental signal through Notch itself. As long as thymic in terms of promiscuity of killing. A recent gene knockout
immigrants and even ETP and DN2 cells are being exposed model suggests that only one transcription factor, Bcl11b,
to Notch ligands in the thymic microenvironment, they may stand between mature CD8 cells and NK-like behavior
cannot exhibit any alternative potentials efficiently. As dif- normally throughout their lifetimes as mature T cells.58 The
ferentiation progresses, however, the cells also give up the loss, or reduction, of access to the NK program occurs at the
ability to access these alternatives under any conditions specific stage of T-cell development, also in the DN2 stage,
(Fig. 13.4). when Bcl11b is turned on for the first time.58–60
Early B-cell development in particular is intensely inhib- Bcl11b is turned on in a process that requires Notch sig-
ited by Notch signaling. Not only are precursors prevented naling and TCF-1,61,62 possibly restrained until DN2 stage
from initiating B-cell development in the presence of Notch- by other timing factors. DN2 cells that cannot turn on
Delta signals, but also exposure to Notch signals rapidly Bcl11b do not reach a normal DN2b or DN3 stage: instead,
strips them of the capability to enter the B-cell pathway even they keep prolonged access to myeloid fates and ongoing
if Notch-Delta signaling is removed.40 Perhaps because early self-renewal programs as well as the NK program as long
B and T cells share many other early developmental require- as cytokine levels permit.59,60 They are profoundly defec-
ments—the cytokine IL-7, the role of stroma, prominent use tive in generating TCRαβ -lineage thymocytes, DP, and
of E proteins among their crucial lymphoid transcription later stages, as well.63,64 Thus, in addition to its special
factors—the Notch signal is uniquely important to make role in suppressing the NK-cell program, Bcl11b also ap-
these two developmental programs mutually exclusive. The pears to work in a gatekeeper capacity at the crux of the
thymus is too good an environment for B-cell development commitment process.
in every other way; in fact it fi lls up with developing B cells Commitment, therefore, involves at least three distinct
if Notch-DLL4 signaling is defective.41,53,54 mechanisms: the suppression of B-cell potential, the silenc-
The choice against other alternatives takes longer and ining of myeloid transcription factors, and the induction of
volves more conditionality. The myeloid and DC-enabling Bcl11b. The last two of these happen roughly at the same
transcription factor PU.1 is normally expressed into the point, in a transition that is now recognized to split the DN2
DN2 stage and probably explains the access to myeloid stage between uncommitted DN2a cells and committed

Phase 1 pro-T Phase 2 pro-T
Environmental
Notch-DL
signaling
β-selection
CLP DN1/
Cell type LMPP ETP DN2a DN2b DN3a DN3b
αβ T cells
Developmental
potentials Thymus γδδ T cells A
100%
T potential
Myeloid/DC NK potential
potential
B potential
B
Gene expression
1000
Bcl11b
100 TCF-1
GATA-3
10
PU.1
1
C
FIG. 13.4. T-Lineage Commitment and Expression Patterns of Contributing Factors. A: Prethymic and intrathymic stages involved in T-lineage
commitment. The thymus can be seeded by cells that retain myeloid potential (lymphoid-primed multipotent precursor) as well as by cells that
have already reduced their access to this option (common lymphoid precursor). As shown in the figure, Notch-Delta signals are delivered
to the cells repeatedly from their time of entry into the thymus until β-selection. B: Approximate times at which distinct cell-lineage options
for T-cell precursors are excluded. Y axis represents remaining ability of T-cell precursor cells to enter alternative pathway at the indicated
stages (relative units). C: Expression patterns of key regulatory genes noted in the text. Factors used in the T-cell program are depicted with
solid lines, while PU.1, a factor used in multipotent progenitor, myeloid, and B-cell programs, is shown with a dotted line. Adapted from figures
in Rothenberg42 and Rothenberg et al.279 Evidence for gene expression patterns reviewed in Rothenberg et al.32
DN2b cells56,65 (see Fig. 13.4). DN2a cells have access to other ligands until they have successfully undergone β -selection
fates and can survive when removed from Notch ligand- or γδ -selection. Transition to the DN2b stage also appears
rich environments; DN2b cells are committed and become to prepare the way for shutting down precursor expansion
acutely Notch signaling–dependent for viability. From com- and establishing the molecular conditions for enforcing the
mitment on, DN cells cannot survive deprivation of Notch β -selection checkpoint.

PROTOTYPE OF A T-CELL RECEPTOR–DEPENDENT mediate spontaneous clustering that may help to trigger
CHECKPOINT: a-SELECTION signaling.67–69 The rate-limiting criterion for successful
β -selection is thus simply the structural integrity of the
a-Selection: Proliferation, Allelic Exclusion, and
TCRβ chain that allows it to participate in a well-formed
Creation of the Double Positive State
signaling complex.
All TCRαβ cells, regardless of functional subset, are sur-
vivors and products of β -selection. The β -selection check-
point is the fi rst TCR-dependent control checkpoint that The a-Selection Response
these cells encounter, and the β -selection process is the Pre-TCR spontaneous signaling triggers a dramatic cascade
fi rst TCR-dependent signaling response that they un- of events. In hours, the cells not only begin to proliferate
dergo. The outcome shapes the whole population upon again but also set in motion a profound transformation.
which later selection and differentiation mechanisms The cells become large blasts, almost 10 times the volume
must operate. of most thymocytes. As they turn on CD4 and CD8αβ ex-
TCRβ rearrangement and TCR-complex signaling are pression, and express the coreceptor CD28 for the fi rst time,
crucial for both the proliferation and differentiation events they shed their DN cell markers and everything that had
of β -selection. Thus recombinase-deficient (Rag1 or Rag2 sustained their viability as DN cells, losing CD25, IL-7R,
knockout), DNA repair-deficient (eg, Prkdcscid, “SCID”), and the last vestiges of Kit expression and also inactivating
TCRβ constant region knockout, or CD3γ or ε knockout the Notch pathway as well.
mice all leave T-cell development blocked at DN3 stage (see The proliferation unleashed by β -selection is intense.
Fig. 13.2). In normal development, successful TCR Vβ join- In their 2 to 3 days of proliferation, the selected DN3 cells
ing to Dβ -Jβ joints in a productive reading frame is the generate approximately 102-fold more DP cells. During this
limiting event, as CD3 chains begin to be synthesized and expansion, there are ample opportunities for clearing the
expressed on the surface at a low level irrespective of rear- cell surface of obsolete membrane proteins (eg, old CD25
rangement. It is the ability to deliver a signal through these molecules), clearing the nucleus of obsolete transcription
complexes that really drives β -selection, as both in vivo and factors (eg, the Notch-induced transcription factor Hes-1),
in vitro a response that strongly mimics natural β -selection and clearing the genome of obsolete states of chromatin
can be triggered even in RAG-deficient thymocytes if mono- modification.
clonal antibodies are used to crosslink the CD3 chains on During the proliferation triggered by β -selection, TCR
their surfaces to deliver an artificial signal. gene rearrangement is temporarily suspended, as the cells
downregulate Rag1 and Rag2 gene expression and inacti-
Fulfilling the Requirements for a-Selection vate RAG2 protein by cell cycle-dependent phosphoryla-
By the DN3 stage, developing T cells are already under tion.70 RAG1 and RAG2 expression is restored by the time
proliferative restraint, in part to make further develop- proliferation arrests in the DP stage, but by that time the
ment conditional and in part to enable the RAG1/RAG2 cells have transformed the chromatin landscape in which
complex to work efficiently. Although many ETPs already the RAG1/RAG2 recombinase must find its targets. TCRβ
have limited Dβ -Jβ TCR gene rearrangements due to is now closed, establishing rigorous allelic exclusion for this
low-level prethymic RAG expression, 66 these partial rear- TCR chain. TCRα is now open for rearrangement, pioneered
rangements cannot encode functional receptors. By the by some germline transcription as early as the late DN3
DN3 stage, however, efficient recombination of V β seg- stage. At the same time, the potential for expressing TCRγδ
ments to the joined Dβ -Jβ products begins, and some receptors is eliminated. TCRγ genes become inaccessible, so
of these rearrangement products now encode functional that both germline transcripts from unrearranged loci and
TCR β chains. transcripts from any previously rearranged TCRγ genes are
The success of TCR β rearrangement is sensed by the silenced.71–73 TCRδ genes, whether previously rearranged or
cell when the newly made β -chain is assembled into a not, are deleted as the TCRα segments that straddle them
complex at the cell surface and transduces back a low- rearrange. Thus, by the time proliferation stops, the DP cell
intensity signal through essentially the same kinase and is irreversibly confirmed in a TCRαβ lineage.
adaptor mechanisms used in mature TCR signaling.
Isolated TCR β chains cannot assemble with CD3 and
TCRζ chains on their own, and there is no TCRα available Creation of the Double Positive State: Setting the
yet, but the invariant pre-TCRα chain provides a surro- Stage for Positive and Negative Selection
gate. The Ptcra gene does not require rearrangement and is With the first productive rearrangement of TCRα , the de-
turned on strongly as a direct Notch signal–induced gene veloping DP cell will acquire the capacity to recognize un-
by the DN3 stage. The pre-TCRα /TCR β heterodimer thus predictable antigenic targets, including self-antigens. Until
forms a “pre-TCR” containing a full complement of CD3 the new receptor specificity has passed quality control, a DP
and TCRζ , which is readily transported to the cell surface cell is therefore functionally disabled and “contained,” both
to trigger signaling. within the thymic cortex and within a short default lifes-
Unlike all later uses of the TCRβ, this signaling does pan. DP cells not only lack emigration receptor expression
not depend on any component of ligand recognition from but also acquire features that would cause them to be killed
its heterodimer partner, but the pre-TCRα chain can quickly if they did escape. As they pass through β -selection,

they downregulate their own MHC class I molecules and thymocytes rather than robustly activated effector T cells
their surface glycoprotein sialylation74 pathways: they can must be explained as the result of the characteristic tran-
even be purified on the basis of their resulting low MHC scriptional and epigenetic regulatory state of the cells in this
class I expression and preferential binding by the lectin pea- stage—the context within which the signaling pathways are
nut agglutinin. These changes make them into potential NK sensed—rather than a unique biochemical pathway for the
cell targets and ensure that they would be scavenged by mac- signals themselves.
rophages via asialoglycoprotein receptors if they escaped.
Also as β -selection begins, the antiapoptotic factor Bcl2 is
repressed, replaced with the weaker or more conditional A Clear and Present Danger: Creation and
survival factors Bcl-xL and Bcl2A1.75,76 Even while they are Enforcement of the a-Selection Checkpoint
still rapidly dividing, the emerging cells acquire an intense The cell biology of the β -selection response is an obvious
vulnerability to death signals, both glucocorticoid signaling hazard for the organism. Intense polyclonal prolifera-
and other strong signals.77 tion of this magnitude brings the cells close to a malig-
Cell surface expression of TCRαβ complexes are kept nant state, and the program must include brakes to stop
low on DP cells even when TCRα chains are first expressed, the proliferation as well as constraints to prevent it from
partly due to uniquely high levels of Src-like adaptor protein being triggered inappropriately. In fact, most T-cell acute
(Sla gene product), which targets TCRζ protein for rapid de- leukemias have precisely the phenotype of cells that can-
struction.78 Nevertheless, DP thymocytes are, if anything, not stop the β -selection response, continuously generat-
more sensitive to TCR-ligand interactions than mature ing immature DP cells or DN-DP intermediates. Under
cells with the same TCRαβ if an appropriate assay is used. normal conditions, a limiting factor for the process is the
Contributing factors may include their relative lack of sialic programmed downregulation of Notch responsiveness, as
acid residues, thus reducing electrostatic repulsion, and their the extent of population expansion during β -selection is
low expression of the negative feedback molecule CD5 that strongly influenced by Notch signaling.102,103 Indeed, by far
could otherwise damp their response to TCRαβ signals.79–82 the most dominant and consistent feature of T-cell acute
Even though it is sensed, TCR signaling cannot turn on lymphoblastic leukemias is the presence of gain of function
functional response genes like Il2 in DP cells. In part, this mutations in the Notch1 locus, especially mutations that
is because these cells have downregulated key activation- make Notch signaling ligand-independent.104 The result is
transducing transcription factors, c-Jun and c-Rel (for more particularly grave when the cells can join viability signals
information, see www.immgen.org/).83 β-selection also in- from pre-TCR complexes and Notch alike to proliferate
duces expression of the transcription factor, RORγ, from a without restraint.
distinctive promoter (RORγ t, also called RORC2), which Another dangerous combination is if the very rapid pro-
actively blocks conventional effector responses even while liferation program induced by β -selection is superimposed
maintaining Bcl-xL expression throughout the DP stage.84,85 on any continuing expression of hematopoietic progenitor
RORγ t expression sets DP cells apart from all other major self-renewal genes, inherited from earlier stages of T-cell
classes of thymocytes; however, it provocatively links them specification, including SCL (Tal1), Lyl1, Lmo1, Lmo2, and
with Th17 cells in the periphery, with lymphoid tissue– Hhex.105,106 These genes are normally turned off during the
inducing cells that organize lymph nodes in fetal life, and lineage commitment transition, between DN2a and DN2b,
certain classes of newly recognized innate effector cells.86–89 but may easily act as T-cell oncogenes if kept on. Thus,
The DP cells that emerge from the complex regulatory events β -selection may not be safe until these genes have been suc-
of β -selection thus acquire a highly specialized physiology, cessfully repressed. The implication is that progressing into
and they are poised for selection. a normal DN2b-DN3 stage sequence may be important to
The β -selection response is driven by the same major sig- establish the correct control over future growth.
naling pathways used in the activation of mature T cells. Lck, Nonmalignant αβ T-cell development depends on a
Ras pathway, PI3-kinase, and protein kinase C activation all robust mechanism that can enforce a clear developmental
play roles in the process, using virtually all the same adap- arrest checkpoint, before the cells have a chance to receive
tor molecules [LAT, Slp76 (Lcp2), GADS (Grap2)] that are the pre-TCR signal that triggers β -selection. The two most
used to coordinate signaling in positive selection later and in important regulators known for this process are transcrip-
mature T cells.90–92 As in later stages of T-cell development, tion factors: E proteins (E2A and/or HEB) and Ikaros.
TCR signaling temporarily antagonizes effective activity of Both types of factors are expressed fairly stably throughout
E proteins (E2A/HEB). A transient wave of immediate-early early T-cell development, but they become indispensable
Egr factor activation results in transient upregulation of the for enforcement of the β -selection checkpoint in addition
E protein antagonist Id3, a response that will be echoed in to their other roles. E proteins promote Notch1 and Rag
positive selection later.93–96 As the cells reach the resting DP gene expression as well as expression of a number of T-cell
state, high-level E protein activity is restored.97 These factors genes, but they also tend to slow or stop proliferation.107
cooperate with some of the same regulatory factors that are Ikaros is important for the initial production of T-cell pre-
used in early T-cell development: initially Notch signaling, cursors but may also serve as a competitor for Notch/RBPJ
then the important T-cell transcription factor TCF-1.98–101 at certain common DNA target sites.108 Losses of function
The fact that the cells emerging from this signaling response of either class of checkpoint enforcer lead to malignancy
are idiosyncratic, functionally disarmed, short-lived DP with very high efficiency.

THE FIRST T-CELL RECEPTOR–DEPENDENT the mouse TCRγδ complex comes from rearrangements of
PROGRAM CHOICE: à VERSUS fc FATES TCRVδ, Dδ, and Jδ segments. However, because all the TCRδ
Making T-Cell Receptor–dependent Developmental elements are sandwiched into the TCRα genomic locus be-
tween the Vα segments and the Jα segments, any TCRα rear-
Decisions: General Questions, Different Answers
rangement on the same chromosome would delete the whole
The problem of how TCR recognition specificity becomes TCRδ complex. Thus γδ cells need both to rearrange δ suc-
linked to irreversible activation of different transcriptional cessfully and to suppress any rearrangement of the TCRα
programs is the core issue of T-cell development. For most locus, immediately and permanently. Second, a special “en-
developing T cells themselves, this problem is first encoun- hanced allelic exclusion” mechanism is needed to establish
tered during the decision to become a TCRαβ cell or a TCRγδ expression of a unique TCRγ in a given T cell, because there
cell. As they mature, both TCRαβ cells and TCRγδ cells fan are actually three different “subloci” that can serve as rear-
out into a variety of distinct effector cell types or committed rangement sites to encode TCRγ in the mouse, albeit closely
lineages. But TCR specificity most commonly emerges as the linked on the same chromosome. Each sublocus has its own
result of stochastic processes that determine not only which Jγ and Cγ that can serve as a rearrangement target for its own
V, D, and J segments will be joined but also exactly which dedicated Vγ segment or segments. Two of the subloci, those
nontemplated sequences will happen to be added at the cod- containing Cγ4 and Cγ 2, each “own” only a single V segment
ing joints, as the latter end up determining antigen-binding and a single J segment, so that the only source of diversity in
specificity as well as whether the rearrangement is in-frame. their rearrangements is junctional. The third sublocus, the
How can the cells coordinate any coherent transcriptional one containing Cγ1, also has only one Jγ but can rearrange it
regulatory programs for development together with such a to any of four Vγ ’s. In order to establish a unique TCRγδ, the
randomly determined outcome? ordering of rearrangements at each of these distinct subloci
Most models that have been considered as possible an- must be controlled so that only one yields a productive rear-
swers include different balances between “instruction” and rangement. This split locus organization also makes it note-
“selection.” Instructive models propose ways that the ran- worthy that TCRγ RNA expression from any of these loci is
domly determined TCR specificity can result in transduc- repressed if the cell should adopt a DP fate instead.71–73
tion of distinct signals back to the cell that actually direct
specific developmental choices. Selective models postulate Special Role for Interleukin-7R Signaling
that developmental choices are made independently, but One major regulatory unifier for the TCRγ complex is the
that the randomly determined TCR specificity may be more positive input all its subloci appear to require from IL-7R
“appropriate” for one choice than another. Then, the signals signaling. In mutant mice defective in IL-7 or IL-7R, one of
transduced back from the TCR need make only a simple de- the first abnormalities noted was their profound defect in
termination: whether or not the cell should be allowed to TCRγδ cell generation. While IL-7R signaling contributes
live or die. As described in the following sections, there is to survival and expansion of αβ -lineage cells generally, it
substantial evidence for roles of both kinds of mechanisms also has an indispensable role in specification of γδ -lineage
in later TCRαβ cell development. cells. At least part of the molecular explanation is that IL-7/
However, with reference to the TCRαβ versus TCRγδ de- IL-7R-activated Stat5 transcription factor is needed to bind
cision, a third kind of model needs to be considered as well. to regulatory sites in the TCRγ locus, to open the gene seg-
The selection of TCR gene segments to be rearranged in this ments for rearrangement.110–113
case is not random; it is itself under significant developmen-
tal and environmental control.109 Specific aspects of T-cell fc Selection
programming not only influence which TCRγ segments
are used, but also may influence whether a cell will choose Like β -selected cells, TCRγδ cells are selected by TCR com-
TCRγ and TCRδ gene segments to rearrange at all, before or plex triggering within the DN2-DN3 stages. Some emerge
in preference to TCRβ gene segments. Thus, developmental from precursors that had also opened their TCR Vβ loci for
programming can directly influence TCR specificity in this rearrangement and might thus have been β -selected instead.
case. The effect of preprogramming must be considered to- Under some conditions, as described subsequently, there
gether with the effect of TCR specificity to explain the ways even seems to be some overlap, with TCRγδ complexes oc-
that β -selection and γδ -selection, in various versions, can casionally being able to trigger a low-level β -selection re-
diverge from each other. sponse, and some “TCRγδ lineage” cells revealing a cryptic
productive TCRβ rearrangement as well.114 Both the eligible
cells and the signals that trigger the two kinds of selection
Cytokine-dependent and Stage-dependent Access to responses can therefore be extremely similar. However, the
the T-Cell Receptor-fc Loci fate of cells undergoing TCRγδ -selection is considerably
different from that of cells undergoing β -selection, and this
T-Cell Receptor-f and T-Cell Receptor-c Gene Loci remains only partially explained.
The ability to make a successful TCRγδ complex actually
depends on the regulated status of rearrangements at as The fc-Selection Program
many as five distinct TCR J-C target loci in the mouse, be- The consequences of γδ -selection for rearrangement and
cause of the structure of the TCRγ and TCRδ gene complexes expression of different TCR loci are reversed from those
(Fig. 13.5A, B). First, much of the combinatorial diversity of of β -selection, with TCRα rearrangement continuing to

A
Dδ1,2
~60 Jα
Vα Vα Vα Vα Vδ Vδ Vδ Vδ Cδ Cα
Jδ1,2
B
Vγ2/4 Vγ3/5
Vγ5/7 Vγ4/6 Vγ1.2/2 Vγ1.1/1
γ pseudogene
J Cγ1 Cγ2 J J Cγ4
C First-wave fetal
Vγ4/6:Vδ1 Vγ3/5:Vδ1
Invariant Invariant
Tongue, (lung), Skin
reproductive (=dendritic
tract epidermal T)
D Late fetal and adult
Vγ2/4 Vγ1.2/2
Diverse Diverse
Vγ5/7:Vδ4,5,6 Vγ1.1/1:Vδ6.3,6.4
Restricted Restricted
Intestinal Liver, thymus, spleen
epithelium Innate-like IFNγ, IL4
FIG. 13.5. Genetic Loci Involved in T-Cell Receptor (TCR)fc Cell Generation and Distinct Developmental Assignments.
TCRγδ cell generation in the mouse depends on the rearrangement status of the TCRδ/TCRα composite locus and the three
functional TCRγ subloci. A: TCRδ/TCRα loci in the mouse: only a few of the many Vα, Vδ, and Jα segments are shown.
B: TCRγ subloci: the V and J segments within the locus are shown in full, though distances are not to scale. Directions of
transcripts are shown by arrows. Two different nomenclatures are commonly used for the TCRγ gene segments, those of
Heilig and Tonegawa280 and of Garman, Doherty, and Raulet.281 Here, V segments are identified with both numbers, sepa-
rated by a slash: (number from Garman et al.281)/(number from Heilig and Tonegawa280). C: Fetal-specific timing of rear-
rangement and specific developmental fates of cells expressing invariant receptors using the indicated Vγ segments from
the JCγ1 cluster. These TCRγ chains are paired with invariant TCRδ chains using only the Vδ1 segment for rearrangement.
D: Specific developmental assignments for cells using other Vγ segments for rearrangement at later fetal and postnatal
stages. Although cells using the Vγ5/7 together with Vδ4, 5, or 6 and cells using Vγ1.1/1 together with Vδ6.3 or 6.4 TCRs
show some variations in specificity, they are considerably restricted compared to other adult γδ T cells, correlating with
their specific functional properties.109,138

be blocked and TCRγ transcription being allowed to con- progeny only.123 These highly efficient TCRγδ progenitors
tinue in the γδ -selected cells in contrast to DP cells. Unlike are mostly DN2, not DN3. Thus, by the time they reach DN3
β -selected cells, TCRγδ cells undergo only restricted prolif- stage, even before productive TCRβ rearrangement, many
eration, with no induction of RORγ t, much rarer induction cells have already lost much of their potential efficiency as
of CD4 or CD8αβ genes, and maintenance of a much higher γδ precursors. In part, this may reflect the declining IL7R
ratio of Bcl2 to Bcl-xL. Selection of most γδ cells in mice is signaling intensity that cells appear to experience from DN2
also much less dependent on sustained Notch signaling at to DN3 stage, as a precondition for TCRαβ lineage develop-
the DN3 stage than β -selection, and entails much stronger ment.100,124,125 Not just signal strength, therefore, but also de-
upregulation of immediate-early response transcription fac- velopmental stage and cytokine signaling context at the time
tor Egr3 (a zinc finger transcription factor) and the E protein when the first TCR signal is received may distinguish most
antagonist Id3.72,102,103,115 The consequence may be greater or β -selected cells from most γδ -selected cells. It is interesting
more durable viability: in one transgenic model TCRγδ - that the time at which the “TCRγδ window” closes, together
selection depends on the presence of the presumed target with the relative effects of Notch signaling on the two fates,
antigen in the environment116,117—a situation that would are evolutionary variables that are different in human and
normally induce deletion of nascent β -selected cells. mouse αβ /γδ lineage choice.21,22,122
Strength of Signal Distinct Genetic Requirements for a-Selection

One important model to explain these program divergences and fc-Selection
focuses on the difference between the TCRγδ complex and This implies that TCRγδ precursor activity is maximal
the pre-TCR complex in terms of the strength of signals that close to the time of T-lineage commitment itself. Recall
they transduce when they are assembled in the developing that one event that normally divides uncommitted DN2a
cells.72,118 In mice, the TCRγδ complex appears to deliver cells from committed DN2b cells is the induction of
stronger spontaneous signals than the pre-TCR complex, Bcl11b. Bcl11b function is indispensable to make cells eli-
and this seems to have an “instructive” effect on the way the gible for β -selection and to sustain their survival into the
cell responds. There is evidence that experimentally reduc- DP stage.63,64,126 However, it is not essential for expression of
ing the signaling through the TCRγδ complex now enables Rag1 and Rag2, nor for the expression of IL-7R. Remarkably,
the TCRγδ to direct DN3 cells to a β -selection program. Bcl11b-knockout cells retain the ability to generate a mod-
Correspondingly, precocious (transgenic) expression of a est number of T-cell progeny, provided that these are TCRγδ
full TCRαβ heterodimer at the DN3 stage delivers stronger cells.63,127 Thus, there must be some pathway to make at least
signals than a pre-TCR complex, and this can direct many certain kinds of TCRγδ cells that is open so early that it can
of the cells into a γδ cell fate, skipping the CD4, CD8, and bypass the standard commitment machinery, perhaps never
RORγ t expressions that define the DP stage.119–121 Strength using the functions that would be needed to set up a true
of signal appears to be more important than the identity of TCRαβ versus TCRγδ lineage choice.
the TCR itself. More recent work has elaborated on this to Indeed, γδ cells can be generated in a number of mutant
imply that the ratio of TCR signal strength to Notch path- conditions that are prohibitive for T-cell development.22
way signal strength may be the true deciding factor.122 Severe inhibition of E protein activity128,129 or withdrawal
These observations apply to the divergence between DP of Notch signaling102,115,130–133 as already noted can block αβ
cell selection and selection of most γδ cells in the young lineage development much more severely than γδ -cell devel-
adult mouse. However, there is more than one way to be- opment in mice. There appear to be broader differences in
come a γδ cell, and the γδ fates have great developmental central features of cell biology and epigenetic regulation be-
and evolutionary diversity. Furthermore, there is evidence tween TCRαβ and TCRγδ programs, as well. Deletion of key
that many of the cells undergoing γδ -selection need not start epigenetic regulators Brg1 and DNA methyltransferase 1 by
from the same developmental state as the cells that embark Lck-Cre (in DN2-DN3 stage) was found to inhibit αβ lineage
on the β -selection program. cells more than γδ lineage cells, though in these cases deletion
might have occurred too late to affect all γδ precursors.134,135
More remarkably, γδ but not αβ lineage cells are generated in
More than T-Cell Receptor Signal Strength:
mice with germline mutations affecting the Polycomb Group
Developmental Factors in T-Cell Receptor-fc Cell chromatin regulator Bmi1 or the ribosomal protein L22.136,137
Specification Some of these differences may reflect the special require-
Stage-dependence of T-Cell Receptor-fc Specification ments for extensive proliferation at β-selection that do not
If all T cells were undecided until the DN3 stage, when apply to the milder response of γδ -selection. However, the
RAG1/RAG2 activity reaches its first peak and most Vβ -DJβ effect of Bcl11b loss can also hint that the differences could
rearrangement occurs, then there should be no difference trace back to different times and modes of commitment.
in the ratio of αβ : γδ progeny from individual T-cell precur-
sors tested from DN2a stage until late DN3 stage. However,
Kang, Volkmann, and Raulet pointed out that a subset of Multiple Lineages of T-Cell Receptor-fc Cells
DN2-DN3 cells with high IL-7R expression was much more TCRγδ fate can be any one of multiple different develop-
likely than the average to produce TCRγδ progeny as well mental programs138 (Fig. 13.5C,D). The first evidence for
as TCRαβ progeny; cells with lower IL-7R made TCRαβ this related to the “first wave” fetal thymocytes. These are

not a generalized TCRγ population; they comprise two sub- could represent an “innate-like” subset of TCRγδ cells with
lineages that share an invariant TCRδ rearrangement and its own distinctive set of roles.
pair it with an invariant TCRγ rearrangement at the Cγ1 su- The demarcation of these different TCRγδ cell subsets
blocus, in one case using Vγ 3/5 and in the other case using is new enough so that how much their different properties
Vγ4/6 in Figure 13.5 (see figure legend for nomenclature of are results of maturation from different precursors and how
Vγ segments). These two TCRγδ receptor T-cell classes have much they are effects of particular TCR-mediated signaling
distinct assignments in the body, the Vγ 3/5 cells emerging experiences during selection and antigen responses is still
from the thymus to patrol the skin epidermis (as “dendritic unresolved.138,144,147–149 To the extent that many TCRγδ cells
epidermal T cells”) and the Vγ4/6 cells to patrol the tongue use TCR of very limited diversity, it is likely that they are
and reproductive tract. Both types of TCR rearrangements pretargeted to interact with particular structures of evolu-
are limited to fetal life. A third Vγ segment at the Cγ1 sub- tionarily selective importance. However, it is clear that the
locus (Vγ 5/7) is used in a similar way in cells targeted to the choice of being a TCRγδ cell rather than a TCRαβ cell opens
gut epithelium, as a major class of TCRγδ + intraepithelial the door to a rich spectrum of developmental pathways and
lymphocytes. These Vγ 5/7 TCRγδ intraepithelial lympho- immunological roles.
cytes are also distinct in that they can arise not only from
intrathymic progenitors but also from progenitors devel-
oping in a nonthymic site.139 Clearly, these early waves of POSITIVE VERSUS NEGATIVE SELECTION OF
TCRγδ cells have distinct homing specificities as well as no- T-CELL RECEPTOR-à CELLS: MORE THAN A
tably sublineage-specific, invariant recognition potentials. BINARY SWITCH
The Vγ 3/5 cells are preprogrammed to use this Vγ seg- The highly diverse TCRαβ repertoires generated in DP cells
ment for rearrangement because the earliest fetal thymocytes pose a different challenge to the goal of producing self-
maintain a high level of the E protein antagonist Id2, and tolerant mature T cells that can recognize foreign antigens
this Vγ segment is specifically opened in cells that have low and produce appropriate effector molecules to orchestrate
E protein activity.140 Normally, a high Id2:E protein ratio is a the immune response. Because TCRαβ specificities are ran-
hallmark of NK cells, which these cells are not. Nevertheless, domly generated, developing T cells must be selected for
they support their growth with another distinctive NK- their ability to respond appropriately to host MHC mol-
like feature, namely the IL-2/IL-15R complex (a dimer of ecules. This is achieved through positive selection in the
CD122:CD132, also called IL-2/IL-15Rβ : γc).141 In the Vγ 3/5 cortex, a process in which cortical thymic epithelial cells
first-wave cells, signals through this receptor complement (cTECs) deliver a survival signal to thymocytes that rec-
and synergize with IL-7R signaling to promote their robust ognize self-peptide and MHC complexes. Thymocytes that
early dominance. As a result, in these cells and in the intesti- are unable to recognize these complexes fail to receive the
nal-targeted Vγ 5/7 cells only, the Stat5 needed for TCRγ rear- survival signal and thus die by a process known as death by
rangement can be activated by the response to IL-15 via IL-2/ neglect. Interestingly, DP cells selected on a particular class
IL-15R: γc, in place of IL-7/IL-7R interaction.142 It is interest- of MHC molecules maintain their recognition specificity as
ing that cells of this type are especially enriched in popula- they mature and migrate to the periphery.
tions derived from Bcl11b-knockout precursors.59 Positive selection is a test for the basic structural integrity
In the past 5 years, distinctive sublineages have also been of the TCRαβ complex made as an outcome of productive
recognized among adult TCRγδ cells (see Fig. 13.5D). Two TCRα rearrangement and for its ability to interact with self-
other TCR Vγ segments, those used with Cγ4 and Cγ 2, can MHC. However, beyond these criteria it is not stringent in
also be associated in a preferential way with particular γδ its requirement of TCR specificity. A single peptide-MHC
functional subsets. The Vγ1.1/1-JCγ4 TCRγ chain is associ- complex can select DP cells with several TCR specificities,
ated with an IFNγ-producing, NK1.1+ functional subset, and several different peptide-MHC complexes can select a
whereas the Vγ1.2/2-JCγ 2 TCRγ is often associated with an single TCR as long as they all have a low-affi nity interac-
IL-17–producing, CCR6 + subset.138 Cells using Vγ1.1/1-JCγ4 tion with the selecting complex.150,151 This is important as all
chain tend to pair it with a TCRδ chain that uses Vδ6.3 or positively selecting cells only see self-antigens in the thymus,
Vδ6.4 with limited junctional diversity, most likely indicat- and it is possible that they could react strongly and be acti-
ing yet another stereotypical antigen-recognition specificity. vated when a foreign peptide is displayed in the same MHC
Strikingly, the cells that use this receptor have a distinctive context in the periphery. The MHC exhibits both polymor-
set of developmental requirements that split them from all phism and polygeny, and this is thought to allow the im-
other TCRγδ cells. In contrast to other murine γδ cells that mune system to respond to several different, fast-evolving
emerge preferentially from conditions of low E protein activ- pathogens. Moreover, the next step of negative selection
ity, due to high Id3 (most γδ cells) or Id2 (fetal first-wave γδ will weed out any cells that are excessively activated in re-
cells), these “Vγ1.1/Vδ6.3” cells cannot develop unless there sponse to self-antigen presentation in the thymic medulla.
is ample E protein activity. Furthermore, they resemble the In fact, as we will see later, positive selection of some self-
NKT cells of the TCRαβ type in other aspects of their devel- reactive cells is imperative for the formation of Tregs. Thus,
opmental requirements. They express both IFNγ and NK1.1 positive selection ensures that only immunologically useful
and also have a specific requirement for the transcription TCRs (ie, those that recognize the particular combination of
factor PLZF (promyelocytic leukemia zinc fi nger [PLZF], MHC molecules expressed in an individual) are selected for
also called Zbtb16).143–146 As discussed in a later section, this further development.

An inevitable outcome of the random generation of TCR Concomitantly, expression of the chemokine receptor CCR7
diversity and the relatively promiscuous positive selection is turned on. This enables the cells to migrate from the cor-
process is the survival of cells that can be highly reactive to tex to the medulla, where they can be screened by negative
both foreign and self-peptides bound by self-MHC. TCR- selection as well as maturing to acquire effector function.2,9
positive cells go through an additional checkpoint known as While in the medulla, thymocytes undergo several
clonal deletion or negative selection in the medulla, which changes that lead to their fi nal maturation. Several pecu-
shapes the final TCR repertoire by pruning out self-reactive liar features of DP cells, imposed during β -selection, are
cells. In the process, they repetitively test their TCRs for reversed. Glycoprotein processing is altered back to a more
self-reactivity through interactions with medullary epi- “normal” pattern, so that new glycoproteins are once again
thelial cells (mTECs) and DCs expressing a huge range of fully sialylated. The cells regain the Bcl-2 and class I MHC
self-peptide-MHC complexes. Negative selection results in expression that they had lost at β selection, and they begin
the apoptotic death of cells with high-affinity TCRs for at to recover the functional responsiveness that had disap-
least one of the self-antigen-MHC complexes. However, not peared at that time. As this proceeds, the cells eventually
all thymocytes that react with self-peptide-MHC with high become resistant to negative selection, and they downregu-
affinity die during negative selection. Instead, some self- late CD24 (heat-stable antigen) as TCR/CD3 levels rise even
reactive thymocytes survive and develop into Tregs that are higher and CD69 expression finally subsides. Importantly,
potent dominant suppressors of autoimmune T cells, which activation of PI 3-kinase and Akt diminishes, allowing the
also arise in the thymus (see following discussion). Negative quiescence factor FoxO1 to localize to the nucleus, where it
selection and Treg development, together known as central can activate an emigration program.2 FoxO1 target genes
tolerance mechanisms, ensure that the final TCR repertoire include the transcription factor Klf2, which turns on ex-
is self-tolerant. pression of L-selectin and the emigration receptor itself, a
G-protein coupled receptor for the glycolipid sphingosine
1-phosphate.153–156 Thus, the cells are enabled to leave the
Timing and Anatomical Sites medulla and migrate out to the periphery. The TCR upregu-
DP thymocytes have only a 3-day lifespan on average before lation, rescue from death by neglect, release to the medulla,
they die by neglect, and this creates an effective deadline for and functional maturation events appear to be common to
positive selection. Throughout this time they actively carry all positively selected thymocytes.
on V-J rearrangement of the TCRα locus. This starts as soon The exact stage at which negative selection is initiated
as RAG expression resumes after β -selection and sometimes has been a matter of debate. Using TCR transgenic model
even before the cells finish proliferating. Expression of an systems, clonal deletion was originally found to occur at a
in-frame TCRα chain replaces pre-TCRα in complexes with range of stages from the DN or DP stage of T-cell develop-
TCRβ, enabling the cells to be auditioned for positive selec- ment in the thymic cortex to the SP stage in the medulla.
tion immediately based on the interaction with ligands on However, the interpretation of these results for normal thy-
the cortical epithelial cells. There is no allelic exclusion of mocytes remained controversial because of the early timing
TCRα gene rearrangement; the process is terminated either and/or high level of transgenic TCR expression relative to
by positive selection signaling, which finally shuts off RAG normal T cells.157,158 Several lines of evidence suggest that
expression, or by cell death. Individual cells can rearrange under normal conditions, negative selection may be pre-
the α-chain genes on both chromosomes, not only once but dominantly separated from positive selection both develop-
many times, because the locus offers more than 50 possible mentally and spatially. In a transgenic TCR model system
Jα segments as well as Vα segments in a permissive topol- where the TCR was expressed with normal timing, deletion
ogy.152 Rearrangement tends to begin with Vα and Jα ele- was found to predominantly occur in SP cells in the me-
ments centrally located near the TCRδ complex, working dulla.159 Moreover, CCR7-directed migration to the medulla
outward to rearrange more 5′ Vα and 3′ Jα segments until a was recently shown to be necessary for negative selection
product of rearrangement is made that mediates positive or against tissue-restricted antigens (TSAs).160
negative selection, or until the cell dies. During positive and negative selection, cells migrate
Although TCRα gene rearrangement is halted by the first through the cortex and medulla in distinct patterns that
successful TCR engagement, the process of positive selection reflect selection and progression to the next stages of devel-
is more than the initial contact with MHC ligand, and in opment. Thus in the cortex, positively selecting cells survey
fact TCR interactions with peptide-MHC must continue for self-antigen presenting cTECs by a “random walk” at low
several days in order for the process to be completed. As we speeds of 3 to 8 μm/minute. Once a positively selecting sig-
will see later, this requirement for ongoing or repeated TCR nal is received and thymocytes upregulate CCR7, they stop
signaling after the first responses to positive selection signals surveying the cortex and, presumably pulled by the chemo-
is a crucial element of the mechanism that determines the tactic attraction to CCR7 ligands expressed by mTECs, mi-
CD4/CD8 lineage choice. grate in a highly directed manner to the medulla. CCR7 is
The sequence of events triggered by positive selection in- not only important for the directional migration of SP thy-
cludes little if any proliferation, in contrast to β -selection. mocytes, but it is also important for the accumulation of SP
CD69 is upregulated, RAG genes are turned off permanently, cells in the medulla. In the medulla, they resume their ran-
CD5 levels increase, and TCR/CD3 complexes are stabi- dom walk but at a much higher speed of 20 to 25 μm/minute.
lized so that their surface expression increases in parallel. Any particular self-antigen may be expressed only by 1% to

5% of mTECs, and this high-speed scanning is thought to Nur77-GFP knock-in reporter mouse that has made it pos-
help the thymocytes in seeing all possible self-peptides ex- sible for the first time to look at TCR signal strength on a
pressed in the medulla. Upon encountering a negatively se- stage-by-stage and cell-by-cell basis.163 Nur77 is an orphan
lecting antigen, SP cells slow down, give up their random nuclear receptor that is a product of the Nr4a1 gene, an im-
walk again, and restrict their movement to confi ned areas mediate early response gene transcribed downstream of
but continue to survey additional peptide MHC complexes. TCR engagement.164,165 Importantly, upregulation of Nur77
Thus, developing thymocytes effectively scan both the cor- is directly proportional to strength of the activating TCR
tex and the medulla to increase their chance of encountering signals. Earlier work has shown that while positive selec-
positively and negatively selecting ligands.161,162 tion causes a 2-fold increase in Nur77 expression, negatively
selecting cells expressed Nur77 at 10-fold higher levels.166
Similarly, Nur77-GFP expression in vivo was found to be
Signal Strength and Signal Quality much higher in cells undergoing negative selection as com-
Both positive and negative selection involve TCR- and Lck- pared to positively selecting cells, thus demonstrating di-
mediated signaling cascades triggered in response to self- rectly that negative selection is the result of very strong TCR
peptide-MHC complexes, but these processes result in very signaling.
different outcomes (Fig. 13.6). While it has been known While it is not fully understood how these signals
for a long time that cells that react with self-peptide:MHC give such different outcomes, several factors contribute.
with low affi nity (and thus receive weaker TCR signal) Combined triggering of TCR and costimulatory recep-
are positively selected while those that react too strongly tors such as CD28 and CD40L strongly favors a cell death
are negatively selected, most of our knowledge has come response by purified DP cells and incompletely mature
through transgenic TCR expression, and it has been diffi- single-positive cells alike.6 Such costimulation may activate
cult to demonstrate the signal strength perceived by poly- distinct signaling mediators in the cell, or provide an altered
clonal thymocytes undergoing selection in vivo. Recently, balance of signals, not only act to amplify perceived TCR
Moran et al. reported the development of a BAC transgenic signal intensity. For example, research over the past several
Death by neglect Positive selection
Negative selection
Treg selection
Cell #
+TGFβ
TCR affinity to self-antigens

TCRs common to
Treg and Tconv cells
FIG. 13.6. T-Cell Developmental Fate is Determined by the Affinity to Self-Antigens. Majority of thymocytes in normal wild-type mice express
T-cell receptors (TCRs) that do not recognize self-antigen:major histocompatibility complex complexes with sufficient affinity and hence die
due to neglect while a minority are positively selected (dashed green line). However, only those positively selected cells that escape later
negative selection can mature into conventional effector T cells (Tconv). During negative selection, thymocytes that recognize self-antigens
with strong affinity (dashed red line) undergo clonal deletion. Treg selection occurs in a narrow window of TCR affinity, (dashed blue lines),
which shows some overlap with net positive selection giving rise to TCRs represented in both conventional CD4+ T cell (Tconv) and regulatory
T (Treg) cell populations.245,246,248 As shown in the figure, TGFβ is important in rescuing Treg cells as well as thymocytes in general from negative
selection.221

years continues to highlight the essential role of differential are important because at the DP stage as already noted, co-
activation of the mitogen-activated protein kinase (MAPK) stimulation leads to clonal deletion instead of activation.
pathways during positive and negative selection. Signals Thus, the cortical epithelial microenvironment provides a
transduced during positive selection appear to be particu- uniquely forgiving testing ground for newly generated TCR
larly dependent upon calcineurin and ERK, while negative recognition specificities. In addition, however, it may pres-
selection depends more upon JNK and p38 activation, al- ent a qualitatively distinctive set of antigens for positive
though ERK also appears to be activated. However, it seems selection. Recent work has revealed that cTECs uniquely ex-
that weak TCR signals during positive selection result in low, press a novel catalytic subunit “β5t” of the 20S proteosome,
sustained ERK activation, while strong negative selection a multisubunit protein complex that cuts proteins into pep-
signals result in a transient burst of ERK activation and JNK tides, some of which will eventually be displayed on MHC
and p38 activation. It is also reported that during positive I molecules. While cTECs also express a low level of the
selection, activated Erk remains compartmentalized in the immune-cell-specific β5i subunit, the expression of the β5t
Golgi membrane as opposed to its negative-selection-specific subunit is essential for the selection of immunocompetent
localization at the plasma membrane. Differential ERK acti- CD8 T-cell repertoire. Peptides that bind MHC I molecules
vation patterns are functionally important, not just report- in other cells are generated by the chymotrypsin activity of
ers for the difference between positive and negative selection the canonical β5 subunit, which generates peptides with hy-
signaling, for double knockout of Erk1 and Erk2 severely re- drophobic residues at their C terminus. This is important
duced positive selection although it did not show any effect because they form a hydrophobic anchor and ensure that
on negative selection.153,167,168,168a Additionally, while the loss the peptides fit tightly in the peptide-binding groove of the
of Grb2, an adaptor protein that mediates Ras-MAPK acti- MHC. The β5t subunit has significantly reduced chymo-
vation downstream of TCR engagement, results in a severe trypsin activity, and it is not clear whether self-peptides that
defect in both positive and negative selection, the absence are processed in cTECs fit snugly in MHC I peptide-binding
of Grb2-associating protein, Gasp (also known as Themis: groove at all.172–176 Additionally, the crystal structure of TCR-
thymocyte-expressed molecule involved in selection) causes MHC complex shows that TCR V regions that make contacts
a defect only in positive selection.169–171 Identifying signaling with MHC are encoded by the germline Vβ and Vα elements
molecules that act exclusively or differentially during posi- themselves, in contrast to the sequences generated during
tive or negative selection will be important for discovering VDJ recombination, which encode the TCR regions that
specific differences in the pathways that lead to such strik- contact the antigenic peptide. These reports have impor-
ingly different fates. tant implications for positive selection, because suboptimal
The difference may not simply be based on signal strength. binding of self-peptides to MHC I molecules on cTECs and
The physiological context changes between positive selec- the possibility that MHC recognition may be encoded in the
tion, occurring in the more immature DP cells in the thymic genomic sequence of the TCRs together imply that the weak
cortex, and negative selection, occurring typically, but not signal required for positive selection could be generated by
exclusively, in more mature positively selected SP T cells in TCR-MHC rather than TCR-peptide-MHC contacts.177 A
the thymic medulla. These SP T cells are negatively selected cTEC-specific difference is also seen in the peptide process-
within a network of mTECs and DCs that not only express ing machinery responsible for loading MHC II molecules.
an expanded array of self-peptides and MHC complexes but Thus, unlike other bone marrow–derived APCs that express
also express costimulatory molecules such as CD80, CD86, the protease cathepsin S, cTECs express cathepsin L, and the
and CD40 that are rare in the cortex. Furthermore, as a reexpression of this protease is crucial for the positive selec-
sult of positive selection, these SP T cells have undergone tion of CD4 cells.178,179
major changes in differentiation programming as CD4 Negative selection depends on the presentation of
or CD8 cells. The higher levels of surface TCR and other self-peptides to deliver a strong TCR signal that initiates
changes as a result of positive selection, such as upregulation apoptosis in self-reactive thymocytes, i.e. cells that might
of CD5 and restored sialylation of surface glycoproteins are react against self peptides if presented in the periphery
likely to change sensitivity of SP cells to high-affinity TCR by professional antigen-presenting cells. Accordingly, it
signals. Thus, the different developmental states (DP versus is not surprising that an mTEC-specific difference in the
SP), levels of surface TCR and other signaling components, peptide processing and presentation machinery has not
gene expression programs, and microenvironment of the T yet been described for the medulla. mTECs do have spe-
cells, as well as the differing arrays of presented antigens in cial properties that make the medulla a highly efficient
the cortex versus medulla, are all factors that likely contrib- site of negative selection, and these are discussed in a later
ute to the different outcomes of high- versus low-affinity section.
TCR-peptide-MHC interactions leading to negative versus Thus it is possible that the mechanisms described pre-
positive selection. viously generate positively selecting peptides that could be
entirely different from those that trigger negative selection,
accounting for the difference in signal strength. A detailed
Anatomical Differences in Antigen Presentation investigation of the nature of MHC class I and II associated
Cortical epithelial cells provide a rich source of MHC class I peptides displayed by the cTECs should therefore help in
and class II surface complexes with a notable lack of costim- a better understanding of the differences between positive
ulatory molecules for T cells. These features of the cTECs and negative selection.

and that the interaction with MHC merely selects the

POSITIVE SELECTION: CD4/CD8 LINEAGE CHOICE precommitted cells by delivering a survival signal. In one
Major Models for Functional Lineage Divergence version, stochastic events were thought to give rise to CD8
The mechanism that regulates the choice between CD4 and and CD4 T cells with mismatched TCR coreceptor pairs,
CD8 lineages has been the focus of intense research as well which would eventually die due to neglect. However, the
as debate for the last two decades. The central question has experimental evidence required to support either of these
been how DP cells, which express both CD4 and CD8 co- models was lacking. Firstly, in certain TCR transgenic mod-
receptors, make the decision to differentiate into CD4 or els, MHC I–restricted cells could develop into CD4 cells
CD8 SP thymocytes. The CD4/CD8 fate decision is probably and conversely CD8 cells could be class II MHC–restricted
made in the cortex, during the iteration of positive selection based on the selecting antigen, thus discounting the ex-
signals. By the time the cells reach the medulla, their CD4 clusivity of the signals these subsets received. Secondly,
or CD8 identities appear to be set. Research from several the efficiency of selection of TCR repertoire can be much
groups has now provided a better understanding of how and higher than that predicted by stochastic matching of TCR
when the CD4 versus CD8 lineage choice is made. specificity and co-receptors.
It is now clear that MHC class I– and II–restricted thy- The experiments that resulted in the abandonment of
mocytes experience positive selection signals differently. these hypotheses gave rise to newer, more refined models
Firstly, it was observed that the affi nity of Lck, a down- that focused on the strength and more importantly on the
stream kinase crucial for transmitting the TCR signal, for duration of TCR signaling. Thus it was demonstrated using
the CD4 and CD8 coreceptors is different. Thus, because fetal thymic organ cultures in which, depending on whether
CD4 has more Lck associated with it than CD8, DP cells ex- TCR signaling was either sustained or interrupted, that thy-
pressing an MHC II-restricted TCR receive a stronger signal. mocytes expressing the same transgenic TCR could be dif-
The second difference lies in the duration of Lck signaling ferentiated into CD4 and CD8 lineage, respectively. Finally,
over subsequent developmental time. All positively selected when the CD4 gene was placed under the control of CD8
cells are preprogrammed to downregulate CD8 irrespective regulatory elements so that CD4 instead of CD8 was down-
of their MHC restriction and progress to the CD4 + CD8lo regulated in postselection thymocytes, MHC II–restricted
stage. These cells are still uncommitted, and their lineage thymocytes developed into CD8 SP cells. These experiments
fate is determined based on whether or not they receive fur- thus showed that signal strength was secondary to the more
ther TCR signals. Downregulation of CD8 results in inter- important factor, the duration of TCR signaling, in deciding
ruption of TCR signaling in MHC I–restricted thymocytes CD4- versus CD8-lineage fate.183,184
because now there is no CD8 to stabilize the class I MHC
and TCR interaction. On the other hand, class II–restricted
cells still express CD4, which allows continued TCR signal- Kinetic Signaling
ing upon subsequent contact with MHC II molecules and The most recent refi nement to the lineage choice hypoth-
results in the development of CD4 SP thymocytes. Thus, esis came in the form of the kinetic signaling model, which
while CD4 cells develop in a straightforward manner by provides an explanation of the pathway (outlined previ-
downregulating CD8, CD8 SP development involves the ously) through which majority of thymocytes make the
downregulation of the CD8 receptor itself. In the subse- CD4/CD8 decision182,185–188 (Fig. 13.7). Importantly, kinetic
quent stages, in the absence of further TCR signaling and signaling agrees with previous models in that the duration
presence of IL-7R-Stat5 signaling, MHC I–restricted cells of TCR signaling is important in lineage choice. It was for-
downregulate CD4, undergo “coreceptor reversal” to up- mulated principally to incorporate the following experi-
regulate CD8 and mature to the CD8 SP stage. A set of five mental observations not explained by previous models: 1)
enhancers, regulated in a stage specific manner, control the the silencing of CD8 transcription in positively selected
expression of CD8α ; it is interesting to note that differ- DP thymocytes resulting in the development of CD4 +
ent enhancers are involved in the CD8 downregulation in CD8lo thymocytes that exhibit unequal surface expression
postselection thymocytes and its re-expression in CD8 SP of co-receptor molecules, and 2) the lineage noncommit-
cells. This allows for coreceptor reversal by the induction ted state of postselection CD4 + CD8lo thymocytes. Thus,
of lineage-specific factors downstream of IL-7R signaling it was proposed that transcriptional silencing of CD8 ex-
that are silent in uncommitted cells and thus adds another pression in thymocytes following a positive selection signal
layer of transcriptional regulation to the lineage choice is employed as a mechanism to assess changes in the TCR
decision.180–182 signaling environment, which would distinguish between
The major hypotheses originally proposed to explain TCR/coreceptor complexes with class I MHC and those
the lineage choice decision can be classified into two main with class II MHC. If TCR/coreceptor signaling persists,
categories. 1) The “instructive” models were based on the the cells will differentiate into CD4 SP. If TCR/coreceptor
proposal that MHC class I and II signals are qualitatively signaling begins strongly but is then reduced, then upon
different and that TCR and CD8 or CD4 interaction with receiving IL-7R or other γc chain mediated signals, posi-
these molecules “instructs” the cells to differentiate to CD8 tively selected MHC I–restricted thymocytes differentiate
and CD4 lineage, respectively. 2) The “selective” or “sto- to the CD8 lineage. Recent experiments have now shown
chastic” models posited that CD4 and CD8 lineage choices that IL-7R-Stat5 mediated signaling is crucial for corecep-
are made even before the cells interact with MHC molecules tor reversal and CD8 development. Elegantly, the emphasis

Continued TCR signaling OFF

TCR signaling + + discontinuous Lck
Lck activation activation + IL-7
Gata3+
Tox+
Th-Pok
Th-Pok Runx3+
CD4+ CD4+ CD4-
CD8- CD8lo CD8+
CD4 CD8
CD8 CD4
Positive Selection
pMHC I/II:TCR
Positive Selection
pMHC I:TCR
CD4+
CD8+
Gata3lo
Tox-
Th-Pok-
FIG. 13.7. CD4 and CD8 Lineage Differentiation.The majority of the positively selected DP cells (peach) differentiate into CD4 (green) and CD8
single positive (SP; blue) cells through the common CD4+ CD8lo (red) uncommitted intermediate.185 Transcription factors and coreceptor gene
expression are shown in blue, whereas increases or decreases in their expression are denoted by upward and downward arrows, respec-
tively. A small number of major histocompatibility complex I–restricted cells directly mature to the CD8 SP stage (dashed arrow) after positive
selection. Transcription factor expression and signaling differences that enable these cells to take a more direct route to the CD8 SP stage
remain to be identified.
on signal continuity or discontinuity rather than absolute Th-POK and GATA3 as CD4-Specific Drivers
strength enables the CD4/CD8 decision-making mecha- Work from two groups independently led to the discov-
nism to operate independently of the decision to be posi- ery of T-helper–inducing POZ/Kruppel-like factor (Th-
tively or negatively selected. POK, also known as c-Krox or Zfp67), a BTB-POZ domain
The understanding of the genetic basis for CD4 com- containing transcription factor encoded by the zbtb7b gene,
mitment came about through the discovery of Th-Pok, a which is crucial for CD4 lineage commitment and fidelity.
gene that enforces CD4 commitment in CD4 + CD8lo thy- Kappes et al. identified a point mutation in the DNA-
mocytes that experience sustained TCR signaling. Study binding Kruppel-like zinc fi nger of Th-POK to be respon-
of mice deficient in Th-Pok function also showed for the sible for the complete absence of mature CD4 + T cells in
fi rst time that positive selection and lineage commitment the “helper-deficient” mouse model.189 In these mice, while
are two separate events189,190 (see Fig. 13.7). The molecu- MHC II–restricted cells were positively selected, they did
lar circuitry that executes the CD4/CD8 lineage choice not mature into CD4 SP cells but were efficiently redirected
is becoming one of the best understood aspects of T-cell to the CD8 lineage instead. Similar results were seen in a
development. Th-Pok knockout mouse, thus identifying a central role for

Th-POK in CD4-lineage commitment and blocking CD8- Runx3 as a CD8-Specific Driver

lineage development. Bosselut et al. also identified Th-POK In the CD8 lineage, Runx3, a Runt domain–containing
as a differentially expressed gene, which is highly expressed transcription factor expressed only in CD8 SP cells in the
in the CD4 lineage as compared with the CD8-lineage thymus, has emerged as the major factor required for lin-
cells.190 Forced expression of Th-POK was able to block eage commitment. Another family member, Runx1, is ex-
CD8 T-cell development completely and redirect MHC I– pressed in DP thymocytes as well as in CD4 SP cells, but its
restricted cells into the CD4 lineage instead. Interestingly, expression is downregulated in CD8 SP thymocytes. Single
lineage diversion in the absence or forced expression of knockout of Runx3 shows only a reduction in the number
Th-POK was accompanied by the acquisition of a gene ex- of mature CD8 SP cells in the thymus, but Runx1 was found
pression pattern specific to the diverted lineage, indicating to be upregulated in these cells, indicating that there may be
that Th-POK does more than simply regulate CD4 and CD8 some functional redundancy among these proteins. Indeed,
expression. double knockout of Runx1 and Runx3 displayed a severe
Th-POK can enforce CD4-lineage commitment in a and complete block in CD8 SP development, highlighting
dominant fashion and is also required to maintain CD4- the crucial role of Runx activity in promoting this lineage.
lineage–specific gene expression even after commitment. In this model, CD8 T cells were redirected to the CD4 lin-
Continuous Th-Pok expression is required for maintaining eage instead, and interestingly Th-Pok expression was found
CD4 lineage fidelity of developing as well as mature cells, to be upregulated in DP cells at a stage even before positive
as its absence leads to the upregulation of CD8-specific selection. Thus part of the mechanism of Runx-mediated
transcription factors and effector genes, including Runx3, CD8 development may be through its role in repressing Th-
eomesodermin, perforin, and granzyme B.191 Thus the results Pok expression to a level that does not allow it to dictate the
obtained so far indicate that Th-POK represses CD8-lineage CD4 fate in these cells (see following discussion). There are
specific gene expression, in addition to providing positive some asymmetries between the roles of Runx3 and Th-POK
input into several helper-specific genes, to enforce the CD4 in CD8 SP and CD4 SP development, respectively. Forced
fate on developing thymocytes.192–197 expression of Runx3 does not redirect committed CD4 cells
GATA-3, another transcription factor that is important to the CD8 lineage. Thus Runx3 seems to promote the CD8
for T-lineage specification of early DN-stage thymocytes, fate by repressing Th-Pok expression, but not after high-level
also plays a major role in CD4 development. GATA-3 is nor- Th-Pok expression is already established.202–205 This asym-
mally downregulated during the DP stage, and its expression metry makes sense in light of the different signal duration
can be reactivated by a strong TCR signal typically experi- requirements for CD4 SP and CD8 SP development.
enced by DP cells expressing an MHC II–restricted TCR. While both mature CD4 and CD8 SP cells express Runx3
Several important observations point toward an important messenger RNA, it is translated much more efficiently in the
role for GATA-3 during CD4 specification and commitment. CD8 lineage. The difference seems to be in promoter usage:
Firstly, DP-specific knockout of Gata3 results in the block- Runx3 is transcribed from the distal promoter in the CD8
ade of cells at the CD4+ CD8lo stage, and these cells showed lineage, which encodes an efficient translational start site,
some low-level redirection to the CD8 lineage even in a β2- while the poorly translated proximal promoter isoform is
microglobulin–deficient environment where MHC class I expressed in the CD4 lineage. The distal isoform of Runx3
surface expression is profoundly impaired. Furthermore, is transcribed in CD4 + CD8lo cells only after they receive
forced expression of GATA-3 blocks CD8 T-cell develop- an IL-7R–Stat5 mediated signal, and this may be important
ment and promotes CD4 development instead. On the other in activating the distal promoter.182,192,202,206 Runx3 and Th-
hand, on its own, it cannot redirect developing class I– POK thus drive cells to the CD8 and CD4 lineages, respec-
restricted CD8 cells to the CD4 lineage, and thus is not suf- tively, and the gene networks that steer precursors to adopt
ficient to mimic Th-POK in gain of function experiments. these fates will be described in the following section.
A link between GATA-3 and Th-POK was established when
positively selected GATA-3 knockout DP cells were found
to be deficient in Th-Pok expression, and GATA-3 appears Architecture of a Signal-Duration–Dependent Lineage
to regulate Th-POK directly because a GATA-3 binding site Choice Network
in the second intron of the Th-Pok gene was also detected. The research that highlighted the role of Th-Pok and Runx3
As expected, strong TCR signals are required in addition as the major drivers of the CD4 and CD8 lineage fates, re-
to GATA-3 expression for optimal upregulation of Th-Pok. spectively, has shown that these factors antagonize each
GATA-3 seems to be upstream of Th-Pok because its in- other’s function and expression to establish lineage-specific
creased expression in positively selected DP cells does not gene expression profi les. However, these are not the only
depend on Th-POK. However, because Th-POK overexpres- players, and indeed several other proteins play an important
sion cannot rescue CD4 development in GATA-3 knockout role in translating TCR and cytokine dependent signals to
DP cells, it is clear that GATA-3 is not just required to up- drive lineage choice4,193,194,196,197,207–210 (Fig. 13.8).
regulate Th-Pok expression. Instead, like Th-POK, GATA-3 For the CD4 lineage, a strong positively selecting TCR
has other important functions of its own in promoting the signal leads to the upregulation of Gata3 in DP cells, which
CD4 fate. Thus, in summary, while GATA-3 acts as a speci- in turn activates the expression of Th-Pok. Gata3 expression
fication factor, Th-POK is the major commitment factor for itself is mediated by Ras/MAPK and calcineurin pathways,
the CD4 lineage.192,198–201 and depends on the activity of Myb, a transcription factor

Upregulation of
CD4-specific genes
Ras/MAPK and CD4 lineage
commitment
Gata3
c-Myb
Sustained
TCR signal
Tox Th-Pok Th-Pokhi
Positively
selecting MAZR
TCR signal
Upregulation of
CD8-specific genes
Runx3
and CD8 lineage
commitment
Socs1
Stat5
IL-7:IL-7R
FIG. 13.8. Gene Network Involved in Regulating Commitment to the Cluster of differentiation (CD)4 and CD8 Lineages. Important compo-
nents of the network that drive CD4 and CD8 lineage commitment following positive selection are shown in blue and yellow boxes, respec-
tively. Red lines show direct or indirect antagonistic or repressive interactions, whereas black arrows denote positive input into specific
genes. The mechanism by which MAZR leads to the repression of Th-Pok (dashed red line) is not well understood. See text for details.
activated downstream of TCR signaling, as in the absence making sure that Runx-mediated repression of Cd4 cannot
of Myb, Gata3 is not upregulated in CD69 + postselection take place. Th-Pok itself has two upstream Runx binding
thymocytes. Myb was also shown to regulate Gata3 expres- sites, through which it can be silenced in the CD8 lineage.
sion directly by binding to the upstream regulatory region of When expressed at a high level, Th-POK binds to this ele-
Gata3.211,212 Together with GATA-3, the high mobility group ment, an event that neutralizes Runx mediated silencing of
protein Tox is also important for optimal Th-Pok upregu- Th-Pok. In addition to this, Th-POK weakens Runx3 gene
lation in CD69 + cells. Tox deficiency blocks development expression by repressing the distal isoform of Runx3 that is
of all CD4 + lineages selectively while permitting CD8 crucial for its optimal translation and CD8 lineage promot-
development. Additionally, positively selected Tox-deficient ing function. Lastly, Th-POK also maintains CD4 lineage
thymocytes upregulate Gata3 normally but fail to maintain fidelity by dominantly maintaining repression of important
normal Th-POK and surface CD4 expression. Thus it seems CD8-lineage–specific genes like CD8α, Itgae, Nkg7, Cd160,
that Tox and GATA-3 work in parallel in the presence of and Prf1, giving rise to stable, lineage-committed CD4 SP
strong TCR signaling to upregulate Th-Pok expression and cells.192,195
advance the cells to the CD4 + CD8lo stage, which as stated It is not clear whether Th-POK expression is ever ini-
previously, is still uncommitted to either lineage.213 tiated in wild-type positively selected MHC I–restricted
From here on, presence or absence of continued TCR thymocytes. However, in TCR transgenic mice, Th-POK
signaling determines the lineage fate. If these cells receive is upregulated to a much higher level in class II–restricted
further TCR signals, it leads to an even greater upregulation thymocytes.190 Even if some Th-POK was induced in class
of Th-POK whose activity ensures a total shutoff of the CD8 I–restricted thymocytes in the wild-type situation, its effects
program. Firstly, Th-POK binds to the CD4 silencer element could be reversible because class II–restricted CD4 + CD8lo

thymocytes that express relatively higher levels of Th-POK the negative signaling window. TGFβRII− / − thymocytes
still remain uncommitted to the CD4 lineage. Thus, posi- showed an upregulation of the proapoptotic Bim, Bax, and
tively selected CD4 + CD8lo thymocytes that do not receive Bak proteins, and this had an adverse effect on conventional
any further TCR signal upregulate Runx3 expression, T cell as well as Treg maturation.221
which drives them to the CD8 lineage. Activation of Runx3
expression in these cells depends on the signaling through
the IL-7R, which is not expressed on DP cells but is upregu- The Special Role of the Medulla
lated in positively selected CD69 + thymocytes. Stat5 (and The medulla is the specific site of two kinds of APCs, which
under some conditions Stat6 also), which are downstream are potent inducers of negative selection: mTECs and DCs.
of IL-7R signaling, were shown to be crucial for induction These cells express high levels of MHC and costimulatory
of the CD8-lineage–specific distal isoform of Runx3. This molecules, both of which greatly facilitate negative selec-
process in turn is negatively regulated by SOCS1, which ex- tion, as well as being highly specialized for the presenta-
tinguishes signaling downstream of the cytokine receptor tion of a large array of peripheral self-antigens acquired by
and may play a part in keeping developing CD4 SP thymo- several unique mechanisms. The critical importance of the
cytes refractory to IL-7R mediated signaling.182 Once acti- thymic medulla to central tolerance, but not SP cell matura-
vated, Runx3 sets off a series of events that in turn block tion, was demonstrated in experiments showing that when
CD4 development. Most importantly, it silences Th-Pok T cells are prevented from migrating from the cortex to the
and Cd4 expression by binding to the silencer elements in medulla by the lack of the chemokine receptor CCR7 or its
these genes.204,214 However, since Runx complexes are bound ligand, cells undergo normal maturation and export from
to the Th-Pok silencer even in Th-POK expressing cells, the thymus, but the animals develop autoimmunity.222 The
other as yet unidentified factors may be crucial to enforce autoimmune phenotype resembles a failure of negative se-
Runx-mediated Th-Pok repression. The transcription factor lection, although faulty regulatory T-cell development (see
MAZR or Zfp238, product of the gene Patz1, may also coop- subsequent discussion) has not been ruled out. This showed
erate with Runx3 in repressing Th-POK expression, as Patz1 that central tolerance is, at least in part, dependent upon
knockout CD8 SP cells re-express some Th-POK and a small T-cell passage through the medulla and interactions with
number of them are redirected to the CD4 lineage.215 Lastly, medullary cells.
Runx3 also promotes the expression of CD8-specific genes, The discovery of the transcription factor gene, Aire, which
thus driving the cells’ commitment to that lineage.192 causes a rare autoimmune disease, autoimmune polyendocri-
Thus, it is the outcome of the antagonistic interplay nopathy-candidiasis-ectodermal dystrophy syndrome, when
between Th-POK and Runx3 that determines the lineage mutated in humans and mice, confirmed the critical role
choice of uncommitted precursors (see Fig. 13.8). However, of the medulla in negative selection.9a,223–226 Aire expression
the execution of these fates depend on some of their col- is restricted to the CD80hi MHCIIhi mature mTECs in the
laborators and target genes, and more factors contributing thymus where it regulates the transcription of several tis-
directly to CD4- and CD8-lineage differentiation probably sue restricted antigen genes and may also contribute to the
remain to be identified. This is prompted by the striking ob- development and turnover of these mTECs.227,228 The impor-
servation that deficiency of both Runx activity and Th-POK tance of Aire function was shown by experiments in which
still allows some CD4 and CD8 development to take place.192 the deletion of a single Aire-regulated tissue-restricted gene,
coding for Interphotoreceptor Retinoid Binding Protein
(Irbp3), led to an eye-specific autoimmune response when
NEGATIVE SELECTION the deficient thymus was transplanted into an athymic,
The phenomenon of negative selection of self-reactive Irbp3-sufficient host.228 AIRE interacts with several other
T cells in the thymus was first demonstrated in response transcription factors and related proteins, but the exact
to endogenous superantigens and self-MHC.216 Strong TCR transcriptional mechanism through which it regulates its
signals transduced by self-reactive TCRs trigger apoptosis target genes is not known. Several lines of evidence indi-
that appears to be mediated by upregulation and/or phos- cate that Aire may control chromatin accessibility to allow
phorylation of a proapoptotic Bcl-2 family member, Bim, transcription of repressed genes. First, the expression of
in collaboration with Bax and Bak.217,218 Upregulation of tissue specific antigens (TSAs) in mTECs does not require
the orphan steroid receptor, Nur77, also plays a role in this transcription factors that are normally essential for their
apoptosis.219 The negative selection program is not only fa- expression, and TSA levels are much lower than in the tis-
cilitated by TCR+CD28 costimulation in vitro; it has also sue to which their expression is normally restricted. Also,
recently been shown to require CD28 in vivo.219a Distinctive AIRE has the ability to bind H3K4me0, a histone mark as-
signaling pathways from TCR and costimulatory molecules sociated with silent chromatin, which may help its recruit-
that may activate these mediators are under intense inves- ment and the subsequent activation of transcription of
tigation. In knockdown experiments using small interfer- silent TSA genes.229–232 Finally, a new population of Aire-
ing (si)-RNA, the serine-threonine kinase MINK has been expressing cells (extrathymic Aire-expressing cells) impor-
reported to be involved in negative selection specifically, tant in maintaining peripheral tolerance in adult mice was
potentially linking TCR signals with JNK and p38, as well recently reported to be present in secondary lymphoid or-
as Bim activation.220 Recent data has also shown that TGFβ gans. Interestingly, AIRE does not upregulate the exact same
signaling provides a survival signal to thymocytes during set of TSAs in extrathymic Aire-expressing cells as it does in

mTECs, indicating that target specificity is determined by The linkage between Treg development and negative selec-
the transcriptional state of the cell in which it is expressed.233 tion is based on the common requirement of both processes
Thus Aire-mediated promiscuous expression of TSA genes for high-affinity TCR interaction with its cognate ligand
by mTECs provides essential self-peptides required for the (see Fig. 13.6). Treg cells do not develop in mice that express
clonal deletion of self-reactive thymocytes. Recently, it was a transgenic TCR specific for an exogenous protein unless
shown that Aire function is also required for the cross-pre- the cognate ligand is also simultaneously expressed, and Treg
sentation of TSAs expressed by mTECs.234 Thus, while the development in these mice is also accompanied by massive
expression level of an Aire-regulated transgenic ovalbumin negative selection of self-reactive thymocytes.242–244 Also, as
construct was directly correlated with the negative selec- compared to conventional T-cells, Treg cells also have higher
tion of CD8 SP T cells with cognate TCRs, the deletion of levels of the surface markers CD5, CTLA4, and CD25, whose
ovalbumin-specific CD4 SP T cells was independent of its expression level is directly proportional to the TCR signal-
activity to regulate transgene expression. Interestingly, Aire ing intensity. In the Nur77-GFP mouse model, Treg cells show
seemed to regulate transfer of expressed ovalbumin from two-fold higher GFP levels as compared to conventional T
mTECs to intrathymic DCs that are important for CD4 SP cells, further supporting the inference that strong TCR sig-
negative selection. nals are needed for Treg development.163 Thus, it is currently
DCs are more abundant in the thymic medulla than in thought that TCR affinities for self-antigen that are required
the cortex and are very potent mediators of negative selec- for Treg differentiation are higher than those required for
tion, even more so than mTECs. They do not express Aire positive selection but lower than the affinity threshold that
and so do not express the huge range of tissue-specific triggers negative selection (see Fig. 13.6). However, this pres-
genes that mTECs do. However, DCs have the ability to ents a dilemma: if negative selection leads to the deletion
cross-present antigens obtained from various other cells. of self-reactive thymocytes, how does this allow for the de-
Medullary DCs can cross-present antigens expressed by velopment of Treg cells? While the mechanism is not fully
mTECS; as stated previously, Aire expressed in mTECs plays understood, recent experiments have revealed many details
an important role in the antigen transfer. DCs are also ca- of this process.
pable of acquiring antigens from the periphery, migrating Sequencing of TCRs from transgenic mice that can gen-
to the thymic medulla for antigen presentation to devel- erate limited TCR diversity has shown that the regulatory
oping T cells and inducing apoptosis in antigen-specific and conventional cells have equally diverse but largely dis-
T cells. Thus, DCs in the medulla have access to antigens tinct repertoires.245–247 These experiments seem to suggest
obtained in the periphery and indirectly from promiscu- an instructive role for TCR signals in the development of Treg
ous gene expression by mTECs, as well as those they express cells.248,249 However, competition between developing T cells
themselves, for presentation to T cells. with different TCRs and a limiting supply of selecting ligands
In addition to their extraordinary capacity for acquir- may also play a role. Transgenic mice that only express a sin-
ing and presenting self-antigens, mTECs and DCs also gle TCR that was isolated from a mature naturally occurring
express costimulatory molecules such as B7 and CD40. Treg develop only very low numbers of Treg cells. However,
CD28/B7 or possibly CD40/40L interactions appear to play when these TCR-transgenic bone marrow cells were injected
an important role in negative selection, although neither into Rag knockout mice together with wild-type precur-
is absolutely required for deletion. It is possible that in the sor cells, Treg frequency was inversely proportional to the
absence of these molecules, very strong TCR signals and/or fraction of injected cells that were transgenic. Thus, intra-
other costimulatory molecules may be able to provide the clonal competition within transgenic thymocytes for rare
additional signal.6,235,236 selecting ligands could limit normal Treg differentiation.250
Furthermore, Nur-77 GFP induction in individual cells was
also shown to be inversely proportional to the fraction of
Treg Selection as an Alternative transgenic cells in a similar competitive assay as previously
Treg cells are a naturally occurring population of rare Foxp3 + discussed.163 This indicates that low number of thymocytes
CD25 + thymocytes that are activated by self-antigens in the expressing a particular self-reactive TCR normally facilitates
periphery and function to suppress autoimmune responses their interaction with rare Treg-selecting ligands, and that
in a dominant fashion. Foxp3 is the lineage-determining fac- availability of selecting ligands seems to be another impor-
tor for these cells, and its expression in conventional Foxp3- tant factor in Treg development. Finally TCR avidity (TCR
peripheral CD4 + T cells lead to their differentiation into affinity × number of interactions) may also play a role in Treg
functional Treg cells. Foxp3, a forkhead winged-helix tran- differentiation, as reducing the self-antigen–presenting ca-
scription family member, was initially identified as the mu- pacity of mTECs by knocking down the expression of MHC
tated gene causing a severe autoimmune disease in humans, II transcriptional activator CIITA resulted in decreased ef-
immunodysregulation, polyendocrinopathy and enter- ficiency of negative selection but an increase in the number
opathy, X-linked syndrome, and a similar disease in scurfy of regulatory T cells.251
mice. These mutations, as well as knockouts of the Foxp3 Putting several lines of evidence together, a two-step
gene, result in complete loss of Treg cells.237–241 Although the model was proposed for intrathymic Treg development252,253
cortex can fully support the development of Treg cells, it is (Fig. 13.9). The simultaneous delivery of TCR and CD28 sig-
thought that Treg-cell differentiation correlates temporally nals to CD4 + Foxp3 − Treg precursors advances their devel-
and spatially with negative selection in the medulla.225 opment to the CD4 + CD25 + Foxp3 − stage. This stage is the

CD80/86:CD28
+ TGFβ + IL-2
CD25+ CD25+
CD4 SP Foxp3- Foxp3+
pMHC:TCR
- TGFβ ?
DC/mTEC
apoptosis apoptosis
FIG. 13.9. Development of Regulatory T (Treg) Cells. Surface receptors and cytokines (blue) important at each step in the
two-step model of Treg development are shown.248,249 The efficiency of Treg selection and the time and location of initial
major histocompatibility complex contact on Treg maturation are not known; although Treg cells are found mainly in the
medulla, the cortex has also been shown to fully support Treg development.282 Additionally, the question of whether and
how much negative selection and Treg developmental pathways overlap also remains unanswered.
immediate thymic precursor to the fully mature Foxp3 + Treg to produce effector cytokines including IFNγ, IL-17, and
cells. CD28 expression, CD28-enhanced Lck recruitment, IL-4 etc. is restored 265–267 Thus akin to the role of Th-POK in
and downstream signaling in the CD4 SP cell are impor- conventional CD4 + T cells, Foxp3 acts as a dominant factor
tant for this first step because deletion of CD28 or its ligands in the maintenance of the regulatory T-cell lineage.
B7-1 and B7-2 leads to a severe reduction in Treg numbers
and proportion.254–256 The NF-κ B subunit c-Rel, which is ac-
“AGONIST SELECTION” OF INNATE-TYPE
tivated by TCR and CD28 signaling, binds directly to the
Foxp3 promoter and a conserved enhancer element, CNS3
LYMPHOCYTES
(conserved noncoding DNA sequence 3), implying a direct Distinctive Features of Innate-Type Lymphocytes
transcriptional role.257–261 c-Rel also appears to promote the The precedent of Treg selection shows that strong TCRαβ
earliest steps in Treg development even before Foxp3 is de- interactions with agonist ligands can promote something
tectably turned on.262 The survival of cells that have received other than death in the thymus, parallel to the TCRγδ
the TCR-CD28 signal also depends on TGFβRII-mediated lineage maturation programs that are triggered by strong
signaling, which rescues these intermediates from Bim- agonist-mediated TCR signaling. However, the special con-
mediated apoptosis.221 ditions of DP cell physiology do not apply to DN2-DN3
At this stage, these cells require a second signal through thymocytes becoming TCRγδ cells nor to medullary CD4
cytokines like IL-2, IL-7, and IL-15 that signal through SP cells becoming Treg cells. Therefore, it has still been a
the γc chain–containing receptors, with IL-2 playing the surprise that a major class of T cells, the “innate-type” cells
major role in the final maturation to Foxp3 + Treg stage (see that include NKT cells, are also positively selected by strong
Fig. 13.9). Stat5b, a major transcription factor activated by agonist ligand and selected apparently from the DP popula-
signaling of the γc cytokine receptor chain, binds directly tion itself.
to the promoter and CNS2 of Foxp3, and mice expressing The distinguishing features of iNKT cells are a pairing
a transgenic, constitutively active version of Stat5 show a of canonical effector function with canonical TCR rear-
dramatic increase in Treg numbers and percentage. This rangement and recognition specificity.268 These cells use a
cytokine-dependent final maturation step is thought to be limited set of possible Vβ rearrangements—Vβ2, Vβ8.2, or
largely independent of TCR stimulation.263,264 Once induced, Vβ7 (Vβ11 in humans)—together with an invariant pair-
Foxp3 suppresses alternate lineage genes and permanently ing of Vα14 (Vα24 in humans) with Jα18. From the start,
establishes a gene expression pattern characteristic of Treg they appear to observe a different set of rules from most
cells. Although Foxp3 is not required for commitment to TCRαβ cells. Although their positively selecting ligands are
the Treg lineage, its continuous expression is important for nonclassical class I MHC rather than class II MHC, they
Treg function as a loss of its expression in mature Treg cells predominantly become CD4 + or revert to a CD4 − CD8 −
leads to a loss of their suppressor function and their ability phenotype rather than becoming CD8 SP. They acquire a

chronically activated phenotype marked by CD122 (IL2Rβ) signals through a distinctive pathway. Whereas conven-
expression, CD44 expression, and NK receptor expression tional positive selection is mediated by cTECs presenting
in a stepwise process from the DP stage, starting with the peptides on classical class I and class II MHC antigens, NKT
first cells that acquire the canonical receptors. Even within cell selection is based on recognition of glycolipids pre-
the thymus, these cells are spontaneously triggered to ex- sented on nonclassical class I MHC antigens that are only
press cytokines that normally require multiple rounds of expressed by hematopoietic cells, not thymic epithelial cells.
activation to be induced, IL-4, and IFNγ, and they can ex- In general, the main APCs for NKT cell selection are other
press them simultaneously, combining Th2- and Th1-like DP thymocytes themselves. The homotypic cell-cell inter-
functions. action that this entails also makes it possible for the cells
to use a special kind of coreceptor, receptors of the SLAM
family (mostly Slamf1 = CD150, or Slamf6 = Ly108), to me-
Promyelocytic Leukemia Zinc Finger and the Id-Driven diate homotypic adhesion between the future iNKT cell
Maturation Pathway and its developmentally similar APC. The engagement of
DP cells need to survive in a DP state for a longer-than- SLAM family receptors recruits an adaptor that is not used
average amount of time in order to acquire the ability to in conventional positive selection, SLAM-associated pro-
express the invariant iNKT TCR.269 The particular TCRα tein (Sh2d1a), which efficiently couples TCR signaling with
segments used in the canonical receptor are among the last Fyn tyrosine kinase activation and protein kinase C . This
ones that most DP cells use for rearrangement, because of is a fast track to NF-κ B activation,273 which may even occur
the 5′ position of Vα14 and the 3′ position of Jα18. In any preferentially with respect to conventional Ras/MAP kinase
mutant that shortens DP lifespan (eg, loss of RORγ t, HEB, pathway activation.274,275 The potentially altered quantitative
BclXL), the iNKT TCRα chain does not have a chance to be balance between signaling pathways is a likely origin for the
made.268,270 Once the receptor is expressed, however, and if its distinctive program that iNKT selection activates.
ligand is present in the thymic environment, the cells embark iNKT cells are only the best-studied representative of a
on their new program. In the process, they will make use of number of distinct agonist-selected innate-like T-cell lin-
the signature transcription factors of Th1, Th2, and NK cells, eages. It appears that other subsets share many aspects of the
as well as a canonical regulatory factor of their own, the zinc iNKT selection pathway, including use of PLZF and a depen-
finger-BTB domain factor PLZF (or Zbtb16). This is the cru- dence on SAP and Fyn that separates all of them from most
cial regulator that confers on them an activated CD44-high conventional T cells.273,276 Thus, this is an alternate modality
phenotype as well as their distinctive “ready-to-go” priming for positive selection that violates the simple picture that fate
for effector cytokine production,271,272 the same factor that is might be determined only by TCR avidity and developmen-
also required by the “innate-like” Vγ1.1/Vδ6.3 γδ cells. tal stage of contact.
DP cells, upon expressing the invariant iNKT receptor,
embark on a pathway first resembling standard positive se-
lection to a CD4 SP lineage, including use of Runx1, GATA- CONCLUSION
3, and the transcription factor TOX. In the process, they The remarkable diversity of mature T-cell types and their
downregulate CD24 normally, but activate a “survival-asso- potential for long proliferative lifespans offers a challenge
ciated” pathway module including Egr2, Myc, and NF-κ B to understanding their development in terms of a single
activation, which drives a burst of proliferation (unlike the commitment pathway. T-lineage commitment is a process
conventional positive selection program). The proliferation that instills a profoundly durable and renewable cell type
and emergence of the NKT cells is dependent on IL-15 and identity via stable expression of multiple T-cell genes, but
CD122 (IL-2/IL-15Rβ), unlike the conventional SP cells that enables the cell to keep exploring different variants of func-
primarily express IL-7 receptors. Following proliferation, tion. Past views of T-cell development centered on the need
the cells turn on PLZF expression, which endows them with to develop a diverse and self-tolerant recognition reper-
effector function. After this, the cells can turn on T-bet, the toire, but increasing appreciation for the sophistication of
NK-, CD8-, and Th1-associated factor that promotes IFNγ innate immunity casts the T-cell agenda in a different light.
expression and enables them to acquire their distinctive NK Signaling does not need to be dependent on a single recep-
receptor–positive phenotype. With the completion of the tor in a plus/minus way, nor is the TCR unique in its ability
iNKT program, the cells leave the thymus, turning on Id2 to transmit quantitatively graded signals; innate immune
that maintains their homeostasis in the periphery. cells use integrative signaling to adapt to various circum-
stances very effectively without clonally unique receptors.
T-cell development itself produces many lineages that use
Special Selection Mechanism: Hematopoietic stereotyped, practically invariant TCRs, and as these cells
Antigen-Presenting Cells and Distinctive Signaling must consistently be positively selected by agonist in the
The iNKT pathway offers a striking view of how use of thymus, it follows that the receptors that evolution has con-
distinctive signaling mediators to interpret TCR signaling served for use by these cells must be self-reactive ones. Seen
can instructively determine the developmental outcome. in this light, many T cells like innate cells must be impor-
The difference between positive selection of iNKT cells tant for the functions they link to their TCR specificities
and other conventional TCRαβ cells starts with the APCs and not only for the diversity of non-self–structures that
they interact with and also includes the channeling of TCR they can tell apart.

This chapter has touched on a number of cases where temporal resolution, in order to understand their different
parts of the mechanism that endow T cells with a func- outputs.277,278 But the outlines of the answers are coming
tion to match their specificity can be discerned. However, into view.
these are mostly at the frontiers of current understanding.
Simple knockout approaches can be foiled by the way a key
signaling molecule or transcription factor is often shared ACKNOWLEDGMENTS
between different developmental contexts where it may play The authors gratefully acknowledge the important intellec-
different roles. Stage-specific analysis is crucial to reveal tual contributions to the development of this chapter from
the regulatory combinations that are actually guiding each Dr. Mary A. Yui. Support was from NIH grants CA090233,
developmental decision. New tools will be needed to dis- AI095943, and AI083514, and from the Albert Billings
sect the operation of signaling pathways with quantitative, Ruddock Professorship to E. V. R.

Peripheral T Lymphocyte Responses

CHAPTER
14 and Function
Marc K. Jenkins
INTRODUCTION occurred on one chromosome, further rearrangement on the

This chapter focuses on the processes by which conven- other chromosome is suppressed, although not completely.
tional thymus-derived (T)-lymphocytes that express αβ These processes lead to the situation where most developing
T-cell antigen receptors (TCRs) become activated by their thymocytes express a single unique type of αβ TCR.
cognate peptide (p): major histocompatibility complex Thymocytes that successfully express a TCR must then
(MHC) ligand. It will cover how TCR ligation by foreign pass positive and negative selection based on the p:MHC
antigen-derived p:MHC ligands causes rare naïve T cells specificity of that TCR.4 Positive selection occurs just after
from the preimmune repertoire to proliferate in the sec- completion of TCR gene rearrangement at the time when
ondary lymphoid organs (sometimes referred to by immu- thymocytes first express TCRs and coexpress both cluster of
nologists as the periphery) to become effector cells capable differentiation (CD)4 and CD8 coreceptors. To pass positive
of producing microbicidal cytokines and eliminating for- selection, a thymocyte must receive a TCR signal transduced
eign antigens from nonlymphoid tissues. It will also de- by low affinity binding to a self p:MHC ligand on radio-re-
scribe how the innate immune response and signals from sistant thymic epithelial cells in the cortex. It was thought
costimulatory and cytokine receptors shape the functional that this process was inefficient due to the improbability
and migratory properties of effector T cells, and ends with a of VDJ segment rearrangement by chance producing TCRs
discussion of how effector cells give rise to diverse memory with any affinity for a self p:MHC ligand.5 However, recent
cell subsets. An emphasis is placed on the dynamics and work from Marrack and colleagues has shown that all TCRs
anatomy of naïve to effector to memory-cell conversion. have germ-line encoded complementarity determining re-
The chapter will focus on conventional T cells that ex- gion (CDR) 1 and CDR2 domains within their V segments
press αβ TCRs. The reader is directed to other chapters in that bind MHC molecules.6 This binding has been postu-
the book for details concerning natural killer T cells, γδ T lated to cause most thymocytes to undergo negative selec-
cells, and regulatory T cells. tion,7 which is a desirable outcome for the host because these
cells could cause autoimmunity if allowed to mature. In this
model, the only thymocytes that receive the weak positive
NAÏVE T CELLS selection signal are those that produce a CDR3 through VDJ
To understand effector T-cell generation, it is important to recombination that partially impairs CDR1 and 2 binding
consider the naïve T cells from which they are derived. The to MHC and allows weak binding to a self-p:MHC complex.
capacity of the host to make effector and memory cells in Cells that undergo positive selection by recognition of a self-
response to new foreign proteins depends on the existence of p:MHCI ligand lose CD4 and retain CD8, whereas cells that
naïve T cells that express αβ TCRs specific for MHC-bound undergo positive selection by recognition of a self-p:MHCII
peptides derived from that protein.1 Naïve T cells with this ligand lose CD8 and retain CD4, in a process controlled by
particular p:MHC specificity are rare members of a vast ThPOK and Runx3 transcription factors.8 In contrast, thy-
repertoire also containing naïve T cells specific for other mocytes that undergo negative selection either die by apop-
p:MHC ligands. Each naïve T cell expresses a randomly tosis or differentiate into Foxp3 + regulatory T cells.9
generated TCR, which is produced regardless of whether the
relevant foreign antigen is in the host’s body. Thus, naïve T
cells are defined as T cells that have not yet been stimulated
Recirculation
by a p:MHC ligand for which their TCR has a high affinity. CD4 + and CD8 + T cells that complete positive selection
exit the thymus and enter the secondary lymphoid organs
(lymph nodes, spleen, and mucosal lymphoid organs). Just
Generation before leaving the thymus, these cells turn on the KLF2
A naïve T cell is the end result of a complex developmental transcription factor, which promotes expression of several
process that occurs in the thymus.2 To complete develop- gene products that control thymus exit, such as S1P1, and
ment, each thymocyte must produce an αβ TCR by ran- subsequent circulation through secondary lymphoid organs,
domly recombining one of multiple variable (V), diversity such as CD62L and CC chemokine receptor (CCR) 7.10 T
(D), and joining (J) deoxyribonucleic acid segments at the cells also express low levels of CD44 and CD45RO and high
Tcrb locus and one of multiple V and J segments at the levels of CD45RB and CD45RA as they leave the thymus.
Tcra locus.3 Once an in-frame rearrangement at a locus has Over the first 2 weeks after leaving the thymus in adults,
355

T cells increase the levels of expression of CD28 and the in- Self p:MHC ligand presentation by dendritic cells has dif-
terleukin (IL)-7 receptor α chain, and become better able to ferent consequences for naïve T cells than foreign p:MHC
proliferate and differentiate if confronted with the relevant ligand presentation. For example, self p:MHC ligand pre-
foreign p:MHC ligand.11,12 After completing this maturation sentation results in only a subset of the signals that ema-
process, T cells express a CD44low CD45RO − CD45RBhigh nate from the TCR when bound by a high affinity foreign
CD45RA+ CCR7+ CD62L + CD28high IL-7Rα high phenotype, p:MHC ligand, including partial phosphorylation of the
which is used to identify naïve T cells. TCR-associated CD3-zeta chain.20 In addition, although
Naïve T cells spend their lives (T1/2 of 2 to 5 years in hu- signals through the TCR and IL-7 receptor are required for
mans,13 and 50 to 100 days in mice14) recirculating through the survival of naïve T cells,21 these signals do not cause the
secondary lymphoid organs. Expression of trafficking mol- T cells to proliferate in hosts containing normal numbers of
ecules that bind ligands that are only expressed on the spe- T cells due to competition for IL-7. The Tsc122,23 and Foxp124
cialized blood vessels and sinuses of these organs explain transcription factors, the second of which regulates IL-7Rα
this behavior. For example, the extravasation of naïve T levels, enforce the quiescent survival of naïve T cells in this
cells through the high endothelial venules (HEVs) of lymph situation. In contrast, naïve T cells proliferate when the num-
nodes and mucosal lymphoid organs depends on CD62L ber of T cells is very low and IL-7 becomes more available,
and CCR7, the ligands for which are only expressed on for example, early in life or after radiation or chemotherapy.
HEVs.15 After passing through the HEVs, naïve T cells enter This “homeostatic” proliferation also depends on IL-7 and
the T-cell–rich paracortical areas of the lymph nodes and low-affinity TCR recognition of self p:MHC complexes,21
mucosal lymphoid organs and are constrained in these re- but differs from proliferation in response to foreign p:MHC
gions by sensing CCL19 and CCL21 with CCR7.16 Similarly, ligands by being independent of the CD28 costimulatory re-
naïve T cells migrate from the marginal zone sinuses of the ceptor.25 Thus, the same signaling events that cause naïve T
spleen into the T-cell–rich periarteriolar lymphoid sheathes cells to survive in interphase in T-cell–sufficient hosts cause
(PALSs) via an unknown G-protein–coupled receptor- these cells to proliferate in T-cell–deficient hosts, but using
dependent mechanism. Naïve T cells again migrate within a program different from that engaged during the T-cell re-
the PALSs through CCR7 sensing of CCL19 and CCL21. In sponse to high affinity TCR ligands.
all secondary lymphoid organs, naïve T cells are excluded Survival and proliferation could both contribute to con-
from the B-cell–rich follicles due to lack of expression of trol of the number of naïve T cells in normal hosts. In adults,
CXC chemokine receptor (CXCR) 5, which binds to CXCL13 new naïve T cells are exported from the thymus into the sec-
produced in follicles and guides cells to this location. ondary lymphoid organs that are already full of other T cells.
Naïve T cells survive in interphase under these conditions.
However, in neonates26 and perhaps the very aged, new naïve
Survival T cells enter relatively lymphopenic secondary lymphoid or-
The T-cell areas within lymph nodes contain a network of gans and undergo homeostatic proliferation to fi ll the space.
thin collagen tubes called conduits, which carry lymph- Notably, this proliferation causes the T cells to lose many
borne antigens and chemokines from the subcapsular sinus, of the markers that define the naïve phenotype and express
where lymph enters the lymph nodes through afferent lym- surface molecules that are characteristic of memory cells.21
phatic vessels, to sinuses surrounding HEVs.17 The conduits Thus, although cells with the naïve phenotype dominate
are wrapped with fibroblastic reticular cells, which produce the preimmune repertoire, it can contain some cells with a
IL-7. Naïve T cells migrate along the conduits, placing them- memory cell phenotype.
selves in a good position to receive IL-7, on which they de-
pend for survival.18
To survive for a normal lifespan, naïve T cells must also Abundance
receive TCR signals through weak recognition of self p:MHC An adult mouse has about 108 αβ TCR+ T cells in the
ligands, perhaps the ones that caused the T cells to undergo secondary lymphoid organs and another 5 × 106 in the blood.1
positive selection in the thymus.1 Dendritic cells are prob- The naïve T-cell population is split about 60:40 in favor of
ably the important antigen-presenting cell (APC) for this CD4 + T cells over CD8 + T cells. About 70% of the cells in
process because they are constantly in contact with T cells in both subsets are naïve phenotype cells, at least in young mice
the secondary lymphoid organs, and expression of MHCII housed under specific pathogen-free conditions. Thus, an
molecules under the control of the dendritic cell–specific adult mouse has about 7 × 107 naïve phenotype T cells in its
CD11c (Itgax) promoter is sufficient to maintain the sur- body, which extrapolates to about 3 × 1011 naïve phenotype
vival of naïve CD4 + T cells. On the other hand, mice lack- T cells in an adult human. Because the number of potential
ing dendritic cells by Itgax promoter–directed expression TCR amino acid sequences that could be produced by VDJ
of diphtheria toxin have normal numbers of naïve T cells.19 recombination (> 1015)27 is greater than the number of naïve
This result does not rule out a role for dendritic cells in naïve cells in the body, it is possible that each naïve T cell has a dif-
T-cell maintenance, however, because it is possible that the ferent TCR. Thus, it is possible that each of the 7 × 107 naïve T
naïve T cells die at a higher than normal rate without den- cells in a mouse has a TCR capable of binding to one of 7 × 107
dritic cells but are also produced at a higher rate. Studies on different foreign p:MHC ligands. This would mean that the
the turnover rate of naïve T cells in dendritic cell–deficient frequency of naïve cells expressing TCRs specific for a single
mice will be needed to address this question. p:MHC ligand could be as low as 1/7 × 107 naïve T cells.

CHAPTER 14 PERIPHERAL T LYMPHOCYTE RESPONSES AND FUNCTION | 357
The actual number of naïve T cells specific for single from the body. These more dramatic biological effects occur
foreign p:MHC-specific ligands has been measured by flow because the responding T cells receive much stronger or
cytometry following labeling with fluorochrome-labeled more durable signals through the TCR upon recognition
pMHC tetramers and enrichment with anti-fluorochrome of high-affinity foreign p:MHC ligands than they do when
antibody-coated magnetic beads. C57BL/6 mice contain recognizing low-affinity self p:MHC ligands.35 In addition,
about 200 naïve CD4 + T cells specific for an immunogenic foreign antigens naturally enter the body during infection
peptide called 2W bound to I-Ab but only about 20 cells spe- or tissue damage, which triggers dendritic cells and other
cific for peptide 427–441 from the FliC protein of Salmonella innate immune cells to stabilize foreign p:MHC ligands and
typhimurium bound to I-Ab.28 Thus, the frequencies of cells induce costimulatory receptors and cytokines.36,37 The na-
capable of binding to these foreign p:MHCII ligands ranges ture of the signals from the TCR, costimulatory receptors,
from about 1:200,000 to 1:2,000,000 of the naïve phenotype and cytokine receptors then influences the type of effector
CD4 + T cells in mice that were never exposed to these pep- cells that naïve T cells differentiate into. These three signals
tides. Similar analyses have been done for foreign p:MHCI- and how they vary depending on the nature of the antigen
specific CD8 + murine T-cell populations.29–31 The number will be described.
of naïve CD8 + T cells in C57BL/6 mice ranged from 15
lymphocytic choriomeningitis virus (LCMV) L338–346 :Db-
specific cells to 1,100 vaccinia virus B8R:K b-specific cells Dendritic Cells as Initiating Antigen-Presenting Cells
per mouse.31 Arguably, the most important of the three signals is trans-
An important conclusion from these studies is that naïve duced by the TCR pursuant to bind to a foreign p:MHC
populations specific for different foreign p:MHC ligands ligand on an APC. It is clear from ablation experiments
vary in size in a predictable fashion. There is evidence that that dendritic cells are the only APCs capable of initiating
size differences between foreign pMHC-specific populations the TCR signal in CD8 + and CD4 + T cells in the spleen.38
in the preimmune repertoire are caused by negative selec- Dendritic cells are also important for initiating APCs for
tion.32 In other words, naïve foreign pMHC-specific popula- CD4 + T cells in lymph nodes, although other MHCII + cells
tions can be small because their TCRs by chance bind with can do this job.38 It has also become clear that many den-
high affinity to self p:MHC complexes. Another possibility dritic cell subsets exist.38,39 A brief synopsis is presented here
is that peptides that are recognized by larger naïve popula- to give the reader a sense of which dendritic cell types func-
tions have structural features that are conducive to recognition as APCs for different antigens. Refer to Chapter 16 on
tion by a more diverse set of TCRs (eg, amino acids with dendritic cells for more detailed information on their biol-
prominent side chains).32,33 ogy. Dendritic cells are defined in the mouse as cells that
Another important conclusion from these experiments is express the CD11c integrin, have large amounts of MHCI
that the size of naïve p:MHC-specific populations can pre- and MHCII molecules, and reside in or have the capacity to
dict the magnitude of the effector cell response after certain migrate to the T-cell zones of secondary lymphoid organs.40
forms of antigen administration. For example, the presence The spleen and lymph nodes contain two types of dendritic
of about 200 2W:I-Ab- and 20 FliC:I-Ab-specific naïve CD4 + cells that develop from monocyte-dendritic cell precursors
T cells correlates with the 100,000 2W:I-Ab- and 15,000 in these organs. One type expresses the myeloid marker
FliC:I-Ab-specific effector T cells induced by injection of CD11b and is often referred to as the myeloid dendritic cell.
the relevant peptides.28 Similarly, the size of naïve LCMV These dendritic cells are found mainly in the red pulp or
p:MHCI-specific T-cell populations plays a role the mag- marginal zones of the spleen and outer T-cell–rich regions of
nitude of the primary CD8 + T-cell response to the virus. the lymph nodes. A second type depends on the Batf3 tran-
LCMV infection of B6 mice activates CD8 + T cells specific scription factor, expresses CD8α , and is referred to as the
for at least 28 different p:MHCI.31 However, about one-third CD8α + dendritic cell.41 These dendritic cells are located pri-
of the total response is directed against three p:MHCI com- marily in the PALSs of the spleen and the T-cell–rich regions
plexes. The feature of these dominant p:MHCI ligands that of the lymph nodes, and are the major IL-12–producing cells
correlates best with their potent immunogenicity is the large in these locations. Myeloid and CD8α + dendritic cells turn-
size of their naïve populations.31 Therefore, although anti- over rapidly with the latter population having only a 3-day
gen abundance, efficiency of peptide generation by antigen half-life.
processing, and MHC binding affinity are important factors Another type of dendritic cell is called the plasmacy-
in immunodominance,34 so is naïve T-cell population size. toid dendritic cell because of its morphology.42 These cells
develop in the bone marrow and then seed the spleen and
lymph nodes from the blood. Plasmacytoid dendritic cells
GENERAL ASPECTS OF EFFECTOR in mice express low amounts of CD11c and the B220 and
T-CELL FORMATION Gr-1 molecules normally expressed by B cells and granulo-
Unlike the presentation of low-affinity self p:MHC ligands cytes, and are the most potent producers of type 1 interfer-
that maintains the survival of naïve T cells in interphase ons (IFNs)43 when stimulated through pattern recognition
under nonlymphopenic conditions, the presentation of high- receptors (PRRs).44 These cells therefore play a key role in
affinity foreign p:MHC ligands induces the specific naïve T inducing an antiviral state in many cell types.
cells to produce lymphokines, proliferate, and differentiate The lymph nodes contain several additional migratory
into effector T cells that aid in elimination of the antigen CD11c + dendritic populations that move to this location

from tissues through afferent lymphatic vessels.39 All induction of the IL-2 receptor and is required for T-cell
lymph nodes contain CD11b + dendritic cells that also ex- acquisition of the capacity to cause a later delayed-type hy-
press F4/80 and SIRPα , distinguishing them from myeloid persensitivity reaction.
dendritic cells and other CD103 + dendritic cells that often Other types of antigens are accessed and processed by
express an intracellular protein called langerin. In the case different dendritic cells. Given their location in the spleen,
of the skin, these two types of dendritic cells migrate to myeloid dendritic cells probably play a key role as initiat-
lymph nodes from the dermis and are called dermal den- ing APCs for CD4 + T cells in the case of antigens that are
dritic cells and dermal langerin-positive dendritic cells. present in the blood.48 Dendritic cells that migrate from
The skin-draining lymph nodes also contain epidermal the relevant tissue (eg, submucosal dendritic cells during
Langerhans cells, which express langerin and high levels vaginal infection are likely the only cells capable of pro-
of CD11b and EpCAM. Epidermal Langerhans cells are ducing p:MHCII ligands from microbes that are too large
generated from a local radio-resistant precursor in the skin to enter the conduits.43 For unknown reasons, dermal
and are more long-lived than myeloid and CD8α + den- Langerin-positive dendritic cells are the most important
dritic cells. APCs for induction of contact hypersensitivity-causing
The dendritic cell type that is most important for the CD8 + T cells by protein-modifying chemicals applied to
presentation of p:MHC ligands depends on the nature of the skin surface.49 Surprisingly, epidermal Langerhans cells
the antigen. Resident CD8α + dendritic cells and CD103 + are not required for contact hypersensitivity, although the
dendritic cell migrants are critical for p:MHCI ligand pro- sensitizing chemicals are applied to the skin surface. Rather,
duction from exogenous antigens because these are the only p:MHCII presentation by Langerhans cells inhibits the in-
cells in the body that are capable of a process called cross duction of delayed-type hypersensitivity by suppressing the
presentation.41,45 These dendritic cells express receptors such priming of CD4 + Th1 cells and promoting the priming of
as CD36 that mediate uptake of apoptotic cells. After taking Th17 cells.50
up apoptotic or other extracellular material, these dendritic
cells have the unique capacity to move proteins from the
phagosome directly into the cytoplasm. Once in the cytosol, T-Cell Antigen Receptor Signaling
the translocated proteins can be cleaved by the proteosome Once dendritic cells displaying foreign p:MHC ligands
into peptides, which are then pumped by the transporter as- appear in the T-cell area of a secondary lymphoid organ,
sociated with antigen processing into the endoplasmic re- they can be recognized by naïve T cells expressing comple-
ticulum where binding to MHCI occurs. Other cells of the mentary TCRs. In vitro experiments have shown that high-
body cannot translocate ingested proteins into the cytosol affi nity TCR ligation by p:MHC ligands causes the TCR to
and thus are only capable of producing p:MHCI complexes concentrate in a stable central supramolecular activating
from proteins that are translated in their cytosols (eg, their cluster (cSMAC) structure at the point of contact between
own proteins or proteins from cytosolic microbes that infect the T cell and the APC.51 cSMAC formation is often pre-
them). Thus, the cross presentation pathway is important ceded by the formation of TCR microclusters at the pe-
for initiating the CD8 + T-cell response to viruses that do riphery of the T cell-APC contact zone, which are capable
not infect dendritic cells but kill their host cells. In contrast, of the transducing signals.52 TCR clustering then activates
any dendritic cell including plasmacytoid dendritic cells can protein tyrosine kinases such as lck, which stimulate sig-
likely serve as an initiating APC for CD8 + T cells specific for naling cascades that trigger protein kinase c theta, elevate
MHCI-binding peptides derived from proteins from viruses intracellular calcium, convert ras into its active form, and
that directly infect it. activate the extracellular signal-regulated kinases (ERK1
Early activation of CD4 + T cells specific for p:MHCII and ERK2) and stress-activated protein kinases (jun kinase
ligands derived from soluble antigens can occur in two and p38 mitogen-activated protein kinase).53 These path-
waves in lymph nodes.46 Soluble antigens (eg, toxins se- ways culminate in the nuclear translocation and binding of
creted by bacteria in a subcutaneous infection site) rapidly transcription factors such as NFAT and NF-κ B to deoxy-
flow through lymphatic vessels to the draining lymph nodes ribonucleic acid sequences that regulate lymphokine gene
and into the conduit network. Resident dendritic cells as- expression.54
sociate with the conduits, then take up the antigen, perhaps Very little is known about early TCR signaling events in
from small gaps between the fibroblastic reticular cells that naïve T cells in vivo because the assays used to measure most
encircle these tubes.47 These dendritic cells process the anti- of these events rely on cell lines and in vitro culture meth-
gen, produce p:MHCII complexes, and display them for rec- ods. Interestingly, in vivo imaging of naïve T cells showed
ognition by naïve CD4 + T cells. This process occurs within rapid, p:MHC-dependent TCR internalization that was not
several hours of antigen deposition in the skin and results in contingent on prolonged contacts between T cells and APCs
CD69 induction and proliferation in the T cells. or cSMACs.55 These transient interactions must be sufficient
The initial wave of p:MHCII presentation is followed by for TCR signaling, however, because intracellular staining
a second wave mediated by dermal dendritic cells that take with antibodies that recognize the active forms of the c-jun
up the antigen at the site of deposition and then migrate to transcription factor and the p38 mitogen– activated protein
the draining lymph node.46 These dendritic cells arrive in kinase showed that both of these molecules are phosphory-
the T-cell areas 12 to 24 hours after antigen enters the tissue. lated in p:MHC-specific naïve T cells in the spleen within
p:MHCII presentation by these dendritic cells prolongs minutes of intravenous injection of the relevant peptide.56

This rapid response is likely explained by the fact that the primary response.76,77 These molecules bind ligands of the
majority of naïve T cells are constantly contacting dendritic TNF family on the surface of APC and transduce signals that
cells in the T-cell areas.56 sustain the proliferation or survival of p:MHC-stimulated
T cells.
Enhancement of costimulatory signals may underlie the
Proliferation and Costimulation observation that in vivo T-cell proliferation is also influenced
Initial TCR binding to p:MHC ligands on dendritic cells by inflammation at the time of initial p:MHC presentation.
causes naïve T cells to begin proliferating in vivo about This effect is observed in the case of soluble antigens, where
2 days later.57 In vitro experiments indicate that this cell di- the magnitude of T-cell proliferation is several-fold greater
vision is driven by production of IL-258 and induction of the if antigen is administered with a microbial substance78,79
IL-2 receptor alpha chain (CD25).59 Surprisingly, however, containing a pathogen-associated molecular pattern43 such
p:MHC-driven proliferation of naïve T cells is minimally as lipopolysaccharide, which is recognized by a PRR such as
dependent on IL-2 in vivo.60,61 Therefore, other signals or toll-like receptor 4.80 PRR signaling stimulates tissue macro-
growth factors must be capable of driving T-cell prolifera- phages to produce tumor necrosis factor– α ,81 which in turn
tion in vivo, although IL-2 may contribute. The role of IL-2 stimulates dendritic cells to migrate from nonlymphoid tis-
as a T-cell growth factor may be difficult to reveal because sues into the T-cell areas and express higher levels of CD80
IL-2 is also required to maintain regulatory T cells62 and and CD86.82 In addition, these signals result in a maturation
can promote death of activated T cells.60 Thus, a reduction process that changes the antigen processing and presentation
in IL-2–dependent effector T-cell proliferation could be potential of all dendritic cell types.83 When inflammation is
masked by a removal of regulatory T-cell–mediated suppres- not present, dendritic cells efficiently engulf extracellular
sion and activation-induced cell death. In addition, although fluid and produce p:MHC complexes from the ingested pro-
IL-2 is not essential for initial p:MHCI ligand-driven T-cell teins. However, these p:MHC molecules turn over rapidly on
proliferation, it does appear to be essential for memory cell the dendritic cell surface and are presented in the context of
differentiation.63 low amounts of CD80 and CD86. These factors are thought
The initial proliferation of naïve T cells is followed by to be part of the explanation for why p:MHCII presentation
an exponential increase in the number of p:MHC-specific by dendritic cells in uninflamed secondary lymphoid organs
T cells over the next several days.64,65 Depending on the leads to poor T-cell priming and can result in tolerance.84
stimulus, the number of p:MHC-specific T cells reaches its In contrast, inflammatory signals cause dendritic cells to
highest level in the relevant secondary lymphoid organs, 5 reduce antigen uptake and processing, stabilize p:MHCII
to 7 days after antigen enters the body. As mentioned previ- molecules, and induce expression of CD80 and CD86.
ously, naïve mice contain about 200 CD8 + T cells specific for Therefore, inflammatory signals enhance proliferation by
a given p:MHC I complex. Because p:MHCI-specific CD8 + driving more dendritic cells into the T-cell areas to present
T cells can increase to 107 cells at the peak of the primary re- p:MHC ligands and by increasing the costimulatory capacity
sponse,64,65 it follows that CD8 + T cells can expand 500,000- of these dendritic cells.
fold in vivo. Although naïve CD4 + T cells are also capable PRR signaling in dendritic cells or macrophages can also
of dramatic clonal expansion when stimulated appropri- trigger the production of proinflammatory cytokines, which
ately, their burst size appears to be less than CD8-positive enhance effector cell proliferation. For example, IL-1 from
T cells.64 For both CD4 + and CD8 + T cells, the amount of several sources enhances the proliferation of CD4 + T cells
cell division is inversely proportional to the number of naïve through an early indirect effect on APCs,85 and a later more
precursors,14,66,67 indicating that in vivo proliferation is lim- potent direct effect on the T cells themselves.86 PRR signaling
ited by competition between p:MHC-specific T cells. causes CD8α + dendritic cells to produce IL-1241 and plasma-
In vivo T-cell proliferation is regulated by signals from cytoid dendritic cells to produce type 1 IFNs,42 both of which
the costimulatory CD28 molecule, which is triggered by enhance the proliferation of CD8 + effector T cells.87
binding to CD80 and CD86 on APCs.68 The proliferation of
antigen-stimulated CD4 + or CD8 + T cells is greatly reduced EFFECTOR T-CELL DIFFERENTIATION
in mice in which CD28 cannot interact with its ligands.
The proliferation of p:MHC-specific effector T cells during
CD40 ligand (CD154) deficiency also affects T-cell expan-
the primary response is linked with the acquisition of func-
sion,69 which may be related to the fact that CD40 signaling
tions that affect the elimination of antigen such as micro-
in APC induces CD80 and CD86.70 Although costimulatory
bicidal cytokine production and cytolysis. The functional
signals enhance TCR-driven IL-2 production, the in vivo
properties that effector cells acquire are influenced by the
significance of this effect for T-cell proliferation is unclear,
presence of inflammatory cytokines and costimulatory li-
as described above. Although it has been proposed that
gands on APCs present at the time of initial p:MHC presen-
CD28 acts by promoting TCR aggregation in the cSMAC,71
tation in secondary lymphoid organs.
enhancing lymphokine messenger ribonucleic acid produc-
tion72 and stability,73 and/or promoting T-cell survival by
augmenting Bcl-X L production,74 the bulk of recent evidence CD4+ T Cells
indicates that it acts by NF-κ B signaling.75 Members of the Many variations on this theme exist in the case of CD4 +
TNF receptor family, such as OX40, CD27, and 4-1BB are T cells. Effector cells that are generated in the presence of
induced in T cells by CD28 signaling several days into the IL-12, IL-4, IL-6 and TGF-β, or IL-6 and IL-21 become

IFN-γ –producing Th1 cells, IL-4–producing Th2 cells, and the cells acquire a CCR7+ CXCR5intermediate P-selectin
IL-17–producing Th17 cells, or IL-21–producing follicular ligand (PSGL-1) + PD-1− phenotype and live in the outer
helper cells (Tfh), respectively. These effector T cells play T-cell area. As discussed in more detail below, some of these
specialized roles in the elimination of intracellular mi- effector cells yield multipotent central memory cells.
crobes, worms, and extracellular microbes. Polarization of In contrast, if early CD25 − CXCR5 + effector cells receive
Th cells is covered in detail elsewhere in this volume. strong TCR and ICOS signals when interacting with B cells,
Signaling through the IL-2 receptor plays an essential and perhaps because of expression of TCRs with very high affin-
early role in the formation of Th1, Th2, and Th17 effector ity for the inducing p:MHCII ligand,91,104 then they become
cells. IL-2 receptor alpha chain (CD25) expression is induced CCR7low CXCR5high PSGL-1− PD-1+ Tfh cells. The cytokines
in naïve T cells by TCR signaling, which then pairs with IL-2/ IL-6 and IL-21 are also required for this process,105,106 with
IL-15 receptor β (CD122) and IL-2 receptor γ (CD132) chains, IL-21 probably coming from the developing Tfh cells them-
to produce the high-affinity IL-2 receptor.88 Under Th1 selves. The CCR7low CXCR5high phenotype causes Tfh cells to
priming conditions, the IL-2–bound receptor recruits Jak1 localize in germinal centers, which are rich sources of CXCR5
and Jak3 that phosphorylate the STAT5 transcription fac- but not CCR7 ligands.16 Here, Tfh cells produce IL-21 and
tor, which enters the nucleus and enhances expression of the CD40 ligand that promote germinal center B-cell differentia-
IL-12 receptor β2 chain and the T-bet and Blimp-1 transcription, antibody isotype switching, and plasma cell formation.98
tion factors.89 IL-12 receptor signaling through STAT4 then
enhances T-bet expression and enforces the Th1 fate,90 while
Blimp-1 suppresses another transcription factor Bcl6 that is CD8+ T Cells
needed for other fates.91 Under Th2 priming conditions, IL-2 As in the case of CD4 + T cells, IL-2 receptor signaling is not
receptor signaling through STAT5 regulates the Th2 cytokine essential for the initial proliferation of p:MHCI-stimulated
gene cluster and expression of the IL-4 receptor α-chain.89 In naïve CD8 + T cells.63 However, IL-2 receptor signaling plays
contrast, IL-2 receptor signaling inhibits Th17 differentiation a key role in effector cell differentiation and memory cell for-
by suppressing components of the IL-6 receptor.89 mation. Naïve CD8 + T cells that are stimulated by p:MHCI li-
A variation on this theme was recently described for gands on dendritic cells in secondary lymphoid organs during
CD4 + T cells during acute systemic Listeria monocytogenes acute LCMV infection rapidly express CD25 to produce high-
or LCMV infections.92–94 In this case, CD25 is induced by affinity IL-2 receptors.96 However, as in the case of CD4 + T
TCR signaling in microbe p:MHCII-specific CD4 + T cells cells, a subset of the CD8 + T cells sustains CD25 expression
1 day into the infection. However, by day 3, about half of longer than the other cells in the population. The CD25low
the responding T cells lose CD25 expression and begin to cells preferentially upregulate the IL-7 receptor and the cen-
express CXCR5, while half retain CD25 and do not express tral memory T-cell marker CD62L and produce long-lived
CXCR5.92,94 The mechanism underlying this bifurcation is memory cells. In contrast, the CD25high cells proliferate more
unknown, although asymmetric division of CD25 + mother rapidly and become apoptosis-prone terminally differentiated
cells into CD25 + and CD25 − daughter cells is a possibility.95 effector cells, although it is likely that some of these cells pro-
Another possibility is that very strong TCR signaling favors duce effector memory cells. Thus, for both CD4 + and CD8 +
CD25 expression.96 In any case, the CD25 + cells go on to T cells, a lack of IL-2 receptor signaling early in the primary
form T-bet high Th1 effector cells in the T cell areas, red pulp, response produces effector cell precursors of central memory
and probably nonlymphoid organs where they could pro- cells, whereas IL-2 receptor signaling promotes highly differ-
duce IFN-γ in response to p:MHCII presentation by mac- entiated effector cells and like effector memory cells.
rophages. It is likely that early CD25 + effector cells become Again like CD4 + T cells, costimulatory signals and cy-
Th2 cells during infections that promote IL-4 production by tokines produced by innate immune cells are needed for
innate immune cells.97 optimal differentiation of CD8 + effector cells. Naïve CD8 +
The CD25 − CXCR5 + cells formed early during acute in- T cells that are stimulated by p:MHCI ligands also require
fection yield two T-bet low effector cell populations,92,93 both CD28 signals and either IL-12 or type I IFN (IFNα / β) for
expressing the B-cell–follicle guiding CXCR5 receptor but maximal proliferation and development of cytolytic ac-
only one expressing the Tfh marker PD-1.98 Both of these tivity.87 The effects of IL-12 and type 1 IFN are mediated
CXCR5 + effector cell populations depend on the Bcl6 tran- directly on the CD8 + T cells. When either IL-12 or type 1
scription factor for their formation.92 The CXCR5 + PD- IFN are delivered with antigen, the responding CD8 + T cells
1− and CXCR5 + PD-1+ effector cell populations probably form a memory cell population, while immunization with
differentiate depending on quantitative differences in sig- antigen alone induces some clonal expansion, but very few
nals from the TCR and the inducible costimulator (ICOS) cells survive long-term and those that do are anergic. Thus,
receptor.68,94,99,100 Following stimulation by p:MHCII ligands IL-12 and type 1 IFN provide a “third signal” that is nec-
on dendritic cells, the naïve T cells that quickly lose CD25 essary, along with TCR and CD28 signals, for development
and express Bcl6 migrate toward the follicles under the guid- of optimal effector functions and formation of a responsive
ance of CXCR5 to interact with, and provide helper signals memory population. Dendritic cells activated by PRRs, or
to, antigen-specific B cells that display the relevant p:MHCII CD40 signals pursuant to interaction with CD40 ligand
complexes and ICOS ligand.101–103 In return, the T cells re- (CD154) + CD4 + T cells, produce IL-12 and type I IFNs, and
ceive signals from the TCR and ICOS. If these signals are are thus equipped to provide all three signals needed by a
relatively weak, then Bcl6 expression may not be maintained naïve CD8 + T cell to become a cytotoxic effector cell.

IL-12 and type I IFN are largely redundant in their most cell divisions in the secondary lymphoid organs. Thus,
capacity to support a productive CD8 + effector T-cell re- only the most-divided subset of Th1 effector cells in lymph
sponse, and which cytokine is critical depends on the infec- nodes acquire the appropriate trafficking receptors and have
tion. Studies using CD8 + T cells lacking receptors for one or the capacity to migrate to nonlymphoid organs.
both of the cytokines have shown that type I IFN is essen- The induction of skin-homing capacity is controlled by sev-
tial for response to LCMV,107 while IL-12 signaling plays the eral factors produced by dendritic cells that migrate from the
major role in responses to vaccinia virus and Listeria mono- skin and present p:MHCII ligands to naïve T cells in the drain-
cytogenes.108 In a minor histocompatibility antigen trans- ing lymph nodes. One of these factors is IL-12, which induces
plant model, rapid graft rejection depends on CD4 + T cells the expression of enzymes that fucosylate PSGL-1 and convert
stimulating dendritic cells in a CD40-dependent manner to it into the CD62P-binding form.118 Dendritic cells from the
produce IL-12 needed for generation of cytotoxic CD8 + ef- skin also produce 1,25 dihydroxy-vitamin D3, which induces
fector T cells.109 Thus, it appears that one way in which CD4 + the putative skin-homing chemokine receptor CCR10 on T
T cells can provide help for CD8 + T-cell responses is by cells.119 It should be noted, however, that regulation of CCR4,
stimulating dendritic cells to produce third signal cytokines. which is probably the most important skin-homing chemo-
CD8 + effector T cells differ from their naïve precursors kine receptor on T cells,120 is not understood in this context.
regarding surface markers, function, and trafficking prop- The generation of gut-homing CD4 + effector T cells is
erties. CD8 + effector T cells that are generated in second- also controlled by factors produced by dendritic cells at the
ary lymphoid organs during microbial infections express time of T-cell priming. Presentation of p:MHCII ligands to
slightly lower levels of CD8 and more surface O-glycans than CD4 + T cells in mucosal lymphoid organs induces effector
naïve cells,110 and in the human some CD8 + effector T cells cells that express large amounts of α4β7 integrin, which fa-
lose CD27 and CD28 but retain CD45RA.111 Unlike naïve cilitates T-cell migration into tissues containing mucosal ad-
cells, these cells express perforin and granzymes,110 which dressin cell adhesion molecule-1+ blood vessels such as the
are required for efficient cytolytic function. Expression of intestines.15 CD103 + dendritic cells in the mucosal lymphoid
perforin and granzymes contributes to the defining feature organs produce all-trans retinoic acid, which enhances the
of cytotoxic T cells,110 that is, the ability to directly kill tar- expression of α4β7 integrin and CCR9 on CD4 + T cells.121
get cells that display the appropriate p:MHC I complexes. Mycobacterium tuberculosis (TB) infection by inhalation
Interestingly, although large numbers of antigen-specific provides another example of CD4 + effector T-cell migration
CD8 + T cells accumulate in mice injected with heat-killed to nonlymphoid tissue.122 Inhaled TB organisms are taken
Listeria monocytogenes bacteria, these T cells do not acquire up in the lungs by CD11b + dendritic cells, which carry the
cytolytic function.112 This situation may come about because bacteria to the mediastinal lymph nodes and present TB
heat-killed Listeria monocytogenes bacteria are poor inducers p:MHCII ligands to naïve CD4 + T cells expressing relevant
of IL-12 by innate immune cells compared to live organisms. TCRs. This presentation occurs in an IL-12–rich context,
and Th1 effector cells are produced, many of which traffic
back to the lungs. The molecules that control this migration
EFFECTOR T-CELL FUNCTION IN NONLYMPHOID are unknown although CCL27, MIP-1α, IP-10, CCR4, CCR5,
ORGANS and CXCR3 may be involved. Once in the lungs, effector T
Some CD4 + effector T cells migrate into nonlymphoid tis- cells produce IFN-γ in response to TB p:MHCII presenta-
sues at the peak of the primary response.113 Separate homing tion by infected dendritic cells and macrophages, which
pathways to the intestines and skin have been characterized triggers the microbicidal functions of these cells. During
in detail,114 although other pathways probably exist. Injection this process, the lung effector T cells start out as PD-1high
of antigen and cholera toxin into the skin induces two popu- CD69high cells that convert to KLRG1high CD69low short-lived,
lations of CD4 + effector T cells in the draining lymph nodes terminally differentiated cells as they become activated by
that do or do not express the fucosylated form of P-selectin local TB p:MHCII presentation. Because TB organisms are
glycoprotein-1 (fPSGL-1), which binds to CD62P on inflamed not completely eliminated by this process, it goes on for pro-
blood vessels.115 The fPSGL-1− cells are potent helpers of anti- tracted periods of time as new effector cells migrate to the
body production by B cells and are likely one of the CXCR5 + lungs from the draining lymph nodes, eventually damaging
effector cell populations described above. In contrast, the the lung tissue. Thus, TB p:MHCII-specific effector T cells
fPSGL-1+ cells are poor helpers of antibody production by and their effects on infected phagocytes knock the TB organ-
B cells but are capable of IFN-γ production and are likely isms down but not out, and in the process protect the host in
similar to the T-bethigh Th1 effector cells induced during the relative short-term and damage it in the long-term.
acute infections. After transfer into naïve recipients, these CD8 + effector T cells also migrate out of the T-cell areas
cells migrate to the skin as evidenced by a capacity to cause and into many nonlymphoid tissues, particularly inflamed
delayed-type hypersensitivity reactions. Injection of antigen sites of antigen deposition (eg, the lungs during influenza
into the skin with complete Freund’s adjuvant also induces infection123,124 and the gut during vesicular stomatitis virus
Th1 effector cells in draining lymph nodes, which migrate infection125). Strong or prolonged IL-2 receptor signaling is
to the skin injection site in a CD62E- and CD62P-dependent an important factor in the generation of highly differenti-
fashion.116 Interestingly, in this case and another involving ated nonlymphoid tissue-homing effector CD8 + T cells.96
lung migration,117 the CD4 + T cells in the lymph nodes that The migratory capacities of effector CD8-positive T cells
migrate to nonlymphoid tissues are those that undergo the correlate with loss of receptors involved in lymph node

migration (CCR7 and CD62L) and acquisition of recep- cells, which are capable of rapid secondary responses that
tors such as α4β7 integrin,126 following instructions from can produce protective immunity to a later challenge from a
dendritic cells as described above for CD4-positive T cells. microbe.134 Memory cells can be distinguished from effector
The migration of CD8 + effector T cells with cytotoxic po- cells in that most memory cells are not blasts, are not in the
tential into nonlymphoid organs is an effective way of elimi- cell cycle, and many are not directly cytolytic or producing
nating cells that display p:MHC I complexes from all parts lymphokines.135
of the body. Memory cells are heterogenous, however, and exist in at
Several CD8 + effector T-cell–specific homing phenom- least two subsets: effector memory (Tem) cells and central
enon are worth mentioning. One is that CD8 + T-cell recruit- memory (Tcm) cells.136 Tem cells express homing receptors
ment to nonlymphoid tissue depends on prior entry of CD4 + T that facilitate migration to nonlymphoid sites of inflamma-
cells in certain situations. Recognition of p:MHCI complexes tion113 and produce microbicidal cytokines such as IFN-γ
on dendritic cells in the secondary lymphoid organs induces within several hours of TCR stimulation. In many ways,
the expression of CXCR3 on effector CD8 + T cells, which fa- these cells resemble effector cells (eg, Th1 or Th2 cells), with
cilitates entry into nonlymphoid tissues through blood vessels the exceptions that they are no longer blasts and express IL-7
displaying CXCL9 and CXCL10.127 During certain viral infec- receptors. Tcm cells do not produce any of the prototypic ef-
tions, CD4 + effector T cells first enter the relevant nonlym- fector cell lineage cytokines immediately after stimulation
phoid tissue and are stimulated by their p:MHCII ligands to through the TCR, although they secrete IL-2 and proliferate
produce IFN-γ, causing local epithelial cells to secrete CXCL9 extensively and acquire effector lymphokine production later.
and CXCL10, which then recruit the CD8 + effector T cells.128 These cells express CD62L and CCR7, which are involved
Another special homing property of some CD8 + effector in migration through lymph nodes and mucosal lymphoid
T cells relates to the capacity to enter a nonlymphoid tissue organs and positioning in the T-cell areas of these organs,16
and never leave.129 Viral infections in the skin, brain, and and IL-7 receptors, which are critical for their survival.133
intestinal mucosa result in expansion of viral p:MHCI-spe- The Tem and Tcm model fits well for CD8 + memory
cific CD8 + effector T cells in the draining lymph nodes and T cells. A subset of human CD8 + memory cells lacks the
then migration of some effector cells to the site of infection. naïve cell marker CD45RA but expresses CCR7.111,137 These
Some of these cells can be found at this site long after the memory cells also express CD62L and lack perforin and thus
infection is cleared. Parabiosis experiments indicate that would not be expected to be directly cytotoxic. Mice contain
these T cells are not constantly leaving and entering the a comparable CD44high memory cell population after clear-
site from the blood as expected, but rather remain in the ance of acute viral infection.125 These similar populations in
original site without leaving. These resident CD8 + T cells mice and humans fit the description of Tcm cells. In mice,
express CD103, which may tether them in the site by binding the CD62L + Tcm cells undergo slow IL-15– dependent,
to E-cadherin on local epithelial cells. Tissue-resident CD8 + MHCI-independent homeostatic proliferation, which is
effector T cells also express CD69, a marker of acute acti- thought to account for their numerical stability.138
vation, even under conditions where local relevant p:MHCI Mice and humans also contain CD45RA− CCR7−
complexes cannot be detected. Thus, CD69 expression may CD62L− CD8 + memory T cells. These cells express high
be driven in these cells by cytokine receptor rather than levels of β1 and β7 integrins, fPSGL-1, and CCR5, which are
TCR signaling. It should be noted that not all CD8 + effector predicted to facilitate migration into nonlymphoid tissues,139
T cells in nonlymphoid tissues are noncirculating residents and produce IFN-γ rapidly after TCR stimulation.125,137
as evidenced by the presence of p:MHCI-experienced CD8 + These cells therefore fit the description of Tem cells. In
T cells in efferent lymphatic vessels, which carry cells from mice, CD8 + Tem cells undergo much less IL-15–dependent,
tissues to the blood.130 The factors that determine whether MHCI-independent homeostatic proliferation than Tcm
naïve CD8 + T cells will become nonrecirculating CD103 + cells,140 and convert into Tcm cells over long periods of time
tissue-resident effector cells or effectors cells that recirculate in some141 but not all142 situations. Murine CD8 + Tem cells
through nonlymphoid tissues are not known. can recirculate through nonlymphoid organs or reside per-
manently in nonlymphoid organs, as described above.
Humans also contain a CD45RA+ CCR7− CD8 + T-cell
GENERATION OF MEMORY T CELLS FROM subset that contains especially high levels of perforin and
EFFECTOR CELLS direct ex vivo cytotoxic activity.111,137 These cells are likely ef-
Another key function of effector T cells is the production of fector cells that were recently stimulated by p:MHCI ligands.
memory cells.131 The number of antigen-derived p:MHC-spe- Memory CD4 + T-cell populations also contain Tcm
cific effector T cells in the body peaks about a week into the and Tem subsets. The microbial p:MHCII-specific CD4 + T
primary response, and then falls rapidly over a 2-week con- cells that survive in secondary lymphoid organs long after
traction period to about 10% of the peak value.132 This decline clearance of Listeria monocytogenes or LCMV infections
must be due to cell death because the total number of cells in consist of two subsets.92,93 One subset has the characteris-
all parts of the body declines shortly after the peak.113 The T tics of Tem cells including a T-bet high CCR7− phenotype,
cells that survive the contraction phase can persist stably for expression of nonlymphoid tissue homing receptors, and
the life of the host even when the antigen is cleared from the immediate IFN-γ production after stimulation with the
body due to the expression and function of IL-7 and IL-15 relevant p:MHCII ligand. These cells correspond to human
receptors.133 These long-lived cells are known as memory CD45RA− CCR7− CD4 + Tem cells, which express low or

variable levels of CD62L and high levels of fPSGL-1, and/or become memory cells,150 although some evidence against
β1 and β7 integrins and produce IFN-γ or IL-4 rapidly when this idea has accumulated.151,152 Thus, it is possible that 10%
stimulated with anti-CD3 antibody in vitro.137 In the mouse, of the effector cells present at the peak of the T-cell response
CD4 + Tem cells are also found in liver, lungs, and gut long retain the IL-7 receptor or some other survival factor, which
after intravenous injection of antigen plus adjuvant.113 allows these cells to become memory cells.
Because CD4 + memory T cells are constantly coming out Another possibility is that metabolic rate determines
of tissues and into afferent lymphatic vessels,130,143 it is likely whether an effector cell will survive to become a memory cell.
that CD4 + Tem cells are not fi xed in nonlymphoid organs Quiescent naïve and memory T cells produce adenosine-5′-
but recirculate through these sites. triphosphate by mitochondrial oxidative phosphorylation.153
The other microbial p:MHCII-specific CD4 + T-cell In contrast, activation causes effector cells to shift to glyco-
population that survives after clearance of acute infections lytic adenosine-5′-triphosphate production by a pathway that
in mice has the characteristics of Tcm cells. This population depends on mammalian target of rapamycin. Inhibition of
has a T-bet low CCR7+ CXCR5 + Ly6C − PD-1− Tfh-like phe- mammalian target of rapamycin with rapamycin early during
notype and produces IL-2 but not IFN-γ immediately after the primary response promotes the formation of CD8 + mem-
stimulation with the relevant p:MHCII ligand.92,93 These ory T cells by inhibiting T-bet expression in favor of another
cells have been found to be potent helpers of antibody pro- transcription factor, eomesodermin.154–156 This result fits with
duction by B cells.143–145 In humans, this population can give a model in which most effector cells are killed by products
rise to Tfh cells when stimulated in vitro.146 In mice, these of their own high rate of glycolysis through a T-bet–depen-
cells also generate Th1 cells and themselves during second- dent program, which when blocked allows a return to oxi-
ary immune responses and thus have the capacity of Tcm dative phosphorylation and memory cell formation. In this
cells to generate diverse effector cells.92 model, memory cells arise from effector cells with the lowest
Evidence suggests that Tem and Tcm memory cells derive glycolytic rates in the population. Again, increased glycolytic
from like effector cells. IFN-γ-producing CD62L − CD4 + rate and the favoring of effector cell death over memory cell
Tem cells that survive after clearance of acute LCMV infec- formation could involve IL-2 receptor signaling.
tion are derived from similar IFN-γ –producing CD62L− ef- Tumor necrosis factor receptor family proteins also pro-
fector cells present at the peak of infection.147,148 Similarly, mote CD4 + T memory cell survival by inducing antiapop-
the T-bet low CXCR5 + CCR7+ Tfh-like Tcm cells that survive totic factors by TRAF-dependent triggering of NFκ B.77 For
after clearance of Listeria monocytogenes infection are de- example, effector T cells that do not express CD27 survive
rived from similar T-bet low CXCR5 + CCR7+ effector cells poorly in vivo.157,158 Other work indicates that OX40 is dis-
present at the peak of infection.92 pensable for effector cell generation but required for mem-
Even though the various types of memory cells are derived ory cell formation.159
from effector cells, they must be subsets of their effector cell Recent work indicates that Id transcription factors also
precursor populations because they are 10 times smaller than regulate effector-memory cell conversion. Id2 is required for
those populations. In other words, even though Th1 effector the maximal generation of CD8 + effector T cells and per-
memory cells are derived from similar Th1 effector cells, the haps the formation of Tem cells.160–162 Conversely, Id3 ap-
former cells are 10 times less abundant than the latter cells. pears to be critical for CD8 + Tcm cell formation.162
Thus, to understand how memory cells emerge at the end
of the primary response, it is useful to consider the factors
that cause 90% of the effector cells to die. Effector cells pro- CONCLUSION AND MODELS
duce reactive oxygen species, which can cause cell death by What follows is an attempt to unify the information pre-
damaging mitochondrial membranes. Bcl-2 family proteins sented in this chapter into a theoretical sequence of events
regulate this type of death. Naïve and memory T cells express that occur after naïve CD4 + or CD8 + T cells encounter the
anti-apoptotic Bcl-2 molecules in their mitochondria, which relevant p:MHC ligands during skin or mucosal infections,
inhibit the formation of apoptosis-inducing complexes con- respectively. Because these processes are not completely un-
sisting of Bim and Bak. In mice overexpressing Bcl-2 or derstood, certain aspects of what follows are speculative.
lacking Bim, a greater fraction of the effector cell population Naïve T-cell populations that are specific for single
survive contraction and differentiate into memory cells.149 p:MHC epitopes exist at a frequency of about 1:200,000 in
Thus, it is conceivable that under normal circumstances, the the preimmune repertoire.1 These T cells, like other naïve
effector cells that are destined to become memory cells pro- T cells, spend their life spans (∼2 months in mice, ∼2 years
duce less reactive oxygen species or protect themselves better in humans) in a series of 1-day stops in the T-cell areas of
than other effector cells in the population. different secondary lymphoid organs with intervening trips
Another possibility relates to expression of the IL-7 re- through the blood. While in the T-cell area, these naïve T
ceptor. The survival of naïve T cells depends on IL-7 recep- cells receive survival signals through the IL-7 receptor as it
tor signaling, which maintains Bcl-2 expression and limits binds to IL-7 made by stromal cells and through the TCR as
Bim expression.149 TCR-driven activation causes effector it binds to the relevant selecting self p:MHC ligand on the
cells to lose the IL-7 receptor, leading to the fatal situation surface of an APC, probably a dendritic cell.
where Bcl-2 is reduced and Bim is increased. It has been Consider first the case of the CD4 + T-cell response to a
reported that a subset of CD8 + effector cells retains IL-7 bacterial infection in the dermis of the skin (Fig. 14.1). In
receptor expression and survives the contraction phase to this case, local dermal dendritic cells take up the microbe,

CCR10
Blood Ly6C Th1
vessel effector memory
T-bet
T-bet cell
IL-7R T-bet
fPSGL-1 exit
CCR4
Th1 effector cell enter

LV
SKIN
CCR10
IFN-γR IFN-γ
iNOS T-bet Blood

Skin Dendritic Cell MHCII:p TCR NFAT T-bet
T-bet vessel
enter
fPSGL-1
Macrophage
CCR4
MHCII:p
CCR7
Th1 effector cells LV

Lymphatic
LV
vessel exit Ly6C CCR10
IL-7R Th1
IL-2 T-bet
CD25 CCR10 T-bet effector memory
Ly6C IL-2/IL-15Rβ T-bet
IL-12 IL-12βR cell
IL-12βR STAT5 T-bet fPSGL-1
STAT4
Blimp1 CCR4 T-bet CCR4
T-bet T-bet
T-bet Bcl6
T-bet
CD25 PSGL-1
Vit D3
IL-12
Tcm precursor cells Tcm cell
IL-12Rβ2
CD25 IL-2 PSGL-1
CD62L PSGL-1
CXCR5
MHCII:p TCR NFAT CXCR5
CD62L
CD86 CD28 NFκB Bcl6 T-bet
Bcl6 IL-7R
B cell CXCR5 CD62L T-bet
CCR7
ICOSL ICOS CCR7
CCR7
CD4+ T cell CD62L Bcl6
enter
CCR7 T-bet exit
asymmetric cell
CCR7
division?
IL-12 IL-12βR
STAT4
Bcl6
HEV
T-bet
CD62L
LV
CXCR5
T Cell Area
Follicle
IL-21 IL-6 GC
LYMPH NODE IL-21R IL-6R
CXCR5
B cell
STAT3
Bcl6
Bcl6 Bcl6 TCR p:MHCII
MHCII:p TCR Bcl6 Bcl6 Bcl6
PD-1 AID
Ag-specific ICOSL ICOS
T-bet T-bet
CD154 CD40
Bcl6
B cell CXCR5
IL-21 IL-21R
CXCR5
Tfh cells
Germinal Center
FIG. 14.1. CD4+ T-Cell Response to Bacterial Infection of the Skin. See the text for description. Bacterial antigen is indicated with red rectangles.
Surface receptors are shown attached to the cell surface. Transcription factors are shown in the nucleus. Signals are shown with the smallest
arrows. Sequential cell fates are shown with the next largest arrows. Migration routes are shown with the largest arrows. LV, lymphatic vessel;
R, receptor.

CCR9 CD69 CD103

α4β7
Recirculating
Tem cell
T-bet T-bet Resident
T-bet T-bet
IL-7R T-bet T-bet Tem cell
exit
KLRG1
α4β7 CCR9 LV MUCOSAL
CD103+
MHCII:p TCR
GzB
T-bet
NFAT
Blood
vessel
TISSUE
Perforin
Dendritic Cell T-bet
enter
α4β7
apoptoic
CCR7 virus-infected
epithelial cell
CD103
LV LV
Cytotoxic effector cells
Plasmacytoid KLRG1 exit

dendritic cell T-bet
GzB T-bet IL-7R

IL-2
CD25 Perforin Id2 Tem
T-bet cell
IFNα/β IFNAR1 Id2 T-bet
STAT5
STAT1 Id2
CD25 Blimp1
T-bet
all trans T-bet Id3
retinoic acid T-bet
CD103 IFNAR1
CD25 IL-2
CD62L
MHCI:p TCR NFAT

CD86 CD28 NFκB
Id3 Tcm precursor cells
Id2
CCR7
M
C
H
CD8+ T cell
D
C
C
40
II: CD
Id3
D
p
86
IL-7R
C
Id2
D
TC 8
STAT1 Id3
15
Eomes
Tcm
R
IFNα/β IFNAR1
4
2
Eomes Id2
T-bet T-bet T-bet
Id3 Eomes
cell
Id2
CCR7
CCR7
CD4+ T cell enter exit

T Cell Area
HEV
LV
LYMPH NODE
FIG. 14.2. CD8+ T-Cell Response to Viral Infection of a Mucosal Tissue. See the text for description. Virus particles are indicated with hexagons.
Surface receptors are shown attached to the cell surface. Transcription factors are shown in the nucleus. Signals are shown with the smallest
arrows. Sequential cell fates are shown with the next largest arrows. A virally infected cell killed by a cytotoxic clust of differentiation 8–positive
T cell is indicated with a skull and cross bones. Migration routes are shown with the largest arrows. LV, lymphatic vessel; R, receptor.

produce microbe p:MHCII complexes,45 upregulate CD28 Bcl6.92 These Tcm cells recirculate through secondary lym-
ligands,163 begin producing IL-12164 and vitamin D3,119 and phoid organs like naïve cells.
migrate through an afferent lymphatic vessel to the drain- Now consider the CD8 + T-cell response to viral infec-
ing lymph node. Naïve T cells interact with these den- tion of a mucosal tissue (Fig. 14.2). Local CD103 + dendritic
dritic cells165 and those CD4 + T cells that express bacterial cells with cross-presentation capacity take up apoptotic
p:MHCII-specific TCRs receive TCR and CD28 signals56 infected cells,41 produce viral p:MHCI and p:MHCII com-
and induce the translocation of NFAT and NF-κ B to the plexes, upregulate CD28 ligands, produce all-trans retinoic
nucleus,54 which in turn drive expression of CD25, IL-2, acid,121 and migrate through an afferent lymphatic vessel to
Bcl6, and the IL-12 receptor β2 chain, and loss of the IL-7 the draining lymph node. Plasmacytoid dendritic cells are
receptor.61,89,166 The T cells then divide in response to IL-2 recruited to the lymph node and produce type I IFNs in re-
and unknown growth factors,57 perhaps asymmetrically,95 sponse to PRR recognition of viral pathogen-associated mo-
to produce CD25 + and CD25 − effector cell progeny.92,94 The lecular patterns.42 Interaction with CD154 + virus p:MHCII-
CD25 + cells activate STAT5 through IL-2 receptor signal- specific CD4 + T cells increases the activation state of the
ing,167 which further increases expression of the IL-12 recep- CD103 + dendritic cells through CD40 signaling.175,176 Naïve
tor β2 chain,89 which combines with the IL-12 receptor β1 virus p:MHCI-specific CD8 + T cells interact with these den-
chain, binds IL-12, and activates STAT4168 to amplify T-bet dritic cells and produce IL-2 and induce expression of CD25.
expression.169 IL-2 receptor signaling and STAT5 activa- Unknown T-cell growth factors and perhaps IL-2 then stim-
tion also stimulates Blimp-1 expression,170 which represses ulate the CD8 + T cells to divide to produce progeny that ex-
Bcl6,171 thereby enforcing the Th1 fate and preventing Bcl6- press small or large amounts of CD25.96 The CD25high cells
dependent cell fates. The Th1 effector cells also express experience IL-2 receptor signaling and STAT5 activation,
CCR10, CCR4, and fucosylated PSGL-1 due to activation in which stimulates Blimp-1 expression, which cooperates with
the presence of IL-12 and vitamin D3.114,118 Some of these ef- STAT1 activation from the type I IFN receptor to upregulate
fector cells then leave the lymph nodes through efferent lym- T-bet.177 T-bet and Blimp-1 drive the cells to become termi-
phatic vessels, enter the bloodstream, and migrate into the nally differentiated KLRG1+ cytotoxic effector cells loaded
infection site through skin blood vessels displaying CD62P, with granzyme B and perforin-containing granules.170,178
CCL17, and CCL27.116,119,172 After entering this site, the Th1 Blimp-1 represses another transcription factor Id3, which is
effector cells interact with infected macrophages displaying required for Tcm cell formation,162 thereby reinforcing effec-
bacterial p:MHCII ligands.173 TCR signaling then causes the tor cell formation. The effector cells also express α4β7 inte-
T cells to secrete IFN-γ, which triggers production of induc- grin and CCR9 due to activation in the presence of all-trans
ible nitric oxide synthase and nitric oxide as well as killing retinoic acid,121 and migrate to the site of infection through
of intracellular bacteria.173 Some of the Th1 effector cells in mucosal blood vessels displaying mucosal addressin cell ad-
the lymph nodes survive the contraction phase, re-express hesion molecule-1 and CCL25.179 After extravasating into
the IL-7 receptor, and become Th1 effector memory cells,147 this site, the cytotoxic effector cells interact with infected
which recirculate through the skin. cells displaying virus p:MHCI ligands. TCR signaling then
The early CD25 − progeny do not experience IL-2 receptor causes the T cells to expel granzyme B and perforin, which
signaling and STAT5 activation, do not upregulate the IL-12 enter the target cell and kill it.180 Some of the KLRG1+ effec-
receptor β2 chain, and are left with a low amount of T-bet. tor cells in the lymph node survive the contraction phase,
Blimp-1 is also not turned on, allowing the cells to retain perhaps through the action of Id2,161 and then either become
Bcl6 and express CXCR5. The CXCR5 + T cells then migrate CD103 − Tem cells that recirculate through the mucosal
toward the follicles.103 If they interact with an antigen-spe- tissue or CD103 + Tem cells that enter the tissue and never
cific B cell, then they receive TCR and ICOS ligand signals98 leave.129 The recirculating Tem cells may convert to Tcm cells
and produce IL-21, which binds the IL-21 receptor and acti- over long periods of time.141
vates STAT3.174 The combination of TCR, ICOS, and STAT3 The early CD25 − progeny do not experience IL-2 re-
signals then amplifies Bcl6 expression, stimulating the cells ceptor signaling and STAT5 activation and do not turn on
to express PD-1 and more CXCR5, and to commit to the Tfh Blimp-1.178 Without Blimp-1, the cells retain Id3 expression162
lineage. The Tfh cells then migrate into germinal centers, and are prevented from becoming terminally differentiated
interact with antigen-specific germinal center B cells, induce Tbethigh cytotoxic effector cells. Instead, type I IFN receptor
expression of CD154, which engages CD40 on the B cells and signaling facilitates expression of the Eomes transcription
stimulates them to express activation induced cytidine de- factor,156 retention of CCR7 expression, and survival as Tcm
aminase, and undergo antibody isotype switching and so- cells, which recirculate through the secondary lymphoid or-
matic hypermutation.98 If the early CXCR5 + T cells interact gans like naïve cells.
with B cells that are not specific for bacterial antigens, then
they receive an ICOS signal that maintains their survival.92
However, without additional TCR signaling, the T cells do ACKNOWLEDGMENTS
not further upregulate CXCR5, do not lose CCR7 or CD62L, I am grateful to Antonio Pagán, Matthew Mescher, and
and some of these cell become Tcm cells, eventually losing David Masopust for their helpful suggestions.

The Intersection of Innate and

SECTION
V Adaptive Immunity
CHAPTER
15 The Innate Immune System
Luke A. J. O'Neill
INTRODUCTION includes processes such as fever, aches and pains, and drowsi-
Every organism has to contend with infection by microbes. ness, which are thought to allow the infected host to rest and
Infectious agents such as bacteria, viruses, and fungi recuperate.2 They also present a disadvantage, however, be-
threaten health and viability by competing for nutrients or cause they leave the host vulnerable to predators and decrease
more dangerously by killing the infected host as a conse- reproductive potential. Adaptive immunity occurs later in
quence of infection or in order to spread to another host. the cycle of infection, typically becoming evident several
This threat occurs largely because microorganisms divide days after infection, and probably only after innate immune
at a faster rate than the host and because of the production defenses have been breached. It is needed for sterilizing im-
of toxins that the microorganism uses for its own viability. munity. The fact that memory is a key feature of adaptive im-
The job of the immune system is firstly to sense the danger munity, however (which is evident because of the expansion
posed and secondly to mount an appropriate response. This of specific B and T cells), means that upon re-infection, the
response can be to tolerate the microorganism or to resist it innate response is limited because the pathogen is more rap-
such that at best sterilizing immunity occurs. idly cleared and there are therefore fewer microbial products
The immune system is traditionally split into two catego- present to activate innate immunity. In addition, there is evi-
ries: innate and acquired (or adaptive) immunity. The term dence that adaptive immune cells actually inhibit the innate
“innate” refers to the feature that the sensors involved are immune response, which in the first infection acts in a nega-
encoded by genes that do not undergo any rearrangement tive feedback manner.3 In subsequent infections, the mem-
to generate variants. Adaptive immunity, on the other hand, ory cells probably inhibit innate immunity from the start of
refers to host defense proteins encoded by genes that un- the infection. In essence, therefore, it appears that adaptive
dergo rearrangement to generate great diversity. This rear- immunity evolved to promote sterilizing immunity and also
rangement process is not driven by the infectious agent, but to prevent innate immunity happening upon reexposure to
happens by genetic programming. Antigens then lead to an pathogen, given its potential disadvantages in terms of sys-
expansion of antigen-specific B cells or T cells that help clear temic inflammation. Figure 15.1 illustrates this principal.
the infection and importantly persist in the form of memory Previously thought to be crude and nonspecific, in-
cells. This means that when the host is re-infected, there are nate immunity is now known to involve a range of recep-
substantially greater numbers of the specific B- and T-cell tor families that recognize diverse microbial products.
populations that lead to clearance of the infectious agent. Importantly, these receptors program gene expression
There are therefore insufficient numbers of microbes to changes in target cells. The products of these genes drive
trigger substantial systemic symptoms. inflammation and are also required for adaptive immu-
It is estimated that more than 99% of life on earth only nity. These proteins include cytokines that promote expan-
has innate immunity.1 What advantage would the evolution sion of B and T cells. Other innate immune proteins have
of adaptive immunity bestow on an organism? The answer direct antimicrobial effects or act as opsonins, coating
lies in part in the nature of innate immunity. As will be dis- bacteria and promoting phagocytosis. Innate immunity
cussed, the effector mechanisms here involve the inflam- has therefore gone from being viewed as a process lacking
matory process, which when out of control can harm the sophistication, to one determining for both inflammation
host. In mammals, a “sickness behavior” also occurs, which and adaptive immunity.4
367
Paul_Ch15_final.indd 367 9/17/12 5:41 AM

368 | SECTION V THE INTERSECTION OF INNATE AND ADAPTIVE IMMUNITY
T lymphocyte
FIRST EXPOSURE TO PATHOGEN
FIG. 15.1. The Relationship between
Dendritic cell Innate and Adaptive Immunity. A:
Upon exposure to a pathogen, the
innate immune response is triggered
mainly by phagocytes and dendritic
cells. Inflammation ensues, which
B lymphocyte leads to sickness behavior and host
defense. The adaptive immune sys-
tem is then triggered leading to
1. Innate immunity activated 2. Adaptive immunity activated. 3. Memory cells emerge effector T- and B-lymphocytes, some
Inflammation and sickness behavior. Innate response limited. of which become memory cells.
RE-EXPOSURE TO PATHOGEN Adaptive immunity also has a nega-
tive feedback effect on innate im-
munity, preventing damage. B: When
an organism is exposed to the same
pathogen again, the memory cells
can handle the infection, limiting in-
nate immunity such that there are
only minor symptoms. The evolu-
tion of adaptive immunity therefore
provides an advantage to the host:
adaptive immunity evolved to clear
1. Memory cells limit infection 2. Pathogen cleared: limited symptoms. pathogens and limit innate immunity.
THE FRONTLINE OF INNATE IMMUNITY seen in cystic fibrosis, where unusually thick mucus is made
The Skin in lungs that limits expulsion and gives rise to growth of
bacteria in the airways.8 This chronic infection in turn leads
Innate immunity at its simplest involves barriers that prevent to the lung pathology in this disease. In the stomach, acid
entry of microorganisms in the fi rst place. In mammals, the production by the epithelium leads to low pH that limits
skin is an obvious barrier that when broken in injury allows growth of microbes. Enzymes that digest microbial proteins
pathogens to enter. Many pathogens use insects that can de- are also produced, including pepsin in the gut. A phospho-
liver them across the skin into the bloodstream. A good ex- lipid-based substance termed pulmonary surfactant is made
ample here is Plasmodium falciparum, which causes malaria. in the lungs that acts to trap microbes and can also directly
It is transmitted by mosquitoes feeding on blood. If the skin lyse bacteria. Similarly, in skin the production of fatty acids
is broken, pathogens can enter. The importance of the integ- in sebum limits bacterial growth. Lysozyme is a prominent
rity of the skin barrier can be seen in atopic dermatitis, an enzyme in tears and saliva that breaks down peptidoglycan
inflammatory skin disease. Mutations in the gene encoding in the cell walls of gram-positive bacteria. Lysozyme is in
fi llaggrin, which is important for skin integrity, associate fact a prototype of an innate immune protein because it
with atopic dermatitis.5 The altered fi llaggrin is less able to does not affect the host cells, as they lack peptidoglycan. It
maintain the skin barrier, and the dermatitis is likely to be is made by epithelial cells, but also phagocytes and Paneth
caused by microorganisms penetrating the skin or environ- cells in the crypts in the small intestine.
mental irritants that provoke allergy.
The Epithelial Barrier Antimicrobial Peptides

Equally important is the epithelial barrier that lines surfaces Another important example of innate immune proteins pro-
inside the body in the respiratory, gastrointestinal, and uro- duced at barriers is antimicrobial peptides.9 Epithelial cells
genital tracts. Microbes gain access to these surfaces via pro- are a major source of these peptides, but phagocytes can also
cesses such as inhalation and digestion. The epithelial cells make them. They are an ancient form of defense because
that make up the epithelium form tight junctions that keep they are found in most species, including insects and plants.
the microorganisms out.6 Specialized structures called cilia There are three classes: defensins, catelicidins, and histatins.
also expel microbes, and damage to cilia, for example caused Defensins are 30 to 40 amino acid peptides that usually have
by cigarette smoke, promotes respiratory tract infections.7 three disulphide bonds. Their key property is amphipathicity
(they contain both polar and nonpolar properties), and it is
this property that is thought to underly their specificity for
DIRECT-ACTING ANTIMICROBIAL FACTORS the membranes of bacteria, fungi, and certain viruses. Their
The epithelia also produce a range of chemicals that limit hydrophobic part inserts into these membranes, which are
microbial growth. Mucus is a key substance that traps mi- more positively charged than host cell membranes. A pore is
crobes and is expelled. The importance of mucus can be formed, rendering the membrane leaky and leading to lysis.

CHAPTER 15 THE INNATE IMMUNE SYSTEM | 369
There are three subfamilies of defensins-: α, β, and θ, which bacteria live in our intestines and serve several functions, in-
have different specificities in terms of target microorganisms.10 cluding providing vitamin B12 and other nutrients. However,
They are produced as propeptides that require processing by they also maintain epithelial barrier integrity, most likely
proteases. Neutrophils produce α-defensins and store them because they are sensed by innate immune receptors that
in so-called primary granules from which they are secreted will be described in subsequent sections, and these receptors
upon neutrophil activation. Paneth cells in the gut produce a seal the epithelial barrier against infection.14 However, im-
class of α-defensins called cryptidins, which are processed by portantly they also compete with any pathogenic bacteria in
trypsin prior to secretion into the gut., β-defensins are made the intestines limiting their growth. Similarly, on skin and
in the urogenital and respiratory tracts, skin, and tongue. in the urogenital tract, there are competing microbes.
Keratinocytes in skin are also a key producer of β-defensins.
Catelicidins lack the disulphide bonds in the α-defensins
and are produced by epithelial cells, phagocytes, and ke- THE INITIAL RESPONSE ONCE BARRIERS ARE
ratinocytes. They are also produced as propeptides and in BREACHED
neutrophils are processed by elastases in secondary gran- What happens when skin or epithelial barriers are breached?
ules and then secreted in response to microbial stimulation. Pathogens have mechanisms to get across the barriers, or if
Histatins are produced in the parotid, sublingual, and sub- the barrier is damaged by trauma, pathogens will traverse
mandibular glands in the oral cavity. They are histidine-rich the broken barrier. Two important cascades are triggered
and target fungal membranes. in blood that will generate inflammatory factors or act to
wall off pathogens and prevent them from spreading. The
first cascade to be characterized in detail is the comple-
Phospholipase A2 ment cascade, described in detail elsewhere in this volume.
Phospholipase A2 (PLA2) is another antimicrobial protein Complement generates the membrane attack complex that
found in primary granules of neutrophils and similar to will lyse bacteria. Complement factors such as C1q also bind
defensins, is secreted during degranulation in response to to pathogen surfaces and act as opsonins, promoting uptake
bacterial activation. It is also secreted by epithelial cells and by neutrophils and phagocytes. The complement cascade
Paneth cells. PLA2 can directly kill gram-positive bacteria also generates chemotactic factors such as C5a, which at-
by hydrolysing phospholipids in the microbial membrane.11 tracts neutrophils to the site of infection. Coagulation is the
The means by which it distinguishes host membranes from other cascade that is triggered, largely by bacteria. This will
bacterial membranes is because its effect on phospholipids is form clots that prevent bacteria from spreading.
strongly potentiated by another protein termed bactericidal
permeability increasing (BPI), which binds to lipopolysac-
charide (LPS) in gram-negative bacterial membranes.12 The Neutrophil
As will be discussed subsequently, LPS is one of most potent A key cell type, however, that is engaged early in infec-
innate immune activators known; it is sensed by a protein tion is the neutrophil, again described in detail elsewhere.
termed toll-like receptor-4 (TLR4). LPS does not occur in Chemotactic peptides present in bacteria, termed fMLP,
host membranes, and so BPI only facilitates the effect of act as chemoattractants for neutrophils. Both C5a and
PLA2 on bacterial membranes. BPI disrupts the integrity fMLP also trigger an important process in the neutrophil
of the bacterial membrane and presumably allows PLA2 termed the respiratory burst.15 This involves an enzyme
to gain access to phospholipids, where it cleaves fatty acids complex termed the nicotinamide adenine dinucleotide
from the sn-2 position, leading to lysis. Similar to PLA2, BPI phosphate-oxidase. This is also present in macrophages and
is found in primary granules in neutrophils. is activated either by the G-protein coupled receptors that
bind fMLP or C5a, or by the process of phagocytosis. It is
a multicomponent enzyme complex that contains three cy-
Lactoferrin tosolic subunits: p40phox, p47phox, and p67phox. It also
Lactoferrin is an 80 kDa protein that belongs to the transfer- contains a membrane-associated flavocytochrome complex
rin family and therefore acts to sequester iron. This effect is comprising p22phox and p91phox. The subunits form the
thought to limit the availability of iron to bacteria, altering enzyme complex following activation of the low molecular
their metabolism and thereby compromising their ability to weight G protein Rac. The nicotinamide adenine dinucleo-
replicate.13 Lactoferrin is found in the secondary granules of tide phosphate-oxidase generates reactive oxygen species by
neutrophils and occurs at a high level in breast milk, its cel- transferring an electron from its flavin adenine dinucleotide
lular origin there being epithelial cells. Lactoferrin can also cofactor to molecular oxygen forming the superoxide anion
be processed to generate an antimicrobial peptide called lac- and other oxygen radicals. This all occurs in the so-called
toferricin that, similar to other such peptides, can lyse bac- phagolysosome, which comprises a fusion between the
terial membranes. phagosome that contains the ingested bacteria with the lyso-
some. The oxidizing environment will be antibacterial, but
in addition, will lead to acidification of the phagolysosome
The Microbiota due to potassium and hydrogen ions being drawn in to neu-
A final front-line mechanism to keep microbes at bay is tralize the charged superoxide ion. The acidification, how-
commensal bacteria in the healthy intestine. Commensal ever, has been shown to lead to dissociation of proteolytic

enzymes such as elastase from the proteoglycan matrix in domain similar to a Drosphila melanogaster protein
the phagolysosome.16 These proteases will also have a direct Toll.19 The Toll protein had been discovered as being
effect on bacteria. important in the generation of dorsal-ventral polarity
In addition, superoxide is a substrate for myeloperoxi- in the developing Drosophila embryo. At fi rst glance,
dase, which is stored in primary granules and generates hy- this seemed odd. What was a receptor important for
pochlorous acid and chloramines, as well as other reactive inflammation doing with a signaling domain highly
oxygen species that are bactericidal and fungicidal. A final similar to a protein in fruit fly development? Toll
oxygen radical generated in neutrophils is nitric oxide. It is drives a transcription factor termed dorsal, however,
produced by the enzyme nitric oxide synthase, which is in- which is highly similar to NFκB so it seemed as if a
duced by cytokines such as interferon-γ. Nitric oxide is also similar “machine” was being used in two different
microbicidal. contexts: one for inflammation in the case of IL-1
and the other for development in the fruit fly.
2) In 1994, a protein termed N protein was described in
THE CONCEPT OF PATTERN RECOGNITION: tobacco by Whitham and colleagues, which provided
HISTORICAL PERSPECTIVE resistance to tobacco mosaic virus.20 This protein
Much of the forgoing components of innate immunity had also had the domain found in the type I IL-1 receptor
been worked out at least in broad terms by the mid-1990s. and Toll, which was then named the Toll-IL-1 recep-
However, the issue of the exact recognition mechanism dur- tor–resistance (TIR) domain. Clearly, the TIR do-
ing innate immunity by the host was poorly understood. In main was therefore important for innate immunity
1989, Janeway made the point that the continued focus on in very diverse species.
adaptive immunity, which was evident at the time, would 3) This was confi rmed when in 1996, Lemaitre and col-
lead to an asymptote being approached in our understand- leagues reported that Toll also has a role in innate
ing of immunity.17 He speculated that receptors must exist immunity in Drosophila , being activated in re-
for microbial structures that were known to be proinflam- sponse to fungi and driving the antifungal peptide
matory, such as LPS from gram-negative bacteria. Receptors drosomycin.21
such as mannan-binding lectin (MBL), which binds man- 4) Mammalian proteins even more similar to Toll than
nose in microbial membranes (and which does not exist in the IL-1 receptor, and termed TLRs, were then re-
host membranes) and CD14 (which binds LPS), had already ported 22 and one, termed Toll (subsequently renamed
been described at that stage, but importantly these proteins TLR4) when overexpressed in dendritic cells, was
did not signal because they possessed no intracellular do- shown to drive costimulation of T cells, as reported
mains. MBL was shown to be an opsonin, and CD14 had by Medzhitov and colleagues.23
no clear function. Janeway therefore proposed that pattern 5) In 1999, Poltorak and colleagues reported that the
recognition receptors (PRRs) must exist that recognize mi- lack of responsiveness of a strain of mouse termed
crobial “patterns” such as LPS, termed pathogen-associated C3H/HeJ to LPS was due to a mutation in the gene
molecular patterns (PAMPs), and trigger activation of cells encoding TLR4, such that a proline in the signaling
such as dendritic cells, which are prodigious producers of domain was converted to a histidine.24 This rendered
cytokines and also present antigen to T cells. At around the TLR4 unable to signal, identifying TLR4 as the long-
same time, Matzinger was proposing that the job of the im- sought signaling receptor for LPS. It was therefore
mune system is to sense “danger.”18 The danger can be in the clear that TLRs were ideal examples of the PRRs
form of a microbe or in the form of tissue injury. Again, the proposed by Janeway. Subsequently, 10 TLRs were
idea that specific receptors would exist to recognize products described in human, although other species have
of damaged tissue was proposed. In essence, what was being very many more, notably the sea urchin, which has
hypothesized was the existence of receptors that would drive over 150.25 TLRs were shown to recognize a diverse
the production of cytokines, as several cytokines had been range of PAMPs and also in some cases products of
found that could provoke inflammation and also adaptive damaged tissue, termed danger-associated molecu-
immunity. LPS was a powerful inducer of such cytokines lar patterns (DAMPs). Complex signaling pathways
and yet no receptor had been convincingly described for were elucidated and links to disease established. The
LPS that would activate transcription factors such as NF-k B, discovery of TLRs also galvanized researchers to seek
which was known to be required for increased transcription other PRR types, and there are now several families,
of genes such as that encoding the potent proinflammatory notably the retinoic acid-inducible gene I (RIG-I)-
cytokine tumor necrosis factor (TNF). like receptors (RLRs), which sense viral ribonucleic
How did Janeway’s insight that PRRs and PAMPs would acid (RNA), the nucleotide oligomerization domain
form the core of innate immunity lead to the discovery of (NOD)-like receptors (NLRs), which sense bacteria
the TLRs, which became the prototypical PRRs? Five im- and DAMPs, and the C-type lectin receptors, which
portant fi ndings were made. sense fungi.26 Figure 15.2 illustrates the main families
1) The receptor for the proinflammatory cytokine in- of PRRs. A level of complexity in innate immunity
terleukin (IL)-1, which is a key mediator of many was therefore revealed that has provided great insight
aspects of inflammation and which, similar to LPS, into the overall workings of the immune systems of
could activate NFκB, was shown to have a signalling many organisms.

phosphocholines
MBL
Bacterial surface
Fungal surface (mannose)
Pentraxins Bacterial cell walls (eg LPS)
Flagellin
Fungal cell wall (eg Zymosan)
Fungal cell wall (beta-glucan)
LRR Toll-like receptors CRD
TIR ITAM
Dectins
Microbial genomes Bacterial cell wall fragments

(dsRNA, ssRNA, dsDNA) Bacterial toxins NOD-like receptors
Uric acid
Flagellin LRR
Viruses
Potassium efflux NACHT
Toll-like receptors
ROS
Lysosome disruption CARD - PYRIN
endosome
Viral RNA DNA

RIG-I-like receptors AIM2 like receptors HIN200 Pyrin
Helicase CARD
FIG. 15.2. Innate Immune Sensors. Several classes of innate immune sensors have been found. Soluble factors
such as pentraxins and mannoan-binding lectins bind pathogen structures and promote phagocytosis. Toll-like
receptors (TLRs) sense bacteria, fungi, and protozoans from the plasma membrane, whereas endosomal TLRs
sense microbially derived nucleic acids. C-type lectin receptors sense fungal cell wall components and myco-
bacteria. In the cytosol nucleotide oligomerization domain–like receptors sense microbial products and also,
in the case, of Nlrp3 particles such as uric acid and IAPP. Retinoic acid-inducible gene I–like receptor sense
viral ribonucleic acid, whereas AIM-2-like receptors sense microbial deoxyribonucleic acid. All lead to signaling
pathways culminating in the induction of immune and inflammatory genes.
Toll-Like Receptors with all of the TLRs with the exception of TLR3 signalling
TLRs are type I transmembrane receptors that have leucine- via MyD88, to NF-κ B and mitogen-activated protein (MAP)
rich repeats (LRRs) and a TIR domain. LRRs are found in kinases such as p38. TLR2 and especially TLR4 require a
many proteins, and that they occur in TLRs is of interest bridging adapter termed Mal to recruit MyD88. TLR3 sig-
given their capacity to interact with different kinds of li- nals via Trif to NF-κ B and another transcription factor
gands. Also of note is that in agnathans, the LRR is used to IRF3. TLR4 can also signal via Trif and does so via the sec-
build a repertoire of variable lymphocyte receptors, which ond bridging adapter Tram.
are analogous to the antibody repertoire in human.27 The
LRR domain is therefore of evolutionary interest in that it Toll-Like Receptors 2, 1, and 6
is used for TLRs but also generation of diversity in the lam- TLR2 recognizes multiple ligands. From bacteria, these in-
prey and the hagfish. The immunoglobulin domain serves clude lipopeptides, lipoproteins, lipoteichoic acid, and myco-
this purpose in vertebrates and in fact is fused to the TIR bacterial lipoarabinomannan. From yeast, TLR2 recognizes
domain in the type I IL-1 receptor. Figure 15.3 illustrates the the cell wall structure zymosan. TLR2 also recognizes gly-
relationship between these domains. cosylphosphoinositol from the parasite Tryanosoma cruzi.
The 10 TLRs in human have different ligand specifici- Alone amongst the TLRs, TLR2 can heterodimerize with
ties, and several ligand/TLR structures have been solved.28 other TLRs, and this provides a degree of specificity in li-
Broadly speaking, TLRs can be split into two families. TLRs gand recognition.30 The TLR1/TLR2 dimer binds triacylated
1, 2, 4, 5, and 6 are expressed in the plasma membrane, and lipopeptides whereas TLR2/TLR6 recognizes diacylated lipo-
their job is to sense bacterial, fungal, and protozoal prod- pepides. In the case of TLR2/6, one acyl chain fits into a hy-
ucts. TLRs 3, 7, and 9 are all expressed in endosomal mem- drophobic pocket in one TLR, whereas the second acyl chain
branes with their LRRs in the lumen and their TIR domains fits into the other TLR.31 This allows the TLRs to dimerize,
pointing into the cytosol. Their role is to sense viral nucleic an important event for signaling. There is also a coreceptor
acids. There are also differences in relation to signaling,29 for TLR2/6 in the form of CD36, which appears to deliver the

VERTEBRATES LAMPREY
Toll-like receptors Leucine-Rich Repeat domain
Variable lymphocyte receptors
TIR domain
IL-1 receptor family
MULTICELLULAR
ORGANISMS FIG. 15.3. Protein Domains in Innate and Adaptive
Immunity. Three protein domains occur in several
Toll-like receptors immune system proteins. The leucine-rich repeat
Immunoglobulin domain occurs in toll-like receptors in vertebrates and all
multicellular organisms combing with the toll-IL-1
receptor–resistance (TIR) domain. The leucine-rich
repeat also forms the basis of the repertoire of vari-
Antibodies/T cell receptors
able lymphocyte receptors in the hagfish. The immu-
noglobulin domain only occurs in vertebrates and is
found in antibodies and T-cell receptors. However, it
also combines with the TIR domain to form interleu-
kin-1 receptor family members.
lipopeptide to the TLR dimer.32 TLR2 is expressed on macro- analogue polyIC. PolyIC had been known for some time to
phage, dendritic cells, B cells, and also on regulatory T cells. be a potent inducer of the antiviral type I interferons. Similar
to LPS, therefore, the discovery of TLR3 as the receptor for
Toll-Like Receptor 4 polyIC provided a molecular explanation for how polyIC
TLR4 is the receptor for LPS and is the most studied TLR. induces this response. TLR3 is expressed on macrophages
It acts as a homodimer but on its own is unable to recog- and dendritic cells and importantly on CD8 + dendritic cells,
nize LPS. Similar to TLR2 and CD36, a coreceptor is needed, which are potent type I interferon producers.
termed MD2.33 LPS is a hexacylated lipid, and five of the acyl
chains are buried in MD2, whereas the sixth is on the sur- Toll-Like Receptor 5
face and interacts with TLR4.34 However, two other proteins TLR5 is the receptor for bacterial flagella.40 Flagellin is a key
are needed to deliver LPS to MD2. LBP occurs in serum and protein in flagella, which occur in multiple bacteria species.
binds LPS monomers, transferring them to CD14.35 CD14 can Flagellin is in many ways a prototypical PAMP because it
occur as a soluble protein or as a glycerophosphatidylinositol does not occur in humans and can bear few mutations in
anchored protein.36 The job of CD14 appears to be to concen- bacteria. TLR5 is therefore required for host defence against
trate LPS and deliver it to MD2. In B cells, there are two other flagellated bacteria such as Legionella and Salmonella.
accessory molecules: RP105 and MD1.37 These appear to serve
an analogous function to CD14 and MD2. Recognition of Toll-Like Receptor 7
LPS is therefore highly complex. This is presumably because TLR7 recognizes single-stranded RNA from viruses. It is
once TLR4 is activated, there can be lethal consequences highly expressed in plasmacytoid dendritic cells, where
in the form of septic shock; therefore, several proteins are when activated it induces type I interferons. TLR7 was
needed for LPS recognition. Similar to TLR2, a dimerization first described as the receptor for a small molecule termed
event occurs upon ligand binding that allows the intracellu- imiquimod, which is used to treat genital warts. This effect
lar TIR domains to come into juxtaposition and trigger sig- is mediated by type I interferons. TLR7 ′s expression on
naling pathways. TLR4 has been shown to recognize other plasmocytoid dendritic cells reveals a key function in anti-
ligands, notably F protein from respiratory syncytial virus,38 viral immunity.
although the basis for this recognition is not known. Another
example is the metal nickel. In contact dermatitis, there is an Toll-Like Receptor 9
inflammatory reaction to nickel, and it has been shown that TLR9 is the deoxyribonucleic acid (DNA) sensing TLR that,
nickel coordinates two key histidines on different TLR4 pro- similar to TLR3 and TLR7, is found in endosomes. The type
teins, causing a cross-linking event that activates TLR4 and of DNA recognized is of interest as it binds to hypomethylated
drives the inflammatory process in skin.39 TLR4 is expressed CpG-rich DNA, commonly found in bacteria and viruses.
on multiple cell types, notably on macrophages and dendritic TLR9 is also expressed in plasmacytoid dendritic cells as well
cells but also on neutrophils and in the endothelium. as in macrophages and in the epithelium. Once activated,
again by dimerization, TLR9 induces a strong proinflamma-
Toll-Like Receptor 3 tory signal, increasing TNF production as well as type I inter-
TLR3 is the receptor for double-stranded RNA, found in ferons. A protein termed Unc93b is required to traffic TLR9
viruses or more commonly in the form of the viral RNA from the Golgi to intracellular vesicles. The CpG-rich DNA

from pathogens is taken up by cells in an endosome, which involves the IL-1 receptor–associated kinases (IRAKs).
fuses with the TLR9-expressing endosome. In this way, the There are four of these: IRAK-1, IRAK-2, IRAK-4, and
receptor can interact with its ligand. TLR9 has no known co- IRAK-M. They interact with MyD88 via their death do-
receptors, but it has been shown that it requires cleavage by mains, because along with the TIR domain, MyD88 has a
cathepsins for full activation.41 death domain. The structure of this multicomponent signal-
ing machine, termed the MyDDosome, has been solved.45,46
The Recognition of Danger-Associated Molecular Strikingly, it has a stoichiometry of 14. SIx MyD88 subunits
Patterns by Toll-Like Receptors assemble and bring four IRAK-4s to the complex. These sit
TLRs have also been implicated in the recognition of prod- on top of the assembled MyD88 subunits in a helical struc-
ucts of damaged tissues.42 This is an important area as the ture. Four IRAK-2s or IRAK-4s are then recruited. All of the
inflammatory process can occur under conditions where death domain interfaces in this complex have been solved,
there is apparently no infection. DAMPs include a range of and it is predicted to have a high degree of cooperativity.
factors released by damaged cells and include high-mobility The so-called MyDDosome launches signaling pathways ac-
group protein 1, heat shock proteins such as hsp90, prod- tivated by TLRs.
ucts of damaged connective tissue such as hyaluronic acid Once assembled, IRAK-4 phosphorylates IRAK-1, and
fragments, versican and tenascin-C, the acute-phase reac- this event appears to be key for signaling. The IRAKs then
tant serum amyloid A, pyrrole-based lipids such as CEP, and dissociate from the complex, and IRAK-1 and/or IRAK-2
coagulation products such as fibrinogen. This area has been then interact with the ubiquitin ligase Traf6, a member of
somewhat controversial, as in vitro there is a risk of LPS con- the Traf family. Once activated, Traf-6 signals in associa-
tamination. However, the consensus has emerged that at least tion with two noncanonical E3 ubiquitin ligases, Ubc13 and
some of these are indeed sensed by TLRs although what the Uev1A.47 This results in polyubiquitination of Traf6. This
mechanism might be is not fully understood. Many of these is a type of ubiquitination termed K63-linked ubiquitina-
are sensed by TLR4, notably high-mobility group protein-1, tion. This covalently modifies Traf-6 causing oligomeriza-
hsps, hyaluronan fragments, and tenascin-C. TLR2 has been tion and downstream activation. This activation involves
shown to sense versican and CEP. Importantly, these factors the IkappaB kinase (IKK) complex for the NF-κ B pathway.
are only produced by damaged cells, as would occur during The main form of NF-κ B in cells is a p50/p65 dimer. It is,
tissue injury. There may be coreceptors involved here, which however, kept in check by an inhibitory protein I-κ B. Traf6
explains the recognition process, because it is difficult to activates the kinase TAK-1, which in turn activates the IKK
imagine how a single TLR can recognize multiple ligands. complex leading to I-κ B degradation (via K48-linked polyu-
A clearer example is the sensing of host nucleic acids in the biquitination). This releases the p50/p65 dimer, exposing a
case of cell damage. Necrotic cells have been shown to re- nuclear localization sequence and allowing NF-κ B to bind
lease double-stranded RNA, which is sensed by TLR3. TLR7 its consensus site in target genes. In addition, TAK-1 can
can sense host RNA, and TLR9 can sense host DNA in the activate MAP kinase cascades, notably via MKK3/6, which
context of immune complexes. This has particular relevance leads to p38 MAP kinase activation, and MKK7, which leads
to autoimmune diseases such as systemic lupus erythema- to activation of Jun N-terminal kinase . Both of these MAP
tosus (SLE).43 kinases can activate transcription factors, such as ATF3 in
the case of p38 and Jun in the case of Jun N-terminal kinase.
Toll-Like Receptor Signaling Pathways This in turn will modulate inflammatory gene expression.48
The key output from TLR signaling is cytokines, and TLRs MyD88 is also involved in activation of IRF7 by TLR7
fi lled the missing gap in our understanding of what actually and TLR9.49 This is an important transcription factor for the
induces cytokines. Because they bind directly to microbial induction of interferon–alpha, a key antiviral cytokine but
ligands, TLRs bring us close to the initiation of host defense. also a cytokine implicated in the pathogenesis of SLE.
The type of cytokine being induced is dependent on the The importance of the MyD88 pathway can be seen in
TLR. TLR4 is the most prodigious activator of macrophages MyD88-deficient mice, which are unable to mount an effec-
known, with many genes induced by TLR4. Other TLRs are tive host defense response to multiple pathogens, including
less prolific but all share the feature of driving inflamma- a wide range of bacteria, as well as fungi, protozoans, and
tion. Cytokines such as IL-6, TNF, and the chemokine IL-8 certain viruses.50 In addition, MyD88-deficient mice are re-
are typical responses for plasma membrane receptors. For sistant to inflammation in a number of disease models in-
the endosomal TLRs, type I interferons are the typical read- cluding models of SLE and arthritis. However, in the case
out. What gives rise to signaling specificity? As stated previ- of humans, the situation is not as clear-cut. This is because
ously, TLRs use different adapter proteins. All have a TIR of reports of MyD88 deficiency in human.51 People with
domain, and upon TLR dimerization, adapter TIR domains this deficiency show higher mortality in childhood but sus-
are likely to be recruited to the interface formed between ceptibility to a restricted set of pathogens, the main defect
two TLRs. This then launches the signaling pathways. in host defense being against pyogenic infections such as
Broadly speaking, TLR signaling pathways can be divided Streptococci. This is also the case in subjects with IRAK-4
into two: MyD88-dependent and MyD88-independent. All deficiencies.52 Once people with these deficiencies reach
TLRs, with the exception of TLR3, signal via MyD88. IL-1 adulthood, however, no immunodeficiencies are evident. It
receptor family members, including the receptor for IL-18, is likely, therefore, that the MyD88 pathway is somewhat re-
also signal via MyD88.44 The MyD88 dependent pathway dundant in children and is not required in adults probably

because of the emergence of adaptive immunity. This sup- LPS

Plasma membrane
ports the concept raised at the start of this chapter, that once
memory is established, innate immunity is not as important. Recycling if
PIP2 Myr
The second TIR domain adapter to be found was Mal empty of ligand Mal Tram
(also termed TIRAP).53,54 This protein is required for MyD88
Early Late
MyD88-dependent signaling specifically by TLR2 and es- Golgi endosome
endosome
pecially TLR4. It appears to act as a bridging adapter for NF-κB
MyD88 to be recruited to the receptor TIR domains and No PIP2 EEA1 Rab7
has a PIP2 binding domain which localizes it to the plasma
membrane.55 It is also a substrate for IRAK-1 and IRAK-4, Myr Myr Myr
E.R.
and upon phosphorylation undergoes ubiquitination and Tram Tram Tram
degradation,56 providing a negative feedback loop in TLR Trif TAG TAG
signaling.
The MyD88-independent signalling pathway involves IRF3 ‘OFF’ degraded
the third adapter, Trif.57 This adapter is used by TLR3 to
signal and activates NF-κ B as well as another transcrip- FIG. 15.4. Subcellular Toll-Like Receptor 4 Signaling. The signal-
tion factor IRF3 via activation of the kinase TBK-1.58 TLR4 ing pathway, activated by lipopolysaccharide (LPS) via toll-like re-
can also activate Trif, but this happens in endosomes. The ceptor-4 (TLR4), is highly complex. In resting cells, TLR4 recycles
from the endoplasmic reticulum to the plasma membrane. An LPS/
fourth adapter, Tram, mediates this effect, acting as a bridge MD2 complex binds TLR4, stabilizing it transiently at the plasma
for Trif in a manner similar to Mal acting as a bridge for membrane. The activated TLR4 dimer recruits the adapters Mal
MyD88.59 Tram has a myristoylation motif that localizes and MyD88, which then lead to activation of NF-κB, which induces
it to the plasma membrane.60 However, upon activation of multiple immune and inflammatory genes, such as tumor necrosis
TLR4, Tram undergoes phosphorylation by protein kinase factor, interleukin-1beta, and interleukin-6. Mal is retained in the
C-epsilon.61 This leads to dissociation of Tram from the membrane by binding the phospholipid PIP2. Th adapter Tram is
also membrane-localized via a myristate in its N-terminus. Tram
plasma membrane, and it is likely that a TLR/Tram com-
then undergoes phosphorylation by protein kinase C-epsilon. TLR4
plex localizes to the endosome, where Trif is then recruited, then internalizes in combination with Tram and localizes to an early
again leading to TBK-1 activation and induction of type I in- endosome where it encounters Trif. This adapter activates TBK-1
terferons. The TLR4 signal is terminated by a splice variant and promotes IRF3 activation, which regulates the expression of
of Tram termed TAG, which displaces the Trif/Tram com- type I interferons and other gene products. The endsome containing
plex.62 It is this dual pathway, Mal/MyD88 and Tram/Trif, TLR4 then matures to a late endosome where TAG is engaged. This
occurring at different subcellular localizations, that prob- protein displaces Trif from Tram, and TLR4 is ultimately degraded.
ably make LPS such a powerful innate immune activator, as
a wide range of genes can be induced. Figure 15.4 illustrates
the TLR4 signaling pathway. flax has a TIR domain and is required for resistance to flax
The final TIR domain adaptor to be found is SARM. This rust. Plants also have IRAK-like kinases activated by these
is a large protein with a TIR domain and SAM and ARM receptors, notably BAK1 and BIK1 in Arabidopsis. The re-
domain. SARM is a negative regulator of Trif-dependent sig- ceptor for flagellin in Arabidopsis is sensed by a leucine rich
naling, acting as another modulator of TLR4 action.63 It is repeat protein termed FLS-2. This receptor has an IRAK-like
of interest that there are several mechanisms to inhibit TLR kinase domain actually attached and also activates BAK-1
signaling, including microRNAs, decoy receptors, ubiqui- and BIK-1. FLS-2 activates MAP kinases such as MEKK1,
tin regulators (eg, A20), and splice variants of MyD88 and which probably lead to activation of the AtWRKY family of
IRAK2 that are inhibitory.64 This tells us that TLRs, and in transcription factors. In addition, ubiquitin ligases termed
particular TLR4, must be kept under control otherwise the PUBs ubiquitinate FLS-2, downregulating it.70 The paral-
host can be damaged or even succumb. lel in human is Triad3a, which ubiquinates TLR4 leading
Viruses have also been shown to target TLRs. The best to its degradation.71 These conserved elements confirm the
examples are two vaccinia virus proteins A46R and A52R.65 importance of the TLR pathways across evolutionary time.
A46R has been shown to block recruitment of TIR domain– Table 15.1 illustrates the conservation between mammals,
containing adapters, particularly Mal and Tram.66 A52R insects, and plants.
inhibits activation of the IRAKs.67 Deletion of A52R leads
to attenuation of the virus, confirming its importance for Therapeutic Targeting of Toll-Like Receptors
viral evasion.68 TLRs are being targeted clinically, either in the form of new
vaccine adjuvants or with inhibitors that would be anti-in-
Conservation of the Toll-Like Receptor System flammatory.72 The best example of an adjuvant approach is
Across Nature imiquimod, a compound found to be antiviral by activat-
Every multicellular organism has proteins with TIR domains, ing TLR7. This induces type I interferons, and imiquimod is
attesting to their evolutionary importance. A particular case used to treat genital warts. Another example of an adjuvant
is in plants, which have a number of TLRs.69 As mentioned approach is the use of monophosphoryl lipid A in vaccines
previously, N protein from tobacco has a TIR domain and is for such pathogens as human papilloma virus. This acts via
required for antiviral immunity. Similarly, L protein from TLR4 but is not as toxic as LPS. In terms of anti-inflammatory

TABLE 15.1 Conservation of Innate Immunity across Species
Human Fruit Fly Plant

Receptor (TIR domain) Toll-like receptors Toll Tobacco: N protein Flax: L6
Adapter protein (TIR domain) MyD88, Mal, Trif, Tram dMyD88
Non-RD kinase IRAK-1 Pelle Rice: XA21
RD kinase IRAK-4 Arabidopsis: BIK1
Ankyrin repeat IkappaB NF-κB Cactus Relish Rice: NH
RD, Arginine-Aspartic acid; TIR, toll-IL-1 receptor–resistance.
Notes: Each receptor senses microbial products and activates the adapters and kinases indicated. Human and insects have the NF-κB family transcription factors on these
pathways. Plants lack NF-κB but have other transcription factors such as the WRKY family.
approaches, eritoran, an antagonist toward TLR4, has been latter component is more common in gram-negative bacte-
tested in septic shock, but with only limited success. As re- ria. Peptidoglycan is constantly remodeled during bacterial
ferred to previously, TLR4 has also been implicated in the growth, and the component parts are produced constantly,
pathogenesis of contact dermatitis. It is also implicated in making them ideal PAMPs. How the ligands get into the cy-
rheumatoid arthritis and inflammatory bowel disease. TLR2 tosol is not wholly clear, however. Transport systems such as
has been shown to be important for ischemia/reperfusion in- pannexin, PepT1, and PepT2 have been shown to carry mur-
jury and may prove interesting as a target to limit this process amyl peptides across the plasma membrane, and there is also
in transplantation.73 There is promise that targeting TLR7 evidence of endocytosis.76 NOD1 and NOD2 signaling leads
and/or TLR9 might be effective in SLE, as in this disease to NF-κB and MAP kinase activation in a manner similar to
TLR7 senses host ribonucleoprotein complexes (known to be TLRs. There are no adapters, however, and they recruit RIP2
implicated in SLE) while TLR9 senses DNA/chromatin com- kinase to activate these downstream pathways. NOD1 and
plexes (again implicated in SLE). Whether inhibiting TLRs NOD2 agonists can synergise with TLRs, which is likely to be
will be prove to be to immunosuppressant remains to be seen, important in vivo.77
although given the redundancy in the TLR system in relation
of host defense, this may not pose a problem, particularly if a Role of Nucleotide Oligomerization Domain 1 in
given TLR has a nonredundant role in inflammatory disease. Antimicrobial Responses
The first pathogen shown to be sensed by NOD1 in vivo
was Helicobacter pylori ; NOD1-deficient mice were demon-
NOD-Like Receptors strated to be more susceptible.78 Key outputs from NOD1
The second major family of PRRs are the NLRs.74 Nod-like in this model are chemokines, antimicrobial peptides, and
receptors undergo oligomerization in response to activa- type I interferons. NOD1 has also been shown to be im-
tion. They also have LRRs, similar to TLRs. There are sev- portant in the host defense response to other pathogens,
eral subfamilies that can be defi ned by additional domains.
The main members are summarized in Table 15.2. Of par-
ticular interest are the NLRC family, which have antibacte-
rial NOD1 and NOD2 as members and are defined by the Notable Nucleotide Oligomerization
caspase recruitment domain (CARD). The Nlrp subfam- TABLE 15.2 Domain–Like Receptor Family
ily contains a pyrin domain, and the prototypical member Members
is Nlrp3, which is a key component of the inflammasome
multiprotein complex that regulates caspase-1 and drives Subfamily N-terminus Function
production of the proinflammatory cytokine IL-1beta. The NLRB (NAIP) BIR Binds flagellin
NAIP subfamily contains a Bir domin. All NLRs are intra-
cellular proteins, implying that ligands have to be endocy- NLRC
tosed in order to activate them. NOD1 CARD Binds ieDAP
NOD2 CARD Binds MDP
Nucleotide Oligomerization Domain 1 and NLRC4 CARD Binds flagellin/NAIP
Nucleotide Oligomerization Domain 2 NLRP
NOD1 and NOD2 sense breakdown products of peptidoglyan, NLRP1 Pyrin domain Binds anthrax lethal toxin
which is a major component of the cell wall of gram-positive NLRP3 Pyrin domain Mutiple activators (eg, uric
and gram-negative bacteria. The finding that NOD2 muta- acid, IAPP, cholesterol
tions strongly associate with Crohn disease was an important crystals)
finding. It supported the hypothesis that aberrant sensing of BIR, Baculo Virus Repeat; CARD, caspase recruitment domain; IAPP, islet
intestinal microbes is important for the pathogenesis of in- amyloid polypeptide; ieDAP, Diaminopimelic acid; MDP, Muramyl Dipeptide;
NAIP, Neuronal Apoptosis Inhibitory Protein; NLRB, Nod-like receptor with
flammatory bowel disease.75 NOD2 detects muramyl dipep- BIR domain; NLRP, Nod-like receptor with Pyrin domain; NOD, nucleotide
tide, whereas NOD1 detects meso-diaminopimelic acid. This oligomerization domain.

including Clostridium difficile, Legionella pneumophilia, and There is additional evidence that NOD1 and NOD2 pro-
Listeria monocytogenes.79 In all cases, impaired recruitment mote adaptive immunity. This was first observed when it
of neutrophils was a feature of the NOD1-deficient mice, was realized that muramyl dipeptide was a key component
highlighting again the importance of neutrophils in anti- in Freund’s adjuvant, a standard adjuvant used in experi-
bacterial host defense. NOD1-deficient mice have also been mental immunology to drive antibody production. This is
shown to be impaired in their response to the intracellular probably due to NOD2 inducing important T-cell–acti-
parasite Trypanosome cruzi.80 Although the T. cruzi ligand vating cytokines, but there is also evidence for NOD2 in
is not known, macrophages from the deficient mice have an T cells themselves where it promotes interferon-gamma
impaired ability to kill intracellular parasites. production.
Role of Nucleotide Oligomerization Domain 2 in the

Antimicrobial Response Inflammasomes
Similar to NOD1, NOD2 has been shown to be important There are 14 members in the Nlrp branch of the NLRs,
for host defense against multiple bacteria. One interesting and those with a known function are components in
aspect, however, was that NOD2 was only shown to be re- inflammasomes. The term inflammasome comes from the
quired for host defense against L. monocytogenes when given fact that the regulation of caspase-1 was shown to be due
by the oral route.81 Intravenous and intraperitoneal infec- to assembly of a multiprotein complex involving Nlrps and
tions were handled normally in NOD2-deficient mice. This also the scaffold protein Asc.85 This complex is needed for
suggests that other PRRs can compensate for the lack of the activation of caspase-1, the enzyme that processes the
NOD2 depending on the route of infection. There is also re- pro-forms of IL-1beta and IL-18. Given the importance of
dundancy in relation to NOD1 in the response to Salmonella these cytokines in inflammation and inflammatory diseases,
typhiurium infection. NOD1- or NOD2-deficient mice are much attention has focused on inflammasome regulation.
normal in their response, but the double knockout is im-
paired, with decreased inflammation and enhanced coloni- Nlrp3
zation of the intestine.82 The Nlrp3 inflammasome has been characterized in detail.
Nlrp3 has a carboxy-terminal LRR domain, a central NOD
Autophagy, Adaptive Immunity, and domain, and an amino terminal pyrin domain. The pyrin
Nucleotide Oligomerization Domain 1 and domain mainly interacts with the key scaffold protein Asc.
Nucleotide Oligomerization Domain 2 It is triggered by agents phagocytosed by macrophages, in-
Another interesting aspect of NOD1 and NOD2 biology con- cluding bacteria and viruses, but also particles such as uric
cerns a process termed autophagy. This is a process whereby acid, silica, asbestos and cholesterol crystals, and amyloid-
an organelle called the autophagosome digests material in containing proteins such as beta-amyloid and islet amyloid
the cytosol of cells. Autophagosomes are related to lysosomes polypeptide (IAPP).86 Uric acid, cholesterol crystals, beta-
and were originally shown to be important in development amyloid, and IAPP all constitute DAMPs. Activation mech-
and in starvation where in effect a cell will digest some of its anisms are not fully worked out, although potassium efflux
own contents to provide nutrients. Autophagy has also been is a common event that is required for oligomerization of the
found to be important in antimicrobial immunity. It is an Nlrp3/Asc complex. Roles for reactive oxygen species and
important part of intracellular innate immunity and may in cathepsins have also been suggested.87,88 CARD domains
fact be the means by which cells such as macrophages clear in Asc and caspase interact, whereas the pyrin domain in
bacteria in a process that was originally thought to involve Nlrp3 interacts with the pyrin domain in Asc, allowing for
the lysosome. Autophagosomes can be viewed as a subtype the complex to assemble. This is needed to activate cas-
of lysosome. These insights came from the uncovering of pase-1. Extracellular adenosine triphosphate is also able to
proteins required for the formation of the autophagosome, activate the Nlrp3 inflammasome. Also of note is the fact
notably ATG16, which is recruited to membranes and al- that the system must be primed by other innate receptors,
lows formation of the autophagosomal double membrane notably the TLRs, which induce Nlrp3 and also pro-IL-1beta
structure. NOD1 and NOD2 activation have been shown expression.
to lead to recruitment of ATG16L to the plasma membrane The effect of uric acid on Nlrp3 is of interest because uric
sites where S. typhimurium and Shigella flexneri intrude.83 acid may be a common danger signal from damaged cells,
This leads to degradation of bacteria in autophagosomes and as well as having a clear role in gout.89,90 Similarly, the in-
importantly to the processing of bacterial antigens for pre- flammation associated with silica and asbestos is likely to
sentation on major histocompatibility complex class II mol- be due Nlrp3 activation.91 The effect of cholesterol crystals,
ecules. This is a classic example of the link between innate which form in atherosclerotic plaques, on Nlrp3 implicates
and adaptive immunity. Also of interest is the observation the inflammasome in heart disease.92 Similarly, the effect of
of mutations in ATG16L in Crohn disease and the observa- beta-amyloid suggests a role in Alzheimer disease.93 In the
tion that the mutant NOD2 also in Crohn disease is less able case of IAPP, a role in type 2 diabetes is suggested.94 IL-1
to drive ATG16L recruitment.84 This could mean that im- has been implicated in insulin resistance and beta cell loss
paired antigen presentation to T cells (eg, regulatory T cells in type 2 diabetes, but s there is no infection, what drives
that produce IL-10) might be important in Crohn disease IL-1beta is not clear. Hyperlipidemia is now known to acti-
pathogenesis. vate the Nlrp3 inflammasome in adipose tissue.95 This will

lead to IL-1beta release, which causes insulin resistance. The Bacteria such as Salmonella are sensed by IPAF. The inter-
beta cells in the pancreas respond by secreting insulin, but action between IPAF and caspase-1 again involves CARD
also IAPP, which modulates insulin action. However, IAPP domains. More generally, IPAF is activated mainly by gram-
can form amyloid, and macrophages in the pancreas take negative bacteria that contain bacterial type III or type IV
up IAPP; the Nlrp3 inflammasome is activated leading to secretion systems. These include Salmonella, Legionella,
IL-1beta production. These findings support the clinical tar- Shigella, Pseudomonas, and Yersinia.107 Several of these
geting of IL-1beta in several diseases notably type 2 diabetes strains when deficient in flagellin do not activate caspase-1.
and atherosclerosis.96 Flagellin added extracellularly does not activate the IPAF in-
Another important aspect of the Nlrp3 inflammasome flammasome and is likely to enter the cytosol via the type III
concerns genetic mutations. Activating mutations in Nlrp3 or type IV secretion systems. IPAF has a coreceptor that is
lead to autoinflammatory diseases such as Muckle Wells dis- another NLR family member NAIP5. This protein interacts
ease and familial Mediteranean fever.97 Autoinflammatory directly with flagellin and is the actual sensor, the signaling
diseases such as these are prototypical genetic diseases of in- occurring via IPAF.108 Similar to NODs and TLRs, there is
nate immunity because there are only inflammatory symp- also evidence for an interaction between IPAF and TLR5.
toms with no roles for T and B cells and autoimmunity. In a model of immunization with ovalbumin and flagellin,
The diseases are systemic, involving skin inflammation, ar- it was shown that IPAF- or TLR5-deficient mice are normal
thropathy, and amyloidosis. They are treatable with anti-IL- in the humoral response, but a double knockout mouse was
1beta blocking agents, attesting to the potency of this proin- deficient in its response.109 This implies that both IPAF and
flammatory cytokine. TLR5 are acting in a redundant manner in this system. More
generally, it is no surprise that innate systems would be re-
Nlrp3 in Host Defense dundant in terms of host defense, as a lack of redundancy
Several bacteria have been shown to activate the Nlrp3 would leave the host open to manipulation by pathogens.
inflammasome. A common feature is pore-forming tox-
ins that cause potassium efflux. Listeria produces a toxin Nlrp1
called listeriolysin that is required for Nlrp3 activation.98 The Nlrp1 inflammasome was the first to be characterized
Secretion of pneumolysin by S. pneumoniae or streptolysin biochemically, but its activators are not fully understood.
O from S. pyogenes both result in activation of the Nlrp3 It is expressed in many cell types, including granulocytes,
inflammasome.99 monocytes, dendritic cells, B cells, T cells, and neurons. It
Nlrp3 is also activated by viruses. Influenza is a potent also interacts with Asc, and there is evidence for both cas-
Nlrp3 activator, and Nlrp3-deficient mice are less able to pase-1 and caspase-5 in the Nlrp1 inflammasome. Anthrax
survive influenza infection.100 Viral RNA was shown to toxin has been shown to be sensed by Nlrp1, and in mice the
prime the inflammasome via TLR7, whereas activation of Nlrp1b isoform has been shown to be a key determinant of
Nlrp3 was mediated by the influenza-encoded M2 ion chan- susceptibility to infection by Bacillus anthracis.110 A role for
nel, which led to potassium efflux.101 Other viruses such as Nlrp1 in antiviral immunity is suggested from the obser-
the DNA viruses vaccinia and varicella zoster, and RNA vation that Kaposi sarcoma herpes virus encodes a protein
viruses such as encephalomyocarditis virus and vesicular called Orf63 that blocks Nlrp1-dependent responses and is
stomatits virus all activate the Nlrp3 inflammasome.102 It is required for reactivation and generation of progeny virus.111
therefore likely that the Nlrp3 inflammasome is key for the Finally, mutations in Nlrp1 are associated with the skin dis-
fever response to all these viruses, as IL-1beta is a key pyro- ease vitiligo.112
gen during infection. AIM2
The Nlrp3 inflammasome can also be activated by fungi, The final inflammasome to be characterized contains AIM2,
notably Candida albicans and Saccharomyces cerevisiase.103 which also couples to caspase-1 via Asc. AIM2 is a member
Again, there is interplay with TLRs, as zymosan (a PAMP of the PYHIN family of proteins, which contain a pyrin do-
in the cell wall of yeast) acting via TLR2 primes the Nlrp3 main and HIN200 domain, which binds DNA.113 The pyrin
inflammasome for activation. Aspergillus fumigatus can also domain in AIM2 allows it to interact with Asc. It is activated
activate the inflammasome.104 directly by DNA and has been shown to be activated by DNA
Finally, parasites can also activate the Nlrp3 inflamma- from bacteria such as Listeria monocytogenes and Francisella
some. Plasmodium falciparum, the causative agent of ma- tularensis, and viruses such as vaccinia virus and cytomega-
laria, secretes a crystal called hemazoin, which activates lovirus.114 Responses to these pathogens in vivo are all im-
Nlrp3.105 Malaria severity is less in Nlrp3-deficient mice. paired in AIM2-deficient mice.
Schistosomes have also been shown to activate Nlrp3.
Overall, therefore, fever, a common response to infection,
is likely to involve the Nlrp3 inflammasome in many cases. Retinoic Acid-Inducible Gene I–Like Receptors
RLRs are a family of RNA helicases that contain a DExD/H
IPAF box as the defi ning protein domain.115 They occur in the
Another inflammasome complex comprises the NLR IPAF, cytosol of most cell types and sense viral RNA. The key
also known as NLRC4. IPAF contains an N-terminal CARD, output from RLRs is type I interferons. There are three
a central NOD domain, and a C-terminal LRR. It responds RLRs: RIG-I, melanoma differentiation associated fac-
to bacterial flagellin, which activates caspase-1 via IPAF.106 tor 5 (Mda5), and laboratory of genetics and physiology 2.

RNA ligand
miochondrion
Viral infection
CARD CARD
CARD
K63
Uq FIG. 15.5. Retinoic Acid-inducible Gene I (RIG-I)
RIG-I MAVS Signaling. In response to viruses such as influenza,
Riplet
Trim 25 RIG-I is activated by viral ribonucleic acid. This leads
to a dramatic conformational change, such that the
TBK-1 caspase recruitment domain (CARD) domains (or-
ange) move out from the protein structure and en-
gage with the CARD domain in MAVS, which is in the
mitochondrial outer membrane. RIG-I also undergoes
IRF-3
K63-linked ubiquitination by Riplet and TRIM25, an
event required for activation. MAVS then activates
TBK-1, which phosphorylates IRF3, leading to induc-
Type I interferons tion of type I interferons and an antiviral response.
Figure 15.5 illustrates RIG-I signaling. RIG-I and Mda5 An excellent example of immune evasion by viruses is
sense a wide range of viruses and have shared structural the targeting of MAVS by the hepatitis C protease NS4.124
features. They both have two CARD domains, a DEAD box This leads to MAVS degradation and impaired RIG-I sig-
Helicase domain, a repressor domain, and a C-terminal do- naling. NS4 is currently the focus of targeting by a prote-
main. Structural insights have revealed that upon binding ase inhibitor, which could have utility in the treatment of
RNA, a major conformational change occurs releasing the hepatitis C.125
repressor domain and shifting the two CARDs out of the RIG-I signaling is a tightly controlled process as
protein to engage with the downstream signaling molecule overactivation could kill the host; in fact, this may be the
MAVS.116–118 Laboratory of genetics and physiology 2 lacks reason for the lethal effects of influenza and other viruses.
the CARDs and is thought to function as a regulator of Ubiquitination is a key process here. K63-linked ubiquitina-
RIG-I and Mda5 signaling. tion of RIG-I by the ubiquintin ligase Riplet is required for
Viruses detected by RLRs include Sendai, respiratory RIG-I signaling.126 TRIM25 also causes K63-linked ubiq-
syncytial virus, measles, rabies, influenza, Ebola, hepatitis uitination, and this has been shown to enhance the inter-
C, myxoma, West Nile virus, dengue, and vaccinia. RIG-I action with MAVS.127 RNF125, on the other hand, causes
preferentially recognizes RNA sequences that contain 5′tri- K48-linked ubiquination of RIG-I, leading to its degradation
phosphorylated ends, which serves to distinguish them from in the proteosome.128 The deubiquitinase CYLD removes the
host RNA sequences.119 Polyuridine motifs in the RNA are K63-linked ubiquitin from RIG-I, inhibiting RIG-I activ-
also important for recognition, and such motifs occur in ity.129 Reversible ubiquitination is therefore a key control
hepatitis C, Ebola, and influenza virus. mechanism for RIG-I.
Retinoic Acid-Inducible Gene I–Like Receptor Signaling Cooperation between Retinoic Acid-Inducible
The key process in the RLR signaling mechanism is the Gene I–Like Receptors and Other Pattern Recognition
recruitment of the adapter protein MAVS (also known as IPS- Receptors in Antiviral Immunity
1, Visa, and Cardif). MAVS is a membrane-bound CARD- As is the case with NLRs and TLRs, RLRs and other PRRs
containing protein that occurs in the outer mitochondrial also cooperate in host defense, in this case against viruses.
membrane and also on the membrane of peroxisomes.120 A good example is the response to West Nile virus. This
Activation of RIG-I and Mda5 leads to a translocation to virus is transmitted by mosquitoes; upon injection into
these membranes via the interaction with MAVS. The tran- skin, RIG-I triggers an interferon response from resident
scription factors IRF3 and IRF7 are activated by MAVS via Langerhans cells and skin fibroblasts.130 Some virus will
the kinases TBK-1 and IKK-epsilon.121 These in turn lead to then disseminate to the draining lymph node where it in-
induction of type I interferons and other anti-viral genes, fects resident macrophages and dendritic cells. The TLRs
with the type I interferons then activating subsequent antivi- are then engaged, particularly TLR7, and induce cytokines
ral responses via the JAK/STAT signaling pathway. RIG-I can to promote adaptive immunity.131 Some virus will escape to
also interact with Asc and activate caspase-1, placing RIG-I as the spleen, and RIG-I will again be engaged here. The situa-
another inflammasome component.122 Another key protein tion for TLR3 is more complex because TLR3-deficient mice
that interacts with RIG-I is STING. This protein was found have improved survival rates.132 SARM knockout mice, on
to be involved in DNA-dependent signaling but also appears the other hand, have increased lethality,133 consistent with
to be involved in RNA-dependent signaling via RIG-I.123 the inhibitory effect of SARM on TLR3 signaling.

Retinoic Acid-Inducible Gene I–Like Receptors and interferon-inducible gene and was then shown to bind a
Adaptive Immunity 70-base-pair double-stranded DNA sequence from vac-
As with other PRRs, RLRs have also been shown to be key cinia. Of particular interest is the structure of IFI16 be-
for adaptive immunity via the induction of cytokines, in cause, similar to AIM2, it also senses DNA but activates
this case type I interferons. Interferons are key regulators of caspase-1. IFI16 has a pyrin domain and two HIN200 do-
T cells, promoting their survival and clonal expansion after mains, placing it in the PYHIN family. AIM2 and IFI16
antigen presentation. Interferons also promote the activity are therefore part of the AIM2-like receptor family.
of cytotoxic T cells and promote antibody production in B IFI16 signals via a protein termed STING, which is found
cells. They also promote major histocompatibility complex in the endoplasmic reticulum membrane.143 DAI has also
class I expression as well as costimulatory molecules on den- been found to signal via STING. The function of STING is
dritic cells. to interact with TBK-1 and promote IRF3 activation. STING
is in fact essential for TBK-1 activation. STING-deficient
PKR, 2′-5′-Oligoadenylate Synthase, and RNaseL mice have impaired interferon induction in response to bac-
Prior to the RLRs, PKR and 2′-5′-oligoadenylate synthase terial, viral, and mammalian DNA, as well as to several DNA
were known to sense viral RNA. PKR is a serine/threonine viruses, including herpes simplex virus-1, human cytomeg-
protein kinase that contains three double-stranded RNA alovirus, vaccinia virus and baculovirus, and Plasmodium
binding domains in the N terminus and a kinase domain falciparum.144,145 STING is also required for the immune re-
in the C terminus.134 It is activated by double-stranded RNA sponse to DNA vaccines, presumably because of the essential
and therefore acts as an intracellular PRR. A key substrate role it plays in the induction of interferons.
for PKR is the translation initiation factor eIF-2alpha.135 STING also plays a role in the induction of interferon
This directly restricts both cellular and viral protein synthe- by Francisella tularensis and Listeria monocytogenes. Cyclic
sis. PKR can also induce apoptosis in virally infected cells. dinucleotides produced by Listeria have been shown to be
Oligoadenylate synthase (OAS) is another double- directly sensed by STING, implying that it is actually a PRR
stranded RNA sensor.136 It is highly induced by Type I inter- for these dinucleotides.146
ferons. Oligoadenylate produced by OAS activates RNAseL.137 There is also evidence for viral evasion of DNA-sensing
This enzyme degrades viral RNA but can also degrade ribo- pathways. Cytomegalovirus encodes a protein termed M45
somal RNA, limiting translation of host and viral mRNAs. that blocks DAI signaling.147 Another cytomegalovirus pro-
Of particular note is the observation that RNAseL can also tein is pUL83, which inhibits IFI16.148 The pox virus protein
generate ligands for RIG-I from viral RNA.138 This indicates M13L has been shown to target AIM2.149
that the OAS/RNAseL axis can serve to amplify antiviral
responses triggered by double-stranded RNA. There is also
evidence that RNAseL can also promote Mda5 activation. C-Type Lectin Receptors
Based on the nature of RNAseL-generated products, both The term C-type lectin was first used to describe a family
RIG-I and Mda5 can therefore recognize fragments of RNA of calcium (hence “C”)-dependent carbohydrate-binding
that are specific cleavage products from viral RNA. (lectin) proteins.150 The carbohydrate binding region was in
a carbohydrate recognition domain and the more general
term C-type lectin domain is used to describe this fam-
Cytosolic Sensors of Dexoyribonucleci Acid that ily of receptors. Dectin-1 is the best characterized mem-
Induce Type I Interferons ber. It recognizes β -1,3-linked glucans, which occur in the
The process of DNA sensing in the cytosol that leads cell wall of fungi, some bacteria, and plants.151 It binds a
to type I interferons is not as well understood as RNA range of fungal pathogens, including Candida, Aspergillu,
sensing. As mentioned previously, TLR9 can sense hypo- Coccidiodes, Pneumocystis, and also bacterial pathogens such
methylated CpG motifs in DNA, but this occurs in endo- as Mycobacteria.151 Two related proteins are CLEC-2 and
somes. Intracellular delivery of mammalian or bacterial DNGR-1.152,153 CLEC-2 has been shown to bind HIV, whereas
DNA can, however, induce type I interferons in a TLR9- DNGR-1 has been shown to bind an unknown ligand seques-
independent manner. The pathway was shown to involve tered in living cells and exposed upon cell death. All three
IRF3 and TBK-1, as with RLRs. Three mechanisms have of these C-type lectin receptors (CLRs) signal via a tyrosine
been proposed. The fi rst is DNA-dependent activator of kinase termed Syk.154 This protein couples to CARD9, which
IFN-regulatory factor (DAI) that interacts with DNA and in turn can activate NF-κ B and promote inflammatory
enhances type I interferons.139 The precise role of DAI, gene expression. It has also been shown to activate Nlrp3
however, is not clear because DAI-deficient mice have a and promote IL-1beta production,155 These CLRs can also
normal response to infection with DNA viruses.140 The promote phagocytosis. Similar to other PRRs, there are also
second mechanism involves RNA polymerase III. This synergies. Dectin-1 and TLR2 when both activated strongly
host enzyme has been shown to convert viral DNA into promote the production of proinflammatory cytokines.156
RNA, which can then be detected by RIG-I.141 This process DC-SIGN is another CLR but instead of signaling via Syk,
has been shown to occur during infection with Epstein- it activates Raf-1 and promotes MAP kinase activation.157 It
Barr virus, herpes virus. and the bacterium Legionella binds to mannose and fucose and has been shown to bind
pneumoniea . The third process involves DNA sensing with HIV, measles virus, severe acute respiratory syndrome, den-
IFI16.142 This protein was identified as the product of an gue virus, and mycobacteria. Raf-1 is able to modulate the

NF-κ B pathway by regulating phosphorylation of the p65 classes of pathogens. The main function of MBL is to acti-
subunit of NF-κ B. vate the lectin pathway of complement. It is associated with
two serine proteases, MASP-1 and MASP-2.164 These prote-
Phagocytic Receptors ases, when activated by ligand-bound MBL, cleave the C2
The PRRs clearly form a major class of innate immune and C4 complement proteins initiating the complement cas-
receptors. What distinguishes them from other innate rec- cade. MBL can also act as an opsonin, binding to C1qRp.
ognition systems is their ability to change gene expression MBL deficiency has been found in humans and leads to in-
in target cells, the complexity of the gene expression profiles creased susceptibility to various infections.165
ultimately giving rise to the specific effector mechanisms re-
quired to clear the provoking pathogen. Another group of Pentraxins. C-reactive protein (CRP) and serum amyloid P
pathogen sensors, however, are proteins that recognize the both belong to the pentraxin family.166 They are acute phase
pathogen but whose function it is to promote phagocytosis. protein in human and mouse, respectively. They are pro-
duced mainly by hepatocytes in response to cytokines such
Scavenger receptors. Scavenger receptors (SRs) are in the as IL-1 and IL-6. They bind to bacterial surfaces and rec-
plasma membrane and are defined by their ability to bind ognize phosphorylcholine. They promote opsonization and
modified low-density lipoprotein.158 Six classes are known but can activate the classical complement pathway via C1q. CRP
most is known about two: SR-A and macrophage receptor with is a classical diagnostic indicator of inflammation. Another
collagenous structure. Both contain a collagenous region and pentraxin is PTX3, which plays an important role in host
an SR cysteine-rich domain. Both are trimers. They have a role defense against bacteria and fungi.167 It is expressed on mul-
in low-density lipoprotein uptake, but also recognize LPS and tiple cell types, including dendritic cells, macrophages, and
lipoteichoic acid from bacteria. SR-A–deficient mice are more neutrophils. It is strongly induced by TLRs. Similar to CRP
susceptible to infection by Listeria monocytogenes, herpes sim- and serum amyloid P, it binds directly to bacterial and fun-
plex, and malaria.159 They are also more susceptible to LPS- gal membranes and promotes phagocytosis.
induced sepsis, implying a role in LPS clearance. Macrophage
receptor with collagenous structure can bind gram-positive Peptidoglycan recognition proteins. Peptidoglycan recogni-
and gram-negative bacteria and promote their phagocytosis.160 tion proteins are a family of PRRs that bind peptidogly-
can.168 They are conserved across evolution and were first
Macrophage mannose receptor. The macrophage mannose
described in Drosophila. They all contain a highly conserved
receptor (MR) is a protein found in the plasma membrane
peptidoglycan domain, and some also have transmembrane
of macrophages and contains cysteine-rich and fibronectin
domains. Most are secreted, however. Four have been de-
type 2 domains, followed by eight carbohydrate recognition
scribed in humans. Their functions have yet to be fully elu-
domains of the C-type lectin family.161 MR recognizes man-
cidated, although they can clearly bind peptidoglycan. There
nose-rich carbohydrates from bacteria but can also promote
is some evidence that they might compete with other pepti-
phagocytosis of high-mannose glycoproteins from the host.
doglyan recognition PRRs, such as TLR2, possibly acting to
MR has been shown to be required for phagocytosis of mul-
limit inflammation, although this has to be demonstrated.
tiple bacteria, fungi, and protozoans. There is evidence that
MR can complex with TLRs and promote TLR signaling,162
yet another example of cooperation between PRRs. CONCLUSION
Secreted pattern recognition molecules. There are also a Clearly our knowledge of innate immune processes has
range of secreted pattern recognition molecules that have greatly improved in the past few years. Multiple mechanisms
various functions, although a unifying feature is opsoniza- exist that keep pathogens at bay, and several receptor classes
tion. They are secreted into the circulation during infec- can drive innate defence mechanisms via the production of
tion and can reach high concentrations, effectively acting effector cytokines and other mediators. The innate immune
as acute phase proteins. Three classes have been particularly system provides an important link to the adaptive response,
well-studied. mainly via the activation of dendritic cells. Other cell types
that participate include mast cells, basophils, and natural
Mannan-binding lectin. MBL is the best characterized of killer cells, which are discussed at length elsewhere. Innate
the collectin family, which have a C-type lectin domain.163 immunity therefore provides us with a first line of defense,
It binds to the terminal mannose and fucose residents on but via the inflammatory response and the activation of
microbial carbohydrates and importantly can recognize the adaptive immunity orchestrates the major mechanisms that
spatial arrangement of these sugars, which differs accord- lead to restoration of the host after infection.
ing to the host forms. It recognizes a wide range of patho-
gens, including Staphylococci, Streptococci, Pseudomonas,
Klebsiella, Salmonella, Mycobacteria, Candida, influenza, ACKNOWLEDGMENTS
human immunodeficiency virus, and trypanosomes. As This work is supported by Science Foundation Ireland and
can be seen from this list, MBL can therefore recognize all the European Research Council.

CHAPTER
16 Dendritic Cells
Kang Liu • Michel C. Nussenzweig
INTRODUCTION density centrifugation in bovine serum albumin gradients

Immunology interfaces with medicine at many points, the and adherence to glass were originally used to purify DCs.3
most prominent being infectious diseases, cancer, trans- At the time, macrophages were thought to be the key ac-
plantation, allergy, and autoimmunity. Antigen-specific cessory cell because they composed a major population of
immunity is critical for vaccination and for protective adherent cells and also because their role in innate immu-
immunity against pathogens and tumors. Moreover, when nity had been long appreciated. Moreover, purified antigen
control over immune responses is lost, the result is autoim- loaded macrophages induced immune responses when re-
mune disease. T cells are an essential element that regulates injected into their hosts.4 However, macrophages failed to
the balance in immunity by killing infected cells, helping an- show robust activity in induction of antibody responses in
tibody formation, and suppressing autoimmune responses. vitro, and they rapidly degraded ingested antigens suggest-
However, T cells are incapable of recognizing native anti- ing that they would be unable to present it to lymphocytes.5–7
gens. Instead, they recognize processed peptides presented
by major histocompatibility complex (MHC) molecules. Dendritic Cells Initiate T-Cell Responses
Dendritic cells (DCs) are professional antigen-presenting
In addition to their morphology, the cell surface composi-
cells that inform the fight against invasive pathogens while
tion of the newly discovered DCs proved to be distinct. DCs
enforcing tolerance to self- and harmless environmental
were found to express high levels of MHC antigens.8,9 At the
antigens. They capture pathogens and receive signals from
time, the precise function of the MHC in antigen presenta-
pathogens that influence the outcome of immune responses.
tion was not known, but it was already clear that the MHC
On the basis of these signals, DCs orchestrate antigen-
was a key genetic determinant of immune responses and that
specific T-cell differentiation toward Th1, Th2, Th17, or Tfh
it encoded many of the antigens involved in graft rejection.
pathways. Alternatively, they can silence self-reactive T cells
The mixed leukocyte reaction (MLR) was considered an
by inducing deletion, anergy, or active regulation (via regu-
in vitro model system to study graft rejection and was used
latory T [Treg] cells). This chapter will focus on the discov-
by Steinman and Cohn to examine the function of DCs.10
ery, function, and development of DCs, and the mechanisms
They found that DCs were nearly two orders of magni-
by which they link innate immunity to adaptive immunity.
tude more potent than unfractionated spleen cells, B cells,
or macrophages in stimulating allogeneic T cells in the
DISCOVERY MLR.10,11 On the basis of these experiments, they specu-
DCs were discovered as part of an effort to understand the lated that DCs were the accessory cells that present antigen
initiation of immunity. The mouse as an experimental ani- to T cells to initiate immune responses.12 However, their
mal was critical because of a system, developed by Mishell conclusions were not widely accepted by immunologists be-
and Dutton,1 in which mouse spleen cell suspensions could cause unlike traditional immune responses, the MLR did
be stimulated to generate antibody responses in culture. One not require priming.13
of the early observations was that lymphocytes alone were Proof that DCs are antigen-presenting cells came from
not sufficient to induce antibody-forming cell responses and experiments performed by Nussenzweig and Steinman
that an adherent accessory cell was required. The search for who cocultured DCs with responding T cells and hapten-
the accessory cell led to the discovery of DCs. In this sec- modified thymocytes. They showed that DCs induced
tion, we will review the early methods used to identify and MHC-restricted cytotoxic T cells specific to the hapten and
study DCs. that macrophages and other purified populations of lym-
phoid cells were nearly inactive as accessory cells.14
Identification and Isolation
DCs were discovered by Steinman and Cohn when they Monoclonal Antibodies to Dendritic Cells
examined spleen adherent cells by phase contrast micros- Although the morphologic and functional differences
copy.2 They then employed physical techniques to fractionate between DCs and other leukocytes were striking, few
spleen cells and purify DCs. These cells were found to adhere immunology laboratories were equipped to perform the
to plastic or glass, had low buoyant density, and did not bind purification procedures required to study DCs until mono-
to erythrocytes coated with antibody. Sequential steps of clonal antibodies to DCs became available. A series of
381

mouse DC-restricted monoclonal antibodies were produced antigens targeted to one or the other cDC in vivo. Antigens
by Steinman and others starting in 1980s, including 33D1 targeted to CD8α + and CD8α− DCs by chimeric monoclo-
(specific for the cell surface receptor DCIR2/Clec4a4),15 nal antibodies that bind endocytic receptors Dec205/CD205
NLDC-145 (specific for the adsorptive endocytosis recep- or DCIR2/Clec4a4 are biased to strong CD8 or CD4 T-cell
tor, DEC205/cluster of differentiation [CD]205),16 and N418 responses, respectively.28
(specific for the CD11c integrin).17 These monoclonal anti- In conclusion, the two major subsets of DCs have dis-
bodies were used to establish the unique functions of DCs tinct microanatomic locations, gene expression profi les, and
within heterogeneous mixtures of leukocytes and to iden- functions. However the CD8α + cDC in lymphoid tissue and
tify DCs in situ. For example, 33D1 was used to selectively CD103 + subset in nonlymphoid tissue DCs subsets are far
deplete DCs from mouse spleen suspensions and shows that better characterized than the CD8α− cDC or CD103 − sub-
this depleted MLR stimulating activity12 and accessory func- set, which are heterogeneous and difficult to distinguish
tion in the Mishell-Dutton system.18 from activated monocytes.
Dendritic Cell Subsets ANATOMIC DISTRIBUTION

Shortman and colleagues first showed that DCs are het- Lymphoid Organs
erogeneous and suggested the existence of DC subsets.19,20 Peripheral lymphoid organs (ie, spleen, lymph nodes, and
These subsets differ in their anatomic distribution, cell sur- mucosal associated lymphoid tissues) are the sites where
face marker expression, and function. In the mouse, three primary immune responses develop, including activation
major groups of DCs exist in the steady state: plasmacytoid of helper, killer, and antibody-forming cells. Initiation of
DCs (pDCs), conventional DCs (cDCs), and migratory DCs immune responses is facilitated by the highly organized
(mDCs). pDCs are important mediators of antiviral immu- structure of the peripheral lymphoid organs, with lym-
nity through their ability to produce large amounts of type I phocytes segregated into B- and T-cell areas. In addition to
interferons (IFNs) upon viral infection (see following dis- lymphocytes, the T-cell area also contains large interdigi-
cussion). cDCs are composed of two major subsets, namely tating cells that were initially thought to be macrophages.
CD8α + and CD8α−.21 mDCs are present in nonlymphoid However, cytologic and functional studies of spleen and
tissues such as the liver, gut, skin, lung, and aorta, and they lymph nodes in rat and mice revealed that interdigitating
are composed of two main subsets CD103 + and CD103 −.22 cells lack phagosomes and lysosomes, and are poorly phago-
These nonlymphoid tissue DCs are referred to as mDCs be- cytic.29–31 These features distinguished interdigitating cells
cause they transit from tissue to lymphoid organs.21,23 from macrophages and suggested that they might be related
The two major subsets of mouse spleen cDCs have to the cDCs isolated from mouse spleen.
some overlapping functions. Both can process and pres- Mouse interdigitating cells were later proven to be cDCs
ent antigens to T cells and also secrete cytokines such as based on immunohistochemistry using DC-specific mono-
interleukin (IL)-12, which can influence the ultimate po- clonal antibodies. In human lymphoid organs, interdigitating
larization of the T-cell response to pathogens. However, the cells in the T-cell area express CD83, a member of immuno-
two subsets also have unique functions in vivo and are not globulin superfamily. This marker is not conserved in the
redundant.24–26 mouse where the interdigitating cells in the T-cell area of
Some of the differences in cDC subset function may be lymph nodes express CD11c and DEC205/CD205. In con-
accounted for differential expression of toll-like receptors trast, the two cDC subsets in the mouse spleen segregate
(TLRs) or other mediators of cDC activation. For example, into distinct microanatomic compartments. CD8α + cDCs
spleen CD8α + cDCs express TLR3 (recognizing double- in spleen T-cell areas resemble those in the lymph nodes in
stranded ribonucleic acid [RNA]) but lack TLR5 (recogniz- that they express DEC-205, whereas those in the bridging
ing flagellin) and TLR7 (recognizing single-stranded RNA), channels and red pulp express DCIR2.28
whereas CD8α – cDCs express TLR5 and TLR7 but have low In the T-cell area, cDCs form a dense network of cells
levels of TLR3.27 Comparative analysis of messenger RNA that are primarily sessile but constantly protrude and
expression profi les between the two DC subsets revealed retract large membrane folds into their environment.32 T
that the two are as different from each other as are T- and cells actively migrate through the DC network in search of
B-lymphocytes.28 This is in part reflected in the antigen pro- antigen, and when they find a cDC presenting cognate anti-
cessing capacity of the two cDCs types. CD8α + cDCs are gen, they pause for a prolonged interaction.33 The large DC
specialized for MHCI cross-presentation, and enriched in membrane protrusions are thought to increase the surface
Tap1, Tap2, calreticulin, calnexin, Sec61, ERp57, ERAAP, as area for antigen presentation and facilitate antigen detection
well as cystatin B and C, all of which are involved in MHCI by migrating T cells.34
presentation or inhibition of enzymes that process peptides
for MHCII presentation. In contrast, CD8α – cDCs, which
are biased for MHCII presentation, were enriched in ca- Nonlymphoid Organs
thepsins C, H, and Z, asparagine endopeptidase, GILT, and In addition to the lymphoid organs, DCs are also found
H2-Mbeta 1, all of which are implicated in the MHCII anti- in most nonlymphoid organs and in all epithelial surfaces
gen-processing pathway.28 The immunologic consequences that contact the environment. In organs such as heart, lung,
of this specialization can be seen after immunization with kidney, the dermal layer of skin, and meninges and choroid

CHAPTER 16 DENDRITIC CELLS | 383
DENDRITIC CELL DEVELOPMENT

Cell Surface Comparison between
TABLE 16.1 Origin of Conventional Dendritic Cells in the
Dendritic Cells and Macrophage
Lymphoid Organs
DC Macrophage DCs, like all other leukocytes, develop from bone-marrow–
MHC II + −/lo derived hematopoietic stem cells. Monocytes, macrophages,
FcR (CD16/CD32) − + granulocytes, megakaryocytes, and erythrocytes differenti-
CD11b −/lo + ate from a common myeloid progenitor (CMP) whereas B,
F4/80 − + T, and natural killer (NK) cells differentiate from a common
CD115 −/lo + lymphoid progenitor.43–45 Although early cell transfer and
CD, cluster of differentiation; DC, dendritic cell; MHC, major histocompatibility complex.
genetic experiments were interpreted to indicate that DCs
originated from lymphoid and myeloid progenitors,46–49
subsequent work showed that all DCs are derived from my-
eloid progenitors.50
plexus in the brain, they are found in interstitial spaces that
are drained by lymphatics.35–37 Like their lymphoid coun-
Relationship between Dendritic Cells and Monocytes
terparts, interstitial DCs can be distinguished from macro-
In the steady state, DCs and monocytes can readily be dis-
phages by abundant expression of MHCII and low levels of
tinguished based on cell surface marker expression. For ex-
lysosomal hydrolases. In comparison to macrophages, which
ample, DCs are CD11c + CD115−/loMHCII+ and monocytes
are also abundant in the interstitial spaces, interstitial DCs
are CD11b + CD115 + F4/80 + (see Table 16.1). In contrast,
are less phagocytic and more sensitive to radiation. Tissue
under inflammatory conditions or in tissues such as lung
macrophages and DCs can be distinguished by immunohis-
and intestine, where they can be exposed to pathogenic and
tology; the former typically express F4/80 and lysosomal hy-
nonpathogenic microbes, monocytes develop many of the
drolase, whereas the latter express MHCII and CD11c. The
characteristic features associated with DCs, and the distinc-
standard markers for distinguishing DCs and macrophages
tion becomes far more difficult. Moreover, the question of
in interstitial tissues are listed in Table 16.1. Importantly,
how DCs are related to monocytes and macrophages during
although these markers help to differentiate macrophage
development was a vexing problem that was only resolved
and cDCs in most tissues, they are not sufficient in defining
very recently.
cell lineages unless used in combination with functional and
The idea that monocytes might be the direct precursors
developmental analysis. This is because CD11c and MHCII
of DCs in vivo was suggested by elegant experiments with
can also be expressed by lung macrophages and induced on
human CD14 + monocytes.51 Monocytes undergoing reverse
monocytes during inflammation.38
transmigration, which simulates their entry into lymphatic
vessels from tissues, lose expression of monocyte markers
Afferent Lymphatics CD14 and CD64 and upregulate human leukocyte antigen-
Afferent lymphatics in rats, rabbits, guinea pigs, sheep, DR and CD54.51 Phagocytosis further stimulates expression
and humans contain cells with motile processes that were of CD80, CD86, human leukocyte antigen-DR, DC-LAMP,
named veiled cells because of their large membrane protru- and CD83, all of which are expressed by DCs.51 Consistent
sions.34,39–41 These cells were found to be poorly phagocytic, with these cell surface changes, activated monocytes stimu-
express high levels of MHCII, and are potent stimulators of late allogeneic T-cell proliferation in MLRs.
the MLR in vitro. Lymphadenectomy and thoracic duct can- Similarly, in mice, microspheres injected intracutaneously
nulation in larger animals such as rats and sheep made it are phagocytosed by CD11b + F4/80 + monocytes that can
possible to collect these cells from afferent lymph. Using this then be found in the draining lymph nodes 3 to 4 days after
technique, Huang et al. revealed that veiled cells are con- the injection.52 These cells display high expression of MHCII,
stantly carrying intestinal epithelial contents via lymphatics MHCI, CD86, and stimulate allogeneic T-cell proliferation;
to the cDC network in the T-cell area of the mesenteric however, unlike DCs, they express low levels of CD11c and
lymph nodes.42 Therefore, in the healthy individual, these no CD8α . Therefore, monocytes in the skin did not appear
veiled cells continually carry and present self- and harmless to become classical cDCs upon migration to lymph nodes;
antigen to the adaptive immune system, thereby contribut- nevertheless, these activated monocytes shared many of the
ing to the induction of peripheral tolerance (see following features associated with DCs. Finally, activated monocytes
discussion); during infection, they carry pathogen-derived that appear in the spleen during Listeria monocytogenes in-
antigens and induce specific T-cell activation. Thus veiled fection expressed many of the cell surface features of cDCs.53
cells correspond to mDCs. These cells produce tumor necrosis factor (TNF) and induc-
ible nitric oxide synthase (iNOS) and were named TNF/
iNOS–producing DCs (Tip-DCs).53 Tip-DC deficiency led
Dendritic Cells in the Skin and Other Body Surfaces to lethal L. monocytogenes infection; however, this effect
Several types of DCs, monocytes, and macrophages exist in was independent of the development of adaptive CD8 T-cell
the skin and at other body surfaces. Professional antigen- immunity.53
presenting cells in the skin include Langerhans cells (LCs) in The idea that DCs develop from monocytes was tested
the epidermis, and CD103 + and CD103 − DCs in the dermis. by several laboratories in direct transfer experiments in

mice without clear success.54–57 Indeed, it was always the the combination of these markers and persistent expres-
nonmonocyte fraction of the bone marrow that produced sion of Flt3 defined pre-DCs and allowed for their isolation
cDCs.54,56 The only exception to this rule occurred after in- in mice.60
flammation induced by Freund’s Complete Adjuvant (CFA), Pre-DCs migrate from the bone marrow to the blood and
when monocytes differentiated into CD11c + CD11b + Mac3 + then to peripheral lymphoid organs and nonlymphoid tis-
cells, which again are phenotypically distinct from cDCs.56 sues.60 These cells comprise ∼0.5% of all leukocytes in bone
The inability of monocytes to produce cDCs in steady- marrow, 0.02% in blood, 0.05% in the spleen, and 0.03% in
state lymphoid organs was further confirmed in a series the lymph nodes. Pre-DCs have a short half-life in the blood
of elegant genetic studies using reporter mice carrying of < 1 hour65 ; this, together with the small number of these
Rosa26-Stopflox EGFP.58 In these mice, a lox-flanked Stop cells in blood and tissues, may explain why previous efforts
transcription signal must be deleted by Cre expression to identify pre-DCs failed and why human pre-DCs have yet
before GFP can be transcribed. Breeding LysM-Cre and to be identified.
Rosa26-Stopflox EGFP mice results in lysozyme promoter– Pre-cDCs isolated from the bone marrow, blood, or
driven Cre expression in monocytes and neutrophils, as well spleen give rise to both CD8α + and CD8α− cDCs in lym-
as deletion of the Stop sequence from Rosa26-Stopflox EGFP phoid and nonlymphoid organs.60 Thus, the pre-DC is a pro-
in these cells. As a result, monocytes and their progeny are genitor with significant plasticity. In conclusion, the DC and
irreversibly marked with enhanced green fluorescent pro- monocyte lineages split in the bone marrow, where MDPs
tein expression. Classical DCs in the peripheral lymphoid give rise to both monocytes and CDP; the latter produce
organs remained enhanced green fluorescent protein nega- pre-DCs, which migrate from bone marrow through the
tive in these mice; therefore, they cannot be derived from blood to the periphery to give rise to DCs (Fig. 16.1).
monocytes.58
In conclusion, experiments in humans and mice show
that monocytes can develop some of the features of DCs Origin of Nonlymphoid Tissue Dendritic Cells
under conditions of inflammation in vivo, or when cultured Skin
with cytokines in vitro, but they are not precursors of cDCs. The skin contains LCs in the epidermis and CD103 + and
CD103 − DCs in the dermis.
Dendritic Cell Progenitors in the Bone Marrow
If monocytes do not produce DCs, the question remains: Epidermis . LCs were discovered in 1868 by Paul Langerhans.
what is the origin of DCs and where does commitment to Although LCs were initially thought to be of neural crest or-
this lineage occur? The first clue in solving this problem came igin,66,67 this view was changed with the discovery that LCs
from the identification of a common precursor for mono- are closely related to leukocytes,68,69 express Fc receptors70
cytes, macrophagesm and classical DCs (macrophage-DC and MHCII antigens,71 and are competent to present antigen
progenitors [MDPs]54). MDP are Lin− CX3CR1+ CD11b − CD to primed T cells.72
115 + cKit + CD135 + and account for 0.5% of all bone marrow Experiments using bone marrow chimeras and parabiotic
mononuclear cells in mice.54,59 When cultured with granu- mice demonstrated that LCs are long-lived cells that di-
locyte-macrophage–colony stimulating factor (GM-CSF) in vide in situ in the skin.73 These cells originate from fetal
vitro or adoptively transferred into mice, these cells produced liver– derived progenitors in an Flt3L-independent manner,
macrophages and DCs, but not neutrophil granulocytes, B- and they are only replaced by bone marrow–derived
or T-lymphocytes, or NK cells.54,59,60 Therefore, MDPs are hematopoietic cells during inflammation.73,74 Thus, the ori-
more restricted than CMPs, from which they are derived. gin of the LCs is different from that of cDCs.
A DC-restricted progenitor that produces cDCs and Nevertheless, mouse LCs have been valuable as models
pDCs but not monocytes in vitro61 or in vivo62 was then for studying the differentiation and migration of antigen-
identified based on reduced cKit (CD117) and residual Flt3 presenting cells.75–81 LCs in the epithelial layer of the skin
expression and named common-DC progenitor (CDP; can take up particles82 and contain distinct granules, known
Lin− CD115 + Flt3 + CD117lo). The CDP is downstream of as Birbeck granules, which are a type of endocytic organelle
CMP and MDP because adoptive transfer of either CMP or enriched in langerin.83,84 While LCs express Fc receptors
MDP gives rise to CDP and monocytes.60 More importantly, and MHCII in situ, the expression of Fc receptors drops and
these experiments showed that the split between the mono- MHCII molecules redistribute from intracellular compart-
cyte and DC lineages occurs in the bone marrow between ments to the cell surface in cultured LCs that become more
the MDP and CDP stages of development.60 highly immunostimulatory.75,76,85 Cells that resemble LCs
are also found in other stratified squamous epithelia such as
Migratory Pre-Dendritic Cells the vagina, cervix, anus, pharynx, and esophagus.
cDC precursors must migrate from the bone marrow to the LCs show many of the phenotypic features of cDCs, in-
lymphoid organs through the blood. However, MDPs and cluding morphology, ability to redistribute large amounts of
CDPs are restricted to the bone marrow.60 The identity of MHCII from the endocytic system to the cell surface, and
the DC precursor was suggested by the finding that low- capacity to stimulate allogeneic T-cell proliferation in vitro
density CD11c + MHC II− SIRPα lo cells isolated from blood, after activation.75,78,86–88 Importantly, however, LCs differ
bone marrow, and periphery had the potential to produce from DCs in that they are resistant to irradiation, they are
cDCs.56,63,64 Although this group of cells was heterogeneous, self-renewing in situ and not of pre-DC origin in the steady

Lymphoid tissues
CD8α+ CD8α–
cDC cDC
Pre-DC Pre-DC Pre-DC
3L
Flt
CDP
3L
Flt Non-lymphoid tissues
Flt3L
MP MDP PDC
M
-C PDC CD103+ CD103–
SF
Mono DC DC
Mono
Mono
Tip-DC
Bone marrow Blood Periphery
Commitment Migration Differentiation
FIG. 16.1. Dendritic Cell and Monocyte Origin and Development.
state, and finally, like monocytes and macrophages, their sion of the IFN regulatory factor (IRF)8, Id2, and Batf3 tran-
development is dependent on macrophage-colony stimu- scription factors (see following discussion). Finally, both
lating factor (M-CSF) and not on Flt3L.89 LC precursors, dermal CD103 + dermal DCs and CD8α + cDCs specialize
which are CX3CR1+ CD45 +, colonize the dermis during in antigen cross-presentation.28,96 The CD103 − dermal DC
late embryonic development.90,91 In mice, these precursors compartment appears to be a heterogenous mixture of cells,
complete their differentiation into LCs by postnatal d2, some of which are Flt3 dependent and of pre-DC origin and
whereupon they undergo a burst of cell division between others M-CSF–dependent and of monocyte origin.37
postnatal d2 and d7, which accounts for the dramatic in- Skin-draining lymph nodes contain subpopulations of
crease in LCs during this developmental window. The LC DCs that resemble CD103 + and CD103 − dermal DCs.92,95
precursor in skin appears to resemble the MDP.90 Recently, Their migration from the skin to the draining lymph nodes
fate-mapping analysis revealed that LCs derive from primi- is dependent on CCR7 but independent of CD62L. In the
tive macrophages during yolk sac stage of embryogenesis, steady state, these cells account for 20% of lymph node DCs
and not from postnatal hematopoietic progenitors.91 These while the remainder of the DCs are derived from blood pre-
two lineages can be differentiated by their dependence of DCs.60 Inflammation dramatically enhances the migration
Myb gene. The absence of Myb impaired HSC development, of dermal DCs to the lymph node.52,58
but did not affect development of the yolk sac macrophage
progenitor that gives rise to LCs, microglia and liver Kupffer Gut, Kidney, Lung, and Liver
cells. This finding places the LC in its own differentiation Similar to the skin, CD103 + and CD103 − DCs are found
pathway distinctive from cDC and monocyte.91 in nearly all mouse nonlymphoid tissues.37,97 However, the
origin of these cells was not established until very recently.
Dermis . The dermal layer of the skin contains two sub- CD103 + nonlymphoid tissue DCs express Flt3, a marker ex-
sets of DCs: CD103 + and CD103 − DCs. CD103 + der- pressed on most cells in the DC lineage but not on mono-
mal DCs are langerin + CD11blo, whereas CD103 − dermal cytes. Furthermore, mice deficient in Flt3L or its receptor,
DCs are langerin− CD11b +.92–95 Although both LCs and Flt3, showed significant decreases in CD103 + nonlymphoid
CD103 + dermal DCs express langerin, CD103 + dermal tissue DCs but normal numbers of monocytes.37,98 Finally,
DCs are more closely related to spleen CD8α + cDCs than adoptive transfer of pre-DCs into naïve recipients, or mice
to LCs.25,37,96 CD103 + dermal DCs, like spleen cDCs, are in which CD11c cells had been depleted, resulted in recon-
M-CSF–independent and Flt3-dependent.37 Unlike LCs that stitution of the CD103 + nonlymphoid tissue DC compart-
can self-renew in situ, dermal CD103 + DCs are continually ment in the liver, lung, kidney, and intestine.37,57 In con-
replenished from the same bloodborne pre-DCs that give rise trast, adoptive transfer of 50 times more monocytes failed
to lymphoid organ DCs.95 In addition, like CD8α + splenic to show detectable DC progeny in naïve mice.37,57 Thus, in
cDCs, development of CD103 + dermal DCs requires expres- the steady state, CD103 + DCs in all tissues tested are pre-DC

and not monocyte-derived. Only after depletion of endog- lymphocytes, polymorphonuclear leukocytes, monocytes,
enous CD11c + cells were monocytes able to produce some and DCs express one or more of the receptors for these
CD103 − DCs.37,57 The idea that all CD103 + nonlymphoid growth factors. For example, Flt3 is expressed very early in
tissue DCs are derived from pre-DCs and not from mono- hematopoiesis on shared progenitors of lymphocytes and all
cytes is further supported by genetic experiments that show myeloid cells.109 However, with the exception of pre-DCs,
a common requirement for Batf3, Id2, and IRF8 for lym- Flt3 expression is lost by nearly all leukocytes by the time
phoid tissue CD8α + DCs, and CD103 + nonlymphoid tis- they enter circulation.60 Conversely, M-CSF receptor expres-
sue DCs, but not monocytes or macrophages (see following sion is retained by cells in the monocyte lineage but lost on
discussion).24,37,99,100 pre-DCs and DCs.60
In conclusion, in addition to DCs in the lymphoid organs, As might be predicted from their patterns of expression,
pre-cDCs also contribute to CD103 + DC development in M-CSF and Flt3L are essential for normal development of
nonlymphoid tissues. CD103 − DCs appear to be more heter- the monocyte and DC lineages, respectively. For example,
ogenous; some of these cells are derived from pre-DCs while mice deficient in M-CSF or its receptor show deficiencies
others may be of monocyte origin. in monocyte, macrophage, and LC development but have
normal spleen and lymph node DCs.110 In contrast, DC de-
velopment is impaired in mice deficient in either Flt3L or
Plasmacytoid Dendritic Cells Flt3, but monocyte development is normal.59,111 Consistent
pDCs were described by pathologists as a unique popula- with these observations, administration or overexpression
tion of cells in the peripheral lymphoid organs that resemble of Flt3L results in selective expansion of DC populations,
plasma cells and monocytes and were referred to as plas- including fully differentiated DCs in lymphoid organs.59,112
macytoid monocytes. In 1990s, Svensson et al. described
a population of cells in human blood that produces large
amounts of type I IFN upon exposure to herpes simplex Dendritic Cell Cultures
virus, and they named these cells IFN-producing cells.101 In the early 1990s, Banchereau and Schuler and their col-
Both of these populations correspond to the pDCs described leagues established in vitro culture systems to produce
by Liu and colleagues.102–104 human monocyte–derived DCs (moDCs), and Inaba and
pDCs were first purified from human tonsil and blood on Steinman for the mouse equivalent.76,113,114 These culture
the basis of being CD4 + CD3 − CD11c−. They were then shown systems are widely used today for basic and clinical studies.
to develop DC-like morphology upon culture with IL-3 and Bone marrow cells from mice or monocytes from humans
anti-CD40. Liu and colleagues discovered that purified pDCs are cultured for 6 days in medium containing GM-CSF
produced 100 to 1,000 times more type I IFN than the other to produce cells with dendritic morphology that exhibit
blood cell types following viral activation.104 Subsequently, modest phagocytic activity, express CD11c and MHCII, and
pDCs were identified and isolated from lymphoid tissues in stimulate the MLR. Differentiation to moDCs is further
mice, and later in rats, monkeys, and pigs. In humans, the stimulated by addition of lipopolysaccharide, which induces
pDCs express CD4, MHCII, CD68, CD123, and blood DC a series of changes that were originally described as matura-
antigen 2 (BDCA2), but do not express other lineage marker tion. Notably, monocytes do not give rise to DCs in the lym-
such as CD3 (T cell), CD14 (monocyte), CD19 (B cell), CD16 phoid organs in vivo (see previous discussion), and GM-CSF
and CD56 (NK cell), or CD11c, BDCA1, and BDCA3 (cDC). is entirely dispensable for DC development in vivo115 ; thus,
They are identified as CD4 +, CD11c−, and Lin− (CD3, CD14, the moDCs generated in GM-CSF culture are not entirely
CD16, CD19, CD56) cells.105 In mice, pDCs express CD11c representative of the cDCs found in the lymphoid tissues.
(lower levels than cDCs), Gr1, B220, and PDCA1; low levels moDCs may be more closely related to activated monocytes
of MHCII; do not express CD11b or CD19;and they are fre- than to steady-state DCs in the lymphoid organs.
quently isolated by virtue of CD11c and B220 expression.106 Authentic cDCs can be produced in in vitro cultures of
pDCs are closely related to cDCs; they originate from bone marrow cells supplemented with Flt3L, which is an es-
the same progenitor, the CDP. Development of pDCs is also sential growth factor for DCs in vivo.116 Flt3L cultures produce
critically dependent on Flt3L. Unlike cDCs, however, pDCs all of the known subsets of DCs, including cells with features
do not express high levels of MHC II in the steady state and of pDCs, CD8α+/CD103+ cDCs, and CD8α− cDCs.116
therefore, their primary function is not antigen presentation
to CD4 T cells. Instead, pDCs focus on sensing infection
and producing type I IFN. Depletion of pDC blocks antivi- Transcriptional Regulation of Dendritic
ral IFN responses but also results in loss of Treg homeostasis, Cell Development
impaired differentiation of Th cells, and initiation of CD8 A number of different transcription factors are known to
T-cell responses (see following discussion).107,108 regulate DC development, but to date none of these are cDC
specific. For example, signal transducer and activator of tran-
scription (STAT)3 is required for cDC development in part
Growth Factors for Dendritic Cells and Monocytes because this transcription factor is downstream of Flt3.117
GM-CSF, M-CSF and Flt3L are essential growth factors for Binding of Flt3L to its receptor on the surface of DC pre-
myeloid cell development in vitro and in vivo. During the cursors or cDCs induces activation of the phosphoinositide
early stages in myeloid development, shared progenitors of 3-kinase–Akt–mammalian target of rapamycin signaling

pathway; this leads to activation of STAT3.118 Activation in more mature cells reversed the development, suggesting
of STAT3 is also regulated by the transcriptional repres- differentiation plasticity that needs to be constantly reen-
sor, Gfi-1, deletion of which leads to reduction of lymphoid forced.126,127 Similar to mice, E2-2 haploinsufficient humans
DCs and increase of LCs.119 PU.1, an E-twenty-six family with Pitt-Hopkins syndrome show a pDC defect.126
transcription factor, is a third transcription factor that is
downstream of Flt3 signaling.120 Although PU.1 deficiency is Batf3
lethal, mice reconstituted with PU.1-deficient hematopoietic Batf3 is expressed at high levels in DCs and is specifically re-
cells show a number of hematopoietic defects, including quired for development of CD8α + cDCs and CD103 + non-
strong reduction in CD8α + and CD8α− cDCs. lymphoid tissue DCs.24,128 Batf3 − / − mice failed to mount
antitumor and antiviral immune responses, demonstrating
The Interferon Regulatory Factors a key role of CD8α + cDCs and CD103 + nonlymphoid tissue
IRF2, 4, and 8 are prominent transcription factors that DCs in priming CD8 T cells.
regulate DC diversification. IRF2− / − mice exhibit reduced
CD8α− cDCs, IRF4 − / − mice display reduced CD8α− cDCs Notch
and slightly reduced pDCs, and IRF8 − / − mice have reduced Terminal DC differentiation appears to depend on signaling
CD8α + cDCs, pDCs, and LCs.100,121–123 through Notch receptors and their transcriptional effector
RBP-J. Inactivation of RBP-J in DCs results in severe reduc-
NF-kB/Rel tion and functional impairment of splenic CD8α− cDCs.129
Members of the NF-kB/Rel family also contribute to DC de- In addition, Notch2 deletion leads to the loss of CD103 +
velopment. CD8α− cDC numbers are dramatically reduced nonlymphoid tissue DCs in the intestinal lamina propria
in RelB − / − mice.124 TRAF6 belongs to TNF receptor– and to a corresponding decrease of IL-17–producing CD4 T
associated factor family and acts upstream of the NF-kB cells in the intestine.130
cascade. Consistent with a role for NF-kB in cDC develop- In conclusion, DC differentiation is dependent on a
ment, TRAF6 − / − mice also show a defect in the CD8α− number of interacting transcription factors (Table 16.2).
cDC compartment.125 Together, they turn on expression of lineage-specific genes
and suppress alternative developmental programs.
Basic Helix-Loop-Helix Transcription Factors
Basic helix-loop-helix factors (E12, E47, HEB, E2-2) interact
with helix-loop-helix Id (inhibitors of deoxyribonucleic acid Human Dendritic Cell Development
[DNA] binding) proteins to control DC development. Loss of Research on human DCs is difficult because it is largely
Id2 leads to a severe reduction of CD8α + cDCs and absence limited to peripheral blood, whereas in the mouse, most of
of LCs. Another helix-loop-helix family member, E2-2, plays the work has been done on DCs residing in lymphoid or non-
a crucial role in guiding pDC development and maintain- lymphoid tissues. As a result, human DC research has relied
ing stability of the pDC lineage. Deletion of E2-2 in develop- on peripheral blood monocytes cultured in the presence of
ing precursors blocked pDC development, whereas deletion cytokines, which generate moDCs and not cDCs.
TABLE 16.2 Transcription Factors Determine Dendritic Cell Development and Homeostasis in Vivo
Classical DC Tissue DC
Plasmacytoid DC CD8a+ CD8a- CD103+ CD103- LC
PU.1−/− −−− −−−−
Irf8−/− −−−− −−−− n.a. −−−− −−
Irf2−/− n.a. −−− −
Irf4−/− − n.a. −−−
RelB−/− n.a. −−−−− n.a.
Gfi1−/− −− −− −− ++
ID2−/− n.a. −−−− n.a. −−−− −−−−−
E2-2−/− −−−− n.a. n.a.
Stat3−/− −−−− −−−−
Stat5−/− −−− −−− −−−
Ikaros−/− −−−− −−−−−
Xbp1−/− −−− −− −−
Batf3−/− n.a. −−−−− n.a. −−−−− n.a.
Runx3−/− ++ − −−−−−
RBP-J−/− n.a. −−−−
Notch2−/− − −−−− −−−− n.a.
−, decrease; +, increase; CD, cluster of differentiation; DC, dendritic cell; LC, Langerhans cell; n.a., not altered.

Despite these obstacles, three different subsets of DCs increased serum Flt3L, which induces expansion of myeloid
have been found in human blood. These subsets are referred progenitors. In humans, as in mice, loss of DCs is associated
to as BDCA1, 2, and 3 based on their expression of cell with loss of Treg cells, confirming the existence of a homeo-
surface markers.131 BCDA1+ DCs resemble mouse CD8α− static feedback loop these two cell types141 (see following
cDCs, BDCA2 + CD11c− DCs are equivalent to mouse pDCs, discussion).
and BDCA3 + DCs are equivalent to mouse CD8α + cDCs
(Fig. 16.2). These interspecies associations were initially
based on similarities in gene expression between human DENDRITIC CELL HOMEOSTASIS
and mouse DC subsets,132 and were recently confirmed in Lymphoid Organs
functional experiments.133–137 Moreover, human DCs can be Based on their rapid labeling kinetics and loss of label in
obtained from cultures of human cord blood supplemented pulse chase experiments with BrdU, it was proposed that
with Flt3L and GM-CSF.134 DCs have a very short half-life.136,137 However, these early
The human equivalents of the bone marrow–derived experiments are difficult to interpret in view of the more
DC precursors, MDP, CDP, and pre-DC remain to be iso- recent finding that DCs in lymphoid and nonlymphoid
lated. However, recent clinical studies revealed genetically tissues divide in situ.37,65,142
defined syndromes associated with DC deficiency, which The half-life of cDCs in tissues was measured in para-
may shed some light on human DC development. Three bionts and shown to vary from 5 to 7 days in the spleen,
different genetic lesions have been associated with the triad lymph nodes, liver, and kidney, and as many as 25 days in
of DC deficiency, monocytopenia, and opportunistic infec- the lung.37,65 cDC homeostasis in all organs is maintained
tions. GATA2 mutation leads to a loss of DC, monocytes, B through a dynamic balance of three parameters: continu-
cells, and NK cells.138,139 These patients have normal LCs and ous input of pre-DCs from the blood, limited cDC divi-
macrophages, suggesting that LCs and tissue macrophage sion in situ, and cell death. Fitting BrdU incorporation and
have different origin from DCs and monocytes in humans. parabiosis separation data into mathematical equations
Mutations in IRF8 are associated with disseminated bacille produced a numerical estimate of the rate of DC precur-
Calmette-Guérin infection.140 Two different mutations in sor input (∼4,300/hour) and DC death (∼9,600/hour) in the
IRF8 have been identified. The K108E variant is associated mouse spleen.65
with an autosomal recessive severe immunodeficiency with
a complete absence of circulating monocytes and DCs; the
T80A variant is an autosomal dominant, milder immu- Regulation of Dendritic Cell Division
nodeficiency and is associated with selective depletion of cDCs in mouse lymphoid and nonlymphoid tissues divide
circulating DCs. Loss of DCs due to IRF8 mutation also in situ for 10 to 14 days before being replaced by pre-
leads to myeloproliferation in humans and in mice due to DCs.65 In the steady state, their division is regulated by
Myeloid DC
Human Mouse Human Mouse
CD141+ CD8α+ (lymphoid) CD11chi CD11c+
CLEC9A+ CD103+ (non- CD11b+ CD11b+
CD11b– lymphoid) CD1c+ M-CSFR
XCR1+ CLEC9A+ CD1a– (blood) CX3CR1
TLR3+ CD11b– CD1a+ (tissue,
FLT3+ XCR1+ lymph node)
CD11clow TLR3+
FLT3+
CD11c+
- Cross-presentation - MHC class II-restricted antigens
- MHC class I-restricted antigens - CD4 T cell responses
- CD8 T cell responses
Monocyte-derived DC Plasmacytoid DC
Human Mouse Human Mouse
CD14+ CD11b+ CD123+ BST2+
CD11b+ CX3CR1+(intestine) CD303+ B220+
CX3CR1 CD304– Siglec-H4+
CD209+ Tip-DC TLR7 TLR7
CD11b+
TLR9 TLR9
CD206+
- TNF CD209+
- iNOS - Type I interferons
FIG. 16.2. Functionally Correlated
- Bacterial antigens - Durable memory responses
Dendritic Cell Subsets in Human and
- Secondary immune responses Mouse.

lymphotoxin-beta and Flt3L. Flt3L impacts nearly all stages Even in the absence of invading pathogens, some DCs are
of cDC development in lymphoid and nonlymphoid tissues, always migrating from tissues to lymph nodes. Studying af-
and is essential for their development.59,112,143 In contrast, the ferent lymph is difficult in mice, but in rats, for example,
effects of lymphotoxin-beta appear to be limited to CD4 + DCs move along liver sinusoids in hepatic lymphatics to ce-
spleen DCs.142 How the levels of circulating Flt3L are regu- liac lymph nodes,150,151 and DCs from intestine migrate to
lated is not known, but there appears to be a feedback loop mesenteric lymph nodes.42,152 These cells are not found in
between tissue cDCs and Flt3L that results in upregulation the efferent lymph, indicating that most of the migrating
of Flt3L in the serum in response to cDC depletion,144 loss DCs die after their arrival in lymphoid tissues. The DCs that
of Tregs,60,145,146 or TLR ligand injection.147 In all instances, migrate in the steady state might have several functions: to
increased Flt3L production leads to increased export of DC replenish immature populations, to transport self- or envi-
progenitors from the bone marrow and DC division in the ronmental antigens, or to be on patrol to identify invaders.
periphery. DC migration is a regulated process, controlled at the
Conversely, increasing the number of DCs by Flt3L injec- level of chemokine production and chemokine receptor ex-
tion leads to an increase in the number of Treg cells. Thus, pression and function.153 Some immature DCs can express a
the feedback loop between DCs and Tregs is mediated by repertoire of receptors (eg, CCR1, CCR2, CCR5, CCR6, and
Flt3L, and that maintains the physiologic numbers of these CXCR4) that bind inflammatory chemokines (eg, CCL5,
two cell types in the steady state (Fig. 16.3).146 Alterations in CCL2, CCL3, CCL4, CCL20, and CXCL12). DC activa-
this mechanism lead to immune imbalance and can alter the tion due to inflammatory signals is usually associated with
course of autoimmune disease in mice.146,148 For example, downregulation of chemokine inflammatory receptors and
increasing the number of DCs in diabetes prone nonobese the de novo expression of CCR7, the receptor for CCL19
diabetic mice leads to increased numbers of Tregs and a delay and CCL21. In mice homozygous for an autosomal recessive
in disease onset. Similarly, Flt3L injection can protect mice mutation in CCL21 (paucity of lymph node T cells or plt/plt
from inflammatory bowel disease by increasing the number mice), naive T cells fail to home to secondary lymphoid or-
of cDCs and thereby the number of Tregs. Thus, the Flt3L- gans. DCs in these mice also fail to accumulate in the spleen
mediated homeostatic feedback loop between Tregs and DCs and in the T-cell areas of lymph nodes.154 Similarly, CCR7− / −
has clinical implication for vaccine design as well as the con- mice show defective secondary lymphoid organ architecture,
trol of autoimmunity. Finally, this mechanism is entirely defective homing of DCs and lymphocytes, and defective
consistent with the proposed role of DCs in maintaining entry of DCs into lymphatic vessels at peripheral sites both in
tolerance and regulating immunologic responses in vivo.149 the steady state and inflammation conditions.155 CCL19 and
CCL21 also increase the maturation and proinflammatory
differentiation of cDCs.156 Therefore, chemokine/chemokine
Dendritic Cell Migration receptor interactions not only orchestrate the DC migration
DC migration from peripheral tissues into lymphoid organs but also influence their immunogenic potential.
is key to their sentinel and antigen trafficking functions. There are many other examples in which specific che-
Upon microbial contact or stimulation by inflammatory cy- mokines control the traffic of select populations of DCs
tokines, nonlymphoid tissue resident cDCs traffic through (Table 16.3). In skin exposed to ultraviolet light, LCs dis-
afferent lymphatics to the T-cell areas of the lymph node appear and are replaced in 2 weeks. The recruitment of LC
where they can participate in the initiation of immune precursors from blood is dependent on their expression of
responses. Migration is dependent on CCR7, which is up- CCR2.73 During murine listeriosis, CCR2 is also required
regulated by stimuli such as TLR ligands that activate DCs. for TNF-inducible nitric oxide synthase–producing DCs to
migrate into the spleen.53 Migration of pDCs into inflamed
lymph nodes and their redistribution during inflamma-
MHC-II tion depend on CXCR3 and CCR7. CCR6 is used by cDCs
to populate epithelial surfaces during inflammation.157–159 In
DCs Tregs the steady state, CXCL14 is important for LC progenitors to
establish themselves in the skin.160 All these findings indi-
cate that the differential expression of chemokine receptors
by DCs (and their subsets) at different stages of their life his-
IL-10, TGFβ tory determines their location in vivo.
FL
CTLA-4
ANTIGEN PRESENTATION
Antigen Capture
Tregs DCs Macropinocytosis
DCs in culture continuously form 0.25 to 1.0 um pinocytic
MHC-II vesicles. These vesicles allow DCs to sample a large volume
of extracellular fluid and soluble proteins that are present at
FIG. 16.3. Homeostatic Feedback Loop between Dendritic Cells and low concentrations. Expression of aquaporin 3 and 7 on DCs
Regulatory T Cells. may also contribute to macropinocytosis.161

TABLE 16.3 Animal Models Demonstrate Dendritic Cell Migration Depending on Chemokine Receptor
DC migration
CCR2−/−, CCR5−/− The recruitment of new Langerhans cells is dependent on their expression of the CCR2 chemokine
receptor and on the secretion of CCR2-binding chemokines by inflamed skin.73 Recruitment of TNF/
inducible nitric oxide synthase–producing DC subset in spleens of Listeria monocytogenes–infected
mice is dependent on CCR2 and independent of CCR5.53,246
CCR6−/− CCR6 mediates DC localization, lymphocyte homeostasis, and immune responses in mucosal tissue. In CCR6−/−
mice, DCs expressing CD11c and CD11b are absent from the subepithelial dome region of Peyer patches.247
CCR7−/−, plt/plt CCR7 coordinates the primary immune response by establishing functional microenvironments in second-
(CCL19 and CCL21−/−) ary lymphoid organs.155 Plt/plt mice lacking CCR7 ligands, CCL19 and CCL21, have defects in lymphocyte
homing and DC localization.154 CCR7 also governs skin DC migration under inflammatory and steady-state
conditions.248 CCR7 ligands, CCL19 and CCL21, induce a proinflammatory differentiation program in DCs.156
CCR8−/− CCR7 and CCR8 pathways are used by monocyte-derived DCs during mobilization from skin to lymph nodes.249
CXCR3−/− Plasmacytoid DCs migrate to inflamed lymph nodes, produce interferon-α, and help lymph node DCs to
induce antiviral CTLs.250,251
CTL, cytotoxic T-lymphocyte; DC, dendritic cell; plt, paucity of lymph node T cells; TNF, tumor necrosis factor
Receptor-Mediated Phagocytosis DCs is tolerance and not immunity. Steady-state DCs

Phagocytosis is triggered by the attachment of extracellular express high levels of pattern recognition and activation
particles to surface receptors, which mediate particle up- receptors, allowing them to sense changes in the environ-
take. Multiple receptors on cDCs enhance recognition and ment, including pathogens and inflammatory cytokines.
ingestion of particulates including pathogens and dying These signals induce extensive differentiation to a mature
cells. Importantly, many of the receptors also recognize mo- or activated state characterized by increased levels of cell
lecular patterns on pathogens and as a result, activate DCs. surface MHC and coactivator expression, as well as cyto-
In addition, DCs also express Fc receptors that mediate in- kine secretion. In the activated state, DCs initiate potent
gestion of opsonized particles and delivery to intracellular and specifically polarized T-cell immune responses. Thus,
compartments that facilitate TAP-1–dependent cross-pre- pathogen sensing by DCs is an important mechanism that
sentation to CD8 T cells. DCs express both activating and links innate pathogen recognition to the adaptive immune
inhibitory forms of Fc receptors.162 The inhibitory receptors response.
help maintain DCs in an immature tolerogenic state. DCs
that lack the Fc γRIIB inhibitory receptor are more read- Toll-Like Receptor Signaling
ily activated because the immunoreceptor-based tyrosine Microbes and viruses induce DC activation in part by en-
activation motif–associated Fc γRIIA receptor is no longer gaging TLRs. DCs express nearly all known TLRs and also
subject to inhibition.163 These observations have been ex- express additional cytoplasmic receptors that recognize
tended in mice and humans through the identification of pathogen patterns such as melanoma differentiation associ-
monoclonal antibodies that selectively block inhibitory ated protein 5 (MDA-5), RIG-I, and DDX41, and nucleotide
receptors.164,165 oligomerization domain–like receptors, although expression
Other than TLRs, DCs express many different types of of specific receptors is restricted to distinct DC subsets.166–168
pattern recognition receptors including those of the spe- In many cases, a single microbe will trigger several differ-
cific intercellular adhesion molecule-3-grabbing nonin- ent TLRs expressed by DCs; therefore, a single TLR may
tegrin (SIGN) family, and multiple C-type lectins such be redundant for inducing immunity to many, but not all,
as macrophage mannose receptor (MMR), DEC205, and pathogens. Thus, in humans, loss of TLR3 results in suscep-
DCIR2. MMR binds a range of bacteria, yeasts, and viruses tibility to herpes simplex virus-1 infection but not to other
through interactions between a mannose-type carbohydrate microbes that also contain the TLR3 ligand.169 Pathogen rec-
recognition domain and pathogen-associated high man- ognition by TLR ligation induces rapid DC activation and
nose structures. Notably, DCs deficient in any single C-type upregulation of MHC and costimulatory molecules, and
lectin receptor, such as MMR, DEC205, or DCIR2, develop also triggers secretion of inflammatory cytokines such as
normal adaptive immune response to dying cells, or patho- IL-1α , IL-6, TNFα , IL-18, and IL-12, which are critical in
gens, suggesting that this important class of receptors is driving T-cell differentiation.170–172
redundant.
Cytokines, T, and Natural Killer T Cell Signals
DCs can also sense proinflammatory cytokines such as
Dendritic Cell Maturation TNFα and IL-1β, and direct cellular contact with activated
DCs exist in two functionally distinct and phenotypically T or NKT cells by ligation of CD40, a TNF receptor family
different stages. In the steady state, DCs in most tissues molecule that is highly expressed on the surface of DCs.
are equipped to capture and present antigens to T cells, Many of these receptors, including TLRs, IL-1R, and TNF
but the outcome of antigen presentation by steady-state receptors, activate DCs through the NF-κ B pathway.

Antigen Processing to the plasma membrane. In addition, DC activation reduces

DCs process ingested antigens for presentation on both lysosomal cystatin C, which leads to increased activation of
MHC class I and MHC class II. Although DCs are excellent cathepsin S and more efficient processing of Ii.85
MHCII antigen-presenting cells, they are not unique in this
respect. Other cells, including activated B cells and mono-
DENDRITIC CELLS LINK INNATE AND
cytes, can also process and present antigens for presenta-
tion by MHCII efficiently. For example, in germinal centers,
ADAPTIVE IMMUNITY
activated B cells are the key antigen-presenting cell. Plasmacytoid Dendritic Cells
Macrophages endocytose antigens and rapidly digest Nucleic acid–sensing pDCs are activated when they sense
them. In contrast, DCs sequester and preserve the captured pathogen or self-derived nucleic acids and are involved in an-
antigen for later presentation. Preservation of antigens is tiviral immunity and autoimmune diseases. Whereas mono-
critical for immunogenicity and depends on two DC spe- cytes express TLR4 (lipopolysaccharide sensor) and cDCs
cializations. The first is low lysosomal protease activity. express TLR3 (double-stranded RNA sensor), pDCs selec-
Macrophages contain high levels of lysosomal proteases, in- tively express TLR7 (single-stranded RNA sensor) and TLR9
cluding cathepsin S, cathepsin L, cathepsin K, and asparagine (DNA sensor). pDCs endocytose viruses and sense their nu-
endopeptidase. These proteases enable rapid degradation of cleic acids using TLR7 and TLR9, both of which reside in
internalized proteins to single amino acids. Compared to the endosomal compartment. TLR7 detects single-stranded
macrophages, DCs express low amounts of proteases, result- RNA viruses, such as influenza, respiratory syncytial virus,
ing in a limited capacity for lysosomal degradation.173 This Sendai, and vesicular stomatitis virus, and synthetic single-
low-level proteolytic capacity appears to be crucial as the stranded RNA analogs such as R848/Imiqimod; TLR9 de-
peptides loaded on MHC and recognized by T cells consist tects DNA viruses, such as herpes simplex virus-1, herpes
of peptides between 8 to 17 amino acids. Consistent with simplex virus-2, and murine cytomeglavirus, or cytosin-
this idea, mutations in amino acids in antigens that increase phosphatidyl-guanosin oligodeoxynucleotides. Viruses that
their resistance to lysosomal proteolysis increase immuno- enter the cytoplasm of pDCs are detected after autophagy,
genicity by decreasing the rate of proteolysis.173 a conserved cell-autonomous process involving lysosomal
A second feature of DC lysosomes that facilitates anti- degradation of cellular organelles to deliver cytoplas-
gen presentation is that DC lysosomes are less acidic com- mic RNA to TLR7-containing endosomal compartments.
pared to those in professional phagocytes. In macrophages Both TLR7 and TLR9 use MyD88 as the adaptor protein
and neutrophils, the low pH of the phagosome and endo- for activation; thus, MyD88-deficient pDCs have defective
some activates lysosomal proteases delivered to these en- responses to a wide range of viruses that enter the cyto-
docytic vesicles. In DCs, the pH in these compartments is plasm. Recruitment (binding) of MyD88 to the TLR leads
less acidic because the assembly of V-ATPase, an adenosine to assembly of a signal-transducing complex that includes
triphosphate–dependent vacuolar proton pump, appears to IRAK4, TRAF6, Bruton’s tyrosine kinase, and IRF7.177 This
be incomplete in the lysosome of steady-state DCs, leading signaling complex is critical for IFN production by pDCs.
to alkalinization of the endocytic compartment.174 In addi- In most cells, production of IFNα depends on binding of
tion, DCs show efficient recruitment of Nox2 to endosomes IFNβ to interferon-α/β receptor (IFNAR), which induces
and phagosomes. Nox2 produces reactive oxygen species, IRF7 expression. In pDCs, however, constitutive high-level
which consumes protons and thereby reduces the acidi- expression of IRF7 facilitates a rapid IFNα response that is
fication of the endocytic compartment. NOX2-defective independent of IFNAR signaling.105
DCs exhibit a higher level acidification of the phagosome IFN production by pDCs is rapid, starting 4 hours after
and increased proteolysis and decreased antigen presenta- exposure to TLR ligands and peaking after 24 hours when
tion.175 Furthermore, Nox2-deficient patients, who suffer pDCs become refractory. Type I IFN produced from acti-
from chronic granulomatous disease, show impaired cross- vated pDCs not only directly inhibits viral replication, but
presentation.176 Together, the lower levels of proteolytic ac- also activates NK, B cells, and cDCs.178 For example, human
tivity and decreased acidity in endocytic compartment lead immunodeficiency virus (HIV) or herpes simplex virus in-
to a decrease in the rate of antigen digestion and increased fection stimulates pDCs to produce type I IFN and CD40L,
availability of partially processed peptides for loading on which in turn activates cDCs and facilitates antigen presen-
MHC. This unique feature of DCs may also help preserve tation and development of antiviral immunity. Similarly,
the antigen during the migration of nonlymphoid tissue pDC nucleic acid sensing and cDC activation occurs in
DCs from the site of antigen capture to the lymph nodes. autoimmune diseases. In patients with systemic lupus ery-
Antigens are actively taken up by immature DCs and tar- thematosus, pDCs are continuously activated by circulating
geted to MHCII-positive lysosomes. However, these antigens immune complexes composed of self-DNA or RNA leading
are sequestered in an intact form and are not efficiently uti- to high circulating levels of IFNα .178
lized for formation of MHCII-peptide complexes. Immature Given the potential pathologic consequences of high-level
DCs do form stable MHCII dimers, but their presence does IFNα production by pDCs, it is not surprising that this re-
not result in immunogenic complexes. DC activation by mi- sponse is highly regulated. pDCs express an array of negative
crobial products or proinflammatory cytokines results in regulators to downregulate IFNα production. In humans,
redistribution of MHCII from intracellular compartments two surface molecules, BDCA2 and immunoglobulin-like

transcript 7 (also known as LILRA4), were found to sup- lular compartments to the plasma membrane; the upregula-
press TLR-induced IFN production from pDCs. Both recep- tion of costimulatory molecules such as CD40, CD80, and
tors mediate their inhibitory effects through cytoplasmic CD86; and change in profi les of cytokine and chemokines
immunoreceptor-based tyrosine activation motifs, and both such as TNFα and IL-12. All these changes likely contribute
BDCA2 and immunoglobulin-like transcript 7 associate to the initiation of T-cell immunity.
with the γ-chain of the high-affinity Fc receptor for im- The communication between DCs and T cells is a dia-
munoglobulin E (Fc εRIγ).179 Other immunoreceptor-based logue in which the DCs also respond to T cells. For example,
tyrosine activation motif–mediated inhibitors of pDC IFN CD40 and TRANCE/RANK receptor on DCs are ligated by
production include NKp44 and sialic acid binding Ig-like the corresponding TNF family member expressed on acti-
lectin (Siglec)-H, which recruits DAP12.180 vated and memory T cells (ie, CD40L and RANK-L).193–195
This leads to increased DC survival and in the case of CD40
ligation, upregulation of CD80 and CD86, secretion of IL-12,
Interactions with Innate Lymphocytes and release of chemokines such as IL-8 and MIP-1α and β.
DCs interact with innate lymphocytes (eg, NK, NKT, and γδ
T cells) in ways that enhance the functions of both cell types.
DCs produce IL-12, IL-15, IL-2, and IFNα /β that affect dif- Controlling the Quality of the T-Cell Response
ferent facets of NK cell function, whereas the innate lympho- DCs are involved in critical T-cell fate decisions such as
cytes induce DC activation.181,182 Injection of activated cDCs clonal selection, tolerance versus immunity, Th1 versus Th2,
leads to the recruitment of NK cells into the draining lymph and even memory. In the presence of mature DCs produc-
nodes while exposure of cDCs to TLR stimuli induces IL-2 ing IL-12 or IFNs (as might occur when DCs are ligated by
and IL-12 production, which in turn activate NK cells to pro- CD40L or infected with viruses), CD4 T cells differentiate
duce IFNγ. DCs also present different glycolipids on CD1d along a Th1 pathway for IFNγ production. The latter in turn
molecules to the invariant T-cell receptor on NKT cells. activates the antimicrobial activity of macrophages and pro-
These include endogenous lysosomal glycosphingolipids motes killer T-cell differentiation. However, in the presence
(eg, iGb3),183 microbial lipids,184,185 or synthetic glycolipids of exogenous IL-4, DCs induce T cells to differentiate into
(eg, alpha-GalCer)186,187 to activate NKT cells in vivo. Th2 cells, which secrete IL-4, IL-5, and IL-13. These cyto-
Type I IFN appears to be one of the essential mediators of kines help B cells to make antibodies of the IgG1 and IgE
DC activation in vivo. For example, antigen targeted to cDCs isotypes, activate eosinophils, and stimulate fibrosis. A new
can elicit strong Th1 immunity in the presence of PolyI:C. and striking pathway that was first discovered with human
PolyI:C binds to TLR3 on the cell surface and intracellular monocyte–derived DCs involves the epithelial– derived cy-
sensor MDA-5. However, PolyI:C does not directly stimulate tokine thymic stromal lymphopoietin (TSLP). This cytokine
DC maturation because deficiency of TLR3 or MDA-5 on stimulates DCs to induce “inflammatory Th2 cells” that
DCs does not abrogate immunogenicity.188 Instead, PolyI:C produce TNFα (rather than IL-10) in addition to IL-4, IL-5,
stimulates nonhematopoietic cells to produce large amounts and IL-13.196 Recent studies have shown that TSLP receptor
of type I IFN.188 High levels of systemic type I IFN stimulate knockout mice exhibit strong Th1 responses, with high lev-
functional maturation of cDCs, which consequently induce els of IL-12, IFNγ, and IgG2a, but low production of IL-4,
a robust Th1 response. Thus, microbial stimuli–induced DC IL-5, IL-10, IL-13, and immunoglobulin E.197 DCs that are
maturation in vivo engages a cascade of cellular and molecu- activated with either CD40L or TSLP are similar in appear-
lar mechanisms, which may amplify the environmental sig- ance, being rich in MHCII and CD86 co-stimulatory mol-
nals to facilitate optimal immunity. ecules. However, they differ significantly in cytokine and
chemokine production, and the functional consequences for
T cells vary.196
Initiating Adaptive T-Cell Immunity
In the steady state, DCs can endocytose a diverse array of
antigens through multiple receptors on their cell surface. Cross-Presentation by Conventional Dendritic Cells
However, steady-state DCs typically express relatively low CD8 T cells are critical for protective immunity against
levels of surface MHC class I and II products and only low intracellular pathogens and malignant tumor cells. When
levels of costimulatory molecules (eg, CD80, CD86). Upon autologous T cells are incubated with influenza-infected
receipt of an activation stimulus, DCs undergo extensive DCs, strong proliferative and cytotoxic T-lymphocyte (CTL)
differentiation, and they migrate in increased numbers to responses develop within a week. Thus, DCs present endoge-
secondary lymphoid tissues. nous antigens on MHCI to prime CD8 + T cells. Interestingly,
DC–T-cell interactions in the lymph nodes have now DCs can also present exogenous antigen on MHCI, an un-
been studied in the living state with two-photon micros- conventional presentation path called “cross-presentation.”
copy.33,189–192 Naïve antigen-specific T cells arrest on anti- The earliest experiment demonstrating the capacity of
gen-presenting DCs, and this stable interaction lasts for DCs in cross-presentation was done in vitro, when trini-
at least 18 hours. In the steady state, this stable DC–T-cell trophenyl-modified cells were taken up by DCs. DCs can
interaction leads to tolerance while an activation signal present trinitrophenyl, an exogenous antigen, and induce
such as TLR ligation allow DCs to initiate immunity. cDC antitrinitrophenyl CTLs in vitro in a MHC-restricted man-
activation results in redistribution of MHCII from intracel- ner.14 Later, DCs were found to take up exogenous influenza

antigen in the form of nonreplicative virus or apoptotic cells, and present them to T cells in lymphoid organs.206 A criti-
and present them on MHCI to prime CD8 T cells.198 In the cal observation was that bone-marrow–derived cells in
absence of DCs, mice are unable to process several different the pancreatic lymph nodes present peptides derived from
antigens through the exogenous pathway, indicating that insulin-producing β cells from pancreatic islets, leading to
DCs were a major cell type for cross presentation to CD8 T tolerance.207,208 Although DCs were not implicated directly, a
cells in vivo.199 DCs cross present exogenous antigen taken direct role for DCs in maintaining tolerance was established
up as soluble form or via multiple receptors, including FcR, by Hawiger and colleagues by targeting antigens directly
CD91, Lox-1, DEC205/CD205, and DNGR1. CD8 + DCs in to DCs.209 Furthermore, CD8α + cDCs are particularly ef-
mice and their functional equivalent BDCA3 + DCs in hu- ficient in capturing dying cells201 and inducing tolerance.202
mans excel in cross-presenting exogenous antigens.24,134,200 Likewise, when DCs capture innocuous proteins from the
Exogenous presentation or cross-presentation is essential for airway, profound tolerance develops even though the T cells
protective immunity against viruses and tumors. can initially proliferate extensively to the antigen-capturing
DCs in the draining lymph nodes.210
DENDRITIC CELLS CONTROL TOLERANCE Deletion and Anergy
The mechanisms used by DCs to initiate immunity against Mechanisms for peripheral tolerance can be intrinsic (dele-
pathogens pose a major risk for the development of auto- tion and anergy) or extrinsic (through suppressive Treg cells).
immunity, allergy, and chronic inflammatory disease. For Interestingly, the former requires expression of B7 family
example, infection frequently induces some cell death; members on the steady state DCs (eg, PD-L1 and CD86),
therefore, DCs at the site of infection capture pathogens which then ligate PD-1 and CTL antigen-4 on the T cells to
and dying cells, and the two are processed and presented be tolerized.
similarly.201,202 Likewise, at body surfaces, maturing DCs are
also capturing environmental antigens to which the body Regulatory T cells
must remain unresponsive. How do maturing DCs focus the Autoreactive T cells can remain quiescent in the presence of
immune response on antigens derived from the pathogen Treg cells. There are two types of Tregs: naturally occurring
and avoid inducing immunity to self- and nonpathogenic Tregs derived from thymus (natural Tregs) and Tregs induced
environmental antigens? from Foxp3 − CD4 + T cells in the periphery (induced Tregs).211
The groups of Belkaid and Powrie found that a subset
of gut DCs expressing CD103, the αEβ7 integrin, are spe-
Thymic Dendritic Cells Contribute to cialized at inducing Foxp3 + Tregs and maintaining oral
Central Tolerance tolerance.212 CD103 + DCs from mesenteric lymph nodes
Self-reactive thymocytes are deleted by antigen-presenting or gut-associated lymphoid tissue produce transforming
cells during negative selection. The thymus contains two growth factor-β, a cytokine critical for induction of Foxp3 +
major populations of antigen-presenting cells that express Tregs.212 Additionally, CD103 + gut DCs use retinal dehydro-
MHCII, namely, medullary thymic epithelial cells and cDCs. genase to metabolize vitamin A to bioactive retinoic acid,
Both cell types are required for efficient negative selection. which acts as a cofactor for transforming growth factor-β
Medullary thymic epithelial cells express a panoply of self- to induce Foxp3 + Tregs.212,213 Similarly, skin DCs use vitamin
antigens under the control of the autoimmune regulator D3 to induce Tregs.214,215 Therefore, DCs are able to employ
“AIRE.”203 Self-antigens for negative selection in the thymus environmental signals, vitamin A in the gut, and vitamin
also include antigens expressed by DCs, and antigens that D3 in the skin to induce tolerance to harmless foreign anti-
enter the thymus through the bloodstream and are captured gens. The CD8 + and CD103 + DC subset appears to be spe-
by DCs. In the absence of antigen presentation by DCs, nega- cialized at Treg induction. When antigens were targeted to
tive selection of CD4 + thymocytes is impaired.204 In addition CD8α + DEC-205 + or CD8α− DCIR2 + DC subsets in vivo,
to negative selection, thymic cDCs also support the devel- only CD8α + DEC-205 + DCs were able to induce Foxp3 +
opment of Foxp3 + Tregs.205 Thus cDCs contribute to central Tregs from Foxp3 − CD4 + T cells.216
tolerance in the thymus by more than one mechanism. DCs also maintain the homeostasis of Tregs. Loss of DCs
leads to a loss of Treg cells, and the remaining Treg cells ex-
hibit decreased Foxp3 expression. The DC-dependent loss
Dendritic Cells Mediate Peripheral Tolerance in Treg cells leads to an increase in the number of T cells
DCs also mediate tolerance in the periphery. Because cen- producing inflammatory cytokines, such as IFNβ and IL-
tral tolerance alone is incomplete, the immune system must 17.217 Conversely, increasing the number of DCs leads to in-
continually establish tolerance to harmless or “noninfec- creased Treg-cell division and accumulation by a mechanism
tious” antigen in the environment. Therefore, the effec- that requires MHCII expression on DCs.146 Activation of
tive control of self-reactive T cells depends on peripheral β-catenin is essential for DC to control Treg and peripheral
tolerance. For example, mice and humans deficient in tolerance. DCs lacking β-catenin show decreased production
Tregs, which suppress autoreactive T cells in the periph- of immunosuppressive cytokines and a decreased ability to
ery, succumb to autoimmunity at an early age. DCs con- support the differentiation of naïve T cells into Tregs in vitro.218
stantly carry innocuous antigens from the periphery (eg, In summary, DCs have the capacity to induce tolerance
from the skin, airways, stomach, intestine, and pancreas) by several mechanisms.

DENDRITIC CELLS IN CLINICAL IMMUNOLOGY with hemorrhagic fever. Dengue virus targets DCs directly
This section focuses on diseases in which DCs play a patho- through DC-SIGN but also enters DCs as a passenger in
genic role or can be targets for therapy. immune complexes that are taken up by Fc receptors.231,232
When infection occurs through antibody enhancement
mediated by Fc receptor, the infected DCs are involved in
Dendritic Cells in Transplantation induction of the T-cell cytokines that mediate the vascular
DCs play a key role in the outcome of organ and hemato- leak syndrome associated with the infection.
poietic transplantation. In organ transplantation, both
donor and recipient DCs contribute to graft rejection.
Donor DCs in grafted organs migrate to lymphoid organs
Dendritic Cells in Cancer
where they stimulate alloreactive T cells in the recipient to Tumors can suppress immunity in part through their effects
induce organ rejection219,220 ; recipient DCs can also capture on DCs. DC differentiation and activation can be suppressed
antigens from the graft and elicit organ rejection. For exam- by cancer-derived cytokines, such as IL-6, vascular endothelial
ple, in hematopoietic transplantation, recipient DCs initiate growth factor, and IL-10.233 In contrast to their normal coun-
T-cell–induced graft-versus-host reactions.74,221 terparts that activate immune responses, DCs derived from
tumors induce Foxp3 + Tregs234 and IL-13–producing CD4 +
T cells,235 and suppress proliferation of CTLs236 and NKT cells.
Dendritic Cells in Autoimmune Disease
Several human autoimmune diseases appear to involve DCs.
Abundant DCs are found in the synovial exudates of rheu- Dendritic Cell–Targeted Vaccines
matoid arthritis, and TNF-α produced by DCs contributes DC-based vaccines are currently being used in the clinic,
to the severity of the disease. Similarly, in psoriasis, DCs and DC-targeted vaccines are being tested.
infi ltrate lesional skin, produce TNF- α , and polarize T DC-based immune therapy is currently available for treat-
cells toward Th1/Th17222,223 ; additionally, pDCs in psoriatic ing prostatic cancer, but it is far from optimized and is not
lesion become activated to produce type I IFN, which also curative. Monocyte-derived DCs are generated ex vivo, loaded
contributes to the inflammatory response. with tumor cells or tumor antigens, and reinjected into the
In systemic lupus erythematosus, two subsets of DCs patient.237 Scientific and practical problems exist with this ap-
contribute to the onset and severity of the disease; pDCs in proach, including limited responses possibly due to inefficient
patients with systemic lupus erythematosus overproduce migration of monocyte-derived DCs from injection site to the
IFNα , which activates cDCs and interferes with their ability draining lymphoid organs and inefficient antigen presentation.
to maintain peripheral tolerance.178 Finally, DCs have been DC-targeted vaccines are based on the idea that antigens
implicated in mouse models of type I diabetes by carrying delivered specifically to DCs in conjunction with the appro-
self-antigens from the pancreas to draining lymph nodes priate adjuvants will produce strong and lasting immunity.209
where diabetogenic T-cell responses are initiated. In addi- DC-targeted vaccines require that antigens be delivered
tion, DCs cultured from nonobese diabetic mice show an specifically to endocytic receptors on DCs together with the
activated phenotype with increased IL-12 and costimulatory appropriate stimuli to induce DC activation. For example,
molecule expression.224 antigens have been incorporated into antireceptor mono-
clonal antibodies, which are then injected into the vaccine
recipient.209,238,239 This paradigm was established using anti-
Dendritic Cells in Viral Infections bodies to DEC205/CD205, which is abundant on DCs and
DCs mediate antiviral immunity by priming T-cell re- delivers antigens to both MHC class I and II antigen-process-
sponses. However, a number of viruses have evolved ing compartments.240 In mice, antigens targeted to DEC205/
strategies to subvert DCs, and thereby, the immune system. CD205 on activated DCs induce strong immunity against tu-
The interaction between HIV and DCs is a fascinating mors and a number of intracellular pathogens, including ma-
example of viral immune subversion. DCs carry HIV from laria and Leishmania.238,241,242 Importantly, these responses
peripheral tissues into draining lymph nodes where the are broad, often generating immunity against multiple epit-
virus is transmitted to CD4 T cells. HIV binds to DCs by opes243 in mice of several MHC haplotypes and the responses
CCR4, CXCR5, or DC-SIGN, but productive infection is elicit protection from mucosal infections.244 Other poten-
restricted by SAMHD1, a protein encoded by an Aicardi- tial DC targets include LOX-1/OLR1, MMR/CD206, DCIR/
Goutières syndrome susceptibility gene.225 Transmission CLEC4A, DC-SIGN/CD209, DNGR1, langerin (CD207), and
to CD4 T cells is dependent on DC-SIGN, a C-type lectin CD40.28,245
pathogen-recognition receptor expressed on the surface of In conclusion, DCs play important roles in a number of
DCs that retains the attached virus in an infectious state. different diseases. Moreover, they are excellent targets in
In the lymphoid organs, close interaction between DCs and designing new approaches to prevention and treatment of
CD4 + T cells facilitate HIV transmission.226,227 these diseases.
Interestingly, DC-SIGN also serves as receptor for sev-
eral other viruses including hepatitis C virus, Ebola virus,
cytomegalovirus, dengue virus, and the severe acute respi- ACKNOWLEDGMENTS
ratory syndrome coronavirus.228–230 Dengue is a mosquito- We thank Drs. Heidi Schreiber and Jakob Loschko for their
borne flavivirus that causes a disease that can be associated comments on the manuscript.

CHAPTER
17 Natural Killer Cells
Wayne M. Yokoyama
INTRODUCTION NK cells do not express the TCR on the cell surface, do not
Natural killer (NK) cells were initially described because produce mature transcripts for TCR chains, and do not re-
they spontaneously kill certain tumor targets.1–4 However, arrange TCR genes.12,13 Mice with the scid mutation or defi-
they are now recognized to play important roles in early in- ciencies in Rag1 or Rag2 lack TCR gene rearrangements and
nate immune responses, especially to viral infections. They mature T cells but possess NK cells with apparently normal
interact with other innate immune components and modu- function.14–17 Several CD3 components may be found in the
late the subsequent adaptive immune response. These effects cytoplasm of NK cells, particularly immature NK cells, but
are due to NK cell responses to proinflammatory cytokines they are not displayed on the cell surface18 with the exception
or susceptible targets, which stimulate NK cells to secrete of CD3ζ. But CD3ζ is expressed in association with Fc γRIII
other cytokines and/or kill targets. In this chapter, we will (CD16) and other NK cell activation receptors instead of the
consider all of these issues in detail by describing how they TCR/CD3 complex.19,20 Whereas mice lacking CD3ζ lack
differ from other lymphocytes, their functions, unique most T cells, NK cell number and function are minimally
target recognition strategies, tolerance, development, and affected.21 On the other hand, NK cells are completely ab-
role in immune responses and human diseases. There will sent in mice with only partial defects in T-cell subsets, such
be an emphasis on their target recognition receptors because as in mice lacking components of the IL-15R (see following
their discovery made it possible to understand NK cell biol- discussion). NK cells do not require the presence of major
ogy more precisely. histocompatibility complex (MHC) class I (MHC-I) mol-
ecules on their targets for lysis in an important functional
distinction with CD8 + MHC-I–restricted T cells. Instead,
GENERAL DESCRIPTION NK cells kill more efficiently when their targets lack MHC-I
Developmental studies have provided strong evidence that expression. Thus, NK cells can be clearly distinguished from
NK cells belong to the lymphocyte lineage (discussed in de- T cells, even from so-called CD3 + NKT cells that express
tail in the following). Morphologically, NK cells are typi- NK cell markers (see following discussion).
cally large lymphocytes containing azurophilic granules.5
However, the large granular lymphocyte morphology is not
Natural Killer Cells, Lymphokine-Activated Killer
invariably associated with NK cells because small, agranular
Cells, and Interleukin-15
lymphocytes may display natural killing,6 activated cyto-
toxic T-lymphocytes (CTLs) can display this morphology,7 Another area of overlap between NK and T cells concerns
and human large granular lymphocyte leukemias contain cytokine responses. When mouse splenocytes or human
NK- and T-cell variants.8 Among lymphocytes, NK cells peripheral mononuclear cells are exposed to high concen-
more closely resemble T cells than B cells. Thus, it is use- trations of interleukin (IL)-2 (800 to 1000 U/mL), robust
ful to compare and contrast these two lymphocyte popula- lymphocyte proliferation ensues (ie, LAK cells are gener-
tions as well as consider another enigmatic cell termed the ated).22–25 Although most are CD3 − NK cells, TCR/CD3 + T
“lymphokine-activated killer” (LAK) cell. cells are also produced. To distinguish NK cells within this
population, they are sometimes called “CD3 – LAK” cells or
“IL-2– activated NK cells.”
Natural Killer Cells versus T Cells In a related phenomenon, NK cells are activated when
NK cells are most often confused with T cells because they mice are injected with polyinosinic-polycytidylic acid
may have similar morphologies, express several cell surface (poly-lic) or other agents that trigger through Toll-like
molecules in common,9,10 and share functional capabilities. receptors (TLRs), often on plasmacytoid dendritic cells
While this confusion was frequent before the molecular (DCs).26 NK cells can also be activated in vitro with inter-
description of the T-cell receptor (TCR)/cluster of differen- feron (IFN) α / β, IFNγ, or low concentrations of IL-2 that are
tiation (CD)3 complex, their similarities remain a potential insufficient to induce proliferation.3,4,27
source of uncertainty. However, mature NK cells are clearly NK cells activated in these various ways, with or with-
not T cells by several criteria.11 Conventional NK cells do out proliferation, display enhanced killing of typical NK-
not require a thymus for development and are normal in sensitive targets. They also kill a broader panel of targets,
athymic nude mice (though this is not the case for the newly including those that are generally resistant to freshly isolated
described “thymic” NK cell subset, discussed subsequently). NK cells, such as the murine P815 mastocytoma cells and
395

freshly explanted tumors. Many agents that enhance killing IL-15 does not bind to IL-2Rα but instead utilizes a unique
by NK cells may also activate T cells, such that even T-cell IL-15Rα chain to form a high-affinity complex with IL-
clones may display promiscuous killing of targets that is no 2Rβγ.51,52 The IL-15Rα chain does not directly signal. Its dis-
longer MHC-restricted.28 This phenomenon was a frequent tribution is widespread on numerous cell and tissue lineages
source of confusion between NK cells and T cells during the including NK cells.
initial characterization of both cell types. IL-15 has a number of effects on NK cell biology. It is re-
Why cytokine activation leads to enhanced killing is quired for NK cell development; mice lacking IL-15 or any
incompletely understood. Interestingly, mouse NK cells component of the trimeric IL-15R complex lack NK cells.44–
46,53,54
constitutively express messenger ribonucleic acid (mRNA) Not surprisingly, mice deficient in other components of
for cytotoxicity components, perforin, and granzymes, but the IL-15R complex and its signaling pathway (IL-2Rβ, Jak3,
no protein.29 Cytokine activation enhances expression of and STAT5α / β) exhibit similar defects in NK cell develop-
mRNA for perforin and granzymes,30 and translation into ment.44,55–57 Depending on its relative concentration, IL-15
expressed proteins that contributes to more lytic capac- has an antiapoptotic or proproliferative effect.58,59 When NK
ity,29,31 but this effect may not explain the capacity of LAK cells are transferred to NK cell– deficient mice, they undergo
cells to kill a broader panel of targets. Activated NK cells “homeostatic” proliferation,60,61 akin to T-cell homeostatic
express additional receptors that may deliver stimulatory proliferation.62 Like memory CD8 + T-cell homeostasis, NK-
signals,32,33 but their contribution to the LAK phenomenon cell homeostasis is IL-15 – dependent,60,61 to a more or less
remains unclear. In mice, cytokine activation of NK cells degree.63 Finally, LAK cells can be generated with IL-15.64
results in expression of an alternatively spliced form of a re- These studies strongly suggest that LAK cells are generated
ceptor termed NKG2D (see subsequent discussion), which because high-dose IL-2 acts through the IL-2Rβγ that is nor-
may play a role,34,35 and IL-15 contributes to LAK-like activ- mally expressed with IL-15Rα as components of the consti-
ity of CTLs, although how much this applies to conventional tutively expressed trimeric IL-15R complex on resting NK
NK cells is not understood.36 Enhanced killing may also be cells. Thus, the LAK cell phenomenon is related to the role
due to effects on adhesion molecules.37,38 Nonetheless, it of IL-15 and its receptor in NK cell biology.
remains unclear how each of these factors contribute to the Interestingly, IL-15 is expressed at very low levels and is
LAK phenomenon of enhanced and broader killing capacity difficult to detect in vivo.65 The IL-15Rα chain can pres-
as compared to resting NK cells. ent IL-15 in trans to NK cells that can respond through IL-
It seems unlikely that high concentrations of IL-2 can 2/15Rβγ alone.66 For example, IL-15Rα – deficient NK cells
be achieved, even locally, to stimulate NK cells in vivo. develop in bone marrow (BM) chimeric mice in which
Furthermore, NK cells tend to be early responders in im- IL-15Rα – deficient BM was used to reconstitute IL-15Rα –
mune responses whereas the prime reservoir of large sufficient animals,67 indicating that IL-15Rα on another
amounts of IL-2 is the activated T cell that produces it some- cell can allow IL-15Rα – deficient NK-cell development. In
what later. Although naïve T cells can make IL-2 soon after certain inflammatory situations in vivo, DC presentation of
stimulation39,40 and DCs can produce IL-2 to enhance NK IL-15 in trans can enhance NK cell responses (also known
lytic activity,41 whether the resultant IL-2 concentrations are as priming).31,68 In DCs expressing both IL-15 and IL-15Ra,
sufficient to generate LAK cells in vivo is unclear. the IL-15Ra chain provides a chaperone function to stabi-
Interestingly, NK cells are apparently normal in mice lize receptor-cytokine complexes on the cell surface.69 Taken
with a targeted mutation in the IL-2 gene or the IL-2Rα together, trans presentation of IL-15 may be physiologically
chain,42,43 indicating that IL-2 itself is not required for nor- important to NK cell function.
mal NK cell development. Paradoxically, NK cells are defi-
cient in mice with a mutation in either IL-2Rβ or IL-2Rγ.44–46
The discrepancy in NK cell dependence on IL-2 versus IL-2 SELECTIVE NATURAL KILLER CELL SURFACE
receptor (IL-2R), as well as the LAK cell phenomenon, may MARKERS
be best understood by comparing the components of the The constitutive expression on NK cells of IL-15R complex
IL-2 and IL-15 receptors. with IL-2Rβ has practical usefulness because anti-IL-2Rβ
In brief, the high affinity IL-2R is a heterotrimeric (CD122) is sometimes used to identify naïve CD3 − NK cells
receptor complex comprised of α (p55), β (p75), and γ (p64) or deplete them in mice,70 but anti-CD122 is less useful dur-
chains.47 Though individual components may bind IL-2 ing an ongoing immune response and CD122 is expressed on
with low affinity, only the intermediate-affinity βγ receptor regulatory T cells. Anti-IL-15Rα antibodies have not been
(Kd ∼1 nM) and the high-affinity αβγ receptor (Kd ∼10 pM) widely used. Other markers have proven to be more useful
are capable of signaling. Resting NK cells constitutively ex- for analysis of NK cells.
press IL-2Rβγ47,48 and upon activation, may induce IL-2Rα In the mouse, the NK1.1 molecule is an especially impor-
and further upregulate IL-2Rγ chain expression.47 In con- tant marker on NK cells in C57BL strains.11 NK1.1 is an ac-
trast, resting T cells generally do not express any functional tivation receptor encoded by Nkrp1c (Klrb1c),71 a member of
IL-2 receptors, and most naïve T cells do not respond to high the Nkrp1 gene family (see following discussion). In FACS
concentrations of IL-2.49 The IL-2Rγ chain is also termed the sorting experiments, the NK1.1+ fraction contained all of the
common γ subunit (γc) because it is a required component natural killing activity in the spleen.72 In vivo administration
of the multimeric receptor complexes for other cytokines, of the anti-NK1.1 mAb PK13673 completely abrogated natu-
including IL-15,50 that is particularly relevant to NK cells. ral killing but did not affect adaptive immune responses74

CHAPTER 17 NATURAL KILLER CELLS | 397
(mAb PK136 is available from the American Type Culture The glycolipid determinant asialo-GM1 is expressed
Collection [ATCC], Manassas, VA [HB-191] and is an IgG2a by most if not all murine NK cells and a subpopulation of
isotype [ATCC, and data not shown], not IgG2b as originally T cells.99–101 Although the functional significance of this
described73). While mAb PK136 is very efficient at NK-cell molecule is unknown, polyclonal rabbit anti-asialo-GM1
depletion and is widely used for this purpose, unfortunately (Wako Chemicals USA, Richmond, VA) has been used to ef-
it recognizes an epitope on NK1.1 that is confined to fectively deplete NK cells. In more recent studies, the anti-
C57BL/6, C57BL/10, and a few other strains.73 Moreover, in NK1.1 mAb PK136 has become the reagent of choice for
Swiss, NIH and SJL/J mice, mAb PK136 recognizes another NK-cell depletion because of an available defined mAb and
NKRP1 family member, NKRP1B.75,76 However, there are its more restricted reactivity with NK cells.11,72–74 However,
now available NK1.1+ congenic strains, such as BALB.B6- anti-asialo-GM1 remains in use for NK-cell depletion when
Cmv1r (catalogued as C.B6-Klra8Cmv1-r /UwaJ, stock number anti-NK1.1 cannot be employed.73
002936 at The Jackson Laboratory, Bar Harbor, ME) in which In addition to NKG2D and NKp46, human NK cells se-
the C57BL/6 allele of NK1.1 has been genetically bred onto lectively express CD56. Although it is also found on neural
the BALB/c background that otherwise lacks the NK1.1 epi- tissues and some tumors, CD56 is generally not expressed by
tope.77 Similarly, the NK1.1 allele has been introgressed onto other hematopoietic cells or lymphocytes.102–104 This 140 kDa
the nonobese diabetic (NOD) background.78 molecule is derived from alternative splicing of the gene en-
A subpopulation of T cells expresses NK1.1, described in coding neural cell adhesion molecule (NCAM) involved in
detail in another chapter 18. These “natural killer T (NKT) nervous system development and cell-cell interactions.105,106
cells” express the TCR/CD3 complex and typically are re- CD56 may be involved in adhesion between NK cells and their
stricted by the nonclassical MHC-I molecule, CD1, which targets,107 but this function is controversial. Curiously, mouse
presents glycolipid antigens to NKT cells. NKT cells respond CD56 is not expressed on hematopoietic cells,108 indicating
early during the course of an immune response and may po- that its role on NK cells is not conserved. Nevertheless, CD56
tently activate conventional NK cells.79 Nonetheless, NKT is particularly useful as a pan-NK– cell marker in humans.
cells can be distinguished from conventional NK cells by Human NK cells can be functionally divided according
expression of the CD3 complex (ie, conventional NK cells to the level of CD56 expressed.103,109 Most human periph-
are NK1.1+ CD3 −). eral blood NK cells are CD56dim, a phenotype associated
The NKG2D (Klrk1) activation receptor is expressed on with more cytotoxicity and less cytokine production than a
all NK cells in human and all strains of mice examined.80 In smaller subset of NK cells that express CD56 at higher lev-
humans, NKG2D is also expressed on all γδTCR+ and CD8 + els (CD56bright). These cells also tend to differentially express
T cells, whereas in mice, NKG2D is expressed on most NKT receptors involved in target recognition as well as CD16. The
and γδTCR+ T cells but not on resting CD8 + T cells.81–83 CD56bright cells may undergo a maturation process to be-
However, essentially all activated mouse CD8 + T cells express come CD56dim cells110 and may be related to a subset of NK
NKG2D. In both humans and mice, CD4 + T cells do not ex- cells identified in mice, termed “thymic” NK cells.111
press NKG2D, but it is found on a subset CD4 + CD28 − T cells Other molecules selectively expressed on NK cells are
in patients with rheumatoid arthritis.84 Blocking anti-NKG2D better discussed below under the general topic of NK cell
mAbs and NKG2D-deficient mice have been described.80,85–87 receptors because they are molecularly defined and their
Regardless, conventional NK cells are NKG2D + CD3 −. ligands are known.
The NKp46 (Ncr1) activation receptor appears to be ex-
pressed on all CD3 − NK cells in humans and all strains of
mice. However, recent reports indicate expression of NKp46 A MOLECULAR DEFINITION OF
on immune cells in the gut that may be developmentally NATURAL KILLER CELLS?
distinct from conventional NK cells.88–92 Moreover, deplet- A precise molecular definition of NK cells has been elusive.
ing anti-NKp46 mAbs have not been described, limiting its There are no known molecules that are exclusively expressed
usefulness for in vivo functional experiments. Nonetheless, on NK cells and are responsible for critical functions only
recently developed mice may allow other approaches, such displayed by NK cells. The NK cell is therefore still defined
as a mouse where a green fluorescent protein (GFP) cassette by function to the exclusion of other cells, a concept first
was inserted into Nkp46 and two different transgenic (Tg) articulated 25 years ago.11
mice with a Nkp46 promoter contruct for expression of Cre The defining functional feature of NK cells remains their
or diptheria toxin receptor.93–95 intrinsic ability to perform natural killing (ie, they spontane-
The mAb DX5 recognizes a molecule that is coexpressed ously lyse certain tumor cells in a perforin-dependent man-
on most NK1.1+ CD3 − cells and on small populations of ner). Unlike other lymphocytes, NK cells do not express surface
splenocytes in NK1.1− strains, consistent with identification immunoglobulin or the TCR/CD3 complex, and generally do
of NK cells in all strains. However, mAb DX5 recognizes not require MHC-I expression on targets for lysis. Therefore,
the α2 integrin that is widely expressed on other leuko- a current working definition is that an NK cell is a sIg−, TCR/
cytes, not just NK cells,96,97 and its expression is regulated.98 CD3− lymphocyte that can mediate perforin-dependent natu-
Nevertheless, the DX5 epitope has been used to identify NK ral killing against targets that may lack MHC-I expression.
cells in mouse strains that do not express NK1.1, but it has It is noteworthy that T cells were historically defi ned by
been largely supplanted by other nonpolymorphic markers an awkward functional definition (thymus-derived, sIg−
such as NKp46. lymphocytes responsible for cell mediated immunity).112

With the molecular definition of the TCR and coexpressed in which enlarged lysosomes are observed apparently due
CD3 molecules, immunologists can now define a T cell as to decreased lysosome fission.123 Normal granules con-
a cell expressing the TCR/CD3 complex.113 The availabil- tain perforin and granzymes (granule enzymes). Perforin,
ity of molecular probes and mAbs directed against this a pore-forming protein, is rendered inactive by associa-
complex provides precise definition even in pathologic tion with calreticulin and serglycin, and is activated by a
tissue sections, without the need for functional analysis (cell cysteine protease.124 Granzymes are first produced as in-
mediated immunity, thymus dependence). Similarly, a mo- active proenzymes that are activated by N-terminal cleav-
lecular definition should permit unequivocal identification age by dipeptidyl peptidase I, also known as cathepsin C.
of NK cells to define their role in normal immune responses However, granzymes are rendered inactive by the acidic pH
and pathologic settings. of the granules. Upon activation by a sensitive target, NK
Presumably, such a definition will require further knowl- (and T) cells are triggered to rapidly polarize the granules
edge of the molecular basis for NK-cell function, such as the and reposition the microtubule organizing center toward
receptors involved in natural killing. On the other hand, one the target in a dynein-dependent manner.125 The granule
difficulty is that the function that is most attributed to NK membrane ultimately fuses with the plasma membrane, and
cells, natural killing, can be displayed by other cells, such as externalizes, releasing granule contents. Calcium-dependent
cytokine-treated T cells. Moreover, NK cells can utilize more polymerization of perforin results in “perforation” of the
than one receptor for target recognition, and individual NK target cell plasma membrane, and granzyme entry by an as
cells can simultaneously express several of these receptors. yet incompletely understood process. A recent study sug-
Thus, there is, as yet, no consensus on the elusive “NK-cell gests that perforin induces a plasma membrane repair pro-
receptor” analogous to the TCR, and the general sentiment cess that results in endocytosis of perforin and granzymes
in the field is that there is unlikely to be such a receptor. into enlarged endosomes, called “gigantosomes.”126 Perforin
In the absence of a precise definition, most investigators pores in the gigantosome membrane then allow delivery of
currently consider the following phenotypes to be surrogate granzymes that mediate cleavage of caspases and Bid, ulti-
markers of bona fide NK cells. Mouse NK cells are typically mately leading to target cell apoptosis.127
NK1.1+ (in appropriate strains), Fc γRIII + (CD16), CD122 +, Recently, many details of the granule exocytosis pathway
and CD3 −. Human NK cells are generally CD56 + and CD3 −. have come from studies of the heterogeneous human disor-
In general, mouse and human NK cells also express NKG2D der, hemophagocytotic lymphohistiocytosis (HLH).128,129 In
and NKp46, with caveats as elaborated previously. particular, genetic studies of heritable HLH, termed familial
Note that these markers are generally correlated with cells HLH (FHL), led to identification of the first described FHL
having natural killing capacity but the markers themselves mutation in the perforin gene (PRF1), responsible for FHL2.
are not required for natural killing. It should be emphasized Studies of patients with FHL without PRF1 mutations led
that these phenotypes can lead to some confusion due to ex- to discovery of other genes whose products (MUNC13-4,
pression of other molecules on NK cells that are used to help syntaxin 11, MUNC18-2) affect granule exocytosis by cy-
define other immune cells. For example, NK cells express totoxic lymphocytes. Fusion of the cytolytic granule with
B220 (CD45R) that is often used as a B-cell– specific marker; the plasma membrane requires vesicular RAB27a, a mem-
CD19 is more reliable to distinguish B cells from NK cells.114 ber of the small GTPase superfamily. Defects in RAB27a
Similarly, NK cells express CD11b, first described as Mac-1 are associated with the human disorder Griscelli syndrome,
on macrophages.98 Moreover, NK cells express CD11c, a type 2. Mice have been described with defects in granule
marker used to define certain DC populations, leading to exocytosis components including LYST (beige), perforin
publications describing a novel type of DC, termed killer (Pfn1−/ −), Unc13d (equivalent to MUNC13-4, also known
DCs.115,116 However, detailed investigation suggests that these as Jinx), and Rab27a (ashen). As highlighted by the names of
cells are developmentally unrelated to DCs and are actually the mutant mice, many mutations of molecules in the gran-
activated NK cells.117–119 Therefore, markers associated with ule exocytosis pathway are associated with skin pigment
NK-cell function have been extremely useful in shaping our changes because these molecules also affect melanosomes in
current concepts of NK cell biology and elaborate their ef- melanocytes.130
fector functions, but caution may be necessary to avoid con- Human T and NK cells also express another pore-form-
fusion with other immune cells. ing molecule, granulysin, that is related to a family of sapo-
sin-like proteins.131 Based on crystallographic studies, these
molecules appear to be active against bacteria, fungi, and
EFFECTOR FUNCTIONS OF NATURAL KILLER CELLS tumor cells by charge association with target membranes
Cytotoxicity and subsequent disruption, leading to target cell lysis.132
A hallmark of NK-cell effector function is target killing, me- Granulysin is contained in cytolytic granules containing the
diated by a process termed granule exocytosis that can be other cytolytic proteins, such as granzymes.133 In a perforin-
initiated by exposure to susceptible targets or cross-linking dependent manner, granulysin causes target apoptosis but is
of specific activation receptors. Like CTLs, NK cells possess not expressed in mouse cytotoxic lymphocytes.134
preformed cytoplasmic granules that resemble secretory ly- NK-cell cytotoxcity can be demonstrated in several re-
sosomes with properties of both secretory granules and lyso- lated ways. Natural killing refers to the process by which
somes.120 Granule formation is affected by Lyst, the molecule NK cells kill certain tumor targets without need for prior
defective in humans with Chediak-Higashi syndrome121,122 host sensitization with the target. Natural killing was first

assessed with a simple in vitro assay for target membrane In addition to natural killing, cytotoxicity by NK cells
integrity that is still used today, the standard 51Cr-release can be triggered by deliberate cross-linking of activation
assay.135 The prototypical NK-sensitive tumor target for receptors (discussed in greater detail in the “Activation
mouse NK cells is YAC-1 (TIB-160 from ATCC), a thy- Receptors” section). Plant lectins can also trigger target
moma derived from Moloney virus –infected A strain mice, killing.159 In general for all stimuli, cytotoxicity occurs via
whereas the standard human target is K-562 (CCL-243 from granule exocytosis and can be measured with the same
ATCC), an erythroleukemic cell line derived from a human assays for natural killing.
patient with chronic myelogenous leukemia in blast crisis.136
Maximal killing by enriched, IL-2–activated NK cells usu- Cytokine Production
ally occurs with effector:target (E:T) ratios of <10:1 whereas
When exposed to NK-sensitive targets or cross-linking of
unfractionated, freshly isolated peripheral blood or spleno-
receptors, NK cells also produce cytokines, including IFNγ,
cyte preparations usually require E:T ratios of >100:1. Even
TNFα, and granulocyte-macrophage colony stimulating fac-
at high E:T ratios, not all targets are killed, with percentage-
tor (GM-CSF).160–162 They can also be similarly triggered to
specific cytotoxicity typically ranging from ∼10% with fresh
produce chemokines, such as RANTES, lymphotactin, MIP-
NK cells to ∼80% with activated NK cells. Note that per-
1α, and MIP-1β.163 Moreover, NK cells produce cytokines
forin-dependent leakage of 51Cr from the targets is mostly
in response to other cytokines. For example, in response to
complete within about an hour; 4-hour assays are standard.
IL-12, NK cells produce IFNγ.164 Similarly, NK cells respond
Longer periods may reflect other apoptotic processes, such
to type I IFNs (IFNα/β) produced by DCs stimulated by in
as Fas-induced apoptosis.
vivo administration of poly-I:C and other ligands for TLR
While 51Cr release is still the gold standard, there are also
and nucleic acid sensors.165 Cytokine-stimulated responses
numerous nonradioactive tests for perforin-dependent kill-
may obscure detection of specific activation by activation re-
ing, including release of intracellular enzymes or use of flu-
ceptors in vivo.163,166
orochromes for target labeling.137–139 The release of granule
While cytokine production can be indirectly measured
components, including granzymes, into the supernatant
with RT-PCR for mRNA, it should be noted that resting NK
can be determined by conversion of an appropriate sub-
cells typically already express abundant levels of cytokine
strate, such as granzyme A-mediated cleavage of alpha-N-
mRNA even though the proteins are not synthesized,167 as
benzyloxy-carbonyl-L-lysinethiobenzyl ester (also known as
described previously for granule components in mouse NK
BLT-esterase activity).140
cells.29 Enzyme-linked immunosorbent assays (ELISA) of
A particularly useful new flow cytometric assay exploits
tissue culture supernatants are often used, but recent studies
the orientation of lysosomal-associated membrane pro-
have utilized intracellular staining of cytokines, such as IFN,
tein-1 (LAMP-1, CD107a) on the luminal side of cytotoxic
for analysis of individual NK-cell responses that may be more
granules in unactivated NK cells. During granule exocyto-
informative, akin to use of the CD107a degranulation assay.168
sis, the granule fuses with the plasma membrane, resulting
In immune responses, NK-cell production of cytokines
in externalization of the granule membrane and exposing
should occur relatively early and may thereby influence the
CD107a on the external surface of the plasma membrane as
subsequent adaptive immune response. Moreover, their re-
an indicator of NK-cell activation.141–143 By contrast to other
sponses to cytokines are regulated by complex interacting
methods, the CD107a assay provides the opportunity for
pathways.169 A fuller description of NK-cell cytokine re-
measuring NK-cell responses at the single cell level, isolating
sponses and production is provided in the following sections
triggered NK cells,144 and possibly simultaneously assessing
on NK cell responses during infections and interactions
other NK-cell functions.
with DCs.
Activated NK cells and CTLs also induce perforin-
independent target cell killing by expressing Fas ligand
(tumor necrosis factor [TNF] superfamily 6) that binds Fas NATURAL KILLER CELL RECOGNITION OF TARGETS
(TNF receptor superfamily 6, TNFRSF6) on the target, trig- Molecular dissection of NK-cell recognition of their targets
gering apoptosis.145–150 Similarly, other TNF superfamily opened new frontiers in NK-cell biology because it not only
members, such as TNF-related apoptosis-inducing ligand explained target recognition but it led to identification of
(TRAIL TNFSF10), can be involved in related processes.151 receptors that are selectively expressed on NK cells. In addi-
However, mice deficient in TNF family members or their tion to providing molecular tools for detailed studies of NK
receptors may manifest significant alterations in lymphoid cell function, this analysis yielded several surprises. In con-
organogenesis and splenic architecture, and NK cell num- trast to CTL recognition: 1) NK cell receptors are germline
ber and function,152–154 such that the relative contributions of encoded and are not strictly “clonotypic” as defined in terms
these pathways to NK-cell function are incompletely under- of clonotypic TCRs (unique receptor only expressed by the
stood. Moreover, NK cells from mice deficient in perforin, rare effector clone and its progeny); 2) individual NK cells
granzymes, or molecules involved in granule formation or express both inhibitory and activation receptors for target
exocytosis demonstrate profound defects in natural killing in recognition, and often simultaneously express several dif-
vitro.155–158 Similar defects have been found with NK cells de- ferent receptors of each type; 3) the receptors are often pro-
rived from patients with deficiencies in granule exocytosis.128 miscuous and may have overlapping ligand specificities; and
Thus, the available data strongly suggest that granule exocy- 4) NK-cell receptors specifically bind MHC-I molecules but
tosis is the predominant mechanism for natural killing. they are functionally and structurally distinct from other

receptors that bind MHC-I (ie, TCR and CD8). In the fol- of individual human NK cell clones were not easily assign-
lowing sections, we will discuss NK-cell receptors involved able to specific MHC-I alleles.175 Thus, the MHC-I effect on
in target recognition by first considering the relationship be- natural killing was controversial for some time.
tween target susceptibility to natural killing and expression Several groups, however, observed that MHC-I–express-
of MHC-I. ing parental targets were resistant to natural killing, whereas
mutants selected for absence of MHC-I expression became
susceptible. The parental (resistant) phenotype could be re-
Target Cell Major Histocompatibility Complex-I and stored by reconstitution of MHC-I expression by transfected
Natural Killer Cells: The “Missing-Self” Hypothesis expression of molecules to correct the defect, be it β2-micro-
Whereas initial studies suggested that natural killing was globulin (β2m)176 or transporter associated with processing
“non-MHC-restricted,”11 substantial progress in under- (TAP).177,178 Studies utilizing mice with a targeted mutation
standing NK-cell recognition began with ascertaining in the β2m gene added substantial support to the MHC-I
the role of MHC-I molecules in natural killing (Fig. 17.1). protective effect, as normal expression of MHC-I heavy
Kärre and colleagues discovered that MHC-I–deficient tu- chains requires β2m.179,180 β2m-deficient lymphoblasts were
mors remained susceptible to in vivo rejection, apparently susceptible to lysis by normal NK cells. Moreover, β2m − / −
by NK cells.170 Conversely, target cell expression of MHC-I BM transplanted into otherwise syngeneic normal hosts was
molecules appeared to have a protective effect against NK- rejected by recipient NK cells.180,181 These results resembled
cell–mediated lysis in vitro. A number of methods, such as hybrid resistance whereby NK cells in irradiated F1 hybrid
IFNγ treatment, to upregulate MHC-I correlated with target mice reject parental BM transplants.182 Hybrid resistance
protection but other effects could not be excluded.171 There is regulated by parental determinants that are genetically
was significant variability in capacity of specific MHC-I linked to the MHC-I region, H-2D.183 Thus, in several dis-
molecules to protect targets172,173 ; in vitro culture conditions tinct NK-cell recognition systems, the target cell expression
could influence NK-cell specificities,174 and the specificities of certain MHC-I molecules correlated with resistance to
FIG. 17.1. Major Histocompatibility Complex (MHC)-I Expression on Targets is Inversely Related to Natural Killing. Targets expressing
MHC-I are more resistant to lysis by natural killer (NK) cells than targets lacking MHC-I expression. This is the exact opposite of the require-
ments for MHC-I–restricted cyotoxic T-lymphocytes that recognize foreign peptides presented by MHC-I. As depicted, T cells can recognize
virus-infected cells, but some viruses may evade T cells by downregulating MHC-I. These infected cells then become more susceptible to NK
cells, which generally tend not to discriminate between self- and viral-peptides, though there are some peptide contributions to NK recogni-
tion as described in the text.

natural killing whereas absence of MHC-I was associated surveillance.187–189 In either case, however, the MHC-I–defi-
with susceptibility to NK cells. cient cells should become more susceptible to natural killing.
NK cells, therefore, have a different relationship to tar- The host, therefore, is endowed with two components (T and
get cell MHC-I molecules than MHC-I–restricted CTLs (see NK cells) with opposing requirements for self–MHC-I ex-
Fig. 17.1). Strictly speaking, NK cell lysis is “non-MHC– pression. This fail-safe system should eliminate pathologic
restricted,”11 at least as far as MHC restriction is precisely de- processes that might otherwise evade immune responses
fined for T cells having a requirement for specific self-MHC by any alteration of MHC-I expression (either increased to
molecules presenting a given peptide antigen.184 However, avoid NK cells or decreased to avoid T cells). The missing-self
the term “non-MHC–restricted” (and its synonyms) is now hypothesis thus provided a tentative physiologic explanation
somewhat outdated because it implies, when viewed in a for MHC-I–associated resistance, creating a framework for
broader sense, that MHC plays no role in NK-cell cytotoxic- initial attempts to define NK-cell recognition of their targets.
ity. Avoidance of these terms will minimize confusion con-
cerning the relationship of target cell MHC-I molecules with
NK-cell specificity. Current Principles of Target Recognition by Natural
As initially observed and discussed by Kärre in the Killer Cells
“missing-self” hypothesis, the relationship between tar- The MHC-I–associated resistance to natural killing inspired
get expression of MHC-I and resistance to natural killing a panoply of models and their variants to explain not only
highlights a fundamental distinction between NK and T resistance but also natural killing.172 The target interference
cells185 (see Fig. 17.1). Whereas T cells are triggered by de- or masking model predicted that a single NK-cell receptor
tection of “foreign” epitopes, Kärre proposed that NK cells activates natural killing when it engages its putative target
are equipped to detect the absence of “self” epitopes. The cell ligand.185 MHC-I molecules mask the putative target
[missing-self] hypothesis suggests that NK cells survey tis- cell ligand and block its recognition by the NK-cell recep-
sues for expression of MHC-I molecules that are normally tor. This hypothesis was initially favored because it was the
ubiquitously expressed and that somehow prevent NK-cell simplest.190 Moreover, it made the most sense if one consid-
activity. If MHC-I molecules are downregulated or mutated, ered that NK cells should have only one defined receptor
NK cells can then lyse the target. The generally opposite re- analogous to the TCR. The effector inhibition or inhibitory
quirements of NK and T cells for target cell MHC-I expres- receptor model suggested that NK cells are inhibited from
sion may be physiologically important. Several pathogens, natural killing by an NK-cell receptor that binds MHC-I on
including herpes viruses, possess mechanisms that prevent the target and delivers negative signals overriding a default
the normal expression of MHC-I molecules on infected pathway of activation.185
cells, providing means to avoid MHC-I–restricted T cells.186 Although the target interference model has not been
Moreover, tumorigenesis is frequently associated with altera- refuted, it is now known that NK cells express inhibitory
tions in MHC molecules, either mutation in structural genes receptors that physically bind MHC-I in a manner that
or decreased expression, again leading to escape from T-cell may be influenced by MHC-bound peptides191 (Fig. 17.2).
FIG. 17.2. Current Principles of Target Recognition by Natural Killer (NK) Cells. Pictured are several scenarios of interactions between
ligands on targets (top row) and receptors on NK cells (bottom row). Successful activation of NK cells is shown by the dashed upward line.
A: Targets expressing majoe histcompatibility complex (MHC)-I are resistant to lysis by NK cells because of MHC-I–specific inhibitory recep-
tors. B: The absence of MHC-I (or lack of receptors specific for target MHC-I, not depicted) does not automatically result in target killing. C:
Activation receptor engagement is required to trigger target killing in absence of MHC-I. D: In the situation where both inhibitory and activa-
tion receptors are engaged, the inhibitory receptor effect often dominates and no killing occurs. E: In the induced-self model, induced expres-
sion of NKG2D ligands can overcome the inhibitory influence of MHC-I, resulting in NK cell activation. F: Normal epithelium masks ligands
for NK-cell activation receptors at the tight junctions. G: Under pathologic situations, the epithelial architecture may be disrupted, leading to
ligand exposure. (Not depicted are inhibitory ligands at the epithelial tight junctions that may inhibit NK cells when they transmigrate through
epithelial barriers.) NK cells can also recognize pathogen encoded ligands on infected cells (not shown, but similar to C or E).

All known inhibitory receptors contain cytoplasmic immu- against cells expressing these ligands is limited by inhibi-
noreceptor tyrosine-based inhibitory motifs (ITIMs) con- tory receptors (Fig. 17.2D). Another group of ligands is
sisting of V/I/L/SxYxxL/V (single amino acid code where x characterized by their relatively low expression on normal
is any amino acid).192 Ligand engagement leads to phosphor- tissues and induced expression under “stress” conditions
ylation of the ITIM, which leads to inhibition, traditionally (Fig. 17.2E). Other ligands become exposed when tissue
thought to be due to recruitment and activation of intracel- architecture is altered (Fig. 17.2F,G). Because the ligands
lular phosphatases, although recent evidence suggests more are encoded in the normal host genome, they would be
complexity. recognized by the NK cell as indicators of pathologic con-
MHC-I inhibitory receptors on NK cells have either of ditions, either as “induced-self ” or “exposed-self,” respec-
two general structures193 : 1) C-type lectin-like receptors that tively. Another group of ligands is found on infected cells
are disulfide-linked dimers with type II transmembrane and is encoded by the pathogen (not shown but similar to
topology (extracellular carboxyl termini). These receptors Fig. 17.2C or E).
are encoded in the NK gene complex (NKC) and were first Finally, many other NK-cell receptors have been discov-
described in mice. 2) Immunoglobulin (Ig)-superfamily re- ered that do not fall neatly into the categories described
ceptors that have type I transmembrane orientation. These here. Some appear to have similar inhibitory function
molecules are encoded in a different genetic region, termed as the MHC-specific inhibitory receptors but bind non-
the leukocyte receptor complex (LRC), and were first de- MHC ligands, strongly suggesting MHC-independent self-
scribed in humans. Although ongoing studies indicate that recognition. The function of these and other receptors
both structural types of receptors are expressed on mouse remains under intense investigation.
and human NK cells, the lectin-like receptors (Ly49 recep-
tors) are the major MHC-specific inhibitory receptors in NATURAL KILLER CELL RECEPTORS
mouse whereas the Ig-like receptors (killer Ig–like receptors
[KIRs]) predominate in human. In the following sections, we will describe the major
The absence of MHC-I does not always result in killing, receptors on NK cells in detail by fi rst discussing the MHC-
indicating that release from inhibition does not result in ac- specific inhibitory receptors that helped elucidate NK rec-
tivation by default (Fig. 17.2B). Instead, it was suggested that ognition paradigms before delving into MHC-independent
NK cells express two functionally different receptors for tar- inhibitory receptors, activation receptors, and other recep-
get cell ligands.194,195 In this two receptor model, one receptor tors found on NK cells. Given the large number of recep-
triggers activation upon ligand binding (Fig. 17.2C) whereas tors now identified (Table 17.1), this section will primarily
the MHC-I–specific receptor inhibits activation by negative discuss work on the receptors that have been studied most
signaling. In many circumstances, the inhibitory receptor extensively. This summary will illustrate the experimen-
effect dominates over the activation receptor (Fig. 17.2D), tal approaches that led to identification of these receptors,
but the outcome usually reflects the integration of signals their features, and outline general principles applicable for
from both types of receptors, which can be affected by li- study of other receptors that will not be discussed in detail
gand expression or affinities (not shown). due to space constraints. Nonetheless, description of the
Many, but not all, NK-cell activation receptors are en- major activation receptors and their ligands help illustrate
coded in the NKC and LRC, having similar structural and provide molecular handles on the various functions of
properties as their inhibitory receptor counterparts except NK cells.
for absence of cytoplasmic ITIMs. The activation receptors
typically do not have signaling motifs in their cytoplasmic Inhibitory Natural Killer-Cell Receptors Specific for
domains but contain charged transmembrane residues that Major Histocompatibility Complex–I Molecules
facilitate association with reciprocally charged residues in The mouse and human MHC-specific inhibitory receptors
the transmembrane domains of signaling chains having im- are remarkably different in protein structure. Each will be
munoreceptor tyrosine-based activation motifs (ITAMs) discussed separately.
analogous to ITAMs in TCR and B-cell receptor (BCR)
complexes (D/ExxYxxL/Ix6–8YxxL/I). NK cells express three Mouse Ly49
ITAM-containing signaling chains: CD3ζ, Fc εRIγ, and The Ly49A receptor was the first inhibitory MHC-I–
DAP12 (DNAX associated protein of 12 kDa, also known as specific NK-cell receptor to be described in molecular
killer activating receptor–associated protein, KARAP; Ly83; terms.191,196 Ly49A was originally identified as a molecule
tyrosine kinase binding protein, Tyrobp). NK cells also ex- of unknown function on a T-cell tumor.197,198 It is a disul-
press DAP10 (hematopoietic cell signal transducer, Hcst) fide-linked homodimer (44 kDa subunits) with type II
that lacks ITAMs and instead contains a motif for recruit- membrane orientation, and C-type lectin superfamily ho-
ment of phosphatidylinositol 3-kinase (PI3K) and Grb2. mology.199,200 Previously termed Ly49, it is now appreciated
The signaling chains typically provide two major functions: that Ly49A (Klra1) belongs to a family of highly related
facilitate cell surface expression of the associated activation molecules.195,201–203 Indeed, genetic analysis revealed that the
receptor, and transduce signals. genes for Ly49A and NK1.1 are linked in the NKC (Fig. 17.3),
To date, the ligands for activation receptors fall into leading to studies indicating that Ly49A is constitutively ex-
several major groups. One group is encoded by the host pressed on a distinct subpopulation (20%) of NK cells in
and is expressed normally. Presumably, NK-cell attack C57BL/6 mice.203

TABLE 17.1 The Panoply of Receptors Expressed by Natural Killer Cellsa

Inhibitory (I) or
Receptor H M Activation (A) Other Names Ligand
Ly49A X I Ly49, Ly-49, Klra1 H2Dd, Dk, Dp, alleles in H2r,
and H2q
Ly49C X I Klra3 H2Kb, numerous
Ly49E X I Klra5 Urokinase plasminogen
activator
Ly49G2 X I LGL-1, Klra7 H2Dd
Ly49I129 X I Klra9 m157
Ly49Q X I Klra17 H2Kb
KIR2DL1 X I CD158a, p58.1, EB6 HLA-C2 (HLA-Cw2, Cw4,
-Cw5, -Cw6)
KIR2DL2/KIR2DL3 X I CD158b, p58.2, GL183 HLA-C1 (HLA-Cw1, -Cw3,
-Cw7, -Cw8)
KIR2DL4 X I CD158d HLA-G
KIR3DL1 X I CD185e1, NKB1, p70, NKAT3 Bw4 (HLA-A and B)
KIR3DL2 X I CD158k, p140, NKAT4 HLA-A3, -A11
Lilrb4 X I gp49 αvβ3 integrin
CD94/NKG2A X I Kp43 HLA-E
X I Qa-1
LILRB1 X I CD85j, ILT2, LIR1 Folded HLA, UL18
LILRB2 X I ILT4, LIR2 Folded, free HLA
LAIR-1 X I Collagen
Siglec-7 X I P75, AIRM1 Carbohydrates
Siglec-10 X I Carbohydrates?
Siglec-E X I ?
PILRa X I CD99
PILRb X I CD99
FcγRIII X X A CD16 Fc of IgG
Ly49D X A Chinese hamster MHC-I, H2Dd
Ly49H X A m157
Ly49PMA/My X A M04 + H2Dk
KIR2DS1 X A CD158h, p50.1 HLA-Cw7
KIR2DS2 X A CD158j, NKAT5, p50.2,
clone 49
KIR2DS3 X A NKAT7
KIR2DS4 X A CD158i, NKAT8, clone 39 HLA-Cw4
KIR2DS5 X A CD158g, NKAT9
KIR3DS1 X A CD158e2
CD94/NKG2C X A HLA-E
CD94/NKG2C X A Qa-1
NKG2D X A, costimulation KLRK1 MICA, MICB, ULBP/RAET1
NKG2D X A, costimulation Klrk1 H60, RAE1, MULT1
Nkrp1c X A NK1.1 ?
Nkrp1b(d) X I Clrb
Nkrp1f X A? Clrg
NKRP1A X I LLT1
NKp80 X A AICL
NKp65 X A KLRF2 CLEC2A
2B4 X X A,I SLAMF4 CD48
CD2 X X A CD48
NTBA X X A SLAMF6, Ly108 NTBA
Ly9 X X A SLAMF3 Ly9
CD84 X X A SLAMF5 CD84
CRACC X X A SLAMF7 CRACC
NKp46 X X A Hemagglutinin
NKp44 X A ?
(continued)

TABLE 17.1 The Panoply of Receptors Expressed by Natural Killer Cellsa (Cont.)
Inhibitory (I) or
Receptor H M Activation (A) Other Names Ligand
NKp30 X A B7-H6
CD69 X X A ?
Ly6 X A ?
Gp42 Rat A ?
Klrg1 X I Caherins
CEACAM1 X I CD66a CEA
CD226 X X A DNAM-1 necl-5 (CD155, PVR), nectin-2 (CD112, PVRL2)
CD96 X A Tactile necl-5
CRTAM X A necl-2
TIGIT X I PVR, PVRL2
AICL, activation-induced C-type lectin; AIRM1, adhesion inhibitory receptor 1; CRTAM, class I-restricted T-cell–associated molecule; LGL, large granular lymphocyte; NK, natural
killer; TIGIT, T-cell immunoglobulin and immunoreceptor tyrosine-based inhibitory motif domain.
a
These are the major receptors discovered on NK cells in humans and mice, listed in order of appearance in the text.
Multiple lines of evidence indicate that Ly49A is an of H-2Dd reversed resistance in killing experiments (per-
inhibitory receptor specific for MHC-I, particularly H2Dd : mitted lysis) and blocked the cell binding assay.191,204,206,207
1) Functional analysis: the Ly49A+ NK-cell subset were 4) In vivo expression: the apparent level of Ly49A expressed
equivalent to Ly49A− NK cells in killing several targets, but per NK cell was downregulated in MHC congenic and Tg
they could not lyse a large panel of targets that were read- mice expressing H-2Dd.208–210 This was not due to nega-
ily lysed by Ly49A− NK cells. This phenotype was related tive selection because the percentage of Ly49A + NK cells
to MHC-I expression of certain H2 haplotypes on target was unchanged. 5) Gene transfer: primary NK cells and T
cells.191,204,205 Transfected expression of H-2Dd selectively cells expressing a Ly49A transgene and a Ly49A-transfected
rendered a susceptible target resistant to natural killing NK cell line were specifically inhibited by H-2Dd.211,212 6)
by Ly49A+ NK cells. Moreover, killing through disparate Inhibition by Ly49A is ITIM-dependent, based on gene
stimuli by Ly49A+ NK cells was also inhibited. 2) Cell transfer of mutant Ly49A molecules.212 7) H2Dd tetra-
binding: Ly49A+ tumor cells bound specifically to immo- mers bind Ly49A transfectants.213 8) Ly49A tetramers bind
bilized MHC-I molecules206 and to H-2Dd-transfectants.207 H2Dd on transfected cells.214,215 9) Biophysical studies:
3) Antibody blocking: F(ab ′)2 fragments of mAb directed recombinant Ly49A binds recombinant H2Dd in surface
against either Ly49A or the α1/ α 2 (but not the α 3 domain) plasmon resonance (SPR) studies with K D = ∼2.0 μ M.216
NK Gene Complex (NKC)
Leukocyte Receptor Complex (LRC)
FIG. 17.3. The Genomic Organization of the Natural Killer (NK) Gene Complex and Leukocyte Receptor Complex in Humans and Mice. The
figures are not drawn to scale with precise gene locations being modified as new sequence information becomes available (genome.ucsc.
edu/ and www.ebi.ac.uk/ipd/kir/). The grey shading is coordinated to represent related genes. Question marks indicate genes whose precise
location is not known. An “X” indicates genes not homologous to other aligned genes. Double slashes represent large genomic distances.
Note that most but not all genes are expressed on NK cells. Many remain orphan genes because the functions of their gene products have
not been determined. Modified from Kelley et al.312

10) Crystallography: the structure of Ly49A complexed carbohydrates could affect the interaction, such as affi ni-
with H2Dd was determined.217 11) Less extensive studies ties or kinetic parameters, but this has not been studied
also indicate that Ly49A recognizes H-2Dk, H2Dp, and al- in depth. These studies also provide a structural explana-
leles in H2r and H2q.191,208,213,218,219 Therefore, Ly49A is an tion for species-specific β2m requirements as revealed by
MHC-I– specific receptor for H-2Dd, H-2Dk, and H2Dp, functional studies.226 Thus, Ly49A recognizes site 2 in the
and alleles in H2r and H2q. MHC molecule in terms of trans recognition between the
The nature of the Ly49A interaction with MHC-I, NK-cell receptor and target cell MHC-I molecule.
however, is fundamentally different from TCR/MHC-I Most but not all other Ly49 receptors are MHC-
interactions because the former appears to be relatively I−specific inhibitory receptors. First noted by Southern
independent of the specific peptide bound by H2Dd.220,221 blot analysis and cDNA cloning, genome sequence analysis
However, bound peptides are required for appropriately revealed 15 complete Ly49 genes in C57BL/6 mice.199–202,227
folded MHC-I molecules that can be recognized. Despite There is evidence for alternative splicing and alterna-
its structural homology to C-type lectins that are carbo- tive transcriptional start sites for the Ly49 genes, though
hydrate-binding proteins, 222,223 Ly49A does not have the their importance has not been elucidated.202,228,229 Notably,
residues for coordinate binding to calcium that is required Ly49C has broad specificity for H2 alleles, as revealed by
for lectin binding. Moreover, Ly49A binding to its MHC tetramer staining, and is the only known inhibitory NK-
ligands is not carbohydrate-dependent based on func- cell receptor specific for an H2b haplotype allele (H2K b)
tional, SPR, and crystallographic analyses.216,224 In the in C57BL/6 mice, notwithstanding unconfi rmed reports
crystallographic structure of Ly49A complexed to H2Dd that Ly49I also binds H2K b.168,213,230 Ly49C is recognized
(2.3Å resolution), the lectin-like structure of Ly49A was by two mAbs: 5E6, which also binds Ly49I, and 4LO3311,
confi rmed 217 (Fig. 17.4). Two interaction sites were seen which has exquisite specificity for Ly49C.231–233 X-ray crys-
between the lectin-like domain of Ly49A and H2Dd : site tallographic studies have indicated that Ly49C binds H2K b
1 involved the “left” side of the peptide-binding cleft of in a manner similar to Ly49A interaction with H2Dd (see
H2Dd and a wedge-like site 2 involved the undersurface of Fig. 17.4).234,235 Only a site 2 interaction was seen with the
the peptide-binding cleft. The residues in Ly49A involved contact residues, showing a similar but distinct topology
in binding either ligand site are overlapping. Mutational to the Ly49A-H2Dd interaction. Interestingly, however,
analysis revealed that Ly49A binds site 2 where it con- peptides bound to H2K b clearly affect functional interac-
tacts α1, α 2, and α 3 of H2Dd and β2m.214,216,225 This site tions with Ly49C236 and affi nities as measured by SPR.234
is near Asn80, an Asn-linked glycosylation site conserved However, Ly49C does not directly engage the peptide, in-
in all MHC-I molecules, leaving open the issue of whether dicating long-range effects. Finally, both Ly49A and Ly49C
A B
FIG. 17.4. Crystal Structures of Natural Killer (NK)-Cell Receptors in Complex with Their Ligands. A: Mouse Ly49C bound to H2Kb at site 2
(PDB ID = 3C8K).193,235 Ly49A interaction with site 2 of H2Dd is very similar.170 Site 1 of Ly49A-H2Dd interaction is approximately located where
the H2Kb label is placed. B: Human LILRB1 (LIR1, ILT2) bound to human leukocyte antigen-A2 (PDB ID = 1P7Q).266 Figures were produced
using the University of California, San Francisco Chimera package from the Resource for Biocomputing, Visualization, and Informatics at the
University of California, San Francisco (www.cgl.ucsf.edu/chimera).1030 The structures are viewed from the side with the NK cell positioned at
the top of the figure and the target cell surface at the bottom. The major histocompatibility complex molecules are oriented similarly.

appear to undergo conformational changes upon ligand direction. Directionality and monoallelic expression may
binding, and a Ly49 dimer can engage two MHC-I mol- also be controlled by DNA methylation.251 A “probabilistic”
ecules.235,237 Thus, Ly49 receptors bind their MHC ligands model has been proposed to explain these fi ndings that may
in a structurally related manner. also explain the stochastic expression of Ly49 genes and
Recent studies suggest that Ly49 molecules also can bind their stable expression. However, recent studies of Ly49 in-
MHC in a cis interaction between receptor and ligand on dicate highly variable transcriptional start sites, suggesting
the NK cell itself.168,238 For example, an MHC ligand for that the probabilistic model may not be correct.229 Other
Ly49A or Ly49C on the same cell prevents binding of MHC data indicate that TCF-1 but not LEF-1 in the T-cell factor/
tetramer.168,238 If the cells are briefly exposed to mild acidic lymphoid enhancer family of DNA-binding proteins affects
conditions, MHC-I expression is lost (due to disruption of some but not all Ly49 receptor expression.252–254 Thus, the
the noncovalently linked MHC-I heterotrimer), and Ly49A elements controlling Ly49 gene expression are incompletely
binding to cognate MHC-tetramers is restored. This cis understood.
interaction is dependent on site 2 residues in the MHC mol- Analysis of the Ly49 receptors thus far is largely based
ecule. These findings may help explain the observation that on examination of the C57BL/6 alleles, but the Ly49 re-
the presence of self-MHC ligands leads to downregulation of ceptors display extensive polymorphism. The Ly49 fam-
Ly49 expression, as previously noted on primary NK cells in ily is encoded in the NKC located on mouse chromosome
MHC congenic mice.208–210 Cis interactions may also explain 6 with the syntenic human region being chromosome
functional differences in NK cells in MHC-congenic mice or 12p13.2195,203,255,256 (see Fig. 17.3). While the NKC also con-
NK cells that do or do not express MHC ligands in Tg mice tains genes for other lectin-like receptors, the Ly49 genes
that are mosaic for MHC expression.239–241 Finally, there are are clustered with the exception of Ly49b. Corresponding to
biophysical data supporting a role for the relatively long, restriction fragment length polymorphic (RFLP) variants
flexible stalk region of Ly49 receptors in allowing either originally detected with the Ly49A cDNA,203 there is sig-
trans or cis interactions.242 At the moment, the physiologic nificant allelic polymorphism of the Ly49 cluster between
importance of cis interactions is incompletely understood inbred mouse strains with differences in gene number as
but may be relevant to NK-cell tolerance and education, as well as alleles for the Ly49 genes.227,257–259 In contrast to
discussed below. C57BL/6J mice, genomic sequence analysis shows 8 puta-
Although they have not been studied as extensively as tive Ly49 genes in BALB/c mice, 19 in 129 mice (of which
Ly49A and Ly49C, other inhibitory Ly49 receptors and at least 9 appear to be pseudogenes) and 22 in NOD mice.
their ligands have been identified202,230 and characterized Array-based comparative genomic hybridization analysis
with other MHC allele specificities, as detailed in a re- of 21 mouse strains compared to the reference C57BL/6J
cent review.243 Moreover, they are structurally related.235,244 strain indicated that these mice could be grouped into five
Individual NK cells may express multiple Ly49 receptors clusters that correspond to or are predictive of restriction
simultaneously,210,233,245 often (but not always) two or more, fragment length polymorphic patterns on Southern blot
suggesting that individual NK cells may be inhibited by analysis.203,260 There are also multiple alleles for individual
more than one MHC-I molecule. Ly49 family members.227,249,257,258,261 Thus, there is significant
Ontogenetic studies demonstrate that the total repertoire polymorphism of the Ly49 molecules at both the haplotype
of Ly49 expression does not reach adult levels until sometime (gene numbers) and individual gene (alleles) levels, not un-
after 3 weeks of age, concomitant with attainment of full expected because the Ly49 molecules bind highly polymor-
NK-cell cytolytic activity.246 Thereafter, the expression of phic MHC-I molecules.
Ly49 receptors is generally thought to be fi xed and stable on The MHC-I specificities have generally been well char-
an individual NK cell. Ly49E is expressed only on fetal NK acterized for only a few Ly49 alleles. Interestingly, mAbs
cells, but NK cells in mice deficient in Ly49E are otherwise specific for one Ly49 allele may bind another molecule
normal.247 In adult mice, developmental studies indicate with a different function or specificity in another mouse
that Ly49 receptors are first expressed on immature NK cells strain,262–264 similar to what was recognized for mAb reactiv-
in the BM, before a phase of constitutive proliferation98 that ity with different MHC alleles.265 Thus, the polymorphisms
appears to be modestly affected by MHC haplotype.168 There also raise practical issues when studying Ly49 molecules in
are only modest effects on the fi nal “repertoire” of MHC- different mouse strains.
specific receptors expressed by splenic NK cells in different Finally, it should be noted that Ly49 receptors may be
MHC environments.210,248 expressed by other cells; some are selectively expressed on
The expression of inhibitory Ly49 receptors appears to non-NK cells and some may have specificities for non-MHC
occur in a stochastic manner. There is evidence for mono- ligands. NKT and other T-cell subsets may express Ly49
allelic expression of Ly49 receptors (expression from one receptors but they have not been thoroughly studied.266,267
chromosome), initially described as “allelic exclusion,” a Ly49B and Ly49Q are not expressed on NK cells; rather,
term that has fallen out of favor because it has a specific they are expressed on myeloid cells.268,269 Interestingly,
meaning and mechanism for TCRs and BCRs.249 At least Ly49Q recognizes H2K b and positively regulates TLR sig-
some Ly49 genes possess bidirectional, overlapping pro- naling.270,271 On the other hand, Ly49E appears to recognize
moters directed in opposite orientations.250 Transcription urokinase plasminogen activator, though physical binding
factors driving transcription in one direction prevent bind- has not been established.272 Interestingly, Ly49B, Ly49E, and
ing of other factors driving transcription in the opposite Ly49Q are predicted to be distinct in fi ne structure from the

known MHC-I– specific Ly49 receptors on NK cells.235 Thus, informative HLA-B alleles showed that this specificity was
while Ly49 receptors are predominantly NK-cell inhibitory conferred by a region in the α1 domain overlapping the
receptors for MHC class I, they may have other roles that are area on HLA-C recognized by p58 molecules.282 Finally,
less well understood; still other Ly49 receptors have activa- HLA-A3, -A11– specific receptors have similar properties
tion function, as discussed subsequently. to p58 and NKB1 except that they appear to be disulfide-
linked dimers termed p140, 283 whereas others have found
Human Killer Immunoglobulin-like Receptors that a monomeric HLA-A3 – specific receptor resembles
In contrast to mouse NK cells, human NK cells can be NKB1.284 Thus, representative alleles of all classical HLA
cloned by limiting dilution in the presence of irradiated class I loci are capable of inhibiting NK cells through p58/
feeder cells, phytohemagglutinin, and IL-2, leading to NKB1/p140 receptors although HLA-B and -C alleles dom-
establishment of short-term NK-cell clones that have dif- inate human NK-cell specificities, and it is not yet known if
ferences in target killing and surface molecules. mAbs were there are receptors reactive with each HLA allele.
isolated that reacted specifically with these clones; reactiv- When the cDNAs for the p58 and NKB1 molecules were
ity correlated with the capacity of the clones to kill certain cloned, they were surprisingly found to encode type I in-
tumors, and the mAbs affected cytotoxicity. This general tegral membrane proteins with Ig-like domains285–287 unlike
approach led to the identification of the human NK-cell the lectin-like Ly49 family of type II receptors. The Ig-like
receptors. receptors are now collectively known as killer Iq-like recep-
A series of studies273–275 showed that the mAbs GL183 tors (KIRs) or CD158.288,289 The KIR nomenclature is based
and EB6 identify serologically distinct 55 kDa or 58 kDa on whether the receptor has two or three Ig-like external do-
molecules, initially termed p58. These molecules had sev- mains (KIR2D or KIR3D, respectively), and possession of a
eral features: 1) selective expression on overlapping NK cell long (L) or short (S) cytoplasmic domain. In general, the L
subsets; 2) expression on NK cell clones correlated with ex- forms are inhibitory because they contain ITIMs, whereas
pression of certain human leukocyte antigen (HLA) class the S forms appear to be activation receptors (see following
I alleles on resistant targets; 3) a target susceptible to a discussion). Each distinct receptor is also designated by a
given NK-cell clone bearing p58 molecules reactive with number. The KIR2DL1 (CD158a, p58.1) molecule bears the
either mAb was made resistant by transfection of cDNAs original EB6 epitope and is specific for HLA-C (Lys80, speci-
encoding certain HLA-C molecules; 4) the otherwise re- ficity 2), whereas KIR2DL2 (CD158b1, p58.2) and KIR2DL3
sistant, HLA-C−transfected targets could be lysed in the (CD158b2, p58) have the GL183 epitope and are specific for
presence of the appropriate anti-p58 mAbs. The mAb ef- HLA-C (Asn80, specificity 1). (As detailed in the follow-
fect occurred with F(ab′) 2 fragments, suggesting that the ing, structural analysis supports grouping of HLA-C alleles
interaction between p58 and an HLA class I molecule on into two mutually exclusive groups, HLA-C1 and HLA-C2,
the target cell inhibits the NK cell. Thus, the p58 mole- based on direct interaction of KIR2DL2/3 and KIR2DL1,
cules displayed features consistent with a role as inhibitory respectively, with residue 80 of HLA-C, validating original
human NK-cell receptors specific for MHC-I, analogous to functional groupings but simplifying HLA-C groupings to
the mouse Ly49A receptor that was being studied in paral- just residue 80.276,277) KIR2DL4 (CD158d, p49) reportedly
lel, as described previously. binds HLA-G290 but displays both inhibitory and activation
Other studies noted that NK-cell specificity was skewed functions.291–293 KIR3DL1 (CD158e1, NKB1, NKAT3) is spe-
when the NK cells were grown in the presence of cells bear- cific for HLA-A and HLA-B molecules with the Bw4 epit-
ing allo-MHC determinants.276,277 This specificity correlated ope.282 KIR3DL2 (CD158k, p140, NKAT4) has HLA-A3 and
with reactivities that mapped to paired residues at position HLA-A11 specificity.283,284
77 and 80 in the .α1 domain of HLA-C. All known HLA-C There is unequivocal evidence that the KIR2DL and
molecules could be divided into two groups, one with KIR3DL molecules are inhibitory HLA class I–specific recep-
Asn77-Lys80 (HLA-Cw2, -Cw4, -Cw5, -Cw6) and the other tors. In addition to the data with NK cell clones and mAbs
with Ser77-Asn80 (HLA-Cw1, -Cw3, -Cw7, -Cw8). Indeed, mentioned previously, the following have been described:
transfection analysis showed that p58 specificity for HLA-C 1) KIR bind directly to HLA class I: soluble KIR2DL-Fc fu-
molecules was related to expression of the EB6 epitope for sion proteins bind cells expressing the appropriate transfected
the former (specificity 1, now termed HLA-C1), whereas the HLA class I alleles.294,295 In addition, a soluble KIR2DL mol-
latter was related to the GL183 epitope (specificity 2, HLA- ecule containing only the extracellular domain binds specifi-
C2) on the NK-cell clones.276,278,279 Thus, human NK-cell re- cally to its HLA-C ligand in solution.296 2) Gene transfer of
ceptors showed promiscuous specificity that was dependent KIR: KIR2DL specificity and inhibitory function were trans-
on residues 77 and 80 in HLA-C. ferred when KIR2DL cDNAs were transiently expressed with
The NKB1 (p70) molecule was serologically similar vaccinia constructs in human NK-cell clones.294 Similarly, Tg
to p58 molecules with regard to subset expression, and expression of KIR2DL2 in mice conferred inhibition of rejec-
correlation of expression on NK-cell clones to specificity tion of BM expressing Tg HLA-Cw3.297 3) SPR measurements
for HLA class I.280,281 In contrast to p58 molecules, how- indicate that the KIRs bind their HLA ligands with Kd = ∼10
ever, NKB1 had a distinct Mr (70 kDa) and specificity for μM.298–301 Binding is affected by peptide bound by HLA mol-
HLA-B. The NKB1+ clones were specifically inhibited by ecule.300,301 Through histidine-rich domains, the KIRs bind
targets expressing transfected HLA-Bw4 molecules, and Zn ++, which affects KIR multimerization and binding ki-
the anti-NKB1 mAb reversed the inhibition. Analysis of netics to HLA ligands.302,303 4) Crystallographic studies dem-

onstrate KIR2DL1 (2.8 Å resolution) and KIR2DL2 (3.0 Å Ly49s bind their MHC ligands in markedly different ways,
resolution) interactions with their cognate HLA ligands.299,304 despite their analogous functions as MHC-specific inhibi-
Thus, KIR molecules are clearly MHC-I–specific inhibitory tory receptors.
receptors on NK cells. Also unlike the Ly49s, the KIRs are encoded in the LRC
Interestingly, structural studies indicate that KIR mol- on human chromosome 19q13.4 that encodes many other
ecules bind HLA class I molecules in a manner analogous Ig-like receptors (see Fig. 17.3)311; the KIR genes are clus-
to recognition of MHC by TCRs (Fig. 17.5). In particular, tered toward the telomeric end of the LRC.312 Interestingly,
both KIR2DL1 and KIR2DL2 use surface loops near their the mouse LRC on chromosome 7qA1 does not include
interdomain hinge regions to bind their cognate HLA-C genes for KIR-like molecules that instead are encoded on
ligands (Cw4 and Cw3, respectively) with a footprint over- the X chromosome.313,314 On the other hand, like the Ly49s,
lying the “right” side of the peptide-binding cleft (when the KIRs display remarkable polymorphism with at least 11
viewed from the “top” in standard depictions of MHC-I genes315–317 (see www.ebi.ac.uk/ipd/kir/index.html for up-
molecules).299,304 The receptors bind both α1 and α 2 he- dated database).
lices with interactions between KIR2DL1 and Lys80 of The human KIR locus also demonstrates consider-
HLA-Cw4 and between KIR2DL2 and Asn80 of HLA-Cw3. able haplotype diversity with at least 27 different haplo-
These interactions with residue 80 of the HLA-C mole- types,315,316,318 recently defined at the sequence level.317 While
cules were lost when the reciprocal residues were swapped, there has not been a consensus defi nition, two major types
accounting for the previously described HLA-C groupings of haplotypes have been described.289 Group B contains one
and KIR specificities in functional studies276,277 and muta- or more of the following: KIR2DL5, KIR2DS1, KIR2DS2,
tional analysis indicating that residue 80 is more signifi- KIR2DS3, KIR2DS5, and KIR3DS1, whereas group A hap-
cant for KIR interaction than residue 77.305–308 Although lotypes have none of these genes. Reflecting extensive allelic
neither KIR2DL molecule has extensive contacts with pep- polymorphism of individual genes, a large number of
tides bound to HLA-C, KIR2DL interactions with HLA-C different KIR genotypes have been described, and they are
imposes physical constraints on the p8 position of the pep- distributed differently in the various ethnic populations.
tide. This may account for observed peptide preferences As with the Ly49s, only a few KIR alleles have been well
in functional studies and antagonism of certain peptides characterized with respect to HLA class I specificities.
on KIR inhibitory function.309,310 A recent structure (1.8 Nonetheless, these genetic variants have not only provided
Å resolution) of KIR3DL1 complexed to HLA-B*5701 re- clues to new receptors and ligand specificities but also valu-
vealed a similar recognition strategy.301 Thus, KIRs and able links to the role of NK cells and their receptors in
A B C
FIG. 17.5. Additional Structures of Natural Killer–Cell Receptors in Complex with Their Ligands. A: Human KIR2DL2 with HLA-Cw3 (PDB ID =
1EFX).236 Two KIR molecules are apparent in the crystal structure with only one molecule (killer Ig–like receptor [KIR] A) contacting the human
leukocyte antigen (HLA) molecule. In this view, KIR B molecule is not shown. B: Human NKG2D with MICA (PDB ID = 1HYR).412 C: Human
CD94/NKG2A with HLA-E (PDB ID=3CDG).382 The figures were produced and oriented as described in Figure 17.4.

disease pathogenesis (discussed in the subsequent clinical 7) a single NK cell simultaneously expresses one or more
section), and more broadly, human evolution. of either type of inhibitory receptor (each may be
functional);
Convergent Evolution of Major Histocompatibility 8) once they are expressed, their expression appears to
Complex–Specific Natural Killer-Cell Receptors be stable;
Despite the controversy surrounding the initial cloning of 9) both are germ-line encoded by small families of genes
mouse Ly49s and human KIRs and leukocyte Ig-like recep- that are clustered in the genome;
tors (LILRs), additional data have provided new interpreta- 10) both display impressive polymorphism, in terms of
tions of the distinctly different receptors used by mouse and gene number and alleles for each gene;
human NK cells, respectively, to recognize MHC-I. Detailed 11) both are related to molecules that lack ITIMs and
genome sequence information is available on the NKC and instead are activation receptors; and
LRC in mice, humans, and other species. In the mouse, 12) both are involved in NK-cell education by self-MHC-
MHC-specific NK-cell receptors with Ig-like domains have I (see following discussion).
not yet been described, although there is conservation of
several genes in the broader LRC on mouse chromosome Thus, the mouse Ly49 receptors and human KIRs are
7312 (see Fig. 17.3). Activated mouse NK cells do express analogous receptors in an apparently striking example of
gp49b (Lilrb4), an Ig-like inhibitory receptor also expressed convergent evolution,331 whereby each species came up with
on other leukocytes, including mast cells.319–322 However, it a different genetic solution to provide extremely important
is not expressed by resting NK cells and is not specific for functions for species reproduction and survival.
MHC-I. Instead, it binds the integrin αvβ3 and appears more In other species, Ly49 and KIR genes have been analyzed
important for responses of other cells, such as neutrophil, eo- primarily with respect to sequence and gene number.332 For
sinophil, and DCs.323–328 Mouse Kir3dl1 is on the X chromo- example, the LY49L gene in baboons appears to be func-
some (see Fig. 17.3) and expressed in NK and T cells, but its tional but the putative polypeptide lacks an ITIM.333 In
function and ligand remain unknown.313,314 The Ly49 locus rats, the Ly49 cluster appears to have markedly expanded
in humans consists only of LY49L (KLRA1) that is a pseu- with at least 25 genes, 334,335 demonstrating one of the most
dogene because of a point mutation that gives rise to a splic- rapid rates of gene expansion.336 Dog, cat, and pig appear
ing abnormality.329 Thus, current data indicate that mouse to have only one Ly49, whereas horse represents the only
NK cells do not express functional KIR orthologues while known nonrodent mammal with several Ly49 genes.337
human NK cells do not express functional Ly49 orthologues. The chicken genome has several lectin-like receptor genes
One reason for this discrepancy may be that the corre- that are genetically linked to the MHC.338–340 On the other
sponding orthologue is present in the genome but has not hand, multiple KIR genes have been described in primates
been identified. Indeed, genomic sequencing has revealed a and cattle.341–344 Rhesus macaques have a profound plas-
multitude of candidate orphan receptors in the genome and ticity of KIRs with multiple genotypes and haplotypes345
specifically the LRC and NKC that have yet to be studied encoding receptors that bind MHC-I.346,347 In rat, a KIR-
carefully.311,330 In that regard, identification and functional like sequence has been reported, 314 and dogs and cats lack
analyses of NK cell receptors led to the recognition that they functional KIRs.348 Interestingly, pigs and marine carni-
are encoded in genomic regions containing gene clusters for vores each possess a single Ly49 and KIR gene, but it is
related receptors that are expressed on other leukocytes, not not clear if these are functional.348 Finally, several species
just NK cells. Dissection of the expression and function of have no readily identifiable Ly49 or KIR genes332 (eg, tele-
these receptors is a rich area of research that is beyond the ost fish instead possess a large number of novel immune-
scope of this chapter. type receptors with sequence homology to mammalian
The alternative and currently favored view for the dis- LRC-encoded receptors349). Perhaps these species have had
crepancy is that mice and humans independently evolved alternative convergent evolutionary stratetgies to preserve
analogous receptors to serve the same function. While both inhibitory MHC-I– specific receptors.
human and mouse NK cells express a conserved relatively Additional studies of the Ly49-like and KIR-like mol-
nonpolymorphic lectin-like receptor, CD94/NKG2, it does ecules as well as potential new orthologues in other species
not possess many of the features that are shared by mouse will be of interest to evolutionary biologists for several
Ly49 and human KIRs: reasons including prior description of NK-like cells in
1) both Ly49s and KIRs are constitutively and selectively lower vertebrates and MHC genes that are coevolving.332,350
expressed on naïve, unstimulated NK cells (with excep- Moreover, in primitive chordates, NK-like, missing-self–like
tions for rare populations of T cells); recognition affects histocompatiblity reactions351 but the
2) both bind MHC-I molecules with intermediate affinity molecular determinants of histocompatibility involves mol-
(KD = ∼2-10 μM); ecules unrelated to mammalian Ly49, KIR, or even MHC
3) binding to MHC is promiscuous; itself.352–354 Thus, evolutionary studies of NK-cell receptors
4) MHC-bound peptides have only a modest effect, if at and their ligands may provide unique insight into missing-
all, on recognition; self and other histocompatiblity reactions.
5) both use ITIMs to inhibit NK-cell activation; While we have focused thus far on mouse Ly49 and
6) they are expressed in a stochastic fashion on overlapping human KIR as the major MHC-I– specific receptors, there
subsets of NK cells; are other well-described NK-cell receptors belonging to

either the lectin-like receptor or Ig-like receptor super- other hand, this binding interface is similar to binding of
families. These receptors also have specificities for MHC-I another NKC-encoded, lectin-like receptor, the NKG2D
molecules. activation receptor, to its MHC-like ligands (see follow-
ing discussion). Thus, NKC-encoded, lectin-like receptors
Human and Mouse CD94/NKG2 surprisingly use different strategies to contact their li-
The analysis of CD94 (Klrd1) and NKG2 (Klrc, excluding gands, even though these ligands have structurally related
NKG2D [Klrk1]) family of molecules was especially chal- MHC-I folds.
lenging and required insightful investigations. Identified by Despite its conservation between mice and humans, the
subtractive hybridization, the human NKG2 molecules are role of CD94/NKG2 receptors in NK-cell function is still
type II integral membrane proteins with external C-type lec- incompletely understood. For example, viruses encode
tin domains355 encoded in the NKC356 (see Fig. 17.3). Initial peptides that bind and enhance expression of HLA-E, pro-
attempts to express NKG2 molecules on the cell surface were viding a CD94/NKG2A-dependent mechanism to avoid
thwarted. Meanwhile, mAb reactivity suggested that CD94 NK-cell attack.386,387 By contrast, human NK cells expressing
was variably expressed on human NK cells as a disulfide- CD94/NKG2C (an activation receptor) expand in response
linked dimer (70 kDa NR, 43 kDa R),357 and both activa- to cytomegalovirus (CMV)-infected targets.388 When a large
tion and inhibition functions for CD94 were described.358–361 number of human NK-cell clones were obtained from two
Surprisingly, cDNA cloning revealed that CD94 has a short normal individuals, CD94/NKG2 seemed to account for
seven amino acid cytoplasmic domain, suggesting that the majority of self-MHC – specific receptors on clones from
it cannot signal on its own.360 Furthermore, anti-CD94 one individual whereas KIRs dominated the self-specific
immunoprecipitates were not detectable from radiolabeled receptors on clones from the other individual, suggesting
CD94 transfectants, despite easily detectable expression that some individuals may depend on CD94/NKG2 for self-
on FACS analysis with the same mAbs.362,363 These appar- tolerance.389 Qa-1 and HLA-E can present peptides derived
ent discrepancies were resolved when it became clear that from other molecules, including the signal sequence of heat
CD94 heterodimerizes with NKG2 molecules362 ; NKG2A shock protein 60 (Hsp60) that is induced by a number of
is the 43 kDa molecule previously identified as Kp43 with stimuli,390,391 a multidrug resistance transporter,392 or blas-
anti-CD94 mAbs.364,365 While CD94 may be expressed as a tocyst MHC expressed in embryonic tissues.393 This may re-
homodimer, the NKG2 partner provides the signaling motif, sult in loss or gain of recognition by the inhibitory CD94/
whether activation or inhibition.364,366 (NKG2B is an alterna- NKG2A receptor, suggesting intrinsic mechanisms to per-
tively spliced form of NKG2A. The rest of the NKG2 family turb inhibition by CD94/NKG2A in certain circumstances.
is discussed in the following.) Yet, CD94 appears to be dispensable in certain strains of
The ligand specificity for CD94/NKG2 receptors was also mice, such as DBA/2J, that do not appear to have any untow-
initially thought to be promiscuous as interactions with many ard NK-cell phenotype.394 This finding has been recapitu-
classical (class Ia) and nonclassical (class Ib) HLA molecules lated in studies of a CD94 knockout mouse on the 129 strain
had been described.359,362,364,367–370 However, human CD94/ background.395 Interestingly, CD94 knockout mice are more
NKG2 receptors directly recognize HLA-E, a MHC-Ib mol- susceptible to ectromelia virus,396 as detailed below.
ecule homologous to mouse Qa-1.371–373 HLA-E (and Qa-1) CD94/NKG2 molecules may be important in T-cell
is widely expressed with limited polymorphism.374–376 While function. CD94/NKG2A is rapidly induced on antigen-
HLA-E heavy chain is expressed with β2m and a peptide specific CD8 + T cells during polyoma virus and other
occupying its peptide-binding cleft, its peptide repertoire is infections,397–399 and CD8 + T cells expressing CD94/NKG2A
largely derived from the leader sequences of MHC-Ia mol- preferentially proliferate during persistent infection, sug-
ecules, as previously noted for mouse Qa-1.377,378 HLA-E (or gesting that CD94/NKG2 receptors may play a role in
Qa-1) expression thus requires normal production of HLA-E memory T-cell responses,400 and that TCR specificity is cor-
(or Qa-1) and synthesis of certain MHC-Ia molecules. Mouse related with CD94/NKG2A expression by human CTL.401
CD94/NKG2 recognizes Qa-1 that shares many features Indeed, CD94/NKG2A inhibits antigen-specific cytotoxicity
with HLA-E.379,380 These findings need to be considered in in polyoma virus responses, although this effect is patho-
the context of the prevailing view at the time that mouse and gen-dependent. CD94/NKG2A has been studied with other
human NK cells use structurally different receptors to rec- viral infections, including herpes simplex virus,402 murine
ognize MHC-I molecules.381 Clearly, the CD94/NKG2 recep- CMV,403 gHV68, and influenza,404 for example. Although not
tors and their ligands are conserved in humans and mice. all studies revealed functional consequences, the recurring
The crystal structure of human CD94/NKG2A bound theme is the appearance of CD94/NKG2A on previously
to HLA-E was resolved to 2.5Å and 4.4Å resolution.382,383 activated CD8 + T cells. Thus, CD94/NKG2 receptors may
Remarkably, CD94/NKG2A interfaces with HLA-E in a regulate T-cell responses.
manner analogous to TCR recognition of peptide-loaded
MHC-I, 384 including TCR binding to HLA-E itself. 385 Both Other Human Immunoglobulin-like Receptors Specific
innate and adaptive receptors for HLA-E lay across the pep- for Major Histocompatibility Complex-I Molecules
tide-binding cleft, though peptide itself plays a relatively The human LILR family is encoded in the LRC (see
minor role in binding CD94/NKG2A.382,383 Strikingly, this Fig. 17.3), just centromeric to the KIR genes. There are two
binding is distinct from that of Ly49 receptors to their general forms of these receptors, subfamily A that appears to
classical MHC-I ligands (see previous discussion). On the be activation receptors, and subfamily B that has the ITIMs

characteristic of inhibitory receptors. The best character- be involved in inhibitory signaling. Although the in vivo con-
ized members, LILRB1 (also known as CD85j, Ig-like tran- text for functional interaction awaits further characteriza-
script 2 [ILT2], or leukocyte IG-like receptor 1 [LILR1]) is tion, it is reminiscent of the broader reactivity of the Siglecs.
broadly expressed whereas LILRB2 (CD85d, ILT4, or LIR2) The CD33-related sialic acid binding Ig-like lectins
is not expressed on NK cells but is expressed by myelomono- (CD33rSiglecs) are type I receptors with varying numbers of
cytic cells, including DCs and monocytes. Both recognize Ig-like domains expressed on a broad array of cells and en-
HLA class I molecules405,406 with LILRB1 exclusively bind- coded in the “extended” LRC312,424,425 (see Fig. 17.3). Despite
ing folded HLA class I molecules with β2m, whereas LILRB2 having sialic acid recognition in common, the Siglecs appear
can bind both folded and free HLA class I heavy chains.407 to show differences in carbohydrate recognition, depending
Interestingly, a human CMV protein, UL18, binds LILRB1 on the specific glycan context.424 Human NK cells express
with 1000-fold higher affinity than HLA molecules, impli- Siglec-7 (p75, adhesion inhibitory receptor 1 [AIRM1])
cating a role for LILRB1 in host defense.408 The ligands and and Siglec-10, whereas some mouse NK cells express a re-
functions of other LILRBs (LILRB3 [CD85a, ILT5, LIR3], lated Siglec-E.425 Siglec-7 has been most extensively studied.
LILRB4 [CD85k, ILT3, LIR5], LILRB5 [CD85c, LIR8], and As expected, its cytoplasmic ITIM can recruit SHP-1 and
LILRB6 [CD85b]) are as yet unknown but they may not be inhibit NK-cell functions.426,427 Moreover, expression of its
able to bind HLA due to structural constraints.409 ligand on targets inhibits NK cells in a Siglec-dependent
LILRB1 has four Ig-like domains and binds a conserved manner.428 However, the effects appear to be modulated by
region in the α3 domain of most, if not all, classical and cis interactions between the Siglec receptor and its carbohy-
nonclassical HLA class I molecules (HLA-A, -B, -C, -E, -F, drate ligands on the NK cell itself,424,428,429 reminiscent of cis
and -G).408 Interestingly, the crystal structure of LILRB1 interactions between Ly49 and MHC ligands, as previously
bound to HLA-A2 (3.4Å resolution) reveals that it binds discussed.
MHC molecules under the peptide-binding domain where The Ig-like receptors, termed paired Ig-like recep-
it contacts α3 and β2m, more akin to Ly49 engagement of tor (PILR)α and β, are not encoded in the LRC, rather on
MHC than KIR409 (see Figs. 17.4 and 17.5). Even though human chromosome 7 (mouse chromosome 5).430,431 Mouse
LILRB1 and LILRB2 have differing capacities to bind free PILRα is an ITIM-containing inhibitory receptor, whereas
HLA molecules, LILRB2 has an HLA class I binding site that PILRβ is an activation receptor that couples to DAP12.430,432
overlaps with but is distinct from that for LILRB1.410 LILRB1 Both receptors are expressed on NK (and other immune)
binds UL18 in a manner structurally similar to HLA class I cells and recognize CD99. Interestingly, sialylated O-linked
binding.411 glycans on CD99 are involved in recognition by PILRs.433
Interestingly, LILR genes demonstrate allelic polymor- Thus, NK cells (and other leukocytes) express multiple
phisms,412 though less so than the adjacent KIR cluster.413 inhibitory receptors that are capable of MHC-independent
Nonetheless, polymorphisms in LILRB1 may affect receptor recognition. How they participate in NK-cell responses and
expression.414 Moreover, polymorphisms in LILR genes are contribute to MHC-dependent effects are beginning to be
associated with certain autoimmune diseases, such as rheu- elucidated, and some appear to play a specific role in the
matoid arthritis.415 context of activation receptors, as discussed in the following.
Major Histocompatibility Complex–Independent Natural Killer-Cell Activation Receptors

Natural Killer-Cell Inhibitory Receptors NK cells clearly kill MHC-I– deficient targets more ef-
As already mentioned, NK cells also express inhibitory re- ficiently than MHC-I– sufficient targets. However, this
ceptors for non-MHC ligands, such as mouse gp49b, which enhanced killing does not occur simply because a nonspe-
binds the αvβ3 integrin,323 and there is a growing list of cific default pathway is unleashed when MHC-I is absent.
other such molecules.416 Some of these receptors will be dis- Instead, it is clear that susceptible targets express ligands
cussed in a more appropriate context in the following sec- for NK-cell activation receptors. In general, these receptors
tions. Most of these receptors contain cytoplasmic ITIMs and their ligands were defined following description of the
so their inhibitory function can be predicted even if not di- inhibitory receptors.
rectly tested, though some caution is required because the
motifs may be involved in other signaling processes. Approaches to Identification of Activation Receptors
Human and mouse NK cells express the ITIM-bearing Initial progress in elucidating NK-cell activation receptors
leukocyte-associated Ig-like receptor 1 (LAIR-1, CD305), was difficult. The approaches that yielded the molecular def-
which is an Ig-like molecule broadly expressed by most leu- inition of the TCR, such as subtractive hybridization, mu-
kocytes and encoded in the LRC.417,418 Initial reports indicat- tagenesis of T-cell tumors, and anticlonotypic mAbs,434–436
ing that LAIR-1 binds epithelial cellular adhesion molecule were of limited success.437,438 Unlike the working paradigm
were irreproducible.419,420 Instead, LAIR-1 binds multiple of MHC restriction that guided the molecular identification
forms of collagen,421 which has been validated in crystal- of the TCR, the principles guiding NK-cell activation by
lographic and biochemical studies.422 Interestingly, LAIR-1 targets were unclear. Breakthroughs in identifying NK-cell
mediates inhibition that is independent of Src homology 2 activation receptors thus required other approaches.
(SH2)-domain– containing phosphatases, and instead re- Some activation receptors were recognized because they
cruits C-terminal Src kinase (Csk),423 suggesting that Csk may were first identified on other cells, such as Fc γRIII. Other

activation receptors were identified by genetic means, facilitates expression and provides signal transduction ca-
such as cDNA clones for molecules resembling the inhibi- pacity.447–450 Although Ly49H will be considered below in
tory receptors but lacking cytoplasmic ITIMs. Others were the context of viral infection, Ly49D has no known role in
identified by a genetic positional cloning approach. Specific viral defense. A positional cloning approach indicated that
stimulation of NK cells through mAbs proved useful for the Ly49D is the product of the Chok locus, which controls NK-
initial identification of candidate activation receptors and cell specificity for killing of a xenogeneic target, Chinese
to validate the activation function of receptors identified by hamster ovary cells,159,451 due to recognition of a Chinese
other means. hamster MHC-I molecule.452 Interestingly, when Ly49D was
NK cells can be stimulated to mediate antibody-depen- transfected into RNK-16 cells, Ly49D can recognize H2Dd,440
dent cellular cytotoxicity (ADCC) through the Fc γRIII but H2Dd tetramers do not bind Ly49D for unclear reasons,
(CD16) receptor that binds the Fc portion of the IgG coat- although potentially reflecting lower avidity.213,262 Several
ing a target. In a related way, anti-Fc γRIII can also trigger other Ly49 receptors have been identified in non-C57BL/6
through CD16 in a process termed “redirected lysis” or “re- mouse strains that have properties of activation receptors
verse ADCC” because the antibody binds in the opposite (charged transmembrane residues, no ITIMs) but they have
orientation to ADCC. A few mAbs against other NK-cell been less well characterized,257,258 except for Ly49P in MA/
surface molecules also can activate in the redirected lysis My mice, which is involved in controlling viral infection
assay, highlighting a relatively unique functional property that is related to MHC-I recognition453 (see subsequent de-
of the recognized molecules (and mAbs), because activa- tails). Thus, some members of the Ly49 family are activation
tion does not occur when most NK-cell surface molecules receptors with apparent specificity for MHC-I, potentially
are cross-linked. First, popularized for analysis of anti-TCR with less avidity.
antibodies,439 redirected lysis occurs when IgG reacts spe- Molecular cloning of the KIR family also led to the iden-
cifically with the NK-cell receptor, and its Fc portion binds a tification of two domain receptors (also known as p50) or
target cell Fc receptor (Fc γR) that apparently provides bridg- three domain Ig-like receptors with short cytoplasmic do-
ing and cross-linking effects.439 Target lysis does not occur if mains lacking the ITIM.454,455 These molecules are now
Fc γR binding on the target is prevented with Fc γR-deficient known as the KIR2DS or KIR3DS, respectively, with num-
targets, F(ab ′)2 fragments of the anti–NK-cell receptor anti- bers for specific molecules (ie, KIR2DS1 [CD158h, p50.1],
body, or anti-target cell Fc γR Ab blockade. In the latter case, KIR2DS2 [CD185j, NKAT5, p50.2, clone 49], KIR2DS3
Fc regions must be removed to prevent inadvertent trigger- [NKAT7], KIR2DS4 [CD158i, NKAT8, clone 39], KIR2DS5
ing of conventional ADCC via CD16 on the NK cell. Thus, [CD158g, NKAT9], and KIR3DS1 [CD158e2]). Expression
the redirected lysis assay is a helpful experimental tool. of these molecules may be difficult to determine because
Gene transfer studies have been helpful adjuncts to study mAbs for KIR2DL molecules cross-react with KIR2DS mol-
NK-cell receptors. NK cells are difficult to transfect, and ecules.456 KIR2DS molecules can associate with DAP12 and
there are few useful tumors with the notable exception of activate NK cells in the redirected lysis assay.454,457 KIR2DS
RNK-16, a rat NK tumor line.440 Viral vectors, such as vac- and KIR3DS molecules can recognize HLA class I molecules
cinia virus, have been useful for gene transfer with the caveat with specificities similar to corresponding inhibitory KIRs,
that functional experiments have to be performed within a apparently with lower avidity.458–461 However, KIR2DS4 and
small time frame before untoward effects occur.441 Recent HLA-C alleles and KIR3DS1 and HLA-B alleles influence
use of lentiviral vectors also show promise for gene transfer their respective interactions.462,463 Thus, further analysis is
into primary NK cells.442 needed to support the hypothesis that activating forms of
Recent studies have exploited reporter cell assay systems KIRs bind HLA alleles less well than the inhibitory receptors
similar to those used to identify TCR ligands.443 ITAM- as a potential explanation for dominance of inhibition over
mediated signaling leads to inducible, nuclear factor of ac- activation (see Fig. 17.2D).
tivated T cells (NFAT) – dependent expression of a reporter It remains possible that the activating forms of the Ly49s
molecule, such as β -galactosidase or GFP. Even an inhibitory and KIRs have other ligands and perhaps their MHC speci-
receptor can be used to activate the reporter cell by fusion of ficities are instead somehow related to their physiologi-
its extracellular domain to a suitable transmembrane and cally relevant ligand. Indeed, this has been demonstrated
cytoplasmic domain containing ITAMs. Such reporter cells for Ly49H, an activation receptor that recognizes a virus-
can then be used to detect ligands.444–446 encoded ligand with an MHC-I–like fold.444,446,464 Moreover,
In the following sections, we will describe NK-cell KIR2DS4 may recognize a non-MHC ligand.465 Thus, fur-
activation receptors with an emphasis on those with known ther analysis is required for understanding the role of Ly49
ligands. and KIR activation receptors in MHC-I recognition and
with respect to their inhibitory counterparts.
Activation Receptors Related to Major Histocompatibility In addition to NKG2A, the NKG2 family also contains
Complex–Specific Inhibitory Receptors NKG2C, NKG2E, and NKG2F, which are products of differ-
Despite initial characterization as inhibitory receptors for ent genes.355,466 (NKG2B is an alternatively spliced isoform of
MHC-I, the Ly49 family contains other members (ie, Ly49D NKG2A, whereas NKG2H is an alternatively spliced isoform
and Ly49H in C57BL/6 mice) without cytoplasmic ITIMs. of NKG2E.) NKG2C and NKG2E lack cytoplasmic ITIMs
Instead, they are activation receptors containing charged and contain charged transmembrane residues for associa-
transmembrane residues for association with DAP12 that tion with DAP12.467 While the role of NKG2F is unknown

because it lacks an external domain and remains inside the ligation but not when NK cells are exposed to NK-sensitive
cell, associated with DAP12 but not with CD94,468 NKG2C targets. Deficiency of γ chain abrogated ADCC but not natu-
and NKG2E form functional heterodimers with CD94.364 ral killing.474 Thus, CD16 is not involved in natural killing.
Like CD94/NKG2A receptors, these heterodimers recog- CD16-related artifacts must be considered when study-
nize HLA-E or Qa-1, but unlike CD94/NKG2A receptors, ing NK cells. Flow cytometry experiments may be flawed if
they activate NK cells.469,470 Interestingly, the inhibitory CD16 binding is not taken into account. To eliminate this
form binds with higher affinity to HLA-E than the activat- possibility, F(ab′)2 fragments should be used. Alternatively,
ing form.382,383,471,472 There also appears to be some peptide blockade of Fc γRIII binding may be sufficient with protein
preference between the different functional forms382,383,473 A or G (that bind Fc region on Ig) or anti-Fc γRIII mAbs,
that may be relevant in certain physiologic situations. Thus, such as unlabeled mAb 2.4G2 (ATCC HB 197) that reacts
the CD94/NKG2 receptors may discriminate between subtle with both mouse Fc γRII and FcγRIII.478 Similarly, as dis-
differences in their MHC-Ib ligands. cussed previously, antibody blockade experiments should be
done with caution if the antibody specifically reacts with the
Fcg RIII (CD16) target because ADCC may be stimulated.
Frequently overlooked but perhaps the first molecularly
defined activation receptor on NK cells is Fc γRIII (CD16), NKG2D
through which NK cells mediate ADCC against IgG-coated NKG2D (KLRK1) was first cloned from human NK cells
targets.474,475 Unlike other Fc γ receptor–bearing effector as a cDNA related to NKG2A and C.355 However, NKG2D
cells, NK cells are generally thought to express only one of is distinct from other NKG2 molecules for several reasons.
the known Fc γ receptors that binds IgG with low affinity,476 There is only limited sequence homology between NKG2D
although others suggest that human NK cells may express and other NKG2 molecules (28% amino acid identity for
Fc γRII isoforms.477 There are two human Fc γRIII isoforms the lectin-like domain) whereas other NKG2 molecules
with identical extracellular domains.476 Human NK cells are closely related to each other (70% identity). Rather
express only Fc γRIIIA., which is a transmembrane molecule, than heterodimerizing with CD94, NKG2D is expressed
whereas Fc γRIIIB has a glycosylphosphatidyl-inositol (GPI) as a disulfide-linked homodimer on all NK cells in hu-
linkage and is expressed by neutrophils. In mice, only the mans and mice. In humans, NKG2D is also expressed on
transmembrane isoform (FcγRIII) is present478 and dis- all γδTCR+ and CD8 + T cells, whereas in mice, NKG2D
plays 95% sequence conservation with muFcγRII. The re- is expressed on most NKT and γδTCR+ T cells but not on
cently described FcγRIV is a newly recognized orthologue of resting CD8 + T cells.81–83 However, essentially all activated
human CD16A but is not expressed on mouse NK cells.479,480 mouse CD8 + T cells express NKG2D. In both humans and
There are species differences in CD16 binding to mouse IgG mice, CD4 + T cells do not express NKG2D, but it is found
isotypes; mouse IgG3 mAbs bind human CD16 the most on a subset of CD4 + T cells in patients with rheumatoid ar-
efficiently (3 > 2a > 2b >> 1), whereas they bind mouse CD16 thritis.84 Finally, NKG2D has functional properties and li-
with the lowest affinity (2b > 2a > 1 >> 3).476 In the labora- gand specificities that distinguish it from the other NKG2
tory, a rabbit antimouse Ig polyclonal Ab, whose Fc portion molecules, indicating that NKG2D should not be considered
binds strongly to both human and mouse CD16, could be as a member of the NKG2 family.
added to facilitate Fc receptor binding. NKG2D does not have any known cytoplasmic motif and
The transmembrane Fc γRIII molecules are physically was first shown in humans to preferentially associate with
associated with Fc εRIγ and less commonly with CD3ζ.481 a signaling chain termed DAP10, encoded by a gene local-
Fc γRIII can also associate with γζ heterodimers. The associ- ized 130 bp away from the gene for DAP12.487 DAP10 does
ated chains are required for optimal cell surface expression not have any ITAMs; instead, it contains a YxxM motif for
of Fc γRIII and for signal transduction. After cross-linking, recruitment of PI3K.487 This motif is similar to that found
Fc γRIII activates biochemical events that are reminiscent of in CD28, and functional studies indicate that NKG2D can
T-cell activation, leading to granule exocytosis and cytokine act as a costimulatory molecule on T cells.488–490 Moreover,
production.161,482–485 In vivo, ADCC may be useful in host DAP10 has a site for recruitment of Grb2.491 Thus, NKG2D
defense against pathogens or infected cells if Abs are bound may provide qualitatively different signals, resulting in dif-
to their surface, triggering not only killing but also cyto- ferent cytokine production, for example,490 than activation
kine production and other NK-cell responses. Although NK receptors associated with ITAM-signaling chains.
cells are generally thought to participate early in a primary Other studies have suggested that NKG2D functions as a
immune response (see subsequent discussion), the delay re- primary activation (triggers alone) rather than costimula-
quired for isotype switching to IgG production suggests that tory receptor (does not stimulate unless it synergizes with
CD16 cross-linking on NK cells plays a role in secondary another receptor) on NK cells.80,492 Such studies need to
immune responses in vivo. be reconsidered in light of several factors. 1) Many stud-
Regardless, ADCC is remarkably similar to natural kill- ies of NKG2D function use targets that are poorly killed by
ing and was important for the initial establishment of the NK cells. When these targets are transfected with NKG2D
concept of NK-cell activation receptors.161 Yet, CD16 is not ligands, killing is enhanced in an NKG2D-dependent man-
required for NK-cell target recognition because human ner. However, such studies do not distinguish whether
CD16 − CD3 − lymphocytes can still mediate natural kill- NKG2D functions as a primary activation receptor or as a
ing.486 Moreover, CD3ζ is phosphorylated upon CD16 costimulatory receptor (analogous to CD28 requirement for

full activation of T cells) as the same experimental outcome Where studied, all NKG2D ligands have distant amino
is anticipated in either case. 2) When cross-linking is done acid similarity to MHC-I, although they share only about
with immobilized mAbs alone and CD16 coengagement on ∼25% amino acid identity with each other. Within a family,
NK cells is avoided, NKG2D functions as a costimulatory however, they may be much more closely related (up to 90%
receptor on mouse IL-2-activated NK cells.80 3) In mice but identity). None of the NKG2D ligands associates with β2m
not humans, there are two alternatively spliced isoforms of or binds peptides.504–507 Many contain only the α1/α2 plat-
NKG2D.34,493 A long form (NKG2D-L) contains a 13 amino forms and many are GPI-linked to the plasma membrane.
acid extension at the amino terminus (cytoplasmic domain) The ligands display binding to NKG2D in two ways based
as compared to the short form (NKG2D-S). Resting NK cells on affinity.501,502,504,508,509 For example, RAE1α , β, γ, and δ,
predominantly express NKG2D-L that preferentially associ- and H60b and c demonstrate low-affinity interactions with
ates with DAP10. However, activation of NK cells with cyto- mouse NKG2D, similar to MICA to human NKG2D (K D =
kines causes a transient increase in NKG2D-S that associates ∼300 to 8700 nM). In contrast, RAE1ε (also described as
with ITAM-containing DAP12 as well as DAP10. However, RAE1B6),509 H60a and MULT1 show a much higher affinity
others have found that NKG2D-L can associate with DAP12, interaction (K D = 6 to 30 nM). Thus, closely related mole-
albeit to a lesser degree, and that both isoforms are present cules within a family may display wide disparity in affinities
in resting NK cells.494 Regardless, in the absence of DAP10, whereas distantly related molecules can share high affinity
mouse NKG2D can associate with DAP12,35 allowing it to with NKG2D.
signal akin to a primary activation receptor. Thus, mouse Structural studies at 2.6 to 3.5 Å resolution indicate
NKG2D is an unusual example of a receptor with the same that the NKG2D ligands have MHC-I–like folds, though
extracellular domain but with potentially different func- the “peptide-binding cleft” is closed.504,505,510 NKG2D binds
tional outcomes (primary activation versus costimulation), its ligands more analogous to TCR docking on MHC and
depending on its associated partner chain. unlike Ly49 recognition (see Figs. 17.4 and 17.5), despite
There is remarkable plasticity for NKG2D in that it can the relationship of NKG2D and Ly49 receptors as NKC-
bind many apparently disparate ligands that are only superfi- encoded, lectin-like homodimers that bind MHC-related
cially related to each other by sequence alignment, and human molecules.504,510,511 Indeed, NKG2D interactions with its li-
NKG2D can bind mouse ligands and vice versa. Based on gands are much more aligned with CD94/NKG2A recogni-
soluble MICA binding, the first human NKG2D ligands were tion of HLA-E, as described previously. Moreover, NKG2D
found to be MICA and MICB (MHC-I chain-related, A and uses largely nonoverlapping patches to engage a similar or-
B) encoded on chromosome 6p21.3, centromeric to HLA-B thogonal footprint on its disparate ligands. Interestingly, the
locus.81 Subsequently binding studies of UL16 from human putative immunoevasion molecule, UL16, binds MICB in
CMV (HCMV) identified the ULBP (UL16 binding protein) essentially the same manner as NKG2D is predicted to bind,
family that in turn binds NKG2D.495 The UL16-binding even though UL16 has a different three-stranded β-sheet
proteins (ULBPs) are encoded by genes in the retinoic acid structure.512 Additional structures of NKG2D complexed
expressed transcript (RAET1) gene family (official HUGO with other ligands should yield additional insight as to how
nomenclature) encoded on chromosome 6q24.2-q25.3 and NKG2D can bind a panoply of distantly related ligands with
discovered by genomic mining. Whereas RAET1F, RAET1J, varying affinities.
RAET1K, and RAET1M are pseudogenes, other RAET1 NKG2D function has been described in the “induced-
genes give rise to ULBP1 (RAET1I), ULBP2 (RAET1H), self” model (see Fig. 17.2E) because expression of its ligands
ULBP3 (RAET1N), ULBP4 (RAET1E), ULBP5 (RAET1G), can be inducible and can override inhibitory influences of
and ULBP6 (RAET1L), some of which have been extensively MHC-I.513–515 In many cases, low basal transcriptional lev-
studied as ligands for NKG2D.495–498 MICA, MICB, ULBP4, els are markedly upregulated in pathologic conditions. For
and ULBP5 are considered to be transmembrane proteins example, MICA and MICB expression is markedly enhanced
whereas the other RAET1 proteins are GPI-linked.499 on epithelial tissues in inflammatory bowel disease.516
Mouse NKG2D ligands were discovered with soluble Whereas this notion was initially thought to be related to
NKG2D that was used for expression cloning.82,83 Two types heat shock elements in MICA promoter, recent detailed
of ligands were originally defined: the minor histocompati- studies indicate that NKG2D ligand transcription in some
blity antigen H60 and members of the retinoic acid early in- cells is not affected by heat shock or hypoxia in vitro.517
ducible gene-1 (RAE-1) family. It is now clear that the mouse Instead, transcription of NKG2D ligands can be induced
NKG2D ligands are the H-60 family, consisting of H60a, b, by several other stimuli. For example, DNA damage from
and c, the RAE-1 family, consisting of RAE-1α , β, γ, δ, and ε ionizing radiation or chemotherapy agents results in up-
(also known as Raet1a-e), and murine ULBP-like transcript regulation of human and mouse NKG2D ligand transcripts
(MULT1).82,83,500–502 There are strain-specific differences in and concomitant surface protein expression.517 This process
ligand expression. For example, BALB/c mice express H60a, was dependent on activation of the ataxia telangiectasia,
RAE-1α , β, and γ, whereas C57BL/6 mice do not express mutated (ATM), and ATM- and Rad3 (ATR)-related path-
these molecules and express RAE-1δ and ε.503 In addition, ways. Interestingly, several micro-ribonucleic acids (miR-
there are allelic forms of NKG2D ligands (eg, 80 different NAs) can regulate MICA and MICB expression, and they
alleles of MICA and 33 alleles of MICB [hla.alleles.org/ are decreased following heat shock, allowing upregulation
classo.html]), but these have not been fully characterized in of MICA and MICB.518 On the other hand, Dicer knock-
humans (or mice). down also results in a DNA damage response that results

in enhanced MICA and MICB expression.519 Regardless, as a result of proteolytic cleavage of membrane-expressed
NKG2D ligand expression due to genotoxic responses may ligands,533,536 though the enzyme required is unclear, and
be relevant to chemotherapy effects in cancer, such as mul- the process apparently requires palmitoylation of MICA.537
tiple myeloma.520 Mice Tg for NKG2D ligands show impotent NKG2D re-
Other external stimuli, such as phorbol ester, retinoic sponses,534,538 a topic better discussed subsequently in the
acid, cytokines, and TLR stimulation, can induce NKG2D context of NK cell tolerance and education. On the other
ligand expression on mouse and human cells.500,521,522 The hand, acute upregulation of an NKG2D ligand alone can
phorbol ester effect is regulated by the transcription fac- rapidly induce immune responses539 (albeit effects not ex-
tor JunB523 and retinoic acid stimulation was the impetus amined for NK cells). Taken together, these findings may
for original identification of the RAE-1 family.524 Cytokines explain why tumors frequently express NKG2D ligands that
can also regulate NKG2D ligands. For example, IFNγ can would otherwise enhance their susceptibility to NK-cell at-
downregulate transcripts for H60a expression on tumor tack, supporting a role for NKG2D in tumor surveillance
cells and alter NK-cell susceptibility.525 For MICA, the IFNγ and also the general concept that NKG2D-mediated activa-
effect is mediated through miRNAs.526 Interestingly, TLR- tion is balanced between acute, induced-self, and chronic
induced primarily MICA and not MICB, through pathways expression of its ligands.
dependent on ATM, ATR, and miRNAs.522 Viral infection Additional studies support a role for NKG2D in tumor
can also induce NKG2D ligands,527 perhaps not surprisingly surveillance. For example, mice are more sensitive to de-
because viruses target NKG2D and its ligands (described in veloping methylcholanthrene-induced fibrosacromas when
the following). For example, murine CMV (MCMV) infec- NKG2D is neutralized by chronic anti-NKG2D mAb admin-
tion induced RAE-1 mRNA and surface expression within istration86 or targeted deletion.540 Moreover, mice lacking
18 hours. This effect was independent of the DNA damage γδT cells are more susceptible to carcinogenesis apparently
response but required PI3K activation.527 However, PI3K ac- due to an NKG2D-dependent effect.500 IFNγ is a known host
tivation alone was insufficient to induce ligand expression, mediator that shapes the tumor phenotypes in a broader
suggesting that additional signals are required for NKG2D process known as “immunoediting,”541,542 and IFNγ also
ligand expression by viral infections and perhaps other mediates downregulation of H60 expression on tumors.525
stimuli. Thus, NKG2D and its ligands are important in the host re-
For other NKG2D ligands, transcripts are constitutively sponse (or lack thereof) to tumors.
expressed in a wide variety of tissues495,501; posttranslational A role for NKG2D in antiviral responses is indicated by
modifications alter surface protein expression levels. For studies that show that viruses use different strategies to inter-
example, mRNA for MULT1 is widely expressed in normal fere with ligand recognition by NKG2D. This was fi rst noted
tissues501 and MULT1 surface expression is regulated by when the HCMV protein UL16 was found to bind ULBPs.495
ubiquitination.528 MULT1 has a long cytoplasmic domain, In infected cells, UL16 retains some (MICB, ULBP1, and
unlike most other NKG2D ligands, which are GPI-linked. ULBP2) but not all NKG2D ligands in the endoplasmic re-
Under normal circumstances, MULT1 is ubiquitinated on ticulum and cis-Golgi, preventing their expression on the
its cytoplasmic Lys residues and is targeted for degrada- cell surface and protecting from NK cell lysis.543–545 HCMV
tion. However, cellular stress, such as heat shock or ultra- also encodes UL142, which retains certain alleles of MICA
violet irradiation, reduced its ubiquitination and allowed in the cis-Golgi.546 In mice, MCMV encodes four molecules
surface expression. In particular, MULT1 expression is that downregulate expression of all NKG2D ligands. gp40
regulated by the E3 ubiquitin ligases, membrane-associated from the m152 open reading frame (ORF) downregulates all
RING-CH (MARCH) 4 and membrane-associated MARCH five RAE-1 ligands but has no effect on H60 or MULT1.503,547
9.529 A human virus, Kaposi sarcoma– associated herpesvirus H60 is downregulated by the product of the m155 ORF at
(KSHV), exploits this pathway to avoid NK-cell attack (de- a post-Golgi level, perhaps by targeting H60 for protea-
tailed in the following). somal degradation548,549 and MULT1 expression is affected
Despite the nearly universal acceptance of the induced- by m145.550 The herpes virus Fc receptor (fcr-1), the product
self model for NKG2D function, it is not clear if this hy- of the m138 ORF, targets RAE-1ε, H60, and MULT1.551,552
pothesis is applicable to all of its ligands.530,531 It remains Interestingly, KSHV encodes an E3 ligase, termed K5, that
possible that constitutive expression or affinity of some can downregulate MICA and MICB expression.553 This ef-
NKG2D ligands may be too low to permit NKG2D activa- fect was due to ubiquitinated internalization of MICA, but
tion to override MHC-I– dependent and –independent in- not degradation, and protected cells from NK-cell cytotox-
hibitory receptors. Continued analysis at the protein level is icity, suggesting that KSHV targets a normal pathway regu-
anticipated, as well as study of NKG2D ligands in specific lating NKG2D ligand expression. HCMV and other herpes
tissues, exemplied by analysis of H60c in the skin where it viruses (KSHV, Epstein-Barr virus) encode miRNAs that
is selectively expressed and can also costimulate dendritic affect MICA and/or MICB expression.554,555 Non–herpes vi-
epidermal T cells.532 ruses also target the NKG2D pathway; adenovirus E3/19K
On the other hand, chronic exposure to membrane- sequesters MICA and MICB in the endoplastic reticulum,556
bound or soluble NKG2D ligands results in downregulation whereas the orthopoxviruses, cowpox, and monkeypox vi-
of NKG2D expression and lower functional responsive- ruses encode a secreted, high-affinity NKG2D ligand that
ness.85,533–535 Patients with tumors expressing MIC frequently antagonizes NKG2D recognition and activation.557 Thus, the
contain soluble MIC in their peripheral blood, presumably multitude of viral strategies that affect NKG2D recognition

strongly suggests that NKG2D plays an important role in the associated with Fc εRIγ, although its deficiency curiously
host response to viral infections. does not affect NK1.1 expression.579 An Nkrp1a loss mutant
Surprisingly, an NKG2D knockout mouse displayed of rat RNK-16 cells failed to kill certain targets; transfec-
enhanced resistance to MCMV infections.87 This mouse tion restored killing capacity, suggesting that rat Nkrp1a is
also displayed modest changes in NK-cell development with an activation receptor specific for target determinants.580
faster NK-cell maturation that were not demonstrated in Thus, Nkrp1 molecules were among the first described NK-
another NKG2D knockout,540 even though both were gen- cell– specific activation receptors though inhibitory forms
erated directly in C57BL/6-derived embryonic stem cells. were later described.
A subsequent retargeting of the Klrk1 locus recapitulated the Inasmuch as Nkrp1 molecules are homologous to C-type
enhanced resistance to MCMV,558 which for the moment is lectins, the ligands for Nkrp1 molecules were initially
at odds with viral neutralization of NKG2D. presumed to be carbohydrates but Nkpr1 molecules lack
Finally, NKG2D may play a role in autoimmune dis- residues for coordinate binding of calcium that is required
orders. In autoimmune type I diabetes mellitus (TIDM), for authentic C-type lectin recognition of carbohydrates.222
MICA polymorphisms are reportedly associated with in- While the ligands for Nkrp1c have not yet been identi-
creased risk,559 but MICA is closely linked with HLA, which fied, Nkrp1f recognizes C-type lectin related g (Clrg, also
clearly imparts significant risk so additional analysis of large known as Clec2i, Dcl1).445,581–583 Nkrp1f is presumed to be
patient cohorts is needed to segregate the MICA effect from an activation receptor because it has a charged transmem-
that of HLA alleles in TIDM or any other HLA-associated brane residue and no cytoplasmic signaling motifs. The
autoimmune disease.560 Nonetheless, NKG2D ligands are inhibitory receptor, Nkrp1d, is expressed on all NK cells
expressed in the prediabetic NOD mouse pancreas and anti- in C57BL/6 mice and is specific for Clrb (Clec2d; osteo-
NKG2D blockade prevents TIDM.561 In human patients with clast inhibitory lectin, Ocil). Thus, the Nkrp1 family is an
inflammatory syndromes, such as rheumatoid arthritis, MHC-independent system that is presumably involved in
Wegener granulomatosis, and unstable angina, there is an self-tolerance and is the first example of a lectin-like recep-
unusual population of CD4 + CD28 − NKG2D + T cells that tor recognizing a lectin-like ligand unlike MHC-like ligands
is expanded.562–564 NKG2D can provide a costimulatory sig- for other NKC-encoded receptors.
nal for T cells, at least for CD4 + CD28 − T cells in rheuma- Genome sequence analysis indicates that Nkrp1d
toid arthritis and rheumatoid arthritis synovium expresses in C57BL/6 mice is represented by Nkrp1b in other
NKG2D ligands.84,489 IL-15 induces NKG2D expression on strains.75,76,569,584 The Nkrp1b molecule from SJL and SW
intestinal epithelial T-lymphocytes that then display a LAK mice was discovered to be reactive with the anti-NK1.1
cell–like promiscuous killing capacity against enterocytes mAb, highlighting the close similarities of these molecules
which upregulate NKG2D ligands from gliadin exposure, and serologic cross-reactivity. Whereas the older literature
suggesting that NKG2D may play a pathogenic role in celiac suggested that there are also strain differences in transcript
sprue.36,565,566 On the other hand, others have postulated expression with markedly lower expression in BALB/c NK
that CD4 + NKG2D + T cells may be normally immunosup- cells,585 more recent data demonstrate abundant expres-
pressive as they are inversely correlated with disease activ- sion of Nkrp1 transcripts.584 Moreover, the BALB/c allele of
ity in patients with systemic lupus erythematosus.567 Thus, Nkpr1b also reacts with Clrb, indicating conserved specific-
NKG2D may play a pathogenic role, albeit somewhat unity and function.
clear, in several autoimmune disorders in humans and mice. The Clr molecules belong to an NKC-encoded family
with type II orientation and C-type lectin homology.581 They
Natural Killer Gene Complex–encoded Lectin-like are most closely related to the CD69 molecule encoded by an
Receptors Recognize Natural Killer Gene adjacent gene. In C57BL/6 mice, seven Clr genes have been
Complex–encoded Lectin-like Ligands identified at the genomic level. RT-PCR analysis indicates
First identified by functional studies in the rat, Nkrp1 that Clrb is broadly expressed, whereas Clrg and Clrf genes
(Klrb1) molecules belong to a family of lectin-like, disulfide- are present in restricted and nonoverlapping tissues, includ-
linked homodimers with type II orientation encoded in the ing NK cells, and Clra and Clrc transcripts have not yet been
mouse and rat NKC71,255,330,568–571 (see Fig. 17.3). Expression is identified. Clre, a probable pseudogene, has numerous stop
relatively selective for NK cells, although Nkrp1 molecules codons in its expected open reading frame. Genomic analy-
are also expressed by NKT cells in rats and mice.572,573 There sis indicates the existence of a rat Clr family,570 whereas in
is only a single gene (NKRP1A, CD161) in humans that is humans, three genes, LLT1,586 AICL,587 and DCAL-1588 are
expressed on a subpopulation of NK cells.574 The continu- localized next to CD69 and are related to the Clr family of
ing allure of the Nkrp1 family stems from observations that genes.
NK1.1 (the most widely known specific serologic marker of Human NKRP1A binds LLT1, a functional homolog of
NK cells) is encoded by Nkpr1c (Klrb1c) in C57BL/6 mice,71 mouse Clr.589,590 Moreover, activation-induced C-type lec-
Nkrp1 receptors are conserved, and they have interesting tin (AICL CLEC2B) is the ligand for the NK-cell receptor
genetics with respect to their ligands.575 activation receptor, NKp80 (KLRF1), which is also encoded
Initial functional studies indicated that rodent Nkrp1 in the NKC but is absent in rodents.591–593 Finally, CLEC2A
molecules can activate NK cells through redirected (keratinocyte-associated C-type lectin, KACL) is primarily
lysis,32,571,576–578 which can be prevented by inhibitory Ly49 reexpressed in the skin and is recognized by another NK-cell
ceptor engagement.140 Mouse Nkrp1c (NK1.1) is functionally activation receptor, termed NKp65 (KLRF2).594 All of these

receptors and their ligands are lectin-like molecules encoded redirected lysis assay 605 and granule exocytosis,606 but CD2−
in the human NKC. NK cells can still mediate natural killing.9 The cytoplasmic
The genetics of the Nkrp1 and Clr loci is especially inter- domains of SLAM, 2B4, NTBA, Ly9, CD84, and CRACC
esting from several viewpoints. 1) These loci are co-mingled contain a motif with sequence similarly to the ITIM,
in the NKC, from rodents to humans445,570,575,584,589,590 (see termed the immunoreceptor tyrosine-based switch motif
Fig. 17.3). 2) In mice, there is limited allelic polymorphism, (ITSM) consisting of TxYxxV/I consensus sequence.607 The
with conservation of gene order and content, despite genetic ITSMs allow interactions with a signaling adapter, SLAM-
proximity to the highly polymorphic Ly49 cluster.445,584,595 associated protein (SAP, also known as SH2D1A). The im-
For example, compare the conservation of the Nkrp1 gene portance of SAP and its associated receptors is highlighted
cluster with polymorphism of the Ly49 cluster in C57BL/6 by the X-linked lymphoproliferative (XLP) syndrome, a
and 129 strain mice.227,257,260,445,595 3) At the individual gene human immunodeficiency involving abnormal prolifera-
level, Nkpr1 alleles appear to be much less divergent than tion of T and B cells during Epstein-Barr virus infections
the corresponding Ly49 alleles in the same inbred mouse due to mutations in SAP.608
strains.257,445,584,595 4) The Nkpr1-Clr interval of the NKC SAP is related to Ewing sarcoma– associated transcript
appears to be genetically protected with suppression of re- (EAT2, also known as SH2D1B1) and EAT2-related trans-
combination.596,597 Genetic linkage of Nkrp1 and its ligands ducer (ERT, also known as SH2D1B2), which is present
thus resembles the tight genetic linkage of receptor and in rodents and only expressed in NK cells.604 SAP and the
ligand genes and recombinational suppression of the self- SAP-related molecules contain a single SH2 domain for
incompatibility loci in plants to prevent self-fertilization and interaction with the ITSMs of presumably all SLAM fam-
related mating loci in other species.598,599 Furthermore, the ily members except CRACC.609–613 Instead, CRACC re-
coevolution of such linked genes for receptors and ligands cruits EAT2, which functionally substitutes for SAP.612
also represents an interesting issue in evolutionary biology Subsequently, the SAP-related proteins can recruit a wide
because reciprocal mutations in both receptor and ligand variety of downstream signaling molecules including Fyn, a
genes are simultaneously needed to generate new specifici- Src-family tyrosine kinase, and tyrosine phosphatases, ino-
ties.598 Nevertheless, the genetic pressure to conserve the sitol phosphatases, and adaptor molecules.614
Nkrp1-Clr gene order and coding sequences may therefore The complexities of this receptor-ligand signaling path-
reflect a critical role for Nkrp1 and Clr molecules in innate way are evident from the previous description. Receptors
immune cell interactions and functions.575 have overlapping ligand specificities; they recruit different
While the in vivo role of Nkrp1 and Clr molecules adaptors that in turn can recruit downstream signaling
is incompletely understood, the best evidence for their molecules, even those with opposing functions. In this re-
functional importance comes from studies in rats. Rat gard, functional studies of SLAM family molecules on NK
CMV encodes RCTL (rat CMV C type lectin-like), which cells have often been ambiguous, often depending on the
closely resembles rat Clrb and regulates its expression on experimental approach.602 For example, 2B4 cross-linking
infected cells.600 RCTL interacts with the Nkrp1b inhibi- with mAb can both stimulate as well as inhibit mouse NK
tory receptor and inhibits NK cell killing of infected cells. cells,601 whereas a 2B4-deficient mouse primarily displays
An RCTL loss mutant rat CMV is attenuated in vivo in an effects consistent with 2B4 being primarily an inhibitory
NK-cell– dependent manner. These studies, indicating that receptor.615 However, the inhibitory effect of 2B4 is more
viruses encode decoy ligands for Nkrp1 receptors, sup- apparent in mice than humans.604 Some of these outcomes
port the importance of Nkpr1-Clr interactions in immune may depend on spatial distribution, 2B4 isoforms, or differ-
responses. ential recruitment of SAP or its related molecules.602 Further
analysis should enlighten NK-cell function and signal-
2B4, CD2, and Signaling Lymphocytic Activation ing because 2B4 and related receptors demonstrate MHC-
Molecule Family of Receptors independent regulation of NK cells, and mediate inhibition
The 2B4 (CD244, SLAMF4) molecule was originally identi- and activation in a functionally distinct manner from the
fied on mouse NK cells with a mAb that perturbed mouse MHC-specific inhibitory receptors and ITAM-signaling
NK-cell function.601 2B4 is a type I integral membrane pro- chain associated receptors, respectively.
tein that belongs to the family of Ig-like molecules, includ-
ing signaling lymphocytic activation molecule (SLAM) Natural Cytotoxicity Receptors
(SLAMF4, CD150), NK, T, and B cell antigen (NTBA, also A series of mAbs that redirected lysis of human NK cell
known as SLAMF6 or Ly108), Ly9 (SLAMF3, CD229), CD84 clones lacking KIRs led to the identification of the “natural
(SLAMF5), and CD2-like receptor activating cytotoxic cells cytotoxicity receptors” (NCRs), NKp46 (NCR1, CD335),
(CRACCs; SLAMF7, CD319).602 Although these receptors NKp44 (NCR2, CD336), and NKp30 (NCR3, CD337).616–
620
may be broadly expressed, 2B4, NTBA, Ly9, CD84, and These molecules are selectively expressed on NK cells,
CRACC are expressed on NK cells. NTBA, Ly9, CD84, and though NKp44 is expressed only upon activation. Whereas
CRACC are involved in homophilic interactions whereas NKp46 is encoded in the LRC (see Fig. 17.3), NKp44 and
2B4 recognizes CD48, a GPI-linked molecule expressed on NKp30 are encoded in the class III region of the MHC
hematopoietic cells.603,604 CD48 itself is also recognized by on human chromosome 6.621,622 cDNA cloning revealed
CD2, albeit at a ninefold lower affi nity than by 2B4 (Kd = that they are type I integral proteins with one (NKp30,
∼16 μM). Anti-CD2 mAbs can stimulate NK cells in the NKp44) or two (NKp46) Ig-like extracellular domains.

They contain charged transmembrane residues for asso- cells.634,637–639 Thus, there may be heterogeneity among
ciation with ITAM-signaling chains (ie, ζ γ heterodimers IL-22 – producing NKp46 + cells.
(NKp46, NKp30) or DAP12 (NKp44). In the mouse, only
the gene for NKp46 is present in the syntenic region of Other Activation Receptors
chromosome 7 with the genes for other NCRs being absent Several other NK-cell– expressed molecules can activate cy-
(NKp44) or a pseudogene (NKp30).623 The NCRs appear to tolysis. For example, mAbs against mouse CD69 and Ly-6
play a role in cytotoxicity against tumors of varying ori- and rat gp42 can trigger killing.32,33 CD69 is encoded in the
gins because anti-NCR antibodies block target killing.624 A NKC and is structurally related to other NKC-encoded re-
mouse NKp46-Ig fusion protein binds RMA-S targets and ceptors. Interestingly, CD69 expression is upregulated upon
deficiency of NKp46 leads to impaired in vivo clearance of stimulation,322 and its expression is not confined to NK
RMA-S cells.93 Interestingly, human NCRs apparently rec- cells.640 CD69 is functionally active on a large variety of he-
ognize mouse tumors and vice versa, suggesting conserva- matopoietic cells when cross-linked by anti-CD69 mAbs.
tion of these receptor-ligand pairs across species but iden- Its ligand is unknown, but it regulates the function of the
tity of their target ligands were elusive,624 until recently for sphingosine 1-phosphate receptor-1 through a membrane
NKp30 and NKp46. interaction.641 Ly-6 belongs to a large family of small (15 to
In functional and structural studies, NKp30 clearly 18 kDa) GPI-anchored molecules that can activate lympho-
recognizes B7-H6, a member of the CD28 family that also cytes when cross-linked.642,643 Rat gp42 is a GPI-anchored
includes CTLA-4 and PD-1.625,626 B7-H6 is not expressed protein with two Ig-like domains that was originally identi-
normally, is found on tumor cells, and sensitizes targets fied on IL-2–activated NK cells and the rat RNK-16 NK cell
to NKp30-dependent cytotoxicity by NK cells. In a 2.0 Å line.33 Anti-gp42 can activate RNK-16 but not IL-2-activated
structure of NKp30 complexed to B7-H6, NKp30 engages NK cells. However, CD69, Ly-6, and gp42 are not expressed
its ligand in an antibody-like manner distinct from the way on freshly isolated NK cells and are expressed only after
other receptors bind B7 family molecules. Other potential activation through other pathways; they are therefore not
NKp30 ligands include pp65 of HCMV, though it is a tegu- involved in triggering natural killing by freshly isolated NK
ment protein and not expressed on the plasma membrane of cells. Although their physiological role is unknown, their
infected cells.627 HLA-B – associated transcript 3 (BAT3) was activation potential and enhanced killing by IL-2–activated
also proposed, but it is a nuclear protein that is released with NK cells suggest that these molecules may contribute to this
DNA damage or endoplasmic reticulum stress.628 Finally, phenotype.
NKp30 reportedly binds poxviral hemagglutinin indepen-
dent of B7-H6 and BAT3.629 Nonetheless, at least one ligand
is now well characterized for NKp30. Receptors Involved in Recognition of
Human NKp46 appears to recognize influenza hemag- Epithelial Tissues
glutinin on infected cells.629,630 The interaction is depen- NK cells possess receptors that are specific for ligands
dent on sialic acid residues on oligosaccharides on NKp46 expressed at cell-cell junctions in epithelial tissues.
itself,630 but this specificity is difficult to explain due to the Interestingly, ligands, such as cadherins, carcinoembryonic
ubiquitous expression of sialylated saccharides. NKp46 and antigen (CEA)-related adhesion molecules (CEACAMs),
NKp30 may recognize heparan sulfate proteoglycans on nectins, and nectin-like proteins (necls), are involved in
their cellular targets, though this is also controversial.631,632 forming adherens junctions and are engaged in homotypic
Nevertheless, NKp46-knockin (GFP)/knockout mice are or heterotypic interactions, and thus, are not normally
susceptible to influenza, consistent with a role for NKp46 exposed.644 KLRG1 (also known as mast cell– associated
in defense against influenza,93 albeit possibly in an indirect function antigen, MAFA) is a lectin-like receptor with a sin-
manner.633 These studies may provide important clues to gle cytoplasmic ITIM.645–647 Unlike human KLRG1, mouse
understanding of a receptor (NKp46) conserved in humans Klrg1 resides relatively distant and centromeric from the rest
and mice. of the NKC, consistent with gene duplication and chromo-
NKp46 Tg mice have been proposed to be useful for NK- somal inversion events. First discovered on rat mast cells, it
cell analysis.95 These mice are Tg for a construct consisting has a broader distribution in human and mouse, including
of 400 bp from the human NKp46 promoter driving ex- NK and T cells but not mast cells. Expression of Klrg1 on
pression of enhanced GFP and the human diptheria toxin NK cells is downregulated in MHC-I– deficient mice, un-
receptor. However, NKp46 + cells in the gut may be derived like the Ly49s that are modestly upregulated, but no MHC
from a non–NK-cell lineage. binding has been observed for Klrg1.648 Instead, mouse
Human NK-22 cells were identified in the gut as being Klrg1 binds three of the seven “classical” cadherins, E-,
CD3 − CD56 + NKp44 + cells that produce IL-22 but N-, and R-cadherin, leading to inhibition of NK lysis.649,650
not IL-17 upon stimulation with IL-23.88 In the mouse, Biophysical studies show the K D = 120 mM for mouse
a similar functional population producing IL-22 was KLRG1-mouse E-caherin.651 The structure of KLRG1 at 1.8 Å
identified among NKp46 + cells and to a lesser degree, resolution revealed a typical lectin-like domain for KLRG1,
among NK1.1+ cells.89–91,634 Immature human NK cells in which binds E-cadherin at a site distinct from where the
secondary lymphoid tissue and the uterus may produce IL- integrin a Eb7 (CD103) binds E-cadherin. A KLRG1-deficient
22.635,636 However, other reports indicate that NK-22 cells mouse demonstrates that it is dispensable for NK- and T-cell
are more closely related to lymphoid tissue inducer (LTi) differentiation and antiviral responses, but it has not been

exhaustively studied, especially for epithelial responses.652 This issue is further complicated by the capacity of receptors
Human CEACAM1 (CD66a) contains a cytoplasmic ITIM like DNAM-1 to modulate LFA-1 association and function
and can directly bind CEA, thereby potentially regulat- on NK cells.660 Moreover, NK cells express multiple recep-
ing NK-cell activities in human MHC-I– deficient patients tors capable of binding target ligands, with each receptor po-
and during pregnancy.653–655 Mouse NK cells also express tentially affecting different downstream signaling pathways.
CEACAM1.656 Inasmuch as CEACAM1 binds homophilic When pairs of these receptors (eg, CD16, NKp46, NKG2D,
and heterophilic ligands expressed on epithelial and other 2B4, DNAM-1, or CD2) are cross-linked, synergistic NK-cell
cells,657 NK cells thus express several inhibitory receptors for activation in terms of cytokine production and target kill-
ligands normally expressed on epithelial cells. ing can be seen for some but not all pairs.674 However, this
NK cells also express cell-cell adhesion Ig-like receptors approach has not been used for the panoply of NK-cell re-
for molecules located at the adherens junction (ie, nectins ceptors. Thus, how “accessory” molecules or other receptors
and necls).658,659 Specifically, they express CD226 (DNAM-1, contribute to each activation receptor function on NK cells
DNAX accessory molecule-1) that recognizes necl-5 (CD155, remains to be systematically determined.
poliovirus receptor, PVR) and nectin-2 (CD112, PVRL2).660
NK cells also express CD96 (Tactile, T-cell activation, in-
creased late expression) that binds necl-5 but not nectin-2, Ligands Recognized by Both Activation and
and class I-restricted T-cell– associated molecule (CRTAM), Inhibitory Receptors
a receptor that recognizes necl-2.661–663 DNAM-1 associ- Evident from the previous discussion is the general concept
ates with lymphocyte function– associated antigen (LFA-1) that NK cells usually express both activation and inhibitory
and regulates its capacity to bind its ligand, ICAM-1.660,664 receptors simultaneously. Moreover, some of these receptors
In addition, DNAM-1 can recruit actin-binding proteins with opposing functions bind essentially identical ligands.
to form an adhesive complex with other cells. DNAM- For example, both activation and inhibitory receptors in the
1– deficient mice have defective NK and T cells, and fail to Ly49 and KIR families reportedly bind MHC-I molecules,
control tumors, both transferred and de novo carcinogen- and CD94/NKG2A and CD94/NKG2C bind HLA-E/Qa1.
induced.665,666 Although these receptors can activate Src and In general, where studied, the inhibitory receptors tend to
other downstream pathways,659 they are thought to function bind ligands with higher affinities than their corresponding
on NK cells as adhesion receptors that enhance cytotoxicity, activation receptor counterpart, perhaps accounting for the
rather than as primary activation receptors. Conversely, NK common observation that inhibition tends to dominate over
cells also express T-cell immunoglobulin and ITIM domain activation. However, the overall relevance of paired activa-
(TIGIT), which recognizes PVR and PVRL2 but not PVRL3, tion and inhibitory receptors to NK-cell immune responses
and inhibits NK-cell killing.667,668 remains to be elaborated.
These studies suggest that these receptors may affect dif-
ferent NK-cell functions as related to epithelial tissues.644,649
For example, the inhibitory receptors may regulate NK-cell SIGNAL TRANSDUCTION IN NATURAL
transmigration across an epithelial barrier or prevent NK- KILLER CELLS
cell attack against normal tissues.649,669 While the activation NK cells receive two basic types of external stimuli from
receptors could also affect NK-cell transmigration,649,669,670 cytokines and chemokines, and from target cell recognition.
they also could be poised to attack cells that have disordered In general, their cytokine and chemokine responses are re-
cell-cell junctions, such as tumors that typically lose cad- lated to those found in other cells responding to the same
herin expression and expose nectins and necls, potentially pathways. However, as can be appreciated from the previous
making them more susceptible to NK attack, consistent description of individual target recognition receptors, the
with an “exposed-self” model for NK-cell activation (see pathways leading to target killing are complicated because
Fig. 17.2F, G). On the other hand, tumors may become re- some receptors are coupled to ITAM-containing signaling
sistant to NK cells by altering expression of these molecules. chains (CD3ζ, FcεRIγ, or DAP12). Indeed, some receptors
Current evidence provides some support to these complex use more than one of these signaling chains, which are gen-
scenarios that will require further investigation.644 erally thought to be equivalent but not dissected in detail.
Other receptors, such as NKG2D (through DAP10), have
Tyr motifs for recruitment of PI3K and Grb2. Still others
Accessory Molecules (2B4 and related receptors) have ITSMs. Moreover, integ-
The role of accessory molecules in NK-cell activation has rin signaling also affects NK-cell activation by targets, and
been difficult to address without knowledge of the “NK-cell it was reported that NKG2D signaling through DAP10 can
(activation) receptor.” Nevertheless, NK-cell cytotoxicity affect IL15 responses and vice versa,675 which highlights the
is critically dependent on classical accessory and adhe- possible cross-talk of signaling pathways more broadly in
sion molecules, such as LFA-1.671,672 Insect cells expressing immune cells.676 Finally, because individual NK cells can ex-
ICAM-1, a ligand for LFA-1, can be killed by human NK cells press (and use) multiple different receptors simultaneously,
through an LFA-1– dependent process, suggesting that LFA-1 and some but not all work synergistically,674 the outcomes of
alone is sufficient to trigger killing.673 LFA-1 cross-linking these activation pathways may reflect profound complexity,
does not lead to granule polarization, suggesting that it may even without consideration of inhibitory receptor signaling
be insufficient for triggering the killing process by itself.143 events.677

The study of NK-cell signal transduction also comes with they are required for NK-cell activation. In contrast,
some caveats. Most studies have been performed on bulk deficiencies in Vav family members lead to T- and B-cell
populations of NK cells, NK cell clones, or in vitro adapted development defects.687 These observations also indicate the
cell lines that may not be representative of pathways trig- redundancies in NK-cell receptor signaling.
gered in individual NK cells which may express only some On the other hand, Vav1 appears to be central to human
but not all signaling molecules detected in the entire NK- NK-cell activation by targets688,689 even though the role of
cell population. When NK cells are triggered by target cells, other Vav isoforms in human NK cells is less clear. Vav1 is also
there may be unknown stimulation through other receptors involved in macromolecular complexes with c-Cbl that are
because unknown target ligands may inadvertently trigger formed during NK-cell activation.690 Interestingly, c-Cbl pro-
or modulate functions of other NK-cell receptors. Finally, vides an inhibitory influence on NK-cell activation.689 During
there may be species (human versus mouse) differences in NKG2D and 2B4 synergistic signaling, strong Vav1 signals
signaling. Despite these complexities, themes have emerged override the c-Cbl effect to allow NK-cell activation, again
in understanding of NK-cell signal transduction. placing Vav1 as a central player in human NK-cell activation.
For the signal transduction pathways stimulated by Recruitment and activation of PI3K appears to be critical
NK-cell activation receptors coupled to ITAM-containing to NK-cell activation for target killing.691,692 How PI3K is acti-
signaling chains, downstream events resemble those found vated depends on the receptor and its proximal signal trans-
in TCR and BCR signaling.678 Activation receptor cross- duction events.677 In the case of NKG2D, the YxxM motif in
linking leads to ITAM phosphorylation by Src family ty- DAP10 is phosphorylated and recruits PI3K directly,693 which
rosine kinases, recruitment and activation of Syk family appears to be sufficient to activate NK-cell killing in the ab-
tyrosine kinases, and subsequent downstream activation sence of DAP12 or Syk family tyrosine kinases.694 The path-
events. Inasmuch as TCR and BCR signaling is covered in way may be more complex in human NK cells where evidence
detail elsewhere, we will highlight here notable differences suggests that DAP10 recruits a Grb2-Vav1 intermediate for
with TCR and BCR signaling. 1) Individual NK cells may activation.491 For other activation receptors, Syk activation is
simultaneously express multiple activation receptors, each of upstream of PI3K695 and may directly recruit PI3K.696 PI3K in
which may associate with different ITAM-signaling chains turn activates a mitogen activated protein kinase (MEK) for
that could be phosphorylated by different Src family tyrosine stimulation of extracellular signal-regulated kinase (ERK).691
kinases and/or lead to different downstream signaling events. Ultimately, actin polymerization occurs, granules become
2) NK cells express multiple Src kinases, including Lck, Fyn, polarized toward the target, and exocytosis occurs, resulting
Src, Yes, Lyn, and Fgr,679 with redundancies in their contri- in target apoptosis, as discussed previously. Interestingly, in
butions to ITAM phosphorylation. For example, Fyn is ac- human NK cells, engagement of LFA-1 by ICAM-1 on insect
tivated following Ly49D (DAP12) cross-linking but NK cells cells was sufficient to induce granule polarization but not de-
deficient in Fyn, Lck, or both still kill in a Ly49D-dependent granulation,143,674 suggesting additional complexity regarding
manner.680 3) NK cells express both Syk family tyrosine ki- the signals required for polarization versus degranulation.
nases, ZAP-70, and Syk itself, whereas T and B cells express High-resolution microscopy has enabled visualization
either, respectively, but not both. Deficiency of either ZAP-70, of events following NK-cell activation. Like CTLs, NK cells
Syk, or both has only minimal effects on NK cell killing,681 form discrete protein clusters at the site of contact with their
suggesting other pathways for transmitting NK activation targets, termed the immunological synapse,697,698 though
signals. 4) NK cells express many adapter molecules found obviously the TCR is not involved. Following target con-
in T and B cells, but their contributions to NK signal trans- tact, the NK immunological synapse (NKIS) is formed at
duction are less well defined. For example, SLP-76 and linker the NK-target interface in discrete stages whereby receptor
for activation of T cells are dispensable for NK-cell signaling, signaling, fi lamentous (F)-actin rearrangement, and po-
although they are both required for T-cell signaling.682,683 5) larization of granules occur.699,700 CD2, F-actin, LFA-1, and
Different isoforms of signal transduction molecules are re- Mac-1 accumulate in the peripheral supramolecular activa-
sponsible for NK-cell activation. For example, mouse NK tion complex (pSMAC), whereas granules (perforin) and the
cells utilize phospholipase C-γ 2 (PLCγ 2) as a critical signal- activation receptor accumulate centrally (cSMAC).701 Other
ing mediator684 more like B cells than T cells.685 6) In contrast peripheral supramolecular activation complex and centrally
to T and B cells, NK-cell signaling components are generally supramolecular activation complex components and pat-
not required for normal NK-cell development. For example, terns are found depending on whether or not the contacts
NK cells are not deficient in number or maturation state in lead to cytolysis.698,702 With this basic outline of microscopic
PLCγ 2-deficient mice, unlike B cells, which require PLCγ 2 events, molecules required for NKIS formation and granule
for development.684,685 Taken together, these findings high- polarization are being dissected.125,692,703–705
light differences between NK-cell and T- and B-cell signaling.
Illustrating some of the aforementioned issues are stud-
ies on the Vav family of guanine nucleotide exchange fac- Mechanism of Inhibition by Natural
tors, Vav1, Vav2, and Vav3, which are all expressed in mouse Killer–Cell Receptors
NK cells.686 Vav2 and Vav3 are required for FcRγ and DAP12 The now widely accepted view is that the NK-cell inhibitory
signaling, whereas Vav1 is required for DAP10 signal trans- receptor mechanism consists of ligand binding and receptor
duction. Additionally, NK-cell number and differentiation cross-linking followed by ITIM phosphorylation, and
are not apparently affected by Vav deficiencies even when preferential recruitment of SHP-1, and that this mechanism

operates with both structural types of inhibitory receptors. Inhibitory signaling also induces other events. Normal
A typical human KIR molecule has two ITIMs separated by human NK-cell activation by susceptible targets led to Tyr
~24 residues. In contrast, Ly49A has only one ITIM per chain, phosphorylation of Vav1 and its association with c-Cbl,
but Ly49 molecules are normally expressed as homodimers. which was also Tyr phosphorylated.690 In turn, c-Cbl was as-
CD94 has a minimal cytoplasmic tail, but NKG2A has two sociated with a signaling complex consisting of an adaptor
ITIMs. Each ITIM can be phosphorylated upon receptor protein (Crk, either CrkII or CrkL, depending on the cell),
cross-linking (though difficult to visualize) or tyrosine a scaffold protein (p130CAS, 130 kDa Crk-associated sub-
phosphatase inhibition,212,706–710 presumably by a Src family strate), and C3G (Crk SH3 domain-binding guanine-nucle-
tyrosine kinase as in B cells,711 although redundant expres- otide exchange factor). These c-Cbl complexes have the po-
sion and function of these kinases have precluded identifica- tential to signal a wide variety of cellular pathways including
tion of a single kinase required for ITIM phosphorylation Rap1 GTPase activation, actin reorganization, cell adhesion,
in NK cells. ITIM phosphorylation results in recruitment and mitogen-activated protein kinase activation,729 which
and activation of the intracellular tyrosine phosphatase, appear to be critical for NK killing. By contrast, ligand en-
SH2-containing protein tyrosine phosphatase (SHP)-1 (also gagement of either KIR or CD94/NKG2A blocked the for-
known as hematopoietic cell phosphatase (HCP), protein mation of Cbl-Crk-p130CAS-C3G complexes.690 Moreover,
tyrosine phosphatase 1C (PTPIC), and protein-tyrosine ligand engagement of either KIR or CD94/NKG2A induced
phosphatase, nonreceptor-type, 6(PTPN6)).212,706–708,712 Two the association of the tyrosine kinase c-Abl with Crk, and
ITIMs are required for sequential binding of the two SH2 Crk phosphorylation that was shown to be required for ac-
domains in SHP-1.710,713 The NK-cell MHC-specific inhibi- tive dissociation of Crk from Cbl and for inhibition. Thus,
tory receptors preferentially recruit SHP-1 rather than SH2- these studies showed that inhibitory signaling ironically
containing inositol polyphosphate 5-phosphatase (SHIP), involved Tyr phosphorylation and disruption of macro-
though they may occasionally bind SHP-2.706,714,715 Inhibitory molecular signaling complexes, which in turn, suggest that
receptor engagement results in early KIR and SHP-1 recruit- NK-cell activation (and inhibition) may be subject to more
ment to areas surrounded by LFA-1.716 Moreover, early events qualitative controls than previously recognized.
in NK-cell activation are blocked by inhibitory receptors, Another possible explanation for inhibitory signaling is
such as formation of the NKIS.716–718 On the other hand, KIR related to the inhibitory influence that c-Cbl has on NK-cell
phosphorylation occurs within clusters at the NKIS, which activation689 and other immune cells, such as T and B cells.730
may serve to focus inhibition on downstream targets,719 such Although Cbl family molecules exhibit E3 ubiquitin ligase
as Vav1, which is recruited into the mature synapse during activity, and may exert their negative effects on lymphocyte
NKG2D/DAP10 signaling.704 Indeed, the downstream target signaling by enhancing degradation of signaling compo-
of the ITIM-recruited tyrosine phosphatase appears to be nents,731 there are as yet little data supporting this mecha-
Vav1 in human NK cells activated by targets.688 Thus, the nistic role for Cbl in NK-cell activation.689,690 Nonetheless,
general consensus has long been that the inhibitory recep- current information suggests additional aspects of NK-cell
tors recruit tyrosine phosphatases that dephosphorylate inhibitory receptor signaling are yet to be discovered.
molecules in the activation receptor cascade, such as Vav1. Finally, Ly49 receptors on NK cells can acquire their
On the other hand, related ITIM-containing inhibitory re- cognate MHC-I ligands from surrounding cells, resulting
ceptors bind other effector molecules, such as Csk by LAIR-1 in display of both molecules on the NK cells.732,733 Broadly
and LILRB1,423,720 suppressor of cytokine signaling 3 (SOCS3) speaking, this effect is probably related to trogocytosis, a
by Siglecs,721 and SHIP by KLRG1722 and FcγRIIB.723 Even a poorly understood phenomenon of intercellular transfer of
KIR ITIM may recruit other molecules such as β-arrestin724 surface molecules.734 How much this contributes to cis ef-
and PI3K.725 Morever, a SHIP-deficient mouse demonstrates fects on inhibitory receptor function and its physiological
abnormalities in NK-cell function and Ly49 receptor expres- relevance require further evaluation.
sion,726 even though Ly49 receptors recruit SHP-1. The func-
tional significance of these signaling effector molecules in
NK-cell inhibition and the specificity for ITIM recruitment of NATURAL KILLER–CELL TOLERANCE
selective molecules are poorly understood and is potentially NK cells display potent effector functions that must be con-
related to the complexity of inhibitory receptor signaling. trolled to prevent inadvertent damage to normal tissues (ie,
Emerging data suggest that inhibitory receptor signaling NK cells must demonstrate tolerance to self). Although the
may not be as simple as phosphatase recruitment to dephos- missing-self hypothesis provides a rationale and starting
phorylate proximal molecules in the activation pathway that point for dissection of NK-cell tolerance, it has a number of
then prevents all downstream signaling events. For example, caveats that need to be considered before delving into cur-
inhibitory receptors efficiently blocked granule polarization rent understanding of this aspect of NK-cell biology.
but inefficiently prevented degranulation.727 During ongoing
NK-cell activation, KIR2DL2 engagement by photoactivated
HLA-C ligand disrupted the NKIS but had little effect on on- Influence of Host Major Histocompatibility
going calcium flux.728 Thus, some but not all of the signaling Complex–I Environment
events downstream of NK-cell activation receptors may be The missing-self hypothesis implies that there should be
effectively regulated by the inhibitory receptors, directing fu- overt NK-cell autoreactivity in MHC-I– deficient hosts, but
ture studies to selected aspects of NK-cell signal transduction. this was not observed in humans or mice.179–181,735–738 Instead,

NK cells from MHC-I– deficient mice demonstrate poor cells from MHC-I– deficient mice were functionally defec-
killing of MHC-I– deficient targets or rejection of MHC-I– tive in triggering by immobilized antiactivation receptor
deficient BM, even though they appear normal in number, antibodies.168 Conversely, in wild-type mice, functional
tissue distribution, and expression of activation receptors. competence correlated with expression of a Ly49 inhibitory
Another very difficult issue is related to hybrid resistance receptor for self-MHC-I.168 CD94/NKG2 receptors appeared
whereby host NK cells in an F1 hybrid animal can reject BM not to be relevant,168 as confirmed by studies of a CD94-
grafts from either inbred parent.739,740 It was difficult to un- deficient mouse.395 Particularly informative were studies
derstand how NK cells discriminate between cells express- with a single chain trimer MHC-I molecule, consisting of
ing the full complement of self-MHC alleles versus those antigenic peptide-linker-β2m-linker-H2K b as a single poly-
otherwise normal cells expressing only some self-MHC peptide that binds only Ly49C.168 In Tg mice expressing only
molecules. NK-cell–mediated rejection of MHC-I– sufficient this MHC-I molecule, only Ly49C + NK cells were function-
BM grafts also depends on which host MHC-I allele is pres- ally competent. Although crystallographic studies revealed
ent.741,742 Thus, NK cells are regulated by host MHC environ- two potential sites on MHC-I that could be recognized by
ment, not just by the target cell in effector responses. Ly49 receptors, studies of Tg mice bearing mutant MHC
The MHC-specific inhibitory receptors were likely to be in- molecules indicate that the Ly49 receptors recognize the
volved in self-tolerance because they explain the influence of same site on their MHC ligands (site 2) for both education
MHC expression on NK-cell effector functions against target and effector inhibition.749 Thus, NK cells use their MHC-
cells. However, how these receptors are related to self-toler- specific receptors interacting with self-MHC-I to become
ance in vivo was a challenging issue. Several hypotheses were competent to be triggered through their activation receptors
proposed. An individual NK cell can simultaneously express by (Fig. 17.6), an education process termed “licensing.”
multiple inhibitory receptors but some may not recognize self- Despite initial confusion over the term licensing,750 most
MHC. The “at least one receptor” model suggests tolerance is investigators now concur that licensing or education by self-
achieved as long as each NK cell expresses at least one receptor MHC – specific receptors results in fuctionally competent
with self-MHC specificity.389,743 There are changes in the rep- NK cells.751 Moreover, human NK cells are also subjected to
ertoire of MHC-I–specific inhibitory receptors, depending on a similar process involving KIR and self-HLA ligands752,753
the MHC haplotype, but these differences are modest, regard- that may also explain clinical studies relating KIR and HLA
less of whether the receptor is self-specific or not.210,248,389,743,744 alleles and disease and even human immunodeficiency virus
As determined by antibody reactivity, the expression level (HIV) control.754,755 Such KIR-HLA relationships frequently
of an NK receptor on an individual NK cell decreases when involve pairs with high affinities and resolution of chronic
the host expresses the MHC ligand for that receptor.208,209,745 infections756 that are difficult to explain when only consider-
Functional analyses of NK cells with “downregulated” recep- ing effector inhibitory function of the KIRs. Thus, licensing
tor expression suggest that such NK cells are more sensitive may be clinically relevant and may be applicable to other
to small changes in MHC ligand expression on the target, as clinical uses of NK cells (see following discussion) because it
explained by the “receptor calibration” model.744,746 However, is a second function of the NK-cell receptors for MHC.
these studies involved in vitro conditions that could alter Licensing leads to appropriate pairing of inhibitory recep-
their intrinsic functional capacities. Thus, several hypotheses tors with self-MHC and strongly suggests that there are two
based on the MHC-specific inhibitory receptors were pro- types of self-tolerant NK cells (see Fig. 17.6).750,757 Regardless
posed to account for NK-cell self-tolerance. of the MHC-I environment, licensed NK cells are tolerant
Also unexplained was the observation that NK cells were because they have inhibitory receptors for self-MHC, the
found in wild-type mice expressing none of the known self- same receptors involved in licensing. Unlicensed NK cells
MHC –specific inhibitory receptors. These cells were func- are also tolerant because they are not functionally compe-
tionally “hyporesponsive,” showing defective antiactivation tent and have no need for inhibition by self-MHC under
receptor cross-linking and killing of NK-sensitive targets,747 steady-state conditions. In hosts heterozygous for MHC
but it was possible that other, yet to be defined receptors were alleles, each MHC allele could potentially license different
playing a role. Another issue was the appropriate pairing of NK-cell populations. This aspect of licensing is relevant to
polymorphic inhibitory receptors with their highly polymor- hybrid resistance because an (A × B)F1 hybrid animal should
phic cognate MHC ligands.260,317 The genes for the receptors have NK cells that are separately licensed on different MHC
and their ligands are located on different chromosomes (ie, alleles.750 F1 hybrid NK cells that were licensed by MHC al-
the receptor and ligand genes segregate independently). Thus, leles from parent A should be inhibited by A alleles but not
there must be mechanisms to provide NK cells with the ap- B alleles, and thus reject BM from parent B. The converse
propriate inhibitory receptors with specificity for self-MHC should also be true. In the F1 animal itself, NK cells are li-
because the appropriate pairs are not inherited together, un- censed by either parental allele so all NK cells should be
like the closely linked Nkpr1-Clr gene pairs.575 inhibited by normal tissues that codominantly express both
MHC alleles. Licensing potentially explains how NK cells
distinguish cells expressing the full complement of MHC-I
The Licensing Hypothesis from those expressing only some alleles.
Studies with target cell-free stimulation and single cell as- Although initial studies on licensing were focused on
says of NK-cell responsiveness revealed insight into NK- only a few receptors and MHC alleles to demonstrate con-
cell tolerance.168,747,748 Freshly explanted, resting murine NK vincing effects, subsequent studies have demonstrated that

FIG. 17.6. Licensing of Natural Killer (NK) Cells by Host Major Histocompatibility Complex (MHC)-I. Depicted is the situation in a mouse
expressing only H2Kb that is a ligand for Ly49C but not Ly49A. Cells expressing Ly49C therefore have a self-specific receptor, whereas those
expressing Ly49A do not. Ly49C engagement by self-MHC-I results in a licensed NK cell that has functional competence to be triggered
through its activation receptors. These cells can be inhibited by self-MHC through the same receptor that conferred licensing. Ly49A+ NK
cells remain self-tolerant because they are unlicensed. For clarity, only one MHC allele and cells expressing one Ly49 receptor are shown.
Activation receptors are also not shown for the target killing depiction. Normally, the host has multiple MHC alleles, each of which could
license different NK cells, and individual NK cells express multiple Ly49s simultaneously. Each NK cell is likely to be licensed separately,
potentially by different MHC alleles, and through different Ly49s. Modified from Elliott and Yokoyama.750
the licensing status of an individual NK cell is dependent on an MHC-sufficient environment results in licensed donor
the Ly49 receptor, MHC allele, and the number of expressed NK cells.761,762 The converse was also observed (ie, loss of
self-MHC – specific Ly49 receptors.219,758 For example, dif- function when wild-type NK cells were transferred into an
ferent MHC alleles appear to be more potent in licensing MHC-deficient environment). These studies add to the dis-
Ly49A+ NK cells, in a hierarchical manner.219 The same hi- cussion of whether an NK cell interacts with self-MHC in
erarchy of MHC alleles was seen with Ly49A downregula- trans (on another cell) on in cis (on the same NK cell).763
tion and inhibition of Ly49A+ NK cells in effector responses. What is difficult to exclude is the possibility that NK cells
An MHC allele was more potent at inhibition as compared may take up MHC-I from other cells via their MHC-specific
to licensing, providing a margin of safety against NK-cell receptors,732,733 but the transfer studies indicate that the NK
autoreactivity.219,759 Interestingly, MHC heterozygosity was cell itself does not need to synthesize MHC-I. Thus, licens-
comparable to homozygosity with capacity of an MHC al- ing may be a dynamic process in which NK cells constantly
lele to license.219 Studies of mice bearing one, two, or more test the same site on their MHC-I ligands to simultaneously
MHC alleles also show similar effects.758 These results have maintain functional competency and effector inhibition.
prompted a refinement of the licensing hypothesis (ie, the Although licensing is a positive outcome (ie, the acquisi-
“tuning” or “rheostat” model) whereby the strength of in- tion of functional competence), it requires the ITIM of the
hibitory receptor engagement by self-MHC determines the self-specific Ly49 receptor to deliver signals that ultimately
degree to which an NK cell is capable of being activated.750,760 result in a licensed NK cell168 (see Fig. 17.6). The nature of
Where and with whom licensing occurs is currently un- such signals is not yet known, and at least two qualitatively
clear. While original studies suggested that licensing occurs different models are being considered. The MHC-specific
during development in the BM,168 adoptive transfer of other- receptor could confer licensing in a positive way by delivery
wise unlicensed NK cells from an MHC-deficient host into of signals that directly induce a differentiation process and

license NK cells, akin to a stimulatory receptor and also requires an intact microenvironment because BM ablation
known as the “arming” hypothesis764 In the second model, or congenital BM defects lead to abnormal NK cells.776–780
also known as the “disarming” hypothesis,764 the MHC- Indeed, certain aspects of NK-cell development are regu-
specific receptors decrease the effect of a putative, as yet lated by direct interactions between developing NK cells
uncharacterized, self-specific activation receptor. If unim- and stromal elements, such as interactions between mem-
peded by self-MHC-I, NK cells become hyporesponsive due brane lymphotoxin-α (LTα) – expressing NK-cell precursors
to overstimulation through the activation receptor. This (NKPs) and LTα –responsive stromal cells that are necessary
model is somewhat akin to T-cell anergy and places the self- for normal development.154,781 In vitro, NK cells can be gen-
MHC – specific receptor in a familiar light as an inhibitory erated from early hematopoietic cells in a cytokine cocktail
receptor. Both models are similar to those previously pro- consisting of stem cell factor (c-kit ligand), IL-7, flt-3 ligand,
posed,765 but engagement of the MHC-specific receptor with and IL-15.782–789 Direct contact with stromal cells appears to
self-MHC is now known to be a required key step168 and is be required for acquisition of Ly49 receptors by developing
common to both models. Also, the models are not mutually NK cells. Moreover, the Tyro3 family of receptors (Tyro3,
exclusive and either could be affected by coreceptors or ad- Mer, Axl) on NK cells and their ligands (Gas6, protein S) on
hesion molecules. Regardless, licensing may be explained by stromal cells are apparently necessary for expression of NK-
either the arming or disarming hypothesis.750 cell receptors and functional differentiation in vitro and in
For the arming hypothesis, it is difficult to reconcile the vivo.790,791 Thus, NK-cell development requires certain cyto-
involvement of the ITIM in the self-MHC specific receptor kines and direct stromal cell contact.
in licensing,168 though recent analysis shows that inhibitory Although the general topic of NK-cell development is cur-
receptor signaling is more complex than previously recog- rently an active area of investigation, and is thus subject to
nized.689,690 Recent studies also suggest that there are few dif- future modifications, several major themes have emerged that
ferences in genes expressed by licensed verus unlicensed NK will be summarized here. The earliest step involves the com-
cells.766 Instead, in licensed NK cells, the activation recep- mitment of hematopoietic stem cells in the BM to the com-
tors appeared to be localized in nanodomains in the plasma mon lymphoid progenitor that can give rise to NK, T, and B
membrane, whereas in unlicensed cells, the activation re- cells, but not to myeloid cell lineages.792,793 Mice deficient in
ceptors were apparently confi ned to the actin meshwork.766 various transcription factors, including Ikaros and PU.1, dis-
Future studies will be needed to validate this correlation and play severe defects in the development of all lymphoid cells
how the self-MHC – specific receptors mediate these changes. including NK cells, while myeloid and erythroid lineages
For the disarming model, studies on mice constitutively are less affected.794–798 Thus, NK-cell development appears to
expressing activation receptor ligands by transgensis or ret- share a common pathway with other lymphocytes, providing
roviral transduction of hematopoietic stem cells have been evidence that NK cells belong to the lymphocyte lineage.
informative.534,538,767,768 In support of the disarming hypoth- Commitment to the NK-cell lineage from the common
esis, continuous engagement of NKG2D or Ly49H resulted lymphoid progenitor (CLP) appears to involve an interme-
in hypofunctional NK cells, akin to anergy. Indeed, m157 diary cell with T- and NK-cell potential (T/NK progenitor),
Tg mice could not resist MCMV infections normally, and perhaps reflecting the close resemblance of T- and NK-cell
the Ly49H + NK-cell subset was selectively hyporesponsive. effector functions. Consistent with this bipotentiality, CD3ε
However, the anergic phenotype was unaffected by expres- and Fc εRI γ Tg mice exhibit selective defects in both NK- and
sion of a self-MHC – specific receptor.767 Although it remains T-cell development.799,800 Moreover, deficiencies in several
possible that receptor-ligand interactions that normally gov- genes, including transcription factors and IL-15,54,801 affect
ern NK-cell self-tolerance were not well represented in these development of NK-cell and certain T-cell subpopulations,
Tg mice, the current data nonetheless do not support the particularly NKT cells and memory T cells. T/NK progeni-
disarming hypothesis as an explanation for licensing and tor populations, including immature thymocytes, can give
suggest that “anergy” may be a distinct tolerance mecha- rise to T and/or NK cells but not to other lineages, depend-
nism for NK cells, perhaps analogous to separable tolerance ing on culture conditions, such as Notch signaling.802–808
mechanisms for T and B cells.769 Interestingly, the transcription factor Bcl11b is required to
In addition to bone marrow rejection, licensed NK cells maintain T-cell fate by apparently repressing NK-cell fate.
may provide primary protection in viral infections and may Specifically, Bcl11b is required for T-cell specification,809–812
enhance CD8 + T-cell responses.770–772 On the other hand, and its deletion, in committed T-cell precursors or even ma-
unlicensed NK cells may gain functional competence by ex- ture T cells, results in loss of T-cell phenotype and instead
posure to cytokines.168 During MCMV infection, unlicensed acquisition of NK cell-like phenotype,809 highlighting the
NK cells (ie, cells without self-MHC – specific receptors) ap- developmental pathways shared by these cell lineages.
pear to control viral replication in both MHC-deficient and NK cells also have a developmental relationship to
-sufficient mice.773–775 Thus, the role of self-tolerant NK cells Lymphoid Tissue inducer (LTi) cells. In addition to pheno-
in immune responses requires additional study. typic similarities as previously discussed for NK-22 cells, LTi
and NK cells also share a relatively unique feature in express-
ing surface LTα, which influences development of lymphoid
NATURAL KILLER–CELL DEVELOPMENT tissues and NK cells, respectively.154,781 Moreover, deficien-
Early evidence indicated that the complete phenotypic and cies of either Ikaros or Id2 result in defects in LTi and NK
functional maturation of NK cells occurs in the BM and cells, though other cells are also affected.813,814 Interestingly,

mice lacking the thymocyte selection-associated high mo- expression of Mac-1 (α Mβ1) and CD43, proliferation
bility group box protein DNA-binding protein lack both markedly decreases unless challenged by pathogens such
LTi and CD122 + NK cells.815 The NK-cell defect is NK cell– as viruses that can stimulate mature NK-cell prolifera-
intrinsic, and Id2 overexpression could not compensate for tion.98,166 NK cells then gain functional activities associated
thymocyte selection-associated high mobility group box with splenic NK cells, indicating that Mac-1 expression
protein (TOX) – deficiency. correlates with a late maturation stage. A selective NK-cell
Until recently, the acquisition of IL-2/15Rβ subunit deficiency in a Tg mouse is manifested by a failure to be-
(CD122) was considered to mark the earliest identifi- come mature, functional Mac-1hi cells,828,829 consistent with
able committed NKP.805,816 However, an earlier progenitor this differentiation step. Finally, while CD27 is expressed
population (pre-NKP) has been identified that has not yet from the very earliest stages of NK-cell development,817 its
expressed CD122.817 Upon adoptive transfer, these lineage- diminished expression within the Mac-1+ subset is associ-
negative (lin−) cKit− Flk2− CD27+ CD244 + IL-7Rα + ated with a more responsive capacity.830 Thus, Mac-1 and
CD122− cells from the BM could give rise to NK cells in CD27 expression are now commonly used to ascertain NK-
the BM, blood, and spleen but no other cell lineages. Thus, cell maturation in the mouse.
IL-15 is not required for commitment to the NK cell lineage, In addition to the transcription factors mentioned pre-
even though it is required for subsequent NK-cell develop- viously, several other transcription factors, including Ets-
mental stages as illustrated by the persistent expression of 1, Id2, IRF-1, MEF, and GATA-3, have been implicated in
CD122.44,53–57 NK-cell development and function.813,831–836 The need for
CD122 expression is regulated by the combined ac- Id2 in NK-cell development can be overcome by deletion
tion of the T-box transcription factors, T-bet (Tbx21) and of the gene for the E protein, E2A.814 IRF-1– deficient mice
Eomesodermin (Eomes) that have redundant function in lack NK cells, and this defect can be overcome by the ad-
NK-cell development.801,818 Specifically, Tbx21− / − Eomes + / − dition of IL-15, suggesting that stromal cells produce IL-15
mice have marked decreases in CD122 expression and NK in an IRF-1– dependent manner.813 Expression of Bcl2 re-
cells, whereas Tbx21− / − Eomes + /+ mice manifest no defect stored the CD8 + T-cell deficiency but not the NK-cell de-
in CD122 expression and more modest defects in NK-cell ficiency in IRF-1– deficient mice.837 IRF-2 – deficient mice
number. Moreover, Eomes targets the promoter of Cd122. manifest a late defect in NK-cell maturation with normal
IL-15 appears to regulate the basic leucine zipper (bZip) constitutive proliferation but increased apoptosis there-
transcription factor E4BP4 (E4 binding protein 4,819 also after.838,839 In contrast to the other transcription factors,
known as nuclear factor IL-3820) in NK-cell development. GATA-3 does not affect NK cell number; rather, it controls
E4bp4 − / − mice display profound NK-cell deficiency.821–823 their IFNγ production and their homing to the liver.836
Gene transduction of E4bp4 − / − hematopoietic stem cells Similarly, CCAAT/enhancer binding protein (C/EBP) γ-
with E4bp4 rescued NK-cell development in vitro even in the deficient NK cells display defective effector functions.840
absence of IL-15, suggesting that E4BP4 acts downstream of The relationships of the transcription factors to each other
IL-15 signaling. E4bp4 − / − mice show no alterations in other are just beginning to be elaborated in terms of regulating
hematopoietic cells except for a defect in CD8a + cell devel- NK-cell development and function.
opment.824 This latter finding may be relevant to phenotypic As for mouse NK cells, human NK-cell developmental
overlap between NK cells and DCs.115–119 Regardless, Eomes stages have been identified by correlating markers on in
affects NK-cell development by promoting CD122 expres- vivo subsets. 841–844 In general, these stages have been re-
sion, whereas E4BP4 affects commitment to the NK-cell lin- capitulated by in vitro development of functional human
eage by apparently acting downstream of IL-15. NK cells from hemapoietic stem cells from umbilical cord
Committed NKPs next differentiate into mature NK blood or differentiated embryonic stem cells.845,846 As in
cells in a series of putative developmental intermediate the mouse, a combination of cytokines and stromal cells
stages that occur in the BM.98,816 Originally defi ned by cor- is necessary for in vitro differentiation.
relating phenotypic markers, including NK-cell receptors, These studies suggest that human NK cells differenti-
integrins, and other molecules with apparently distinct de- ate in secondary lymphoid tissues as revealed by studies of
velopmental stages,825,826 these stages have been confi rmed CD56bright and CD56dim NK cells.110,847 CD56bright NK cells
by studies following serial acquisition of markers during preferentially express CD94/NKG2 receptors (versus KIRs),
NK-cell differentiation in vitro.789 Regardless, these stages are CD16 −, and are better cytokine producers and less ef-
of committed NKPs are undergoing constant refi nement ficient killers than the CD56dim subset that tend to be KIR+
as new rate-limiting steps are being uncovered, such as in and CD16 +.109 A CD34 + CD45RA+ hematopoietic precursor
studies of mice lacking specific transcription factors or as cell (HPC) was identified that expressed the integrin α4β7 at
new markers and tools are developed.827 Nonetheless, a few high levels.847 This population was found at low levels (< 1%
highlights should be noted. of HPCs) in the BM but was markedly enriched in lymph
NK-cell receptors involved in target recognition are ex- node (LN; > 95% of HPCs), and could be differentiated into
pressed at immature stages, with NK1.1 and CD94/NKG2 CD56bright NK cells in vitro. On the other hand, in nonreac-
being among the fi rst, followed by NKp46, and NKG2D tive LNs, NK cells tend to be CD56bright whereas in reactive
then the Ly49s.827 Following Ly49 acquisition, develop- LNs, they tend to be CD56dim, the predominant population
ing NK cells undergo spontaneous proliferation at an im- in peripheral blood.110 CD56bright cells acquire the CD56dim
mature stage.98 Thereafter, as NK cells acquire high-level phenotype when stimulated in vitro with proinflammatory

cytokines, such as IL-12, or with IL-15 in trans in a human- and liver. Interestingly, NK cells are localized to the red pulp
ized mouse model.848 Taken together, these data suggest that of the spleen.849,850 In the liver, they are in the sinusoidal
human NK cells differentiate in secondary lymphoid tissues regions rather than the parenchyma, and few NK cells are
in response to inflammation by first becoming CD56bright present in other solid organs. In viral infections, NK cells
then completing maturation to the CD56dim phenotype. infi ltrate the liver parenchyma in the vicinity of infected
Although the maturation of mouse NK cells in second- foci.849,851 Several chemokines, including MIP1α , have been
ary lymphoid tissues has not been studied carefully, there implicated in NK-cell localization to liver parenchyma dur-
is an abundance of NK cells with an immature phenotype ing immune responses,852 for example, but the NK cell “che-
in organs, such as the liver.98 On the other hand, recent mokine code” needs to be clarified.853 Surprisingly, there are
studies also provide evidence for thymic maturation of a relatively few NK cells in naïve lymph nodes,849,854 and the
subpopulation of mouse NK cells, termed “thymic” NK thymic NK-cell phenotype is overrepresented among the few
cells111 These cells are characteristically CD127+ CD69high resident NK cells,111 although NK cells can be recruited to
Ly49low CD11blow in contrast to conventional resting splenic draining LNs.855,856 However, human NK cells are relatively
NK cells that are CD127− CD69 − Ly49hi CD11bhi. Thymic abundant in LNs where they appear to mature, as discussed
NK cells were enriched in LNs and are absent in GATA-3- previously. Study of human NK-cell responses in solid tis-
deficient and in athymic nude mice, indicating that these sues has been limited for obvious reasons. Nonetheless,
cells require GATA-3 and an intact thymus for their de- NK cells seem best suited for surveying the blood for trans-
velopment. Indeed, thymic NK cells can develop in vivo formed or infected cells and pathogens during acute im-
and in vitro from double negative (CD4 − CD8 −) 1 (DNI) mune responses but can be recruited to pathologic sites as
subsets of immature thymocytes.789 Interestingly, mouse needed.
thymic NK cells bear some phenotypic markers that tend As early innate immune responders, NK cells have the ca-
to be found on human CD56bright NK cells.111 The contri- pacity to shape the adaptive immune response. For example,
bution of secondary lymphoid tissues to NK-cell develop- early NK-cell control of viral infection essentially limits an-
ment, the role of thymic NK cells in immune responses, tigen load and presentation by DCs thereby limiting CD4 +
and the relationship of these mouse and human NK-cell T-cell responses.857 By contrast, early control also reduces
subsets remain to be clarified. activation of plasmacytoid DCs, thereby lessening the det-
rimental effects of type I IFNs, preserves the conventional
DC compartment, and accelerates CD8 + T-cell responses.858
ROLE OF NATURAL KILLER CELLS IN Early IFNγ production by NK cells can prime and polarize
IMMUNE RESPONSES T-helper responses.855,859–861 In leishmania infections, later
Mature NK cells in the periphery are involved in rapid in- production of IL-10 by NK cells has an inhibitory effect on
nate defense. However, they constitute only a small popula- T cells.862 In addition, NK-cell lysis of targets also enhances
tion of cells (about 2.5% of splenic leukocytes in C57BL/6 T-cell responses.863 Thus, NK cells have the capacity to en-
mice). How can this small population quickly respond with hance, suppress, and polarize adaptive immune responses.
enough of a critical mass to effect significant innate defense?
One mechanism involves the expression of multiple activa-
tion receptors by individual NK cells.449 By contrast to clon- Natural Killer Cells and Tumor Surveillance
ally distributed TCRs endowing the individual T cell with An abundant early literature supported a role for NK cells
the ability to respond only to one antigen, an individual NK in resisting tumor growth and metastasis.3 Most prior work
cell appears capable of responding to multiple activation utilized experimental protocols involving adoptive trans-
receptor ligands. Furthermore, the naïve T-cell population fer of tumor cells into mice where NK cells may eliminate
contains only rare cells with a TCR for the relevant antigen, > 90% of tumor cells within the first 24 hours.864 Several
whereas large percentages of the NK-cell population express long-term assays of in vivo tumor clearance are available,
any given activation receptor in an overlapping fashion. This such as survival or lesion size, but T-cell responses need
multiple activation receptor expression on sizeable subpop- to be excluded even in syngeneic hosts due to the possibil-
ulations would allow a substantial number of NK cells to ity of tumor-specific peptides. Clearance of intravenously
quickly respond to a given specific insult. Another mecha- administered radiolabeled tumor cells can be measured
nism is related to their constitutive expression of cytokine with radioactivity of the lung as an index of tumor bur-
receptors that permit many NK cells to be stimulated by den. Because this lung clearance assay can be performed
proinflammatory cytokines produced early in the course as early as 4 hours after tumor inoculation, it is relatively
of an immune response. Finally, NK cells appear poised confined to innate NK-cell responses in the unimmunized
to respond rapidly. They constitutively express mRNA for host.15 In these assays, NK-deficient mice are unable to clear
effector molecules, such as granzymes and cytokines, but adoptively transferred tumors.828
no protein due to translational control, and can rapidly pro- On the other hand, few studies are available on control
duce these molecules upon stimulation.29 Thus, large num- of primary tumor formation by NK cells (ie, tumor surveil-
bers of NK cells can rapidly respond to a particular stimulus lance).865 Probably the most systematic studies in a single
through their activation or cytokine receptors. experimental model involve fibrosarcoma development
Where do NK cells respond? In the mouse, mature con- after subcutaneous methylcholanthrene challenge.866 −868
ventional NK cells are found primarily in the spleen, blood, In brief, NK cells collaborate with NKT cells in preventing

fibrosarcomas through mechanisms involving perforin, of T cells,888 due to early NK-cell production of IFNγ.889
TRAIL, and IFNγ. There is also evidence for involvement of However, NK cells do not appear to respond directly to
NKG2D86 and its ligands with respect to the immunoediting Listeria. Rather, macrophages produce IL-12 that then stim-
of tumors by NK cells.525,542 ulate NK-cell secretion of IFNγ and infection control.164,890
In humans, epidemiological data indicates a strong cor- Furthermore, TNFα can synergize with IL-12 to induce
relation between high cytotoxic activity of peripheral blood NK-cell production of IFNγ, whereas IL-10 is antagonistic.164
lymphocytes and reduced cancer risk.869 In addition, ongo- The increased susceptibility of mice lacking IL-12 receptor,
ing studies of BM transplantation for leukemia demonstrate IFNγ, or the IFNγ receptor signaling pathway891–894 are con-
a relationship between KIR, HLA alleles, and reduced risk sistent with macrophage production of IL-12 that stimulates
of relapse from leukemia in some circumstances.870,871 Thus, NK cells to secrete IFNγ in listeriosis.
current data support the initial concept that NK cells are While a similar IL-12-IFNγ pathway is also operational in
involved in tumor immunosurveillance. MCMV infections,886,895,896 IL-18 also contributes somewhat
to NK cell control of MCMV.897 However, IL-12 appears to
be more critical than IL-18 because uniform lethality was
Natural Killer Cells in Host Defense observed in IL-12p35–/–mice challenged with MCMV while
Against Pathogens all IL-18–/– mice survive.897 On the other hand, neutraliza-
While NK cells reportedly respond to a wide variety of mi- tion of IL-18 is a common feature of orthopox-viruses.898 For
croorganisms including viruses, bacteria, parasites, and example, ectromelia virus (mousepox) contains an ORF for
fungi,872,873 the role of NK cells in infections is probably an IL-18 binding protein (IL18BP) that effectively neutral-
best illustrated by human patients with selective NK-cell izes the effects of IL-18 on NK cells.899 In addition to aug-
deficiencies.874,875 Although the molecular basis for most mented IFNγ production, IL-18 enhances perforin-depen-
human NK-cell deficiencies is unknown, several points can dent cytotoxicity.900–902
be gleaned. 1) NK-cell deficiency is associated with a pro- Importantly, not all viral infections are controlled by NK
pensity for severe or recurrent virus infections, particularly cells. For example, NK-cell depletion has little effect on lym-
herpes viruses. 2) Difficulty with tumors is not a common phocytic choriomeningitis virus (LCMV) infections.883,896,903
feature, except for virus-related lesions. 3) The disorder is Yet, during infections, even with LCMV, cytotoxicity of NK
rare, perhaps because patients succumb to overwhelming in- cells is enhanced and proliferation ensues. These events
fection before the syndrome is recognized. 4) Defective NK constitute systemic effects directly or indirectly mediated
cells can be found in other genetic and acquired immuno- by cytokines, such as IL-12, IL-18, and IFNα / β.904 However,
deficiency disorders that affect other immune components. IFNγ. production is not seen in LCMV infections.895,905 This
For example, NK cell activity is significantly diminished in apparent paradox is due to an inhibitory effect of IFNα / β on
AIDS.876,877 NK-cell infection with herpesvirus 6 induces de IL-12− dependent IFNγ. production.905 Inhibition by IFNα / β
novo expression of CD4 rendering susceptibility to HIV-1 is mediated through the STAT1 signaling pathway. In the
infection,878 perhaps accounting, in part, for increased sus- absence of STAT1, IL-12 responsiveness is restored and
ceptibility of patients with acquired immunodeficiency syn- IFNα / β induces IFNγ production. These studies indicate
drome to opportunistic infections, such as severe CMV.879 that the NK-cell cytokine response to infection varies with
Immature NK-cell number and function in the developing the pathogen even though many responses may appear to
fetus may be clinically relevant to the classic “TORCH” syn- be similar.
drome, birth defects associated with maternal toxoplasma, A challenging issue is the role of IL-15 in NK cell re-
rubella, CMV, and herpes virus infections.880 Thus, NK cells sponses in vivo906 because IL-15 and IL-15Rα-deficient mice
appear to be especially important in controlling infections, lack NK cells.53,54 Nevertheless, NK cells can be stimulated
especially from herpes viruses. during infection by IL-15 and control viral replication dur-
In mouse models, detailed evaluation of NK-cell re- ing in vitro cultures.906–910 Moreover, IL-15 can provide
sponses against Listeria and viral infections have been es- protection to herpes simplex viral infections.910 IFNα / β can
pecially revealing.881,882 In vivo antibody depletion of NK stimulate IL-15 production that can drive NK-cell prolif-
cells results in marked viral replication in internal organs eration.911 Interestingly, however, NK-cell proliferation can
(spleen, liver), and lethality with MCMV, vaccinia virus, occur in an IFNα / β - or IL-15 −independent manner.912,913
or mouse hepatitis virus.883–885 A similar phenotype was NK-cell responses to cytokines are also regulated by
observed in Tg mice lacking NK cells.886,887 Interestingly, if other innate lymphocytes. For example, in TCRδ − / − mice,
depleting antibody was given to wild-type mice later in the Listeria infection is enhanced compared to TCRβ − / − mice
infection, there was no untoward effect.884 Thus, NK cells and is associated with diminished production of IFN γ by NK
are significant in early, innate immunity to infections. cells, suggesting that γδT cells regulate NK-cell responses.914
Similarly, NKT cells can regulate NK cells because adminis-
Natural Killer–Cell Cytokine Responses and tration of α-galactosylceramide, a potent ligand for the TCR
Production during Infection on NKT cells, results in nearly concomitant activation of
During infection, NK cells can respond to several different NK cells79 that may be exploited as a potential cancer im-
cytokines resulting in production of other cytokines. In lis- munotherapy.915 The mechanism behind the NK and NKT
teriosis, the classic model for T-cell– dependent resistance, cell cross-talk involves IFNγ, apparently produced initially
scid mice achieve acute control of infection despite absence by an activated macrophage.916 Inasmuch as NKT cells can

recognize glycolipid antigens in mycobacteria,917,918 these parainfluenza virus, suggesting it may be involved in resis-
studies indicate a potential physiologically important mech- tance to these viruses.630 However, this interaction is depen-
anism and therapeutic intervention for NK-cell activation in dent on sialic acid residues that are widely expressed, and
innate control of these organisms.919 the in vivo significance of these findings is difficult to assess
in humans. Nevertheless, susceptibility of NKp46-deficient
mice to influenza infections corroborates these findings.93
Natural Killer–Cell Activation Receptors in Infection The autosomal dominant Cmv1 gene in the NKC is
There are several observations from studies of viral evasion responsible for resistance of certain mouse strains to
indicating the importance of NK-cell activation receptors MCMV596,597,936,937; MCMV-resistant C57BL/6 mice are sus-
in infection. Because viruses have evolved numerous strat- ceptible when depleted of NK cells.938 Extensive genetic and
egies to downregulate MHC-I molecules on infected cells immunological evidence established that Ly49h is responsible
to avoid MHC-I–restricted cytotoxic T-lymphocytes,186 vi- for genetic resistance to MCMV.449,938–942 A DAP12-signaling
rally infected cells should have enhanced susceptibility to mutant mouse could not resist MCMV,943 consistent with in
NK-cell lysis. However, viruses also encode proteins that vitro signaling studies showing that Ly49H signals through
evade NK cells.920 In many cases, viral interference of NK DAP12.448,939 Thus, Ly49H is an NK-cell activation receptor
cells is related to enhanced function of inhibitory MHC- responsible for genetic resistance to MCMV.
I– specific NK-cell receptors. For example, murine and rat The ligand for Ly49H is encoded by the m157 ORF in
CMV contain ORFs m144 and r144, respectively, that en- MCMV.444,446 m157 and 11 other putative MCMV molecules
code molecules with sequence and structural homology have predicted MHC-I folds, now validated by crystallog-
to MHC-I and enhance in vivo virulence presumably by raphy.464,926 Interestingly, in mice lacking adaptive immu-
interacting with as yet unidentified NK-cell inhibitory re- nity, m157 mutant MCMV clones emerge during MCMV
ceptors.921–925 MCMV also encodes m157 that can bind the infection, indicating selection pressure from Ly49H + NK
inhibitory receptor Ly49I in 129 strain mice (see following cells result in escape mutant viruses.944 The reasons for
discussion), as well as other MHC-I−like molecules of un- maintenance of m157 in the MCMV genome are still under
known function.444,446,464,926 HCMV encodes an MHC-I−like investigation, but several observations suggest m157 may
molecule (UL18) that binds LIR1 (ILT2), an Ig-like inhibi- be advantageous to the virus under certain circumstances.
tory receptor on NK cells,405,411,927 though its functional 1) m157 binds to the Ly49I inhibitory receptor in 129
significance is somewhat unclear.928 HCMV also encodes mice.446 2) m157-deletion viruses cause a modest decrease in
a peptide that binds and enhances expression of HLA-E viral titers in mice lacking both Ly49I and Ly49H, suggesting
that in turn binds CD94/NKG2A, a lectin-like NK-cell in- yet another immune evasion role.945 3) A unique aspect of
hibitory receptor.386,929–931 Another example is the selective herpes virus biology is the capacity to become latent. If the
downregulation of MHC-I by HIV-1.932,933 In this case, the virus kills the host during the acute viral replicative phase
virus downregulates HLA-A and B but not HLA-C or E; the during which Ly49H mediates its control, then there will
former HLA molecules tend to be restricting elements for be no latent phase. 4) Host-virus coevolution often results
MHC-I−restricted CTLs, whereas the latter are selectively in attenuated viruses,946 and other MCMV proteins have
recognized by human KIRs and CD94/NKG2A. Therefore, positive effects on host responses.947 Indeed, as one aspect of
viruses have evolved mechanisms to selectively engage in- this coevolution, natural variants of m157 show differential
hibitory receptors that presumably prevent the action of binding activities to a broader range of inhibitory receptors
NK-cell activation receptors. in different mouse strains, including inhibitory Ly49C from
Viruses can also directly block triggering of NK-cell ac- B6 mice.948,949 Thus, studies of m157 reflect the coevolution
tivation receptors. For example, both HCMV and MCMV of the host and virus.
encode multiple proteins that block NKG2D ligand expres- Interestingly, other Ly49 activation receptors also recog-
sion by intracellular retention with functional consequences nize MCMV-encoded proteins. In MA/My mice, the Ly49P
in vitro and in vivo,920 as previously detailed. HIV-1 Nef activation receptor is responsible for resistance in the con-
also blocks NKG2D ligand expression.934 Cowpox and mon- text of H2k.453,950–952 This receptor-ligand interaction is more
keypox viruses use a different strategy; they encode a high- complex than Ly49H-m157 because Ly49P appears to rec-
affinity, soluble antagonist that binds NKG2D and blocks ognize the cell surface combination of H2Dk in association
its ligand recognition.557 HCMV, Epstein-Barr virus, and with m04/gp34 encoded by MCMV. However, there appears
KSHV use yet another strategy by encoding miRNAs that to be another molecule required for Ly49P recognition as
target MICB mRNA.554,555 More generally, KSHV also avoids m04 expression on H2Dk-expressing cells was insufficient for
NK-cell activation through K5 that downregulates expres- recognition; infection with Δm04 MCMV was required.453
sion of ICAM-1 and B7-2, ligands for NK cell coreceptors Recent studies demonstrate that other Ly49 activation recep-
involved in target-induced stimulation.935 Thus, viruses use tors from different mouse strains (BALB/c, NOD/LtJ, PWK/
multiple strategies to specifically thwart NK-cell responses Pas) can also recognize MCMV-infected cells in an m04-
through their activation receptors. H2 allele-dependent manner, providing in vivo control.953
The viral evasion strategies implicate NK-cell activation Taken together, these remarkable studies indicate NK cells
receptors that specifically recognize infected cells; several recognize MCMV in an “MHC-restricted” manner that is
have been identified. Human NKp46 binds hemagglutinin fundamentally different than MHC-restriction as defined
of influenza virus and hemagglutinin-neuraminidase of for T-cell biology.

Loci for resistance to other pathogens, such as ectromelia HCMV.963,964 Moreover, expansion of CD94/NKG2C + NK
virus and herpes simplex virus, have also been genetically cells also occurs in the setting of acute Chikungunya virus
mapped to the NKC.954,955 These loci are termed Rmp1 and infection.965 In vitro studies revealed that human NK cells
Rhs1, respectively, and C57BL/6 mice are resistant. Rmp1 can respond to HCMV-infected fibroblasts by producing cy-
was initially mapped using DBA/2 mice that lack a func- tokines and degranulating. These responses were enhanced
tional Cd94 gene.394,954 Recent studies of a CD94-deficient by IL-12 and type I IFNs, and there was evidence for stimu-
mouse show ectromelia virus susceptibility, suggesting lation through NKp46 and DNAM-1. However, there was
that Rmp1 is encoded by Cd94.396 CD94/NKG2E appears no apparent involvement of the CD94/NKG2C activation
to recognize ectromelia virus − infected cells via Qa-1 and receptor,966 despite expansion of CD94/NKG2C + NK cells
synergizing with NKG2D. Thus, NKC-encoded NK-cell with prolonged culture.388 Thus, the factors driving human
activation receptors mediate NK-cell – dependent resistance NK-cell expansion and control of viral infection are not
to viruses. completely understood.
In addition to target killing, NK-cell activation recep-
tors also trigger cytokine production in vitro. For ex-
ample, Ly49H + NK cells can be selectively activated to MEMORY-LIKE NATURAL KILLER RESPONSES
produce IFNγ and chemokines within 6 to 8 hours after Although innate immunity is typified by the absence of
coculture in vitro with MCMV-infected macrophages or memory responses, several recent studies suggest that NK
m157 transfectants,444,956 and m157 itself is expressed soon cells have memory-like responses, more typical of adaptive
after infection.957 However, during early MCMV infection immunity, apparently blurring the distinction between in-
in vivo, IFNγ production was not confined to the Ly49H + nate and adaptive immune responses.751 In classic contact
NK-cell subset in C57BL/6 mice, indicating early relative hypersensitivity (CHS) assays in vivo, NK cells appear
“nonspecific” stimulation of NK cells.166 Moreover, infection to mediate hapten-specific responses in a T- and B-cell–
stimulates NK-cell proliferation958,959 but early (days 1 to 2 independent manner.967,968 Adoptive transfer of Ly49C/I +
postinfection) in vivo NK-cell proliferation was nonselective liver NK cells from sensitized C57BL/6 mice transferred
with respect to Ly49H expression. This proliferation resem- CHS. In addition, hapten-specificity was long lived, as NK-
bled the cytokine-driven “bystander proliferation” observed dependent responses lasted at least 4 weeks. Subsequent
in T cells in response to viral infections or stimulation with studies indicated that exposure to virus and viral particles,
IFNα / β,960 suggesting that the initial phase of viral-induced specifically influenza virus, vesicular stomatitis virus, HIV-
NK-cell proliferation represents a nonspecific response to 1, and MCMV, can also induce virus-specific memory-like
proinflammatory cytokines (IL-12, IFNα / β) and prolifera- NK cells.969,970 Upon adoptive transfer, NK cells can then
tive cytokines such as IL-15.912 The nonselective proliferation protect against lethal challenge with the sensitizing virus.
phase was followed by a period of preferential proliferation While MCMV protection apparently occurred with sen-
of Ly49H + NK cells peaking at days 4 to 6 of MCMV infec- sitized splenic NK cells, protection from other viral chal-
tion.166 The selective phase of NK-cell proliferation reflected lenges required liver NK cells expressing CXCR6. Thus,
the augmentation of pro-proliferative cytokine stimulation emerging studies suggest that NK cells can display mem-
by Ly49H signaling mediated via DAP12.59 Other activation ory-like responses.
receptors also trigger specific proliferation in MCMV infec- The role and basis for NK-cell memory-like responses will
tion, such as Ly49L+ NK cells in BALB/c mice.953 However, require detailed investigation. It is not yet clear if this aspect
in infection with vaccinia virus, which lacks a ligand for of NK-cell function is relevant to “vaccination” and how
Ly49H, the initial phase of nonspecific NK-cell prolifera- much it contributes to immune protection when immune
tion was found but the later specific proliferation of Ly49H + T and B cells are present. It is unclear how NK cells can dem-
NK cells was absent.166 Thus, initial virus-specific NK-cell onstrate “antigen” specificity in the absence of the somatic
responses may be masked by generic cytokine responses and rearrangement mechanism underlying T- and B-cell anti-
only later become detectable. gen receptor gene recombination because NK memory-like
In humans, how NK cells participate in viral infections responses were seen in Rag2– / – mice.967 Moreover, cytokine
has been challenging to understand. However, an unprec- stimulation alone, without “antigen” stimulation, can in-
edented opportunity to study human NK-cell responses duce NK cells to have more robust responses later.971 Finally,
during an epidemic of Puumala hantavirus revealed rapid there appear to be significant differences in the inflamma-
and sustained expansion of NK cells.961 Most cells expressed tory characteristics and histological features of CHS related
CD94/NKG2C and a self-HLA– specific KIR, consistent with to NK-cell responses,972 suggesting that there may be aspects
expansion of licensed NK cells. Interestingly, expansion of to be considered that differ from conventional concepts of
CD94/NKG2C + NK cells occurs in HCMV+ individuals but T-cell–mediated CHS.968 Thus, although mechanistic details
not with Epstein-Barr virus or herpes simplex virus infec- should be forthcoming, this topic has generated intense in-
tions,962 though this could not explain the finding in the han- terest as indicated by review articles which now outnumber
tavirus-infected individuals. Regardless, children infected the primary literature by about three to one, because the
with HCMV show a similar expansion of CD94/NKG2C + concept of NK-cell memory breaks traditional views on in-
NK cells; in one case of a T-cell– deficient infant, the ex- nate and adaptive immunity, and potentially opens the door
panded NK-cell population was associated with resolution for examination of related processes in other innate immune
of infection, suggesting that human NK cells can control components.

NATURAL KILLER CELL AND DENDRITIC migration to the paracortex where CD4 T-cell activation
CELL INTERACTIONS occurred, indicating dynamic interactions in the LN not
Given that both NK cells and DCs, when separately stud- previously appreciated. In humans, DCs in LNs may be in-
ied, are critical early responders in host immune defense, it volved in differentiation of CD56bright NK cells into CD56dim
was not surprising that these cells could communicate in a NK cells.110
bidirectional manner.973 Since the initial description of this
“cross-talk,” innumerable studies have been published indi- NATURAL KILLER CELLS AND MATERNAL-FETAL
cating that NK cells can respond to signals derived from the INTERACTIONS
various DC subsets and vice versa, in a variety of different One enigmatic area of continuing interest concerns NK cells
immune responses in mice and humans. in maternal-fetal interactions.993–995 Several observations are
Data support both cell contact-independent and -depen- related: 1) Initially described as granulated metrial gland
dent pathways. The contact-independent pathways are per- cells, maternal uterine NK (uNK) cells accumulate in the
haps better understood. For example, a variety of stimuli, uterus, near the fetal trophoblast layer, in all mammalian
such as certain TLR ligands,26,974,975 can activate DCs to pro- species examined thus far. 2) At the time of implantation,
duce cytokines. Indeed, activated DCs can produce cyto- uNK cells are the most abundant leukocyte present. 3) In
kines, including IFNα / β, IL-12, IL-15, IL-18, and apparently humans, uNK cells closely resemble CD56bright subpopula-
IL-2,26,975–979 that can stimulate NK-cell production of other tion, but there are phenotypic differences, including ex-
cytokines, such as IFNγ. Moreover, these cytokines enhance pression of activation markers, and different repertoire
NK-cell cytotoxicity by increasing protein expression of per- of MHC-specific receptors. 4) In some mice lacking NK
forin and granzymes29 as well as perforin-independent cyto- cells, reproductive defects have been described.996,997 5)
toxic pathways. The cytokines can also enhance responses Trophoblasts can stimulate mouse uNK cells.998,999 6) There
mediated by ITAM-signaling chain–associated receptors.59 are significant differences in the anatomy of the maternal-
The in vivo relevance of TLR-induced, DC-produced cyto- fetal interface that may limit extrapolation of experimental
kines is evident in studies showing that TLR9-dependent results from one species to another,994 although recent stud-
activation of plasmacytoid DCs in MCMV infections is ies suggest that rodent uNK cells may interact with tropho-
critical for appropriate Ly49H-dependent NK-cell control of blast tissues in a manner similar to human uNK cells.1000,1001
infection.974 Thus, uNK cells may provide important clues to under-
Direct NK-DC cell-cell contact can also enhance NK- standing maternal-fetal interactions.
cell activation.973 Several molecules have been implicated Interestingly, counterintuitive insight has come from epi-
in this process, including cytokine-induced DC expression demiological data regarding KIR and HLA genotype asso-
of ligands for human or mouse NKG2D.980,981 Human DCs ciations with reproductive failure.1002,1003 The role for HLA-G
can also activate resting NK cells via the NKp30 receptor.982 in these disorders is unclear, though early studies suggested
Other studies indicate that mouse NK cells can be activated that human trophoblasts express the nonclassical MHC-I
by DCs in a TREM-2– dependent manner through the DAP12 molecule, HLA-G,1004 but not other MHC-I or -II molecules;
signaling pathway.983 DCs can also present IL-15 in trans HLA-G may protect the fetus from uNK cell attack.1005
that promotes NK-cell development and primes NK cells to However, HLA-G is relatively nonpolymorphic, so this may
respond to bacterial and viral pathogens.31,984 This process not be able to explain alloantigenicity of the fetus. By con-
may also be relevant to human NK-cell differentiation.848 trast, recent studies, using newly available staining reagents,
On the other hand, NK-DC interactions can induce DC indicate that trophoblast cells predominantly express HLA-
maturation.985 For example, MHC-I– deficient targets can C.1003 Furthermore, protection from reproductive failure
activate NK cells that then induce DC maturation.986 NK-DC is associated with maternal KIR haplotypes encoding the
interactions also enhance TH1 polarization apparently KIR2DS1 activation receptor that is specific for HLA-C2,
through NK-cell production of IFNγ.855 During MCMV whereas unopposed inhibitory receptors are associated with
infections, recognition of virus-infected cells by NK cells increased risk. In addition, protection increased when the
is critical for maintenance of DC subsets in the spleen.987 fetus carried HLA-C2. The implication of this work is that
While NK-DC cell contact is also subject to inhibitory ef- uNK cells sense fetal tissues via their activation and inhibi-
fects and even killing of DCs by NK cells,988,989 the capacity tory receptors to guide the dynamic process of placental re-
of NK-DC interactions to induce T-cell responses855,981,986,990 modeling,1006 and suggest a mechanism for selection of KIR
has led to interest in exploiting these interactions for thera- and HLA genotypes in different populations.
peutic vaccines.991 Indeed, uNK cells appear to regulate the placental vascu-
A hallmark of DC maturation is migration to a draining lature. Mice lacking NK cells show defects in placental blood
LN. Interestingly, however, deliberate introduction of DCs vascular remodeling that appears to be partially IFNγ-
into the lymphatics leads to robust recruitment of activated dependent.1007 In addition, human uNK cells produce an-
NK cells.855 Real-time imaging indicates that NK cells con- giogenic factors, particularly early in pregnancy.1008 Recent
tact DCs in the superficial regions of the LNs where they studies in mice and rats recapitulate some of the fi ndings
are less motile and their interactions with DCs are more observed in humans,1000,1001,1003 providing new experimen-
extensive than T-cell motility and contacts with DCs.992 tal approaches to better understand the role of uNK cells in
Leishmania infection led to NK-cell secretion of IFNγ and maternal-fetal tolerance.

NATURAL KILLER CELLS IN HUMAN DISEASE cases of complete remission were reported, but the treat-
Throughout this chapter, we have used human disorders ment required intravenous administration of high doses of
involving NK cells to help illustrate their biology. Whereas IL-2 that has significant toxicity. Improved understanding
NK-cell dysfunction has been associated with a large num- of NK-cell biology and their receptors and specificities has
ber of other diseases, it is not yet clear if they are clearly renewed interest in using NK cells for adoptive immuno-
involved in a pathogenic manner. However, KIR and HLA therapy.1021,1022 In addition, clinical use of IL-15 may allow
genotypes have been linked with an increasing number of for more specific NK-cell activation while sparing the toxic-
human ailments,754 providing strong evidence implicating a ity seen with IL-2.1023 However, these approaches are still in
pathogenetic role for NK cells in these diseases. For example, their infancy and will require controlled studies.
hepatitis C virus can cause a chronic, persistent infection, Finally, NK-cell proliferative disorders are associated with
although some patients resolve the infection. Interestingly, both chronic viral infections (Epstein-Barr virus) and auto-
patients who are homozygous for a KIR gene (KIR2DL3) are immune phenomenon.8,1024,1025 NK-cell malignancies are par-
more likely to clear hepatitis C virus if they are also homo- ticularly aggressive and can appear as lethal midline granu-
zygous for the HLA-C alleles recognized by KIR2DL3.756 In loma due to the propensity of NK lymphomas to present in
papillomavirus infections that can go on to produce cervical the sinus and nasopharyngeal passages where they are espe-
carcinomas, resistance to developing neoplasia is similarly cially destructive.1026 NK cells are also associated with hydroa
associated with genotypes encoding certain KIR-HLA re- vacciniforme and hypersensitivity to mosquito bites.1027
ceptor-ligand pairs.1009 In HIV-1 infections, certain KIR and
HLA alleles are associated with reduced risk of infection and CONCLUSION
slower progression.1010–1013 These associations are even more Detailed studies have moved NK cells from the fringes of
compelling given a recent report indicating that the KIR immunology into the mainstream. Along the way, there
genotypes are associated with amino acid polymorphisms have been innumerable challenges, some from “mistaken
in HIV-1 itself, suggesting that NK cells through their KIRs notions” based on existing paradigms for other immune
selected specific HIV-1 genotypes.1014 These sequences en- cells, because NK cells have novel mechanisms of partici-
hance binding of inhibitory KIRs to HIV-1–infected CD4 + pating in immune defense and tolerance.1028 Nonetheless,
T cells and reduce the antiviral activity of NK cells, pro- as immunologists and physicians gain more appreciation
viding a putative mechanism to explain the association of for the molecular basis for NK-cell biology, the future holds
specific KIR and HLA genotypes with HIV-1 progression. promise for insight into a number of enigmatic and unusual
Thus, while the means by which other KIR-HLA combina- human diseases as well as common disorders, such as hep-
tions modify disease progression is a topic under current atitis C virus and HIV infections, and exploitation of NK
investigation, the emerging data nonetheless demonstrate cells for immunotherapy.
the importance of NK cells and KIR specificities in human Finally, the elucidation of KIR genotypes and their role
disorders and treatments. in infection and pregnancy support a broader view. The net
Studies also suggest that KIR specificities may be clini- outcome of selection for certain KIR and HLA genotypes
cally useful to guide treatment. For example, in BM trans- may reflect a short-term survival advantage in terms of in-
plantation for leukemia, donor NK cells may help provide fection and a longer-term survival advantage as determined
as in anti-leukemia effect against residual malignant cells by reproductive influences. Similar processes must also be
in the recipient if there is a mismatch between the donor ongoing in other mammalian species and support the basis
and recipient HLA alleles.870 While this observation was not for convergent evolution of the NK-cell inhibitory receptors
uniformly observed in other transplant centers,871,1015,1016 im- for MHC-I. Thus, “NK cells are centrally involved in both
provements in determing KIR polymorphisms have guided immunity and reproduction.”1029
donor selection strategies, leading to superior outcomes
in BM transplant therapy for leukemia and neuroblas-
toma.1017,1018 Thus, KIR genotyping is clinically useful. ACKNOWLEDGMENTS
Dysfunctional NK cells are found in patients with FHL I thank members of my laboratory and our collaborators for
who typically demonstrate HLH, as described previously. their continued interest in dissecting the molecular basis
NK-cell dysfunction is also found in systemic onset juvenile for NK-cell activities and tolerance. I thank Megan Cooper,
idiopathic arthritis and macrophage activation syndrome Tony French, and Dorothy K.Sojka for their review of the
associated with that disease.1019 However, it is not yet clear if manuscript. Investigations in the author’s laboratory are
NK dysfunction is a cause or effect of systemic illness. supported by the National Institutes fo Health, the Barnes-
Early studies used LAK cells for adoptive immunother- Jewish Hospital Research Foundation, and the Howard
apy of cancers refractory to conventional therapy.1020 Several Hughes Medical Institute.

CD1d–Restricted Natural Killer

CHAPTER
18 T Cells
Albert Bendelac
INTRODUCTION AND DEFINITION Other semi-invariant TCRs have been identified among
Natural killer T (NKT) cells are cluster of differentiation mouse CD1d-restricted NKT cells, including Vα10-Jα50/
(CD)1d-restricted T cells that use semi-invariant αβ T-cell Vβ8,4 Vα3.2-Jα9/Vβ8, and Vα8/Vβ8,5 but their combined
receptors (TCRs) and reside as long-lived effector cells in frequency is modest compared to the dominant Vα14-Jα18
lymphoid tissues and in the microvasculature of organs TCR. Nevertheless, these cells appear to adopt a similar
such as the lung and the liver. They promptly release a broad NKT effector phenotype, likely because they follow the same
range of cytokines and chemokines in response to microbial thymic developmental pathway.
and self-lipid antigens presented by CD1d glycoproteins, ex-
erting a protective or pathogenic role in infection, inflam-
mation, allergy, and cancer. The NKT lineage is best defined ANTIGENIC LIGANDS OF NATURAL KILLER T CELLS
by expression of Zbtb16, which encodes promyelocytic leu- Cluster of Differentiation 1d
kemia zinc factor (PLZF), a master transcription factor that
CD1d is one of five members of the mammalian family of
directs acquisition of innate-like effector properties during
lipid-presenting, β2-microglobulin–associated, major histo-
thymic development.
compatibility complex (MHC)-like CD1 molecules, and the
The NKT lineage is a member of the larger family of
only one conserved in mouse.6,7 The mouse CD1 locus con-
so-called innate-like lymphocytes, which includes B1
tains cd1d1 encoding a functional surface glycoprotein and
B cells, marginal zone B cells, intraepithelial γδ T-cell
a duplicated gene cd1d2, which is generally poorly expressed
sublineages, the CD8 α α TCRα β population of intesti-
and, in the C57 background, is inactivated by a frameshift
nal lymphocytes, and the MR1-restricted mucosal asso-
mutation preventing surface expression.8 CD1d is found
ciated invariant T cells.1 Like NKT cells, these lineages
constitutively on the cell surface of most antigen-presenting
express semi-invariant TCRs or B-cell receptors encoding
cells, including dendritic cells (DCs), macrophages, and B
specificity for conserved microbial and self-ligands. Their
cells, with particularly high levels on marginal zone B cells.9
“canonical” antigen receptors are sufficient to instruct
Of relevance to NKT-cell development, CD1d is also promi-
lineage differentiation during development in the thymus
nently but transiently displayed on cortical thymocytes. It is
or bone marrow, in a process matching antigen specificity
also expressed by endothelial cells and hepatocytes.
with specialized effector functions and homing to dedi-
CD1d is assembled in the endoplasmic reticulum by as-
cated tissue environments. These stereotypical proper-
sociation with β2-microglobulin before reaching the cell
ties, which are reminiscent of truly innate lymphoid cells,
surface and undergoing extensive rounds of internalization
such as natural killer (NK) cells, represent a distinct host
and recycling between the late endosome/lysosome and the
defense strategy, perhaps corresponding to an early phase
plasma membrane10–12 (Fig. 18.1). The rapid rate of internal-
of evolution of adaptive T- and B-cell immunity.
ization depends upon a tyrosine motif encoded in its cyto-
plasmic tail, which binds adaptor protein (AP)-2 and AP-3
in mouse, and AP-2 in human. This intense recycling ac-
CANONICAL NATURAL KILLER T T-CELL RECEPTORS counts for the steady state accumulation of CD1d in the late
In mice, the vast majority of CD1d-restricted NKT cells endosome/lysosome for mouse or late endosome for human.
express the semi-invariant Vα14-Jα18 TCRα chain paired While a diversity of exogenous and endogenous lipids can
with β chains made of Vβ8, Vβ7, or Vβ2 joined to variable load CD1d in various compartments,13–15 the late endosome/
DβJβ segments, whereas the homologous Vα24-Jα18 chain lysosome environment is optimized for lipid antigen acqui-
associated with Vβ11 is used in humans2 (Table 18.1). The sition due to the presence of efficient glycolipid processing
“canonical” sequence of the TCR alpha chain is entirely enzymes and lipid transfer proteins such as saposins,16–18
encoded in the genome, although alterations due to nucleo- Gm2 activator,16 Niemann-Pick type C2 protein,19 and, in
tide trimming and N additions can be tolerated if they pre- humans, CD1e.20 In this acidic compartment, short or poly-
serve the antigenic specificity of the TCR. These canonical unsaturated lipids loaded at the cell surface are quickly re-
rearrangements arise randomly and at very low frequency moved and replaced by long and saturated lipids that bind
in both fetal and adult life, but massive thymic expansion more stably to CD1d but require lysosomal transfer pro-
post-TCR expression ensures a high frequency of NKT cells teins for loading.21 Unlike MHC class II, CD1d expression
among recent thymic emigrants.3 and its presentation of lipid antigens at the cell surface are
432
Paul_CH18_final.indd 432 9/17/12 6:52 PM

CHAPTER 18 CD1d–RESTRICTED NATURAL KILLER T CELLS | 433
TABLE 18.1 Semi-invariant T-Cell Receptors of Cluster of Differentiation 1d–Restricted Natural Killer T Cells
Human
TRAV10 GTG GTG AGC G
TRAJ18 C GAC AGA GGC TCA ACC
GTG GTG AGC GAC AGA GGC TCA ACC Variable hVβ11
Canonical hVα24
V V S D R G S T
Mouse
TRAV11 GTG GTG GGC G
TRAJ18 TA GAT AGA GGT TCA GCC
Canonical mVα14 GTG GTG GGC GAT AGA GGT TCA GCC Variable mVβ8, mVβ7, mVβ2,
V V G D R G S A
independent of toll-like receptor (TLR) signaling and DC The marine sponge lipid turned out to be closely related
maturation.22 Thus, immature DCs can readily present pu- to microbial lipids found in some gram-negative lipopoly-
rified lipid antigens to NKT cells and become activated upon saccharide (LPS)-negative bacteria. Notably, Sphingomonas,
CD40 engagement by NKT cells. a member of the class of α-proteobacteria and a ubiquitous
bacterium found in terrestrial and marine environments
(including as a bacterial symbiont of sponges), uses alpha-
Lipid Ligands of Natural Killer T cells branched glycuronylceramides as a substitute for LPS in the
bial `-glycosylated Lipid Ligands outer membrane of its cell wall.30 It can infect dendritic cells
The first universal NKT ligand emerged from studies of ma- and macrophages to activate NKT cells.31,32 Other structurally
rine sponge extracts that prolonged survival of mice bearing related α-glycosylated lipids, such as α-galactosyldiacylglyc-
B16 melanoma.23 The active principle of the sponge Agelas erol, found in Borrelia burgdorferi, the agent of Lyme disease,33
mauritianus was an alpha-branched galactosylceramide, and α-glucosyldiacylglycerol, found in Streptococcus pneu-
which after slight modification led to the synthesis of an moniae,34 also activate mouse and human NKT cells. Such
extremely potent variant named KRN7000, commonly re- α-branched glycolipids have not been reported in vertebrates,
ferred to as αGalCer24,25 (Fig. 18.2). Over 95% of mouse suggesting that they represent a pathogen signature that the
and human NKT cells recognize αGalCer, irrespective canonical NKT TCR may have evolved to recognize.
of their variable CDR3 β sequence.26,27 Furthermore, the
mouse CD1d-αGalCer tetramers stain the NKT cells of both Endogenous a-glycosylated Lipid Ligands
human and nonhuman primates,28,29 attesting to the high Vα14 NKT cells are spontaneously autoreactive to CD1d-
degree of conservation of this recognition system. expressing cells.35 For example, NKT hybridomas or clones
Vα14 TCR Self Exogenous

or VLDL lipid
microbial
lipid LDL &
scavenger
receptors Bacteria
AP-2/AP-3
Golgi
LTP? LTP
CD1d
β2m
ER Late endosome/lysosome
FIG. 18.1. Presentation of Lipid Antigens by Cluster of Differentiation (CD)1d. CD1d acquires lipid antigens after biosyn-
thesis in the endoplastic reticulum and Golgi prior to reaching the cell surface. Multiple rounds of AP-2/AP-3 clathrin-
mediated internalization to the late endosome/lysosome and recycling to the plasma membrane allow lipid exchange.
Lysosomal degradation pathways provide lipid antigens for exchange in a process assisted by lysosomal lipid transfer
proteins such as saposins, Gm2 activator, NPC-2, and CD1e. Exogenous lipids bound to lipoproteins enter the cell through
scavenger receptors and low-density lipoprotein receptor. Microbial organisms access the lysosome after phagocytosis.

OH
HO
O
KRN7000 O
HO HN OH
HO O
OH
O
HOOC
HO O
GSL-1 HN
HO
HO O OH
OH
OH
HO O
BbGL-II O
HO O
HO O O
O
OH
O
HO O
Glc-DAG-s2 O
HO
HO O O
O
OH
HO
O
O OH
HO OH OH
iGb3 O HN
HO
O O O
O
FIG. 18.2. Self- and Foreign-Lipid
OH HO Antigens Recognized by Natural
HO OH
Killer T Cells. From top to bottom,
O KRN7000, or reference α-GalCer,
OH
HN synthetic version of the original ma-
βGlcCer HO O
O rine sponge antigen; Sphingomonas
HO
HO OH
GSL-1; Borrelia burgdorferi BbGL-II,
α-galactosyldiacylglycerol; Strepto-
coccus pneumoniae, α-glucosyldi-
H3C CH3 O O O acylglycerol; isoglobotrihexosylce-
Lyso-PC H3C
N
O
P
O O ramide iGb3; β-glucosylceramide;
OH and lyso-phophatidylcholine.
can be activated to secrete modest amounts of cytokines self-ligand recognized by NKT cells led to the identification of
when cultured in the presence of fresh thymocytes, DCs, a trihexosylceramide, iGb3, which is synthesized in the Golgi
and various tumor cell lines.8,35–38 This autoreactivity and as an intermediate in the biosynthetic pathway of iGb4 and is
the agonist ligands involved appear to be central not only also generated in the lysosome after degradation of iGb4 by β-
for NKT-cell thymic development,36 but also for their re- hexosaminidase AB.43 iGb3 is further degraded into LacCer by
sponse to various microbial infections, especially when the α-galactosidase A (Fig. 18.3). Mice lacking β-hexosaminidase
pathogens themselves do not express NKT ligands.31,39 The B lacked 95% of their NKT cells, a defect associated with the
identification of self-ligands is therefore of major relevance impaired ability of their thymocytes to stimulate autoreac-
to key aspects of the biology of NKT cells. tive NKT hybridomas. Conversely, unlike αGalCer, iGb3 is
In the mouse system, recognition of self-ligands by Vα14 recognized with a Vβ7 > Vβ8 > Vβ2 hierarchy of affinities
NKT cells is largely dependent upon CD1d endosomal traffick- that mirrors the relative selection of NKT thymic precur-
ing,40,41 suggesting the recognition of endosomal/lysosomal sors.44 Furthermore, DCs lacking α-galactosidase A exhibited
lipid antigens. This may not be the case for their human coun- greater spontaneous stimulation of NKT cells,45 consistent
terpart, Vα24 NKT cells.42 The search for a lysosomally-loaded with the demonstrated accumulation of iGb3.46–48
(iGb4) GalNacβ1-3Galα1-3Galβ1-4Glcβ1cer
Hexb
(iGb3) Galα1-3Galβ1-4Glcβ1cer
A3galt2 Gla
(lactosylcer) Galβ1-4Glcβ1cer
B4galt6 Galc
(glucosylcer) Glcβ1cer FIG. 18.3. Toll-Like Receptor (TLR)-Mediated Regulation
Ucgc of the Synthesis and Degradation of Natural Killer T Cell
(ceramide) cer Glycosphingolipid Antigens. TLR signaling downregulates
ER/GOLGI LYSOSOME B4galt6 encoding lactosylceramide synthase and Gla encod-
ing α-galactosidase A, but upregulates Ucgc encoding β-
glucosylceramide synthase. These changes result in the net
TLR accumulation of β-glucosylceramide and iGb3.

Among several genes encoding Galα1-3Gal transferases, Lysophosphatidylcholine was also reported to stimulate
a3galt2 is considered more specialized for iGb3 synthesis. As a fraction of human but not mouse NKT cells,54,55 suggest-
a3galt2 null mutant mice did not show NKT cell defects49 ing that upregulation of this ligand in multiple myeloma
and the human homologous gene appeared dysfunctional,50 and other inflammatory conditions might also contribute to
iGb3 may not be the sole endogenous glycosylceramide rec- NKT-cell activation.
ognized by mouse and human NKT cells. Other reports have
suggested that β-galactosylsylceramide51 and β-glucosylce-
ramide52 might also function as natural ligands of NKT cells, Structural Basis of Lipid Recognition by
but these claims have not yet been widely replicated. While Natural Killer T cells
the respective importance of all these candidate ligands re- The lipid-binding pocket of CD1d is particularly well
mains to be further studied in physiological and pathologi- adapted to bind self- and microbial glycosphingolipids, with
cal conditions, it is possible that NKT cells might recognize the acyl chain in the A’ hydrophobic pocket and the sphingo-
multiple endogenous β-glycosylceramides as weak agonists. sine chain in the F’ hydrophobic channel.56–58 The hydroxyl
Importantly, several enzymes involved in the synthesis and groups of the sphingosine emerge from the groove to establish
degradation of β-glycosylceramides appear to be coordinately hydrogen bonds with the α1 helix Arg79 and Asp80, while
regulated in conditions associated with NKT-cell activa- the galactose is stabilized though hydrogen bonds between its
tion45,52,53 (see Fig. 18.3). For example, TLR signaling was shown 2- and 3-hydroxyl groups and Asp153 of the α2 helix. Thus,
to specifically downregulate α-galactosidase A in the degrada- the protruding sugar is solidly anchored in a position parallel
tion pathway,45 leading to increased iGb3.46 In the biosynthetic to the plane of the α-helices, explaining the exquisite stimu-
pathway, TLR signaling upregulated β-glucosylceramide syn- latory properties of several carbohydrate hydroxyl groups.
thase while downregulating lactosylceramide synthase, thus In contrast, the trisaccharide chain of iGb3 protrudes in or-
increasing β-glucosylceramide.52 Collectively, these findings thogonal orientation to the plane of the α-helices, with the
suggest that microbial organisms lacking NKT ligands can proximal glucose forming hydrogen bonds with the α2 helix,
nevertheless activate NKT cells through TLR-mediated accu- in particular Asp153 and Thr156, whereas the position of the
mulation of endogenous glycosphingolipid ligands. ceramide backbone is similar to αGalCer (Fig. 18.4).
A
FIG. 18.4. Natural Killer T (NKT) T-Cell Receptor (TCR)
Recognition of `GalCer and iGb3. A: Conserved footprint
of the mouse NKT TCR on the surface of cluster of differ-
entiation (CD)1d-αGalCer and CD1d-iGb3 (the lipid shown
is αGalCer): orange, CDR3α; cyan, CDR2β; green, CDR3α
and CDR2β. B: Binding mode of αGalCer and iGb3 to CD1d
and TCR. Left, αGalCer (same structure before and after
TCR binding); right, iGb3 before (green sticks) and after
(yellow sticks) TCR binding. Note that the distal sugar
is not represented before TCR binding. Courtesy of Dirk
Zajonc (La Jolla Institute of Allergy and Immunology, San
Diego, CA). B

How could such highly dissimilar structures be rec- DEVELOPMENT AND HOMEOSTASIS OF NATURAL
ognized by the same TCR? In the ternary structure of the KILLER T CELLS
human TCR-CD1d-αGalCer complex, the TCR docked
Development
parallel to and at the extreme end (F’ pocket) of the CD1d
binding groove, enabling a lock-and-key type of interac- The major determinant of NKT cell development is the
tion with the solidly anchored, rigid αGalCer antigen. semi-invariant TCR, which, upon binding to CD1d ligands
The docking was parallel to the long axis of the binding expressed by cortical thymocytes, provides the signals re-
groove, contrasting with the diagonal footprints observed quired for induction of the lineage-specific transcription
for MHC-restricted TCRs.59 The CDR3α loop encoded by factor PLZF. Therefore, insights into the peculiar nature and
the conserved germline Jα18 segment provided a majority of context of TCR engagement and its downstream signaling
contacts, straddling the antigen binding cleft and engaging have considerably advanced our understanding of the mo-
electrostatic interactions with α-GalCer, the α1 and the α2 lecular mechanisms of NKT-cell development.
helices of CD1d. CDR1α interacted solely with αGalCer, and T-Cell Receptor Expression and Positive Selection
CDR2β formed a stretch of interactions with the α1 helix. The use of CD1d-αGalCer tetramers specific for the ca-
Consistent with the great diversity of CDR3β usage, this nonical mVα14-Jα18/hVα24-Jα18 TCRs has revealed a se-
loop contact was limited to a single van der Waals contact quence of selection, expansion, and differentiation events
with Gln150 of the α2 helix. preceding the terminally differentiated NK1.1+ stage.3,63,64
The ternary structure of the TCR-CD1d-iGb3 complex was As illustrated in Figure 18.5, NKT cells originate from
recently elucidated. Unsurprisingly, given the conservation of mainstream thymocyte precursors that transit through the
Jα18 and CD1d in both structures, the overall footprint of the pre-TCRα +TCRβ + stage to reach the CD4 + CD8 + double
TCR on the CD1d-glycolipid surface was unchanged compared positive (DP) stage, where stochastic Vα-Jα rearrangements
with CD1d-αGalCer. Strikingly, however, the TCR appeared to lead to expression of canonical Vα14-Jα18/Vβ8, Vβ7 or Vβ2
simply “squash” the trisaccharide headgroup of iGb3 over the TCRs. These rearrangements involve distal gene segments
α2 helix of CD1d, with the proximal β-linked glucose molded that require prolonged cell survival mediated by Bcl-xL,
into an orientation similar to the α-linked galactose of αGal- whose expression depends on HEB-induced RORγ t.65–67 All
Cer (see Fig. 18.4). Each of the three sugars made stabilizing Vβ families can pair with the Vα14-Jα18 chain, but only
polar and van der Waals interactions with CD1d residues. The the biased set of Vβs confers specificity for the endogenous
proximal and second sugars also contacted the TCR CDR2α ligands. Although Vβ8 predominates among mature NKT
loop, which is not involved in CD1d-αGalCer recognition. The cells, the most enriched Vβ family relative to preselection
importance of the distal sugar is reflected by the lack of detect- frequency is Vβ7, followed by Vβ8 and Vβ2.44 This is consis-
able stimulation by lactosylceramide where the third sugar is tent with the observation that endogenous ligands displayed
absent. These conformational changes imply a highly dynamic on the surface of thymocytes activate the Vα14-Jα18 TCRs
interaction process during the association phase. The energy with a Vβ7 > Vβ8 > Vβ2 hierarchy of affinity, and that in
penalty incurred for binding self-ligands could be overcome mice expressing very low levels of CD1d, NKT cells showed
in specific conditions that strengthen the immune synapse. increased representation of Vβ7.44,68–71 Notably, this hierar-
For example, this may occur during TLR-induced inflamma- chy of affinity parallels that of iGb3 but is different from
tion, which upregulates lymphocyte function–associated an- αGalCer, which favors Vβ8. Thus, the naturally selected
tigen (LFA-1) and intercellular adhesion molecule (ICAM-1) Vβ repertoire seems to precisely reflect the affinities of the
integrin interactions, or during thymic development, which preselected repertoire for endogenous ligands, suggesting
involves homophilic engagement of signaling lymphocytic ac- that negative selection plays little role in shaping the Vβ
tivation molecule (SLAM) family receptors. repertoire. However, this does not imply that NKT cells are
Additional crystallographic studies further illustrated the impervious to negative selection because transgenic over-
difference between self- and foreign-antigen recognition. expression of CD1d decreased the relative frequency of the
The α-glycosyldiacylglycerol antigens of Borrelia burgdor- high affinity Vβ7+ cells and exposure to αGalCer-induced
feri33,60,61 and Streptococcus pneumoniae34 adopted configura- deletion of NKT thymocytes.69
tions relatively similar to αGalCer, although some induced
fit was observed for both CD1d and the glycolipid upon TCR Developmental Stages
binding. In the case of Streptococcus pneumoniae, an un- The earliest signaled NKT precursors are rare CD4 + CD8 +
usual sn2 alkyl chain, vaccenic acid (with an unsaturation DP cells that express the semi-invariant TCR. These cells
at C7), was important for favorable positioning of the glu- become CD4 + single positive (SP) cells expressing high lev-
cose. After insertion of a stretch of hydrophobic aminoacids els of CD24, a marker of immature cortical thymocytes, as
in the CDR3β loop in order to enhance binding to CD1d, well as CD69, a marker of positive selection.64 This is the so-
other self-lipids such as β -GalCer, Galα1-4GlcCer (lactosyl- called stage 0 of NKT development. As CD24highCD4 + cells
ceramide), and phosphatidylinositol could be recognized.62 mature into the next CD24lowCD4 + stage 1, they undergo
Although somewhat contrived, this system revealed a sim- several rounds of cell division.3 Cell cycle may be initiated as
ilar bending of the β -linkage between the proximal sugar early as stage 0, as suggested from the analysis of cyclin D2-
and the ceramide backbone as seen for iGb3, raising the pos- or c-Myc-deficient mice.72,73 The lineage expansion following
sibility that several β -linked self glycolipids might serve as positive selection ensures the high frequency that is critical
autoantigens. for innate immunity. During this phase, a fraction of CD4 +

SLAM/SLAM
EC
DP DP DP
Vα14-Jα18 /
NFAT
CD1d-lipid
Egr2 hi
CD4 CD24hi
trans CD69hi
CD4 Stage 0
PLZF
CD44lo PKC θ
CD24lo
CD4 CD8
CD69lo
CD4 CD62Lhi Bcl10 Stage 1
IL-4 NF-κB
IL-4
CD4
CD44hi
CD4 CD62Llo
CD4 IL-4
Stage 2
DN IFN-γ
IL2Rβlo Tbet
exogenous IL-15
CD44hi
CD62Llo
innate- innate- CD4 CD4
CD4 CD8 like CD4 like CD8
IL-4
DN
Stage 3
DN IFN-γ
IL2Rβhi
emigrant resident
NK1.1+
FIG. 18.5. Development of Natural Killer T (NKT) Cells. Thymic stages of NKT-cell development are aligned with
corresponding stages of cluster of differentiation (CD)4 and CD8 T-cell development. Stage 0 NKT cells cor-
respond to the transitional CD4 cells. Note the cross-talk between NKT cells and CD4 and CD8 T cells, which is
mediated by interleukin-4, and results in the differentiation of “innate-like” CD4 and CD8 T cells in Balb/C mice.
DP, CD4+CD8+ double positive; DN, CD4-CD8- double negative; EC, epithelial cell.
cells downregulates CD4 to become CD4-CD8- double nega- Recent thymic NKT emigrants and their immediate
tive (DN) T cells,64 acquiring a more restricted Th1 cytokine precursors, stage 2 NKT thymocytes, express neuropilin
and chemokine profi le and an enhanced ability to reject tu- 1 (Nrp-1), a transmembrane receptor for vascular endo-
mors.74–76 Dividing cells first activate their interleukin (IL)-4 thelium growth factor and semaphorin family members.82
locus (stage 1) then interferon (IFN)-γ (independently of Nrp-1 represents a convenient marker of recent thymic NKT
both stat6 and stat4) as they upregulate CD44 and down- emigrants, as it remains expressed for a few days and is
regulate CD62L (stage 2).3,63,77 Thus, unlike other αβT cells, downregulated after terminal maturation to stage 3.
postselection NKT lineage cells undergo a sequence of events
that is reminiscent of the antigen-driven activation, expan- Homotypic Thymocyte-Thymocyte Interactions
sion, and effector differentiation of mature T cells.78 The thymic cell types involved in presenting NKT ligands
NKT lineage cells emigrate from the thymus after upregu- have been thoroughly investigated. Unlike MHC class I or
lation of the S1P1 receptor,79 as dividing effector-type cells. class II, CD1d is prominently expressed on cortical thymo-
They represent up to 5% of recent thymic emigrants in the cytes, which is consistent with the ability of cortical thy-
mouse spleen.3,80 Within a couple of days after emigration, mocytes to stimulate NKT hybridomas.36 Bone marrow
they express NK lineage receptors, including the activat- chimera experiments demonstrated that NKT-cell develop-
ing NK1.1 and NKG2D receptors and the inhibitory CD94/ ment required CD1d expression by cortical thymocytes but
NKG2A, Ly49A, C/I, and G2 receptors. Acquisition of this not radioresistant stromal cells.83–85 This is radically differ-
terminal differentiation program is associated with cessation ent from conventional T-cell development that is driven by
of division and with upregulation of IL2Rβ, which is neces- MHC expression on thymic epithelial cells. Transgenic ex-
sary for IL-15 signaling. Intriguingly, a fraction of NKT thy- periments using promoters for Lck, MHC class I, or MHC
mocytes downregulate the S1P1R and remain as permanent class II to redirect CD1d to various cell compartments in
residents in the thymic medulla, where they undergo the same CD1d-deficient hosts suggested that expression on cortical
terminal maturation program.79 These thymic residents ap- thymocytes was necessary and sufficient for lineage devel-
pear to be absent in humans.81 opment.70,86–88 This conclusion was recently confirmed after

conditional deletion of CD1d using Cd4-Cre.89 Mixed chi- lineage-determining consequences. Egr2 directly binds to
mera experiments using CD1d knockout pLck-CD1d trans- the promoters of NKT lineage–specific genes such as Zbtb16,
genic TCR Cα knockout bone marrow (where thymocyte encoding PLZF, and Il2rb, encoding the β chain of the IL-15
development is arrested at the DP stage) as the sole source receptor, and it is required for their induction. This sug-
of CD1d expression demonstrated that expression of CD1d gests a direct connection between the peculiar signaling
solely on cortical thymocytes was sufficient for the full dif- emanating from the TCR synapse and these NKT lineage
ferentiation of NKT cells.87 Intriguingly, however, the transi- checkpoints.100 The sustained elevated Egr levels likely result
tion to the NK1.1 positive stage was partially impaired after from the TCR recognition of agonist ligands36,43 or from the
emigration to peripheral tissues, although it was preserved Slamf-SAP–mediated stabilization of the immune synapse.96
for the cells remaining as residents in the thymus. Terminal As some Zbtb16 messenger ribonucleic acid induction was
differentiation was fully restored if CD1d was re-expressed detected in SAP-deficient NKT precursors, SAP may not be
on MHC class II expressing cells or if the cells were trans- absolutely required for PLZF induction.104
ferred into wild-type recipients.80,87 TCR and Slamf-SAP-Fyn signaling both involve the ca-
nonical NF-κ B pathway through PKCθ and Bcl-10.105,106
Homophilic Signaling Lymphocytic Activation Mice lacking these downstream signaling components
Molecule-f Family Interactions showed partial defects in NKT-cell development,107,108 as fur-
A major pathway that is selectively recruited by the thymo- ther detailed in the following.
cyte-thymocyte interactions involves homophilic binding of Thus, NKT-cell development is tightly dependent on
SLAM family receptors, mainly Slamf1 (SLAM) and Slamf6 the signaling elicited through the TCR and Slamf-SAP-Fyn
(Ly108), which are expressed by cortical thymocytes, but pathways, the specific contributions of which remain to be
not by thymic epithelial cells.90 These partially redundant dissected. Notably, redirecting the expression of MHC class
receptors signal through the adaptor SLAM-associated pro- II proteins on cortical thymocytes through ectopic expres-
tein (SAP) and the kinase Fyn, explaining earlier reports sion of the transcription factor CIITA led to the differentia-
of developmental arrest in SAP- or Fyn-deficient NKT pre- tion of “innate-like” CD4 + thymocytes in a SAP-dependent
cursors.91–95 SAP- and Fyn-deficient NKT thymocytes were manner. These cells resembled stage 2 NKT cells and ex-
blocked at stage 0 and could not be rescued by expression pressed PLZF, reinforcing the notion that homotypic thy-
of Bcl-2 or Bcl-xL. They showed lower induction of CD69,90 mocyte-thymocyte interactions, Slamf-SAP-Fyn signaling,
suggesting defective signaling by the TCR, perhaps due to a and PLZF define a dedicated thymic pathway for the pro-
role of Slamf-SAP-Fyn signaling in stabilizing the immune duction of innate-like effector T cells.109–112 In that context,
synapse.96 Individual ablation of Slamf1 or Slamf6 resulted the reciprocal expression patterns of CD1 and Slamf recep-
in a modest twofold defect in the expansion of the NKT lin- tors by thymocytes and of MHC proteins by epithelial cells
eage between stages 1 and 2. However, in “pseudo-double may well serve the primary purpose of creating different
mutant” mixed chimeras where Slamf1/CD1d double de- niches for different thymic lineages.90
ficient NKT precursors were forced to see their ligands on
Slamf6-deficient thymocytes, a > 10-fold reduction of NKT Cluster of Differentiation 4 and Cluster of Differentiation
cells was observed, demonstrating a requirement of Slamf 8 Coreceptor Expression
receptors at the time of TCR engagement by CD1d ligands.90 NKT cells originate from the same pool of DP precursors as
Notably, the Slamf locus exhibits considerable polymor- MHC-restricted T cells and emerge from thymic selection
phism, which may underlie some of the reported variations as CD4 SP cells expressing the CD4 lineage factor ThPOK/c-
in NKT-cell frequencies in different mouse strains and in Krox in a Gata-3– dependent manner.113,114 A fraction goes
humans. Genetic studies have provided support for this hy- on to downregulate CD4 and acquire the DN phenotype,
pothesis by identifying the lack of expression of Slamf1 by but they still stably express ThPOK, which is essential to
cortical thymocytes in the nonobese diabetic (NOD) strain, downregulate CD8. In humans, some NKT cells can express
which is spontaneously NKT deficient.97 CD8αα homodimers. ThPOK-deficient NKT cells did not
express CD4, and some of them also failed to repress CD8,
Signaling in the Natural Killer T-Cell Microenvironment but otherwise they appeared to develop normally.113 Their
Engagement of the semi-invariant TCR activates the same functional properties, however, have not been fully assessed.
Ras/MAP kinase and calcineurin pathways as reported for CD8α-deficient mice showed a modest but significant bias
MHC-restricted TCRs during positive selection.98,99 Notably, toward the selection of high-affinity Vβ7 TCRs compared
NKT thymocytes show elevated and sustained expression of with littermate controls.83 A similar bias was observed after
the early growth response (Egr) factors 1 and 2 compared anti-CD8 antibody treatment, suggesting a minor role of
with the weak and transient expression observed in MHC- CD8 as a coreceptor for CD1d. Transgenic expression of
restricted thymocytes.100 Egr1 is thought to be downstream CD8α resulted in the disappearance of NKT thymocytes,
of the Ras/MAP kinase pathway, whereas Egr2 is mainly suggesting a role in negative selection,83 a conclusion subse-
induced by the calcineurin/nuclear factor of activated T quently challenged when the transgenic model was found to
cells (NFAT) pathway. Egr1 and Egr2 mediate the survival have impaired Vα-Jα rearrangements.114
of MHC-restricted T-cell precursors after positive selection Thus, while NKT-cell development does not appear to
through induction of Bcl2 and Bcl-xL.101–103 In the case of rely on CD4 or CD8 coreceptors, the induction of ThPOK
NKT cells, however, sustained Egr elevation has specific and CD4 in this lineage may simply reflect the path of high

affinity TCRs whose signaling is coreceptor independent.115 Other Transcription Factors

However, the subsequent CD4 downregulation in up to 50% The induction of Tbet represents an essential step at the
to 70% of NKT cells, despite persistent ThPOK expression, transition between stage 2 and stage 3,125–127 coinciding with
remains to be explained. the relative decrease in PLZF expression. Both Tbet and
Egr2 contribute to Il2rb induction,100 allowing responsive-
Cluster of Differentiation 28-B7 Interactions ness to IL-15. While the pathways leading to Tbet induction
A modest decrease in the thymic expansion of NKT cells was independently of Stat1 or Stat4 remain unclear, a role for
reported in mice lacking CD28 or B7,116,117 but the nature of Ets1 or MEF has been suggested.127,128
the B7-expressing cell type (epithelial cells or DCs) has not Mice lacking Gata-3 exhibited severe but complex NKT-
been determined. cell defects129 consisting of a considerable reduction of most
of their thymic and peripheral NKT cells, with the strange
Transcriptional Control of Natural Killer T-Cell exception of their stage 3 thymocytes which accumulated
Development: The Central Role of Promyelocytic normally. Gata-3 is required for ThPOK induction, explain-
Leukemia Zinc Factor ing the loss of CD4 by residual NKT cells.113
The transcription factor PLZF encoded by Zbtb16 was identi- Mice lacking TGFβRII had severe NKT-cell developmen-
fied as the signature master transcription factor of the NKT- tal defects, particularly in a competitive chimera setting.130
cell lineage.104,118 PLZF is a member of the BTB-ZF family of Mutant NKT cells had defective expression of the IL-7 re-
transcription factors, which also includes Bcl6 and ThPOK. ceptor and increased apoptosis at stage 1, but a normal rate
It is composed of a bric-a-brac, tramtrack, and broad (BTB) of cell division at stage 2. Tbet and IL2Rβ were expressed by
homodimerizing domain and nine Kruppel-type Cys2His2 a larger than normal fraction of stage 2 cells. These effects
zinc fingers. The BTB domain binds a corepressor complex were mediated by different branches of TGFβ signaling, in-
made of histone deacetylases, N-Cor, and SMRT, whereas de- cluding Tif-1γ, Smad4, and the Smad4/Tif-1–independent
oxyribonucleic acid binding is mediated by the zinc fingers. pathway.
PLZF is induced in NKT precursors just after TCR signaling, Complex and somewhat conflicting results have been re-
with high amounts detected on nearly half of stage 0 and all ported regarding the role of different NF-κ B factors. Mice
of stage 1 and 2 cells, and lower levels found in stage 3 cells. lacking IKKβ or expressing a degradation resistant form
Mice lacking PLZF through either the luxoid mutation, of I-κ Bα generally exhibited severe cell-intrinsic defects at
which induces a frameshift leading to a truncated BTB pro- stage 1 and 2, due in part to apoptosis. These effects ap-
tein without zinc fingers, or deletion of exon 2 encoding peared to be mediated by the partly redundant functions
the BTB domain, exhibited a block in NKT-cell develop- of NF-κ B1, RelA, and c-Rel.131–134 The upstream activator
ment,104,118 as well as defects in spermatogonial cell main- of NF-κ B may include signaling through TCR, SLAM/SAP/
tenance and osteoblast differentiation.119,120 PLZF-deficient Fyn, or CD28/B7. In contrast, ablation of RelB in the alter-
NKT precursors identified by CD1d-αGalCer tetramers were nate pathway impaired early NKT cells development non-
blocked at stage 1, unable to acquire effector characteristics. specifically by disrupting the thymic stroma.
Instead, they maintained a naïve phenotype and function, While early reports suggested some cell-intrinsic role for
recirculating between blood, lymph node, and spleen, rather lymphotoxin signaling in NKT-cell development,135 more
than homing to liver and lung, and producing IL-2 instead recent analysis indicated that the major attrition in lympho-
of IL-4 and IFNγ upon stimulation. Intriguingly, however, toxin-deficient animals occurred during the transition from
they incorporated BrdU at the same high rate as wild type, the thymus to the periphery, possibly related to a defect in
but without showing expansion, suggesting aborted division NKT thymocyte emigration in the absence of interaction
or increased cell death. between LTβR on thymic stromal cells and lymphotoxin on
Ectopic expression of PLZF under the CD4 or Lck pro- bone marrow–derived cells.136
moter induced a typical stage 2 NKT program in CD4 T cMyc is induced as early as stage 0, when NKT cells enter
cells, with downregulation of CD62L, upregulation of S phase. Its conditional ablation resulted in developmental
CD44 and LFA-1, dual secretion of IL-4 and IFNγ, and arrest at stage 0,72 similar to cyclin D2-deficient mice (un-
homing to liver and lung.104,121–123 Notably, the full effector published data), suggesting that entry into cell cycle oc-
conversion depended on expression of PLZF at high levels curred at stage 0, one stage earlier than previously thought.64
comparable to stage 1 and 2 NKT thymocytes, which was
only achieved in the CD4 promoter transgenic model.124 In Micro-Ribonucleic Acids
these mice, nearly 100% of CD4 cells acquired the effec- Conditional ablation of Dicer in thymocytes led to a mas-
tor phenotype at the CD4 SP stage. This effector conversion sive loss of NKT thymocytes but the micro-ribonucleic acids
was also observed in MHC class II–restricted TCR/RAG involved have eluded identification.137–140
knockout transgenic models. Thus, while the induction of
PLZF requires the signaling environment of NKT cells, the
NKT effector program can be transferred by PLZF alone. Other Promyelocytic Leukemia Zinc Factor–Positive
The molecular basis of the extensive gene reprogramming Lineages
induced by PLZF, in particular the direct target genes and γδT cells expressing the semi-invariant Vγ1Vδ6 receptor
the biochemical mechanisms of gene activation or repres- were previously shown to express the NK1.1 marker, home
sion, are currently being studied. to the liver, and produce IL-4, like Vα14-Jα18 NKT cells.141

These cells were not only dependent on SAP for their differ- at the surface of IL-15–producing cells, revealed that the sole
entiation, but also expressed PLZF.142–144 In OP-9 supported source of IL-15 for development was a nonhemopoietic radi-
thymocyte cultures, PLZF could be induced after cross- ation-insensitive stromal cell type, most likely a medullary
linking with Vγ-specific TCR antibodies. Likewise, TCR epithelial cell.157 In contrast, both radiation-resistant and he-
αβ thymocytes upregulated Zbtb16 messenger ribonucleic mopoietic cells served as partly redundant sources of IL-15
acid after injection of anti-TCRβ antibodies in vivo.100 These for peripheral NKT cells.
findings support the notion that agonist signaling induces
PLZF to generate innate-like αβ and γδ lineages. MR-1–spe- CXCL16
cific Mucosal-associated invariant T cells were also reported CXCR6 signaling, in response to CXCL16 produced by en-
to express PLZF104 and home to liver and gut.145 Together dothelial cells, is critical for the survival of liver NKT cells in
with experiments showing that CD4 T cells selected by basal conditions, but not for their specific accumulation or
MHC class II expressed on thymocytes also acquired PLZF their crawling behavior in the liver sinusoids.158
in a SAP-dependent manner,109–112 these observations reveal
Cluster of Differentiation 1d Ligands
an authentic cellular and molecular pathway dedicated to
In mice lacking CD1d in peripheral tissues, the full terminal
the generation of innate-like lymphocytes.
maturation of NKT cells after thymic export is partially al-
Deficiencies or mutations affecting certain genes associ-
tered but the overall frequency, tissue distribution, survival,
ated with TCR signaling produced a phenotype characterized
and functional properties of NKT cells are not significantly
by a cell-intrinsic expansion of the PLZF-expressing thymic
impaired.80,87 Thus, peripheral NKT cells do not require
lineages, including Vγ1Vδ6 T cells and Vα14-Jα18 NKT cells.
contact with self-antigens for homeostasis.
This expansion was associated with a bystander conversion
of MHC-restricted thymocytes, mainly CD8 T cells, into in- Microbiota
nate-like CD62Llo CD44hi CD122hi eomesoderminhi effectors, NKT cells did not exhibit major disturbances in their devel-
caused by IL-4 secreted by the PLZF-expressing cells. This opment or function in germ-free mice.78,159,160 However, mice
complex phenotype was observed in mice lacking the Tec ki- lacking NKT cells were reported to harbor changes in the
nases Itk or Rlk,146–148 or bearing the Y145F mutation in the composition of their microbiota.160
scaffold protein SLP76, which impairs binding of these kinases
after TCR signaling.143 Mice lacking the inhibitor of E proteins
Id3, which can be induced downstream of the TCR-activated TISSUE DISTRIBUTION AND RECIRCULATION
MAP kinases, showed a strikingly similar phenotype,143,149,150 In C57BL/6 mice, NKT cells represent ∼0.2% of the T-cell
as did mice after conditional ablation of CREB binding pro- population in blood and peripheral lymph nodes; ∼2.5% of T
tein,151 a general transcriptional coactivator, and KLF2, a tran- cells in spleen, mesenteric, and pancreatic lymph nodes; 5%
scription factor that controls CD62L and S1P1 receptor upreg- in the lung; and up to 40% in the liver. Furthermore, parabi-
ulation after positive selection.148,152 otic experiments have shown that, unlike most lymphocytes,
The convergent impact of these different genes suggests NKT cells do not recirculate and are lifelong residents of lym-
the existence of a negative regulatory pathway that physi- phoid tissues, lung, and liver.161
ologically contains the expansion and function of innate-
like T cells. Notably, wild-type BALB/c mice spontaneously
exhibited a similar, albeit milder, phenotype as mutant Tissue and Intravascular Residents
C57BL/6 mice, including a relative expansion of PLZF- This tissue residency is particularly remarkable in the liver
expressing NKT thymocytes and a bystander induction of and lung where NKT cells are mainly found within the cap-
CD44, CD122, eomesodermin, and the rapid production illary microvasculature.158,161 Intravital imaging using con-
of cytokines by CD8 SP thymocytes.148 This phenotype was focal microscopy of the liver of CXCR6GFP mice, where a
dependent on klf13.153 Thus, there is a natural crosstalk be- majority of brightly fluorescent liver lymphocytes are Vα14
tween innate-like and conventional T cells during thymic NKT cells, demonstrated their presence inside the sinusoid
development, whose variations may have a profound impact capillaries, adhering to the luminal side of endothelial cells
on the immune system. and crawling, without directional bias, at a speed of ∼10 μm/
min (Fig. 18.6). While other effector T cells and monocytes
can adopt a similar crawling behavior transiently, before
Homeostasis they extravasate to enter inflamed tissues, NKT cells are
NKT cells are produced at a similar rate from fetal to adult constitutively and permanently crawling in steady-state
life, and they persist indefinitely after adult thymectomy. conditions.
The mechanisms ensuring their long life in peripheral tis- The expression of high amount of CXCR6 by NKT cells
sues are detailed in the following. matches the production of the corresponding chemokine
ligand CXCL16 by endothelial cells lining the sinusoids.
Interleukin-15 CXCR6 appeared to be important for the survival of liver
Like NK cells and CD8 memory cells, mature NKT cells criti- NKT cells158 as well as NK cells,162 but it was not required for
cally depend on IL-15 for their terminal maturation as well as liver accumulation. Furthermore, the adhesion and crawl-
their survival and homeostatic renewal.154–156 Cell-type–spe- ing processes were unaltered by pertussis toxin treatment,
cific ablation of IL-15Rα , which is essential to present IL-15 suggesting independence from Gαi protein signaling.163

NATURAL KILLER T-CELL FUNCTIONS

Functional Properties
DC activation and the explosive release of cytokines and che-
mokines by NKT cells have well-documented functional con-
sequences in a variety of pathological or therapeutic conditions
involving lymphoid tissues and body organs. Important factors
biasing the Th1, Th2, or Th17 outcome of NKT-cell activation
in vivo have recently emerged. They include the structure of
the lipid antigen, the antigen-presenting cell type, the tissue,
and the coexposure to TLR ligands. These will be discussed
in two separate contexts: the administration of synthetic NKT
ligands, which function as potent vaccine adjuvants, and the
activation of NKT cells in the context of microbial infection.
Vaccine Adjuvant Properties

Dendritic Cell–based Network of Activation
Several studies have characterized a cascade of activation
FIG. 18.6. Intravascular Patrolling by Liver Natural Killer T (NKT) events following the exogenous administration of NKT li-
Cells. Fluorescent NKT cells in CXCR6-GFP “knock-in” mice are crawl-
ing on the luminal side of sinusoid endothelial cells. Dashed lines show gands such as αGalCer (Fig. 18.7). The central feature was a
their path over a period of 10 mn in a live imaging study. (Courtesy of cross-activation between NKT cells and DCs initiated upon
Frederic Geissmann, King’s College, London, UK, and Dan Littman, the presentation of αGalCer by resting DCs to NKT cells.
New York University, New York, NY, USA) Activated NKT cells upregulated CD40L, Th1 and Th2 cyto-
kines, and chemokines; CD40 crosslinking induced DCs to
upregulate CD40, B7.1, and B7.2 and IL-12p40, which in turn
enhanced NKT-cell activation and cytokine production.168,169
Thus, the constitutive activation of LFA-1 was neither a DCs upregulated MHC class I– and class II–mediated anti-
consequence of TCR signaling, as NKT cells normally ac- gen presentation, particularly cross-priming, and secreted
cumulated in the liver of mice lacking CD1d outside of the CCL17,170 a chemoattractant for naïve CCR4 + CD8 T cells.
thymus,80,87 nor of chemokine-induced Gαi protein signal- IL-12 released by DCs promoted prolonged IFN-γ produc-
ing. Notably, transgenic expression of PLZF in MHC class tion by NK cells and activated their cytolytic properties.171,172
II–restricted CD4 T cells was sufficient to induce their liver Thus, αGalCer promotes robust CD4 and CD8 T-cell–medi-
retention, indicating that the adhesion and crawling behav- ated adaptive immune responses against coadministered non-
ior are an integral component of the PLZF-induced func- replicating protein antigens.173–175 Furthermore, although TLR
tional program.161 signaling is not involved in the innate response to αGalCer,
the combination of TLR and NKT ligands was synergistic.176
Organ-specific Natural Killer T-Cell Sublineages Th1 and Th2 Variants
DN NKT cells expressing Ror γ t and producing IL-17 and Different synthetic agonists showed biased Th1 versus Th2
IL-22 but not IL-4 or IFN-γ selectively resided in skin and outcomes mainly due to their differential ability to induce
draining axillary and inguinal lymph nodes.164,165 While IL-12 from DCs and to transactivate NK cells. Variants of the
they shared the usual CD44hiCD62L lo effector phenotype original αGalCer (KRN7000) with short or unsaturated lipid
of other NKT cells, they did not express the stage 3 markers chains induced similar Th2 but decreased Th1 cytokines.177–179
Tbet, IL2Rβ, CD69, or NK receptors and instead displayed While early reports pointed to the lower TCR affinity of a
a CD103hi CCR6 + CD121a + phenotype. The identification variant with truncated phytosphingosine chain (psC9:0,
of a minor cell population in the thymus with a similar called “OCH”),180 several other Th2 variants, including
phenotype suggested that they constituted a separate de- acylC8:0 or acylC20:2, demonstrated identical interactions
velopmental sublineage.166 Intriguingly, their expression based on affinity measurements and crystal structures.56,181,182
of Nrp-1, a marker of recent thymic emigrants and their All these Th2 variants had increased solubility in water and
rapid disappearance after adult thymectomy, suggested could be rapidly loaded onto CD1d molecules at the cell sur-
that, unlike most NKT cells, they represented a short-lived face. In contrast, αGalCer was almost exclusively loaded in
population.82 the lysosome due to a requirement for lipid transfer proteins.
A subset of NKT cells expressing the IL-25 receptor Whereas αGalCer remained stably associated with CD1d,
and biased toward production of IL-4, IL-3, and IL-9 was the short or unsaturated variants were rapidly ejected upon
found in the thymus, spleen, and lung, but was poorly rep- recycling to the acidic endosome.21 Interestingly, αGalCer
resented in the liver and bone marrow.167 Whether these was found associated with surface CD1d molecules on lipid
cells constitute a tissue-specific sublineage that differ- rafts, whereas the short or unsaturated variants were loaded
entiates in the thymus or in the periphery remains to be to CD1d molecules outside of these rafts, perhaps as a conse-
determined. quence of their different loading compartments.182 However,

BCR
CD4 CD8 CD8

B
CD28
B7
IFN-γ CCL17
TCR CD1d IFN-γ

iNKT DC
CD40L CD40
NK
IL-4,-13 IL-12
IL-12
TLR
FIG. 18.7. Cellular and Molecular Network Activated by the Natural Killer T (NKT) Ligand `GalCer.
Cross-activation between NKT cells and dendritic cells (DCs) lead to cluster of differentiation
(CD)40L-CD40 interactions, activating the DC to present peptide antigens to CD4 and also to CD8 cells
(cross-priming). NKT cells promptly release Th1 and Th2 cytokines. Naïve CD8 T cells are chemoat-
tracted through CCL17 secreted by DCs. NK cells are “transactivated” by DCs and interleukin-12 to
release interferon-γ in a prolonged manner. B cells seeing antigens can receive cognate help from
NKT cells and/or from CD4 cells.
this is unlikely to explain the Th2 bias, as the variants in- Sphingomonas, which carries abundant glycosphingolipid li-
duced the same balance of cytokines as αGalCer when tested gands in its cell wall.198 The production of immunoglobulin
in vitro with purified antigen-presenting cells. A more likely (Ig)G antibodies against the bacterial membrane appended
explanation for the Th2 bias was suggested by in vivo stud- enzyme pyruvate dehydrogenase complex-E2 depended on
ies with mice carrying a floxed CD1d allele. Indeed, while CD1d expression by B cells, as evidenced in mixed bone
αGalCer presentation was mostly limited to lysosomally ac- marrow chimeras where CD1d expression was limited to
tive DCs and macrophages, which produced IL-12, the Th2 one Ig allotype–marked, B-cell compartment. NKT cells
variants were broadly presented by all CD1d-expressing cell engaged in cognate and prolonged interactions with B cells,
types, most of which did not produce IL-12.89 Thus, struc- after acquiring a stereotypical Bcl-6 dependent follicular
tural differences in lipids resulted in the involvement of helper program with induction of CXCR5, PD1, and IL-21,
different antigen-presenting cell types, altering the cellular and migration to the germinal center.196,197 Intriguingly, im-
network and the outcome of NKT-cell activation (Fig. 18.8). munization with αGalCer-nitrophenol (NP) induced faster
but less sustained germinal center formation with fewer
Anergy long-lived plasma cells and inferior memory compared
A limitation to the adjuvant properties of NKT ligands is the with keyhole limpet hemocyanin-NP or ovalbumin-loaded
rapid NKT TCR downregulation and apoptosis, followed by (OVA)-NP,196,197 suggesting that NKT cell help to B cells was
a long-lasting depletion and a state of anergy.183–185 Anergy intrinsically inferior to conventional T-cell help. However,
could be partially avoided by CD28 costimulation or PD1 other variables might have contributed to these differences,
blockade,186,187 or by the administration of αGalCer-pulsed including the monovalency of NP in the case of αGalCer,
DCs, or αGalCer conjugated to nanoparticles.188–190 The differences in bioavailability, cellular targeting, half-life,
mechanism of NKT-cell anergy remains obscure. and processing of the different immunogen preparations. It
is also possible that the rapid induction of NKT-cell anergy
Natural Killer T-Cell Help to B Cells after exposure to αGalCer might limit the interactions re-
NKT-cell activation by agonist ligands promoted antibody quired for effective B-cell help.
production, affinity maturation, and switching against Splenic marginal zone B cells constitutively express ap-
associated antigens. Help could be indirect, through proximately eight times more CD1d on their surface than
CD40L-CD40 mediated activation of the DCs presenting other B-cell subsets.9 They responded to antigen in an ac-
peptide antigen to CD4 T cells, which in turn provided MHC celerated manner with increased sensitivity conferred by the
II–restricted help to B cells recognizing a linked antigen. complement receptor CD21, which lowers the B-cell recep-
Alternatively, direct NKT-B cell cognate interactions have tor signaling threshold to complement-coated antigen.199
been demonstrated in vitro and in vivo.191–197 The relevance These characteristics have long suggested a special func-
of this direct cognate help was illustrated by infection with tion in CD1d-mediated B-cell responses. In support of such

Long Lipid Short/Unsaturated Lipid

IFN-γ IL-4 IFN-γ IL-4
B
iNKT iNKT
TCR
CD40L
FIG. 18.8. Th1 and Th2 Agonist Ligands of Natural MΦ
Killer T (NKT) Cells. Left, long and saturated αGal- CD40 CD1d
Cer variants (including KRN7000) load cluster of dif-
ferentiation (CD)1d in the lysosomal compartment
due to their requirement for lipid transfer proteins.
The main antigen-presenting cell is a dendritic cell
(DC), which produces lots of interleukin (IL)-12 and
transactivates NK cells. Right, short or polyunsatu-
rated lipids load CD1d directly at the cell surface, IL-12 DC IL-12
but are rapidly ejected upon recycling to the lyso-
some due to displacement by longer, hydrophobic
peptides at acidic pH. These short or polyunsatu- NK NK
rated lipids are promiscuously presented by all
CD1d expressing cells, limiting DC interactions, in-
terleukin-12 production, and natural killer transac-
tivation. CD1d complexed with long and saturated
variants sit in lipid rafts, whereas CD1d complexed
with short or polyunsaturated variants are excluded IFN-γ
from lipid rafts.
NKT-MZB interactions in vivo, depletion of marginal zone LPS-negative member of α-proteobacteria, an abundant class
B cells impaired the IgM anti-NP antibody response after of bacteria. These glycosphingolipids included the dominant
systemic administration of αGalCer-NP.200 α-branched glucuronyl and galacturonyl ceramides (GSL-1)
and the less abundant di- (GSL-2), tri- (GSL-3), and tetra-
Natural Killer T-Cell Help to Macrophages and glycosylated (GSL-4) species. Although they form structures
Myeloid Suppressor Cells that are reminiscent of LPS, their synthesis pathway and
Subcapsular CD169 + macrophages efficiently captured and their role in the microbial cell wall are not well understood.
presented αGalCer to NKT cells in lymph nodes after footpad GSL-1 activates large proportions of mouse and human
injection of 200 nm silica beads coated with the lipid antigen.201 NKT cells,31,32,206,207 but it is unclear at present whether the
This selective capture further emphasizes the importance of more complex GSL-2, -3, and -4 can be recognized by NKT
the formulation of αGalCer in targeting different antigen- cells or even whether they can be efficiently processed into
presenting cell types. Notably, agonist-mediated NKT-cell GSL-1 by host antigen-presenting cells. During infection,
interaction with myeloid suppressor cells could reverse their Sphingomonas was phagocytosed by macrophages and DCs,
suppressive properties. Furthermore, natural interactions be- and elicited an activation cascade similar to exogenous αGal-
tween NKT cells and myeloid suppressor cells have been sug- Cer. NKT-deficient mice had 15- to 1000-fold more residual
gested in transplanted cancers and influenza infection.202–204 bacteria than their wild-type littermates in the first few days
after infection.31,32 High doses of Sphingomonas induced a
Microbial Infections lethal toxic shock in wild-type but not NKT-deficient lit-
The recognition of either microbial α-linked glycolipid termates. These observations have led to the hypothesis that
antigens or self-β -linked antigens induced by TLR signal- NKT cells and their canonical TCR specificity might have
ing ensures the recruitment of NKT cells in most microbial evolved to meet the challenges of these gram-negative LPS-
infections (Fig. 18.9). Which of these modes of activation negative bacteria. While Sphingomonas can cause severe in-
predominates for individual microorganisms is somewhat fection, particularly in immunocompromised hosts, other
controversial, as TLR signaling induces IL-12, which is a po- more deadly members of the class of α-proteobacteria may
tent amplifier of IFN-γ secretion, and NKT cells upregulate have provided stronger evolutionary pressures on the NKT-
their IL-12 receptor upon TCR signaling. cell system. Particularly interesting is the case of Ehrlichia, a
tick-borne pathogen and member of the Rickettsiales, which is
Direct Microbial Lipid Recognition of widespread significance for mammals, including wild and
Glycosphingolipids closely related to αGalCer were found in domesticated ruminants, dogs, and humans in some regions
the cell wall of Sphingomonas,30,205 a prominent gram-negative of the world such as Africa and east Asia. Ehrlichia muris

Direct Indirect
Microbial Recognition Microbial Recognition
IFN-γ IFN-γ
IL-4
iNKT IL-12 IL-12 iNKT
CD40L TCR
TCR LFA-1
CD40 CD1d: CD1d: ICAM-1
bacterial self-lipid
lipid
ICAM-1
self-lipid
DC
IL-12 IL-12
Dectin-1 TLR4
β1,3-Glucan LPS
FIG. 18.9. Dual Recognition of Self- and

Microbial Ligands in Microbial Infections.
Left, microbial organisms expressing agonist
NK NK ligands can elicit direct activation of natural
killer T (NKT) cells. Right, microbial organisms
lacking NKT ligands activate NKT cells through
the enhanced recognition of self NKT ligands.
These self ligands are upregulated by toll-like
receptor– or dectin-1–mediated signals, and
their recognition is enhanced by upregulation of
intercellular adhesion molecule at the immune
synapse. Interleukin-12 secreted by dendritic
IFN-γ cells amplifies interferon-γ produced by NKT
cells and by natural killer cells.
activated NKT cells independently of MyD88, and its clear- Kupffer cells and the pertussis toxin–sensitive response of
ance was profoundly impaired in CD1d- or Jα18-deficient CXCR3-expressing NKT cells.
animals.31 However, the cell wall composition of Ehrlichia,
a gram-negative, LPS-negative obligate intracellular bacte- Self-Lipid Recognition
rium, has not been elucidated. Many bacteria that do not harbor NKT ligands never-
Other bacteria such as Streptococcus34,208 and Borrelia33 theless induce massive amounts of IFN-γ in a NKT- and
expressed α-linked diacyglycerol lipids that bound CD1d CD1d-dependent manner.31,39 This secretion of IFN-γ is
and could directly engage the mouse and human NKT TCR, considerably amplified through TLR-induced IL-12 released
resulting in accelerated microbial clearance. Borrelia was by DCs and macrophages and the transactivation of NK
normally cleared by NKT-deficient mice, except at later time cells (see Fig. 18.9). TLR signaling regulates several glyco-
points in the joints.163,209 Intravital microscopy of the liver sphingolipid enzymes, leading to the accumulation of stim-
response to an intravenous inoculation of Borrelia burgdor- ulatory glycosphingolipids (see Fig. 18.3), and activation of
feri showed rapid microbial uptake by stationary Kupffer LFA-1/ICAM-1 interactions further facilitate NKT-cell rec-
cells followed by the attraction and arrest of NKT cells in the ognition of these low-affinity ligands. However, contrary to
sinusoids. Stable, CD1d-mediated contacts between Kupffer an early report,210 NKT cells do not usually constitute the
cells and NKT cells depended on the secretion of CXCL9 by predominant cell-type producing IFN-γ in response to IL-12

in vivo,39,211 which explains why they generally do not ap- Epstein-Barr virus in patients with X-linked lymphoprolif-
pear to play an essential role in many bacterial infections. erative immunodeficiency syndrome due to SAP mutations
Nevertheless, bacterial clearance and neutrophils were re- was hypothesized to result from the absence of NKT cells.93
portedly decreased in the lungs of CD1d-deficient mice after Which of these effects or associations reflect specific viral
infection with Pseudomonas aeruginosa.212 This may not be evasion/immune defense strategies and the nature of the
the case at other sites of infection.213 Different conclusions putative NKT ligands involved in these infectious condi-
were also reported regarding the importance of NKT cells tions remain to be determined.
versus NK cells in lethal LPS-induced toxic shock.214,215
Autoimmunity, Inflammation, and Allergy
Fungal and Parasitic Infections Asthma
The production of IgG antibodies to the malaria circum- Studies in the OVA/Alum model of asthma in mouse and a
sporozoite antigen, a key component of protective immune report that humans with asthma harbored high percentages
responses in humans, was suggested to depend on NKT- of Vα24 NKT cells in their bronchoalveolar lavage suggested
cell recognition of malarial glycosylphosphatidylinositol a broad role of NKT cells in various forms of airway allergic
antigens in a mouse model.216 However, additional experi- inflammation. However, these studies have not been widely
ments failed to detect a CD1d-dependent component to confirmed.230
this antibody response and glycosylphosphatidylinositols Other studies have focused on the presence of airborne
have not been identified as NKT-cell antigens in other re- NKT ligands in natural environments, for example in house
ports.217,218 In the context of helminth infection, DCs pulsed dust or in air samples where Sphingomonas is one of the pre-
with Schistosoma mansoni eggs activated NKT cells to se- dominant microbial species identified.231,232 Airway exposure
crete Th1 and Th2 cytokines in vitro in a β -hexosaminidase to these NKT ligands resulted in the rapid CCL17-induced
B-dependent but MyD88-independent manner, suggesting recruitment of resident intravascular NKT cells into the
recognition of the self-ligand iGb3 in the absence of TLR lung with formation of eosinophilic granulomas. Notably,
signaling.219 a single airway exposure to protein antigen associated with
Intratracheal infection by the fungal pathogen Aspergillus NKT ligands led to massive recall allergic airway inflamma-
fumigatus indirectly stimulated NKT-cell production of tion upon airborne challenge with protein alone. Likewise,
IFN-γ through activation of the dectin-1/MyD88 pathway by coadministration of house dust extracts containing un-
the major cell wall polysaccharides, β -1,3 glucans. Clearance identified NKT ligands promoted allergic sensitization to
of the pathogen was impaired in NKT-deficient mice. airborne OVA in an NKT-dependent manner.
Thus, emerging studies indicate that natural exposure to
airborne NKT ligands in the environment may constitute a
Viral Infections previously unappreciated cause of allergic sensitization.
Viruses can activate TLR signaling through the adaptors
MyD88 and TRIF, which in turn activate autoreactive NKT Diabetes and Autoimmune Diseases
cells. Relatively modest defects in viral clearance have been Many reports have suggested a role of NKT cells, usually
reported in CD1d-deficient mice infected with encepha- of regulatory nature, in autoimmune diseases such as type
lomyocarditis virus220 or coxsackie B3,221 but these defects I diabetes, arthritis, and lupus, and in atherosclerosis and
were not observed in Jα18-deficient mice, ruling out a inflammatory bowel disease.233 These reports mostly relied
specific role of Vα14 NKT cells. Influenza virus clearance on comparisons between wild-type versus Jα18- or CD1d-
was normal or modestly impaired in mice lacking NKT deficient mice but did not offer precise mechanistic insights
cells.222,223 Infections with lymphocytic choriomeningitis into the putative function of NKT cells and their mode of
virus, mouse cytomegalovirus, vaccinia virus, and coro- recruitment. Some of these studies did not use littermates
navirus were unaffected. In humans, a profibrotic role of as wild-type controls and have not been widely reproduced.
Vα24 NKT cells was suggested in hepatitis C224 and non- Furthermore, the recent discovery that NKT thymocytes
Vα24 CD1d-restricted T cells were found in the liver.225 impact the function of MHC-restricted T cells in a by-
Although a specific role of Vα14 NKT cells in herpes sim- stander manner, through the secretion of IL-4,148 suggests
plex virus infection is controversial, 226,227 studies have sug- that alternative, indirect mechanisms might contribute to
gested that viral invasion may be associated with counter- these observations.
measures against CD1d or NKT cells. For example, herpes
simplex virus-1 drastically and specifically impaired CD1d Primary Biliary Cirrhosis
recycling from lysosome to plasma membrane, an essen- Primary biliary cirrhosis (PBC) is an enigmatic disease char-
tial pathway for glycolipid antigen presentation to NKT acterized by the presence of antimitochondrial antibodies,
cells.228 Kaposi sarcoma–associated herpes virus encodes liver lymphocytic infi ltrates, and the chronic destruction of
two modulators of immune recognition MIR1 and MIR2 the biliary epithelium leading to cirrhosis.234 The autoanti-
that downregulated CD1d along with other immunologi- bodies recognize an epitope of the mitochondrial pyruvate
cally relevant molecules such as MHC class I, CD86, and dehydrogenase complex-E2 enzyme that is particularly well
ICAM-1 through ubiquitination of lysine residues in their conserved in Novosphingobium aromaticivorans, a strain of
cytoplasmic tail.229 The lethal outcome of infections with Sphingomonas. Further, patients with PBC, including those

lacking antimitochondrial antibodies, were specifically sero- CONCLUSION

positive against Sphingomonas, which was detected by poly- Recent Advances
merase chain reaction in stool samples of 25% of diseased or
healthy individuals, suggesting that PBC may be induced by In the past 4 years, pivotal discoveries have advanced our
aberrant host reactivity to this bacterium.235 Patients with understanding of the biology of NKT cells, particularly re-
PBC also showed an enrichment of Vα24 NKT cells in liver garding their dual recognition of self- and foreign-lipid anti-
biopsies and a depletion in blood.236 In light of the recent gens; their role in infectious and allergic diseases; their spe-
finding that Sphingomonas cell wall glycolipids specifically cialized effector functions within microenvironments; their
activate NKT cells, these studies suggested that NKT cells expression of a lineage-specific master transcription factor,
might play a key role in the pathogenesis of PBC by promot- PLZF; and their cross-talk with MHC-restricted T cells.
ing aberrant responses to Sphingomonas. An experimental
model of infection induced a chronic liver disease with PBC- Remaining Challenges
like granulomas and antimicrobial autoantibodies, which The nature and the hierarchy of self- and foreign antigens
could be transferred by lymphocytes. These provocative recognized by NKT cells and their proposed role in a variety
findings support the hypothesis that PBC might be the con- of diseases remain a work in progress. Are there other, per-
sequence of cryptic episodes of bacterial infections, perhaps haps more broadly relevant NKT ligands and what controls
triggered by commensals containing NKT ligands, leading their expression?
to NKT cell–assisted breakdown of tolerance to shared mi- Which class of microorganisms might explain the evo-
crochondrial/bacterial antigens. lution of this elaborate lineage? Might NKT cells protect
against diseases that have recently disappeared, at least in
Cancer Western societies? In that regard, typhus-like disease after
infection by tick-borne Ehrlichia may be comparable to ma-
A role of Jα18 and non-Jα18 CD1d-restricted NKT cells larial disease in its devastating impact in endemic regions.
in cancer surveillance has been inferred from reports that What are the signaling events underlying NKT-cell dif-
CD1d- and Jα18-deficient mice seemed to differ from wild- ferentiation, particularly the signals inducing PLZF, the
type mice in their susceptibility to a variety of spontaneous molecular mechanisms of PLZF function, and the acquisi-
and transplanted cancers.237 Vα14 NKT cells were suggested tion of NK receptors? Are there master transcription factors,
to exert protection against spontaneous sarcomas. These similar to PLZF, that control other innate-like lineages such
included fibrosarcomas in mice injected intramuscularly as CD8αα TCRαβ IELs and B1 B cells?
with methylcholantrene,238 osteosarcomas and hemopoietic Finally, will NKT agonists demonstrate efficiency and
tumors in p53 + /- mice,239 and carcinomas in transgenic ad- safety as vaccine adjuvants to combat infections and cancer
enocarcinoma of the mouse prostate mice.240 Some trans- in humans?
planted cancers seemed aggravated in their progression by
the presence of non-Vα14 NKT cells, apparently because of
their secretion of IL-13 and interaction with myeloid sup- Other Cluster of Differentiation 1–restricted T Cells?
pressor cells.202,203 These cancer studies have commonly Although there is strong evidence that the majority of CD1d-
concluded that endogenous ligands might be induced and restricted αβ T cells in mice belong to the NKT lineage,5 little
presented by the tumor themselves or by antigen-presenting is known about the human populations of αβT cells restricted
cells. The crosstalk between myeloid-derived cells and NKT by CD1d or other CD1 isotypes such as CD1a, b, and c.
cells resulted in suppression, or on the contrary, to exacer- Convergent studies reviewed in this chapter suggested
bation of immune responses against cancer. In most cases, that thymic selection by ligands predominantly expressed
however, the nature of these putative ligands has escaped on cortical thymocytes or other hemopoietic cells was a key
precise identification. One exception lies with the proposed determinant of innate-like lineage decision. As CD1a, b, c,
role of lysophosphatidylcholine as a tumor antigen in my- and d are all predominantly expressed on cortical thymo-
eloma patients.54 cytes, they might mainly select PLZF-expressing T cells.
A serious challenge to the interpretation of these stud- This hypothesis remains to be directly tested in fresh human
ies is their near exclusive reliance on comparisons between T-cell populations. Notably, humanized mice expressing a
wild-type and NKT-deficient mice and the lack of direct evi- CD1b-restricted human TCR together with human CD1b
dence of NKT cell involvement in the antitumor response. under its own regulatory elements developed effector-like T
Moreover, some studies did not use proper littermate con- cells that expressed PLZF.242 Other insights have come from
trols, or could not be independently reproduced. For ex- studies of a small subpopulation of human αGalCer-specific
ample, the longstanding but isolated report that methyl- T cells that did not express Vα24, but used the remain-
cholantrene-induced sarcomas were naturally controlled by ing conserved elements of the NKT TCR, Jα18 and Vβ11.
Vα14 NKT cells238 could not be confirmed in recent dou- These cells expressed intermediate amounts of PLZF and
ble-blind studies comparing large groups of CD1d- or Jα18- kept a naïve CD62Lhi CD45ROhi phenotype.124 This natural
deficient mice and their littermate controls.241 In addition, example suggests that PLZF expression may be a reliable
some of the reported results may be explained by an indirect marker of T cells selected by CD1 molecules, but that dif-
rather than a direct function of NKT cells, for example due ferent levels of expression may be imparted depending on as
to their cross-talk with conventional T cells.148 yet unknown factors such as perhaps, the intensity of TCR

signaling. Other studies based on the large scale cloning human CD1-restricted TCRs and CD1 molecules, a more de-
of fresh T cells have suggested that 3% to 10% of the cord finitive characterization of the different populations of CD1-
blood T-cell population may be CD1-restricted and naïve- restricted T cells should emerge. Will they demonstrate the
like, with a larger proportion of memory-type cells appear- heavy contribution of germline TCR segments and the basal
ing in adults.243 These important studies could not, however, autoreactivity that is a hallmark of NKT cells? Will they dis-
determine whether the CD1-restricted T cells were selected play adaptive or innate-like features? Forthcoming answers
by CD1 molecules or were merely cross-reactive cells pri- to these fundamental questions will not only be crucial for
marily selected by MHC. clinical applications, but will also shed light on the evolu-
Most fresh CD1-restricted T cells identified so far in tionary logics of lipid antigen recognition.
mice and humans have shown a CD4 + or CD4-CD8β - DN
phenotype, with occasional expression of CD8α but rarely
CD8β.124,243 This coreceptor expression pattern parallels the ACKNOWLEDGMENTS
one found in mouse and human CD1d-restricted NKT cells I thank members of my laboratory and my colleagues Luc
where it is associated with expression of ThPOK/cKrox, sug- Teyton and Paul B. Savage for their insights, contributions,
gesting that most CD1-restricted T cells might also express and comments on this chapter; Dirk Zajonc for discussions
this transcription factor. and for providing Figure 18.4; Frederic Geissmann and Dan
These observations provide a background for studies Littman for providing Figure 18.6; and Seth Scanlon for
of fresh, unimmunized CD1-restricted T cells in humans, help with the figures. Albert Bendelac is supported by the
which are now possible through the use of tetramers. Howard Hughes Medical Institute and by grants from the
Together with studies of “humanized” mice expressing National Institutes of Health.

CHAPTER
19 Macrophages and Phagocytosis
Siamon Gordon
INTRODUCTION Metchnikoff, a comparative developmental zoologist, is

Macrophages (M f ) represent a family of mononuclear leu- widely credited for his recognition of phagocytosis and
kocytes that are widely distributed throughout the body leukocyte recruitment as a fundamental host defense
within and outside of lymphohemopoietic organs. They mechanisms of primitive, as well as highly developed
vary considerably in lifespan and phenotype, depend- multicellular organisms. 3,4,7 The Nobel awards of 2011 to
ing on their origin and local microenvironment. Mature Bruce Beutler,9 Jules Hoffman,10 and Ralph Steinman11
M f are highly phagocytic, relatively long-lived cells that reflect the paradigm shift of immune recognition from
are adaptable in their biosynthetic responses to antigens lymphocytes to innate antigen-presenting cells (APCs).
and microbial stimuli. The functions of M f within tis- Metchnikoff already clearly stated the link between cap-
sues are homeostatic, regulating the local and systemic ture of infectious microorganisms by the spleen and sub-
milieu through diverse plasma membrane receptors and sequent appearance of reactive substances (antibodies) in
varied secretory products. They react to, and themselves the blood, although mistakenly ascribing their produc-
generate, signals that inf luence growth, differentiation, tion to the phagocytes themselves. The importance of
and death of other cells, recognizing and engulfing se- systemic clearance of particles by Mf, especially Kupffer
nescent and abnormal cells. These activities contrib- cells in liver and other endothelial cells, was enshrined in
ute substantially to recognition and defense functions the term reticuloendothelial system. Although it was re-
against invading microorganisms, foreign particulates, jected by influential investigators in the field in favor of
and other immunogens. Innate immune functions of M f the term mononuclear phagocyte system, the appreciation
complement their contributions to acquired humoral and that sinuslining Mf in liver and elsewhere share common
cellular immunity, in which they regulate activation of T- properties with selected endothelial cells is worth preserv-
and B-lymphocytes; this is achieved in part through their ing. 3 Earlier studies by Florey and his students, including
specialized derivatives, dendritic cells (DCs) of myeloid Gowans, established that circulating monocytes give rise
origin. M f, with or without DCs, process and present anti- to tissue Mf. Van Furth and his colleagues investigated
gen; produce chemokines and cytokines such as interleu- the life history of Mf by kinetic labeling methods; sub-
kin (IL)-1, IL-6, IL-12, IL-18, IL-23, tumor necrosis factor sequently, the development of membrane antigen mark-
(TNF)- a , IL-10, and TGF β ; and phagocytose apoptotic ers facilitated a more precise defi nition of specialized Mf
and necrotic cells. Acting directly or under the inf luence subpopulations in tissues such as brain. The appearance
of other immune cells, M f capture extra- and intracellu- and potential importance of M f during development
lar pathogens, eliminate invaders, and deliver them to ap- also became evident as a result of sensitive immunocyto-
propriate subcompartments of lymphoid organs. As key chemical methods. Morphologic and functional studies
regulators of the specific as well as the natural immune by Humphrey and many others drew attention to strik-
response, M f boost as well as limit induction and effector ing diversity among Mf-like cells in secondary lymphoid
mechanisms of the specific immune response by positive organs, especially within the marginal zone of the spleen,
and negative feedback. where complex particulates and polysaccharides are cap-
The properties and roles of DCs, especially in antigen tured from the circulation.
presentation, are described in detail elsewhere in this vol- The era of modern cell biology impinged on Mf studies
ume. Here, we focus on other members of the Mf lineage, following the studies of Cohn,12 Hirsch, and their colleagues.
consider their interrelationship, and outline specialized Their work touched on many aspects of cell structure and
properties that underlie their roles in the execution and function, including phagocytosis (the zipper mechanism of
regulation of immune responses. A number of texts and Silverstein), fluid- and receptor-mediated endocytosis, se-
presentations deal with the history and broad aspects of Mf cretion, and antimicrobial resistance. Isolation and in vitro
immunobiology.1–8 culture systems became available for cells from mice and
humans, especially after the identification of specific growth
and differentiation factors such as colony-stimulating fac-
SOME LANDMARKS IN THE STUDY tor-1 (CSF-1). It is perhaps fitting that the earliest known
OF MACROPHAGES natural knockout (ko) affecting Mf, a natural mutation
Our understanding of Mf developed in parallel with in the op gene in the osteopetrotic mouse, should involve
the growth of immunology as an experimental science. CSF-1.13 Cell lines retaining some but not all features of
448

CHAPTER 19 MACROPHAGES AND PHAGOCYTOSIS | 449
mature Mf have been useful for many biochemical and tropic for Mf by virtue of their expression of cluster of
cellular studies. Macrophages and dendritic cells can be de- differentiation (CD)4, chemokine coreceptors, and DC-
rived from embryonic stem cells and induced pluripotent SIGN, a C-type lectin also expressed by DCs. Although
cells by growth in appropriate culture conditions and trans- Mf had been implicated by earlier workers such as Mims
fection of selected transcription factors. as important in antiviral resistance generally, their role in
The role of Mf as antigen-processing cells able to initi- this regard was neglected before the emergence of HIV as
ate adaptive immune responses had false trails (“immuno- a major pathogen.
genic ribonucleic acid [RNA]” was thought to be involved Many molecules have been identified as important in
at one time) and encompassed early genetic strategies (Mf Mf functions in immunity and serve as valuable markers
of mice selected for high antisheep erythrocyte antibody to study their properties in mice and humans. These in-
responses by Biozzi and colleagues displayed enhanced clude Fc and complement receptors, which are important
degradative properties; adherent cells from defi ned guinea in opsonic phagocytosis, killing, and immunoregulation;
pig strains were shown to play an important role in major scavenger receptors originally implicated in foam cell for-
histocompatibility complex [MHC] Ia-restricted anti- mation and atherogenesis by Brown and Goldstein; nonop-
insulin responses). For many years, the APC functions of sonic lectin receptors, such as the mannose receptor (MR)
adherent cells were highly controversial as promoted by and b-glucan receptor (dectin-1) and secretory products
Unanue, who concentrated on intracellular processing by such as lysozyme, neutral proteinases, TNFa, chemokines,
Mf, and Steinman, who discovered the specialized role of and many other cytokines. A range of membrane antigens
“DCs” in antigen presentation to naive T-lymphocytes. The expressed by human and rodent mononuclear phagocytes
importance of Mf as effector cells in immunity to intra- has been characterized and reagents made available for fur-
cellular pathogens such as Mycobacterium tuberculosis was ther study of Mf in normal and diseased states. Recently,
recognized early by Lurie and Dannenberg. Mackaness the role of deoxyribonucleic acid (DNA)-binding tran-
used Listeria monocytogenes and bacille Calmette-Guérin scription factors including members of the NF-k B and
(BCG) infection in experimental models and developed the ETS (Pu-1) families has received increased attention in the
concept of Mf activation as an antigen-dependent but im- study of differential gene expression by Mf. Gene inacti-
munologically nonspecific enhancement of antimicrobial vation has confi rmed the important role of many of these
resistance. The subsequent delineation of T-lymphocyte molecules within the intact host, and use has been made of
subsets and characterization of interferon (IFN)- g as the cell-specific or conditional ko to uncover the role of Mf in
major cytokine involved in macrophage activation, includ- immunologic processes. Naturally occurring inborn errors
ing MHC II induction, merged with increasing knowledge in humans such as the leukocyte adhesion deficiency syn-
of the role of reactive oxygen and, later, nitrogen metabo- drome and chronic granulomatous disease have contributed
lites as cytotoxic agents. The role of virus-infected Mf as to the analysis of important leukocyte functions, including
MHC I–restricted targets for antigen-specific CD8 + killer those of Mf, in host resistance to infection. Mutations in a
cells was part of the initial characterization of this phe- monocyte-expressed gene (nucleotide oligomerization do-
nomenon by Zinkernagel and Doherty. D’Arcy Hart was an main [NOD]-2), involved in cytosolic sensing of microbial
early investigator of the intracellular interactions between products and NF-k B activation, have been implicated in
Mf and invaders of the vacuolar system, especially myco- a subset of individuals with an enhanced susceptibility to
bacteria, which survive within Mf by inhibiting acidifica- Crohn disease. The validity of murine ko models for human
tion and phagosomelysosome fusion, thus evading host genetic deficiencies has been confi rmed for key molecules
resistance mechanisms. Mouse breeding studies by several involved in Mf activation, such as IFNg and IL-12. N-ethyl
groups defi ned a common genetic locus involved in resis- N-nitrosourea mutagenesis has begun to reveal new mac-
tance to BCG, Leishmania , and Salmonella organisms. The rophage innate immune functions, as has increasing appli-
host phenotype was shown to depend on expression in Mf cation of system biology tools for microarray, proteomic,
and, many years later, the gene (termed N-ramp for natu- epigenetic, and microRNA analysis.
ral resistance-associated membrane protein) was identified
by positional cloning by Skamene, Gros, and their col-
PROPERTIES OF MACROPHAGES AND THEIR
leagues. Positional cloning by Beutler and associates led
to the identification of the gene responsible for lipopoly-
RELATION TO IMMUNE FUNCTIONS
saccharide (LPS) resistance in particular mouse strains. Introduction
Together with studies by Hoffmann and his colleagues on Mf participate in the production, mobilization, activation,
the toll pathway in Drosophila, this work resulted in an ex- and regulation of all immune effector cells. They interact
plosion of interest in the identification of mammalian toll- reciprocally with other cells while their own properties are
like receptors (TLRs) and their role in innate immunity to modified to perform specialized immunologic functions.
infection. At the same time, it became apparent that some As a result of cell surface and auto- and paracrine interac-
malignant tumors contain macrophage populations that tions, Mf display marked heterogeneity in phenotype,14,15
may favor their growth. a source of interest and considerable confusion to the in-
This brief survey concludes with the identification of vestigator. Increasing knowledge of cellular and molecular
Mf as key target cells for infection, dissemination, and properties of Mf bears strongly on our understanding of
persistence of human immunodeficiency virus (HIV), their role in the immune response. These will be reviewed

briefly, with emphasis on functional significance, and atten- considerable further replication at local sites of inflamma-
tion will be drawn to unresolved and controversial issues. tion.19 Growth and differentiation are tightly regulated by
specific growth factors and their receptors (eg, IL-3, CSF-1,
granulocyte-macrophage [GM]-CSF/IL-34, IL-4, IL-13) and
Growth and Differentiation: Life History and Turnover inhibitors (eg, IFNa / b, transforming growth factor [TGF]-
In contrast to T- and B-lymphocytes, monocytes from b, leukemia inhibitory factor), which vary considerably in
blood give rise to terminally differentiated Mf that can- their potency and selectivity. These processes are modu-
not recirculate or reinitiate DNA replication except in a lated by interactions with adjacent stromal and other cells
limited way. Unlike other myeloid granulocytic cells, Mf (eg, through c-kit/ligand and Flt-3/ligand interactions). The
can be long lived and retain the ability to synthesize RNA growth-response of the target cell to an extrinsic stimulus
and protein to a marked extent, even when in a relatively decreases progressively and markedly (from 108 or more to
quiescent state as “resident” cells. These are distributed 100) during differentiation from stem cell to committed pre-
throughout the tissues of the body and constitute a possible cursor to monoblast, monocyte, and Mf, yet even the most
alarm-response system, but they also mediate homeostatic terminally differentiated Mf such as microglial cells can be
and poorly understood trophic functions. Following “reactivated” to a limited extent by local stimuli. Elicited/
inflammatory and immune stimuli, many more monocytes activated Mf respond more vigorously than resident Mf to
can be recruited to local sites and give rise to “elicited” or growth stimuli in vivo and in vitro, but the molecular basis
“immunologically activated” Mf with altered surface, se- for their enhanced proliferation is unknown.
cretory, and cytotoxic properties. The origins of Mf from Although this general picture of blood monocyte-to-
precursors are well known: from yolk sac (and possibly ear- tissue Mf differentiation has been accepted for some time
lier paraaortic progenitors), migrating to fetal liver, then as a result of parabiosis, adoptive transfer, and irradia-
spleen and bone marrow, before and after birth.16 Yolk sac tion-reconstitution experiments, recent studies in mouse
precursor cells may contribute to the establishment of se- and man have demonstrated monocyte heterogeneity and
lected tissue macrophages such as Langerhans cells in the distinct properties,20–23 with a subpopulation remaining
adult.17 In the fetus, mature Mf proliferate actively during within the vasculature, to perform a patrolling function.
tissue remodeling in developing organs. In the normal adult, Our understanding of DCs and osteoclast differentiation is
tissue Mf do not self-renew extensively except in specialized still compatible with a relatively simple model (Fig. 19.1) in
microenvironments such as epidermis,18 nervous system, which major Mf populations in mouse tissues can be char-
or lung; after TH2-type parasitic infection, there can be acterized by selected antigen markers such as F4/80 (Emrl,
FIG. 19.1. Differentiation of mononuclear phagocytes based on antigen markers FA-11 (macrosialin, murine
CD68) and F4/80. See text for detailed discussion.

a member of a family of EGF-TM7 molecules) and macro- to V-CAM, cannot be induced on terminally differenti-
sialin (CD68), a pan-Mf endosomal glycoprotein related to ated peritoneal Mf if these are placed in the same culture
the lysosome-associated membrane protein (LAMP) family. system. This contrasts sharply with the ready adaptation
The DCs of myeloid origin (see elsewhere in this volume) of many tissue Mf to conventional cell culture conditions,
share many properties with monocyte/macrophages,24 but when the cells often adopt a common, standard phenotype.
are specialized to capture, process, and present antigens to Irreversible stages of Mf differentiation may therefore occur
naïve lymphocytes. Circulating precursors of DCs and and in specialized microenvironments in vitro or in vivo.
macrophages are normally present in the mononuclear frac- Little is known about determinants of Mf longevity and
tion of blood in small numbers25 ; studies in the mouse may turnover. Growth factors such as CSF-1 enhance Mf sur-
not reflect the origin and differentiation of precursor cells vival and prevent induction of an apoptotic program. The
in humans.26 Monocytes that have crossed the endothelium expression of Fas-L and Fas on Mf has been less studied
may be induced to “reverse migrate” into the circulation by than on lymphocytes; they and other members of the TNF
selected stimuli in tissues.27 Finally, the mouse spleen has and its receptor family may play a major role in determin-
been shown to serve as a reservoir of monocyte/Mf for re- ing Mf survival, especially in induced populations, where
cruitment to sites of inflammation.28 cell turnover is markedly enhanced. Tissue Mf vary greatly
Circulating mononuclear precursors for osteoclasts are in their lifespan, from days to months. Apart from inflam-
less defined and differentiate into mononucleate cells, re- matory and microbial stimuli, local and systemic environ-
cruited in response to sphingosine-1-phosphate to bone, mental factors such as salt loading and hormones, including
for example,29 where they fuse to form multinucleate bone- estrogen, are known to influence Mf turnover.
resorbing osteoclasts.30 Local stromal cells, growth factors
such as CSF-1, steroids (vitamin D metabolites), and hor- Tissue Distribution and Phenotypic Heterogeneity
mones (eg, calcitonin, for which osteoclasts express recep- of Resident Macrophages in Lymphoid and
tors) all contribute to local maturation. Osteoprotegerin, Nonlymphoid Organs
a naturally occurring secreted protein with homology The use of the F4/80 plasma membrane antigen made it
to members of the TNF-receptor family, interacts with possible to detect mature Mf in developing and adult mu-
TRANCE, a TNF-related protein, to regulate osteoclast dif- rine tissues and define their anatomic relationship to other
ferentiation and activation in vitro and in vivo. cells in endothelium, epithelium, and connective tissue, as
Use of antigen markers such as CD34 on progenitors, well as the nervous system.32,33 Subsequently, other mem-
CD14 and CD16 on monocytes, and chemokine receptors brane antigens,34 macrosialin, sialoadhesin, and others were
and multichannel fluorescein-activated cell sorter analysis identified as useful markers for Mf in situ (Table 19.1). Mf
have made it possible to isolate leukocyte subpopulations subpopulations in different tissues display considerable het-
and study their progeny and differential responses in differ- erogeneity in expressing these and selected receptor anti-
ent mouse tissues and models of disease.31 The mononuclear gens (eg, complement receptor [CR]3 and class A scavenger
fraction of blood may contain precursors of other tissue receptor [SR-A]), drawing attention to unknown mecha-
cells, including mesenchymal stem cells able to synthesize nisms of homing, emigration, and local adaptation to par-
matrix proteins such as collagen, and some endothelial cells. ticular microenvironments. From the viewpoint of immune
Perhaps the mysterious follicular dendritic cells (FDCs) responses, a few aspects deserve comment.
with mixed hemopoietic and mesenchymal properties fall
in this category. Fetal Liver and Bone Marrow
The large-scale production of immature and mature DC- Mature Mf form an integral part of the hemopoietic mi-
like cells from bulk monocytes in cytokine-supplemented croenvironment and play a key role in the production,
culture systems (IL-4, GM-CSF, TNFa ) has revolution- differentiation, and destruction of all hemopoietic cells. The
ized the study of these specialized APCs. Individually, the fetal liver is a major site of definitive erythropoiesis from
same cytokines give rise to Mf-like cells, and early during midgestation.16 The bone marrow becomes active in the pro-
in vitro differentiation, the cellular phenotype is reversible. duction of hemopoietic cells from shortly before birth, and
Later, when mature DCs with high MHC II, APC function, Mf are a prominent component of the hemopoietic stroma
and other characteristic markers are formed, differentiation throughout adult life. Mature “stromal” Mf in fetal liver
is irreversible. This process is independent of cell division, and adult bone marrow express nonphagocytic adhesion
although earlier progenitors in bone marrow and GM-CSF– molecules such as sialoadhesin (Sn), an immunoglobulin
mobilized blood mononuclear cells can be stimulated to (Ig)-superfamily sialic acid-binding lectin (Table 19.1), and
multiply, as well as differentiate, in vitro. These examples of the EbR referred to previously, which is also involved in ad-
terminal differentiation observed with DCs and osteoclasts hesion of developing myeloid and possibly lymphoid cells
may extend to other specialized, more obvious Mf-like cells. (Fig. 19.2). VLA-4 has been implicated as a ligand for EbR.
Mature Mf can be derived by growth and differentiation in Ligands for Sn include CD43 on developing granulocytes
steroid-supplemented media in Dexter-type long-term bone and on lymphocyte subpopulations. Sn clusters at sites of
marrow cultures that contain stromal fibroblasts and he- contact between stromal Mf and myeloid but not erythroid
mopoietic elements. These Mf express adhesion molecules cells. Chemokines are able to induce polarized expres-
responsible for divalent cation-dependent cluster formation sion of adhesion molecules such as intercellular adhesion
with erythroblasts (EbR). This receptor, possibly related molecules and CD43 in leukocytes, but the significance

TABLE 19.1 Selected Differentiation Antigens Used to Study Murine Macrophage Heterogeneity
Ab Ag Structure Ligands Cellular Expression Function Comment
F4/80 F4/80 (EMR1) EGF-TM7 ? Mature Mf, absent Peripheral Useful marker
T areas tolerance development, CNS
FA-11 Macrosialin Mucin-LAMP OX-LDL Pan-Mf, DC Late endosomal Glycoforms regulated
(CD68) by inflammation
and phagocytosis
5C6 CR3 (CD11b, b 2-integrin iC3b, ICAM Monocytes, mi- Phagocytosis, Important in inflamma-
CD18) croglia, PMN, adhesion tory recruitment,
NK cells PMN apoptosis
2F8 SR-A (I, II) Collagenous, Polyanions, LTA, Adhesion, Protects host against
type II LPS, bacterial endocytosis LPS-induced shock
glycoprotein proteins
Isoforms differ, Modified proteins
cysteine-rich b -amyloid
domain apolipoprotein
A, E
Mf, sinusoidal Phagocytosis Promotes athe-
endothelium of apoptotic rosclerosis
cells and
bacteria
SER-4 Sn (Siglec-1) Ig superfamily Sialyl glycoconju- Subsets tissue Mf Lectin Strongly expressed
gates (eg, CD43)
3D6 Marginal zone
metallophils in
spleen and sub-
capsular sinus of
lymph nodes
CNS, central nervous system; DC, dendritic cell; ICAM, intercellular adhesion molecule; Ig, immunoglobulin; LAMP, lysosome-associated membrane protein; LPS, lipopolysaccharide; LTA,
lipoteichoic acid; Mf, macrophages; NK, natural killer; OX-LDL, oxidised low density lipoprotein; PMN, polymorphonuclear neutrophil; Sn, sialoadhesin; SR-A, type A scavenger receptor.
FIG. 19.2. Associations of tissue macrophages with

other hemopoietic cells to illustrate variations on a
common theme. See text for details.

of altered ligand distribution for interactions between Mf phenotype, life history, and functions (Fig. 19.3). Mf are
and bound hemopoietic cells is unknown. Adhesion of central to antigen capture, degradation, transport, and pre-
immature cells to stromal Mf may play a role in regulating sentation to T- and B-lymphocytes, and contribute substan-
their intermediate stages of development before release into tially to antimicrobial resistance. Recent work has unveiled
the bloodstream, whereas fibroblasts in the stroma associ- an unexpected role in facilitating activation of other lym-
ate with earlier progenitors, as well as with Mf. Discarded phocyte subsets, such as invariant natural killer T cells37; CD
nuclei of mammalian erythroid cells are rapidly engulfed by 169 + macrophages also activate CD8 T cells in response to
stromal Mf, but the receptors involved in their binding and dead cell–associated antigens in lymph nodes and by trans-
phagocytosis are unknown. Mf also phagocytose apoptotic ferring antigen to DCs in the spleen. Because other hemo-
hemopoietic cells generated in bone marrow, including large poietic and secondary lymphoid organs can replace many of
numbers of myeloid and B cells. We know little about the these functions after maturation of the immune system, the
plasma membrane molecules and cytokine signals operating unique properties of the spleen have been mainly recognized
within this complex milieu, but it is clear that stromal Mf in the immature host and in immune responses to complex
constitute a neglected constituent within the hemopoietic polysaccharides. Splenectomy in the adult renders the host
microenvironment. susceptible to infection by pathogenic bacteria such as pneu-
mococci that contain saccharide-rich capsular antigens; the
Thymus marginal zone of the spleen in particular may play an essen-
Apart from their remarkable capacity to remove apoptotic tial role in this aspect of host resistance.
thymocytes, the possible role of Mf in positive and negative The properties of Mf in the unstimulated mature mouse
selection of thymocytes has been almost totally overlooked; spleen are different according to their localization in red
more attention has been given to local DCs and their spe- or white pulp and the marginal zone. Mf are intimately
cialized properties. Mature Mf with unusual features are associated with the specialized vasculature. Species differ-
also present in cortex and medulla. Clusters of viable thy- ences in splenic anatomy and phenotype are well recog-
mocytes and Mf can be isolated from the thymus of young nized, although Mf display broadly common features in
animals by collagenase digestion and adherence to a substra- humans and rodents. Subpopulations of Mf, DCs, and cells
tum (see Fig. 19.2). The nonphagocytic adhesion receptors with mixed phenotypes have been characterized by in situ
responsible for cluster formation are more highly expressed analysis by antigen markers, liposome or diphtheria toxin-
by thymic than other Mf, but their nature is unknown (N. depletion studies, various immunization and infection pro-
Platt, unpublished observations). These Mf also express tocols, and cytokine and receptor gene ko models in the
MHC class II antigens and other receptors such as the SR-A mouse. The results raise questions about the dynamics and
(see subsequent discussion), which contributes to phagocy- molecular basis of cell production, recruitment, differen-
tosis of apoptotic thymocytes in vitro, but is redundant in tiation, emigration, and death within each distinct splenic
vivo; other markers, such as the F4/80 antigen, are poorly compartment. Cell isolation methods are still primitive in
expressed in situ but can be readily detected after cell isola- correlating in vitro properties with those of Mf subpopula-
tion. A striking difference between thymic and several other tions in vivo and remain an important challenge. Detailed
tissue Mf subpopulations is their independence of CSF-1; aspects of splenic architecture, DC origin and function, and
the CSF-1–deficient op/op mouse lacks osteoclasts and some T- and B-lymphocyte induction and differentiation are de-
Mf populations, including monocytes, peritoneal cells, and scribed elsewhere in this volume. Here, some features of Mf
Kupffer cells, but contains normal numbers of thymic Mf, as in the normal and immunoreactive organ are highlighted.
well as DCs and selected Mf in other sites. A second ligand
for the CSF-1 receptor Fms, IL-34, may account for CSF-1 Marginal Zone Macrophages
independence.35 Factors involved in constitutive recruitment The marginal zone of spleen consists of a complex mixture of
of thymic Mf are unknown; following death of thymocytes resident cells (reticular and other fibroblasts, endothelium),
induced by ionizing radiation or glucocorticoids, intensely Mf, DCs, and lymphoid cells, including subpopulations of
phagocytic Mf appear in large numbers; it is not known what B-lymphocytes. It constitutes an important interface with
proportion arises locally and by recruitment. the circulation that delivers cells, particulates, or soluble
molecules directly into the marginal sinus or via the red
Spleen pulp. Resident Mf are present as specialized metallophilic
From the viewpoint of the Mf, the spleen is perhaps the most cells in the inner marginal zone, and other Mf are found
complex organ in the body.36,37 It contributes to hemopoi- in the outer zone; the latter may be more phagocytic. Sn is
esis, which persists postnatally in some species or can be in- strongly expressed by the marginal metallophils, compared
duced by increased demand, can serve as a reservoir as noted with only weak expression in red pulp and virtual absence
previously, and contributes to the turnover of all blood ele- in the white pulp. Sn + cells appear in this zone 2 to 4 weeks
ments at the end of their natural lifespan. The spleen fi lters postnatally in the mouse as the white pulp forms. Liposomes
a substantial proportion of total cardiac output, captures containing clodronate, a cytotoxic drug, can be delivered sys-
particulate and other antigenic materials from the blood- temically and deplete Sn + cells and other Mf ; regeneration
stream, and plays an important role in natural and acquired of different Mf subpopulations in spleen occurs at different
humoral and cellular immunity. The organ is rich in sub- times, and this procedure has been used to correlate their re-
populations of Mf that differ in microanatomic localization, appearance with distinct immunologic functions. Marginal

FIG. 19.3. Microheterogeneity of macrophages in spleen,

resting, and antigen-stimulated lymph nodes. See text
for markers and details.
zone Mf lack F4/80 but may express an undefi ned ligand for can be induced to migrate into white pulp as described after
F4/80 on circulating activated DCs, which mediates periph- LPS injection; alternatively, they may shed complexes of sol-
eral tolerance to anterior chamber or gut-derived antigens.38 uble MR-glycoprotein ligand for transfer to other CR-Fc +
Marginal zone Mf express phagocytic receptors, such as cells, which may be resident or newly recruited mononuclear
SR-A, which is more widely present on tissue Mf, as well as cells. Finally, the marginal metallophilic Mf population de-
MARCO, a distinct collagenous scavenger receptor, which is pends on CSF-1 for its appearance and on members of the
almost exclusively present on these Mf in the normal mouse. TNF receptor family, as shown with op/op and experimen-
The structures and possible role of these pattern recognition tally produced ko mice.
receptors in uptake of microbes are discussed subsequently.
In vivo studies have shown that an Mf lectin, the MR, may White Pulp Macrophages
be involved in transfer of mannosylated ligands to the site of The F4/80 antigen is strikingly absent on murine white
an immune response in the white pulp.39 The MR contains pulp Mf, which do express FA-11 (macrosialin), the murine
a highly conserved cysteine-rich domain, not involved in homolog of CD68. Actively phagocytic Mf express this in-
mannosyl recognition, that reacts strongly with ligands on tracellular glycoprotein in abundance compared with DCs.
a subset of marginal metallophilic Mf, sulfated glycoforms After uptake of a foreign particle (eg, sheep erythrocytes
of Sn, and CD45, among others; this has been demonstrated or an infectious agent, such as BCG or Plasmodium yoel-
with a chimeric probe of the cysteine-rich domain of the lii), white pulp Mf become more prominent, although it
MR and human Fc (CR-Fc) and by immunochemical analy- is not known whether there is migration of cells into the
sis of tissue sections and affinity chromatography of spleen white pulp or transfer of phagocytosed material and re-
ligands. After immunization, this probe additionally labels activation of previous resident Mf. Tingible body Mf ap-
undefined cells in the FDC network of germinal centers, as pear to be involved in uptake and digestion of apoptotic
well as tingible body Mf. It is possible that marginal zone Mf B-lymphocytes.

Red Pulp Macrophages resident and recruited macrophages and DC in the mouse
These express F4/80 antigen and MR strongly and in the intestine.43 The role of the microbiome44 has received a great
mouse include stromal-type Mf involved in hemopoiesis. deal of attention in regard to innate cell phenotype and
Extensive phagocytosis of senescent erythrocytes results in epithelial integrity in the gut.
accumulation of bile pigments and ferritin, and play an im-
portant role in iron turnover40 and tolerance.41 The role of Nonlymphoid Organs
various phagocytic receptors in clearance of host cells and Regional F4/80 + and CD68 + Mf are well described in liver
pathogens by red pulp Mf requires further study. (Kupffer cells), dermis, neuroendocrine and reproductive
There is no evidence that Mf, other than interdigitating organs, and serosal cavities, where they are able to react
DCs, associate directly with CD4 + T-lymphocytes in the to systemic and local stimuli. In the lung, alveolar Mf are
normal spleen. Following infection by BCG, for example, or strongly CD68 + but only weakly F4/80 + and are distinct
by other microorganisms such as Salmonella, there is mas- from interstitial Mf and intraepithelial DCs. In the lamina
sive recruitment and local production of Mf, many of which propria of the intestine, Mf display a downregulated pheno-
associate with T-lymphocytes. Newly formed granulomata type, ascribed to TGFb of local origin.45 In addition, resident
often appear first in the marginal zone (focal accumulations Mf are found throughout connective tissue and within the
of activated Mf and activated T cells). As infections spread interstitium of organs, including heart, kidney, and pan-
into the white and red pulp, the granulomata become con- creas. These cells vary greatly depending on their local mi-
fluent and less localized, obscuring and/or disrupting the croenvironment; for example, in the central nervous system,
underlying architecture of the spleen. The possible role of microglia within the neuropil differ strikingly from Mf in
activated Mf in T-cell apoptosis and clearance in spleen has the meninges or choroid plexus.46 Perivascular Mf in the
not been defined. brain can be distinguished from resident microglia by their
expression of endocytic receptors (eg, the SR-A and MR,
Lymph Nodes and of MHC I and II antigens). Microglia are highly rami-
F4/80 antigen is relatively poorly expressed in lymph node fied, terminally differentiated cells of monocytic origin;
(see Fig. 19.3), but many macrosialin (CD68) + cells are pres- many Mf markers are downregulated. Their phenotype is
ent. The subcapsular sinus is analogous to the marginal zone influenced by the blood–brain barrier, normally absent in
and contains strongly Sn + cells; this is the site where affer- circumventricular organs, and disrupted by inflammatory
ent lymph enters, containing antigen and migrating DCs de- stimuli. Microglia can be reactivated by local LPS and neu-
rived from skin and mucosal surfaces. The medulla contains rocytotoxins; they are then difficult to distinguish from
Sn +, CD68 + Mf, which also express high levels of SR-A. newly recruited monocytes, which acquire microglial fea-
As in the spleen marginal zone, subcapsular sinus Mf are tures once they enter the parenchyma of the brain. Resting
strongly labeled by the CR-Fc probe. Following primary or microglia are unusual among many tissue Mf in that they
secondary immunization, the staining pattern moves deeper constitutively express high levels of CR3 and respond to CR3
into the cortex and eventually becomes concentrated in ger- ligands, such as mAb, by induced DNA synthesis and apop-
minal centers. The kinetics of this process strongly suggests tosis. In other sites, such as lung and liver, CR3 expression
a transport process by Mf-related cells resembling antigen is a feature of recent myeloid recruitment, including mono-
transport cells described previously. CR-Fc + cells can be cytes. Resident Kupffer cells lack constitutive CR3 but ex-
isolated by digestion of lymph nodes and form clusters with press a novel CR implicated in clearance function.
CR-Fc– lymphocytes. Adoptive transfer has shown that flu- Resident tissue macrophages in human tissues express
orescein-activated cell sorter–isolated CR-Fc + cells resemble CD68 antigen, but their phenotypic diversity and microhet-
DCs in their ability to home to T-cell areas and to present erogeneity in different organs remain poorly defi ned. Access
antigen to naive T and B cells. Overall, there is considerable to skin biopsies, bronchoalveolar lavage, and placenta, for
heterogeneity in the population of migratory APCs involved example, provides material for further analysis.
in antigen capture, transport, and delivery to T and B cells,
and it may turn out that specialized tissue Mf as well as
myeloid-type DCs can migrate in response to immunologic Enhanced Recruitment of Monocytes by Inflammatory
stimuli, especially TLR ligands.42 and Immune Stimuli: Activation in Vivo
In response to local tissue and vascular changes, partly
Peyer Patch induced by resident Mf during (re)activation by inflam-
Although less studied, the Mf in Peyer patch resemble the matory, infectious, and immunologic stimuli, monocytes
CD68 +, F4/80– cells described in spleen and white pulp are recruited from marrow pools and blood in increased
and in other T-cell–rich areas. They are well placed to in- numbers; they diapedese and differentiate into Mf with
teract with gut-derived antigens and pathogens taken up via altered effector functions as they enter the tissues.47 These
specialized epithelial M cells in the dome, and deliver an- Mf are classified as “elicited” when cells are generated in
tigens to afferent lymphatics, as myeloid DCs. These cells the absence of IFNg and as “immunologically activated”
are distinct from abundant F4/80 + cells in the lamina pro- after exposure to IFNg. Enhanced recruitment can also in-
pria found all the way down the gastrointestinal tract and volve that of other myeloid or lymphoid cells; selectivity of
may play a role in the induction of mucosal immunity. the cellular response depends on the nature of the evoking
Recent studies have described heterogeneous populations of stimulus (immunogenic or not), the chemokines produced,

and the receptors expressed by different leukocytes. Mf and cell activation along a continuous spectrum rather than true
other cells produce a range of different chemokines and ex- differentiation.
press multiple seven-transmembrane, G protein-coupled The migration and differentiation of newly recruited
chemokine receptors. The chemokines can also act in the monocytes once they have left the circulation are poorly
marrow compartment, especially if anchored to matrix and understood. They are able to enter all tissues, undergoing
glycosaminoglycans, may display other growth regulatory alterations in membrane molecules and secretory poten-
functions, and can control egress. Locally bound or soluble tial under the influence of cytokines and surface interac-
chemokines induce the surface expression and activity of tions with endothelial cells, leukocytes, and other local
adhesion molecules on circulating white cells, as well as di- cells. Phenotypic changes mentioned in the following sec-
recting their migration through and beyond endothelium. tion have been characterized by a range of in vitro and in
Feedback mechanisms from periphery to central stores and vivo studies. Well-studied examples include murine perito-
within the marrow stroma may depend on cytokines and neal Mf—resident, elicited by thioglycollate broth or biogel
growth factors such as macrophage inflammatory protein- polyacrylamide beads, and immunologically activated by
1a and GM-CSF, which inhibit or enhance monocyte pro- BCG infection. The latter provides a useful model of granu-
duction, respectively. The adhesion molecules involved loma formation in solid organs but does not fully mimic the
in recruitment of monocytes, originally defined by stud- human counterpart associated with M. tuberculosis infec-
ies in humans with inborn errors and by use of inhibitory tion. Granuloma Mf vary in their turnover and immune
antibodies in experimental animal models,48 overlap with effector functions and display considerable heterogeneity;
those of polymorphonuclear neutrophils and lymphocytes lesions contain recently recruited monocytes, mature, epi-
and include L-selectin, b2-integrins, especially CR3, CD31, thelioid Mf (described as secretory cells), and Langhans
an Ig-superfamily molecule, and CD99; additional mono- giant cells. Interactions with T-lymphocytes, other myeloid
cyte adhesion molecules for activated endothelium include cells, DCs, fibroblasts, and microorganisms yield a dynamic
CD44, vascular cell adhesion molecules, b1-integrins, and assembly of cells as the granuloma evolves, heals, and re-
newly described receptors such as EMR2 and CD97, mem- solves (see Fig. 19.2). Apoptosis and necrosis of Mf and other
bers of the EGF-TM7 family.33 The mechanisms of constitu- cells contribute to the balance of continued recruitment and
tive entry of monocytes into developing and adult tissues, local proliferation. The emigration of Mf rather than DCs
in the absence of an inflammatory stimulus, are unknown. from sites of inflammation is less evident, although it has
By contrast with the uncertain precursors of resident Mf become clear that elicited Mf within the peritoneal cavity,
and DC populations, distinct monocyte subsets have been for example, migrate actively to draining lymph nodes.49
implicated in the enhanced mobilization and turnover in Gene ko models have confirmed the role of molecules
response to inflammatory, infectious, and metabolic stim- previously implicated in recruitment, activation, and granu-
uli, as noted previously (Fig. 19.4). Differential expression loma formation. These include the adhesion molecules listed
of the fractalkine receptor, CCR2, and other chemokine re- previously, their ligands, such as intercellular adhesion mol-
ceptors, together with antigen markers (Gr-1 in mouse and ecule-1, and key cytokines such as IFNg, IL-12, IL-23, and
CD14 in human), have made it possible to define monocyte TNFa, as well as their receptors. Antimicrobial resistance
heterogeneity. Although such subsets seem to be conserved and Mf cytotoxicity resulting from production of reactive
across several species, their properties may reflect stages of oxygen and nitrogen metabolites are now accessible to study
Bone Marrow Peripheral Blood Tissues
CD14+CD16+ STEADY STATE

Ly-6lo/Gr-1lo
Resident Tissue MØ?
Progenitor CCR2- e.g., Splenic MØ, Kupffer cells,
Alveolar MØ, Microglia, Osteoclasts
CD62L -
CX3CR1hi Resident Tissue DC?
Monoblast INFLAMMATION
Pathogen
CD14++ CD16- MØ clearance
Ly-6hi/Gr-1hi Resolution of
Promonocytes
CCR2+ inflammation
CD62L + Antigen
CX3CR1lo DC presentation FIG. 19.4. Distinct monocyte subsets give rise to
inflammatory and resident macrophages and den-
dritic cells. The origin of different resident subpop-
Monocytes ulations is not clear. See text for further details.

in knockouts of the phagocyte oxidase and inducible nitric removers of dying cells? Do Mf contribute to recruitment,
oxide synthase. kos of membrane molecules of immuno- differentiation, and death of DCs at sites of inflammation
logic interest expressed by Mf and other cells include MHC before their migration to secondary lymphoid organs? Do
class II and I, CD4, and CD40L, other accessory molecules adjuvant-stimulated Mf interact with B-lymphocytes, di-
such as B7–1 and B7–2, and the Mf-restricted intracellular recting their migration into germinal centers? Are inter-
molecule N-ramp. actions of activated Mf with antibody and complement,
The study of inborn errors in humans50,51 and of disease through different Fc and complement receptors, implicated
models in genetically modified mice has brought insight in fine-tuning humoral responses? Are activated Mf them-
into essential, nonredundant contributions of molecules selves cytocidal for infected host cells, and to what extent
that regulate Mf activation in vivo, including immunopa- do they in turn interact with and provide targets for attack
thology syndromes such as septic shock and autoimmunity. by natural killer cells and cytotoxic T-lymphocytes? Study
Examples include myeloid antigens such as TREM-1/2,52 as- of a range of experimental models and disease processes in
sociated with DAP-12, receptor-ligand pairs such as CD200/ vivo should yield new insights, as well as extend and confirm
CD200 receptor,53 and suppressors of cytokine signaling mechanisms already defined in vitro.
proteins.54,55 TNFa is essential for host resistance to infec-
tion56 and also contributes to immunopathology. Highly
effective anti-TNFa therapy for chronic inflammatory dis- Phagocytic Recognition and Intracellular Infection
eases such as rheumatoid arthritis can result in reactivation The initiation and localization of an immune response de-
of latent tuberculosis. pends on recognition by Mf and other cells of particulate
The potential for innate,Th1- and Th2-type regulation agents or soluble proteins that are foreign or modified-self.67
of Mf demonstrated in vitro, and discussed subsequently, Phagocytic and endocytic recognition by Mf and DCs de-
can result in highly complex, often coexistent, heterogene- pends, in turn, on opsonic (mainly antibody, complement)
ity of Mf phenotype in situ (see Fig. 19.4). Although almost and nonopsonic pattern recognition receptors that interact
all granuloma Mf express lysozyme,57 only subpopulations with a range of related ligands (Fig. 19.5). Innate and acquired
express cytokines such as IL-1b, IL-6, and TNFa. Pro- and responses are thus interlinked. Different FcR are involved
anti-inflammatory cytokines, IL-12, IL-18, IL-10, and TGFb, in uptake and destruction of targets as well as in negative
produced by Mf themselves and other cells, modulate the regulation of effector functions.68 CRs69 are also heteroge-
phenotype of Mf in vivo. neous; CR3 interacts with C3-derived ligands formed by
Intravital imaging has provided insights into the dy- activation of the classical, alternate, or lectin pathways and
namics of interactions of myeloid and lymphoid cells in mediate phagocytosis, cell migration, and cell activation.
granuloma formation.58 Studies of human granulomatous, Other ligands include intercellular adhesion molecules. CR3
chronic inflammation provide opportunities to dissect ge- functions are modulated by fibronectin, via integrins, other
netic and acquired influences (eg, in NOD-2 deficiency in adhesion molecules, and inflammatory stimuli. FcR ligation
Blau syndrome59 and Crohn disease60). The mouse resis- and cross-linking activates tyrosine kinases such as syk that
tance protein lrgm1 (LRG-47) 61 has also been implicated in are essential for phagocytosis; CR3 signaling is less defi ned
pathogen defense.62 and may not trigger a respiratory burst or arachidonate re-
Apart from the local interactions outlined, Mf regulate lease, unlike FcR, thus favoring pathogen entry. Antibody-
systemic host reactions to immune and infectious stimuli mediated uptake targets an organism or soluble antigen to a
by producing circulating cytokines such as IL-6 and ara- different, degradative compartment70 (Fig. 19.6) and usually
chidonate- and other lipid-derived metabolites, includ- results in its neutralization and destruction, although en-
ing resolvins, that contribute to the resolution of acute hancement of infection can also occur in Mf.71 Flavivirus
inflammation.49 These mediators also act on neural and infection in the presence of specific antibody can result in
endocrine centers, crossing the blood–brain barrier, or are the dengue hemorrhagic shock syndrome. Immune com-
generated locally by reactive microglia and Mf. MicroRNAs plexes, with or without complement, localize antigens to
have been identified that play a role in resolution.63 FDCs and other FcR+ CR+ cells. Mf themselves are able to
Glucocorticosteroids are powerful immunomodulators64 produce all components of the complement cascade in sig-
and form part of a network that regulates monocyte recruit- nificant amounts at local sites, which may be less accessible
ment and Mf functions through circulating mediators such to circulating proteins made by hepatocytes.
as migration inhibition factor. Mf contain potent enzymes Nonopsonic receptors reacting directly with ligands
involved in steroid biosynthesis and catabolism.65 on microorganisms72 include CR3, lectins, especially the
Although the immunologic relevance of Mf-induced re- MR and b-glucan receptor, the scavenger receptors SR-A
sponses may seem evident, many aspects remain unclear. and MARCO, and the family of TLRs. MRs are present
For example, what is the role and phenotype of monocyte/ on Mf, DCs, and sinusoidal endothelium.39 They mediate
macrophages in the immune reconstitution inflammatory phagocytosis and endocytosis, including macropinocyto-
syndrome, associated with dual acquired immunodeficiency sis, and structurally resemble another multilectin, Dec 205,
syndrome/tuberculosis infection, recipitated by antiretrovi- present on DCs as well as tissue Mf and epithelial cells in
ral treatment66? Do Mf actively suppress or destroy activated thymus; carbohydrate recognition by the latter has not been
T-lymphocytes, thus contributing to regulation of immune demonstrated. The MR has eight C-type lectin domains,
responses and peripheral tolerance, or are Mf only passive homologous to the mannose-binding protein, a circulating

FIG. 19.5. Macrophage activation and the role of microbial stimuli and cytokines. See text for details.
hepatocyte-derived, acute-phase reactant. Mannose-binding phagocytosis of bacteria but is independently regulated, as

protein, also known as mannan-binding lectin, contains a discussed subsequently. MARCO80 and Mincle,81 a lectin-like
single lectin domain per polypeptide, which oligomerizes receptor, contribute to macrophage responses to trehalose
like other collectins to achieve multivalent interactions and dimycolate, a virulence factor of pathogenic M. tuberculo-
activate complement via associated serine proteases. MR ex- sis. An initial report of a phosphatidylserine receptor im-
pression on Mf is selectively down- and upregulated by IFNg plicated in the recognition of novel lipid ligands expressed
and IL-4/13, respectively. A possible role of the cysteine-rich on the surface of apoptotic cells was not confirmed. CD36
domain in transport of immunogenic glycopeptides within (thrombospondin receptor), vitronectin receptors, CD91,
secondary lymphoid organs has been proposed.73 and CD44 have all been implicated in the uptake of senescent
The b-glucan receptor, previously reported as dectin-1, polymorphonuclear neutrophils by Mf.82 Other opsonins
is related to C-type lectins and is responsible for phagocytic for apoptotic cell clearance include milk fat globule protein
recognition of unopsonized zymosan and for Mf activa- (lactadherin). A role for Mf SR-A in immune induction has
tion (Figs. 19.7 and 19.8).74 It contains an immunoreceptor not been demonstrated, but studies in SR-A ko mice have
tyrosine-based activation motif–like motif in its cytoplas- revealed an important inhibitory role in limiting TNFa pro-
mic domain that is essential for phagocytosis and induced duction by immunologically activated Mf. Wild-type, BCG-
secretory responses (eg, TNFa). It cooperates with TLR2 primed mice produce granulomata rich in SR-A+ Mf ; SR-A
and synergizes with other TLRs in cell activation, utilizing ko mice restrict growth of this organism and form normal
syk and CARD9,75 and can promote TH17 differentiation.76 granulomata containing activated, MHC II + Mf ; on addi-
It is essential for resistance to a range of fungal particles tional challenge with LPS, the ko mice die more readily than
in vivo, as shown by studies with ko mice and in human wild-type animals. TNFa levels in the circulation rise mark-
primary innate immunodeficiency.77 Dectin-2, a distinct edly because of unopposed triggering via CD14, a receptor
mannan-binding lectin, has also been implicated in TH17 for the LPS-binding protein, and contribute to septic shock,
differentiation.45,78 because blocking anti-TNF mAb protects these mice.
SR-A mediates endocytosis of modified proteins (eg, The family of TLRs consists of homo- or heterodimeric
acetylated lipoproteins) and selected polyanions, such as transmembrane molecules related to the IL-1 receptor,
apolipoprotein A1 and E,79 LPS, and lipoteichoic acid.72 In which are involved in innate immunity to microbial con-
addition, it can serve as an adhesion molecule and contrib- stituents and activation of Mf responses, and are discussed
utes to phagocytic clearance of apoptotic thymocytes and elsewhere in this volume.83 Downstream signaling depends
gram-negative as well as gram-positive bacteria. MARCO, on association with other soluble and membrane molecules,
a related collagenous receptor,72 mediates cell adhesion and as well as with intracellular proteins. MyD88, for example,

FIG. 19.6. Phagocytic Pathway in Macrophages. A: Heterophagy. Fc receptor-mediated phagocytosis. a: Microbes are coated with a variety
of opsonins, including complement, pentraxin 3, and antibodies. A number of receptors are involved in initial recognition of microbes and
induction of proinflammatory signaling (eg, the toll-like receptors [TLRs], and especially TLRs 2, 4, and 5), but these receptors are not phago-
cytic. Receptors involved in phagocytosis include the complement receptors, Fc receptors, and others. b: Fc receptor ligation initiates a sig-
naling cascade that results in actin polymerization and extension of the plasma membrane. Phagocytosis mediated by this method occurs
via the “zipper” mechanism (ie, sequential binding between the Fc receptors and their ligands along the length of the microbe). c: Fusion
with the early endosome results in a slight drop in pH that results in the uncoupling of receptors with their ligands. Receptor recycling is
facilitated by the Rab proteins, which also confer the ability to undergo subsequent fusion. The developing phagosome now contains major
histocompatibility complex (MHC) class II and TLRs, including those that signal from within the developing endosome (eg, TLR9). d: Fusion
with the late endosome results in the addition of the lysosome-associated membrane proteins, the accumulation of acid-resistant phospho-
lipids, and a subsequent drop in pH. e: On fusion with lysosomes, the low pH results in the activation of a number of proteolytic enzymes.
These are necessary for both direct antimicrobial activity and the creation of peptides for presentation via MHC class II. The phagolyso-
some is a highly oxidative environment, exposing the pathogen to destructive reactive oxygen and reactive nitrogen intermediates. f: Under
certain circumstances, the phagosome may contain markers specific for the endoplasmic reticulum (ER). The ER may contribute directly to
the creation of the phagocytic membrane in some circumstances (eg, phagocytosis of latex beads), or ER-derived vacuoles may contribute
MHC class I and other molecules to the endosomes or the developing phagolysosome. (continued)
has been implicated in many but not all TLR-induced sig- Particle uptake involves multiple membrane receptors, the cy-
naling resulting in transcription factor regulation, cell acti- toskeleton, bulk membrane flow, and remodeling, including
vation, or apoptosis. formation of a phagocytic synapse and signaling pathways85,86
Naturally occurring microbial ligands for these nonopsonic (see Fig. 19.887). Phagosome formation and maturation re-
receptors are still poorly defined; individual receptors mediate semble endocytic uptake, initiating Mf vesicle trafficking and
microbial binding and uptake of microorganisms, although recirculation, fusion with lysosomes, acidification, ion fluxes,
each contributes only part of total binding (see Fig. 19.6A).72,84 and digestion. Table 19.2 lists immunologic and other markers

FIG. 19.6. (continued) B: The autophagy pathway. Autophagy is a homeostatic process that can be further en-
hanced in macrophages in the presence of interferons or by starvation. Cytosolic proteins and organelles are
found in ER-derived cytoplasmic vesicles. These vesicles fuse with lysosomes, and the proteins and organ-
elles are degraded and recycled. Conventional wisdom states that endogenous and cytoplasmic proteins are
presented by MHC class I molecules, whereas exogenous peptides are presented by MHC class II molecules;
however, it has become clear that peptide presentation is altered considerably on induction of autophagy. The
presentation of peptides from intracellular and lysosomal source proteins is increased on MHC-II.
used to identify intracellular compartments.88 Guanosine tri- in sensing by NOD-like receptors, inflammasome assembly,
phosphate–binding proteins and complex signaling cascades activation of caspase-1, and processing and release of IL-1
play an important role in these dynamic events.62 A key issue b.95–97 Nucleic acid recognition results in cytoplasmic and
that needs to be resolved is how cell and receptor functions mitochondrial-associated protein signaling responsible for
are modulated so that microbial phagocytosis or invasion in- type 1 interferon gene expression (see Fig. 19.9).98 A novel
duces inflammatory responses, unlike the uptake of apoptotic ligand, c-di-AMP, secreted by cytosolic Listeria monocyto-
cells. The MHC II biosynthesis and subcellular localization genes, activates a host type I IFN response92 and may serve
and proteolytic processing of peptide antigens in vacuolar and as a sensing mechanism for many intracellular pathogens
cytosolic compartments of APCs are discussed elsewhere in (see Fig. 19.10).99 STING is a direct innate immune sensor
this volume. Cytokines, especially IL-4/13, IL-10, and IFNg, of cyclic di-GMP, another cytosolic metabolite of intracel-
influence endocytosis via MR-dependent and -independent lular microbial infection.100 It has recently become clear
pathways and selectively alter vesicle dynamics. that induction of autophagy and apoptosis by intracellular
Pathogens vary in using Mf plasma membrane mol- pathogens, including Mycobacterium tuberculosis, provides
ecules for entry and evasion of host defenses, and modify important host-protective responses101 (see Figs. 19.6B and
the composition of the resultant phagosome membrane 19.9A, B). For a review on autophagy in immunity and in-
(Fig. 19.9).89 Mycobacteria, for example, employ a range flammation, see Levine et al.102
of mechanisms to evade killing by Mf, including delayed Although the “canonical” entry pathway described here
maturation of phagosomes and inhibition of fusion with and illustrated in Figure 19.6A is used and modified by
lysosomes and acidification (Figs. 19.9 and 19.10).90,91 many pathogens, recent evidence has shown that organ-
Listeria monocytogenes escapes into the cytosol by disrup- isms such as Legionella pneumophila102,103 (see Fig. 19.9C,
tion of the phagosome membrane,92,93 whereas Leishmania D) and Brucella abortus104 induce vacuoles with novel mem-
multiplies in phagolysosomes.94 Humoral (antibody, brane components or colonize compartments derived from
complement) and cellular (IFNg) mechanisms overcome the Golgi apparatus and the endoplasmic reticulum. The
parasitization of Mf by diversion to lysosomes or induce relative contributions of plasma membrane and endoplas-
killing via oxygen/nitrogen (O/N)- dependent and other mic reticulum (see Fig. 19.6A) to vacuole formation varies
mechanisms. considerably, depending on the nature of the phagocytic
Entry of microbial constituents such as muramyldipep- cargo or invading pathogen.105 Opsonins such as antibody
tide from vacuolar compartments to the cytosol can result are able to divert the cargo to lysosomes. IFNg can induce

Membrane or Content Markers

Used to Identify Phagosomes
TABLE 19.2 as Resembling a Given Endocytic
Compartment or the Endoplasmic
Reticulum
Compartment Markers
Early endosome Membrane markers: TfR, Rab5, annexins, I,

II, and III
Proteases: immature cathepsin D
Late endosome Membrane markers: M6PR, Rab7, LAMP1,
LAMP2, CD63, CD68
Hydrolases: acid phosphatase, aryl
sulfatase, trimetaphosphatase
Proteases: cathepsin B, D, H, dipeptidyl
peptidase I and II
Phospholipid: LBPA
Lysosome Membrane markers: LAMP1, LAMP2, CD63
Hydrolases: acid phosphatase, aryl
sulfatase, trimetaphosphatase
Proteases: cathepsin B, D, H, dipeptidyl
peptidase I and II
ER Membrane markers: calnexin, calreticulin
Enzyme: glucose-6-phosphatase
CD, cluster of differentiation; ER, endoplasmic reticulum; LAMP, lysosome-associated
FIG. 19.7. Macrophages Rock! Macrophage engaging zymosan par- membrane protein; LBPA, lysobisphosphatidic acid; M6PR, mannose-6-phosphate
ticles. Macrophages are adherent but motile. They use focal adhe- receptor; TfR, transferrin receptor. From de Chastellier C. Electron microscopy.
sions to anchor themselves to the extracellular matrix. Membrane In: Cossart P, Boquet P, Normark, et al., eds. Cellular Microbiology, 2nd ed.
ruffles and lamellipodia form at the leading edge of the cell to enable Washington, DC: ASM Press, 2005:451, with permission.
them to crawl around, and they extend pseudopodia to explore their
environment and capture microbes and dead cells for phagocytosis.
Consequently, their morphology is highly variable and constantly
changing. © Helen Goodridge and David Underhill
CD148
CD45
β-glucan particle
e.g. yeast
FIG. 19.8. The Dectin-1 Phagocytic Synapse.

A: To permit sectin-1 signaling, cluster of clustered Dectin-1
differentiation (CD)45 and CD148 tyrosine
phosphatases (which negatively regulate
signal transduction downstream of dectin-1) Src Src Src Src Src Src Src Src Src Src
are excluded from the contact site upon Syk Syk Syk Syk Syk Syk Syk Syk Syk
detection of beta-glucan particles by dec-
tin-1 at the macrophage surface.The phago-
cytic synapse is similar to the immunological
synapse that forms between an antigen pre-
senting cell and a T cell. B: Confocal imag- A
C
ing of a macrophage encountering zymosan C
i Dectin-1 ii
particles (i) shows two phagocytic synapses 3 μm
CD45
(arrows), and a three-dimensional isosur-
face model (ii) of one of them (asterisk). *
C: A phagocytic synapse that formed upon
contact of a swollen Aspergillus fumigatus
conidium with a macrophage. Images from
4 μm CD45
Goodridge et al.,87 with permission from Dectin-1
Nature Publishing Group. B

A
A
B
B
FIG. 19.9. Interactions of Selected Intracellular Pathogens with the Phagocytic Pathway. A: Mycobacterium tuberculosis evades destruction
by subverting normal phagolysosome maturation. a: Phagocytosis of M. tuberculosis occurs via the complement pathway (although it may not
require direct binding of complement to the bacterium) and is characterized by (b) “sinking” phagocytosis (ie, very little filopodia formation or
actin polymerization). c: The M. tuberculosis–containing vacuoles contain markers of the early and late endosomes such as Rab5 and Nrampl
but are devoid of most lysosomal markers, including the lysosome-associated membrane proteins, and do not undergo normal acidification.
It has been proposed that the colocalization of some but not the normal allotment of endosomal markers can be explained by the concept of
“kiss and run” fusion. This implies that the early and late endosomes may have transient contact with the M. tuberculosis–containing vacu-
oles, and there may be a selective transfer of markers rather than a complete fusion. The high pH of the M. tuberculosis–containing vacuole
(pH 6.2) does not allow optimum loading of major histocompatibility complex (MHC) class II molecules, and thus they remain loaded with non-
mycobacterial peptides. Elevated pH also inhibits production of inducible nitric oxide synthase, which is required for killing. In the presence of
interferon (IFN)-g, normal acidification may be restored, resulting in destruction of the pathogen. B: M. tuberculosis–containing phagosomes
are targeted to the autophagy pathway on treatment with IFNg. IFNg treatment can both restore the normal process of acidification and alter
the expression of a number of endoplasmic reticulum proteins, the result of which is the targeting of M. tuberculosis–containing phagosomes
to the autophagosomes. The fusion between the M. tuberculosis–containing phagosomes and lysosomes results in an autolysosome with low
pH that destroys M. tuberculosis and results in M. tuberculosis peptides being presented via MHC class II. (continued)

D
D
FIG. 19.9. (continued) C: Legionella pneumophila survives intracellularly by subverting phagosome maturation at an early stage. a: The rec-
ognition and uptake of L. pneumophila are not well characterized, but phagocysis occurs via a “spiral” mechanism. b: Once inside the cell, L.
pneumophila uses a specialized secretion system (ie, Dot/Icm secretion system) to secrete proteins directly into the cytosol. These proteins
alter the morphology of the vacuole in a number of ways, for example, by actively recruiting vesicles in transit from the endoplastic reticulum
to the Golgi apparatus and inhibiting the fusion of lysosomes. c: The L. pneumophila–containing vacuole thus has many similarities to the
Golgi apparatus and endoplastic reticulum, and is rich in peptides, the primary carbon source for L. pneumophila. D: Macrophages enhance
autophagy in response to L. pneumophila. A consequence of residing in a vacuole that so closely resembles the endoplastic reticulum is that it is
subject to autophagy. L. pneumophila–containing phagosomes fuse with lysosomes, resulting in destruction of the pathogen and antigen presenta-
tion. Factors secreted from L. pneumophila cause macrophages to increase the number of autophagosomes, although autophagosomes containing
L. pneumophila mature more slowly, and thus it is believed that the bacteria encode factors to delay normal progression.

L. monocytogenes
LLO
M. tuberculosis
ESX-1
? MdrM
DNA/RNA ? Sting
Y. pseudotuberculosis sensor
nucleus
? ? local & NH2
IRF3 IRF3 IFNαβ systemic
Ysc N
type III TBK1 effects N
secretion IRF3 other genes
OH N N
Dot/lcm OH
autocrine P O
type IV ? effects O O
secretion DNA/RNA type I IFN O
sensor receptor
L. pneumophila O
O
O
O P
type VI secretion OH
OH
N
F. tularensis escape to cytosol N
N
N c-di-AMP
NH2
A B
FIG. 19.10. Host Pathways Leading to the Expression of Interferon (IF) Beta and Coregulated Genes. A: Virulent, but not avirulent, intracellular
pathogens activate a cytosolic surveillance pathway of innate immunity. Induction of IFN beta by all depicted bacterial species is MyD88
and TRIF-independent, indicating that toll-like receptors are not required. Sting and TBK1 are necessary for IRF3 activation.Wild-type Listeria
monocytogenes enters the cytosol and activates the expression of IFN beta, while LLO-minus mutants that fail to enter the cytosol do not.
Mutants in the multidrug efflux pump, MdrM, enter the cytosol, but induce threefold less IFN beta expression. The L. monocytogenes ligand is
c-di-AMP (B), while the receptor remains unknown. Mycobacterium tuberculosis resides in a phagosome that resembles early endosomes.
Mutants in the ESX-1 auxiliary secretion system and phthiocerol dimycocerosate synthesis fail to induce IFN beta expression. Both TBK-1
and IRF3 are required. The infectious spore from Histoplasma capsulatum induces the expression of IFN beta, whereas the parasitic yeast
form does not. Legionella pneumophila resides in a modified vacuolar compartment that has features of the endoplasmic reticulum. The Dot/
lcm-type IV secretion system is necessary for intracellular growth and, independently, for induction of IFN beta expression. Both MDA5 and
MAVS are required, suggesting that the ligand may be ribonucleic acid. Francisella tularensis needs to enter the cytosol to induce expression
of IFN beta. The putative type VI secretion system is required for entry into the cytosol and induction of IFN beta. In each case, nucleic acids
may be the ligands specifically recognized by cytosolic receptors, leading to the activation of IFN beta and co-regulated genes. Courtesy of
D.A. Portnoy and R.E. Vance. For further information see Vance et al.91 and Woodward et al.92
guanosine triphosphate-binding proteins that associate with antigen degradation and loading of MHC molecules? What
vacuoles inhabited by a range of intracellular pathogens (eg, interactions take place between intracellular pathogens and
Toxoplasma gondii), thus marking them for destruction host Mf, especially in regard to hypoxia112 and nutritional
within the macrophage.62,106 requirements of the host and microbe? What is the role of
Clearance of proteinase-inhibitor complexes (eg, by pathogen-derived secretory products in the vacuolar milieu,
CD91) and of haptoglobin-hemoglobin complexes by the in recruitment of organelles such as endoplasmic reticu-
Mf-receptor CD163, and protection by hemeoxygenase-1, lum and mitochondria, and in effects on host cell biosyn-
are essential homeostatic functions of tissue Mf, limiting thesis? What are the intracellular killing mechanisms, and
potentially injurious extracellular molecules.107 Other mol- how can organisms survive, or become latent, within Mf ?
ecules released by injured infected cells can serve as alarm- To what extent do extracellular pathogens interact with
ins108 and danger signals (eg, heat shock proteins109 and S100 macrophages and DCs113? Finally, what receptor-mediated
family members,110 inducers of inflammatory and immune signals induce the secretion of Mf molecules such as IL-12,
responses). IL-23, and IL-10 that direct and regulate the resultant spe-
Major unsolved questions remain concerning phagocy- cific immune response?
tosis, intracellular infection, and immune responses. How
do sterile particulate antigens111 and microbial agents in-
duce T= cell responses, and what are the relative contribu- Gene Expression and Secretion
tions to this process of Mf and DCs, abundant and sparser Knowledge of Mf gene expression and protein synthesis is
professional phagocytes, respectively? What is the role of growing rapidly from the application of gene array and pro-
Scavenger and RAGE72,110 receptors mediating entry of di- teomic technologies. After surface and endocytic stimula-
verse ligands in subsequent adaptive immunity? Does TLR tion, the mature Mf is able to secrete a very large range of
engagement within vacuoles determine the kinetics of high- and low-molecular weight products. These include
phagosome maturation as well as induce local intracellu- enzymes involved in antimicrobial resistance (lysozyme),
lar responses? What determines the balance between total neutral proteinases and arachidonate metabolites that

TABLE 19.3. Modulation of Macrophage Phenotype

Category Stimulus Selected Marker Changes Function
Innate activation Microbial products Costimulatory molecule expression, Phagocytosis, adaptive immunity
(eg, LPS, other TLR ligands) MARCO upregulation
Classic activation Interferon g MHC II upregulation, Cell-mediated immunity
proinflammatory cytokine (eg, intracellular pathogens)
secretion, inducible NO synthase
Alternative activation IL-4/IL-13 Upregulation of MHC II, arginase, Parasitic and allergic immunity,
mannose receptor, Ym1, FIZZ1 repair
(resistin-like), production of
selected chemokines,
macrophage fusion
Innate and acquired Apoptotic cells, IL-10, Various surface and secretory Anti-inflammatory and altered
deactivation glucocorticoids, markers, (eg, anti-TNFa actions) immunity
TGFb, PGE2
IL, interleukin; LPS, lipopolysaccharide; MHC, major histocompatibility complex; NO, nitric oxide; PGE2, prostaglandin E2; TGFb, transforming growth factor-b; TLR, toll-like receptor;
TNFa, tumor necrosis factor-a.
contribute to inflammation and tissue repair, cytokines such Mf-restricted molecules may, in due course, make it pos-
as IL-1 and TNFa that modulate the activities of other leu- sible to direct Mf biosynthetic activities precisely, to boost
kocytes and endothelium, and reactive oxygen and nitrogen or inhibit immune responses.
intermediates implicated in host defense. Proinflammatory
cytokines account for part of the effects of immune adju-
vants in promoting, broadening, and sustaining humoral Modulation of Macrophage Activation in Vitro
responses. The ability to release these products depends on Our understanding of Mf activation derives from studies
the prior history of the Mf, whether resident, recruited, or of induction of MHC II and costimulatory antigens; of ef-
activated (primed), its encounters with microbial wall prod- fector functions such as proteinase, TNFa, reactive oxygen
ucts, including LPS acting via TLRs, or with apoptotic cells, intermediate and reactive nitrogen intermediate release;
and exposure to cytokines and other immunomodulatory of expression of membrane receptors such as MRs; and of
molecules in its immediate environment. Ligation of spe- resistance to infectious agents, for example, Mycobacteria,
cific receptors induces various signaling pathways114 and is Listeria, Candida, and HIV. Generalizations can be made, but
able to alter gene expression in the Mf selectively. NF-kB,115 it must be remembered that organisms vary considerably in
Pu-1 and other transcription factors,116–119 and IFN regu- their ability to evade or survive Mf restriction mechanisms,
latory factor families119,120 contribute to Mφ-restricted or and they interact with Mf in individual ways. Various inhib-
activation-dependent changes in gene expression, as does itory cell surface molecules (eg, CD200, SIRPa,122 and TAM
epigenetic regulation.121 Product expression depends fur- receptors123) are known to regulate Mf activation through
ther on translational regulation, posttranslational modifica- interactions with other activating plasma membrane recep-
tion such as proteolytic processing, intracellularly or at the tors. Receptors including FcRg use paired immunoreceptor
cell surface, and coexpression of inhibitors such as IL-10. tyrosine-based activation motif and immunoreceptor tyro-
Messenger RNA turnover varies greatly for different prod- sine-based inhibitory motif intracellular signaling motifs.
ucts due to the presence or absence of specific 3′ instabil- Figure 19.4 and Table 19.3 illustrate various pathways and
ity sequences. Many Mf products are labile and act close to markers of Mf activation that result from microbial, cellular,
the cell surface; overproduction results in tissue catabolism and cytokine interactions. Knowledge is based mainly on in
and systemic effects associated with widespread infection or vitro experiments and in vivo challenge of selected animal
chronic inflammation, often as a result of an immunologi- models. Innate activation depends on direct stimulation by
cally driven disease process. microbial products, independent of cytokines, although often
Whereas most bioactivities have been defined in vitro, enhanced by concomitant stimulation (eg, by IFNg). Newly
there is evidence that expression of Mf secretory activi- discovered markers of innate activation of mouse peritoneal
ties may be quite different in situ; lysozyme production macrophages include upregulation of MARCO, SR-A, via a
is characteristic of all Mf in culture but is downregu- TLR pathway, and of CD200, a more widely expressed IgSF
lated on most resident cells in vivo, and its expression by membrane glycoprotein able to inhibit TLR, NOD-like recep-
granuloma Mf, for example, depends on induction by tors, and inflammasome activation. Induction of MARCO, a
immune or phagocytic stimuli. 5′ promoter sequences of phagocytic receptor for a range of bacteria, represents an ad-
human lysozyme and CD68 transgenes have been used to aptation of the innate immune response to microbial contact
target tissue- and Mf activation-specific expression of re- Analysis of the actions of individual cytokines (IFNg, IL-10,
porter molecules in vivo. The promoters of these and other IL-4/13) on defined Mf targets (murine peritoneal Mf and

human monocyte–derived Mf ) reveals three characteristic activation of macrophages has recently been implicated
and distinctive in vitro phenotypes across a spectrum of ac- in thermoregulation.130 A great deal remains to be learned
tivation.124,125 IFNg and its production and amplification via about Mf behavior in physiology as well as disease.
IL-12, IL-23, or IL-18 play a central role in MHC II induction,
enhanced antimicrobial resistance, and proinflammatory
cytokine production, characteristic of Th1-type (M1-type) CONCLUSION AND SOME REMAINING ISSUES
responses; conversely, IL-10 suppresses markers of activa- Mf influence and respond to all other cells involved in im-
tion while inducing selective expression of other Mf genes. A munity, during both the afferent and efferent limbs. Many
comparable link between Mf /APC and the induction of Th2- of the molecules that mediate particular functions are now
type responses has proved elusive to identify. IL-4/13 have defined, but their role within the Mf and in intercellular in-
closely overlapping functions and induce an alternative, M2- teractions is often poorly understood. Mf developed during
type activation phenotype in Mf consistent with increased the evolution of multicellular organisms before immunolog-
APC function and humoral responses in allergy and parasitic ically specific, clonotypic responses of B- and T-lymphocytes
infection as well as giant cell formation (see Table 19.3).126 emerged. A recent report of a rearranging, TCR-like recep-
It is important to distinguish modulation of Mf immuno- tor in macrophages needs confirmation.131 Mf themselves
logic properties by IL-4/13 from marked deactivation and diversified in parallel with T-helper lymphocytes, generating
inhibition of proinflammatory and cytotoxic functions by DCs as specialized APCs for naïve T-lymphocytes and yield-
IL-10 and glucocorticosteroid. Immune complexes are also ing a range of effector cell phenotypes in response to diverse
able to induce an analogous, regulatory alternative activation activated T cells, both CD4 + and CD8 +. Mf and their deriv-
pathway, which overlaps with, but differs from IL-4/13– and atives cluster with differentiating hemopoietic cells in fetal
IL-10–induced phenotypes.127 By extension, GM-CSF, gluco- liver and bone marrow, with developing thymocytes, with
corticosteroids, TGF-b, and type I IFN all modulate Mf gene naïve CD4 + T lymphocytes and antigen during immune in-
expression with individual signatures. duction, and with activated T cells and microbial pathogens
The interplay of cytokines derived from Mf themselves, in granuloma formation (see Fig. 19.2). In addition, they
from activated T- and B-lymphocytes, and from other cells associate with antigen-stimulated B-lymphocytes during
(eg, natural killer cells, endothelial cells) results in recipro- cell expansion, diversification, and apoptosis. A major chal-
cal positive or negative interactions and time-dependent lenge will be to define the role of specific and accessory sur-
changes in activating and inhibitory signals. Some predic- face molecules by which Mf discriminate between live and
tions from in vitro studies can be extended to the intact dying cells and to uncover the intrinsic and extrinsic factors
host. For example, IFNg, IL-12, and IL-23 deficiency results that control Mf activities within these diverse immune cell
in inability to restrict opportunistic organisms in murine interactions.
models and in humans, and inducible nitric oxide synthase Our understanding of the multiple roles of Mf and DCs
is important for resistance to a range of infectious agents.128 in immunoregulation is also evolving as we better appreci-
IL-10 deficiency, on the other hand, results in overactive ate their specializations and adaptations. Central issues in
Th1-dependent inflammation, for example, in gut. IL-4 de- the immunobiology of Mf remain interesting topics for fur-
ficiency by itself has little effect on Mf phenotype in vivo ther investigation. These include the following:
because IL-13 mimics many of its actions. These cytokines Mf display broad functions in homeostasis, beyond host
share a common receptor subunit, and its targeted genetic defense and immunity, which may be special instances of a
ablation makes it possible to study Mf that lack the ability more general role in preserving host integrity, comparable
to respond to both IL-4 and IL-13. to that of the central nervous and endocrine systems. Their
The foregoing analysis is oversimplified. Combinations dispersion, plasticity, and responsiveness raise obvious
of cytokines in vitro have different effects on Mf than the questions for the biologist. In particular, what are their roles
sum of the parts. For example, the combination of IL-4 and in development, in trophic interactions within different or-
GM-CSF induces differentiation of human monocytes into gans, in angiogenesis,132 repair, and fibrosis133?
immature DCs, whereas each alone induces cells with dis- The Mf lie at the heart of the classic immunologic ques-
tinctive Mf properties. Furthermore, a particular “Th2- tion of recognition of altered or non-self, especially of par-
type” cytokine such as IL-10 can display radically different ticulates. What are the actual exogenous and endogenous134
effects on antimicrobial (inducible nitric oxide synthase ligands recognized by the diverse range of plasma mem-
dependent) killing, which is markedly suppressed, and anti- brane receptors capable of direct detection, and what deter-
HIV activities of Mf, which are enhanced. Whereas IFNg mines whether uptake of a target is immunologically silent
and IL-4 may have opposing actions on MR expression or productive? How can this information be harnessed for
and phagocytosis of yeast, in combination they synergize vaccine development?
to enhance uptake markedly. Other combinations of cyto- The delineation of further subsets of CD4 + T lympho-
kines, such as IFNa b and IFNg, can antagonize each other, cytes (TH17, regulatory T cells) suggests that it will be useful
presumably by competition for signaling pathways. Selected to define the effects on the Mf phenotype of contact-and
pathogens such as Francisella tularensis are able to modulate cytokine-dependent interactions with these cells. It is likely
macrophage responses to infection from an M1- to an M2- that further distinctive type 2 activation pathways of Mf
type, thus facilitating their survival.129 Finally, alternative will be discovered by microarray and protein analysis.

Once activated, Mf change their ability to recognize and a limited range of cell types, they also express highly re-
destroy targets, directly or in concert with antibody, comple- stricted molecules responsible for unique functions. Can
ment, and other less-defined opsonins. Can Mf directly kill these be harnessed for Mf-specific gene targeting at selected
virus-infected and other immunologically activated cells? If microanatomic sites to deliver functionally precise signals
so, do they use MHC matching, even in a limited way, and at predetermined times? Techniques are becoming available
do they contribute to tolerance and, by implication, autoim- for at least part of this fantasy, and they should provide new
munity by failure to perform such a suppressive function? insights into the multiple roles of the Mf in immunity.
A special case in which Mf are present in large numbers
at a site of “failure” to respond immunologically is the fe-
toplacental unit. CSF-1 is produced locally at high levels; ACKNOWLEDGMENTS
does this deactivate Mf or make them switch to perform a I thank my colleagues for discussions; Drs. Martinez-
trophic role? Do tumors that are rich in Mf adopt a simi- Pomares, Bowdish, Goodridge, Underhill, Portnoy,
lar strategy135–137? Catabolism of tryptophan by Mf enzymes and Vance for providing illustrations; and Dr. Annette
has been put forward as another mechanism for preventing Pluddemann for help in preparing the manuscript. Research
local destruction of an allogeneic fetus.138 This is but one ex- in the author’s laboratory was supported by grants from
ample of a growing interest in immunometabolism.139 the Medical Research Council, UK, the Wellcome Trust,
Although Mf express a large number of genes involved the Arthritis Research Council, and the British Heart
in household functions and share expression of others with Foundation.

CHAPTER
20 Granulocytes and Mast Cells
Patrizia Scapini • Nicola Tamassia • Carlo Pucillo • Marco A. Cassatella
INTRODUCTION neutrophils).1 Nonetheless, extensive research performed in

Polymorphonuclear leukocytes or granulocytes are he- the last 20 years has recognized neutrophils as highly versa-
matopoietically derived blood cells that typically act at tile and sophisticated cells displaying a significant synthetic
the frontline of innate defense in host response to foreign capacity as well as an important role in linking the innate
microorganisms. Granulocytes contain heterogeneous and adaptive arms of the immune response.2,3
cytoplasmic granules, which are storage pools for cell-spe-
cific intracellular enzymes, preformed receptors, cationic Neutrophil Generalities
proteins, and other cell-specific molecules. According to
Neutrophils are the most abundant (40% to 70%) circu-
their granular staining properties, polymorphonuclear leu-
lating leukocyte type in human blood, normally present
kocytes are classified into three different populations: neu-
at 2.5 to 7.5 × 109 cells/L.4 Morphologically, these cells can
trophils, eosinophils (Eos), and basophils (Bas). Neutrophil
be identified by the peculiar shape of their nucleus, which
granules stain preferentially with neutral dyes, eosinophilic
is polymorphous and usually consists of three to five sau-
granules stain with acidic colorants such as eosin, and
sage-shaped masses of chromatin connected by fine threads
Ba granules stain with basic dyes.
(Fig. 20.1). Mature neutrophils are terminally differenti-
Mast cells (MCs), similarly to polymorphonuclear leu-
ated, nondividing cells that develop and mature in the bone
kocytes, represent another crucial effector cell type of the
marrow from pluripotent CD (cluster of differentiation)34-
innate immune system that also stores elevated amounts of
positive (CD34+) stem cells, under the regulatory effects
preformed inflammatory mediators within their cytoplas-
of several colony-stimulating factors (CSFs), including
mic granules. However, while polymorphonuclear leuko-
granulocyte-macrophage CSF (GM-CSF), granulocyte CSF
cytes are mainly peripheral blood– circulating cells, MCs
(G-CSF), interleukin (IL)-3, and IL-6.4 Neutrophils can
are tissue resident cells distributed throughout the vascular-
be cytofluorimetrically identified by their characteristic
ized tissues or serosal cavity. Granulocytes and MCs differ
morphology (high side scatter) and expression pattern of
in their functions and roles during the inflammatory pro-
plasma membrane proteins such as CD66b, CD11b, CD15,
cess, MCs, Bas, and Eos being, for instance, essential com-
and CD16, in conjunction with the lack of expression of
ponents of allergic inflammation. Interestingly, recent data
CD2 and CD19. In humans, circulating polymorphonuclear
have revealed that granulocytes and MCs may also play key
leukocytes are 10 to 20 μm in diameter, display a half-life
roles in orchestrating the transition from innate to adaptive
previously thought to correspond to 7 to 12 hours, but more
immunity. These latter observations have caught the atten-
recently reevaluated and extended to up to 90 hours,5 and
tion of immunologists who are currently reevaluating the
exist in a dynamic equilibrium with a “marginated” pool
importance of granulocytes and MCs, and very intensively
that is sequestered within the microvasculature of many or-
working to clarify their multifaceted aspects in immunity.
gans.1,4 In the resting uninfected host, the peripheral neutro-
phil population is maintained within a constant number by
several mechanisms. One of them consists in programming
NEUTROPHILS neutrophils to spontaneously undergo apoptosis1,3 to be,
Neutrophils are well known to function as the first line in turn, cleared by tissue macrophages located in the bone
of defense against invading pathogens, principally bacte- marrow, spleen, and liver.6 In this latter context, a feedback
ria and fungi, but also viruses.1 These cells, together with loop involving an IL-23/IL-17/G-CSF axis, crucial for the
monocytes, macrophages, and dendritic cells (DCs), feature regulation of neutrophil production, has been recently iden-
the characteristic properties of the “professional” phago- tified in mice.7 According to this model, the uptake of apop-
cytes and utilize several effector mechanisms to destroy totic neutrophils by macrophages and DCs would determine
pathogens, including the generation of massive amounts a downregulation of their IL-23 production. Consequently,
of reactive oxygen species (ROS) in combination with the the Th17 subset of proinflammatory T-lymphocytes would
discharge of many potent antimicrobial enzymes or factors.1 be poorly sustained and thus much less IL-17A would be
Because of their powerful microbicidal equipment, neutro- generated. As IL-17A positively regulates fibroblast- and en-
phils are often depicted as harmful cells that can cause dam- dothelial cell–derived G-CSF, which is essential for control-
age to the surrounding tissues during acute inflammation ling both granulopoiesis and neutrophil survival, the fi nal
(as observed in those inflammatory diseases dominated by outcome of this circuit—triggered by the massive neutrophil
468

CHAPTER 20 GRANULOCYTES AND MAST CELLS | 469
FIG. 20.1. A Polymorphonuclear Neutrophil

Circulating in Peripheral Blood. From
Anderson’s Atlas of Hematology; Anderson,
Shauna C., PhD. Copyright 2003, Wolters
Kluwer Health/Lippincott Williams & Wilkins.
apoptosis at peripheral sites—would consist in a decrease in of the ongoing inflammation, ultimately provoking the
the levels of neutrophils released from the bone marrow.7 onset of chronic inflammatory/autoimmune diseases.9
On the other hand, the number of circulating neutrophils
can dramatically increase (even up to 10-fold) under acute
inflammatory conditions (eg, during a bacterial infection), Neutrophil Microbicidal Mechanisms
from accelerated neutrophil production and release from To destroy and eliminate invading pathogens, neutro-
the bone marrow.4 Moreover, even the lifespan of neutrophils essentially utilize two fundamental mechanisms10 :
phils is significantly extended under inflammatory condi- an oxygen-dependent process that is mediated by the
tions, as various host- and pathogen-derived mediators such generation of ROS, which include O2− (superoxide anion),
as G-CSF, GM-CSF, interferon (IFN)-γ, tumor necrosis fac- hydrogen peroxide, singlet oxygen, and other products de-
tor (TNF)-α , lipopolysaccharide, and nucleic acids inhibit rived from the metabolism of hydrogen peroxide; and an
neutrophil apoptosis and hence prolong their survival.8 oxygen-independent process consisting in the release into
During an acute inflammatory response, neutrophils the phagocytic vacuole of lytic enzymes; and antimicrobial
are rapidly recruited to the site of injury by a coordinated polypeptides stored in their intracellular granules.10 The
sequence of events that begin with the elaboration of vari- oxygen-dependent process, also referred to as the “respira-
ous mediators able to specifically promote their migration tory burst,” is defined as an increase of a mitochondrial in-
from the intravascular compartment. Mediators derive from dependent oxidative metabolism that leads to the generation
numerous sources (tissue macrophages, endothelial cells, ac- of O2−, which occurs through the activation of the phago-
tivated plasma components) and include vasoactive amines, cytic NADPH oxidase, an enzymatic system that is unique
proinflammatory lipids, small polypeptides, chemotactic to phagocytes (neutrophils, monocytes, macrophages, DCs,
factors, and cytokines such as TNFα , IL-1β, or IL-17A (the and also Eos). Nicotinamide adenine dinucleotide phos-
latter being one of the most abundant products of Th17 phate (NADPH) oxidase is a multiprotein complex formed
cells).1,3 Chemotactic factors, in particular, are generated in by a flavocytochrome-b558 (a heterodimer of gp91phox and
temporally distinct waves and include C5a, leukotriene-B4 p22phox chains, where phox stands for phagocyte oxidase),
(LTB4), formyl-Met-Leu-Phe (fMLF), as well as neutrophil three cytoplasmic components (namely p40phox, p47phox,
specific chemokines, such as CXCL8/IL-8, CXCL1/GROα , p67phox) and either Rac1 or Rac2 from the Rho family of
CXCL5/epithelial cell neutrophil-activating protein-78, low-molecular-weight GTPases (Fig. 20.2). Upon cell stim-
etc.1,3 Once recruited at an inflammatory site, neutrophils ulation, the cytosolic components of the complex become
function as mobile arsenals that recognize, phagocytose, phosphorylated and assemble together with the cytochrome
and ultimately destroy their targets. If the acute inflam- and Rac1/2 on the plasma membrane, thus forming the
matory response correctly subsides, then neutrophils may active enzyme that produces superoxide anion radicals, by
actively participate in its resolution (see following discus- catalyzing the transfer of electrons from NADPH to molec-
sion). If not, an uncontrolled and continuous release of the ular oxygen (see Fig. 20.2). O2− is converted by superoxide
proinflammatory cargo (ROS and proteases) by neutrophils dismutase into hydrogen peroxide, which, in the presence of
recruited at the site of infection/injury may eventually lead myeloperoxidase and halogens, is then metabolized into hy-
to destruction of bystander tissue, and thus to exacerbation pochlorous acid. The latter represents one of the neutrophil’s

2O2- hematopoiesis at the promyelocyte stage. As highlighted in

Figure 20.3, granules are released in a hierarchical order and
P22 rac-2 Gp91 P22 under separate control by mature neutrophils, depending on
Gp91 P
P67 P47 the type of stimulus.11
+ +
Azurophil or primary granules undergo limited exo-
P40 NADP + H
rac-2 cytosis in response to stimulation and are packaged with
acidic hydrolases and antimicrobial proteins that contribute
Rho GDI 2O2 + NADPH primarily to the killing and degradation of engulfed mi-
P croorganisms into the phagolysosome.11,12 These granules
P67 P47 contain myeloperoxidase, an enzyme that catalyzes the
P40 formation of hypochlorous acid, hydrolases, lysozyme,
matrix metalloproteinases, and three structurally related
Rho GDI rac-2 serprocidins (serine proteases with microbicidal activity):
proteinase-3, cathepsin G, and elastase. The latter proteins
can degrade a variety of extracellular matrix components,
P47 including elastin, fibronectin, laminin, type IV collagen, and
P67
vitronectin. Azurophil granules also contain antimicrobial
P40
molecules such as bactericidal/permeability-increasing pro-
tein (which is important for killing gram-negative bacteria)
FIG. 20.2. Nicotinamide Adenine Dinucleotide Phosphate (NADPH) and α-defensins, a family of small cysteine-rich antibiotic
Oxidase Assembly. In the resting neutrophil, the cytochrome b sub- peptides with broad antimicrobial activity against bacteria,
units gp91-phox and p22-phox are tightly bound in the membrane. fungi, and certain enveloped viruses.11,12
p47-phox, p67-phox, and rac-s complex are in the cytosol. Upon Specific (or secondary) granules, which are formed at
activation, Rho GDP-dissociation inhibitor (GDI) releases rac-2, and
p47-phox becomes phosphorylated. This causes translocation of the myelocytic stage, are smaller and less dense than the
rac-2, p47-phox, and p67-phox to the membrane and complex for- azurophil ones, and contain unique constituents, such as
mation with the cytochrome components, thereby completing the collagenase, haptoglobin, vitamin B12-binding protein, as
assembly of the active oxidase. (After Burg ND, Pillinger MH. Clin well an extensive array of membrane-associated proteins in-
Immunol. 2001;1:7–17.) cluding cytochromes, signaling molecules, and receptors.11,12
Secondary granules also contain an arsenal of antimicro-
bial substances, such as lactoferrin, neutrophil gelatinase-
major weapons against microbes, as it also synergizes with associated lipocalin, lysozyme, hCAP18 (cathelicidin human
granule proteins to kill pathogens in the neutrophil pha- cationic antimicrobial protein of 18 kDa, the proform of LL-
golysosome. O2− can also reacts with other cellular radicals, 37), and pentraxin-3. The inhibitory effect of antimicrobial
such as nitric oxide, to form different species of cytotoxic proteins include depriving ions essential for microbial sur-
oxidant, such as peroxynitrite. The critical role of NADPH vival, degrading structural components of microorganisms
oxidase and its products in host defense is best illustrated by (eg, peptidoglycan), and disrupting the integrity of target
the plight of patients with chronic granulomatous disease, cell membrane by punching pores in the membrane or by
in which mutations in any of the NADPH oxidase com- perturbing membrane integrity. Neutrophil-specific gran-
plex subunits (gp91phox, p22phox, p40phox, p47phox, and ules also contain an important family of soluble protein-
p67phox) leads to a severe immunodeficiency characterized ases, known as matrix metalloproteinases (MMPs), which
by defective killing of phagocytosed pathogens for the lack include neutrophil collagenase-2 (MMP-8), gelatinase-B
of ROS generation.10 These infections typically involve mi- (MMP-9), and leukolysin (MMP-25). These proteinases
croorganisms for which oxidant-mediated killing is particu- are generally stored as inactive proenzymes and undergo
larly critical for effective host defense, such as Staphylococcus proteolytic activation following granule fusion and interac-
aureus, Aspergillus spp., Nocardia, and a variety of gram- tion with azurophilic granule contents. Neutrophil MMPs
negative enteric bacilli. disrupt major structural components of bacteria and/or ex-
The release of potent proteolytic enzymes contained tracellular membranes, and are therefore crucial not only
in their granules represents the other crucial mechanisms for bacterial killing, but also for neutrophil extravasation
utilized by neutrophils to eliminate pathogens follow- and migration.11,12
ing phagocytosis.10 Neutrophil granules are subdivided Tertiary granules are produced at the metamyelocyte
into peroxidase-positive granules (based on the presence stage of differentiation and are smaller, lighter, and more eas-
of myeloperoxidase, their marker), also called primary or ily exocytosed than the other granule classes.11,12 These gran-
azurophil granules (owing to their affinity for the basic ules are indeed important primarily as a reservoir of matrix-
dye azure A), and peroxidase-negative granules that in- degrading enzymes and membrane receptors needed during
clude the specific (secondary) granules, the gelatinase (ter- polymorphonuclear leukocyte extravasation and diapedesis.
tiary) granules, and the secretory vesicles11 (Fig. 20.3). The The primary constituent of tertiary granules is gelatinase, a
different types of granules appear at progressive stage of latent metalloenzyme with the capacity for tissue destruc-
neutrophil development,11 with the primary granules, as tion. Finally, the secretory vesicles are smaller than the other
suggested by the name, being the first ones to appear during granules, are generated by endocytosis during the late stage

Primary granules Secondary granules Tertiary granules Secretory granules

(azurophilic) (specific) (gelatinase)
Lysozyme Lactoferrin Gp91phox/p22phox Gp91phox/p22phox

Myeloperoxidase (MPO) Neutrophil gelatinase- Lysozyme
Defensins associated lipocalin (NGAL)
Bactericidal/permeability- Pentraxin-3
increasing protein (BPI) B12-binding protein Antibacterial
Lysozyme proteins
Haptoglobin
hCAP18
Gp91phox/p22phox
Prodefensin
Laminin-R CD14
uPAR CD16
Thrombospondin-R CD35 (CR1) Receptors
C1q-R
fMLF-R
CD11b/CD18 CD11b/CD18 CD11b/CD18

Adhesion
CD66 CD67 CD67
molecules
CD67
Cathepsin G uPA Arginase 1 Proteinase-3

Neutrophil elastase Collagenase (MMP8) Gelatinase (MMP9) Leukolysin (MMP25) Proteases
Proteinase-3 Leukolysin (MMP25)
β-Glucoronidase Orosomucoid CRISP3 Nramp1

Heparin-binding protein Secretory leukocyte β2-Microglobulin VAMP-2
(HBP) peptidase inhibitor (SLPI) Nramp 1 SNAP-23
Granulophysin (CD63) CRISP3 VAMP-2 CD10
Sialidase Histaminase SNAP-23 Cd13
Presenilin Heparanase Alkaline phosphatase Other classes
Acid mucopolysaccharides Stomatin DAF of functional
Acid β-glycerophosphatase β2-Microglobulin Heparin-binding protein proteins
α1-Antitrypsin VAMP-2 (HBP)
α-Mannosidase SNAP-23 Plasma proteins (including
N-acetyl-β-glucosaminidase albumin)
Increasing tendency of granule release
FIG. 20.3. Main Constituents of Neutrophil Granules.
of nuclear neutrophil segmentation in the bone marrow, and extracellular space consists in the ability of neutrophils to
are the most readily mobilizable.11,12 These vesicles are pref- form so-called neutrophil extracellular traps (NETs).13 The
erentially directed to the plasma membrane, as reflected in latter structures consist of nuclear chromatin decorated
the density of vesicle-associated membrane protein (VAMP), with antimicrobial peptides and enzymes (eg, bacteri-
a fusogenic protein associated with the granule membrane. cidal/permeability-increasing protein, elastase, pentraxin3,
Secretory vesicles do not contain toxic substances, but cathepsin G, and many others) that lacks membranes and
mainly plasma proteins like albumin and receptors (includ- cytosolic markers.13 NETtosis, a novel type of neutrophil
ing β2-integrins, the complement receptor [CR]1, receptors death mechanism that occurs under settings of extreme
for formylated bacterial peptides [fMLF-R], CD14, the Fc neutrophil stimulation (different from necrosis, apoptosis,
portion of γ-immunoglobulins (Igs) [FcγRIII/CD16], and and also independent from caspase activation), underlies
the metalloprotease leukolysin). Heparin-binding protein the generation of NETs.14 Accordingly, the nuclear enve-
(also known as CAP37 or azurocidin), whose release is essen- lope, granules, and cell membranes gradually dissolve dur-
tial for the polymorphonuclear leukocyte–induced increase ing NETtosis, allowing the nuclear contents to mix and
in vascular permeability at the initial stage of extravasation, condense in the cytoplasm before being released into the
is also stored in the secretory vesicles. extracellular space.14 NETs, in turn, bind to various gram-
An additional nonphagocytic microbicidial mechanism positive and gram-negative bacteria (such as Staphylococcus
used by neutrophils to capture and destroy microbes in the aureus, Salmonella typhimurium, Streptococcus pneumoniae,

and group A streptococci), as well as to pathogenic fungi Neutrophil Extravasation

(such as Candida albicans). Similarly to what happens in In general, leukocyte recruitment during inflammation,
the phagolysosome, the high local concentration of anti- also called extravasation, is a multistep and highly complex
microbial peptides and enzymes is responsible for the kill- phenomenon characterized by a number of predetermined
ing of the pathogens trapped by NETs.13 The observation steps occurring in the vessel lumen, known as leukocyte
that neutrophils from patients with chronic granulomatous capture (or tethering), rolling, activation, and firm adhe-
disease do not form NETs has suggested, on the one hand, sion (arrest)16 (Fig. 20.4). These latter steps are not discrete
that ROS-mediated signaling/cascades are involved in NET phases of inflammation; rather, they simply represent a se-
generation, and, on the other hand, that the lack of NETs quence of events from the perspective of each leukocyte.
might contribute to the pathogenesis of chronic granulo- Moreover, each of these steps appears to be necessary for
matous disease.3,14 Whatever the case, it is noteworthy to the effective leukocyte recruitment; blocking any of them
remark that both the oxygen-dependent and -independent can severely reduce leukocyte accumulation in the tissue.17
effector mechanisms in host defense toward pathogens Adhesion of blood leukocytes to the endothelium during
are also utilized by neutrophils for their cytotoxic and inflammation requires specific leukocyte–endothelial inter-
tumoricidal activities. actions involving different families of adhesion molecules.
The latter include members of the selectin family and their
cognate carbohydrate and glycoprotein ligands, which me-
Neutrophil Receptors diate the initial leukocyte deceleration along the vessel wall
Under inflammatory conditions, neutrophils sense a wide (a process called “rolling”), as well as members of the inte-
range of extracellular ligands that, through the interaction grin family and their cognate Ig superfamily ligands, which
with specific receptors, subsequently trigger a number of mediate the subsequent high-affinity adhesion and arrest of
effector functions, including adhesion, migration, phago- leukocytes to the venules16,17 (Table 20.1).
cytosis, survival, cell activation, gene expression modu- Under noninflammatory conditions, neutrophils (as well
lation, target cell killing, and mediator production and as other leukocytes) travel primarily through the center of
release.1,3,15 For instance, agonist-stimulated neutrophils the blood vessel lumen, where the flow is fastest. In response
may trigger not only degranulation, but also the release of to proinflammatory signals, however, both the neutro-
arachidonic acid and/or other eicosanoids (eg, prostaglan- phils and the blood vessels undergo a series of changes.
din [PG]E2), via the activation and/or the upregulation of As a consequence of vascular dilatation, blood flow first
PLA 2 and COX-2, respectively. Upon appropriate stimula- increases and then slows, thus facilitating the interactions
tion, neutrophils can also generate LTA4 through the ac- between leukocytes and the endothelial cells. The process
tion of 5-lipoxygenase, as well as LTB 4 by converting LTA4 of capture/tethering and rolling then ensues, in which
via the action of LTA4 hydrolase.15 A nonexhaustive list of L-selectins on neutrophils, and P- or E-selectins on endo-
neutrophil receptors includes 1) receptors for proinflamma- thelia, interact with sialyl-Lewisx moieties or PSGL-1 on
tory mediators (eg, the anaphylotoxin complement compo- their respective cell partners. Although these interactions
nent C5a, LTB 4, platelet-activating factor [PAF], substance are reversible and transient, they prepare neutrophils for a
P, and fMLF); 2) receptors for cytokines, such as IFNγ, tighter binding, integrin-mediated step.16,17 The most im-
IL-1, IL-4, IL-6, IL-10, IL-13, IL-15, IL-18, TNFα , G-CSF, portant neutrophil integrins are four, each one composed of
GM-CSF, and many others; 3) receptors for chemokines, an identical β -subunit (β2, known as CD18) noncovalently
including CXCR1 and CXCR2; 4) receptors/adhesion mol- linked to different α-subunits: CD11a/CD18 (αLβ2, lympho-
ecules for the endothelium; 5) receptors for tissue matrix cyte function antigen), CD11b/CD18 (α Mβ2, CR3), CD11c/
proteins; and 6) opsonin receptors, such as Fc γ Rs and CD18 (α Xβ2, CR4), and CD11d/CD18.18 When neutrophils
those for the major cleavage fragments of the complement encounter their specific chemoattractants (eg, CXCL8),
system (see following discussion). Neutrophils also express displayed bound to glycosaminoglycans on the vessel wall,
a variety of pattern recognition receptors (PRRs), includ- neutrophil integrins are converted from an inactive to an
ing all toll-like receptors (TLRs), with the exception of active conformation16,17 (see Fig. 20.4). Activated integrins
TLR3, cytoplasmic ribonucleic acid helicases involved in can interact with their counterreceptors on the surface of
viral ribonucleic acid recognition such as MDA5 and RIG- the TNFα and/or IL-1β–activated endothelium (eg, inter-
I, and deoxyribonucleic acid binding cytoplasmic proteins cellular adhesion molecule-1 [ICAM-1] and intercellular
(IFI16 and LRRFIP1).3 adhesion molecule-2), leading to the strong adhesion and
arrest of the cells. Neutrophils then flatten and transmigrate
between and through the endothelial cells of postcapillary
The Role of Neutrophils in Acute Inflammation venules into the surrounding tissue (see Fig. 20.4), also se-
In order to carry out their prototypical defensive role in creting a broad range of MMPs that degrade the basement
acute inflammation, bloodstream neutrophils must extrava- membrane. Transmigration involves homophilic interaction
sate. To do so, they attach to activated endothelium, trans- of CD31/PECAM-1 and JAM-A on neutrophils and endo-
migrate through postcapillary venules (diapedesis), and thelial cells, where CD31/PECAM-1 and JAM-A act sequen-
then migrate toward a corresponding chemotactic gradient tially to mediate neutrophil migration through the venular
to the injury site where they recognize their target, engulf, walls.17,18 Once in the interstitial compartment, neutrophils
and finally destroy it. migrate along the chemotactic gradient toward the site of

Vessel lumen
Tethering Rolling Activation Adhesion Diapedesis

Inactive Chemokine
integrin receptor
PSGL-1 L-selectin
L-selectin Active LFA-1 PECAM-1
P-selectin
S-Lex S-Lex integrin ICAM-1
JAM-A
Cytokine TNFα CXCL8
receptor IL-1
Macrophage
FIG. 20.4. Leukocyte Transmigration. Following an inflammatory stimulus, tissue-resident macrophages and other cells release inflam-
matory mediators such as tumor necrosis factor α and interleukin-1, which induce the rapid expression of preformed P-selectin (and
transcription-dependent E-selectin expression) on the endothelium. The interaction between selectins and their glycoprotein ligands initi-
ates leukocyte tethering and rolling. Activation by chemokines—and other leukocyte activators (eg, leukotriene-B4 or platelet-activating
factor)—presented on endothelial cells causes leukocyte integrin activation, thus resulting in transition from cell rolling to cell firm
adhesion, in view of the strength of integrin-mediated binding with endothelial immunoglobulin superfamily members. Leukocytes can
then transmigrate through the endothelial monolayer and chemotactically move toward the inflammatory stimulus. Examples of adhesion
molecules involved in each step are depicted.
injury or infection and, once arrived, begin to react with the although infections tend to be less severe in patients with
etiopathogenic agent. LAD-II than in patients with LAD-I. Finally, LAD-III has
The fundamental importance of adhesive glycoproteins been recently described. LAD-III is an autosomal recessive
in vivo is testified by those individuals affected leukocyte disorder caused by mutations in the human kindlin-3 gene,
adhesion deficiency (LAD) disease, who display an abnor- which codes for a protein essential for integrin activation.
mally high susceptibility to bacterial infections.19 Several Consequently, LAD-III is characterized by impaired adhe-
types of LAD disease have been identified. In LAD-I, muta- sion of leukocytes to the endothelium of inflamed tissues
tions of the β2 integrin typically eliminate the expression of and by severe bleeding. Curiously, a number of LAD-III
all four integrin complexes. Because of the multiple defects patients additionally suffer from osteopetrosis.
in adhesion-related functions, patients with LAD-I develop
recurrent bacterial and fungal infections, mostly with Neutrophil-Mediated Phagocytosis
Staphylococcus aureus or gram-negative enteric microbes. Neutrophils play a critical role in host protection as they
Neutrophilia with paucity of neutrophils at inflamed or in- eliminate microorganisms through phagocytosis (the
fected sites is characteristic of LAD-I, while typical clinical cellular process of engulfi ng particles larger than 0.5 μm).
features include frequent skin and periodontal infections, While many cells in our body are capable of phagocytosis,
delayed separation of the umbilical cord and omphalitis, neutrophils do it to an extent sufficient to be considered
and deep tissue abscesses. LAD-II is caused by mutations “professional phagocytes” (eg, a single neutrophil can en-
in the membrane transporter for fucose and thus is associ- gulf up to 10 to 12 particles [eg, bacteria]).20 Phagocytosis is
ated with loss of expression of fucosylated glycans on the cell triggered either through receptor (eg, mannose receptor or
surface. Fucosylated proteins such as sialyl-Lewis X (CD15s) Dectin-1) recognition of certain polysaccharides present on
are ligands for endothelial selectins and are important the surface of some yeast cells and/or upon the binding of
for the rolling phases of leukocyte extravasation. Patients opsonized microorganisms through, for instance, Fc γRs and
with LAD-II also have leukocytosis and form pus poorly, CRs.20 Neutrophils constitutively express the low-affi nity

TABLE 20.1 Main Adhesion Molecules Involved in Leukocyte-Endothelial Cell Interaction
Adhesion Molecule Distribution Ligands and Counterreceptors Function

a a
E-selectin (CD62E) Endothelial cells PSGL-1, ESL-1, CD44 , CD43 Rolling
P-selectin (CD62P) Endothelial cells, platelets PSGL-1, PNAd Rolling
L-selectin (CD62L) All leukocytes except effector and memory PNAd, MAdCAM-1, PSGL-1, Rolling
effector T cells E-selectin, P-selectin
Selectin ligands
PSGL-1 All leukocytes All selectins (essential for Rolling
P-selectin)
sLex Myeloid cells, some memory T cells, HEVs All selectins Rolling
PNAd HEV, some sites of chronic inflammation L-selectin, P-selectin Rolling
Integrins
αMβ2(MAC-1; CD11b/CD18) Granulocytes, monocytes, some activated ICAM-1, fibrinogen, C3b, JAM-C Adhesion,
T cells transmigration
αLβ2(LFA-1; CD11a/CD18) All leukocytes ICAM-1, ICAM-2, JAM-A Adhesion,
transmigration
αDβ2(CD11d/CD18) Monocytes, macrophages, eosinophils ICAM-1, VCAM-1 Adhesion
αxβ2(p150,95; CD11c/CD18) DCs Fibrinogen, C3b Adhesion
α4β1(VLA-4) Most leukocytes VCAM-1, fibrinogen, JAM-B Rolling, adhesion
α4β7(LPAM-1) Lymphocytes, NKCs, mast cells, MAdCAM-1, fibronectin, VCAM-1 Rolling, adhesion
monocytes
Immunoglobulin superfamily
ICAM-1 (CD54) Most types of cells LFA-1 Mac-1, fibrinogen Adhesion,
transmigration
ICAM-1 (CD102) Endothelial cells, platelets LFA-1 Mac-1 Adhesion,
transmigration
VCAM-1 (CD106) Endothelial cells VLA-4, α4β7αDβ2 Rolling, adhesion
MAdCAM-1 HEVs in PP and MLN α4β7, L-selectin Rolling
PECAM-1 Endothelial cells, platelets, leukocytes PECAM-1 Transmigration
JAM-A Endothelial cells, platelets, most JAM-A Transmigration
leukocytes
JAM-B Endothelial cells, HEVs JAM-B, JAM-C Transmigration
JAM-C Endothelial cells, HEVs, platelets, JAM-C, JAM-B Transmigration
monocytes, DC, some T cells
CD, cluster of differentiation; DC, dentritic cell; ESL, HEV, high endothelial venule; ICAM, intercellular adhesion molecule; JAM, LFA, MAdCAM, mucosal addressin cell adhesion
molecule; NKC, PECAM, PNAd, PSGL, sLex, VCAM, VLA.
a
CLA decorated.
Fc γRs (Fc γRIIA/CD32A and Fc γRIIIA/CD16A), and, when space in which proteolytic enzymes and other bactericidal
exposed to IFNγ or G-CSF, the high-affinity Fc γR (Fc γRI/ components are discharged and pathogen degradation oc-
CD64) as well.3 CRs expressed by neutrophils are CR1 curs. At the same time, NADPH oxidase assembles on the
(also known as CD35), which binds to complement com- phagosomal membrane after phagocytosis and starts to gen-
ponents C1q, C4b, C3b, and mannan-binding lectin; CR3, erate ROS into the phagolysosome to kill bacteria by oxidiz-
which binds to iC3b, intercellular adhesion molecule-1, and ing microbial proteins and lipids. The activity of NADPH
some microbes; and CR4, which binds to iC3b. By express- oxidase also leads to the acidification of the phagosome,
ing these latter receptors, neutrophils are able to recognize which enhances the effectiveness of pH-sensitive antimicro-
and bind, in a cooperative manner, IgG-opsonized par- bial compounds. Thus, neutrophil mechanisms of pathogen
ticles and/or complement-opsonized microbes, and then destruction within the phagosome are multiple and involve
activate their phagocytosis. During the phagocytic process, granule fusion, toxic oxygen radical production, activation
the foreign particle is internalized, initially through mem- of latent proteolytic enzymes, and the activity of antibac-
brane recruitment to the site of particle contact, and then terial proteins (see Fig. 20.5). Remarkably, an activation of
via membrane extensions outward to surround the particle gene transcription and a selected generation of cytokines
and form a new vesicle called a cytoplasmic phagosome.20 also occur during phagocytosis, a feature that neutrophils
(Fig. 20.5). The phagosome then undergoes fusion with utilize for boosting a more effective innate immune re-
neutrophil granules to form a phagolysosome, a protected sponse. For instance, recruited neutrophils that phagocytose

Antibacterial antibody a pathogen also respond by producing chemokines, in par-

and complement ticular CXCL8, to amplify their own recruitment, but also
molecules CCL3, CCL4, and CCL19 that serve to recruit monocytes
Bacterium
and DCs.21
Role in the Resolution Phase of Inflammation

Locally activated neutrophils not only amplify the inflam-
matory process, but, surprisingly, actively participate in its
Receptors
resolution phase.22 To do so, during the late, final phases
Granules of a resolving, acute inflammatory reaction neutrophils,
(lysosomes) for instance, switch their eicosanoid biosynthesis poten-
tial from LTB4 to lipoxins, with profound modifications of
Nucleus
their effector functions. In fact, lipoxin A4 (LXA4) and LXB4
stop neutrophil chemotaxis, adhesion, and transmigration
through endothelium (by decreasing P-selectin expression),
A Attachment inhibit Eo recruitment, stimulate vasodilation (by induc-
ing synthesis of PGI2 and PGE2), inhibit LTC 4- and LTD4-
stimulated vasoconstriction, inhibit LTB4 inflammatory ef-
Invagination of
fects, and inhibit the function of natural killer (NK) cells.22
cell membrane
Neutrophils also contribute to the biosynthesis of resolvins
(such as resolvin E1, resolvin E2, resolvin D1, and resolvin
D2) and protectin D1, which all inhibit neutrophil transend-
othelial migration and tissue infi ltration, as well as stimulate
Bacterium
resolution and reduce the magnitude of the inflammatory
response in vivo.22 Furthermore, neutrophils might also
Nucleus serve as major producers of anti-inflammatory cytokines
such as transforming growth factor (TGF) β and IL-1 recep-
Granules tor antagonist, the latter being an endogenous inhibitor of
(lysosomes) IL-1β signaling and mediated effects.21 Finally, neutrophils
must be cleared from the inflammatory site as inflamma-
B Phagocytosis tion resolves. Indeed, neutrophils undergo apoptosis and are
engulfed by tissue macrophages, which then reprogram into
Oxidants and
digestive enzymes the M2 phenotype and start to generate antiinflammatory
cytokines such as TGFβ and IL-10.6,23
Phagocytic
vacuole
Novel Neutrophil Effector Functions
Bacterium The functions described previously for neutrophils in host
defense are fundamental for combating infectious diseases.
However, more recent discoveries on neutrophils as source
of a variety of cytokines21 have revealed that these cells are
not only key components of the inflammatory response,
Granules but also crucial effectors of innate and adaptive immune
(lysosomes)
regulatory networks.23
Nucleus
Neutrophil-Derived Cytokines and Chemokines
Numerous in vitro and in vivo studies, focusing on novel
aspects of the neutrophil biology and function, have re-
cently shed new light on the potential role that neutrophils
C Degranulation
can exert in the modulation of innate and adaptive immune
FIG. 20.5. Phagocytosis. The figure shows ingestion, digestion, and responses.23 It is now unequivocal that neutrophils are not,
destruction of foreign particulate matter (a bacterium, in this exam- as for a long time thought, terminally differentiated cells
ple) by a neutrophil. A: Cell membrane receptors bind to antibody and “devoid of transcriptional and protein synthesis activity.”23a
complement molecules previously attached to the bacterial surface. In fact, besides the several preformed or rapidly generated
B: The cell membrane creeps around the bacterium and envelopes inflammatory mediators described previously, neutrophils
it. C: The bacterium is trapped in a special space, the phagocytic
vacuole, into which lysosomes discharge proteases, which together display the capacity to de novo synthesize and release also
with oxidants kill it. Then, digestive enzymes dissolve it. (Thomas several chemokines and cytokines with immunoregulatory
H. McConnell, The Nature Of Disease Pathology for the Health properties.21,23 It is, however, important to mention that,
Professions, Philadelphia: Lippincott Williams & Wilkins, 2007.) at least in vitro, neutrophils usually produce, on a per-cell

basis, fewer molecules of a given cytokine than mononuclear

leukocytes.21 However, considering that neutrophils clearly Cytokines Expressed in Resting or
TABLE 20.2
predominate over other cell types under inflammatory con- Activated Neutrophils
ditions in vivo, it becomes obvious that the contribution
C-X-C Chemokines Proinflammatory TNF Superfamily
of neutrophil-derived cytokines can be of foremost impor- GRO`/CXCL1 Cytokines Members
tance. To date, a wide range of stimuli able to induce char- CINC-2`/GROb/ TNFa FasL
acteristic signatures of chemokine and cytokine synthesis by MIP-2α/CXCL2 IL-1α, IL-1β CD30L
neutrophils have been identified. Among these, cytokines CINC-2β/GROγ/ IL-6(?),IL-7, IL-9 TRAIL
themselves, chemotactic factors (fMLF, LTB4, PAF, C5a, MIP-2β/CXCL3 IL-16(?), IL- LIGHTa
and CXCL8), phagocytic particles, microorganisms (such as PF4/CXCL4 17A/F(?) Lymphotoxin-a
fungi, viruses, and bacteria), and PRR ligands can all induce ENA-78/CXCL5 IL-18 APRIL, BAFF/BLyS
GCP-2/CXCL6 MIF RANKL
the synthesis and release of chemokines and cytokines by IL-8/CXCL8
neutrophils.21 Considering that neutrophils usually repre- MIG/CXCL9 IP-10/ Anti-inflammatory Colony
sent the first cell type infi ltrating at the site of infections, a CXCL10 I-TAC/ Cytokines Stimulating
stimulus-specific response of neutrophils in terms of cyto- CXCL11 IL-1ra Factors
kine production might direct the evolution of certain types IL-4(?), IL-10(?) G-CSF
of inflammatory and immune reactions to support the tran- C-C Chemokines TGFa1, TGa2 M-CSF(?)
MCP-1/ GM-CSF(?)
sition from innate to adaptive immunity.
CCL2MIP-1`, Immunoregulatory IL-3(?)
Table 20.2 lists all the cytokines that, to date, have CCL3 Cytokines SCFa(?)
been shown to be released by neutrophils in vitro, either MIP-1a, CCL4 IFNα, IFNβ,
constitutively or following appropriate stimulation, or TARC/CCL17 IFNf(?) Angiogenic
in vivo. Numerous in vivo observations, in fact, not only PARC/CCL18 IL-12 and Fibrogenic
have confirmed and reproduced the in vitro findings, but MIP-3`, CCL19 IL-23(?) Factors
often have clarified their biological meaning and implica- MIP-3a, CCL20 VEGF
MDC/CCL22 Other Cytokines BV8/Prokineticin-2
tions. As outlined in Table 20.2, neutrophils can produce
Oncostatin M HB-EGF
proinflammatory, anti-inflammatory, immunoregulatory, GDF (?) FGF-2
angiogenic, and fibrogenic cytokines, chemokines, and li- NGF, BDNF, NT4 TGFa
gands belonging to the TNF superfamily. The role of these PBEF/visfatin/ HGF
molecules in mediating various neutrophil-dependent NAMPT
immunoregulatory functions is partially described in the amphiregulin
following chapter. For instance, chemokines are particularly ?, requires definitive corroboration; GRO, growth regulated oncogene; CXCL, CXC
represented among the cytokines produced by neutrophils chemokine ligand; CCL, CC chemokine ligand; MIP, macrophage inflamma-
and include those primarily chemotactic for neutrophils tory protein; CINC, Cytokine induced neutrophil chemoattractant; PF4, Platelet
factor-4; ENA-78, epithelial-derived neutrophil-activating peptide 78; GCP-2,
themselves, monocytes, DCs, NK cells, and T helper (Th)1 granulocyte chemotactic protein-2; IL-, interleukin-; MIG, Monokine induced by
and Th17 cells. It follows that a role for neutrophils in or- gamma interferon; I-TAC, Interferon-inducible T-cell alpha chemoattractant; IP-10,
Interferon gamma-induced protein 10; MCP, monocyte chemotactic protein; TARC,
chestrating the sequential recruitment to, and activation of, Thymus and activation regulated chemokine; PARC, pulmonary and activation-
distinct leukocyte types in the inflamed tissue is plausible, regulated chemokine; MDC, Macrophage-derived chemokine; TNF, tumor necro-
sis factor; MIF, macrophage inhibitory factor; IL-1ra, IL-1 receptor antagonist;
as already demonstrated to occur in several experimental TGF, Transforming growth factor; IFN, interferon; GDF, Growth Differentiation
models.21,23 factor; NGF, nerve growth factor; BDNF, Brain derived neurotrophic factor; NT4,
Neurotrophin-4; PBEF, pre-B-cell colony-enhancing factor; NAMPT, Nicotinamide
phosphoribosyltransferase; TRAIL, TNF-related apoptosis-inducing ligand; APRIL,
Neutrophils in Immunoregulation a proliferation-inducing ligand; BAFF/BLyS, B-cell activating factor/B lymphocyte
There is now wide experimental evidence that neutrophils stimulator; RANKL, Receptor activator of nuclear factor kappa-B ligand; G-CSF,
macrophage colony stimulating factor; M-CSF, macrophage colony stimulating
have the capacity to modulate the migration, maturation, factor; GM-CSF, granulocyte-macrophage colony stimulating factor; SCF, stem
and function of several leukocyte types including DCs, cell factor; HB-EGF, heparin binding-like epidermal growth factor; FGF, fibroblast
growth factor; HGF, hepatocyte growth factor.
T cells, and B cells.23 Regarding DCs, it is noteworthy to Cytokines in bold refer to neutrophil studies in animal models that confirm human
mention that neutrophils have been shown to produce findings.
a
biologically active CCL20 and CCL19, two structurally Messenger ribonucleic acid only.
related CC-chemokines that have been suggested to play

a fundamental role in trafficking of, respectively, imma-
ture and mature DCs to mucosal surfaces and lymphoid Mac-1 (CD11b/CD18) and DC-specific intercellular adhe-
organs. Likewise, neutrophils release several antimicrobial sion molecule.24 Neutrophils also act as transport vehicle
compounds, such as lactoferrin, LL-37, and cathepsin G, for pathogens and, in turn, deliver antigens to DCs, thus
that have been found to act as chemoattractants for im- playing an important role in the activation of T-cell im-
mature DCs. In addition, neutrophils can proteolytically mune responses controlled by DCs.24
activate prochemerin to generate chemerin, one of the few Concerning the interactions between neutrophils and B
chemokines that attracts both immature DCs and plas- cells, of particular interest are the findings that neutrophils
macytoid DCs.3 Neutrophils can also modulate DC matu- produce significant amounts of BLyS/BAFF (B-lymphocyte
ration and function either through the release of several stimulator/B-cell activating factor) and APRIL (a prolifer-
mediators or through direct physical interaction between ation-inducing ligand), two related members of the TNF

family that are well known to be essential for B-lymphocyte trophils but also other granulocyte receptor-1+ (Ly6C +)
homeostasis.25 Therefore, it is plausible to assume a role cells. Luckily, a more specific anti-Ly6G monoclonal anti-
of neutrophils not only in sustaining B and plasma cell body (1A8) has been raised; it is now preferentially used to
antibody production and survival, but also in promoting B- deplete neutrophils in vivo under different experimental
cell–dependent autoimmune diseases and tumors, as already settings.31 Obviously, it will take some time prior to con-
elegantly demonstrated in the case of B-cell lymphoma.25 trolling and, eventually, revising all data on neutrophil
Cross-talk between neutrophils and T cells has been depletion generated by using RB6-8C5. More importantly,
repeatedly described to occur during infections or other in- the future availability of conditional knockout mice,
flammatory responses and diseases. Current evidence now selectively targeting, one by one, neutrophil function such
indicates that neutrophils exhibit a significant chemotac- as survival, migration, or activation, will help to fi nally
tic effect toward Th1 or Th17 cell subsets, through the re- clarify the specific contributions of neutrophils under
lease of CCL2, CXCL9, and CXCL10, or CCL2 and CCL20, different inflammatory/immune settings.
respectively.26 Neutrophils have also a role in directing T-cell
polarization, for instance through their capacity to produce Role of Neutrophils in Angiogenesis and Tumor Growth
the Th1-inducing cytokine, IL-12. The latter has been There is no longer doubt that, in addition to macrophages,
clearly demonstrated in mouse models, in which strong neutrophils may positively or negatively influence the angio-
Th1-dependent T-cell responses that result in pathogen genic process and tumor growth.32 In the former case, it has
clearance are elicited upon infection with Candida albicans, been observed that neutrophils, via elastase release, may indi-
Helicobacter pylori, or Legionella pneumophila. Strikingly, rectly generate massive amounts of bioactive, angiostatin-like
depletion of neutrophils reverses the Th1 responses into fragments, and thus inhibit fibroblast growth factor (bFGF)
a predominant Th2-response, therefore making the mice plus vascular endothelial growth factor (VEGF)-induced en-
susceptible to infection. Besides the neutrophil’s ability to dothelial cell proliferation.33 However, as described for mac-
modulate T-cell functions through the production of che- rophages, neutrophils can also favor malignant growth and
mokines and cytokines, recent reports suggest that neu- progression, in relation to the type of tumor environment
trophils travel to the lymph nodes during infections and in which they reside, for instance via a remarkable produc-
express both major histocompatibility complex (MHC) tion of proangiogenic molecules such as VEGF and CXCL8.34
II and costimulatory molecules.27 However, whether neu- On the other hand, experimental studies of tumor cure and
trophils directly acquire antigen-presenting functions or prevention have suggested that, at least in some models,
transmit signals to naïve T cells remains still puzzling. engagement of neutrophil functions can be crucial for the
Neutrophils have also been shown to modulate the establishment of an effective antitumoral immune response
maturation, activation, and functions of NK cells, either and immune memory reactions.35 Indeed, neutrophils can
by themselves or in cooperation with other cell types.28 In produce several cytotoxic mediators for tumor and endothe-
this context, it is worth mentioning that neutrophils, by lial cell killing, including TNFα , defensins, proteases (such
interacting with specific subsets of peripheral blood myeloid as elastase and cathepsin G), ROS, nitric oxide, and angio-
DC (eg, 6-sulpho LacNAc + DC, also known as slanDC) can static chemokines (CXCL9, CXCL10, and CXCL11).35 It has
strongly potentiate IFNγ release by NK cells.29 Importantly, been recently found that neutrophils exposed to IFNs ex-
the potential pathophysiologic relevance of a cell network press and produce TNF-related apoptosis-inducing ligand,36
among neutrophils, slanDC, and NK cells has been sug- another TNF superfamily member that selectively stimulates
gested by immunohistochemical studies that have revealed tumor cell killing. More recently, in vivo evidence proving
their colocalization in several chronic inflammatory pathol- that, similarly to M1 and M2 macrophages, neutrophils also
ogies, such as Crohn disease and psoriasis.29 Finally, it has polarize from an N1, proinflammatory and antitumoral
also been proposed that mature postmitotic neutrophils can phenotype, to an N2 anti-inflammatory and protumoral
also “transdifferentiate” into much-longer-lived cells with phenotype, has been provided,37 thus supporting the no-
macrophage- or DC-like characteristics, which might con- tion that the tumor environment can profoundly shape the
stitute a further manner for neutrophils to act as regulatory functional status of neutrophils. Furthermore, it is now well
cells of the adaptive immune response.3,23 established that myelopoiesis can be profoundly modified
It is important to mention that despite of all experi- during inflammation and cancer, releasing altered mature
mental observations in vivo that strongly suggest that myelocytes and myeloid-derived suppressor cells (MDSCs)
neutrophils could potentially act as important players in that exert immunosuppressive and protumoral activity,
the orchestration of immune responses, additional reeval- mainly by inhibiting T-cell functions.38 Although mature
uations and validations are required. Indeed, in order to human neutrophils do not seem to be a major component of
investigate the role of neutrophils in vivo, an antigranu- such MDSC population, several mouse tumor models have
locyte receptor-1 monoclonal antibody, RB6-8C5, has revealed the existence of a granulocytic MDSC population
been extensively used to deplete mice of neutrophils.30 with potent T-cell suppressing activity.38 Nonetheless, low-
However, RB6-8C5 not only binds to Ly6G, which is pres- density granulocytes (so called because of their abnormal
ent on neutrophils, but also to Ly6C, which is expressed behavior upon density centrifugation) able to inhibit T-cell
on neutrophils, DCs, and subpopulations of lymphocytes activation and function have been found in human patients
and monocytes. Therefore, it has recently been shown that with cancer.39 Further research is now needed to better un-
in vivo administration of RB6-8C5 depletes not only neu- derstand the origin, phenotype, and relationship to mature

neutrophils of all these immunosuppressive granulocytic mainly involved in host defense against large parasites that
populations. Such studies will better clarify the real role of cannot be internalized by other professional phagocytes.
neutrophils in cancer as well as in other inflammatory/auto- Eos can directly kill the parasites, whereas Bas and MCs
immune diseases (such as infections, psoriasis, and lupus), preferentially release the contents of their granules into
in which the presence of MDSC- or low-density granulo- the external milieu upon activation, thus creating an envi-
cyte–like cells have been also described.40 ronment hostile to invading organisms. Furthermore, the
unique property of Bas and MCs to express the high-affinity
Fc receptor for IgE (FcεRI) that, after activation by antigen-
Neutrophils in Diseases specific IgE, induces rapid discharge of potent inflammatory
Previous sections have already summarized some of the mediators (histamine being one of them), is central to the
most common inherited disorders of neutrophil function initiation and propagation of immediate hypersensitivity
that impair critical responses for host defense. Dramatic reactions. Coughing, sneezing, and vomiting, all expulsive
clinical consequences may also be observed as a conse- responses that typically are caused by basophil- and MC-
quence of acquired neutropenia (eg, a susceptibility to in- derived mediators and accompany allergic and pathologic
fectious diseases when neutrophil counts fall below 500 inflammatory diseases, may actually reflect mechanisms
cells/μL4). Neutropenia can be due to depressed produc- that evolved to expel parasites. The properties and roles
tion (ie, hereditary neutropenias) or increased peripheral of Eos, Bas, and MCs in pathologic conditions such as al-
destruction.4 Apart from the disorders associated with ge- lergy and autoimmunity are described in detail Chapter 45.
netic dysfunctions or quantitative alterations of neutrophils, Herein, we focus on the specialized properties that highlight
there are other pathologic situations in which neutrophils their role in the innate and acquired immune response.
themselves become the predominant contributors to tissue
injury, especially when the mechanisms supposed to control
and inactivate their hypothetical beneficial and protective Eosinophils
effector functions are deregulated. Examples of such pa- Eosinophil Generalities
thologies, tentatively classified according to the major neu- Eos are end-stage, multifunctional leukocytes involved in
trophil-activating event,1 are 1) diseases caused by ischemia protection against parasitic helminths and bacterial and
reperfusion injury (ie, myocardial infarction); 2) bacterial viral infections, that are also implicated in the pathogen-
infections (endotoxic shock, osteomyelitis, adult respira- esis of numerous Th2-type inflammatory processes and
tory distress syndrome,); 3) cytokine-mediated diseases tissue injury.41 These granulocytic, bilobed, nucleated cells
(rheumatoid arthritis, inflammatory bowel diseases); 4) were named as Eos by Paul Ehrlich in 1879, because of their
diseases caused by crystal deposition (gout); 5) antineutro- intense staining with eosin, an acid aniline dye (Fig. 20.6).
phil cytoplasmic antibody–associated vasculitis (Wegener Eos are approximately 12 to 17 μm in diameter and
granulomatosis, pauci-immune necrotizing crescentic glo- represent 1% to 6 % of the total blood leukocyte popula-
merulonephritis); and 6) airway diseases (chronic obstruc- tion. Besides circulating and massively recruited at sites of
tive pulmonary disease, bronchiectasis, bronchiolitis, cystic Th2-dominated inflammation, Eos also reside in various
fibrosis, and even certain forms of asthma are characterized organs, such as the gastrointestinal tract, mammary glands,
by neutrophil infi ltration of the airway wall). One of the and bone marrow.41,42 Eos derive from bone marrow, CD34 +
key challenges in neutrophil-dominated conditions is how myeloid-committed progenitors, and upon maturation can
to manipulate neutrophil function to abolish their destruc- be cytofluorimetrically identified by their characteristic
tive potential in a way that does not compromise their an- morphology/side scatter and expression pattern of plasma
tibacterial and antifungal capacity. This has been difficult membrane proteins such as CD66b, CCR3, IL-5Ra in con-
to achieve, as successful treatments in animal models have
frequently proven ineffective or limited by side effects when
used in human inflammatory diseases.
EOSINOPHILS, BASOPHILS, AND MAST CELLS

Eos, Bas, and MCs are critical effector cells not only in
allergic inflammation, but also in innate and adaptive
immunity. In addition, they have a crucial role in the sur-
veillance of epithelial tissues, especially of the mucosa of the
gastrointestinal, respiratory, and urogenital tracts: MCs as
tissue resident sentinel cells, while Eos and Bas principally
upon recruitment from the bloodstream. The microbicidal
activity of these cells can be induced either after direct mi-
croorganism recognition or after activation by complement- FIG. 20.6. A polymorphonuclear eosinophil circulating in peripheral
or other leukocyte-derived products. Whereas neutrophils blood. (Reprinted with permission from Cohen BJ, Wood DL.
destroy internalized microorganisms by delivering cytotoxic Memmler’s The Human Body in Health and Disease. 9th Ed.
intracellular compartments to them, Eos, Bas, and MCs are Philadelphia: Lippincott Williams & Wilkins, 2000.)

junction with the lack of expression of CD16 (to better dis- host allergic and other immune responses, for instance in
tinguish them from neutrophils), and other lineage markers the stool or sputum of patients with gastrointestinal or respi-
(such as CD2, CD14, and CD19).41–43 Eosinophil differen- ratory eosinophilia.48,49 Specific granules have a distinctive
tiation and expansion from progenitor cells mainly occurs ultrastructural appearance with an electron-dense core and
in response to three cytokines, namely IL-3, IL-5, and GM- contain cationic proteins that give Eos their unique stain-
CSF, which provide permissive proliferative and differentia- ing properties. The major cationic proteins in the specific
tion signals through GATA-1, PU.1, C/EBPα , and C/EBPε granules are major basic protein (MBP, composed of two
transcription factors.44 IL-5 is, however, the most specific related cationic proteins, MBP-1 and MBP-2), Eo peroxi-
factor for Eo development, differentiation, and release from dase (EPO), Eo cationic protein (the Eo’s two ribonucleases
the bone marrow into the peripheral circulation, especially RNase2 and RNase3, respectively), and Eo-derived neuro-
during helminthic infection or allergic inflammation.41–43 toxin (EDN). MBP is the major component of the crystalloid
Also, allergen challenge or the experimental administration cores of specific granules and accounts for more than 50%
of CCL11/eotaxin-1 (acting through the CCR3 receptor) of the Eo granule protein mass. MBP is highly cationic, it
can cause bone marrow release of Eo precursors and mature lacks enzymatic activity, and its toxicity is mediated by en-
Eos.45,46 However, basal levels of Eo development can also hanced membrane permeability resulting from interactions
occur in the absence of IL-5 as in antigen-induced tissue of the cationic protein with the plasma membrane. MBP has
eosinophilia, as suggested by the presence of tissue Eos in in vitro activity toward parasites, including schistosomes
asthmatic patients treated with anti–IL-5–neutralizing and other helminths. In patients with asthma, serum and
antibodies.45,46 Once released from the bone marrow, Eos bronchoalveolar lavage fluid MBP concentration correlates
enter the circulation (with a half-life of 8 to 18 hours) and with bronchial hyperresponsiveness. EDN and Eo cat-
then traffic to tissues, at mucosal surfaces and/or sites of ionic protein demonstrate in vitro toxicity to parasites and
allergic inflammation, in response to a variety of differ- single-stranded ribonucleic acid pneumoviruses, includ-
ent chemoattractants including CCL11, CCL24/eotaxin-2, ing respiratory syncytial virus. EPO is another highly cat-
CCL26/eotaxin-3, CCL5/RANTES, or LTB4, LTD2, PAF, ionic protein that makes up approximately 25% of granule
and C5a. IL-4 and IL-13 play a central role in promoting proteins and that catalyzes the oxidation of halides, pseudo-
Eo trafficking to mucosal tissue by upregulating eotaxins halides, and nitric oxide to oxidant products that are toxic
(CCL11 and CCL26) and endothelial expression of vascu- to microorganisms and host cells. It has been shown that,
lar cell adhesion molecule (VCAM)-1, the counterreceptor depending on the activation stimulus, Eo granule cationic
for VLA-4, which is expressed on the surface of Eos.45,47 In proteins can be selectively secreted or released into tis-
addition, GM-CSF, IL-3, and IL-5, as well as IL-33 and IFNγ, sues by a number of different secretory pathways, ranging
are important for promoting long-term Eo survival and from classical granule fusion and exocytosis (eg, in killing
activation in vitro and in tissues.41–43,46 parasitic helminths), piecemeal degranulation (eg, vesicu-
lar transport from the granules in the absence of classical
Eosinophil Granules exocytosis), and cytolytic degranulation (release of intact
Human Eos have a bilobed nucleus with highly condensed membrane-bound granules directly into the tissue upon eo-
chromatin and contain up to four different populations of sinophil apoptosis/cell death).41–43,48,49 Studies in the past 15
secretory organelles: primary granules, secondary/specific/ to 20 years of the biochemistry, functions, and localization
crystalloid granules, small amorphous granules (containing in tissues of the unique enzymatic and nonenzymatic cat-
arylsulfatase and acid phosphatase), and secretory vesicles.48 ionic proteins present in Eo granules have provided compel-
These four organelles, along with vesiculotubular struc- ling evidence supporting a pathologic proinflammatory and
tures and small vesicles involved in transport and secretion effector role for the Eo in directly inducing tissue damage.
by the activated cell, serve as the major subcellular sites for
the eosinophil armamentarium of preformed cytotoxic and Eosinophil Mediators
inflammatory mediators.48 Eos also contain lipid bodies, In addition to containing numerous highly basic and cy-
which are non–membrane-bound lipid-rich organelles that totoxic granule proteins that are released upon activation,
are the major site of eicosanoid synthesis as they contain Eos can also synthesize an arsenal of lipid mediators, ROS,
eicosanoid synthetic enzymes including 5-lipoxygenase, and inflammatory/hematopoietic cytokines that mediate
leukotriene C4 synthase, and cyclooxygenase.41–43,48,49 Eos the pathophysiologic role of this granulocyte in health and
contain greater numbers of lipid bodies than neutrophils, disease (Table 20.3). Lipid-derived mediators generated by
which even increase during eosinophil activation in vitro or Eos include PGE2, thromboxane, LTC 4, and PAF, which are
engagement in inflammatory reactions in vivo.41–43,48,49 all considered as responsible for many of the Eo-triggered
Primary granules are similar to those found in other proinflammatory activities, namely the increase in leuko-
granulocyte lineages, are formed early in Eo development, cyte trafficking, endothelial adhesion, smooth muscle con-
and are enriched in Charcot–Leyden crystal protein (which traction, vascular permeability, and mucus secretion.41–43
represent less than 5% of total granules).48,49 Charcot– Eo-derived cytokines (including TGFβ, GM-CSF, IL-3,
Leyden crystal/galectin-10 is a hydrophobic protein of un- IL-4, IL-5, CXCL8, IL-10, IL-12, IL-13, IL-16, IL-18, TNFα ,
known function that is produced and released in high levels CCL5, stem cell factor [SCF], NGF, and CCL11) are in part
by activated Eos. Its characteristic hexagonal bipyramidal stored as preformed in granules and thus can be rapidly
crystals can be detected as a hallmark of Eo involvement in released upon degranulation (similarly to MCs).41–43,50–52

Eosinophils in Host Defense. The earliest recognized Eo

TABLE 20.3 Major Eosinophil-Derived Mediators effector functions were those associated to their role in
host defense against multicellular helminthic parasites,
Class Mediators Physiologic Effects
although it is now clear that Eos are also involved in host
Cationic MBP, ECP, EDN, EPO Cytotoxic and bac- defense against viral and bacterial infections.54 Among the
granule tericidal proper- several mechanisms through which Eos can exert defensive
proteins ties, smooth functions, the best described consist in their release of cy-
muscle cells totoxic cationic proteins (such as MBP, Eo cationic protein,
contraction,
mast cell activa-
EDN, and EPO), production of ROS, antibody- and com-
tion and survival plement-mediated killing, and the expulsion of extracellular
Lipid media- LTD4, LTE4, PAF, PGE1, Leukocyte traffick- deoxyribonucleic acid traps.43,54 Despite substantial in vitro
tors PGE2, 15-HETE ing, endothelial data, there is a lack of convincing evidence of the effective
adhesion, role of Eos in host defense in vivo, in particular during
smooth muscle helminthic infections. Indeed, the only human Eo-specific
cells contrac- condition, known as hereditary EPO deficiency, has not
tion, vascular
permeability, been related to increased susceptibility to helminthic infec-
inflammation tions. Furthermore, the results from infection studies car-
Cytokines TNFα, IFNγ, IL-1α, IL-1β, Inflammation, immu- ried out in mouse models remain unclear and controver-
IL-2, IL-3, IL-4, IL-5, noregulation sial.55,56 Considering that many of these experiments were
IL-6, IL-9, IL-10, IL-11, performed using human pathogens, which do not naturally
IL-12, IL-13, IL-16, IL- infect rodents, and that there are several phenotypic and
17, IL-25 functional differences between human and murine Eos, the
Chemokines CXCL1, CXCL5, CXCL8, Leukocyte che-
CXCL9, CXCL10, moattraction
relevance of these data should be reconsidered.55,56 It is note-
CXCL11, CCL2, CCL3, and tissue worthy to mention that, as described subsequently, some of
CCL5, CCL7, CCL11, infiltration of the mechanisms used by Eos in host defense against patho-
CCL13 leukocytes gens may also produce detrimental effects on the host.42,54
Growth factors HB-EGF, LBP, NGF, GM- Growth of various
CSF, TGFα, TGFβ, cell types Eosinophils in Immunoregulation. Accumulating evidence
SCF, VEGF, APRIL suggest that Eos can perform discrete immune regulatory
ECP, eosinophil cationic protein; EDN, eosinophil-derived neurotoxin; EPO, eosinophil
functions mainly through production and release of cy-
peroxidase; IFN, interferon; IL, interleukin; LTD, leukotriene D; LTE, leukotriene E; tokines and other immunomodulatory molecules, or via
MBP, major basic protein; PAF, platelet-activating factor; PGE, prostaglandin E; antigen presentation. For instance, Eos produce cytokines
TNF, tumor necrosis factor.
that are able to act on Eos themselves, the so-called autocrine
cytokines, including IL-3 and GM-CSF, which function, in
part, to prevent apoptosis and prolong Eo survival, once
Interestingly,GM-CSF is among the cytokines produced in these cells are recruited into sites of tissue inflammation,
greatest quantities by Eos.53 such as the lung in asthma.41–43 Eos can regulate their own
Eos are functionally activated via an array of cell surface recruitment, as well as that of DCs and T cells, through the
molecules, including Ig receptors for IgG (Fc γRII/CD32) secretion of CCL3, CCL5, and CCL11. Eo-derived cytokines
and IgA (FcαRI/CD89), receptors for complement-derived may also influence the functions of other immune cells. For
fragments (CR1/CD35, CR3, and CD88), cytokines (IL-3R, instance, Eos play an important role in the regulation of MC
IL-5R, GM-CSF, as well as IL-1αR, IL-2R, IL-4R, IFNαR, functions, including activation, differentiation, maturation,
and TNFRI), chemokines (CCR1 and CCR3), adhesion mol- and survival, mainly via the release of SCF.50–52 Eos have also
ecules (VLA4, α4β7), leukotrienes (CysLT1R and CysLT2R; been implicated in the regulation of B-cell function, not only
LTB4-R), prostaglandins (PGD2 type 2R), PAF, and ligands under homeostatic conditions, but also during immuniza-
for TLRs (TLR1, TLR4, TLR7, TLR9, and TLR10, the TLR7/8 tion responses with alum.57 Eos seem indeed able to modu-
representing an important mechanism for host defense late early stages of B-cell activation and IgM production.57
against viral infections).41–43 The best characterized role of Eos in immune regulation is
their ability to initiate Th2-type responses by modulating
Eosinophil Effector Functions DC and T-cell recruitment and activation. Eo-derived EDN
Eos have historically been considered as end-stage effec- can recruit and activate DCs and, in turn, skew them to a
tor cells, causing damage to parasitic pathogens during Th2-promoting phenotype.58 In several murine models of
helminthic infections or to host tissues in allergic diseases. parasitic infections and airway allergen challenge, Eos have
However, as outlined in the following, accumulating evi- been shown not only to precede Th2 cells arrivals in tissues,
dence suggest that Eos can perform various immune regula- but also to be able to migrate to draining lymph nodes and
tory functions, pointing to a more complex role of these cells polarize Th2 immune responses through the secretion of
not only in regulating inflammation and bridging innate cytokines such as IL-4 and IL-25.42,58 However, human Eos
and adaptive immunity, but also in maintaining epithelial also express indoleamine 2,3-dioxygenase which, through
barrier functions and affecting tissue remodeling.41–43 kynurenine production, inhibits Th1 effector functions.59

Another important role of Eos in immune regulation is their Mast Cells and Basophils
function as antigen-presenting cells. Eos possess the capac- Mast Cell Generalities
ity to internalize, process, and present antigenic peptides MCs were identified as granular cells in the mesentery of the
within the context of surface expressed MHC II, to provide frog by Dr. Von Recklinghausen in 1863 and were named
costimulatory signals to T cells through surface expression “Mastzellen” by Dr Paul Ehrlich in 1878. Initial studies fo-
of molecules such as CD80, CD86, and CD40, and to physi- cused on their histologic characteristics, distribution, and
cally interact with CD4 + T cells.60 In humans, although abundance in health and disease. The discovery of histamine
circulating Eos from healthy donors are generally devoid in 1910, slow-reacting substance of anaphylaxis (now leukotri-
of surface MHC class II expression, they can be induced to enes) in 1938, and IgE in 1966 provided initial insights into the
express MHC class II and costimulatory molecules upon role of MCs in allergic reactions. MCs are tissue-dwelling in-
appropriate cytokine stimulation and after transmigration flammatory cells widely distributed throughout the body and
through endothelial cell monolayers.42,60 Human Eos also are common at perivascular sites and at mucosal surfaces, par-
constitutively express a Jagged1, a Notch ligand, suggesting ticularly in those tissues that form interfaces with the external
a capacity of Eos to provide a polarization signal to naive environment, such as skin, conjunctivae, and intestinal and
CD4 + T cells.61 airway mucosa.63–65 Thus, MCs are strategically placed so as
to function in a first line of host defense.66,67 Morphologically,
Eosinophil Homeostatic Functions. As mentioned, at base-
MCs in tissues appear as round, spindle-shaped, or spider-like
line conditions Eos are present in tissues such as the gut,
cells, with round or oval nuclei, ranging between 7 and 20 μm
mammary gland, uterus, thymus, and bone marrow, via
in diameter, and readily identified using cationic dyes such as
recruitment by the CCL11/CCR3 axis. Although their ho-
toluidine blue or methylene blue.68 These dyes impart a blue-
meostatic role in all these tissues is not fully understood, Eos
to-purple change in color, known as “metachromasia,” which
seem to be mainly involved in regulating the morphogenesis
occurs as a result of their abundant intragranular content of
and maintenance of mucosal organs, as well as the immune
sulfated proteoglycans (eg, heparin and chondroitin sulfates).68
homeostasis of the thymus and bone marrow.41–43 Thymic
Of all other hematopoietic cells, only Bas share this staining
Eos, for instance, are thought to be involved in the MHC
feature with MCs, along with other properties (see Table 20.4
class I–restricted negative selection of double positive thy-
mocytes. In the bone marrow, instead, Eos colocalize with
plasma cells, and by secreting a proliferation inducing ligand
(APRIL) and IL-6, they play a crucial role in long-term Major Features of Mast Cells and
maintenance and survival of the same plasma cells.41,43,62 TABLE 20.4
Basophils
Eosinophils in Tissue Remodeling. A growing body of
Mast Cells Basophils
evidence has proven the ability of Eos to influence tissue
remodeling and fibrosis in many Eo-associated diseases, Origin Hematopoietic Hematopoietic
mainly due to their ability of secreting high levels of TGFβ. stem cells stem cells
Human Eos also produce other profibrotic and angio- Lifespan Months Days
genic factors, such as osteopontin, VEGF, and MMPs.50–52 Primary location Tissues Intravascular
circulation
Furthermore, Eos produce and release NGF and promote
Site of maturation Connective tissues Bone marrow
the extension of neurites in nerve cells. Differentiation Stem cell factor Interleukin-3
factors
Eosinophils in Diseases Histamine content 1 to 15 pg/cell 1 to 2 pg/cell
Eos are currently thought to participate in the patho- Size 7 to 20 μm 5 to 10 μm
genesis of the chronic phases of allergic diseases such as Nucleus Oval or round Segmented
asthma, allergic rhinitis, atopic dermatitis, and hypere- Granules Smaller and more Larger and fewer
numerous compared with
osinophilic syndromes.41–43 Allergic diseases are associated compared with mast cells
with a mild peripheral blood eosinophilia, although in tis- basophils
sues and in the nasal secretions, sputum, and BAL fluid, Peptidoglycans Heparin and Predominantly
Eos can be more significantly elevated. Studies in murine chondroitin chondroitin
models support a role for Eos in airway remodeling, air- sulfates sulfates
way hyperactivity, and mucous production. Elevation of Chymase content High Absent
peripheral blood Eos is also observed in drug reactions, Tryptase content High Low
Lipid mediators PGD2, LTB4, LTC4, LTC4, LTD4, LTE4
helminthic infections, as well as in specific primary im-
LTD4, LTE4, PAF
munodeficiency diseases, most notably Omenn syndrome
and hyper-IgE syndrome. Eosinopenia is typically seen in Activation by
acute bacterial or viral infections and with systemic cor- Substance P Activation No effect
ticosteroid treatment. Finally, the use of Eo-specific anti- Morphine Activation No effect
bodies has demonstrated that almost all human or mouse fMLF No effect Activation
cancers are associated with an important Eo infi ltrate at fMLF, formyl-Met-Leu-Phe; LTB,leukotriene B; LTC, leukotriene C; LTD, leukotriene D;
some point in tumor growth.43 LTE, leukotriene E; PAF, platelet-activating factor; PGD, prostaglandin D.

ment upon inflammation.63,72 Accordingly, while a range of

chemokines are active on human and mouse MCs in vitro,
no MC-selective chemokine has been identified yet.
MC localization to small intestine is reliant on adhesive
interactions controlled by α4β7 integrin, VCAM-1, and mu-
cosal addressin cell adhesion molecule-1 (MAdCAM-1).63
CXCR2 ligands also play a critical role for the constitutive
localization of MC progenitors to the intestine.63,78 In con-
trast to the small intestine, MC progenitors are not abun-
dant in normal lung. However, MCs are detected in the
bronchial epithelium and airway smooth muscle, associated
with pulmonary inflammation and abundant in human
asthma.79 VCAM-1 interactions with both α4β1 and α4β7 in-
tegrins, but not mucosal addressin cell adhesion molecule-1,
are essential for the trafficking of MC progenitors to the
lung during antigen-induced pulmonary inflammation.79 It
is interesting to note that in a model of Aspergillus fumigatus
FIG. 20.7. Formalin-fixed, paraffin-embedded human prostate tissue extract–induced allergic pulmonary inflammation, IgE can
stained with an anti–mast cell tryptase antibody.
influence the number and function of mature MCs, but not
MC progenitor recruitment.80
On the basis of their location, histochemical staining,
for the major similarities and differences between MCs and content of proteases, and reactivity to selected secreta-
Bas). MCs can be cytofluorimetrically identified by their char- gogues and antiallergic drugs, two major subtypes of MCs
acteristic expression pattern of plasma membrane proteins have been described in rodents: mucosal-type MCs, which
serving as markers, including FcεRI (as Bas and a subpopula- express MC protease-1 and -2; and connective tissue–type
tion of DCs), CD23, CD117, and CD203, in conjunction with MCs, which are positive for MC protease-4, -5, -6, and car-
the lack of expression of lineage markers (such as CD2, CD14, boxypeptidase A.81 In humans, mucosal MCs preferentially
and CD19).69,70 Furthermore, MCs can be identified either by express mouse MC protease-1 and -2, whereas connective
biochemical reagents that detect their intracellular proteases tissue MCs express MC protease-4, -5, -6, and carboxypep-
for their enzymatic properties or by immunostaining using tidase A.81,82 Human MCs also exhibit heterogeneity and are
antibodies toward tryptase (Fig. 20.7) or chymase (the latter thus classified by their content of serine proteases as trypt-
representing the primary method of choice for identifying ase-only MC, chymase-only MC, or both tryptase- and
human MCs in tissues).71 chymase-positive MC.81,82 Both in humans and mice, these
Human MCs originate from CD34 + /CD117+ /CD13 + MC phenotypes are reversible in certain microenvironmen-
multipotent hematopoietic progenitors in bone marrow, and tal conditions, and transdifferentiation between the pheno-
then migrate through blood to tissues where they mature.72,73 types has been shown.81,82 Each MC subtype predominates
Details of their differentiation and phenotypic diversification in different locations: tryptase-only MC cells are promi-
have not been definitively identified, but are likely included nent within the mucosa of the respiratory (nose, lung) and
within the CD34 + /KIT+ cell subpopulation and are clearly gastrointestinal tracts, and increase upon mucosal inflam-
distinct from the basophil lineage (which is KIT/CD117– but mation, whereas tryptase- and chymase-positive MC cells
IL-3R+).74 In mice, a hematopoietic stem cell progresses to are prominent within connective tissues such as the dermis,
a multipotent progenitor, a common myeloid progenitor submucosa of the gastrointestinal tract, heart, conjunc-
and a granulocyte/monocyte progenitor.75,76 A monopotent tivae muscularis of the uterus, and perivascular tissues.81
MC progenitor is found in bone marrow and intestine, and In light of murine studies highlighting a “fi ne tuning” on
a common basophil/MC progenitor is also found in mouse the MC functions and effector properties by cytokines or
spleen, observations that await study in human. After their matrix proteins, it seems likely that human MCs will prove
homing in the tissues, maturation of the MC precursors is to be even more heterogeneous than currently thought.
dependent on SCF expressed on the surface of fibroblasts, Accordingly, MC phenotype, behavior, and responsive-
stromal cells, and endothelial cells.68,77 Accordingly, in vitro ness may be dramatically altered by cytokines, including
studies using mouse and human MCs confirm that soluble IL-4 (which upregulates the expression of Fc εRI and is
SCF protects MCs from apoptosis and induces their prolif- particularly pivotal in regulating functional responses of
eration, chemotaxis, and also some degree of activation and MCs in mucosal inflammation), IL-5 (which promotes MC
secretion.68 Consistently, mutant mice lacking KIT- or SCF- proliferation in the presence of SCF), and IFNγ (which in-
mediated functions are often used as experimental tools to duces the expression of the high-affi nity activating Fc γRI/
implicate a biologic role of MCs in given models.68 While CD64 and also decreases MC numbers).68 Furthermore,
there are several studies on the mechanisms of localiza- MCs express an array of adhesion and immune receptors
tion of MCs to different tissues, there is little information that may assist in the recognition of invading pathogens and
about specific molecules that could regulate MC-progenitor in the sampling of different stimuli coming from the micro-
trafficking to tissues, movement, and incremental recruit- environment.67 This versatility is reflected in the numerous

TABLE 20.5 Major Basophil and/or Mast Cell–Derived Mediators

Class Mediators Physiologic Effects
Preformed mediators Histamine Vasodilation, angiogenesis, mitogenesis, suppressor
of T-cell activation
5-HT Leukocyte regulation, vasoconstriction, pain
Chymase Tissue damage, pain, angiotensin II synthesis
Tryptase Activation of PAR, inflammation, pain, tissue damage,
degradation of antigens and peptides
kininogenases Synthesis of kinins, pain
Nitric oxide synthase Nitric oxide production
Carboxypeptidase A Degrades enzymes
CRH Inflammation, vasodilation, mast-cell VEGF release
Endothelin Sepsis
Kinins Inflammation, pain, vasodilation, mast cell trigger
Somatostatin Anti-inflammatory effects, mast cell trigger
VEGF Neovascularization, vasodilation
Chondroitin sulfate Connective tissue component, anti-inflammatory,
mast cell inhibitor
Heparin Angiogenesis, NGF stabilization, mast cell inhibitor
Lipid mediators LTB4, LTC4, PAF, PGD2 Leukocyte chemotaxis, vasoconstriction, pain,
platelet activation, vasodilation, inflammation,
bronchoconstriction
Cytokines TNFα, TGFβ, IFNα, IFNβ, IFNγ, IL-1α, IL-1β, IL-3, Inflammation, leucocyte migration/proliferation
IL-4, IL-5, IL-6, IL-8, IL-9, IL-10, IL-11, IL-12,
IL-13, IL-15, IL-16, IL-18, IL-25, SCF, MIF
Chemokines CXCL8, CCL3, CCL2, CCL7, CCL13, CCL5, CCL11, Chemoattraction and tissue infiltration of leukocytes
CCL19
Growth factors CSF, GM-CSF, bFGF, VEGF, NGF, LIF Growth of various cell types
Antimicrobic products Nitric oxide, superoxide, antimicrobial peptides Microbial killing
bFGF, fibroblast growth factor; CRH, corticotropin-releasing hormone; CSF, colony-stimulating factor; GM-CSF, granulocyte-macrophage colony-stimulating factor; IFN, interferon;
IL, interleukin; LIF, leukemia inhibitory factor; LTB, leukotriene B; LTC, leukotriene C; MIF, migration inhibitory factor; NGF, nerve growth factor; PAF, platelet-activating fac-
tor; PAR, proteinase activated receptor; PGD, prostaglandin D; SCF; stem cell factor; TGF, transforming growth factor; TNF, tumor necrosis factor; VEGF, vascular endothelial
growth factor.
IgE-dependent and -independent activation pathways that periphery with a half-life thought to be of a few days.85,88
modulate the quality and magnitude of MC responses and Although not predominantly tissue-dwelling cells, Bas are
cytokine release. MCs are endowed with a range of potent able to infi ltrate inflamed tissues, particularly at sites of
proinflammatory effector molecules that may be selectively allergic inflammation within several hours after exposure
released not only through Fc εR1 activation, but also in re- to allergens, and are often accompanied by a simultane-
sponse to a variety of other receptors including C3aR and ous influx of eosinophilic granulocytes and Th2 lympho-
C5aR, Fc γRIIa/CD32a, KIT/CD117/SCF-R, IL-3R, IL-4R, cytes.82,85,89 Bas can be cytofluorimetrically identified by
IL-5R, IL-9R, IL-10R, GM-CSFR, IFNγR, CCR3, CCR5, their selective pattern of plasma membrane proteins serving
CXCR2, CXCR4, NGFR/TRKA, the type 1 receptor for as markers, including Fc εRI (like MCs), CD123 (the α chain
cys-LTs (CysLT1R), the adenosineA3 receptor, and many of the IL-3 receptor), CD11b, and CD13, in conjunction
TLRs, among others.68 Depending on the type, property, with the absence of other lineage markers (such as CD2,
strength, and combination of the stimuli that they receive, CD14, CD16, CD19, and MHC class II).89 They also express,
MCs secrete a diverse and wide range of biologically active and dynamically respond to, ligands for a variety of func-
products (Table 20.5) that can trigger, d irect, or suppress tional cytokine/chemokine receptors (eg, IL-5R, GM-CSFR,
the immune response.68,83,84 CCR2, and CCR3), complement receptors (CD11b, CD11c,
CD35, and CD88), prostaglandin receptors (CRTH2), Ig
Basophil Generalities Fc receptors (Fc εRI and Fc γRIIb), and TLRs. For example,
Bas are the rarest of the granulocytes (typically constitut- following Fc εRI activation (which typically triggers granule
ing 0.5% to 1.5% of peripheral blood leukocytes), are 5 to exocytosis and mediator release), a number of basophil cell
10 μm in diameter, and exhibit a bean-shaped or bilobed surface markers, including CD13, CD63, CD107a, CD164,
condensed nucleus.85 Compared to MCs, they have little and CD203C, may increase.85,89,90
proliferative capacity and possess fewer, but larger (up to 1.2
μm) round metachromatic granules (Fig. 20.8) enriched in Mast Cell and Basophil Mediators
histamine (stored at about 1 to 2 pg per cell).86,87 Bas de- The mediators produced by MCs and Bas are schematically
velop from CD34 + progenitors, differentiate and mature in divided into preformed mediators, newly synthesized lipid
the bone marrow in response to IL-3, and circulate in the mediators, and cytokines/chemokines (see Table 20.5).82,86,89

FIG. 20.8. A Polymorphonuclear Basophil

Circulating in Peripheral Blood. From
Anderson’s Atlas of Hematology; Anderson,
Shauna C., PhD. Copyright 2003, Wolters
Kluwer Health/Lippincott Williams & Wilkins.
Such subdivision is not exclusive as TNFα , for instance, include LTD4 and LTE4).86,92 Activated Bas can produce
occurs as both preformed and as a newly synthesized mol- the same lipid mediators but PGD2 (as they lack the PGD2
ecule.91 Preformed mediators, including histamine or MC synthase).82,86,93 PGD2 functions as bronchoconstrictor and
neutral proteases (tryptase, chymase, cathepsin G, and attracts both Eos and Bas, LTB4 attract neutrophils and ef-
carboxypeptidase) are stored in cytoplasmic granules and fector T-cells, while cysteinyl leukotrienes attract Eos and
form complexes with the negatively charged, very abun- work as potent bronchoconstrictors, in addition to promote
dantly expressed, sulfated proteoglycans.82,92 Upon MC/Ba vascular permeability and to induce mucus production.
activation, the granules fuse with the plasma membrane MCs produce several cytokines, including IL-3, IL-4,
and, within minutes, release their content into the extracel- GM-CSF, IL-5, IL-6, IL-10, and IL-13, growth factors, includ-
lular environment where the various mediators dissociate ing bFGF, SCF, VEGF, and several chemokines, including
from proteoglycans.68,85 Other than being secreted upon CXCL8 and CCL3.82,86,92 In addition, MCs are considered
stimulation, some molecules are also constitutively released, the only cells storing very abundant levels of preformed
as in the case of MC tryptase. Consequently, tryptase levels TNFα , whose critical role in host defense against bacterial
are considered as a parameter that reflects the MC burden, infections has been widely highlighted: in fact, specific ex-
and, in fact, they significantly increase in systemic masto- perimental animal models lacking MC-derived TNFα result
cytosis.82,92 Curiously, the function of tryptase in vivo is in drastically reduced neutrophil influx and significantly
still unknown, even though, in vitro, it digests fibrinogen, increased mortality.91 Bas are a major source of GM-CSF
fibronectin, prourokinase, proMMP-3, protease-activated and VEGF, and release significant amount of chemokines
receptor-2, and complement component C3; it activates (CXCL8, CCL3, CCL11) in response to IgE cross-linking.
fibroblasts as well. The functions of MC/Ba-derived pro- Chief among the cytokines produced by Bas are IL-4 and IL-
teoglycans and proteases are less well understood, but pro- 13, which are released in response to Fc εRI or C5a stimula-
teases provide some protection against snake and insect tion within the range of 20 to more than 600 pg/million and
venoms.82,92 Differently from MCs, humans Bas do not ex- < 20 to 2000 pg/million cells, respectively.82,86,93
press chymase, appear to contain very low levels of hepa-
rin and tryptase (with some exceptions), but contain MBP Mast Cell and Basophil Effector Functions
and Charcot–Leyden crystal (which are typical eosinophil MCs and Bas, although derived from distinct progenitors,
products). Activated Bas have been shown to also contain have been recognized to express partially overlapping func-
granzyme B and to directly kill target cells, at least in part by tions in many aspects of natural and acquired immunity.86,87
this cytotoxic serine protease.82,90,93 More importantly, Bas, Traditionally, MCs and Bas, together with Eos, are consid-
together with neutrophils, seem to be the exclusive periph- ered essential components of IgE-mediated classic type 1
eral blood source of histamine. hypersensitivity (allergic) reactions and allergic inflam-
Newly synthesized lipid mediators by Fc εRI-activated mation (see Chapter 45). Nonetheless, MCs and Bas have
MCs are PAF, PGD2, and LTA4. The latter is then converted also the ability to release and synthesize highly bioactive,
to LTB4 or conjugated with glutathione to form LTC 4 (the proinflammatory, and cytotoxic substances, independent
parent compound to the cysteinyl leukotrienes, which also of signaling through Fc εRI. Therefore, these cells, besides

being important modulators of allergic reaction, can also and -independent inflammatory reactions, mainly by se-
contribute to leucocyte recruitment, stromal and tissue cell creting soluble mediators and cytokines or by functioning
activation, modulation of immune reactions, tissue remod- as antigen-presenting cells. These cells contribute to the ini-
elling, and angiogenesis.84,89,90,94 tiation of acquired immunity by orchestrating DC migra-
tion, maturation, and function and by interacting with T
Mast Cells and Basophils in Allergy. Allergen-specific IgE and B cells.84,89,90,94 MCs have the potential to influence DC
production, with subsequent fi xation of IgE to Fc εRI recep- functions, mainly through the release of histamine, PGD2,
tors on MCs and Bas, followed by rapid degranulation and PGE2, LTB4, or cytokines such as TNF, IL-1, IL-16, IL-18,
release of histamine and specific eicosanoids (eg, LTC4), and CCL5. Generally, both in vitro and in vivo evidence in-
is central to the initiation and propagation of immediate dicate that MCs predominantly inhibit IL-12p70 production
hypersensitivity reactions.65,82 It has been widely accepted and induce a Th2-promoting phenotype in DCs.94,101 MCs
that MCs contribute significantly to acute inflammatory can also modulate B-lymphocyte functions by supporting
reactions to antigens/allergens toward which the host bears their survival, proliferation, and IgA production, mainly
antibodies of the IgE class.95 MCs are responsible for virtually through the expression of CD40L and IL-6.94,102 Evidence of
all of the increased vascular permeability and tissue swelling cell-contact interactions between MCs and B cells in vivo
early in the IgE-dependent passive cutaneous anaphylactic have been found to occur in secondary lymphoid organs and
response.95,96 If the stimulation is of more persistent or of inflamed tissues.94,103
more severe nature, acute response may undergo transition MC surface molecules and secreted products can also
into a late-phase reaction, which, except for the time scale influence various aspects of the biology of T cells, for exam-
ranging from a few to several hours from initial antigen ple, by polarizing naive T cells to Th1, Th2, Th17, and regula-
challenge, is characterized by new recruitment of leukocytes tory T (Treg) cells or by modulating the functions of distinct
(Eos, Bas, and Th2 lymphocytes) to the site of inflamma- T-cell subsets.94,101 Interestingly, MCs through IL-6 secretion
tion.65 Bas play a crucial role in sustaining allergic chronic and via OX40/OX40L contact have recently been shown to
inflammation upon their entering the affected tissue, not counteract Treg-mediated suppression, thus leading to the
only due to the release of stored and newly synthesized me- establishment of Th17-mediated inflammatory responses.104
diators, such as histamine and LTC4, but also via the release MCs can also induce antigen-specific T-cell proliferation and
of high amounts of the cytokines IL-4 and IL-13.89,90,97 activation (of both CD8 + and CD4 + T cells) due to their ex-
pression of MHC class I and II molecules, of costimulatory
Mast Cells and Basophils in Innate Immunity and molecules of the B7 family (such as ICOS-L, PD-L1, CD80,
Host Defense. Recent studies have demonstrated that MCs and CD86), and of members of the TNF/TNF receptor fami-
express many PRRs, including TLRs 1 to 10 and nucleotide lies (such as OXOL and CD40L).84,94,101 Finally, MCs produce
oligomerization domain–like Receptors (NLR) (includ- several mediators with anti-inflammatory activities, includ-
ing nucleotide oligomerization domainreceptors and the ing TGFβ, IL-4, IL-10, and IL-9, and have been shown to play
NLRP3). Expression of various PRRs thus permits MCs a crucial role in transplantation engraftment, for instance to-
to respond to both pathogen-associated molecular and ward allogeneic skin graft.94,101,105 Most likely, MCs manifest
“danger signals” resulting from cell stress or injury. Several predominantly proinflammatory effects in the earlier phase
MC functions triggered by PRR have been implicated in of the immune response, while the anti-inflammatory effects
host defense.84,94,98 These latter include enhancement of the of these cells are more pronounced at the later phase of the
recruitment or function of granulocytes, phagocytosis- response to limit ongoing inflammation.105
dependent bactericidal activities, and proteolytic degra- Bas have the ability to regulate the acquired immune
dation of endogenous mediators, which would otherwise response, mainly by potentiating humoral responses and
be elevated to toxic levels, such as endothelin-1 and neu- promoting Th2 polarization.86,89 Bas can enhance B-cell sur-
rotensin. Secretion of proteases/enzymes and formation vival and proliferation, and Ig production mainly through
of extracellular traps that contain antimicrobial peptides, the release of IL-4, IL-6, BAFF, and APRIL.89,106,107 The abil-
histone, deoxyribonucleic acid, and tryptase have also been ity of Bas to polarize Th2 immune responses is dependent
proposed as potential mechanisms by which MCs exert on their function as antigen-presenting cells and on IL-4
protective functions.94,98 In addition to their role in bacte- secretion. These Th2-promoting functions of Bas have been
rial infections, MCs can promote host resistance to certain demonstrated in responses to protease allergens, helminthic
parasite and virus infections. However, the mechanisms in- parasites, or antigen-IgE complex in vivo.108–110 Recently, the
volved in these functions have not been fully elucidated.94,98 Th2-promoting functions of Bas have been linked to the
Differently from MCs, the role of Bas in host defense is less development of systemic lupus erythematosus in mice de-
studied, especially in bacterial or viral infections. Some ficient of the Src family protein tyrosin kinase Lyn, and in
evidence for a role of Bas in host defense against parasitic human patients.111
infections and tick infestations has been reported, but the
molecular mechanisms underlying the Ba-mediated protec- Mast Cells and Basophils in Disease
tion in these processes remain to be determined.89,99,100 The most striking increase in MCs occurs during parasitic
diseases and mastocytosis (the latter being a pathologic ex-
Mast Cells and Basophils in Immunoregulation. MCs and cess of MCs, most notably in the skin, bone marrow, gas-
Bas manifest immunoregulatory functions in IgE-dependent trointestinal tract, spleen, liver, and lymph nodes, that is

usually caused by gain-of-function mutations of KIT).68,112 involved in cell–extracellular-matrix adhesion and in cell–
The importance of MCs in IgE-allergic reactions (includ- cell contacts, their extensive assortment of soluble pro- and
ing rhinitis, urticaria, and asthma) is emphasized by the anti-inflamatory mediators, may profoundly influence the
increased numbers of these cells seen in affected tissues in development, intensity, and duration of adaptive immune
both mice and humans.68,95 MC hyperplasia and increased responses that ultimately serve for host defense, allergy, and
MC products at sites of tissue injury have also been observed autoimmunity. A representative example of this aptitude is
in chronic inflammatory and autoimmune diseases, such the plasticity that MCs manifest as a result of their recipro-
as rheumatoid arthritis and multiple sclerosis. A pathologic cal interactions with the T-cell populations exerting regu-
role of MCs in these diseases has been suggested mainly latory or activatory functions in normal and/or pathologic
by the utilization of the K/BxN mouse model of antibody- immune responses.101,105 In this context, it has been demon-
induced arthritis and the experimental autoimmune en- strated that CD4 + CD25 + Treg cells establish a cell-cell contact
cephalomyelitis mouse model of multiple sclerosis.94,112 with MCs through the OX40:OX40L axis, which regulates
Moreover, studies in mice indicate that activation of MCs the MC degranulation threshold and contributes to tissue
via the NLRP3 can contribute to IL-1β overproduction and tolerance.104,118 However, in an inflammatory environment,
chronic urticarial rash in subjects with cryopyrin-associ- this MCs/CD4 + CD25 + Treg cell interaction induces a Th17
ated periodic syndrome, a disorder associated with NLRP3 switch by Treg cells.104,118 On the same line, it is remarkable
mutations.113 There is evidence for MCs in promoting, but that, under particular settings, granulocytes and MCs dis-
also in protecting against, tumor growth. The accumula- play functions that in the past were ascribed only to immune
tion of MCs at the periphery of tumors has been observed adaptive cells. For instance, it has been recently demon-
in both rodent models and in a diverse array of tumors in strated that neutrophils, among their various immunoregu-
humans. It is generally thought that MCs promote early an- latory functions,23 can also display a B-cell–helper function
giogenesis events in tumor development through the release that promotes Ig class switching, Ig somatic hypermutation,
of mediators such as VEGF, CXCL8, angiopoietin-1, bFGF, and antibody production by activating marginal zone B cells
heparin, and proteases.114 Finally, MCs could also contrib- through BAFF, APRIL, and IL-21 secretion.119 Similarly to
ute to the pathogenesis of acute or chronic vascular events: neutrophils, MCs are also able to provide costimulatory sig-
accordingly, acute myocardial infarctions show elevations of nals that sustain B-cell expansion and drive the development
both histamine and mature β -tryptase, suggesting that MC of IgA-oriented humoral immune responses,102 thus ascrib-
activation occurs concomitantly with ischemia.94,112 ing a B-cell–helper function that in the past was retained
As far as Bas, it has remained obscure whether these exclusive of T cells.
cells actually play a crucial role or are just redundant with Novel activities have been described for myeloid cells
MCs in allergic reactions in vivo.89,97 Human Bas have been also in the context of host defense, such as for instance in
shown to release IL-1β and TNFα upon the crosslinking of the case of Bas that, during primary and secondary expo-
surface-bound IgD, but not IgE, suggesting a possible role of sure to parasites, might display different roles.93,120 Indeed,
these cells in autoinflammatory disorders such as hyper-IgD in a murine model of primary helminthes infection,121 Bas
syndrome.115 Bas may also contribute to the production of often accumulate in affected tissues where they produce
autoantibodies that cause lupus nephritis.116 large quantities of Th2 cytokines that not only regulate
the eosinophil recruitment to the lung and the IgE produc-
tion, but also favor mucosal MC infi ltration.121 However,
CONCLUSION their major defensive role occurs during the secondary
In the last few years, our perception of granulocytes and MCs infection, in which worm expulsion is more rapid (by 5
has changed dramatically. In fact, there has been mounting days) than in the primary one (by 10 days postinfection),
evidence that the function of these cells is not limited to act- as revealed by Ba depletion studies that cause an impaired
ing as first line of defense against invading pathogens, but is worm expulsion by mice, regardless of the presence or the
extended to perform additional and unexpected activities in absence of MCs.122
strict collaboration with adaptive immune and other non- In view of the continuously emerging fi ndings in the
immune cells. Thus, cells of innate and adaptive immunity field, it is predictable that in the next years there will be the
together orchestrate complex functional programs to pro- discovery of additional, unsuspected biologic features that
mote host defense, control the development of self-tolerance, granulocytes and MCs possess.
and avoid autoimmunity. In this context, the gene expres-
sion pattern and phenotype of the cells discussed in this
chapter must rapidly change in a coordinate, time-depen- ACKNOWLEDGMENTS
dent manner in response to microenvironmental soluble and This work was supported to MAC by grants from Ministero
cellular signals.117 Granulocytes and MCs, in view of their dell’Universita’ e della Ricerca (MIUR-PRIN 2009 MFXE7L),
wide variety of membrane receptor are able to mediate deliv- and Associazione Italiana per la Ricerca sul Cancro (AIRC-IG
ery of costimulatory signals, their broad array of molecules 11782). N.T. holds a FIRC fellowship.

The Major Histocompatibility

CHAPTER
21 Complex and Its Proteins
David H. Margulies • Kannan Natarajan • Jamie Rossjohn • James McCluskey
INTRODUCTION polymorphism of genes of the Mhc and the identification of

The extended family of glycoproteins known as major his- a large number of human diseases that are profoundly in-
tocompatibility complex (MHC) and MHC-like molecules fluenced by either defects of MHC expression or by poly-
is now recognized to comprise a class of receptors, usu- morphic variants. Every student of immunology must be
ally expressed at the surface of somatic cells of vertebrates, acquainted with the basic biology of MHC-encoded mol-
which confer a wide range of functions in the regulation of ecules, and the student’s comprehension of the regulation
the immune response. Their cumbersome moniker, MHC, of the immune response, encompassing the inflammatory,
derives from the early determination that the genes that en- NK, T-cell, and B-cell responses, will be lacking unless the
code these molecules control one of the fundamental mani- function and ongoing evolution of the MHC is understood.
festations of an immune response, skin graft rejection, but We stress that the evolution of the MHC is ongoing be-
contemporary immunologists, virologists, and tumor bi- cause for no other set of genetic markers is there such a large
ologists rarely think about these molecules with allogeneic database cataloguing genetic polymorphism; as a result, it is
(ie, genetically distinct) skin transplantation in mind. The apparent that not only are new alleles being identified, but
name MHC is a general label, and specific immunologic that many of these new alleles have arisen recently. Dramatic
functions or molecular structures invoked when a particu- examples of the ongoing evolution of MHC molecules are
lar MHC is mentioned are dependent on biologic context. In seen not only in new variants of the deoxyribonucleic acid
recent years, our understanding of the structure, evolution, (DNA) sequences observed by those that perform the neces-
and molecular basis of MHC-based interactions has blos- sary typing of Mhc genotypes for transplantation but also
somed, and our ability to exploit engineered forms of these in surveys of the sequences of viruses, particularly the cy-
molecules has also expanded enormously. Our blueprint for tomegaloviruses (CMVs), which appear to have purloined
this chapter is to give an overview of MHC-regulated im- host Mhc genes for immunoevasion. In sequence gazing,
munologic function and then to summarize the explosive we observe the interplay between host and pathogen, as the
growth in understanding of the molecular structures of pathogenic viruses, exploiting their rapid generation times,
MHC molecules and their ligands. Of necessity, we discuss develop new variations on the MHC theme for distinct
briefly some aspects of the cell biology and cellular matura- purposes.
tion of MHC molecules but leave the detailed description of The prototypes of MHC molecules are the MHC class I
these essential aspects of the classic MHC molecules to the (MHC-I) and class II (MHC-II) molecules, obligate cell-
chapter on antigen presentation. Basic aspects of the genet- surface heterodimers that bind and display self- or foreign
ics of MHC molecules not only descriptions of the genetic peptides at the cell surface so that T-cell receptors (TCRs) or
loci that encode them but also discussion of what we know NK cell receptors can interact with the molecular complexes
of the evolution of their encoding genes will be raised. in an MHC- and peptide-dependent manner.
Functional glycoproteins are molecular machines and MHC molecules are crucial for both TCR- and NK-
their cellular and immunologic activities depend on the shape mediated interactions. Our efforts will be to preserve some
of these macromolecules, their surface charge and ability to sense of the historical development of this exciting field
interact with solvent, as well as the flexibility and relationship of study and also to focus on paradigmatic genetic, struc-
of their structured domains and unstructured regions. To un- tural, and functional features that unify this extensive gene/
derstand molecules of the MHC in a rational functional con- protein family. Finally, the Mhc provides a genetic link from
text, we consider it crucial to describe their overall structures immune responsiveness to autoimmune disease—those
as well as their structural interactions with bound ligands and well-known strong associations of particular Mhc genes to
their binding to their cognate receptors found on T cells and particular human diseases—and we will provide an outline
natural killer (NK) cells. Despite a large and ever increasing of the molecular basis for such associations.
database of MHC and MHC-like structures, a number of gen-
eral questions remain unanswered. We will present some of Mhc Nomenclature—Dialects of Mhc
the unsolved mysteries of the evolution and function of the The names of the genes and proteins that are critical to
MHC-like molecules and their viral homologs. understanding the Mhc reflect the historical discovery of
The importance of MHC molecules is underscored their functions and differ among different species. Often
by two general characteristics: the enormous degree of confusing to students of immunology, the nomenclature of
487

the Mhc differs for different species and must deal not only
with a number of distinct, but usually linked, genetic loci Nomenclature of HLA Loci
TABLE 21.1
but also with their encoded molecules, many of which are and Alleles
heterodimeric glycoproteins. Because the first genes of the
Nomenclature Definition
Mhc identified were those that encoded cell surface mole-
cules that could be detected by antibodies or by transplanta- HLA The HLA region and prefix for
tion responses, these are the ones that are referred to as Mhc HLA gene
genes. Now we know of more than 400 genes that map to HLA-DRB1 Particular HLA genetic locus
the human or mouse Mhc, although technically they are all HLA-DRB1*13 Group of alleles encoding DR13
“Mhc” genes, the “MHC” molecules refer specifically to the antigen or with sequence
similarity to other DRB1*13 alleles
MHC-I or MHC-II molecules that are related in structure
HLA-DRB1*13:01 A specific HLA allele
and function. (Genes for several complement components HLA-DRB1*13:01:02 An allele that differs by a synonymous
and related molecules also map here and are occasionally mutation from DRB1*13:01:01 (ie,
referred to as MHC class III.) Other Mhc-encoded molecules DNA sequence difference but no
with distinct structure and function are usually referred to amino acid sequence difference)
by their specific names. Molecules that exhibit structural HLA-DRB1*13:01:01:02 An allele that contains a mutation
similarity with MHC-I molecules, whether they map to outside the coding region
the Mhc or not, may be called as a group “MHC-Ib” to dis- distinguishing it from DRB1*13:01:02
tinguish them from the “classical” MHC-I molecules, also HLA-A*24:09N A “null,” unexpressed allele
referred to as “MHC-Ia.” Another subset of molecules that HLA-A*30:14L An allele encoding reduced or “low”
cell surface expression
exhibit the MHC-I fold but that are expressed by viruses
HLA-A*24:02:01:02L An allele encoding a reduced or
may be called “MHC-Iv” molecules.1,2 “low” cell surface expression,
By convention, genetic loci are indicated by designations where the mutation is outside the
in italics, and the encoded protein products or phenotypic coding region
descriptions are shown in a standard font. The extended ge- HLA-B*44:02:01:02S An allele encoding a protein that
netic region is referred to as the complex; thus the general is expressed as a “secreted”
term used for all species is the Mhc or MHC. The mouse molecule only
Mhc is referred to as H2 (previously called H-2) because it HLA-A*32:11Q An allele for which a previous
was the second genetic locus involved in control of expres- effect on surface expression
has not been confirmed and is
sion of erythrocyte antigens identified by Gorer.3,4 The Mhc
considered “questionable”
in the rat is known as RT1, the human locus is known as
HLA (for human leukocyte antigen), DLA for the dog, GPLA DNA, deoxyribonucleic acid; HLA, human leukocyte antigen.
Source: Adapted from the designations summarized by the WHO Nomenclature
for the guinea pig, SLA for the swine, and RLA for the rabbit. Committee For Factors of the HLA System, Stephen G.E. Marsh, http://hla.alleles.org.
For other species, based on a suggestion by Klein,5 the taxo- Also available at the IMGT website: www.ebi.ac.uk/imgt/hla/nomenclature/index.html.
nomic name forms the basis for the designation, contribut-
ing the first two letters of the genus and the first two of the
species to name the locus. Thus, we have Patr for the chim- Histocompatibility Workshops. The HLA Dictionary is a
panzee, Pan troglodytes ; Gogo for the Gorilla, Gorilla gorilla ; summary of HLA-A, HLA-B, HLA-C, HLA-DRB1/3/4/5,
Mamu for the Rhesus macaque, Macaca mulatta; Mane for and HLA-DQB1 alleles and their association with serologi-
the pig-tailed macaque, Macaca nemestrina, Mafa for the cy- cally defined HLA-A, -B, -C, -DR, and -DQ antigens, respec-
nomolgous monkey, Macaca fascicularis ; and Papa for the tively, available free online at the IMGT/HLA database site.7
Bonobo, Pan paniscus. A single site for curated sequences of As defined by the Nomenclature Committee (http://hla.
many species of MHC molecules is the IPD-MHC Database alleles.org/nomenclature/naming.html), each HLA allele
at www.ebi.ac.uk/ipd/mhc/.6 The most important and name has a unique number corresponding to up to four sets
widely studied models are those of the human, the mouse, of digits separated by colons (Table 21.1 and Fig. 21.1). The
and the rat. For the mouse and rat Mhc, the search engine on digits before the first colon describe the type, which often
the Jackson Laboratory Web site (www.informatics.jax.org/ corresponds to the serologic antigen carried by an allotype
mgihome/nomen) is the most comprehensive. (a genetically distinguishable form of the molecule within
The naming of new HLA genes and alleles and their qual- the same species). The next set of digits lists the subtypes,
ity control are governed by the World Health Organization numbers being assigned in the order in which DNA se-
Nomenclature Committee for Factors of the HLA System. quences have been determined. Longer names are assigned
Software conversion tools to assist in gene/protein identi- only when necessary. Precise designation of human genes is
fication are to be found at the IMGT/HLA database (www by a nomenclature including a number following the locus
.ebi.ac.uk/imgt/hla/dictionary.html). Regular updates on (eg, HLA-A*01:01 and HLA-DRB1*01:01). Summaries of the
HLA nomenclature featuring new alleles can be found in hundreds of human alleles that have been identified are in
the journal Tissue Antigens and via http://hla.alleles.org/ Tables 21.2 and 21.3. Comprehensive databases of human
nomenclature/nomenc_updates.html. The standardization Mhc genes are maintained at the IMGT/HLA database (see
of serologically defined HLA antigens has been achieved by previous link) 8,9 where, at the time of this writing, more than
the exchange of typing reagents and cells in the International 7000 HLA class I and class II alleles have been tabulated.

CHAPTER 21 THE MAJOR HISTOCOMPATIBILITY COMPLEX AND ITS PROTEINS | 489
Hyphen used to separate Suffix used to denote loci with respect to measurable phenotypes and can be
gene name from HLA prefix changes in expression expected to play an important role in studies of immune
recognition.15
Separator Field Separators
The Immunologic Function of MHC Molecules
MHC molecules are a molecular reflection of the health sta-
HLA-A*02:101:01:02N tus of either the cell that synthesizes them (for MHC-I mol-
ecules) or of the local environment in which the cell resides
HLA Prefix Gene Field 4; used (for MHC-II). The structure of the MHC molecule depends
to show
not only on the amino acid sequence of the two polypeptide
differences in
a non-coding chains (α and β for MHC-II; heavy [or α] and β2m [for the
Field 1; allele group
region light chain] for MHC-I) that form the core of the complex but
also of the variable bound peptide that forms an integral part
of the trimer. The MHC molecule, governed by the sequence
Field 2; specific HLA protein
of the encoding structural genes for the Mhc-I heavy chain
and the Mhc-II α and β chains, as well as other genes involved
Field 3; used to show a synonymous DNA
in antigen processing and presentation, must satisfy at least
substitution within the coding region two distinct recognition functions: the binding of peptides
or in some cases nonpeptidic molecules and the interaction
with either T or NK cells via their respective receptors. The
FIG. 21.1. Summary of Current HLA Gene Nomenclature. This illus-
tration explains the nomenclature for HLA alleles as described in TCR may augment its interaction with the MHC molecule
http://hla.alleles.org/nomenclature/naming.html.164 by virtue of interaction with a T cell–expressed coreceptor
(cluster of differentiation [CD]8 for MHC-I and CD4 for
MHC-II). Some NK receptors may also serve as coreceptors
when expressed on T cells.16 The binding of peptides by an
The commonly used term “haplotype” refers to the link- MHC-I or MHC-II molecule is the initial selective event that
age of particular alleles at distinct loci that occur as a group permits the cell expressing the MHC molecule (the antigen-
on a parental chromosome.10 The concept of haplotype is presenting cell [APC] or when this cell is to be the recipient of
important in typing the HLA loci in the human where the a cytolytic signal, the target cell) to sample fragments derived
linked Mhc genes (those genes in “linkage disequilibrium”) either from its own proteins (for MHC-I–restricted antigen
of one chromosome of one parent will generally segregate as presentation) or from those proteins ingested from the im-
a group to the children. Individual haplotypes of the Mhc in mediate extracellular environment (for the case of MHC-II).
the mouse are referred to by a lowercase letter superscript The biochemical steps involved in the production of antigen
as H2b, H2d, or H2k. Thus, the H2k haplotype refers to the fragments from large molecules are collectively known as
full set of linked genes, H2-K k, H2-IA k, H2-IEk, H2-Dk, and “antigen processing,” whereas those that concern the binding
extends to the genes of the Q and T regions as well.11,12 (Some of antigen fragments by MHC molecules and their display at
haplotype designations, such as H2a, refer to natural recom- the cell surface are known as “antigen presentation.”
binants and thus have some of the linked genes from one Specifically, MHC-I glycoproteins gather from the cell’s
haplotype and some from another.) A complication that de- biosynthetic pathway fragments of proteins derived from
mands the precise use of gene and encoded protein names is infecting viruses, intracellular parasites, or self-molecules,
that the number of genes in a particular homologous genetic either expressed normally or in a dysregulated fashion as a
region may differ between strains or between individuals. In result of tumorigenesis, and then display these molecular
the mouse, although some strains have only a single gene at fragments, in complex with the mature MHC-I molecule,
the D locus (H2-Db for instance), other strains may have as at the cell surface.17–21 The cell-bound MHC-I/ β2m/peptide
many as five genes in homologous region (H2-Dd, H2-D2d, complex on the APC is exposed to the extracellular milieu
H2-D3d, H2-D4d, and H2-Ld).13,14 and is available for interaction with either T cells or NK cells.
Table 21.4 summarizes the haplotypes of common mouse The T cell bearing an αβ receptor recognizes the particular
strains. Included in this listing are a number of congenic MHC/peptide complex by virtue of a specific physical bind-
inbred mouse strains, strains that contain the Mhc derived ing interaction. Each T cell is representative of a clonal pop-
from one strain, and the remaining background genes ulation and bears a unique TCR encoded by somatically re-
from another. A regularly updated and comprehensive list- arranged TCR genetic elements. T cells bearing αβ receptors
ing can be obtained at http://jaxmice.jax.org/literature/ undergo a complex selective process in the thymus. Only a
catalog/mhc_h2_haplotypes.pdf. An ongoing project, the small proportion of T cells that enter the thymus ultimately
“Collaborative Cross,” aims to establish a panel of mouse re- reach peripheral lymphoid organs, such as lymph nodes and
combinant inbred lines derived from eight diverse founder spleen. A particular TCR can only bind a very limited selec-
mouse strains as a resource for mammalian genetics. These tion of MHC/peptide complexes. The recognition by T cells
lines, when their breeding is completed, will provide an op- is considered “MHC restricted” in that only a limited set
portunity for mapping a wide variety of quantitative trait of MHC molecules can bind a particular TCR and is also

TABLE 21.2 Listing of HLA Class I Allelesa

HLA-A Alleles HLA-B Alleles HLA-C Alleles
A*01:01:01:01—A*01:104 B*07:02:01—B*07:144 C*01:02:01—C*01:61
A*02:01:01:01—A*02:336 B*08:01:01—B*08:80 C*02:02:01—C*02:55
A*03:01:01:01—A*03:135 B*14:01:01—B*:14:28 C*03:02:01—C*03:139
A*11:01:01—A*11:112 B*15:01:01:01—B*15:238 C*04:01:01:01—C*04:107
A*23:01:01—A*23:50 B*18:01:01—B*18:68 C*05:01:01:01—C*05:72
A*24:02:01:01—A*24:190 B*27:01—B*27:86 C*06:02:01:01—C*06:69
A*25:01:01—A*25:16 B*35:01:01:01—B*35:186 C*07:01:01—C*-7:220
A*26:01:01—A*26:72 B*37:01:01—B*37:31 C*08:01:01—C*08:56
A*29:01:01:01—A*29:32 B*38:01:01—B*38:37 C*12:02:01—C*12:68
A*30:01:01—A*30:58 B*39:01:01:01—B*39:68 C*14:02:01-C*14:34
A*31:01:02—A*31:59 B*40:01:01—B*40:180 C*15:02:01—C*15:56
A*32:01:01—A*32:37 B*41:01—B*41:19 C*16:01:01—C*16:44
A*33:01:01—A*33:54 B*42:01:01—B*42:16 C*17:01:01:01—C*17:11
A*34:01:01—A*34:09 B*44:02:01:01—B*44:139 C*18:01—C*8:05
A*36:01—A*36:05 B*45:01—B*45:13
A*43:01 B*46:01:01—B*46:30
A*66:01—A*66:16 B*47:01:01:01—B*47:08
A*68:01:01:01—A*68:85 B*48:01:01—B*48:27
A*74:01—A*74:15 B*49:01:01—B*49:20
A*80:01—A*80:02 B*50:01:01—B*50:15
B*51:01:01—B*51:126
B*52:01:01:01—B*52:27
B*53:01:01—B*53:27
B*54:01:01—B*54:24
B*55:01:01—B*55:54
B*56:01:01—B*56:32
B*57:01:01—B*57:52
B*58:01:01—B*48:36
B*59:01:01:01—B*59:05
B*67:01:01—B*67:03
B*73:01—B*73:02
B*78:01:01—B*78:07
B*81:01—B*81:05
B*82:01—B*82:03
B*83:01
a
This summarizes HLA alleles as of January, 2012, as described by Robinson et al.8,9 and available at http://www.imgt.org or at www.ncbi.nlm.nih.gov/projects/gv/mhc/main
.fcgi?cmd-init. HLA-Bw4 alleles are generally agreed to include B5, B5102, B5103, B13, B17, B27, B37, B38(16), B44(12), B47, B49(21), B51(5), B52(5), B53, B57(17), B58(17), B59,
B63(15), B77(15), and A9, A23(9), A24(9), A2403, A25(10), A32(19) (http://hla.alleles.org/antigens/bw46.html). Similarly, HLA-Bw6 alleles include B7, B703, B8, B14, B18, B22, B2708,
B35, B39(16), B3901, B3902, B40, B4005, B41, B42, B45(12), B46, B48, B50(21), B54(22), B55(22), B56(22), B60(40), B61(40), B62(15), B64(14), B65(14), B67, B70, B71(70), B72(70), B73,
B75(15), B76(15), B78, B81, and B82.
HLA, human leukocyte antigen.
termed “antigen specific” in that a particular T cell identifies (PLC), which includes TAP, the chaperone tapasin,24 and
a particular peptide. For any given T-cell clone, single amino Erp57,25 are crucial to MHC-I loading. MHC-I molecules
acid substitutions of either the MHC or the peptide may se- are unique among proteins in that their three-dimensional
verely diminish, obliterate, or even augment the functional structure and thermal stability are exquisitely dependent on
interaction of the TCR with the MHC-II/peptide complex. the heavy/light chain heterodimer being bound by an ap-
The MHC-I system draws its spectrum of peptides from pro- propriate peptide.26,27 In mutant cells that lack the necessary
teins in the cytosol that are degraded by the proteolytic pro- apparatus for generating and transporting peptides to the
teasome complex to peptides that are transported from the ER where peptide loading takes place, MHC-I molecules are
cytosol to the endoplasmic reticulum (ER) with the aid of expressed poorly and are inherently unstable.28 MHC-II/
the intrinsic membrane peptide transporter, the transporter peptide loading is controlled in part by the multifunctional
associated with antigen processing (TAP), are then trimmed chaperone/groove protector, Ii, as well as the important
at their amino termini22 and are cooperatively folded as catalytic machinery of the endosomes, molecules known as
an intrinsic component of the newly synthesized MHC-I HLA-DM and -DO in the human, and H2-M and H2-O in
molecule.23 Interactions with the peptide-loading complex the mouse.24 Although there remains some controversy as

TABLE 21.3 Listing of HLA Class II Alelesa
HLA-DR Alleles HLA-DQ Alleles HLA-DP Alleles HLA-DM Alleles HLA-DO Alleles
α chain α chain α chain α chain α chain
DRA DQA1 DPA1 DMA DOA
DRA*01:01:01—DRA*01:02 DQA1*01:01:01—DQA1*01:09 DPA1*01:03:01:01—DPA1*01:10 DMA*01:01:01:01—DMA*01:04 DOA*01:01:01—DOA*01:04N
DQA1*02:01 DPA1*02:01:01—DPA1*02:04
DQA1*03:01:01—DQA1*03:03:02 DPA1*03:01—DPA1*03:03
DQA1*04:01:01—DQA1*04:04 DPA1*04:01
DQA1*05:01:01:01—DQA1*05:11
DQA1*06:01:01—DQA1*06:02
β chain β chain β chain β chain β chain
DRB1 DQB1 DPB1 DMB DOB
DRB1*01:01:01—DRB1*01:45 DQB1*05:01:01:01—DQB1*05:14 DPB1*01:01:01—DPB1:01:03 DMB*01:0101:01—DMB*01:07 DOB*01:01—DOB*01:05
DRB1*03:01:01:01—DRB1*03:77 DQB1*06:01:01—DQB1*06:47 DPB1*02:01:02—DPB1*02:02
DRB1*04:01:01—DRB1*04:107 DQB1*02:01:01—DQB1*02:06 DPB1*03:01:01—DPB1*03:01:02
DRB1*07:01:01:01—DRB1*07:22 DQB1*03:01:01:01—DQB1*03:39 DPB1*04:01:01:01—DPB1*04:02:01:02
DRB1*08:01:01—DRB1*08:49 DQB1*04:01:01—DQB1*04:08 DPB1*05:01:01—DPB1*05:01:02
CHAPTER 21
DRB1*09:01:02—DRB1*09:17 DPB1*06:01
DRB1*10:01:01—DRB1*10:04 DPB1*08:01
DRB1*11:01:01—DRB1*11:121 DPB1*09:01
DRB1*12:01:01—DRB1*12:35 DPB1*10:01
DRB1*13:01:01—DRB1*13:135 DPB1*11:01:01—DPB1*11:01:02
DRB1*14:01:01—DRB1*14:122 DPB1*13:01—DPB1*19:01
DRB1*15:01:01:02—DRB1*15:69 DPB1*20:01:01—DPB1*20:01:02
DRB1*16:01:01—DRB1*16:19 DPB1*20:01:01—DPB1*35:01:01
DRB2 DPB1*35:01:01—DPB1*35:01:02
DRB2*01:01
DRB3 DPB1*35:01:01—DPB1*41:01:01
DRB3*01:01:02:01—DRB3*03:03
DRB4 DPB1*41:01:01—DPB1*41:01:02
DRB4*01:01:01:01—DRB4*01:08
DRB5 DPB1*44:01—DPB1*136:01
DRB5*01:01:01—DRB5*02:05
DRB6
DRB6*01:01—DRB6*02:02
DRB7
DRB7*01:01:01—DRB7*01:01:02
DRB8
DRB8*01:01
DRB9
THE MAJOR HISTOCOMPATIBILITY COMPLEX AND ITS PROTEINS
DRB9*01:01
|
a
This table is based on a listing of alleles maintained by the World Health Organization Nomenclature Committee for Factors of the HLA System, as of January 2012. All new and confirmatory sequences are generally submitted directly to the
committee via the IMGT/HLA database using the sequence submission tool provided. The IMGT/HLA database may be accessed at www.ebi.ac.uk/imgt/hla. Serologic assignment of HLA-DR molecules is largely determined by the DRB1
gene product, whereas assignment of DQ molecules reflects serologic contributions from both DQA1 and DQB1 gene products.
491
HLA, human leukocyte antigen.
9/17/12 5:53 AM
TABLE 21.4 Commonly Used Mouse Strains: H-2 Haplotypesa

H-2 Complex
Strain Haplotype KK Ab Aa Eb Ea D Q T
Common strains
129/J bc b b b b – b b f
AKR/J k k k k k k k b b
ASW/Sn s s s s s – s b b
BALB/c d d d d d d d b c
C3H/HeJ k k k k k k k b b
CBA/J k k k k k k k b b
C57BL/6 b b b b b – b b b
C57BL/10 b b b b b – b b b
C57BR k k k k k k k a a
DBA/2J d d d d d d d b c
NOD/LtJ g7 d g7 d – – b
NON/LtJ nb1 b nb1 ? k k b
NZB/BINJ d2 d d d d d d a a
NZW/LacJ z u u u u u z
P/J p p p p p p p a e
PL/J u u u u u u d
RIII r r r r r r r c(r) b
SJL s2 s s s s – s a a
Congenic strains
B10.BR k2 k k k k k k
B10.D2 d d d d d d d
B10.S s s s s s – s
BALB.B b b b b b – b
BALB.K k k k k k k k
C3H.SW b b b b b – b
Recombinant strains
A a k k k k k k
A.TL tl s k k k k d
B10.A a k k k k k d
B10.A(1R) h1 k k k k k b
B10.A(2R) h2 k k k k k b
B10.A(3R) I3 b b b b/k k d
B10.A(4R) h4 k k k k/b – b
B10.A(5R) I5 b b b b/k k d
B10.T(6R) y2 q q q q – d
B10.S(7R) t2 s s s s – d
B10.S(8R) as1 k k k k/s – s
B10.S(9R) t4 s s s s/k k d
B10.HTT t3 s s s s/k k d
*Adapted from Kruisbeek561 and www.imgt.org/IMGTrepertoireMHC/Polymorphism/haplotypes/mouse/MHC/Mu_haplotypes.html#polymorphism. A dash indicates abnormal gene

expression, although precise mechanism may differ in different strain. Blanks indicate insufficient data for characterization. Additional strains may be identified at http://jax-
mice.jax.org/findmice/index.html.
NOD, nonobese diabetic.
to the precise site of interaction of HLA-DM with HLA-DR, in peptide selection and presentation. In general, they bind
evidence supports the view that this interaction catalyzes peptides derived from the degradation of proteins ingested
peptide interchange, leading to the selection of high-affinity by the APC, and they sort their MHC-II molecules into cel-
antigenic peptides.29 lular compartments where the degraded peptides are gen-
MHC-II molecules, in contrast to MHC-I, are expressed erated and catalytically transferred to the binding site of
on a more limited set of somatic cells—B cells, macro- the MHC-II. The MHC-II antigen-presentation pathway is
phages, dendritic cells, activated but not resting T cells in based on the initial assembly of the MHC-II αβ heterodi-
the human—and have a somewhat more specific function mer with a dual function molecule, the invariant chain (Ii),

which serves as both a chaperone to direct the αβ heterodi- This ability of the NK cell to sense altered levels of MHC-I
mer to an endosomal, acidic, protein processing location on target cells is the basis of the “missing self hypothesis,”
where it encounters antigenic peptides and also serves to which explains that NK cells detect and lyse those cells de-
protect the antigen-binding site of the MHC-II molecule so fective in MHC-I expression due to the loss of the inhibi-
that it preferentially will be loaded with antigenic peptides tory signal that results from engagement of NK receptors by
in this location.30–33 The loading of the MHC-II molecule MHC-I.45–47 The prototype NK receptor in the mouse is the
with antigenic peptide, a process dependent on the release NK inhibitory receptor, Ly49A, a C-type lectin-like molecule
of the Ii derived “CLIP” peptide, in part dependent on the that signals its interaction with a normally expressed MHC-I
MHC-II–like molecule, HLA-DM in the human,34,35 H2-M molecule, such as H2-Dd.48 Distinct clones of NK cells differ
in the mouse,36 then leads to the cell surface expression of in the combinatorial expression of different NK receptors
MHC-II/peptide complexes. The MHC-II–recognizing that have different MHC preferences. Thus, in the mouse,
T cells then secrete cytokines and may also be induced to each distinct NK clone may express a different combination
proliferate or to undergo programmed cell death. Such of NK inhibitory receptors such as Ly49A, Ly49C, Ly49G2,
MHC-restricted cytokine production that facilitates and and Ly49I.49 Because each inhibitory molecule may exhibit
augments the recruitment of additional inflammatory cells slight differences in its MHC-I and/or peptide preference
as well as APCs and antibody-producing cells is a contem- and specificity, this kind of combinatorial expression of NK
porary explanation for what was historically referred to as activity offers a breadth of specificity toward different poten-
“T-cell help.” Under some physiologic circumstances, partial target cells. Recently, it has been shown that developing
ticularly during viral infection, antigens incorporated into NK cells within the thymus are also subject to an education
antigen-presenting dendritic cells via an “outside in” path- process, known as “licensing,” which prevents improper ac-
way may alternatively enter the MHC-I presentation path- tivation of their cytolytic function in peripheral tissues.50–52
way. This phenomenon is known as “cross-presentation.”37,38 NK cells that develop in MHC-I–deficient humans and mice
MHC molecules perform a crucial role in the thymus by fail to kill target cells that lack MHC-I expression. Such tar-
shaping the TCR repertoire as T-cell precursors mature into gets are readily killed by NK cells that develop in MHC-I–
cells that eventually emigrate from the thymus and populate sufficient thymi. Human NK cells seem to go through a
the spleen and other secondary lymphoid organs. In a complex, similar selective/educating process, but the relevant NK re-
multistep process termed “thymic education,”39 developing T ceptors on human cells are generally of the killer cell immu-
cells expressing randomly rearranged TCR αβ receptors are noreceptor (KIR) inhibitory receptor family.53
first subjected to positive selection on MHC-I and MHC-II In addition to showing preference for distinct pathways
molecules expressed by cells of the thymic cortex in order of antigen presentation, the MHC-I and MHC-II molecules
to select for further maturation of only those cells capable of also show preferential restriction to T cells of the CD8- or
recognizing peptides in the context of self-MHC molecules. CD4-bearing lineages. This results from the interaction of
Positive selection may be viewed as setting the minimum CD8 with the nonpolymorphic α3 domain of MHC-I mol-
threshold for T-cell activation. T cells surviving positive selec- ecules,54–58 whereas CD4 binds to membrane proximal do-
tion then migrate to the thymic medulla where those T cells mains of MHC-II.59–64 The CD8 and CD4 molecules serve as
with higher-than-average affinity for self-MHC-II/peptide “coreceptors” on the surface of T lymphocytes, providing both
complexes, and which therefore pose the threat of autoim- adhesion (increase in avidity) and specific activating signals,
mune disease should they be allowed to emigrate to second- mediated through the kinase, lck, which modulate the avidity
ary lymphoid organs, are deleted by negative selection. MHC-I of the T cell in a time-dependent manner.65–68 Additional com-
and MHC-II molecules thus play a critical role in shaping the plexities arise from the interaction of CD8 with MHC-I. CD8
peripheral TCR repertoire. Indeed, mice engineered to lack is expressed as either the CD8αα homodimer or the CD8αβ
MHC-I or MHC-II expression are also deficient in T cells.36,40 heterodimer, which are expressed in a developmentally regu-
Distinct from the recognition of MHC molecules by lated fashion and seem to have distinct functions.69 The nu-
TCRs, a number of NK-cell receptors, both activating and merous interactions of MHC molecules with other cellular
inhibitory, bind MHC-I molecules, and several NK receptors components as well as with the wide variety of peptides and of
interact with MHC-I–like molecules encoded by CMVs.2,41 various immunologic receptors reflect the robust potential of
In general, the NK/MHC-I interaction, as compared to the the MHC structure as a molecular sensor and as a master regu-
TCR/MHC interaction, shows considerably less peptide lator of immune responses. These molecular interactions then
specificity, although the interaction is peptide dependent, read out in different cell trafficking and signaling functions.
and in some cases may exhibit clear-cut peptide prefer-
ences.42–44 The functional purpose of the MHC-I or MHC-I–
THE MAJOR HISTOCOMPATIBILITY COMPLEX
like molecule in NK-cell recognition appears to be more
subtle than that in T-cell recognition. The NK cell is tuned Mhc Genetics
to a balance of inhibitory and activating signals conveyed The Mhc is an extended region of the genome that spans
to it via MHC interaction. In its resting state, the inhibitory some 4 million basepairs (Mb) on the short arm of human
signals predominate. MHC-I is a sensor of the biosynthetic chromosome 6 in the region 6p21.3. The Mhc in the
and metabolic state of the cell in which it is synthesized. human may be considered to map from HLA-F to the gene
When the level of MHC-I is dysregulated by tumorigenesis encoding the MHC chaperone tapasin (TAP binding pro-
or viral infection, this change can be detected by the NK cell. tein), a distance of some 2 Mb, but the “extended” human

Mhc (also known as the xMHC) covers some 7.6 Mb from (Fig. 21.3) and refer the reader to the National Center for
the HIST1H2AA locus to SYNGAP1.70–72 In the mouse, Biotechnology Information homology Web site: www.ncbi.
the Mhc encompasses almost 3.5 Mb on chromosome nlm.nih.gov/projects/homology/maps/ and National Center
17, extending from the H2-Ke1 gene at basepair position for Biotechnology Information Web viewer for access to de-
34056423 to the H2-M2 gene at basepair position 37620474 tailed maps and sequences.
(www.imgt.org/IMGTrepertoireMHC/LocusGenes/index The human Mhc map reveals clusters of genes grouped
.php?repertoire=listIG_TR&species=mouse&group=MHC). roughly into an Mhc-II region covering about 1000 kb, an
Although Mhc genes were among the first to be mapped here, Mhc-III of about 1000 kb, and an Mhc class I region span-
it is now clear that a large number of genes with function ning 2000 kb (see Fig. 21.2). HLA-DP genes (DPA encoding
unrelated to immune recognition also reside in this region. the α chain and DPB encoding the β chain) are proximal to
The interested reader is referred to the regularly updated the centromere on the short arm of the chromosome and
maps and linkages available at various Web sites including are linked to the genes encoding the related HLA-DM mol-
www.ebi.ac.uk/imgt/hla, and the MHC haplotype project at ecule (DMB and DMA). Between these and the DQ genes lie
www.sanger.ac.uk/HGP/Chr6/. LMP (for low molecular weight proteins73–76) and TAP77–81
A schematic of the human Mhc is shown in Figure 21.2. genes that encode molecules that are involved in peptide
For a rough guide, we also provide a simple illustra- generation in the cytosol and peptide transport across the
tive map comparing human, mouse, and rat Mhc regions ER membrane, respectively. Low molecular weight proteins
FIG. 21.2. Map of the Human MHC (HLA Region) on the Short Arm of Chromosome 6. A variety of online tools for seeking and manipulating
MHC sequences can be found at www.ncbi.nlm.nih.gov/gv/mhc/main.cgi?cmd=init.

FIG. 21.3. Comparative Maps of Human, Mouse, and Rat Mhc Regions. This schematic map is not complete nor drawn to scale and is derived
from maps available at www.ncbi.nlm.nih.gov/projects/homology/maps/.
are subunits of the catalytic proteolytic proteasome complex Mhc-III region is important in immunologic terms for sev-
that regulate the specificity of cleavage of proteins, and thus eral reasons—the structural genes for several complement
modulate the repertoire of peptides available for MHC-I– components map here, as well as the structural genes for
restricted antigen presentation.82–85 The TAP genes encode 21-hydroxylase (CYP21A2),96,97 an enzyme critical in the bio-
a two-chain intrinsic membrane protein that resides in the synthesis of glucocorticoids, a deficiency of which can lead to
ER of all cells and functions as an adenosine triphosphate– the genetic disease congenital adrenal hyperplasia. Also lo-
dependent transporter that pumps peptides generated in the cated in the Mhc-III region are the structural genes for tumor
cytosol into the lumen of the ER.86,87 The selective transport necrosis factor (TNF) A (also known as lymphotoxin α) and
of cytoplasmically generated peptides by different TAP pro- B, which are cytokines made by activated T cells.98–100
teins in the rat demonstrates that the spectrum of MHC/ The more distal region of the Mhc encodes other MHC
peptide complexes expressed at the cell surface can be sig- molecules. In humans, the cluster of the major Mhc-I genes
nificantly altered by differences in the antigen presentation lies here, spanning two Mb, including the genes encoding
pathway,88–90 although there is little evidence for this phe- HLA-B, HLA-C, HLA-E, and HLA-A, as well as HLA-H,
nomenon in humans.91 Nevertheless, TAP deficiency syn- HLA-G, and HLA-F. HLA-A, HLA-B, and HLA-C are the
dromes in humans92–95 emphasize the importance of peptide major MHC-I molecules of man. (A summary of these is in
delivery in the antigen presentation pathway for normal im- Table 21.2.) Serologic identification of HLA-C molecules has
mune function. been difficult and imprecise. However, HLA-C molecules
The major Mhc-II genes of the human are HLA-DRA and interact directly with NK receptors of the KIR2D family.
HLA-DRB that encode the chains that form the HLA-DR Direct binding studies have analyzed the kinetics of the in-
molecule, a major antigen-presentation element. Genetic teraction of the KIR2 immunoglobulin (Ig)-like NK recep-
mapping of the human DRB region indicates that several tors,101–103 and three-dimensional structures of KIR2DL2
alternative arrangements of DRB loci explain the varied sein complex with HLA-Cw3 and KIR2DL1 complexed with
rotypes and genotypes observed among different individu- HLA-Cw4 have been published.104,105 Recently, the structure
als (see Fig. 21.2). In particular, note that several HLA-DR of a KIR3DL1 molecule in complex with HLA-B*57:01
“haplotypes,” originally defined serologically as DR52, DR1, emphasizes the unique nature of different KIR/MHC in-
DR51, DR53, and DR8, differ not only in particular allelic teractions.106 The precise functions of HLA-E, HLA-F, and
genes at the DR locus but also in the number and precise HLA-G are not yet clear. HLA-E and its murine analog Qa-1
location of several of the DRB genes and pseudogenes. The bind hydrophobic leader peptides derived from some MHC-I

molecules, forming a complex recognized by the C-type and evolve, we expect that genetic variants should arise but
lectin-like NK receptor CD94/NKG2.107–111 This implies an because of the selection exerted on most gene products, rel-
important function for HLA-E because these molecules are atively few of these genetic variants will persist. A genetic
expressed on placental trophoblast cells and would be ex- locus that exhibits variant alleles at a frequency of more
pected to bind the inhibitory NK receptor, CD94/NKG2A, than 1% is considered “polymorphic.”5 A genetic locus that
preventing NK-mediated rejection of the fetus.112 HLA-E may is relatively invariant is often referred to as “monomorphic,”
serve as a recognition element for some T cells as well, so it even if more than one allele is known. HLA genes exhibit a
also seems capable of a classical role.113,114 Additional evidence high degree of polymorphism; a number of different mecha-
now supports an antigen presentation function of HLA-F and nisms may contribute to the generation and maintenance of
HLA-G,115,116 and the tissue restricted expression of HLA-G as polymorphism. Among these are the selective advantage of a
well as the observation that a soluble form leads to apoptosis heterozygous pool of antigen-presenting elements in a given
of CD8 + T cells suggest that this molecule may be involved individual that might allow the binding and presentation
in the mother’s immunologic tolerance of the fetus.117 HLA- of antigenic peptides derived from a wide variety of envi-
H is a pseudogene mapping to this region.118 This should not ronmental pathogens. Limited polymorphism would make
be confused with the more distantly related HLA-HFE, an the entire population susceptible to a chance infectious
Mhc-Ib gene erroneously called HLA-H by some authors,118,119 agent for which all individuals would be unable to respond,
which controls hereditary hemochromatosis by virtue of the whereas widespread polymorphism would be expected to
interaction of its encoded protein with the transferrin recep- allow the APCs of at least a proportion of the population to
tor.120–125 The observations that extravillous trophoblast cells effectively bind and present antigens derived from invading
express an unusual combination of HLA-C, HLA-E, and pathogens.140 Although such a view was originally based on
HLA-G molecules; that uterine NK cells express NK inhibi- HLA molecules as presenting elements for pathogen-derived
tory receptor (KIR) molecules known to interact with HLA-C peptide fragments to T cells and their antigen-specific
allelic products; and that complex genotypes predispose to TCRs, studies suggest an additional role for Mhc-I–related
preeclampsia raise the possibility that important aspects of resistance to viral infection via NK cell–mediated recog-
fetal rejection and clinical infertility may be related to MHC nition141,142 and altered antigen presentation pathway.143
recognition by NK cells in the placenta.126 Additionally, the presence of a high degree of polymorphism
Comparison of the mouse, rat, and human Mhc maps also implies that there is little or no selective disadvantage to
(Fig. 21.3) reveals several interesting differences.5,127–129 The the expression of new variant MHC molecules.
Mhc genes proximal to the centromere of the mouse and rat The human Mhc-I and Mhc-II genes are clearly polymor-
belong to the Mhc-I family, rather than to the Mhc-II fam- phic, with more than 5000 alleles at the Mhc-I loci and some
ily, as they do in the human. This mapping has suggested 2000 alleles at the Mhc-II loci known (see Tables 21.2 and 21.3)
that an intrachromosomal recombination event that oc- (www.ebi.ac.uk/imgt/hla/index.html).8,9 In experimental an-
curred in some common rodent ancestor relocated some of imals, it is more difficult to demonstrate polymorphism in
the Mhc-I genes from a more distal location to the proxi- terms of population genetics, although typing of wild mice has
mal site.130 Inspection of the current human, mouse, and rat confirmed the existence of natural polymorphism predicted
maps clearly indicates similarities in the relative locations from the analysis of inbred strains and their mutants.144,145
and organization of Mhc-II, Mhc-III, and the distal Mhc-I The polymorphism of Mhc-I and Mhc-II genes, so evident
genes.128 Various genetic expansions and contractions131 are in man and mouse, has also been documented in analyses of
obvious as well. In particular, the mouse Q and T regions cichlid fishes—animals that diverged from the line leading to
have expanded the pool of Mhc-I genes, which are rela- mammals at least several hundred million years ago.146–149
tively few in the human and the rat. Early studies of con-
genic mouse strains mapped multiple genes to the Q and T
Genetic Mechanisms for Mhc Evolution
regions,132–134 and recent evidence suggests significant dif-
ferences in the number of genes of this region in different As both an extended genetic region and a group of genes with
strains. The mouse has some MHC-Ib genes that seem to be many belonging to the Ig supergene family,150,151 the Mhc has
relatively unique in function. In particular, the H2-M3 gene, served as a prototype for elucidating mechanisms that con-
which maps distal to the Q and T regions, encodes a pro- tribute to the evolution of a multigene family and that add
tein that exhibits a preference for binding peptides that have to the polymorphism that is such a dominant characteristic
N-formyl amino terminal modifications. This antigen pre- of the classical MHC molecules.152 The analysis of mutations
sentation function may be geared to bacterial, protozoal, and in the mouse, mostly those of Mhc-I genes, has led to the
mitochondrial antigens.135–137 Rat homologues of the mouse understanding of the mechanisms that give rise to polymor-
H2-M3 and H2-M2 genes have also been identified.138,139 phism. Both induced and spontaneous mutations affecting
skin graft acceptance or rejection have been identified, and
many of these have been mapped to the Mhc. Gross recom-
Mhc Polymorphism binational events have been documented in the Mhc,153,154 as
The Mhc’s function in immune responsiveness is also re- well as more subtle mutations, many of which are multiple
flected in its genetic polymorphism. Polymorphism is the amino acid substitutions in a relatively small part of the pro-
presence at any given time of a larger than expected number tein that seem to derive from nonreciprocal recombinational
of genetic variants in a population. As populations change events. Such recombination that occurs over short sequences

is known as “gene conversion” because of its similarity to the mouse, these were originally identified as genetic loci re-
phenomenon that occurs in yeast.153–161 Convincing evidence sponsible for graft rejection after extended periods of time.
for interlocus gene conversion has been documented in for More recently, several minor histocompatibility loci have
the Mhc-II genes of teleost fish162 and by deep sequencing of been identified as those that encode polymorphic proteins
Mhc-I genes in recently founded bird populations.163 that give rise to peptides presented by MHC molecules,169–173
Some of the polymorphisms of Mhc genes that have been and we now can understand the complexities of transplan-
identified clearly reflect point mutations.164 Structural stud- tation tolerance in terms not only of Mhc genes but also
ies have shown that the profound immunologic effects of in terms of numerous proteins that may give rise to vari-
mutations of the H2 genes H2Kbm1 and H2Kbm8 result from ant peptides for T-cell recognition. Not only can peptides
minimal detectable changes that may affect thermosta- derived from polymorphic genes throughout the genome
bility.165 In addition to such mouse mutants, a number of serve as minor histocompatibility antigens but also defective
somatic cell variants and mutants, some due to major dele- translation products, or peptides resulting from transcrip-
tions or regulatory defects, others clearly point mutants of tion of introns or noncoding strands of DNA may also pro-
structural genes, have been described.166 duce immunologically significant peptides, which, bound
by self-MHC molecules may stimulate T cells.174,175 Whether
such defective, newly synthesized proteins serve as a major
The Mhc and Transplantation source of MHC-I–bound peptides remains controversial.176
Although the early description of the genes of the Mhc was Minor antigens that are confined to hematopoietic cells can
based on identification of loci involved in tumor and allograft be recognized as targets by antitumor cytolytic cells and
rejection, and although these genes clearly play a role in such may explain the lower incidence of relapse in hematopoietic
complex phenomena, a contemporary understanding of the stem cell (HSC) transplant recipients who experience graft-
function of Mhc genes in immunology requires little under- versus-host disease (GVHD).177
standing of the rules of transplantation. The early history of
transplantation is chronicled extensively in several books5,167
and reflects a developing interest in tumor immunology The Mhc and Clinical Transplantation
and congenic mouse strains. The mouse has served as the Processed foreign antigen complexed to HLA class I or class
model system for understanding the genetic underpinnings II molecules is recognized by a specific clonally distributed
of skin, tumor, and organ transplantation, so a brief descrip- TCR for antigen on the surface of T lymphocytes. The T cell
tion of some relevant principles is in order. Comprehensive bearing an αβ receptor is capable of recognizing the unitary
manuals and reviews are available.168 Propagation of a mouse structure of the HLA molecule itself coordinately with the
strain by repeated matings of brothers and sisters leads to the exposed parts of the peptide antigen. Corecognition of HLA
establishment of an inbred strain, a group of animals that and peptide antigen means that TCRs are highly specific
is genetically identical at all loci. More complete descrip- and genetically restricted to recognizing HLA molecules of
tions of the process by which brother–sister mating leads to the individual from which they were derived. Thus, a killer
homozygosity at all loci are given elsewhere.5,167 (cytotoxic) T cell raised against an influenza virus peptide
“Congenic” mouse strains, also known as “congenic rein an individual expressing HLA-A2 will not recognize in-
sistant” or “CR” strains, are those derived by first crossing fluenza-infected HLA-A1. This concept is known as “MHC
two inbred strains that differ in a histocompatibility pheno- restriction” and was described by Shevach and Rosenthal
type such as resistance to a transplantable tumor or ability for recognition of amino acid polymers,178 by Zinkernagel
to reject a skin graft. These are then successively backcrossed and Doherty for recognition of viral antigens,179,180 and by
to one parental strain, and the resistance phenotype is pre- Shearer et al. for recognition of altered self-ligands.181 Given
served. Following at least 10 backcross generations (N10), that T cells are MHC-restricted, it is difficult to under-
a point at which (1/2) 9 = 0.002 of the genes of the selected stand why they should ever recognize a foreign HLA type.
strain should be present, the new strain is propagated by However, in practice they do, data indicate that such allo-
brother–sister mating. Several relevant inbred and congenic reactive T cells arise with remarkably high frequency. It is
mouse strains are listed in Table 21.4 along with their H2 estimated that between 1/10 to 1/1000 activated clonally dis-
designations. tinct T cells are capable of responding to any random alloge-
The early rules of transplantation were determined by neic HLA molecule.182–185 Given the number of T cells in the
observation of the ability of either transplantable tumors or human lymphoid system, this represents a striking tendency
allografts (usually from skin) to survive in a particular infor T cells that are normally restricted to recognizing self-
bred mouse strain host. The graft rejection phenomenon is HLA molecules complexed to foreign peptides to cross-react
an extremely sensitive and specific bioassay that permits the on allogeneic HLA molecules. This cross-reaction can arise
detection of genetic differences as small as a single amino from direct recognition of the allogeneic HLA/peptide com-
acid in an MHC protein. It has been particularly valuable plex, which usually depends on the peptide antigen as well
in assessing spontaneous and induced mutants (see previous as the allogeneic HLA molecule. Alternatively, allorecog-
discussion) and remains the absolute experimental discrim- nition by T cells can occur indirectly.186,187 In such cases,
inator of “histocompatibility.” peptides derived from the allogeneic HLA molecules are
In addition to the Mhc genes, we should not overlook the presented as nominal antigen after processing by the host
genes that encode minor histocompatibility antigens. In the cells bearing self-HLA molecules. In the normal course of

events, T-cell alloreactivity is an in vitro curiosity, although

it is still not entirely clear why the fetal “allograft” does not MHC Matching versus Success of
TABLE 21.5
stimulate the maternal immune system. However, it is the Bone Marrow Transplantation
clinical transplantation of organs and hematopoietic stem
Risk of
cells across HLA compatibility barriers that produces graft Risk of Graft Acute Graft- Survival
rejection or GVHD due to T-cell alloreactivity. Fully alloge- Rejection versus-Host (3 Years)
neic transplants theoretically expose the recipient immune MHC Compatibility (%) Disease (%) (%)
system to up to 12 non–self-HLA allele products expressed
Share two haplotypes 2 40 50
by the allograft. Moreover, the “self-peptides” constitutively (HLA identical
presented by allogeneic HLA molecules are likely to be quite sibling)
distinct from those presented by syngeneic HLA molecules Share one haplotype 7 40 50
because the polymorphisms of the peptide antigen-binding plus phenotypically
cleft of the MHC-I molecule that distinguish HLA alleles identical
alter the spectrum of selected peptides. In the presence of 1 HLA mismatch 9 70 50
appropriate costimuli, allogeneic HLA/peptide complexes >2 HLA mismatches 21 80 15
probably stimulate powerful T-cell responses because of the Share zero haplotypes
high density of unusual determinants and the diversity of (unrelated)
new peptide ligands presented by the allogeneic HLA/pep- “Matched” 3 80 35
tide complexes. Because there are many MHC-linked genes “Mismatched” 5 95 35
encoding a host of proteins, many lacking known immu- HLA, human leukocyte antigen; MHC, major histocompatibility complex.
nologic function, it is likely that polymorphisms in these Reprinted from Christiansen FT, Witt CS. Allogeneic bone marrow transplantation.
In: Bradley J, McCluskey J, eds. Clinical Immunology. Oxford: Oxford University
molecules contribute significantly to the alloresponse (see Press; 1997:445, with permission.
previous discussion of Mhc genetic maps and Mhc alleles).
Accordingly, many studies have demonstrated an incremen-
tal improvement in long-term graft survival with progres- Transplant Research at www.cibmtr.org/pages/index.aspx).
sively higher levels of HLA matching at HLA-A, -B, and -DR The effect of HLA matching on solid-organ transplantation
loci. For this reason, HLA matching is essential in allogeneic has been monitored by the Collaborative Transplant Study
HSC transplantation and highly desirable in solid-organ since 1982 (www.ctstransplant.org/).
transplantation. The degree of HLA matching usually re- In addition to the allogeneic cellular response, the anti-
quired for renal transplantation is shown in Figure 21.4 and body response to HLA molecules and ABO blood groups can
for bone marrow transplantation in Table 21.5. Survival of also cause rejection of certain grafts, especially where these
HSC transplants varies according to the nature of the dis- antibodies are preformed and, therefore, present at the time
ease, disease stage, and age of the patient but can be >70% in of organ transplantation. Antibodies to ABO blood-group
some cases (see Centre for International Blood and Marrow antigens react with these determinants on vascular endo-
thelium, and, therefore, ABO-incompatible solid organs can
be rapidly rejected by humoral mechanisms. In patients who
have been transfused or previously transplanted, or in mul-
tiparous females, exposure to allogeneic HLA molecules can
also result in the production of anti-HLA class I antibodies.
These preformed antibodies can lead to acute and hyperacute
rejection of grafts expressing the particular HLA molecules
recognized by these antibodies. Renal graft survival improves
with fewer HLA mismatches. In Figure 21.4, cumulative data
for graft survival are plotted as a function of time. Curves
represent those groups with the indicated number of mis-
matches. Therefore, for solid-organ transplants, individuals
are not only matched as closely as possible for their HLA types
to avert cellular rejection but, it is also necessary to ensure
ABO compatibility and to exclude preformed antidonor HLA
antibodies in the host. Paradoxically, some patients who have
received multiple blood transfusions prior to transplantation
appear to develop some form of T-cell tolerance to allogeneic
donor HLA alleles; renal graft survival is actually enhanced
in these individuals. This is known as the “transfusion ef-
FIG. 21.4. Effect of HLA Matching on First Deceased Donor Kidnery
Graft Survival. Survival of first kidney transplants based on number
fect,” and in some centers, pretransplant transfusion and
of HLA-A, -B, and -C mismatches as a function of time. Courtesy of even donor-specific transfusions are routinely carried out.
Collaborative Transplant Study, G. Opelz, University of Heidelberg Transfusion of potential renal transplant recipients car-
(www.ctstransplant.org/public/graphics.shtml). ries the risk, however, of inducing undesirable anti-HLA

antibodies in the patient. Testing for anti-HLA antibodies of siblings or other family members. Because unrecognized
is known as the “crossmatch.” In practice, many laboratories HLA mismatching is poorly tolerated in HSC transplanta-
crossmatch only for anti-HLA class I antibodies. Crossmatch tion, high-resolution sequence-based DNA matching (or its
compatibility to exclude anti-HLA class I antibodies is essen- equivalent) is required to avoid GVHD.190–192 The impact of
tial in renal transplantation and is widely practiced in heart/ HLA on the outcome of HSC transplants has been studied
lung transplantation. Crossmatching for liver transplanta- in unrelated donor–recipient pairs by several groups,193,194
tion is practiced at only some centers, and the evidence that and guidelines for matching have been published by the
a positive crossmatch predicts allogeneic liver graft rejection U.S. National Marrow Donor Panel.195 These data show
has not convinced everyone of its importance in routine that unrelated matched transplants can be as effective as
matching. Patients awaiting renal transplantation are usu- sibling matched donor–recipient transplants in donor sur-
ally monitored regularly for anti-HLA class I antibodies be- vival. Mismatching beyond a single, or at most two, HLA
cause the level and specificity of these antibodies can change loci is associated with increased GVHD and inversely with
with time. This monitoring involves regular crossmatching a lower rate of leukemic relapse. Although allogeneic HSC
of patient serum against a panel of randomly selected cells transplantation can be carried out across single-locus mis-
bearing different HLA types. The percentage of positively matches, there is little correlation with the magnitude of
reacting cells is known as the “panel reactivity.” When car- a given mismatch (ie, number of polymorphisms between
rying out a crossmatch between a patient’s serum and donor alleles) and the subsequent immune response. Donor–
cells, many centers test the current as well as “historical recipient MHC differences of just a single amino acid can
peak” serum from the patient. The historical peak is defi ned provoke strong alloresponses comparable to reactions be-
as the patient serum sample giving the highest panel reactiv- tween vastly different MHC products.191 An example of hap-
ity throughout the monitoring period and is thought to be a lotyping in a family study is shown in Figure 21.5. Matching
reflection of previous HLA sensitization. Most centers now for HLA-DP in allogeneic stem cell transplantation appears
prescreen patients on transplant waiting lists for anti-HLA to improve graft success in both stem cell and renal trans-
antibodies using MHC-coated beads as a source of antigen plantation,196,197 but testing for this locus is not routinely
and a highly sensitive flow cytometry technology for their carried out clinically for renal transplants and only for HSC
detection. In highly sensitized patients with multiple host when several donors are available and a rational choice of
antidonor antibodies, these unwanted antibodies can some- the best donor has to be made.198,199 HLA-C matching is
times be functionally eliminated using B-cell ablation ther- also important for chronic myeloid leukemia.200 Typically,
apies (anti-CD20 monoclonal antibodies), plasmapharesis, an HLA typing laboratory will test for HLA-A, -B, -DRB1,
and intravenous gammaglobulin infusions, thus allowing -DRB3, -DRB4, -DRB5, and -DQ loci for HSC transplant
successful transplantation. Similar approaches, coupled with matching.198 Many centers will also insist on HLA-C test-
plasmapharesis, are now taken with increasing frequency in ing. In the family study shown in Figure 21.5, the mother
transplanting kidneys across the ABO barrier. The role of and father are mismatched at both haplotypes. Among the
antibody crossmatching in allogeneic HSC transplantation children, John and Andrew are haploidentical (and there-
is less clear, and many centers do not take the class I or class fore phenotypically identical). Jane and Jim share a single
II crossmatch into account when identifying a bone mar- haplotype, as do Tom and Jim. Jane’s paternal haplotype is
row transplant donor. On the other hand, some large cen- a recombinant involving a crossover event between HLA-A
ters place considerable importance on a positive crossmatch and HLA-B. Recombination is observed between HLA-A/B
as a predictor of bone marrow rejection, and it is therefore and HLA-B and HLA-DR in about 1% of meiotic events.
advisable to crossmatch bone marrow donor–recipient pairs The implications of this family study are that Andrew and
when there is a high risk of rejection (eg, aplastic anemia). John would be ideal bone marrow donors for each other.
Crossmatching is also used to detect anti-HLA antibodies However, none of the other siblings would be suitable as a
that may cause refractoriness to platelet transfusion with donor for these brothers. Even though there is sharing of a
random platelets. single haplotype between Tom and both Andrew and John,
the complete mismatch in the second haplotype would make
Tom unsuitable as a donor for HSC transplantation, which
Family Studies in Histocompatibility Testing requires very close matching of HLA. On the other hand,
The linkage of HLA loci on chromosome 6 means that in- haplotype mismatching is common in renal transplantation,
dividuals will usually inherit a set of nonrecombined HLA in which perfect HLA matching is not absolutely required
alleles encoded at linked HLA loci from each parent. This set or routinely achievable. However, for renal and other solid
of genes (the haplotype) is often identifiable in family stud- organ transplantation, ABO blood group compatibility is
ies, where all the alleles present on one chromosome cosegre- usually essential because these determinants are expressed
gate. In identifying donors for HSC transplantation, testing on vascular endothelium where recognition by isohemag-
of family members is essential to determine haplotypes ac- glutinins leads to rapid intravascular coagulation and organ
curately.188,189 This is because sharing of HLA antigens from failure. When a matched sibling donor does not exist for a
different haplotypes is quite common in families, so mis- patient requiring allogeneic HSC transplantation (70% of
matches within HLA subtypes (ie, allele-level mismatches) cases), searching of the extended family or unrelated bone
are easily overlooked as a result of mistaken haploidentity marrow registries is indicated. The National Marrow Donor

FIG. 21.5. Segregation of Haplotypes in a Family. From McCluskey,558 with permission.
Panel (www.marrow.org) has several million potential do- be beneficial in adult acute myeloid leukemia and childhood
nors suitable for unrelated stem cell transplantation; these acute lymphoblastic leukemia leading to the development of
donors are used in the United States and worldwide. Bone adoptive NK-cell transfers as an experimental therapy for
marrow donor registries also exist in Europe, Australia, graft-versus-leukemia treatment.205 However, these trans-
Hong Kong, and Japan; these registries often provide donors plants must be rigorously T-cell depleted, and variable re-
for patients in other parts of the world. Mobilization of stem sults appear to reflect different transplant protocols, disease
cells in the blood following administration of hematopoietic parameters, and KIR-mismatching assignment.203,204,206
growth factors is now widely used to avoid the need for mar- Cord blood banks have also been established around the
row collections from donors. world. However, cord blood donation of stem cells is often
In the last decade, many HSC transplant protocols have unsuitable for adult transplantation because of limitations in
sought to reduce the pretransplant conditioning of patients. the volume of cord blood collections providing inadequate
There have also been controversial reports indicating dra- numbers of stem cells. Many centers will combine different
matic outcomes of HSC transplantation where deliberate cord bloods that are matched with the recipient as much as
KIR-ligand mismatching leads to donor-versus-recipient NK- feasible to increase the stem cell number. Cord bloods offer
cell alloreactivity toward cells of hematopoietic origin.201–204 the advantage of finding donors faster than adult unrelated
The Velardi group in Perugia has championed this approach registries197 and theoretically providing banked stem cells
providing evidence that in the absence of GVHD, a high fre- from ethnic minority groups that are not well represented
quency of donor NK cells in KIR-ligand mismatched trans- in bone marrow donor registries.198 Cord blood transplants
plants can remove recipient’s target cells.202 In particular, induce less GVHD than bone marrow or peripheral blood
HLA haplotype–mismatched hematopoietic transplants can stem cell transplants, but posttransplant engraftment is

slower.196,207 Cord transplants may also tolerate greater levels they might give a broad measure of the transplant barrier,
of HLA mismatching than unrelated bone marrow trans- but technical improvements will be required before this
plant, but this is dependent on the transfer of sufficient test is widely adopted in clinical practice. 215
CD34 + stem cells.208 The results of transplants using bone
marrow or peripheral blood stem cells are now considered
The Mhc and Disease
comparable clinically,207 but the impact of HLA matching in
patients transplanted at early stage in their disease appears In addition to the control of transplant acceptance or rejec-
to be more marked. Functional tests of HLA compatibility tion and immune responsiveness, the Mhc in the human
testing for HLA identity at all HLA loci is a daunting task plays an important role in the etiology of a number of dis-
for most laboratories because of the very large number of eases, many of which are autoimmune in nature.123,216–218
alleles present in the population. Moreover, in renal trans- Several human diseases are associated with the Mhc-III
plantation, some mismatches appear to be well tolerated and genes because some of the structural genes for enzymes
are associated with long-term graft survival, whereas other involved in the adrenal steroid biosynthetic pathway
mismatches of similar genetic disparity are poorly tolerated (ie, 21-hydroxylase, CYP21A2) map to this region. As many
and are associated with early rejection.209,210 Reliable meth- as 200 diseases have well-established genetic linkages to the
ods for predicting these “taboo” mismatches are not read- Mhc,217,219–222 the most important of which are summarized
ily available. Similarly, high-resolution HLA typing does in Tables 21.6 and 21.7. Recent genome-wide association
not predict all GVHD when selecting suitable unrelated studies confirm the importance of the HLA-DRB1 locus in
donors for HSC transplantation.190,211 Therefore, there has rheumatoid arthritis and type 1 diabetes,71,223–225 and fi ne
been a great deal of interest in developing functional or in mapping analysis suggests that a total of five amino acid
vitro cellular tests of overall donor–recipient compatibility. polymorphisms (three in HLA-DRB1, one in HLA-B, and
Unfortunately, none of the tests so far developed provides one in HLA-DPB1), all located in peptide-binding grooves,
convincing predictability of impending graft rejection or, almost completely explain the MHC association with the
more importantly, GVHD. risk of rhuematoid arthritis.226 The precise mechanisms
Among the tests used historically for assessing func- underlying the association of most of these diseases with
tional compatibility are the mixed lymphocyte reac- the particular Mhc haplotypes are unknown, but several
tivity (MLR) test and allogeneic T-helper or cytotoxic models have been proposed, including the cross-reactivity
T-lymphocyte precursor studies. The MLR or MLC (C = of antimicrobial antibodies and the molecular mimicry of
culture) involves measuring T-cell proliferation of host
T cells in response to donor lymphocytes and vice versa.
In a one-way MLR, the stimulating lymphocytes are ir-
radiated to prevent their proliferation; in the two-way
MLR, both stimulator and responder cells are allowed
TABLE 21.6 Some HLA Disease Associations
to proliferate. In MLR studies, it is necessary to include
controls showing that responder cells can all respond and Strength of
that stimulator cells can all stimulate across an appropri- Disease MHC-I Association
ate barrier such as third-party donor cells. Responses can
Ankylosing spondylitis HLA-B27 +++
vary widely, and individual laboratories use their own
Reiter disease HLA-B27 ++
cutoff values to define negative (ie, nonreactive) and pos-
Psoriasis HLA-C*06 +
itive (ie, reactive) MLR results. 211 Unfortunately, known
Abacavir drug HLA-B*57:01 +++
HLA mismatches can be present in a negative MLR, and hypersensitivity
a positive MLR can be obtained between phenotypi- Behcet disease HLA-B*51 +
cally HLA-identical transplant pairs. Because the MLR Birdshot retinopathy HLA-A*29 +++
is biased toward measuring HLA class II discrepancies
and is notoriously irreproducible, 211 most laboratories MHC-II
have abandoned this test in favor of implementation of Narcolepsy HLA-DQB1*06:02 ++
high-resolution polymerase chain reaction–based HLA Insulin-dependent HLA-DQ8 ++
class II typing. Measurement of allogeneic cytotoxic diabetes mellitus
T-lymphocyte or helper T–cell precursor frequencies is HLA-DQ2 +
carried out at specialized bone marrow transplant cen- HLA-DR2 −
ters but is not universally accepted as being predictive Rheumatoid arthritis HLA-DR4 +
of GVHD. 212,213 The test is labor intensive and requires a Celiac disease HLA-DQ2 +++
skilled technician for reproducibility. Precursor frequen- HLA-DQ8 +
cies are estimated by limiting dilution analysis or ELISpot Multiple sclerosis HLA-DR2 +
of donor-versus-host lymphocytes (ie, T cells expected to
HLA, human leukocyte antigen; MHC, major histocompatibility complex
cause GVHD). High precursor frequencies (up to 1 in 10 4 +++ = very strong association that is clinically useful as a diagnostic tool
cells) are thought to be associated with a greater risk for ++ = strong association with likely primary involvement in disease pathogenesis
+ = clear association with likely role in disease pathogenesis
acute GVHD. 214 It is possible that precursor studies detect − = negative or protective influence on disease probability
major and minor incompatibilities, and so, theoretically, For a more detailed summary of MHC and disease associations see Shiina et al.220

Hereditary hemochromatosis is one of the most common

HLA Disease Associations and genetic disorders in Caucasian populations (with a preva-
TABLE 21.7
Relative Riska lence of 1/300 to 1/400). The gene controlling this condi-
tion (HFE) is Mhc linked, mapping approximately three
Estimated or
Disease Antigen Relative Risk Mb telomeric to the HLA-A locus.122 The HFE protein is a
class I–like molecule, the structure of which has been de-
Graves disease or B*08:01 4 or 2.5 termined.237,238 HFE assembles with β2m and is expressed
myasthenia gravis DRB1*03:01 in the intestinal mucosa and placenta, where it plays a role
DQA1*05:01
DQB1*02:01 in regulating iron uptake and transport.239,240 Mice that are
Multiple sclerosis DRB1*15:01 4 homozygous for an induced defect of β2m as well as those
DQB1*06:02 with targeted inactivation of the HFE gene suffer from iron
Multiple sclerosis DQA1*01:02 4 overload or hemochromatosis.241–244 Mutations at loci other
Psoriasis C*06:02 5 than β2m or HFE also lead to the same disease phenotype.245
Celiac disease DQA1*02:01/DQB* 1 HFE regulates the affinity of the receptor for transferrin, re-
02:02 (DQ2.2) sulting in an alteration of the efficiency of iron transport.
DQA1*05:01/DQB1* 7 The most common molecular defect associated with heredi-
02:01 (DQ2.5) tary hemochromatosis involves a point mutation that results
SLE DRB1*15:01 2 in a Cys282Tyr substitution in the α3 domain of this class
Type 1 diabetes or SLE DRB1*03:0101 4.5 I–like molecule.122 This mutation accounts for > 80% of pa-
a
These examples were taken from single nucleotide polymorphism associations re-
tients with hereditary hemochromatosis.246 The disruption
ported by de Bakker et al.,71 where the tag single nucleotide polymorphisms and of the disulfide bond in the α3 domain at this site prevents
coefficients of determination (r 2) are also reported. For celiac disease, DQ2.2 only efficient folding of the molecule and impairs assembly with
has an effect when in combination with DQ2.5. The relative risk of DQ2.5 changes
depending on the coordinate presence of other DQ types. β2m, resulting in improper HFE protein expression. This
SLE, systemic lupus erythematosus. leads to a failure to downregulate the affi nity of the transfer-
rin receptor for its ligand, transferrin, presumably causing
increased iron uptake by cells and tissue damage as a result
viral antigens that might induce T-cell responses to self- of iron overload. A second common HFE mutation, 187G,
antigens.227–229 The very high incidence of some diseases as- results in a His63Asp substitution and a very slight increase
sociated with certain HLA genes assists in the diagnosis as in susceptibility to developing hereditary hemochromatosis,
well as the counseling of patients and their families. Several depending on the genotype of the individual.247 Incomplete
of these diseases are of particular note. Because virtually penetrance of even the high-risk Cys282Tyr HFE genotypes
100% of patients with narcolepsy have HLA-DQB1*06:02 can be partly explained by natural iron deficiency from lim-
(associated with HLA-DR2),230,231 HLA typing can be used ited dietary intake and menstrual losses in women.
as a test of disease exclusion. Thus, a diagnosis of narcolepsy As summarized in Tables 21.6 and 21.7 and discussed
can be excluded with reasonable certainty if the patient previously, there are a number of autoimmune diseases
does not have HLA-DQB1*06:02. On the other hand, the associated with particular alleles of HLA class II loci,
presence of HLA-DQB1*06:02 is of little predictive value in especially with DR and DQ.221 These diseases include type 1
diagnosis of narcolepsy because this HLA type is relatively (insulin-dependent) diabetes mellitus, rheumatoid ar-
common in many populations and occurs frequently in the thritis, multiple sclerosis, systemic lupus erythematosus,
absence of disease. thyrogastric autoimmunity, Sjögren syndrome, and many
Ankylosing spondylitis is so strongly associated with the others. Rheumatoid arthritis is strongly associated with
Mhc-I allele HLA-B27 and the presence of some bacterial HLA-DR4 subtypes that share a common sequence motif
pathogens that it is a popular hypothesis that ankylosing within the DRβ chain,248 suggesting antigen presentation
spondylitis is due to the stimulation of particular T cells by of self-epitopes by these molecules. The relative risk of se-
HLA-B27–presented bacterial antigens that cross-react on vere rheumatoid arthritis is increased in DR4 homozygotes
self-tissues. These T cells are then thought to initiate an in- and particularly in compound heterozygotes with high-risk
flammatory cascade. Despite the strong association of HLA alleles,249,250 indicating a gene dose effect in susceptibility to
with spondyloarthropathy, critical evaluation of the litera- autoimmune inflammation.
ture brings a postinfectious etiology into question and cer- Although the number of different HLA class I and class II
tainly more studies are indicated.232,233 Recent genome-wide alleles that are associated with type 1 diabetes clearly indi-
studies strongly indicate that a genetic interaction between cates that this relatively common disease has a complex eti-
an enzyme involved in processing MHC-I–associated pep- ology, the identification of a novel Mhc-II haplotype in the
tides, ERAP1, and HLA-B27 contributes to disease suscepti- mutant nonobese diabetic mouse,251–255 and the recognition
bility.234 The tendency of HLA-B27 to form disulfide-linked that particular TCRs can mediate disease,256 suggest that a
covalent dimers raises the question whether the resulting cross-reactive response to a common self- or environmental
cellular pathology related to poor cell surface expression of antigen may play an important role in the etiology of this
this MHC-I molecule may be related to the inflammation of disease as well. In human type 1 diabetes, the incidence of
joints that is characteristic of this disease.235,236 disease is significantly increased in Caucasians with HLA-

DR3-DQ2 and DR4-DQ8 haplotypes.257–259 These haplotypes bonds that bind a dominant deamidated gliadin determi-
impart a synergistically increased relative risk when they nant.268 This leads to enhanced Ag presentation and T-cell
occur as a heterozygous combination compared to the risk recognition with intestinal inflammation in susceptible pa-
of disease conferred by either haplotype alone.258–260 This tients.266,267,269 The protease resistance of the dominant DQ2-
raises the possibility of transcomplementation of HLA-DQ restricted gliadin peptides contrasts with DQ8-associated
gene products producing new molecules involved in anti- disease where the dominant gliadin peptides are protease
gen presentation.261 Analysis of large cohorts of patients in sensitive. The HLA-DQ2– associated and HLA-DQ8–associ-
genome-wide marker studies offers a promising approach ated forms of celiac disease have their own epitope patterns
to identify additional genetic factors that act synergistically of deamidation dependence leading to T-cell reactivity. The
with HLA allelic linkages in raising the odds of disease sus- DQ2-linked peptides possess single residue deamidation at
ceptibility.262 Online genomic maps and correlations of as- P6 as compared with the preferred deamidation at positions
sociations with particular loci further augments our ability P1 and P9 for DQ8-associated celiac disease.270 Deamidation
to understand the relative contributions of MHC-linked and of both P1 and P9 to form glutamate residues engenders
MHC-unlinked genes in autoimmune diseases such as type peptide salt-bridges with HLA-DQ8.271 Whereas the mecha-
1 diabetes. nism of immunopathology of celiac disease is well defi ned,
Celiac disease is an underdiagnosed inflammatory disor- the exact combination of factors that trigger celiac pathol-
der affecting the small intestine263 but resulting from a Th1 ogy is still mysterious.272
hypersensitivity to wheat gluten and related proteins present
in barley and rye. Normally manifest in childhood as a fail-
ure to thrive, the disease may affect adults quite later in life HLA and Drug Reactions
leading to malabsorption and wasting.264 Celiac disease can Recently, a number of adverse drug reactions (ADRs) have
be subclinical, presenting with anemia, osteoporosis, or neu- revealed strong associations with particular HLA types273
rologic symptoms.263 This disease is strongly associated with (Table 21.8). The most notable HLA associations are HLA-
HLA-DQ2 and/or HLA-DQ8 (DQ3)265 as well as with au- B*57:01 with abacavir hypersensitivity syndrome in the
toantibodies to the enzyme tissue transglutaminase (TG2). Caucasian population,274–276 HLA-B*15:02 with carbamaze-
Deamidated gluten peptides are presented to CD4 + T cells pine induced Stevens-Johnson syndrome and toxic epidermal
by HLA-DQ2 or HLADQ8 molecules via macrophages, den- necrolysis in Asian populations,277–279 and HLA-B*58:01 and
dritic cells, and B cells.266,267 HLA-DQ2 preferentially binds allopurinol hypersensitivity syndrome and Stevens-Johnson
peptides that contain glutamic acid residues at P4 or P6/P7. syndrome/toxic epidermal necrolysis.280–282 Odds ratios or
The dominant gluten peptide is thought to have a low affin- relative risks of these associations are greater than 500, 1000,
ity for HLA-DQ2 but this is improved if P6 glutamine is de- and 800, respectively. These ADRs are all associated with
amidated first.266,267 HLA-DQ2 forms a network of hydrogen HLA class I allotypes, with evidence of T-cell involvement.
TABLE 21.8 HLA–Linked Drug Hypersensitivity Reactions

Drug Adverse Drug Reaction HLA Association
MHC class I associations Abacavir HSS B*57:01
Allopurinol SJS/TEN and HSS B*58:01
Carbamazepine SJS/TEN B*15:02
Feprazone Fixed drug eruption B22
Flucloxacillin Hepatitis B*57:01
Sulfamethoxazole Fixed drug eruptions A30, B13, Cw6
Sulfonamides (including TEN A29, B12, DR7
sulfamethoxazole)
Levamisole Agranulocytosis B27
Oxicam SJS/TEN A2, B12
Phenytoin SJS/TEN B*15:02
MHC class II associations Aspirin Asthma DPB1*03:01
Urticaria DRB1*13:02-DQB1*06:09
Hydralazine Systemic lupus erythematosus DR4
Lapatinib Hepatotoxicity DRB1*07:01-
DQA1*02:01-
DQB1*0202/0203
NSAIDs Anaphylactoid and cutaneous reactions DR11
Ximelagatran Elevated alanine aminotransferase DRB1*07, DQA1*02
HLA, human leukocyte antigen; HSS, allopurinol hypersensitivity syndrome; MHC, major histocompatibility complex; NSAID, nonsteroidal anti-inflammatory drug;
SJS, Steven-Johnson syndrome; TEN, toxic epidermal necrolysis.
Adapted from Bharadwaj et al.273

The kinetics of the syndromes are also consistent with Mutations in the H2 Region
T-cell–mediated drug hypersensitivity.276,283–286 The sys- The rapid pace of improvements in deriving mouse and
temic features of many HLA-linked ADRs suggest that gen- other models for genetic disease based on the engineering of
eral mechanisms might mediate these reactions. Abacavir- transgenic, targeted deletion and mutation, and conditional
pulsed APCs expressing HLA-B*57:01, but not closely knockout mice emphasizes the importance of understanding
related allotypes, stimulate CD8-positive T cells that are the valuable reservoir of Mhc-I and Mhc-II mutants origi-
drug-specific.287 It is now known that abacavir specifically nally derived by laborious methods and still available for
binds the HLA-B*57:01 Ag-binding cleft straddling the cen- precise studies of the role of the Mhc in immune responses.
tral pockets and occupying the F-pocket which normally Mutations at the H2 locus have been identified in animals
accommodates a tryptophan anchoring the carboxyl termi- screened by skin grafting in extensive experiments carried
nus of the peptide.287a Consequentially, the chemical envi- out over a 25-year period.294,295 By grafting tail skin of siblings
ronment of the cleft is changed which in turn alters the range to and from each other, spontaneous or induced mutant ani-
of self-peptides that now bind HLA-B*57:01. Accordingly, mals that displayed either “gain,” “loss,” or “gain plus loss”
abacavir induces a dramatic shift in the endogenous pep- transplantation phenotype were identified. Gain mutants are
tides selected for binding by HLA-B*57:01. This leads to a those that express a new transplantation antigen—thus their
state of “altered self” that stimulates a broad range of CD8+ skin is rejected by their nonmutant siblings; loss mutants
T cells reactive in a GVHD-like manner, explaining the have lost a transplantation antigen—thus they recognize the
clinical manifestations of the drug hypersensitivity.287 The skin of their siblings as foreign and reject that graft. Gain
mode of abacavir binding provides a basis for understan- plus loss mutations give effects in both directions—they re-
ding its exquisite specificity in binding to HLA-B*57:01 ject the skin of their siblings, and their skin is rejected by
and not to closely related allotypes with polymorphisms their siblings as well. In classic studies, Melvold, Kohn, and
in the C, D, E, or F pockets. It will be interesting to learn colleagues screened a large number of mouse progeny.294,295
whether this represents a general mechanism for HLA- Homozygous inbred and F1 animals were examined, yield-
linked drug reactions applicable to other small molecule ing a total of 25 H2 mutations identified at K, D, L, and Ab
drugs that might bind specifically to HLA antigen-binding loci, and an additional 80 mutations of non-H2 histocompat-
clefts. HLA testing before administration of these drugs is ibility genes. Although earlier studies suggested that all H2
now recommended by many physicians and the U.S. Food genes might be hypermutable, a more complete retrospec-
and Drug Administration in order to avoid life-threatening tive evaluation of the available data suggests that with the
ADRs.273,288 The very high relative risks of abacavir, carba- exception only of the H2-Kb gene, the spontaneous mutation
mazepine, and allopurinol ADRs and HLA class I allotypes rate for H2 genes was comparable to that for non-H2 genes.
suggests a very high level of specificity in HLA binding, The characterization of these mutant animals, first based
either directly or through a drug-peptide conjugate(s) that on peptide maps and amino acid sequences of the H2 pro-
only binds the associated HLA allotype. Interestingly, such teins,296–298 and later based on the nucleotide sequences of the
high relative risks are not commonplace in HLA allotypes cloned complementary DNAs or genes,159 provided some of
and protective immunity, perhaps because it is unusual for the basic biochemical information on which further studies
any single determinant or HLA allotype to control protec- of structure and function and mechanism of gene evolution
tive phenotypes. were based. X-ray structure determination of the H2-K bm1
In addition to recent progress in recognizing Mhc as- and H2-K bm8 mutant proteins suggests explanations for the
sociations with adverse reactions to small molecule drugs differences in T-cell recognition that result from what might
and a variety of autoimmune and immune deficiency appear to be subtle amino acid substitutions.165
conditions, further analysis has identified relationships
of HLA-linked markers with susceptibility and progno-
sis in a number of infectious diseases. Among the most Control of Expression of MHC Molecules
striking has been the recognition that delayed progres- MHC molecules, synthesized in the ER and destined for cell
sion of human immunodeficiency virus (HIV) infection surface expression, are controlled at many checkpoints be-
to acquired immunodeficiency syndrome correlates with fore their final disposition as receptors available for inter-
possession of HLA alleles that express a molecule with a action with either T cells or NK cells. The classical MHC-I
“HLA-Bw4 motif ” (see Table 21.2) containing isoleucine molecules are trimers, consisting of the polymorphic H2 or
at position 80 along with a gene for the NK-cell receptor HLA heavy chain, the light chain β2m, and the assembled
KIR3DS1. This observation suggested that KIR3DS1 of- self-peptide. Thus, the cell surface expression of the as-
fered protection early in HIV infection for those patients sembled trimer may be reduced by a defect in the synthesis,
bearing the appropriate Bw4 allele.289 More recent obser- steady state, or assembly of any of these requisite protein/
vations indicate that patients with such a KIR3DS1/B*04 peptide chains. The first level of MHC-I expression control
ile80 genotype also exhibit resistance to late opportunistic is transcriptional; interferon-γ –dependent regulation is par-
infections.290 Other studies now show that particular HLA- ticularly important.299 For the most part, MHC-I molecules
DRB alleles as well as MHC-I alleles significantly influence are ubiquitously expressed. Their expression is dependent on
hepatitis B and C virus susceptibility, persistence, and a complex regulatory process that coordinately controls the
response to treatment.291–293 transcription of both MHC-I and β2m.300,301 The basis of the

more limited tissue specific expression of MHC-Ib molecules mic epithelial cells, and B cells, but can also be detected on
is of interest because of the potential importance in the role activated T cells of the human and rat—suggests that tran-
of some of the MHC-Ib molecules in tolerance to the pla- scriptional regulation plays an important role. Extensive stud-
centa. HLA-E and HLA-G, expressed on placenta, and HLA- ies of the promoter activities of Mhc-II genes have defined a
F, another MHC-Ib with limited tissue specific expression, number of specific transcriptional regulatory sequences.336
have been examined in considerable detail.116,302–305 Another One transcription factor, MHC class II transcriptional ac-
critical level of control of MHC-I expression is genetic, that tivator (CIITA) is clearly crucial.300,336–338 The expression of
is, the genes for a particular chain or key components of the CIITA is regulated in a complex fashion, possessing at least
PLC must be present for the trimer to be assembled and then three distinct promoters that function in a cell type–specific
expressed at the cell surface. A decrease in heavy chain syn- manner.339 MHC-II deficiency diseases categorized as bare
thesis as a result of a structural mutation in a coding (such lymphocyte syndrome, type II, result from lesions in at least
as HLA-A*30:14L [see Table 21.1]) or in a noncoding region four different complementation groups, implicating defects
(such as HLA-A*24:02:01:02L) would reveal the simplest in CIITA, RFX5, RFXAP, or RFXANK, all encoding proteins
level of control. In addition, the light chain is absolutely re- regulating MHC-II transcription.338,340–344 Polymorphisms in
quired for cell surface expression of most MHC-I molecules. MHC-II and the resulting differences in tissue and lineage-
Deficiency of β2m in various tumors is a relatively common specific expression of MHC-II suggest a role in the control of
occurrence.306–308 Induced β2m defective animals (B2mo/o, autoimmune disease.145,345–347
“β2m knockouts”) lack normal levels of MHC-I expression A unique aspect of MHC-II regulation is the need to pro-
because of this requirement for cell surface expression.309,310 tect its peptide-binding site from loading with self-peptides
In addition, polymorphism of the MHC-I heavy chain may in the ER and the requirement to traffic to an acidic endo-
influence the efficiency with which the molecule can inter- somal compartment where antigenic peptides, the products
act with the PLC or with which it binds self-peptides, and of proteolytic digestion of exogenous proteins, can be ob-
thus some alleles, because of a decreased ability to be loaded tained. These two functions are controlled by the type II
with self-peptides, may not be well expressed at the surface. membrane protein, invariant chain, Ii,30,348–350 which forms
The rest of the MHC-I biosynthetic pathway is dependent on a nine subunit complex consisting of three Ii and three αβ
proper generation of cytosolic peptides by the proteasome MHC-II heterodimers. The region of Ii that protects the
and delivery to the ER by TAP, appropriate core glycosylation MHC-II peptide-binding groove, the class II-associated in-
in the ER, transport through the Golgi, and arrival at the variant chain peptide (CLIP), is progressively trimmed from
plasma membrane.311 A number of persistent viruses have Ii and is ultimately released from the MHC-II by the cata-
evolved mechanisms for subverting this pathway of expres- lytic action of HLA-DM in the endosome to allow exchange
sion.312 The herpes simplex virus encodes a protein, infected for peptides generated there. HLA-DO, another MHC-II–
cell peptide 47,313–315 that blocks the activity of the peptide like molecule, regulates the repertoire of bound peptides,
transporter TAP.316 Cowpox virus encodes two proteins, presumably by modulating the catalytic activity of the DM
CPXV12 and CPXV203, which contribute to the downregu- exchange reaction. The important role of Ii in regulating
lation of host MHC-I, CPXV12 inhibits TAP, and CPXV203 MHC-II expression has been emphasized by the behavior of
retains MHC-I in the ER.317,318 The human disease bare lym- induced mutant mice lacking normal Ii351–353 that exhibit a
phocyte syndrome, type I, manifested by chronic lung dis- profound defect in MHC-II function and expression.
ease in late childhood, results from genetic lesions in either
TAP1, TAP2, or tapasin, also known as TAP-binding pro-
tein.93,319–323 Several proteins encoded by the human CMV, STRUCTURE OF MHC AND MHC–LIKE MOLECULES
unique short region proteins 2 and 11 (US2 and US11), cause “There is nothing that living things do that cannot be under-
rapid protein degradation of MHC-I molecules.324 Murine stood from the point of view that they are made of atoms acting
CMV encodes an MHC-I-like protein, m152/gp40, which according to the laws of physics.”
inhibits MHC-I expression by diverting MHC-I molecules —Richard Feynman354
to rapid degradation in the endosome.325,326 “Love hides in molecular structures, yeah,
Other molecules that assist the large DNA viruses in evad- Love is the answer.”
ing either the T-cell or NK-cell immune response include —Jim Morrison355
human CMV UL142.327 Murine CMV328 encodes a number
of open reading frames considered MHC-I homologues So central are Mhc genes and their encoded molecules to
that may function to deceive NK cells into the perception both the regulation and the effector function of the immune
of normal MHC-I expression.2,329–331 The adenovirus 2 pro- system that it has been apparent almost since their disco-
tein, E3/19K, also functions to block the transfer of folded very that an understanding of their structure and structural
assembled MHC-I molecules from the ER to the Golgi.332–335 interactions would be fundamental to a comprehension of
MHC-II molecules are also susceptible to regulation at their physiologic effects. Structural relationships of MHC
multiple steps. The clear-cut tissue dependence of MHC-II molecules came first from understanding serologic differ-
expression—MHC-II molecules are generally found on ences, then from analysis of cellular immune responses,
cells, which have specific antigen presentation functions and subsequently from biochemical studies of the MHC-I
such as macrophages, dendritic cells, Langerhans cells, thy- and MHC-II chains. Amino acid sequence comparisons

suggested a domain structure for the MHC-I molecules, the cell with a short carboxyl-terminal tail. The light chain
and an exon/domain correspondence was apparent with the of the MHC-I molecule, β2m, is a single domain soluble
initial identification of genomic and complementary DNA molecule that is encoded elsewhere in the genome.
clones.356 Shortly thereafter, a large number of sequences The MHC-II molecule consists of two chains embed-
were determined leading to their alignment and comparison. ded in the membrane as type I proteins, an α chain and a
Databases of these sequences are maintained and updated as β chain. The α and β chains both consist of two major ex-
indicated in previous sections of this chapter, and searches of tracellular domains, α1 and α2, and β1 and β2, respectively,
standard sequence databases may find Mhc gene and protein each linked to a transmembrane domain and cytoplasmic
sequences of numerous species. Expression of recombinant sequences. An initial analysis of both MHC-I and MHC-II
clones encoding MHC molecules in several systems: mam- molecules suggests that they are noncovalently assembled
malian cells, insect cells, and bacteria, then permitted the heterodimers consisting of four extracellular domains, the
accumulation and purification of molecules for functional, two membrane proximal domains (α3 and β2m for MHC-I
binding and ultimately x-ray structural studies. and α2 and β2 for MHC-II) of each molecule are Ig-like,
A current search of the protein data bank (www.rcsb while the two amino terminal domains (α1 and α2 of
.org/pdb) for the keyword “histocompatibility” yields three- MHC-I and α1 and β1 of MHC-II) possess what is known
dimensional coordinates of more than 500 MHC/peptide, as the MHC-fold. The α1 domains of both MHC-I and
MHC/peptide/TCR, MHC/peptide/coreceptor, and MHC/ MHC-II lack the intradomain disulfide bond characteristic
peptide/NK receptor complexes that allow an understand- of the other extracellular domains. The cytoplasmic domain
ing of the function of these molecules in a detailed structural of MHC-I molecules can be regulated by splicing and differ-
context.357 However, these structures also pose a number of ential phosphorylation or other modification, and is likely
questions that may only be addressed by additional func- to play a role in cell surface stability and cycling between
tional experiments in whole animals complemented by bio- the cell surface and other intracellular compartments.366–372
physical methods applied in vitro. The molecular biologic, However, analysis of directed mutants of MHC-I in some
functional, and structural studies have led to the develop- systems indicates that the cytoplasmic domain is not re-
ment of the use of MHC multimers as extremely powerful quired for cytoskeletal association or surface recycling.373
tools for imaging specific T cells or NK cells. The goal of this (The cytoplasmic domain of MHC-I molecules is the target
section of this chapter is to summarize these developments for the Nef protein of HIV, and the Nef/MHC interaction re-
with an eye toward explanation of function by structure and sults in MHC-I downregulation in HIV-infected cells.374,375
in hopes of revealing some of the current quandaries that The MHC-II transmembrane and cytoplasmic domains
continue to confound our understanding of the function of have clear effects on the level of cell surface expression, the
the Mhc. efficiency of antigen presentation, and the rate of lateral dif-
fusion of the molecules in the cell membrane.376,377 Single
particle tracking of MHC-II molecules indicates that such
Amino Acid Sequences—Primary Structure lateral diffusion is governed, at least in part, by interaction
Before the cloning of Mhc genes, the biochemical purifica- with the actin-based membrane skeleton.378 Amino acid se-
tion and amino acid sequence determination of the HLA-A2, quence alignments of MHC-I and MHC-II proteins, partic-
HLA-B7, and H2-K b MHC-I molecules358,359 indicated that ularly of the human molecules, are available in a number of
the MHC molecules had unique amino terminal “domains” databases such as www.ebi.ac.uk/imgt/hla/ and http://www
and showed similarities to Igs in their membrane proximal .ebi.ac.uk/ipd/mhc/.
regions. Early concerns were to identify the differences be-
tween allelic gene products as well as the differences between
MHC proteins encoded at different loci. With the cloning Identification of Peptides Bound by MHC Molecules
of complementary DNAs and genomic clones for MHC-I Although initial understanding of MHC molecules indi-
molecules156,360–362 and shortly thereafter for MHC-II mol- cated that they are heterodimers, a complete understanding
ecules,151,363–365 the encoded amino acid sequences of a large of their structure and resulting function requires the ap-
number of MHC molecules of a number of species quickly preciation that they are, in fact, heterotrimers, in which the
became available. The comparison of gene and complemen- third component is a short peptide ranging from eight to
tary DNA sequences gave an indication of the exon/intron roughly 15 amino acids in length. This characteristic of ma-
organization of the genes and explained the evolution of the ture MHC-I and MHC-II molecules makes them a unique
MHC molecules as having been derived from primordial class of protein molecules because an integral part of their
single domain structures of a unit size of a single Ig-domain peptide composition and resulting structure derive from a
(such as the light chain β2m) that duplicated to form the heterogeneous array of peptides delivered to the assembling
basic unit of the MHC-II chain (two extracellular domains) molecules during their maturation. Many different lines of
and the MHC-I chain (three extracellular domains).129 The evidence coalesced over a short period of time to demon-
canonical MHC-I molecule has a heavy chain (also known strate that MHC molecules function by binding peptides.
as the α chain), which is an intrinsic type I integral mem- From functional experiments, MHC-II–restricted T-cell re-
brane protein with amino terminal domains called α1, α2, sponses to protein antigens were shown to be dependent on
and α3, is embedded in the cell membrane by a hydrophobic peptide fragments.379 The first direct evidence of MHC/pep-
transmembrane domain and extends into the cytoplasm of tide interactions was that purified MHC-II proteins could

bind synthetic peptides in a specific, saturable, and stable from which the peptides were obtained, and that the length
manner380,381 with measurable affinity and a remarkably of the bound peptides was well defined and short, ranging
slow dissociation rate.381 For MHC-I molecules, the results from 8 to 10 amino acids. From such experiments, a number
were at first less clear, but the realization that some cell lines of peptide “motifs” of peptides bound to particular MHC-I
defective in MHC-I surface expression could be induced to molecules and allelic products were identified. Often, spe-
express higher levels of surface MHC-I molecules by ex- cific amino acids were identified at particular Edman degra-
posure to the appropriate peptides382 led the way for direct dation steps, indicating a common, highly preferred residue
measurement of MHC-I peptide binding.383 at the same spacing from the amino terminus of the pep-
Several laboratories succeeded in developing methods for tide. Thus, the peptide “motif” could be determined even
the partial purification and identification of the peptides from heterogeneous pools of peptides eluted from particu-
that copurified with MHC molecules. A viral, antigenic pep- lar MHC-I molecules. Further refinements in methodology
tide was identified bound to the MHC-I molecule H2-K b by have included the application of mass spectrometry to the
recovering MHC molecules from virus infected cells, frac- identification and sequencing of individual peptides.388,389
tionating the bound peptides chromatographically, identify- Alternative approaches for identifying peptide motifs in-
ing a peak of functional biologic activity in a cytotoxic T-cell clude the use of soluble analogs of MHC-I molecules to ease
assay, and determining the amino acid sequence of the re- the purification390,391 or the use of peptide display libraries to
covered peptide by radiochemical techniques.384 Another identify those peptides that can bind the MHC.392,393 With
method that was applied to the identification of both virus- steady improvements in technology, including the use of
derived peptides as well as the “motif” of self-peptides by transfected soluble versions of MHC-I molecules, immuno-
particular MHC-I molecules involved first the isolation of a affinity purification of the MHC-I molecules of interest, tan-
large amount of detergent solubilized MHC-I using appro- dem reverse phase high pressure liquid chromatography and
priate antibodies, the elution of the bound peptides, their further identification and sequencing by tandem mass spec-
partial purification as pools by reverse phase high pressure trometry, particularly MHC-I motifs, as well as tumor anti-
liquid chromatography, and the determination of the amino gen peptides, may now be identified routinely.394 A summary
acid sequence of the bound peptides by classic amino termi- of MHC-I peptide motifs is given in Table 21.9. Regularly
nal sequencing.385–387 The unpredicted and surprising result maintained online databases of MHC-I peptide motifs
obtained from these studies of MHC-I–derived peptides was may be found at several sites (www.imtech.res.in/raghava/
that specific amino acid residues were favored at particular mhcbn/links.html and www.syfpeithi.de).395 Antigenic
positions of the sequence, depending on the MHC-I molecule peptides that represent those observed as tumor antigens
TABLE 21.9 Peptide-Binding Motifs for Some MHC Class I Moleculesa

Position 1 2 3 4 5 6 7 8 9 (or C-term)
MHC-I
molecule
HLA-A*01 (T) D (L) Y
HLA-A*02:01 L(M) (V) V(L)
HLA-A*03 L(V,M) (F,Y) (I,M,F,V,L) (I,L,M,F) K,Y,F
HLA-B*07 P (R) L,F
HLA-B*08:01 K K,R L
HLA-B*27:05 R L,F (Y,R,H,K)
HLA-B*35:01 P Y,F,M,L,I
HLA-B*53:01 P W,F,L
H2-Kb Y F (Y) L (M,I,V) *prefers 8-mers
H2-Db N M(I,L)
H2-Kd Y(F) I(L,V)
H2-Dd G P (R) I(L,F)
H2-Ld P,S F,L,M
H2-Kk E I(V) *prefers 8-mers
H2-Dk Motif uncertain (R)
Qa-2 (M,L,Q) (N,I,L) (V,I) (K,M,I) H L,I,F
H2-M3 formyl-M
RT1.A1 (A,S,V) F,Y Y,F,L,M
a
Peptide-binding motifs for the indicated MHC class I molecules are shown in the single amino acid code. Position refers to the amino acid position of the peptide from the amino
terminus. Only the most common residues are shown. Assignments in parenthesis are less common than the others. These motifs are taken from the more extensive summary
of www.syfpeithi.de/, www.imtech.res.in/raghava/mhcbn/links.html, and Biddison and Martin.562 Full sequences of proteins may be queried for MHC peptides presented by
particular alleles with http://bimas.dcrt.nih.gov/molbio/hla_bind/, www.immuneepitope.org/, or www.mpiib-berlin.mpg.de/MAPPP/expertquery.html.
HLA, human leukocyte antigen; MHC, major histocompatibility complex.

and available for recognition by T cells are summarized at peptides released from purified MHC-II molecules is re-
www.cancerimmunity.org/peptidedatabase/differentiation. quired. Identification of MHC-II/peptide-binding motifs by
htm. In addition, the immune epitope database (www.im- bacteriophage display is also possible.402 In accord with the
muneepitope.org)396 offers a regularly updated and curated view that MHC-II molecules present peptides derived from
Web site that describes MHC motifs, allows evaluation of an “outside in” pathway, many of the peptides that copurify
new proteins for MHC epitopes, and provides new compu- with MHC-II molecules represent molecules derived from
tational tools for such evaluation. An algorithm that allows the extracellular milieu or the medium in which the cells
the prediction of candidates for MHC-I–restricted peptides were grown. Analysis of MHC-II/peptide complexes with
based on the amino acid sequence of the protein of inter- cloned T cells and monoclonal antibodies with MHC/pep-
est is also available (www-bimas.cit.nih.gov/molbio/hla_ tide specificity reveals that, in part because of the ability of
bind/).397 The distinction between “motif” residues of an MHC-II molecules to accommodate peptides with extensions
MHC-restricted peptide and “anchor” residues is an impor- at their amino and carboxyl-termini, and in part because of
tant one. Motif refers to those amino acid residues that are the smaller role that anchor residues seem to play in peptide
identified based on the sequences of self-peptide or antigenic binding by MHC-II molecules, occasionally even a unique
peptides that have been demonstrated to bind or copurify peptide can bind a particular MHC-II molecule in more
with a particular MHC molecule. Anchor implies a biophys- than one frame.403–406 As a result of structural studies (sum-
ical function of the particular amino acid residue as specifi- marized in the following) and compilation of sequences of
cally interacting with a particular part of the MHC molecule those peptides bound by particular MHC-II molecules, the
itself. The designation of a residue as an anchor residue may general conclusion is that MHC-II molecules all have bind-
be inferred by analysis of binding to peptide variants in the ing pockets identifiable for the particular allelic product. For
context of the parental peptide. Alternatively, anchors can some alleles, such as HLA-DR1 and IAd, these pockets are
be defined from a knowledge of the x-ray structure of the spaced at position P1, P4, P6, and P9 (or i, i + 3, i + 5, i + 9).
peptide complexed to the MHC-I molecule, and predictions (Some analyses, borne out in part by structural studies, sug-
of likely peptide motifs for MHC molecules whose sequence gest that a pocket at P7 [i + 6] may also contribute to the
is known but whose structure has not yet been determined energy of binding.) Because of the complexity and potential
can be made by comparison to known structures. degeneracy of peptide motifs for MHC-II molecules, more
The identification of MHC-II–bound self-peptide-or an- elaborate schemes have been devised for predicting those
tigenic peptides by biochemical methods similar to those peptides that may bind to particular MHC-II molecules.407 A
employed for MHC-I molecules has been proved more dif- summary of identified peptide-binding motifs for MHC-II
ficult because the MHC-II molecules do not have the rigor- molecules is in Table 21.10, but for some alleles, the analy-
ous requirements for a defined amino terminus or restricted sis of MHC-II binding peptides is not straightforward and
length. Whereas MHC-I molecules bind peptides with a demands sophisticated algorithms for identifying potential
particular motif residue at a specific position as defined by antigenic peptides from new proteins. The site www.imtech
the amino terminus, resulting in the ability to identify the .res.in/raghava/mhcbn/links.html tabulates a number of
dominant residue at a particular step in the Edman degra- links to various servers for predicting likely peptide epitopes
dation even amidst a pool of peptides, MHC-II molecules for particular MHC-I and MHC-II molecules. Implicit in the
bind peptides with “ragged ends,” and little information is description of MHC-I and MHC-II peptide motifs is that the
obtained from the standard methods of sequencing of pools bound peptides bind to the MHC molecule only in an amino
of peptides.398–401 Thus, more precise fractionation of the to carboxyl-terminal orientation, that is, that all peptides
TABLE 21.10 Peptide-Binding Motifs for Some MHC Class II Moleculesa

Position i (P1) i+1 i+2 i + 3 (P4) i+4 i + 5 (P6) i+6 i+7 i + 8 (P9)
Allele
DRB1*01:01 YVLFIAMW LAIVMNQ AGSTCP LAIVNFY
DRB1*03:01 LIFMV D KREQN YLF
DRB1*04:01 FYWILVM PWILVADE STQHR DEHKNQR.. DEHKNQR..
DRB1*04:05 FYWVILM VILMDE NSTQKD DEQ
DQA1*05:01/B1*03:01 WYAVM A AIVTS QN
H2-IAb (undefined)
H2-IAd STYEVWMLI.. VLIA AV
H2-IAg7 KHSAV L VA DSE
H2-IAk DN IVLN EQ
H2-IAs (undefined)
H2-IEb WFYILV LIFSA QNASTHRE KR
H2-IEd WFYILV KRIV ILVG KR
H2-IEg7 ILVFWM DESMV QNASTED RKMF
a
Major histocompatibility complex class II peptide-binding motifs are summarized from those of Biddison and Martin562 and www.syfpeithi.de/home.htm.

bound to MHC molecules bind with the amino terminus publications, based on a structure determined to a resolution
at the lefthand side of the peptide-binding groove, in the of 3.5 Å, focused mainly on the general structural outline
canonical structural representation that will be described of the molecule. More recently, structures of a wide variety
in further detail subsequently. However, a recent analysis of MHC-I molecules of different species including human,
employing both x-ray crystallography and nuclear magnetic mouse, rat, and chicken, complexed with specific peptides,
resonance indicates that the CLIP peptide may undergo ori- have been determined. Ribbon diagrams of HLA-A2 as seen
entational inversion, binding dynamically in the carboxyl to from the side (Fig. 21.6A) and from the top (Fig. 21.6B), com-
amino as well as the amino to carboxyl direction.408 plexed with a unique synthetic peptide, indicate the design
of the entire molecule: the peptide-binding site is supported
by the β-sheet floor, and the floor in turn is supported by the
High-Resolution Structures two Ig-like domains, the α3 domain of the heavy chain and
MHC-I Molecules the β2m light chain. (In the canonical view of the MHC-I,
A most graphic description of the relationship of form to HLA-A2 molecule shown in Fig. 21.6B through D, the amino
function of the MHC molecules was made by Bjorkman terminal residue of the bound peptide lies to the left, and the
et al.409,410 who determined the three-dimensional struc- carboxyl terminal residue lies to the right.)
ture of the human MHC-I molecule, HLA-A*02:01, by x-ray The comparison of this structure with that of the
crystallography. For these studies, the extracellular, soluble closely related human MHC-I molecule, HLA-A*68:01 (see
portion of the type I membrane-associated molecule was Table 21.2), suggested that surface depressions in the groove
purified by papain cleavage from the surface of tissue cul- of the MHC-I molecule, now designated pockets A through
ture cells. At the time, there was not a clear appreciation of F, would be available for interactions with some of the side
the role of peptide either in the assembly of the molecule chains of the bound peptide.411,412 These six pockets are il-
or of the nature of the recognition of the MHC molecule lustrated in Fig. 21.6C and with bound peptide in Fig. 21.6D.
by TCRs or, for that matter, NK receptors. Despite the fact Concomitant with the determination of the x-ray structure
that the first purified HLA-A2 molecules possessed a het- of the human MHC-I molecule, HLA-B*27:05,413,414 the
erogeneous mixture of bound peptides, protein from these motif of the peptides that were recovered from this molecule
preparations crystallized readily, and electron density maps was determined, permitting the more precise modeling of
calculated from the diffraction data were interpretable, al- the bound peptide in the cleft of the MHC-I. For HLA-B27,
lowing modelling of the backbone molecular structure. The this was of particular interest because the bound peptides
most important insight in the interpretation of the derived had a strong overrepresentation of arginine at position 2,
electron density map was that part of the electron density,
and thus part of the structure, was due to this mixture of
peptides bound tightly by the molecule, and that this den-
sity could not be modeled based on the known amino acid
sequence of HLA-A2.
This first MHC-I structure clarified several important as-
pects of the mechanism by which the MHC-I molecule car-
ries out its peptide binding function. The amino terminal
domains (α1 and α2) form a unitary binding site for peptide.
This domain unit (also called a “superdomain”) consists of
a floor of eight strands of antiparallel β-pleated sheet that
supports two α-helices, one contributed from the α1 domain
and one from the α2 domain, aligned in an antiparallel ori-
entation. The membrane proximal α3 domain has an Ig
C-type fold and pairs asymmetrically with the other Ig do-
main of the molecule contributed by β2m. The nature of rec-
ognition by T-cells was suggested by comparing the location
of those amino acid residues that had been characterized as
being strong elements in T-cell recognition, residues that dis-
tinguished closely related allelic gene products, and amino
acid residues that had been identified as those that were
responsible for the transplant rejection of the mutants of the
H2-Kb series.409 Amino acid residues of the MHC-I molecule
responsible for T-cell recognition were clearly classified into FIG. 21.6. Structure of HLA-A*02:01. A: Ribbon representation of
one of two categories or an overlapping set: those residues HLA-A2 heavy chain (green), β2m light chain (cyan), and bound
that were “on the top of the molecule,” exposed to solvent peptide (yellow). B: Ribbon representation of the peptide-binding
groove. C: Surface representation of the binding groove with pock-
and available for direct interaction with the TCR, and those ets labeled. D: Surface representation of binding groove colored by
residues whose side chains pointed into the peptide binding electrostatic charge (red, acidic; blue, basic) with peptide shown in
groove and might be considered crucial in the peptide-bind- stick illustration. Figure generated from protein data bank559 struc-
ing specificity of the particular MHC molecule. The original ture 2BSV using PyMOL.560

and scrutiny of the HLA-B27 structure suggested that the expose their carbohydrate moieties to solvent and be avail-
amino acid residues lining the B pocket, particularly glu- able for TCR interaction.422–424 An example of a 13-residue
tamic acid at position 45 as well as cysteine 67, were comple- peptide bound to its MHC-I–presenting element, the rat
mentary to the long, positively charged arginine side chain MHC-I molecule, RT1-Aa, was crystallized and shown to
of the peptide amino acid at that position. Structural studies produce a peptide/MHC complex with a large central bulge.
supported a view of MHC-I/peptide binding in which the Two different complexes consisting of the same MHC and
side chain of the carboxyl-terminal residue of the bound peptide reveal significantly different conformations in this
peptide sits deep in the F pocket. In addition, the amino ter- central bulge region.425 “Bulged” viral peptides have also
minal amino group of the peptide forms hydrogen bonds been characterized in complex with human MHC-I struc-
with the hydroxyl groups of conserved amino acids tyro- tures of HLA-B35 allotypes complexed with 13-mer and
sine 59 and tyrosine 171 that line the A pocket. A hydrogen 11-mer peptides.426,427 TCR recognition of such bulged pep-
bond from the amino group of conserved tryptophan 147 to tides can involve conformational adjustments of the TCR in
the backbone carbonyl oxygen of the penultimate peptide recognizing a fairly rigid peptide426 or “crumplings” of the
amino acid (usually position 8) also seems important, as do bound peptide by a largely rigid face of the TCR upon bind-
charge interactions and hydrogen bonds of the free carboxyl ing the MHC-I/peptide complex.428 Another set of modi-
group at the carboxyl terminus of the peptide with tyrosine fied peptides that can be bound by MHC-I molecules and
84, threonine 143, and lysine 146. presented to TCR includes self-peptides modified by phos-
Other structures of MHC-I molecules were determined phorylation.429,430 Although the fundamental topology of
involving complexes produced with homogeneous peptide, MHC-I structure and its interactions with antigenic peptide
assembled either in vitro from bacterially expressed pro- have been solved by the classic structures mentioned pre-
teins with synthetic peptide415 or exploiting MHC proteins viously, many more MHC-I/peptide complexes have been
expressed in insect cells.416,417 The structures determined studied crystallographically in order to define the details of
with homogeneous peptide confi rmed the impression ob- particular interactions with antigenic peptides. A number
tained from the structures obtained from molecules with of structures at a resolution of 1.40 Å or better have been
heterogeneous self-peptides. Examination of the H2-K b deposited in the protein database. Structure determination
molecule complexed with synthetic, known antigenic pep- of representative single-chain peptide-β2m-MHC-I,431 and
tides derived either from Sendai virus, vesicular stomatitis of MHC-I molecules containing photolabile peptides that
virus, or chicken ovalbumin revealed that the same MHC can be exchanged with other peptides easily,432 reinforce
molecule can bind peptides of different sequence, length, our basic understanding of the structure of MHC-I/peptide
and structure by virtue of their conserved motif residue complexes.
side chains. Although small conformational changes of the
MHC are detectable on binding the different peptides, the MHC-II Molecules
main distinction in the recognition of different peptides Before any MHC-II structure had been determined experi-
bound by the same MHC molecule is due to the location, mentally, a model was constructed based on the alignment
context, size, and charge of amino acid side chains that of amino acid sequences and the available MHC-I three-
are anchored in the MHC pockets. Conversely, solvent- dimensional structure.433 This model made several valid
exposed side chains contribute to the unique structure of predictions that were borne out by the subsequent x-ray
the surface of the molecule available for interaction with structure determination of HLA-DR1.434 MHC-II clearly
TCR and other receptors. showed similarity to MHC-I and formed its binding groove
The most consistent rule learned from the first x-ray by the juxtaposition of the α1 and β1 domains. The posi-
structures and complemented by peptide recovery and early tion of the electron density representing the heterogeneous
binding studies was that the carboxyl-terminal amino acid peptide that copurified with the HLA-DR1 was identified.
side chain of the MHC-I–bound peptide was embedded in Figure 21.7A and C show a ribbon diagram of HLA-DR1
the F pocket. A recent study revealing that the MHC-I mol- with a homogeneous bound peptide. In comparison to the
ecule H2-Db binds a pentapeptide well anchored in the F MHC-I structure (see Fig. 21.6), the peptides bound to the
pocket, but lacking amino terminal interactions, is consis- MHC-II molecule extend through the binding groove rather
tent with this view.418 However, other studies indicate that than being anchored at both ends.
MHC-I molecules may bind longer peptides that extend be- A comparison of the α-carbon backbone of the
yond the residue anchored in the F pocket,419 an observation peptide-binding region of an MHC-I structure with that
confirmed by a crystallographic structure.420 More recently, of an MHC-II structure is shown in Figure 21.8. The struc-
the lack of an absolute requirement for a free C-terminal tures are very similar although the binding domain is built
amino acid has been exploited in the engineering of single of the α1 and α2 domains from the same chain (for MHC-I)
chain molecules in which peptide, β2m, and the MHC-I or of the α1 and β1 domains that derive from two chains
heavy chain are covalently linked. These molecules have (for MHC-II). The location of polymorphic residues can be
unusual thermal stability and are effective MHC/peptide determined by variability plots based on multiple sequence
immunogens.421 An additional variation on the theme of alignments, as originally suggested by Wu and Kabat.435 The
MHC–I binding peptides based on particular anchor resi- current bioinformatics approach to variability prefers to cal-
dues includes the demonstration that glycopeptides, bound culate Shannon entropy, a function that describes the degree
to MHC-I via amino acid side chains and termini, can of unpredictability of an event.436,437 For protein sequence

FIG. 21.7. Color Ribbon Representation

of HLA-DR1. A: Side view. α chain is
in green; β chain is in blue; peptide in
stick representation in yellow. B: Top
view. C: Top view with bound peptide
(PKYVKQNTLKLAT) visualized. The illus-
tration was made with PyMol560 based
on the protein data bank559 coordinates
of 1DLH.
alignments, Shannon entropy provides a sensitive measure show that the bulk of MHC polymorphism derives from
of the diversity of particular amino acid positions.438 Several amino acid variability in regions that line the peptide-
online tools are available for implementing and displaying binding groove. This suggests that MHC polymorphism is
the results of such calculations (http://bio.dfci.harvard.edu/ required to allow the MHC molecules, and as a result, the
Tools/svs_help.html; http://imed.med.ucm.es/Tools/svs_ organism and its species, to respond to a changing antigenic
help.html; and http://consurfdb.tau.ac.il/). Comparative environment.
ribbon diagrams (Fig. 21.9), in which the location of the As with MHC-I, a further understanding of the details of
amino acid residues that are polymorphic for the human the interactions of peptides with the MHC-II molecule came
MHC-I and MHC-II chains are indicated with a color map, from crystallographic studies of molecules prepared with
FIG. 21.8. Comparison of a-Carbon Backbone of MHC-I

and MHC-II molecules. The α1α2 domains of MHC-I
(human leukocyte antgien [HLA]-A2, PDB 3HLA) and
the α1 and β1 domans of MHC-II (HLA-DR1, PDB 1DLH)
were superposed with PyMOL560 and displayed as a Cα
carbon representation. MHC-I is blue, the α1 domain of
MHC-II is red, and the β1 domain of MHC-II is orange.

of hydrogen bonds between conserved amino acids that line

the binding cleft and the main chain (ie, amino nitrogen and
carbonyl oxygen) atoms of the peptide. Although the pocket
designations refer specifically to the features of the MHC-II
molecule surface, occasionally the cognate peptide positions
that reside in such pockets are given the same names.
Among the most provocative observations from the
fi rst MHC-II structures was that the molecule was visual-
FIG. 21.9. Location of Polymorphic Amino Acid Residues in MHC-I ized as a dimer of dimers, and this moved a number of
and MHC-II molecules. Variability plots were calculated as de- investigators to consider the possibility that activation of
scribed by Kabat and Wu435 and level of variability illustrated on the T cell via its receptor might require the dimerization
ribbon diagrams (generated in QUANTA 2000 [Accelrys] of 3HLA412 or multimerization of the TCR, an event thought to be
[HLA-class I], 1DLH439 [HLA-DR], and 1JK8 [HLA-DQ]) where greatest dependent on the propensity of the MHC/peptide complex
variability is red, intermediate is green, and least is blue. to self-dimerize. The simple elegance of this dimer of di-
mers is illustrated in Figure 21.11. Several arguments sup-
port the dimerization hypothesis: the fi nding of a dimer
of dimers in the crystals of HLA-DR that formed in several
homogeneous peptide—in the first case HLA-DR1 complexed different space groups,434,439 the observation that a TCR
with an antigenic peptide derived from the hemagglutinin of Vα domain formed tight dimers and in its crystals formed
influenza virus.439 Based on this structure, a set of pockets dimers of the dimers,440 the demonstration of the abil-
was initially designated, numbered for the peptide position ity to immunoprecipitate MHC-II dimers from B cells,441
that is bound. For the influenza peptide studied in this ex- the apparent requirement for purified MHC-I dimers for
ample, the major interactions were from peptide positions stimulation of a T cell in an in vitro system,442 and the
1, 4, 6, 7, and 9 that are indicated in Figure 21.10. The deep fi nding that MHC-II/peptide/TCR complexes could form
P1 pocket accommodates the tyrosine (the third position of higher order multimers in solution as detected by quasi-
the peptide PKY VKQNTLKLAT) and the pockets indicated elastic light scattering.443 However, a number of strong
by 4, 6, 7, and 9 fit the Q, T, L, and L residues, respectively. counterarguments draw this hypothesis into question.
MHC-II exploits a similar mode of binding as compared Many MHC-II molecules other than HLA-DR1 that have
with MHC-I, but there are key differences. MHC-II lacks the been crystallized do not seem to form the same kind of
requirement for free amino and carboxyl termini of the pep-
tide, the peptide conformation is relatively extended (like that
of a type II polyproline helix), and the MHC-II forms number
FIG. 21.10. Location of Pockets in HLA-DR1 Based on the Cocrystal of FIG. 21.11. Ribbon Diagram of the Structure of HLA-DR1 showing the
HLA-DR1 with a Peptide Derived from the Influenza Hemagglutinin. dimer of dimers and the individual domains of the protein. α chains
An electrostatic surface representation of HLA-DR1 (1DLH) gener- are in green and yellow; β chains are in blue and salmon. Peptide
ated in PyMOL560 is shown with the positions of pockets P1, P4, P6, in stick representation is in red. The illustration was generated from
P7, and P9 indicated. PDB 1DLH with PyMOL.

dimer of dimers in their crystals.444,445 Modes of MHC-II been determined are of particular interest, the CD1 mol-
dimerization with distinct topology that would preclude ecules,453,454 H2-M3,137,455 MICA,456,457 FcRn,458 and Rae-1.459
simultaneous interaction with multiple TCRs on the same Other molecules that are thought to show structural relat-
T cell have been observed. None of the MHC-I molecules edness to the MHC-I fold but whose x-ray structures have
that have been examined by x-ray crystallography show not yet been reported include the stress-induced molecules
dimers in the same orientation as the MHC-II ones re- H-60173 and MULT1,460 and the mucosally expressed MR1
ported. A different Vα domain fails to dimerize even at molecule that selects invariant T cells in the gut.461 H2-M3
high concentration.446 Many reported x-ray structures of is of particular note because of its ability to bind and pres-
MHC/peptide/TCR complexes 447–450 fail to show dimeriza- ent peptide antigens that contain amino terminal N-formyl
tion. Despite the simple elegance of the dimer hypothesis, groups. H2-M3 was originally identified as the MHC-Ib
it is clear that additional experimentation will be required molecule that presents an endogenous peptide derived
to understand the topologic requirements for T-cell acti- from the mitochondrially encoded protein ND1 known as
vation through the αβ TCR. maternally transmitted factor.137,462 Thus, it was of interest
Recent additions to the library of MHC-II structures in- to understand in structural terms how this molecule binds
clude the I-Ag7 molecule, a unique MHC-II that provides one such N-formylated peptides.455,463 The crystal structure of
link in the susceptibility to insulin-dependent diabetes in H2-M3 complexed with an N-formylated nonamer peptide,
the mouse model.451,452 Although the structure fails to pro- fMYFINILTL, revealed that the structure of the A pocket,
vide direct evidence to explain the linkage to diabetes, it highly conserved among MHC-Ia molecules, which have
suggests that the novel repertoire of peptides bound by this tyrosine 7, tyrosine 59, tyrosine 159, tryptophan 167, and
MHC-II molecule reveals unique features of a wider peptide tyrosine 171, is quite different so that it can accommodate
binding groove and resulting relatively low-affinity interac- the N-formyl group in the A pocket. In particular, H2-M3
tion with peptide. Until recently, all of the MHC-II/peptide has an A pocket reduced in size and lined by hydropho-
structures revealed bound peptide in a canonical left to bic residues, leucine at 167 and phenylalanine at 171, and
right, amino to carboxyl orientation. A detailed analysis of leucine 16. These structural features cause the amino ter-
HLA-DR1 molecules bound to the CLIP 106 to 120 peptide minal nitrogen of the formylated peptide to be positioned
revealed crystals in two different forms, which on solution where the peptide position 2 amino nitrogen would lie in a
and refinement indicated that this peptide could interact in MHC-Ia molecule. Thus, the unique peptide selectivity of
either of two distinct orientations (Fig. 21.12).408 This excep- H2-M3 is explained in structural terms.
tion to the general rule of left to right, amino to carboxyl
orientation, may reflect the unique role that the CLIP pep- CD1
tide must play in protecting the MHC-II–binding site while Another MHC-Ib molecule of great interest is CD1, repre-
simultaneously being available for DM-mediated peptide sentative of a class of MHC-I molecules that map outside of
exchange in the endosome. the MHC, that have limited tissue-specific expression, and
that are capable of interaction with both αβ and γδ T cells.464
MHC-Ib Molecules In the human, there are two clearly distinct groups of CD1
H2-M3 molecules: one consisting of CD1a, CD1b, CD1c, and CD1e;
To this point, our description of MHC-I molecules has and another of CD1d alone.465 In the mouse and rat, only
focused on the classical MHC-Ia molecules, represented CD1d is expressed.466 As a group, these are β2m–associated
by HLA-A, HLA-B, and HLA-C in the human, and by chains that bind hydrophobic antigens, primarily glycolip-
H2-K, H2-D, and H2-L in the mouse. Several MHC-Ib ids, with the lipid moiety embedded in the CD1 heavy chain
molecules for which three-dimensional structures have and the carbohydrate portion exposed to solvent. CD1a,
CD1b, and CD1c are capable of binding and presenting vari-
ous nonpeptidic mycobacterial cell wall components such as
mycolic acid containing lipids and lipoarabinomannan li-
poglycans.467,468 NK T cells, a subset of αβ -TCR–bearing T
cells, of independent lineage and defined by the expression of
the NK1.1 marker, are restricted to CD1 recognition.469 The
crystal structures of both mouse and human CD1d1 have
been determined,453,470 revealing classic MHC-I structures
with the basic protein fold and β2m association quite similar
to that of the MHC-Ia molecules. Consistent with its appar-
ent biologic function of binding hydrophobic lipid-contain-
ing molecules, its binding groove is somewhat narrower and
deeper than that of either MHC-Ia molecules or MHC-II
molecules. The backbone configuration of the α1α2 domain
FIG. 21.12. MHC-II Binds Class II-Associated Invariant Chain
Peptide in Either N to C or C to N Orientation. The structures of structure of CD1 is shown in Figure 21.13, where it is com-
HLA-DR1 bound to the CLIP 106 to 120 peptide in either the N to C pared to the homologous region of H2-K b, HLA-DR1, and
(A) or C to N (B) orientation is illustrated. α1 domain is in green; β1 another MHC-Ib molecule, a neonatal Fc receptor (FcRn).
domain is in cyan. Peptide sequences are shown at the bottom. The structure of a number of CD1-glycolipid complexes

FIG. 21.13. Comparison of the Ribbon

Representations and Size of the Potential
Peptide-Binding Cleft. MHC-I molecule H2-Kb
(2vaa), MHC-II HLA-DR1 (1DLH), and MHC-Ib
molecules CD1d (1Z5L) and FcRn (1EXU) are
compared. Approximate maximal distances sep-
arating the two helices are indicated.
have been determined, and human and mouse CD1 mol- The structures were essentially identical for the CD1b heavy
ecules bound to different glycolipids have been compared.466 chain and β2m in the two complexes, and revealed a net-
The depth of the groove of CD1 results from the merging of work of four hydrophobic channels at the core of the α1α2
pockets to form what have been termed the A’ and F’ pockets domain, which accommodate four hydrocarbon chains
in place of the MHC-Ia A through F pockets. This A’ pocket of length from 11 to 22 carbon atoms. These channels are
is about the size of the binding site of a nonspecific lipid called A’, C’, and F’ for the three analagous to the A, C, and
binding protein. F pockets of the MHC-Ia molecules, and a fourth, termed
In an effort to understand more precisely how CD1 T’ which is a distinct tunnel. Illustrations of the binding
molecules bind lipid antigens, Gadola et al.454 crystallized groove with the bound alkyl chains is shown in Figure 21.14,
human CD1b complexed with either phosphatidylinositol which shows views of the binding pockets of CD1a, b, and
or ganglioside GM2 and determined their x-ray structures. d with different ligands. These structures illustrate how the
CD1d/α-gal-cer CD1a/sulfatide
CD1b/phosphatidylinositol CD1b/GM2
FIG. 21.14. Tunnels and Pockets of CD1 Molecules. Structures of (A) CD1d with α-galactosyl ceramide
(1Z5L), (B) CD1a with sulfatide (1ONQ), (C) CD1b with phosphatidylinositol (1GZQ), and (D) CD1b with gan-
glioside GM2 are shown, with pockets A’, C’, and F’ indicated and tunnel T’.

binding site of a classical MHC-I molecule may have evolved serves as an excellent example of similar structures in the
from (or to) the binding site of molecule like CD1 to provide immune system being diverted for alternate purpose. The
antigen selectivity for a distinct set of molecules that would importance of the FcRn has been underscored by observa-
be common to a set of important mycobacterial pathogens. tions of differences in the serum half-life of Ig in animals
Complexes of CD1 with their cognate NK TCR have been that, as a result of an induced deletion of β2m, lack the nor-
reported and will be discussed subsequently.471–474 mal expression FcRn as well and seem to metabolize serum
Ig aberrantly.480
Neonatal Fc Receptor
Another example of an MHC-Ib molecule, noteworthy be- MILL MHC–Like Molecules
cause it exemplifies a novel function of MHC molecules, is A novel set of MHC-Ib–like genes designated Mill (M HC
the FcRn. Originally described in the rat as a molecule of class I-like located near the leukocyte receptor complex)
the intestinal epithelium that is involved in the transport has recently been identified in both mice and rats.481–483
of colostral Ig from the lumen to the bloodstream,475,476 These encode MICA/B-like glycophosphatidylinsol-linked
homologues in the mouse and human have also been de- cell surface molecules that are associated with β2m and do
scribed,477–479 and the structure of the rat molecule has been not require TAP for cell surface expression. Their function
determined crystallographically458 (Fig. 21.15). As suggested remains unclear, although further examination indicates the
by the amino acid sequence similarity of the FcRn to MHC-I presence of two family members, Mill1 and Mill2, observed
proteins, the three-dimensional structure revealed consid- on cycling thymocytes, proliferating smooth muscle cells
erable similarity to MHC-Ia molecules.458 Specifically, α1 and fibroblasts.484
and α2 domains have similar topology to the MHC-I mol- M10 Proteins
ecule, although, as discussed previously, what would be the The robust MHC fold seems to possess both the stability and
peptide-binding groove in the MHC-Ia molecules is closed flexibility requisite for the molecular function of binding
tightly and lacks space sufficient for a ligand. The most pro- ligands such as peptides, glycopeptides, phosphopeptides,
vocative feature of the structure of the FcRn/Fc complex is and glycolipids as well as for subsequent interactions with
that the MHC-I–like FcRn interacts with the Fc through macromolecular receptors such as NK receptors and TCRs.
contacts from the α2 and β2m domains to interact with the Likely, these qualities have allowed its adaptation for distinct
Fc Cγ2-Cγ3 interface. As compared to the structure of the function related to a class of MHC-Ib molecules, the M1 and
unliganded Fc, the complex reveals both conformational M10 families, some members of which are associated with
changes in the Fc and the presence of several titratable the V2R mouse pheromone receptors. The structure deter-
groups in the interface that must play a role in the pH-de- mination of M10.5 revealed a β2m-associated MHC-I–like
pendent binding and release of Ig molecules from the FcRn. molecule with clear three-dimensional similarity to the
The FcRn has taken the MHC-I fold and diverted its func- MHC-I molecule H2-Dd.485
tion for an interaction with the Fc of the Ig. Amino acids at
what would classically be considered the “righthand side” MHC-Iv Molecules
of the peptide-binding groove make contact with the Fc in- Because viruses are constantly engaged in a struggle for
terface that lies between Cγ2 and Cγ3 domains. The FcRn survival in their interplay with the immune system of their
hosts, there is little surprise to note that pathogenic eukary-
otic viruses can evolve numerous approaches to evade the
immune response. MHC and structurally related molecules
are produced not only by vertebrates but also by CMVs,
large DNA viruses of the β -herpesvirus family, which have
coevolved with their vertebrate hosts over millennia and are
exquisitely adapted to persist in the face of host immune re-
sponses.2 Viral MHC-I–like molecules have been identified
bioinformatically from amino acid sequences of the open
reading frames predicted from genome sequence data. A
number of putative MHC-Iv molecules, including UL18 and
UL142 of the human CMV,327,486 members of the “m145 fam-
ily” of the mouse CMV1487,488 the 2L molecule of tanapox,489
and the U21 molecule of human herpesvirus 7,490 may mod-
ulate the immune response of the host. In some cases, such
as the UL18 protein of human CMV and the m144 protein
of mouse CMV, the sequence homology to bonafide MHC-I
molecules is sufficiently strong to unambiguously identify
them as MHC-I like. However, in most cases, such as with
FIG. 21.15. Structure of the Rat Neonatal Fc Receptor Complexed
the m145 family of mouse CMV, the sequence homology to
with Fc. Ribbon diagram of rat neonatal Fc receptor complexes MHC-I is very weak, and relatedness to MHC-I was origi-
with the Fc heterodimer consisting of the wild-type and nonbinding nally based on structure prediction algorithms alone. We
chains. This is from PDB 1I1A. term these virally encoded MHC-I–like molecules MHC-Iv

to reflect their distinct evolutionary history, their struc- evasins that inhibit the NK cell to enable virus survival
tural deviation from typical MHC-I molecules, and their and persistence in the host. Both UL18 and UL142 inhibit
role in evading host immune responses. A survey of DNA human NK function, the former by binding to the inhibi-
sequences of 11 genes obtained from 26 wild murine CMV tory receptor LIR-1. The mode of action of UL142 remains
isolates and laboratory strains indicated that several of the to be determined.327 The m157 binds to the inhibitory NK
MHC-Iv genes (m144, m145, and m155) revealed significant receptor Ly49I in mouse strains susceptible to viral infection
sequence variation, consistent with the view that this varia- but remarkably serves as a target for the activating NK re-
tion may offer some immune-evasive benefit to the virus.491 ceptor Ly49H expressed in resistant strains. The m144 pro-
Structural characterization of MHC-Iv molecules begins tein has also been shown to inhibit NK activation,495,496 but
with an evaluation of the requirement for β2m and/or pep- the ligand remains to be identified.
tide for stable expression. The UL18 is associated with both
β2m and peptide, whereas m144 may be associated only Complexes of MHC Molecules with Ligands
with β2m but does not require β2m for cell surface expres- The most recent and exciting developments in MHC biol-
sion. The structure of m144 has been reported,487 revealing ogy include the detailed description of the interactions of
the preservation of all structural elements of the MHC-I MHC-I and MHC-II molecules with their receptors and co-
fold (Fig. 21.16). The cleft of m144 does not seem capable receptors on cells of the innate and adaptive immune system.
of binding peptides as the groove is narrow and critical ty- Our structure/function survey will be completed by brief
rosine residues are not conserved, results consistent with descriptions of the interactions of MHC-I and MHC-II mol-
earlier biochemical studies.492 A unique disulfide anchors ecules with αβ TCRs, with the T-cell coreceptors CD8 and
the α1 helix to the β sheet. The structure of m157, a CMV CD4, and of MHC-I molecules with NK receptors. Details
MHC-I–like immunoevasin that binds the Ly49H activation of the interactions of a number of CD1 molecules with their
receptor of C57BL/6 mice and the Ly49I inhibitory receptor glycolipid antigens and specific NK T-cell receptors shed
of 129/J mice, has been reported.493 This β2m-independent, light on the biology and evolution of these molecules. A brief
peptide-free MHC-Iv molecule reveals the major features of description of interactions of MHC-II molecules with supe-
the MHC-I fold with some unique aspects, in particular a rantigen will follow. Each of these structural studies com-
novel amino terminal α helix. How it binds its Ly49 ligands plements a host of biologic experiments that have led to an
remains unclear. The m153 protein of mouse CMV, also a appreciation of the importance of understanding the struc-
member of the m145 family, reveals novel adaptations of tural basis of these immune reactions.
the MHC-I fold.1 The m153 protein does not require β2m
or peptide and, in contrast to other MHC-I molecules, is a MHC/T-Cell Receptor Interactions
noncovalent dimer. The monomers are associated in a head- Second to the initial visualization of the structure of the
to-tail fashion. An extended N-terminus contains a unique MHC-I molecule HLA-A2 by crystallographic analysis, the
disulfide that anchors it to the α3 domain. The m153 struc- images of TCR/MHC/peptide complexes initially reported
ture hints tantalizingly of more surprises to come as struc- some 15 years ago clarified the phenomenologic details of
tures of other MHC-Iv molecules become available. Other T-cell recognition, antigen specificity, and MHC restric-
MHC-Iv proteins whose structure have been determined tion. The initial structures were determined in several
include the 2L tanapox virus protein that binds the inflam- systems,447–449,497,498 and there remains ongoing interest in
matory cytokine TNF-α489 and the mouse CMV-encoded the details of particular TCR/MHC interactions. The first
m152, which sequesters both classical MHC-I and the stress- examples were of MHC-I–restricted TCR, one from the
induced NKG2D ligand, RAE-1, from the cell surface.494 mouse448 and one from the human.447 The mouse MHC-I
In those cases where functions have been elucidated, molecule H2-K b was analyzed in complex with a self-pep-
MHC-1v molecules have been shown to act as immuno- tide, dEV8, and a TCR known as 2C,448,499 and the human
FIG. 21.16. Comparison of MHC-Iv Molecules. Structures of MHC-Iv chains of m144 (1U58), m153 (2O5N),
m157 (2NYK), and the 2L protein of tanapox virus (3IT8) were superposed and are illustrated as ribbon
diagrams.

FIG. 21.17. Structure of an MHC-I/Peptide/T-Cell

Receptor (TCR) Complex and the TCR Footprint on the
MHC/Peptide. The structure of the H2-Kb/dEV8/2CTCR
complex (2TCR) is displayed. A: The complex complex.
B: A close-up of the MHC/peptide/TCR interface. C:
The surface of the MHC (magenta)/peptide (yellow)
complex with the complementarity determining region
loop of the TCR Vα (green) and Vβ (blue) shown and
labeled.
HLA-A2 complexed with the Tax peptide was studied with tion of the CDR loops of the TCR, particularly long CDR3
its cognate TCR derived from a cytolytic cell known as loops, is observed in the comparison of TCR free or bound
A6.447 These structures offered a consistent first glimpse at to their cognate MHC/peptide ligands.499,500 A striking ex-
the orientation of the TCR on the MHC/peptide complex, ample of this is illustrated by the structure of the KB5-
but additional structures, including that involving a mu- C20 TCR alone as compared to its complex with H2-K b /
rine MHC-II–restricted TCR suggest that more molecular peptide (Fig. 21.18).
variations may exist. As a canonical example of the MHC/
peptide/TCR complex, we include an illustration of the Structural Insights into Alloreactivity
H2-K b /dEV8/2CTCR complex499 (Fig. 21.17). This illustra- During T-cell development, TCRs destined to be useful to
tion shows that the complementarity determining regions the host are selected for weak reactivity with one or more
(CDRs) of the TCR (labeled 1α , 2α, 3α for CDRs of Vα and self-peptides complexed with cell surface host major histo-
1β, 2β, and 3β for the CDRs of Vβ, respectively) sit sym- compatibility (MHC/peptide) molecules. TCR selection is
metrically on the MHC/peptide complexes. For the H2-K b / customized in each individual because of extensive poly-
dEV8/2CTCR complex, the region contacted by the CDRs of morphism in MHC molecules that is designed to diversify
the Vα domain of the TCR lie to the left and that contacted peptide repertoire and optimize immune responsiveness.
by the CDRs of the Vβ domain lie to the right. The regions Thymic selection and MHC polymorphism conspire to gen-
contacted by CDR3, labeled 3α and 3β, are at the center of erate antimicrobial T-cell responses that are genetically re-
the bound peptide, whereas the regions contacted by CDR1 stricted to recognizing host MHC molecules while retaining
and CDR2 of both Vα and Vβ lie peripherally. Footprints of Ag-specificity. MHC restriction has been a central paradigm
other TCR have been reviewed elsewhere.357 of T-cell immunity and was the basis for the 1996 Nobel
With the publication of additional MHC/peptide/TCR Prize awarded to Peter Doherty and Rolf Zinkernagel.
structures, 357 additional points have emerged: 1) There is Unfortunately for transplant clinicians and their patients,
considerable variability in the orientation of the TCR Vα the rule of MHC restriction is violated when T cells are ex-
and Vβ domains with respect to the MHC/peptide composed to allogeneic MHC/peptide complexes. Remarkably,
plex. Although the fi rst MHC-II/peptide/TCR structure up to 10% of naive T cells react strongly against allogeneic
suggested that an orthogonal disposition, in which Vβ MHC/peptide in vitro (mixed lymphocyte reaction) and in
makes the great majority of contacts with the MHC-II α1 vivo leading to allograft rejection and GVHD. This reaction,
domain and the Vα predominantly interacts with the β1 known as T-cell alloreactivity, is why MHC molecules were
domain, might be the preference for MHC-II/TCR interac- initially called transplantation or histocompatibility mol-
tions,449 additional MHC-II/peptide/TCR structures 450,497 ecules and has puzzled immunologists for decades.
suggest that this disposition is not indicative of MHC-II as There are two main historical theories to explain the
compared to MHC-I but rather reveals the wide variety of high frequency of alloreactive T cells. The first, proposed in
possibilities. 2) Considerable plasticity in the conforma- 1977,183 postulated that a single allogeneic MHC molecule

This PDF was created and uploaded by

www.medicalstudyzone.com which is one the
biggest free resources platform for medical
students and healthcare professionals. You can
access all medical Video Lectures, Books in PDF
Format or kindle Edition, Paid Medical Apps and
Softwares, Qbanks, Audio Lectures And Much
More Absolutely for Free By visiting our Website
https://medicalstudyzone.com all stuff are free
with no cost at all.
Furthermore You can also request a specific

Book In PDF Format OR Medical Video Lectures.
FIG. 21.18. Complementarity Determining Region (CDR)3 Plasticity. Backbone tracings of the KB5-
C20 T-cell receptor CDR loops free (1KB5) or bound to the major histocompatibility/peptide complex
(1KJ2) are superposed. Unliganded forms are in darker colors. The largest conformational difference
is seen in the CDR3β loop.
could give rise to multiple binary complexes with cell sur- self-peptide, dEV8, and the H2-Ld–restricted nonamer, QL9.
face molecules, creating “neoantigenic determinants” recog- Given these differences, it was a fair bet that the cross-reac-
nized by clonally distinct T cells. This has been reinterpreted tivity of 2C on H2-K b-dEV8 and H2-Ld-QL9 would depend
in a “peptide-centric” hypothesis whereby as a single MHC upon plasticity in the CDR3 regions of the TCR as docu-
molecule presents disparate peptides recognized by multi- mented in comparisons of bound and free TCRs, including
ple different T-cell clones. The “multiple binary complex” 2C. Surprisingly, this was not the case as the TCR actually
model implies the TCR interacts with a set of amino acids adopted very similar conformations in the two structures.
shared by self- and allogeneic MHC molecules, so the cross- In contrast to the findings with the alloreactive murine, 2C
reaction depends crucially on the peptide antigen. However, TCR, alloreactivity of the natural human LC13 TCR is based
it was later suggested that alloreactive T cells might focus on on striking molecular mimicry between the cognate and
polymorphic residues exposed on the allogeneic MHC mol- alloantigens.503 The LC13 recognizes an Epstein-Barr viral
ecule itself, the so-called high determinant density model in peptide complexed with self-HLA-B*08:01 but also allore-
which the TCR focus is “MHC centric,” and the peptide is acts with B44 allotypes (HLA-B*44:02 and 44:05) bound to
largely irrelevant.501 an endogenous allopeptide from an adenosine triphosphate–
Colf et al.502 solved the structure of the 2C TCR in com- binding cassette protein ABCD3. HLA-B*08:01, and the
plex with its known allogeneic ligand, H2-Ld-QL9, and then closely related allotypes HLA-B*44:02 and HLA-B*44:05,
compared it to the structure of 2C in complex with its posi- are distinguished by numerous polymorphisms and the
tively selecting ligand, H2-K b-dEV8, and thus were able to LC13-reactive viral and endogenous peptides are unrelated
provide important insight into the structural basis of allo- in their sequence. Nonetheless, the LC13 TCR bound the al-
reactivity. These two TCR footprints were typical of known logeneic and cognate virus-specific HLA-peptide structures
MHC/peptide/TCR structures but differed from each other in a very similar manner.503 Moreover, there was mimicry of
in a number of ways. For instance, each chain of 2C made the viral peptide by the allopeptide. Hence, this study dem-
contacts with one MHC α helix of H2-Ld-QL9, whereas both onstrates that recognition of allogeneic and cognate HLA-
chains contacted both α helices in H2-K b-dEV8. The ge- peptide structures can occur via molecular mimicry, even in
ometry of 2C adopted a more perpendicular orientation on the face of polymorphism and disparate peptide ligands.503
H2-Ld-QL9, and there was a relative rotation of H2-K b-dEV8 Interestingly, another HLA-B*08:01–restricted TCR (CF34)
by 20 degrees. The alloreactive complex reveals that both with the identical Epstein-Barr virus–peptide specificity as
peptide-centric and MHC-centric interactions underpin LC13 lacks B44 reactivity because it arises when HLA-B44
direct T-cell allorecognition by the 2C receptor but with a is coinherited in trans with HLA-B8. Docking of the CF34
heavy emphasis on MHC-centric interactions. Most surpris- TCR focused on the N-terminus of the HLA-B*08:01-peptide
ing, however, was the small number of shared contacts be- while LC13 docked over the C-terminus of the viral peptide
tween the two structures, implying a limited role for mimicry bound by HLA-B*08:01.504 This corresponds with CF34 en-
between cognate and allogeneic MHC/peptide. The H2-K b gagement of a polymorphic region distinguishing HLA-B8
and H2-Ld have 31 amino acid differences, and there is no from HLA-B44, revealing a topographical image of shifting
sequence similarity between the H2-K b –restricted octamer TCR specificity to accommodate T cell self-tolerance.504

The role of molecular mimicry and the relative impor-

tance of peptide-centric versus MHC-centric bias in T-cell
allorecognition is likely to vary in different systems. Hence,
it is likely that the nature of the polymorphisms between
cognate and allogeneic MHC allotypes will affect TCR focus.
Thus, closely related MHC allotypes that differ by as little as
one amino acid (eg, H2-K b mutants in mice; HLA-A2, B44,
and B27 families in humans) are set up for MHC mimicry to
be a key component of T-cell allorecognition where specific-
ity is likely to be peptide centric. For example, HLA-B*44:02
and B*44:03 allotypes differ by only a single amino acid
and yet stimulate strong mutual allogeneic T-cell responses. FIG. 21.19. CD8aa and CD8ab Bind the Same Site on the MHC-I a3
Indeed, the potency of T-cell alloresponses between closely Domain. The MHC-I interactions with the CD8αα (PDB 1AKJ) ho-
related MHC allotypes probably occurs because positive modimer (A) and CD8αβ (PDB 3DMM) heterodimer (B) are shown.
selection of host T cells is purposely designed to create a
repertoire responsive to subtle changes in peptide display.
Therefore, closely related MHC allotypes, with differences
in both peptide repertoire and MHC/peptide conformation thelial lymphocytes in the gut, predominantly express the
of a shared repertoire, play straight into nature’s design for CD8αα homodimer. The CD8αβ heterodimer is considered
T-cell recognition. the relevant functional coreceptor on peripheral T cells.
Important questions concerning the nature of T-cell The three-dimensional structures of human and mouse
signaling mediated by engagement of the TCR by MHC/ MHC-I/CD8αα complexes have been described,508,509 and
peptide complexes remain. What are the molecular and bio- more recently, a complex of mouse MHC-I/CD8αβ has
physical determinants of agonist as compared with antago- been reported.58 These structures localize the binding site
nist activity of particular MHC/peptide ligands? Although it of the CD8 Ig-like αα homodimer or CD8αβ heterodimer
is clear that a number of parameters such as binding kinet- to a region on an exposed loop of the MHC-I α3 domain.
ics, half-life, and affinity (both two dimensional and three This interaction is illustrated in Figure 21.19, which shows
dimensional) as well as coreceptor contributions may play the flexible loop of residues 223 to 229 clamped into the CD8
a role in the outcome of TCR engagement by MHC/peptide combining site. This is one of relatively few examples of an
complexes, and the additional role of receptor clustering non-Ig, Ig-like molecule exploiting an Ig-like heterodimer
may contribute to such signaling, the precise contribution of interface to bind another ligand. The structural identifica-
each of these factors has not been established. A provocative tion of the CD8αα and CD8αβ on the MHC-I provides a
approach to the question of the role of the geometry of dock- context for understanding the contribution of CD8, which
ing of the TCR onto the MHC/peptide complex has been binds the kinase Lck through the cytoplasmic domain of its
addressed in the analysis of the relationship of stimulatory α chain to T-cell signaling.
parameters of four different peptides bound to H2-Ld with The T-cell coreceptor associated with cells restricted to
respect to the three-dimensional structure of the relevant MHC-II antigens, CD4, has also been the subject of detailed
MHC/peptide/TCR complex.505 One of the four peptides structural studies, in part, because of its role as a recep-
failed to induce signaling, and remarkably, the TCR docked tor for attachment and entry of HIV.510 The x-ray structure
onto the MHC/peptide in a unique docking geometry. The determination of the complete extracellular portion of the
authors interpreted these results to demonstrate the rela- molecule (domains D1 through D4) indicates a degree of
tionship between TCR/pMHC docking geometry, peptide segmental flexibility between domains D2 and D3, and both
cross-reactivity, and signaling as well as explaining the basis crystallographic and biochemical data suggest that dimer-
for germline bias in TCR/MHC interactions. ization of cell surface CD4 occurs.511 These results have been
interpreted to support a role for CD4-mediated MHC-II–
MHC/Coreceptor Complexes dependent dimerization in facilitating TCR dimeriza-
The major coreceptors for recognition by αβ TCRs are CD8, tion and signaling. Although a mixed species structure of
which interacts with MHC-I molecules, and CD4, which human CD4 D1-D2 domains with a mouse MHC-II mol-
interacts with MHC-II. Coreceptor function plays a role in ecule was reported at low resolution some years ago,63 only
signaling the T cell in addition to contributions the MHC/ recently, employing an affinity matured human CD4 mol-
coreceptor interaction may provide in increasing apparent ecule have diffraction quality crystals of human MHC-II/
avidity between the MHC/peptide and the TCR. CD8, the human CD4 D1-D2 molecules been obtained64 (Fig. 21.20).
coreceptor on MHC-I–restricted αβ T cells, exists as a cell This recently determined human MHC-II/hCD4 structure
surface homodimer of two α chains or a heterodimer of α is remarkably similar to the low-resolution mixed species
and β chains and plays an important role both in the ac- structure determined more than 10 years earlier. Both struc-
tivation of mature peripheral T cells as well as in the thy- tures reveal the focus of the CD4 interaction on a stretch
mic development of MHC-I–restricted lymphocytes.506,507 of amino acid residues on the Ig-like β2 domain of the
Mature peripheral CD8 T cells express significant amounts MHC-II molecule. Earlier mutagenesis and functional stud-
of CD8αβ, whereas other lymphocytes, such as intraepi- ies of CD4-dependent MHC-II–mediated T-cell activation

human NK receptor/ligand interactions to the NKG2/MIC

and ULBP complexes and to the KIR2D and KIR3D/MHC
interactions.
The activating receptor, NKG2D, interacts with MICA, a
stress-induced MHC-I–like molecule, allowing NK cells to
eliminate virus-infected or tumor cells that are marked by
the surface expression of MICA. The structure of the human
NKG2D/MICA complex reveals that the homodimeric bind-
ing site of NKG2D recognizes the surface α-helical super-
domain, in an orientation similar to that by which TCR
see MHC-I molecules456 (Fig. 21.21). This structure shows
remarkable similarity to that of the hNKG2D bound to
another MHC-Ib protein, ULBP3.515 Similarly, the mouse
NKG2D interacts, in a similar orientation, with the murine
stress-induced molecule, RAE-1β.459
The short KIRs (eg, KIR3DS and KIR2DS molecules) are
considered activating receptors because they have the poten-
tial to interact with the DAP12 signal-transducing molecule,
and the long KIRs (eg, KIR3DL and KIR2DL) are inhibitory
because they have cytoplasmic domains that contain im-
FIG. 21.20. CD4 Interacts Directly with the β2 Domain of MHC-II. A munoreceptor tyrosine-based inhibitory motifs. KIR2DL1
ribbon illustration of a complex of human CD4 with HLA-DR1 (PDB and KIR2DL2 have been studied extensively. They interact
3S5L) is shown. The HLA-DRα chain is in firebrick, HLA-DRβ chain
in blue, peptide in green, and CD4 in magenta.
with the human MHC-I molecules HLA-C*04 and HLA-
C*03, respectively, and show some preferences for MHC-I
molecules complexed with particular peptides. Among the
polymorphic amino acid residues that distinguish HLA-
also implicated a region on the α2 domain of MHC-II.61,512 C*03 and HLA-C*04 are Asn80 of -C*03 and Lys80 of
Although the α2 domain residues do not directly interact -C*04. Thus, it was of interest when the structures of HLA-
with CD4 in these structures, it is conceivable that they play C*03/KIR2DL2104 and HLA-C*04/KIR2DL1 complexes were
some indirect role in MHC multimerization that may con- reported.105 The KIR2DL2/HLA-C*03 interaction is illus-
tribute to signaling. trated in Figure 21.22. The important conclusions from this
structure and those of similar KIR2DL complexes is that
MHC/Natural Killer Receptor Complexes the recognition of MHC-I is via amino acid residues of the
NK cells provide a first line of defense against virus-in-
fected and host cells, and their recognition of such targets
is dependent on a balance of NK activating and inhibitory
receptors.513,514 In addition to their expression on NK cells,
some of the NK receptors are also expressed on subsets
of T cells and other hematopoietic cells. In general, each
of these classes of NK receptors also falls into two differ-
ent structural groups: the Ig-like receptors and the C-type
lectin–like receptors. Because of their functional interac-
tions with MHC-I and MHC-I–like molecules, and because
several different systems have evolved to recognize MHC-I
molecules differently, it is worthwhile to examine the re-
cently determined structures of several MHC-I/NK recep-
tor complexes.
In the human, the major NK receptors fall into several
categories: those of the NKG2 family, the NKp molecules,
and the KIRs. The NKG2 family molecules are C-type
lectin–like molecules that recognize ligands grouped
broadly as the NKG2D ligands, which include the MHC-
like molecules, MICA and MICB, as well as the ULBP fam-
ily of stress-induced molecules. NKp molecules include
NKp30, NKp44, and NKp46, all members of the Ig super-
FIG. 21.21. The NKG2D sees the MHC-Ib MICA a-Helical Superdomain.
family. The KIR molecules, also Ig superfamily members, The structure of the human NKG2D/MICA complex (PDB 1HYR) re-
bind either HLA-C, or for the KIR3 subfamily, HLA-B mol- veals the homodimer NKG2D sitting asymmetrically astride the MICA
ecules. Because the focus of this chapter is members of the α-helical superdomain. The NKG2D domains are in cyan and MICA
MHC protein family, we will restrict our discussion of the in magenta.

makes no direct contact with residues of the MHC-bound

peptide; and 3) in the x-ray structure, there are two poten-
tial sites for Ly49A interaction with the MHC molecule: site
1 at the end of the α1 and α2 helices, and site 2, an extensive
region making contact with the floor of the peptide bind-
KIR3DL1*001 ing groove, the α3 domain of the MHC-I molecule, and the
β2m domain as well. The ambiguity suggested by the x-ray
structure has been resolved by extensive mutagenesis studies
that are consistent with the view that site 2 is functionally
significant.516,517 Recently, several other NK receptors have
been studied structurally. These include the Ly49I inhibi-
tory receptor, which interacts functionally with H2-K b. The
Ly49I has been crystallized without a ligand, revealing a
basic fold similar to that of Ly49A but with a somewhat dif-
ferent dimeric arrangement.518 The Ly49C, another murine
NK inhibitory receptor, has been examined in complex with
its H2-K b ligand.519 Differences between H2-K b /Ly49C and
H2-Dd /Ly49A suggest different modes of NK receptor bind-
FIG. 21.22. KIR2DL2 and KIR3DL1 Interact Similarly with Their ing to MHC-I depending on whether the interaction is “cis”
MHC-I/p Ligands. KIR2DL2 complexed with HLA-C*03 (PDB 1EFX) (ie, between the NK receptor and the MHC-I on the same
(A,C) and KIR3DL1 complexed with HLA-B*57:01 (PDB 3VH8) (B,D) NK cell) or “trans” (between the NK receptor on the NK cell
are shown to illustrate the overall interactions (A,B) and close-ups and the MHC-I on its target).520
of the contact with bound peptide (C,D).
MHC-II Superantigen Complexes
Superantigens are molecules, frequently toxic products of
elbow bend joining the two Ig-like domains of the KIR, and bacteria, which bind MHC molecules on the cell surface
that residues that vary among different KIRs determine the and are then presented to a large subset of T cells, usually
molecular specificity of the interaction with the particular defined by the expression of a particular family of TCR V
allelic product of HLA-C. In addition, the interaction of the regions.521 Most of the known superantigens bind MHC-II
KIR with the HLA-C is also modulated by the particular molecules, although one, the agglutinin from Urtica dioica,
bound peptide, explaining the results of binding/peptide the stinging nettle, can be bound by both MHC-I and
specificity studies. Structural understanding of the KIR3 MHC-II molecules and presented to T cells of the Vβ8.3
molecules, which have three extracellular Ig-like domains, family.522 Its structure in complex with carbohydrate ligand
eluded the field for many years. Recently, a molecular com- has been reported.523,524 MHC-II interactions with superan-
plex of HLA-B*57:01 with KIR3DL1 has been reported (see tigens, such as those derived from pathogenic bacteria, are
Fig. 21.22). This structure reveals the contribution of the D0 the first step in the presentation of the multivalent array
amino terminal domain of KIR3DL1 in recognizing a region of the APC-bound superantigen to T cells bearing recep-
of the MHC-I with limited polymorphism, the D1 domain tors of the family or class that can bind the superantigen.521
accommodates a region with sequence variation, and the D2 A number of structural studies examining the interaction
domain shows high complementarity.
The mouse exploits a set of NK receptors of a different
structural family to provide the same function. In particu-
lar, the predominant, and best studied, mouse NK recep-
tors are those of the Ly49 family. Functional experiments
had demonstrated that Ly49A interacts with the MHC-I
molecule, H2-Dd, and also that for appropriate interaction,
H2-Dd needs to be complexed with a peptide. In contrast to
human NK recognition, however, surveys of H2-Dd–binding
peptides revealed little if any peptide preference or specific-
ity. This would explain the function of the NK inhibitory
receptor in that they are at baseline chronically stimulated
by normal MHC-I on somatic cells, turning off the NK cell.
When MHC-I is dysregulated by tumorigenesis or by patho-
genic infection, the lower level of surface MHC-I diminishes
the NKIR-mediated signal, and the NK cell is activated. The
structure of the mouse Ly49A inhibitory receptor in com- FIG. 21.23. Structure of the H2-Dd/Ly49A Complex Reveals an
plex with its MHC-I ligand, H2-Dd (Fig. 21.23), reveals sev- Alternate Site for MHC/Natural Killer Receptor Interaction. The
eral crucial features of the interaction: 1) the Ly49A C-type structure of the complex (PDB 1QO3) is shown, illustrating both
lectin–like molecule is a homodimer; 2) the Ly49A molecule potential sites of interaction.

of superantigen both with their MHC-II ligands and with Multivalent MHC/Peptide Complexes
TCR have been reported.525 Structural analysis of crystals A major development in the past several years has been the
derived from staphylococcal enterotoxin B complexed with engineering and application of multivalent MHC/peptide
HLA-DR1526 and from toxic shock syndrome toxin-1 com- complexes for the identification, quantification, purifica-
plexed with HLA-DR1527 revealed that the two toxins bind tion, and functional modulation of T cells with particular
to an overlapping site, primarily on the MHC-II α chain, MHC or MHC/peptide specificity. Two general approaches
and indicated that the staphylococcal enterotoxin B site have been exploited: one based on the enzymatic bioti-
would not be expected to be influenced by the specific pep- nylation of soluble MHC/peptide molecules generated in
tide bound by the MHC, although the toxic shock syndrome bacterial expression systems that are then multimerized
toxin-1 site would. The view that superantigens exert their by binding of the biotinylated molecules to the tetravalent
biologic effects by interaction with conserved regions of streptavidin,547 and another based on the engineering of
the MHC-II molecule as well as with conserved regions of dimeric MHC/Ig fusion proteins.548 These reagents can be
the TCR has been challenged by the determination of two used in flow cytometric assays that permit the direct enu-
structures, that of S. aureus enterotoxin H complexed with meration of MHC/peptide-specific T cells taken directly ex
HLA-DR1528 and that of streptococcal pyrogenic toxin C vivo. In either of these methods, multivalent MHC/peptide
in complex with HLA-DR2a.529 Both of these studies indi- complexes are generated, and the relatively weak intrinsic
cate that these superantigens can interact with the MHC-II affinity of the MHC/peptide complex for its cognate TCR is
β chain through a zinc-dependent site that includes supe- effectively magnified by the gain in avidity obtained by the
rantigen contacts to bound peptide. Our understanding of increase in valency. For MHC-I molecules, the technology
superantigen interactions has been augmented by the de- has been so reliable in producing multivalent (tetrameric)
termination of ternary structures of MHC-II/superantigen/ molecules loaded homogeneously with synthetic peptides
TCR complexes for two examples, that of the Mycoplasma that a wide variety of specific MHC-I/peptide multimers
arthritidis mitogen in complex with HLA-DR1 and TCR,530 are available either from a resource facility sponsored by the
and that of Staphylococcus aureus enterotoxin H complexed National Institute of Allergy and Infectious Diseases (www
with HLA-DR1 and TCR.531 In both of these cases, the supe- .niaid.nih.gov/reposit/tetramer/index.html) or from com-
rantigen interacts with the MHC-II and with both TCR Vα mercial suppliers that offer either the tetramer, Ig multimer,
and Vβ domains. or a pentamer preparation. The MHC-I/peptide multimers
have also been exploited for identification of specific popula-
MOLECULAR INTERACTIONS OF MHC MOLECULES tions of NK cells and for assignment of various NK-receptor
specificities.549–552 For some MHC-II molecules, similar suc-
Assays for Molecular Interactions cess has been achieved in the production of such multimers,
Whereas the crystal structures provide a vivid static illus- using insect cell or mammalian cell expression systems for
tration of the interactions of MHC molecules with their molecules produced by either the tetramer or Ig chimera
peptide, FcRn, CD8, CD4, superantigen, NK receptor, and strategy.553–557 MHC-II/peptide tetramers can be effectively
TCR ligands, the dynamic aspects of these binding steps can used for a variety of CD4 T-cell applications, but some TCRs
be approached by a variety of biophysical methods.532 It is (particularly those involved in autoimmune diseases) have
important to note that affinities and kinetics of interaction an affinity for their cognate MHC-II/peptide that is be-
of MHC/peptide complexes for TCR have been determined neath the threshold for detection.545 Also, the potential for
by several methods in a variety of systems.533–540 In addi- the same peptide binding to one MHC-II in multiple frames
tion, MHC interactions with NK receptors101,102,541,542 have generates difficulties in the engineering and preparation of
been quantified by similar techniques. Although there are effective MHC-II tetramers.406
clear differences in the affinity and kinetics of binding of
different TCR and NK receptors for their respective cognate
MHC/peptide complexes, the generally consistent findings CONCLUSION
are that the affinities are low to moderate (ie, Kd = 5 × 10−5 We have surveyed the Mhc as a genetic region and a source
to 10−7 M) and are characterized by relatively rapid disso- for molecules crucial to immune regulation and immuno-
ciation rates (ie, kd = 10−1 to 10−3 sec−1). Recently, attention logic disease. These genes reflect the panoply of mecha-
has been drawn to the differences in measuring interactions nisms involved in the evolution of complex systems and
with one component in solution (three-dimensional affi n- encode cell surface proteins that interact via complex or-
ity) as compared with measurements made between multi- chestration with small molecules including peptides and
valent displays of the interactions on two surfaces, such as glycolipids as well as with macromolecular receptors on T
might mimic the apposition of two cells. Methods exploit- cells and NK cells. The MHC-I molecules provide the im-
ing microscopic techniques543,544 or a micropipette adhesion mune system with a window for viewing the biologic health
frequency assay545,546 suggest that the biologic readout of of the cell in which they are expressed, and MHC-II mol-
T-cell activation may be more readily correlated with two- ecules function as scavengers to taste and display the rem-
dimensional rather than three-dimensional measurements nants of the cellular environment. Viruses and bacterial
of MHC/TCR affinities. pathogens contribute enormously to the genetic dance—

they modulate and compete in the control of MHC expres- prevention, diagnosis, and treatment of immunologic and
sion, sometimes exploiting MHC mimics that they have infectious diseases.
acquired—and the host by adjusting its T-cell and NK-cell
repertoire on the time scale of both the individual organ-
ism and the species, resisting the push to extinction. The ACKNOWLEDGMENTS
immune system, dynamically, resourcefully, and creatively We openly accept the responsibility for errors of fact, in-
provides, through the concerted action of its MHC mole- terpretation, or citation and ask our colleagues to forgive
cules, TCRs, NK receptors, and antibodies as well as a host any failure to appropriately quote their work. We thank the
of other regulatory molecules, an organ system vital not members of our laboratories and our families for their pa-
only to the survival of the individual but also to the suc- tience and understanding. Part of this work was supported
cess of the species. As we understand better the molecular by the intramural research program of the U.S. National
functions of the MHC, we should better understand ratio- Institute of Allergy and Infectious Diseases, National
nal approaches to manipulating the immune system in the Institutes of Health.

CHAPTER
22 Antigen Processing and Presentation
Ted H. Hansen • Paul A. Roche
ANTIGEN PRESENTATION PATHWAYS are activated for detection and elimination of pathogen-
As part of the adaptive immune response, T cells mount infected cells or tumors.
immune responses to diseased cells and abnormal cells.
Remarkably, the mechanism by which T cells discrimi-
ANTIGEN PROCESSING AND PRESENTATION BY
nate self from non-self is based on pathogen- or tumor-
MAJOR HISTOCOMPATIBILITY CLASS I (MHC-I)
derived proteins displayed on the cell surface by self–major
histocompatibility complex (MHC) molecules. More
PROTEINS
specifically, when T cells detect diseased cells displaying Origin of MHC-I binding peptides
MHC molecules loaded with peptides derived from for- As detailed in Chapter 21, MHC-I proteins preferentially bind
eign peptides, effector mechanisms are initiated to elimi- peptides of 8 to 10 amino acid lengths, and which peptides
nate the diseased cell. By contrast, T cells that detect cells bind is determined by MHC-I allele-specific polymorphic res-
displaying MHC molecules loaded with peptides derived idues that form the architecture of its peptide binding groove.
from self-proteins are either eliminated during ontogeny Each MHC-I allele typically binds peptides with a consensus
by negative selection or suppressed by peripheral tolerance binding motif requiring a relatively specific amino acid cen-
mechanisms. Thus, at the molecular level, the T-cell recep- tral anchor and a C-terminal hydrophobic amino acid. Given
tors on T cells discriminates MHC loaded with self- versus the lack of restraints in most peptide positions and the large
non–self-peptides displayed at the cell surface of normal number of proteins synthesized by each cell relative to the
and diseased cells. The mechanisms by which peptide li- number of MHC molecules, the selective processes control-
gands for MHC molecules are generated is referred to as ling which peptides get presented is of considerable impor-
antigen processing, whereas the mechanism by which pep- tance for the detection of pathogens and malignancies.
tide/MHC complexes are displayed at the cell surface is The source of the peptides that get presented during infec-
called antigen presentation. tion has been a question of considerable investigation. It has
To assure the appropriate effector mechanism is gener- been known for some time that MHC-I–binding peptides are
ated to detect cells infected with pathogens in the cytosol derived from membrane-bound, secreted, cytosolic, and nu-
or endocytic compartments, the antigen presentation path- clear proteins that are almost exclusively generated by the pro-
ways for classical cluster of differentiation (CD)8 + versus teasome in the cytosol. However, the uncertainty of peptide
CD4 + T cells are different. More specifically, antigen de- source is based on two apparently incongruent observations.
rived from intracellular pathogens is typically bound to Most proteins from which MHC-I–binding peptides are de-
MHC-I proteins that are uniquely detected by CD8 T cells, rived have slow rates of turnover in cells ranging from hours
whereas antigen derived from extracellular pathogens is to days; viral-derived peptides are presented by infected cells
typically bound to MHC-II proteins that are uniquely to CD8 T cells within a few minutes. This disparity between
detected by CD4 T cells. To preferentially bind peptides the degradation versus presentation kinetics is potentially ex-
derived from proteins synthesized inside the cell, the pro- plained by cumulative observations that a fraction of newly
cessing of peptides that bind MHC-I proteins occurs in the synthesized polypeptides are rapidly degraded, providing a
cytosol and the loading of these peptides occurs in the en- peptide reservoir sufficient to support a robust CD8 T-cell re-
doplasmic reticulum (ER). This pathway is commonly response. Supporting evidence for this conclusion comes from
ferred to as the classical MHC-I antigen-binding pathway. studies that block protein synthesis with cycloheximide and
By contrast, to preferentially bind peptides from proteins detected a rapid decrease in the peptide supply for MHC-I
synthesized outside the cell, the processing and loading of binding.1,2 One of the more elegant ways to monitor peptide
antigenic peptide ligands for MHC-II proteins occurs in transport into the ER was by transporter associated with an-
an endocytic pathway terminating in the lysosomes. This tigen presentation (TAP) mobility in the ER membrane by
pathway is typically referred to the classical MHC-II anti- photobleaching; establishing TAP mobility was proportional
gen-binding pathway. The differences between the MHC-I to the peptides in the cytosol. This assay was used to show a
and MHC-II antigen presentation pathways are determined 90% reduction in peptide supply after 30 m treatment with
by their differential interaction with molecular chaper- cycloheximide, thus kinetically linking presentation closely to
ones. These same molecular chaperones enforce the quality translation.3 Indeed, there are now several reports that within
control of antigen presentation to assure appropriate T cells 30 minutes after synthesis, antigenic peptides are presented at
524

CHAPTER 22 ANTIGEN PROCESSING AND PRESENTATION | 525
the cell surface at levels that activate T cells.4,5 What remained defective translation products and/or a stochastic event of
unclear from these studies is the biochemical and mechanistic normal protein translation remains to be determined.17
basis for the generation of these rapidly transcribed proteins
from which MHC-I–binding peptides are derived.
In a model that has gained wide acceptance, Yewdell and Peptide Trimming in the Cytosol
colleagues proposed that a rapid supply of MHC-I ligands is Degradation of cellular and antigenic peptides bound to
derived from aberrant protein production they termed defec- MHC-I is largely mediated in the cytosol by the protea-
tive ribosomal products or DriPs.6,7 In their model, it was spec- some. The proteasome is responsible for the degradation
ulated that DriPs could be derived from unfolded or misfolded of the majority of cytosolic and nuclear proteins, and in
proteins of the proper sequence and length, proteins with most cases proteasome targeted proteins are ubiquitinated.
errors in sequence or posttranslational modifications, prema- Ubiquitin (Ub) is typically coupled to internal lysine resi-
turely terminated proteins, and proteins translated from the dues of proteins substrates, but coupling can also occur at
wrong start codon.8 Regarding aberrant translations, there is the N terminus or on internal cysteine, serine, or threonine
evidence that MHC-presented peptides can arise from alter- residues.18–21 In this orchestrated process, activated Ub is
native reading frames, generating what has been called cryptic transferred from one predominant Ub-activating enzyme
epitopes.9 For example, Schwab et al. showed that inhibition (E1) to one of 30 to 40 mammalian Ub conjugating enzymes
of the eukaryotic translational initiation factor eIF2α resulted (E2s) and then to the substrate that is bound by one of hun-
in the synthesis of cryptic peptides and intriguingly, viruses dreds of different ubiquitin protein ligases (E3s). The E3s
inhibit host translation by inactivating eIF2α.10,11 However, are the major determinant of substrate specificity. This pro-
most MHC-I–presented peptides have native amino acid se- cess can be repeated to form polyUb chains whereupon the
quence and are derived from standard messenger ribonucleic next Ub moiety is added to one of the seven internal lysine
acid (mRNA) translation products, suggesting cryptic epit- residues of Ub. Proteins coupled with polyUb chains of four
opes represent only a minor component of MHC-presented or more Ub moieties linked through Lys48 Ub residues are
peptides. In addition, the contribution of peptides derived the prototypic signal for proteasome mediated degradation.
from misfolded proteins in the secretory pathway is also likely The cylindrical 20S catalytic particle (CP) of the protea-
modest. Although antigenic peptides from misfolded tyrosi- some is formed by four stacked rings of seven subunits each.
nase were found to be preferentially presented,12 other studies The inner two rings of the 20S proteasome are assembled
show that several mutations causing misfolding do not lead from the beta subunits, three of which have catalytic activ-
to enhanced antigen presentation. Furthermore, ER quality ity with chymotryptic, tryptic, or caspase activity. Thus each
control is typically too slow to account for the rapid supply of proteasome has six active sites with the ability to cleave after
MHC-I–binding peptides.13 The reason for this kinetic delay most types of peptide bonds, although to differing efficien-
is that substrates for ER-associated degradation (ERAD) must cies based on flanking residues. The catalytic sites are exposed
be 1) translocated into the ER, 2) detected presumably by ER to the interior of the central chamber of the 20S cylinder. The
chaperones as being misfolded, 3) ubiquitinated and extracted two outer rings of the 20S proteasome are assembled from
from the ER, and 4) degraded by the proteasome in the cy- alpha subunits forming a pore of 13 Angstrom, mandating
tosol. Thus the predominance of MHC-I–presented peptides protein substrates be partially denatured prior to entering
likely does not come from cryptic epitopes or misfolded ER the interior of the chamber. The gate of the 20S CP is nor-
proteins identified by ER quality control. mally closed, and access to the interior of the 20S proteasome
To obviate trafficking into the ER, misfolded secretory is controlled by the complex of proteins termed the 19S regu-
proteins could be degraded rapidly based on degron detec- latory particle that binds to each end of the 20S cylinder. The
tion such as unshielded hydrophobicity.14 Alternatively, early full assemblage of 19S-20S-19S complex constitutes the 26S
in translation a lack of signal recognition particle engage- proteasome. The lid of the 19S regulator particle binds and
ment could lead to the mistargeting of secretory proteins deubiquitinates protein substrates before they gain entry in
to the cytosol. In support of this model, Schlosser et al.15 the 20S CP; the base of the 19S regulatory particle contains
showed that an epitope encompassing the signal peptidase ATPases that promote substrate unfolding and open the gate
cleavage site was efficiently presented by MHC-I proteins. of the 20S CP to provide access of denatured protein sub-
To rapidly generate peptides, it has been speculated that cells strates to the catalytic activity of the inner chamber.
may have select ribosomes called “immunoribosomes.”16 The evolutionarily conserved function of the proteasome
Such immunoribosomes might be more adept at generating is for recycling amino acids and ubiquitin moieties. However,
peptides with TAP access perhaps by compartmentalization in mammals, modifications to the proteasome have been
and/or targeting proteins to the 20S proteasome for immedi- made to promote antigen presentation. In response to inter-
ate destruction. In any case, there is considerable evidence feron (IFN)γ stimulation that occurs during inflammation,
that antigenic peptides can be rapidly presented by MHC-I three catalytic subunits of the constitutively expressed prote-
within minutes after translation of the protein source. The asome are replaced by IFNγ-induced homologs forming what
physiologic significance of this rapid kinetics of antigen pre- is called the immunoproteasome. More specifically, subunits
sentation is that it allows CD8 T cells to kill virus-infected β1, β2, and β5 of the constitutive proteasome are replaced
cells before viral replication is completed and progeny are re- by subunits β1i (LMP2), β2i (MECL-1), and β5i (LMP7) of
leased. However, whether the production of peptides bound the immunoproteasome. Suggesting an immunologic func-
to MHC class I molecules is a deliberate process for degrading tion, LMP2 and 5 are genetically encoded within a central

region of the mouse and human MHCs juxtaposed to genes complex of TAP1 and TAP2 subunits both encoded within the
encoding the TAP heterodimer. Consistent with their im- central region of the MHC of mouse and human. The translo-
munologic relevance, the immunoproteasomes display en- cation pore of TAP is formed by six transmembrane domains
hanced cleavage of protein substrates after hydrophobic and of each subunit, whereas the remaining transmembrane do-
basic amino acids, thus generating peptides with C-terminal mains (four for TAP1 and three for TAP2) are involved in
residues preferred for binding to many MHC-I alleles. IFNγ interaction with peptide loading complex (PLC).28 Awaiting
stimulation also induces the expression of PA28α and PA28β transport, both the N- and C-termini of peptides are bound
that form the 11S regulatory particle, which can also bind to to TAP29 ; peptides with higher affinity for binding TAP have a
either end of the 20S CP. Interestingly, the PA28 regulatory greater likelihood of transport, MHC binding, and presenta-
particle does not contain ATPases, but it does induce confor- tion to T cells during infection.30,31 TAP typically transports
mational changes to open the 20S CP gate. Thus, PA28 may peptides of 8 to 16 residues, although longer peptides are less
function to prevent the overdigestion of short peptides. This efficiently transported. As discussed in the next section, pep-
is an important issue, as it has been estimated that a nona- tides with N-terminal extensions are trimmed in the ER.
mer peptide in the cytosol has a half-life of only 7 seconds.22
However, studies testing PA28α /β-deficient mice demon-
strated that PA28α /β is not required for overall antigen pre- Peptide Trimming in the Endoplasmic Reticulum
sentation during virus infection, but may affect degradation As noted previously, TAP transports many peptides with
of select substrates.23 In any case, the inclusion of the IFNγ- C-terminal hydrophobic residues required for MHC binding,
inducible β subunits into the 20S CP of immunoproteasome but with N-terminal extensions, which need to be cleaved to
enhances antigen presentation by generating peptides with conform to the MHC-binding motif of the expressed alleles.
preferred C-terminal anchor residues, and in some same It is now clear that N-terminal trimming within the ER is
cases inclusion of the PA28 regulatory particle may prevent carried out by the ER-associated aminopeptidase (ERAAP;
overdegradation of immunologically relevant proteins. ERAP1 in mice and ERAP1 in humans). Interestingly, unlike
The proteasome generates the final C-terminal cleavage of other aminopeptidases, ERAP1 preferentially cleaved pep-
most MHC-I–binding peptides. A recently reported excep- tides of 9 to 16 amino acids thus matching the same peptide
tion to this is the generation of the tumor antigen MAGE-A3 length preference as transported by TAP.32 More specifically,
by insulin-degrading enzyme.23 Whereas insulin degrading ERAP1 was found to degrade a model 13-mer to a 9-mer
enzyme is solely responsible for the generation of the human and then stop, an apparent adaptation to maximize optimal
leukocyte antigen (HLA)-A1–restricted MAGE-A3 epitope, MHC-I binding. The length preference of ERAP1 could be
the HLA-B40–restricted MAGE-A3 epitope was proteasome explained by extended peptides binding to MHC-I before
dependent. Although there are several N-terminal proteases N-terminal trimming. However, this model is not supported
in the cytosol, they have only a limited role in generation of by structural analyses of ERAP1 and the location of its ac-
the MHC-I–binding peptides. More specifically, tripeptidyl tive site.33 Alternatively, recent analyses support a “molecular
peptidase II, leucine amino peptidase, bleomycin hydro- ruler” model. In this model, a hydrophobic pocket of ERAP1
lase, and puromycin-sensitive aminopeptidase have all been binds the C-terminal peptide residue thereby positioning the
found to affect generation of select MHC-I–binding epitopes, peptide so that N-terminal cleavage occurs about nine amino
but these proteases have an overall negative effect on antigen acids away.32,33 Thus, ERAAP/ERAP1 has unique structural
presentation by MHC-I. Interestingly, the metalloprotease features explaining its ability to trim extended peptide pre-
nardilysin that cleaves substrates on the N-terminus of ar- cursors while sparing ones of optimal length for MHC-I
ginine residues in dibasic pairs was implicated in three cyto- binding. As further evidence of their specialized functions
toxic T lymphocyte epitopes and may be more generalizable in antigen presentation by MHC-I proteins, ERAP1 pref-
given the preference for basic N-terminal resides for MHC-I erentially binds and processes peptides with hydrophobic
binding.24 However, as noted in the following, significant N C-terminal residues consistent with the majority of the TAP-
terminal trimming of MHC-I–binding peptide occurs after transported MHC-I–binding peptides. Interestingly, hu-
transport from the cytosol into the ER. mans have a second ER-associated peptidase with homology
to ERAP1 designated ERAP2.34 ERAP2 preferentially binds
peptides with C-terminal basic residues and thus may func-
Peptide Transport into the Endoplasmic Reticulum tion in human cells to provide peptides for HLA-A3, -Aw68,
Landmark studies of the mutagenized and immunoselected and -B27 alleles that bind such peptides. Additionally, in
mouse cell line RMA-S resulted in the discovery of how pep- humans the combined activity of both ERAP1 and ERAP2
tides are transported from the cytosol into the ER lumen.25–27 (most likely as heterdimers) is required for the presentation
The low level of surface expression of MHC-I molecules by of certain epitopes.35 By contrast, in mice ERAAP/ERAP1 is
RMA-S that could be rescued by either low temperatures sufficient to service their MHC-I alleles, all of which bind
or culturing with known MHC-I–binding peptides was peptides with hydrophobic C-termini. As additional evi-
attributed to lack of peptide transport by TAP. TAP is a mem- dence for their tailoring for MHC-I–antigen presentation,
ber of the adenosine triphosphate (ATP) binding cassette the expression and trimming activity of ERAP1 and ERAP2
family. Mechanistically, hydrolysis of ATP is required both for are upregulated upon IFNγ stimulation.
peptide binding to the cytosolic face of TAP as well as peptide The impact of ERAP on specific epitopes and the overall
transport into the ER. Structurally, TAP is a heterodimeric MHC-binding peptide repertoire has been analyzed using

cell lines36,37 and ERAAP-knockout mice.38,39 These studies found to be susceptible to Toxoplasma gondii infection re-
have shown that cytotoxic T lymphocyte detection of some sulting from the fact that the immunodominant epitope
but not all MHC-I/peptide complexes are ERAP1-dependent. presented to CD8 T cells is Ld restricted.40 Interestingly the
For example, in vitro ribonucleic acid interference experi- ERAAP-deficient mice were also found to present several
ments suggested that ERAP1 is involved in formation of one- unique peptides that elicited potent CD8 T-cell responses
third of peptide/MHC-I complexes.36,37 ERAAP-deficient demonstrating ERAAP functions, as a peptide editor alter-
mice had a 20% lower expression of K b, Db, Kd, and Dd alleles ing the repertoire of peptides presented.38
at the cell surface but a 70% lower expression of Ld. Of these
mouse alleles, only Ld preferentially binds peptides with the
motif of Pro in the second position, although about 20% of Chaperone-assisted Peptide Loading in the
human HLA alleles also prefer peptides with a Pro in the Endoplasmic Reticulum
second position. Notably, peptides with Pro in the second Full assembly of the MHC-I heavy chains (HCs) with β2m
position are only transported by TAP with N-terminal exten- and peptide within the ER is orchestrated by molecular
sions and are thus dependent upon ERAP1 for their genera- chaperones that keep folding intermediates in a conforma-
tion. In vivo relevance of this conclusion was demonstrated tion capable of attaining full assembly (Fig. 22.1). Nascent
by the observation that ERAAP-deficient BALB/c mice were HCs are transiently associated with the membrane-anchored
Golgi 4
Apparatus
ERAAP
Endoplasmic
Reticulum CRT
Erp57 Tpn
β2m 3 TAP1
TAP2
CXN 2
HC
1
Proteasome
FIG. 22.1. Schematic Representation of the Sequential Events of Antigen Processing and Presentation by Major Histocompatibility
Complex (MHC)-I Proteins. 1: Peptides are processed by the proteasome and translocated into the ER by TAP1/TAP2 heterodimers.
2: MHC-I heavy chains (HCs) awaiting assembly with β2m are bound to calnexin (CXN)/ERp57 complexes. 3: After assembly with β2m,
HC/β2m heterodimers enter the peptide loading complex wherein tapasin (Tpn) bridges HC with transporter associated with antigen pre-
sentation (TAP)/TAP2, and calreticulin (CRT)/ERp57 complexes bridge HC with Tpn. 4: Following the binding of a suitable peptide, which
may require final N-terminal trimming by ERAAP, fully assembled MHC-I proteins (HC/β2m/peptide) traffic to the plasma membrane via the
secretory pathway.

lectin calnexin (CXN) that, like its soluble paralog calretic- MHC-I then promotes peptide exchange until a peptide
ulin (CRT), functions as a general chaperone for assembly of suitable affi nity binds to complete the folding of the li-
of oligomeric glycoproteins. Quality control of glycoprotein gand binding groove. This peptide-induced folding then in-
assembly by CXN and CRT is carried out by monitoring duces the release of fully assembled MHC-I from the PLC.
terminal glucose residues on their N-linked glycans. More Logistically, how peptide editing by Tpn might occur was
specifically, the folding sensor UDP-glucose glycoprotein provided by mutagenesis studies of the MHC-I/Tpn inter-
glucotransferase allows incompletely assembled substrates action site. Two sites on the HC are critical for Tpn interac-
to reassociate with CXN or CRT, and cycles of glucose ad- tion: one site is in the α3 domain (residues 227 and 229)
dition and removal continue until substrates are either cor- and the other in the α2 domain (residues 128-136)43,58,59
rectly folded and assembled or targeted for ERAD. After the (Fig. 22.2A). Based on the location of the latter site and mo-
MHC-I HC assembles with β2m, the HC/ β2m heterodimer lecular dynamic modeling, it has been theorized that the
exchanges CXN for CRT and enters the PLC consisting of α2 interaction site might function as a folding sensor for
CRT, ERp57, tapasin (Tpn), and TAP.41–44 Optimal antigen C-terminal peptide anchoring in the MHC-I F pocket.60,61
presentation by MHC-I is dependent upon all four compo- Consistent with this model, polymorphisms within the F
nents of the PLC (Tpn, TAP, CRT, and ERp57) that function pocket can have a profound effect of Tpn dependencies of
to improve peptide loading in MHC-I proteins and their different MHC-I alleles.62 Based on mutagenesis and struc-
surface display.45,46 tural analyses, a credible model was constructed by Dong et
Knockout cell lines, genetically deficient mouse strains, al. of the MHC-I/PLC interactions52 (see Fig. 22.2).
and mutagenesis studies have provided key insights in the Although Tpn within the PLC is thought to enforce most
selective roles of PLC components in MHC-I assembly. CRT of the quality control of antigen presentation by MHC-I,
association with MHC-I proteins is dependent upon the there are also adjunct pathways. For example, the sensor
N-linked glycan at residue Asn86 of the HC. Interestingly, UGT1 that adds a terminal glucose to unfolded ER proteins
the location of the glycan is important for CRT but not CXN for rebinding CRT has been recently implicated in the qual-
association with HCs, likely reflecting geometric constraints ity control of antigen presentation by MHC-I. In support of
imposed by PLC assembly. CRT-deficient cells have subop- this conclusion, using in vitro assays with recombinant com-
timal peptide loading resulting in reduced surface MHC-I ponents, UTG1 was reported to preferentially reglucosylate
expression.47 ERp57 is a thiol oxidoreductase that medi- MHC-I proteins loaded with suboptimal peptides.63 Also,
ates disulfide bond formation of different substrates and as a post-ER quality control pathway, CRT was reported to
is commonly associated with CXN and CRT. Interestingly, use a KDEL-dependent mechanism to recycle suboptimally
however, the function of ERp57 as a component of the PLC loaded MHC-I proteins from the early Golgi back the ER for
appears to be independent of the MHC-I redox state. ERp57 improved peptide binding.64
forms a disulfide bond with Tpn that is required for PLC con-
struction.48,49 It was originally proposed that the ERp57-Tpn
complex prevents reduction of the α2 disulfide HC bond, Viral Immune Evasion of Antigen Presentation
thereby keeping HC in a peptide-receptive state.50 However, As noted previously, several molecular components of an-
other studies have detected little evidence the ERp57 con- tigen presentation were coopted from physiologic pathways
trols redox state of MHC-I and more recently, the crystal of protein degradation and quality control. Using elegant
structure of the ERp57-Tpn complex suggests that the role of mechanisms, viruses express diverse proteins that coopt
the ERp57 in the PLC is structural, facilitating recruitment these same pathways of protein degradation and quality
of peptide-accessible MHC-I.49,51,52 As a likely reflection of control to evade immune detection by CD8 T cells.65 Indeed,
its fundamental role in PLC construction, ERp57-deficient the specificity and potency of immune evasion proteins im-
cells were found to have impaired peptide loading, surface pairing antigen presentation makes them efficacious probes
expression, and antigen presentation by MHC-I proteins. for physiologic pathways of relevance for MHC expression.
Tpn, an MHC-I–dedicated molecular chaperone, is re- Most of the well-characterized immune evasion proteins
quired to bridge β2m-assembled HCs with TAP.53 Playing are expressed by deoxyribonucleic acid viruses with large
a redundant role with other members of the PLC, Tpn also genomes, particular viruses capable of latency or host co-
functions in ER retention of MHC with suboptimal pep- existence. Strikingly, immune evasion proteins block sev-
tide cargo.54,55 Perhaps most importantly, Tpn functions as eral different steps of the antigen presentation pathway
a peptide editor by stabilizing peptide-accessible MHC-I by MHC-I proteins (Fig. 22.3). For example, Epstein-Barr
proteins and optimizing peptide cargo before release from virus and Kaposi sarcoma–associated herpesvirus express
the PLC. The mechanism of peptide editing by Tpn of proteins during latency that escape cytotoxic T lymphocyte
MHC-I proteins was revealed using recombinant Tpn teth- detection by containing sequences that inhibit proteasome
ered to HLA,56 and using recombinant Tpn-ERp57 conju- processing.66–70 Alternatively, blocking peptide transport by
gates added to Tpn-deficient cells.57 These and other reports inhibiting TAP function is a commonly used immune eva-
support the model that Tpn promotes the peptide exchange sion strategy. For example, ICP47 of herpes simplex virus
of MHC molecules in an affi nity-dependent manner. The binds the cytosolic side of TAP and blocks peptide and not
consensus model is that Tpn stabilizes the peptide bind- ATP binding,71–76 whereas US6 of human cytomegalovirus
ing groove of MHC-I in an “open” peptide-accessible con- (HCMV) binds the luminal side of TAP and induces a con-
formation. The association of Tpn with peptide-accessible formational change resulting in inhibition of ATP hydrolysis

FIG. 22.2. A Model of the Peptide Loading Complex Based on Known Structures and Mutagenesis
Data of Interaction Sites.52 Heavy chains (HCs), β2m, and ERp57 are shown as ribbons, while cal-
reticulin (CRT) and tapasin (Tpn) are shown as Connolly surfaces (1.4 Å probe). HCs and Tpn are
membrane anchored as indicated with arrows, whereas transmembrane interactions between
Tpn and TAP (not shown) attach the phospholipase C (PLC) components to TAP. HC sites pre-
dicted by mutagenesis studies to interact with Tpn (residues 227 to 229 and 128 to 136) and CRT
(N86) are indicated as spheres. The long flexible P domain of CRT extends above the major his-
tocompatibility complex (MHC)-I peptide binding platform before it noncovalently binds to ERp57,
which is covalently attached to Tpn. A: PLC components are shown from the side to highlight
Tpn contacts to the HC α2 and α3 domains. B: PLC components shown from above reveal co-
ordinated chaperone binding of both MHCI α-helices near the antigenic peptide’s C-terminus.
The figure was rendered by Drs. William McCoy IV and Daved Fremont using Protein Data Bank
(PDB) files kindly provided by Drs. Karin Reinisch and Peter Cresswell. The structures of the Tpn/
ERp57 complex was published in Dong et al.,52 the structure of CRT was modeled using PHYRE2235
on its homolog calnexin,236 and human leukocyte antigen-A2237 is shown as a representative HC.
It should be noted that the HC conformation shown is that attained after peptide occupancy, as
the peptide-accessible HC conformation that binds the PLC has not been resolved.
and peptide translocation.77–81 As a third mechanism of TAP and a highly conserved atypical RING domain that con-
blocking, UL49.5 of bovine herpesvirus induces TAP degra- fers E3 ubiquitin ligase activity.98–100 Viral proteins mK3 of
dation.82 There are also published examples of immune eva- MHV68, and proteins kK3 and kK5 of Kaposi sarcoma–
sion proteins inhibiting Tpn function. For example, US3 of associated herpesvirus that function as immune evasion
HCMV binds to Tpn and impairs optimization of peptide proteins were the founding members of the MARCH protein
loading,83,84 whereas E3-19K of AdV inhibits the ability of family.101–103 Mechanistically, mK3 binds to TAP and awaits
Tpn to bridge MHC-I with TAP.85 the entry of MHC-I into the PLC after which mK3 ubiqui-
Immune evasion proteins also employ strategies to misdi- tinates the cytosolic tail of MHC-I HCs and thus induces
rect trafficking of MHC-I proteins. For example, the afore- their dislocation to the cytosol and degradation in the cy-
mentioned E3-19K protein of AdV binds MHC-I proteins tosol (ie, ERAD).104–107 Of note, the extensively studied US2
and blocks their transport out of the ER.86–90 Interestingly, and US11 proteins of HCMV also target ERAD of MHC-I
cowpox virus expresses two different immune evasion pro- proteins by recruiting cellular E3 ligases.108–113 Indeed, stud-
teins targeting MHC-I proteins that function in tandem. ies of US2, US11 and mK3 continue to provide molecular in-
Cowpox virus 12 is a TAP function blocker that curtails sights into various mechanisms by which ERAD substrates
peptide supply, whereas cowpox virus 203 returns fully as- are detected in the ER and dislocated to the cytosol.114,115
sembled MHC-I proteins from the Golgi back to the ER.91–94 In contrast to ERAD, the kK3 and kK5 MARCH ligases of
To also misdirect assembled MHC-I proteins, the Nef pro- Kaposi sarcoma–associated herpesvirus induce endocytosis
tein of human immunodeficiency virus and gp48 protein of and lysosomal degradation of MHC-I proteins. Interesting,
murine cytomegalovirus shuttle assembled MHC-I proteins kK3 like mK3 appears to ubiquitinate only MHC-I proteins,
from the Golgi to the lysosome.95–97 whereas kK5 targets other surface receptors including the T
Interestingly, the aforementioned immune evasion pro- cell costimulation molecule CD86 (B7.2) as well as natural
teins for the most part do not have cellular homologs, mak- killer cell ligands, MHC class I-related chain A or B (MICA,
ing it difficult to track their evolution. Striking exceptions MICB), and newly defined ligand for NKp80, activation-
to this generality are the viral ER ubiquitin ligases called the induced C-type lectin (AICL).116 Inhibition of natural killer
viral MARCH (membrane-associated RING-CH) proteins. responses is of importance to the virus because downregu-
The extensive family of MARCH proteins includes viral and lation of MHC-I proteins renders cells susceptible to natu-
cellular homologs that share a transmembrane orientation ral killer cell lysis. Mechanistic studies of viral MARCH

Golgi
Apparatus Endosome
5
Endoplasmic 4
Reticulum
Erp57
ERAAP
CRT
Tpn
TAP1
HC TAP2
β2m
3
2
1
Proteasome
FIG. 22.3. Immune Evasion Strategies Used by Viral Proteins to Block Antigen Presentation by Major
Histocompatibility Complex (MHC)-I Proteins. Black arrows demarcate physiologic pathways for anti-
gen presentation, whereas the numbered red arrows demarcate reported immune evasion strategies by
virus proteins. 1: Blocking of antigen processing by the proteasome (examples are the EDNA1 protein of
Epstein-Barr virus or the LANA1 protein of Kaposi sarcoma–associated herpesvirus). 2: Blocking pep-
tide transport by transporter associated with antigen presentation (examples include the ICP47 protein
of herpes simplex virus, US6 protein of human cytomegalovirus, cpxv12 protein of cowpox virus, UL49.5
protein of bovine herpesvirus, and BNFL2a protein of Epstein-Barr virus). 3: Dislocation of MHC-I proteins
from the endoplasmic reticulum (ER) to the cytosol for proteasome-mediated degradation (examples are
US2 and US11 proteins of human cytomegalovirusand mK3 protein of MHV68). 4: ER retention and re-
trieval of MHC-I proteins (examples are E3-19K protein of AdV and the cpxv203 protein of cowpox virus).
5: Mistrafficking of MHC-I proteins from the Golgi to an endosomal compartment (examples include gp48
protein of murine cytomegalovirus and Nef protein of human immunodeficiency virus-1). 6: Rapid internal-
ization of MHC-I proteins from the cell surface to an endosomal/lysosomal compartment (examples are kK3
and kK5 proteins of Kaposi sarcoma–associated herpesvirus).

proteins have provided several molecular insights into lysosome, forming a degradative phagolysosome that neu-
ubiquitin-dependent degradation pathways of particular tralizes and degrades internalized antigens into peptides.127
relevance to antigen processing. In addition, there are about Other surface receptors, such as mannose receptors122,128,129
10 cellular MARCH protein homologs in humans and mice, and DEC-205130 on DCs, also allow for efficient antigen
many of which detect immune receptors.98–100 For example, binding and uptake onto the endocytic pathway, and these
as mentioned in the MHC-II section, MARCHI regulates receptors have been exploited as a vehicle to efficiently de-
MHC-II expression in B cells and dendritic cell (DC) sub- liver bring foreign vaccine antigens into the MHC-II pro-
sets.117–119 Thus, viral MARCH proteins were stolen from the cessing pathway in antigen-presenting cells (APCs).131
host and adapted by the virus to function as immune eva- While antigens entering the endocytic pathway by phago-
sion proteins using similar mechanisms. cytosis, macropinocytosis, and receptor-mediated endocyto-
sis give rise to the majority of peptides that bind to MHC-II,
there is increasing evidence that cytosolic antigens are capa-
ANTIGEN PROCESSING AND PRESENTATION BY ble of entering endosomal MHC-II processing compartments
MAJOR HISTOCOMPATIBILITY CLASS II (MHC-II) and are degraded there into MHC-II–binding peptides. The
PROTEINS most likely cellular mechanism responsible for this phenom-
ena is process termed autophagy.132,133 Macroautophagy is
Source of Antigen for MHC-II Presentation constitutively active in DCs and is a process that leads to the
As a general rule, the antigen-processing machinery that is “engulfment” of aggregated cytosolic protein complexes to
utilized by MHC-I is specialized for the degradation, impor- form an autophagosome. Like phagosomes from extracellu-
tation, and binding of peptides that are derived from antigens lar antigens, the autophagosome fuses with a lysosome and
present in the cytosol of host cells. Although many viruses results in the degradation of autophagosome contents into
and bacteria thrive in the cytosol of infected cells, there are antigenic peptides.134,135 A related process, termed chaper-
pathogens (such as Toxoplasma gondii) that reside in mem- one-mediated autophagy, results from the direct import of
brane-encapsulated intracellular compartments and do not cytosolic proteins into the lumen of endo/lysosomes and is
access the cytosol.120 Furthermore, almost any foreign mate- facilitated by the chaperone protein hsc-70 and the lysosomal
rial (including live or dead pathogens, fragments of apoptotic integral membrane protein Lamp2a.136 Once inside this or-
cells, or soluble proteins) can enter the cell by endocytosis ganelle, the imported protein is degraded into antigenic
and are therefore sequestered from the cytosol. It is for recog- peptides just like “conventionally” internalized antigens.
nition of antigens such as this that MHC-II molecules exist; While this pathway clearly functions in vitro, the extent to
MHC-II binds peptides derived from antigens that gain access which autophagy participates in MHC-II–restricted immune
to intracellular compartments in cells by a variety of endo- responses remains to be established.
cytic pathways.121 Fluid-phase macropinocytosis is a process
whereby extracellular fluid is taken up by plasma membrane
protrusions that bring extracellular fluids and soluble pro- MHC-II Assembly and Transport to the
teins into the cell by endocytosis. Once internalized, mac- Endosomal Pathway
ropinosomes fuse with early endosomes, thereby releasing Because MHC-II is specialized to bind peptides generated in
their contents to the endocytic pathway. Immature (resting) the endocytic pathway, there must be mechanisms in APCs
DCs are particularly efficient in their capacity for macropino- to prevent the premature binding of peptides to MHC-II as
cytosis that enhances their significance as “sentinals” of the the molecules assemble in the ER and traffic through the
immune system.122 Upon activation by encounter with for- Golgi apparatus before their delivery to endosomal pep-
eign organisms, DCs greatly suppress macropinocytosis,122–124 tide loading compartments. In addition to the inability of
thereby limiting the ability of the activated DCs to sample MHC-II to bind to TAP/tapasin and other members of the
their microenvironment and restricting the repertoire of an- MHC-I–peptide loading complex, MHC-II is unique in
tigenic peptides presented to T-lymphocytes. that it binds to a transmembrane chaperone protein termed
As one can imagine, macropinocytosis is not a particu- the invariant chain (Ii). It is the association of Ii with the
larly efficient mechanism of antigen uptake. B cells possess MHC-II–peptide binding site that prevents MHC-II mol-
antigen-specific receptors (immunoglobulins) that permit ecules in the early secretory pathway from binding pep-
specific antigen internalization and presentation to T cells tides.137,138 Indeed, failure of newly synthesized MHC-II to
that can be 1000 times more efficient than fluid-phase en- bind Ii allows for the presentation of ER-imported peptides
docytosis.125 Curiously, the function and fate of B cells that to CD4 T cells by MHC-II.139 A specific region of Ii, termed
generate specific MHC-II peptide complexes following im- the class II-associated Ii peptide (CLIP), serves to block
munoglobulin-mediated antigen uptake versus fluid-phase the peptide-binding groove of nascent MHC-II molecules
uptake of the same antigen differ significantly,126 highlight- until it is ready for removal in endo-/lysosomes. The MHC-
ing the importance of the receptor-mediated antigen uptake II-CLIP complex has been crystallized, and as one might
pathway in B cells. Similarly, macrophages and DCs express expect, the structure of this complex is nearly identical to
Fc receptors and complement receptors that allow efficient that of a MHC-II molecule containing an “antigenic” pep-
phagocytosis of large extracellular particles such as anti- tide.140 As the name implies, unlike the highly polymorphic
body-coated organisms or even apoptotic cells. Following MHC-II molecule itself, Ii is nonpolymorphic. However, the
endocytosis, the phagosome fuses with the membrane of a human Ii mRNA can initiate protein synthesis using one of

two translation initiation sites (Fig. 22.4A), leading to gener- Ii homotrimers and mixed heterotrimers are generated, and
ation of a major 33 kDa form of Ii (Ii-p33) as well as a minor each trimer contains an Ii molecule whose presence in the
population of larger Ii molecules whose translation is initi- trimer is proportional to its level of expression.145–147 Onto
ated using an upstream methionine (Ii-p35141). The use of al- this “scaffold” of an Ii trimer, pairs of MHC-II αβ heterodi-
ternative translation initiation sites appears to be unique to mers assemble, ultimately forming a mature, nine-chain
humans, as it is not seen in rodents.142 Furthermore, alterna- MHC-II-Ii complex.148 The partially formed MHC-II-Ii com-
tive mRNA splicing in all species examined can give rise to plex containing an Ii trimer and only one or two αβ heterodi-
an additional 8 kDa domain in each form of Ii, leading to the mers is retained in the ER by interacting with the chaperone
production of Ii-p41 and (in humans) Ii-p43.143,144 This ad- calnexin,149 presumably allowing the MHC-II molecule addi-
ditional exon encodes a domain (see subsequent discussion) tional time to form a fully functional MHC-II-Ii nonamer. In
that regulates the activity of proteases in endo-/lysosomes. addition, the longer p35 and p43 forms of human Ii contain
The Ii molecule itself forms a trimer even in the absence of an ER retention motif, and free Ii trimers containing one of
MHC-II, and each constituent subunit of this trimer contains these longer Ii forms are retained in the ER.150,151 Curiously,
either Ii-p33, -p35, -p41, or -p43 isoforms (see Fig. 22.4B). completely folded MHC-II-Ii nonamers containing one of
p33 H2N TM CLIP COOH

1 216
p35 H2N TM CLIP COOH

-16 1 216
p41 H2N TM CLIP exon 7 COOH

1 280
p43 H2 N TM CLIP exon 7 COOH

-16 1 280
B C C(216)
C
C CH C
193
CLIP
118
104
81
Membrane Membrane
N N N N
N N
FIG. 22.4. Structure of Invariant Chain (Ii) and Major Histocompatibility Complex (MHC)-II–Ii Complexes. A: The major 216 amino acid
form of human Ii is the 33 kDa Ii-p33 isoform. The use of an upstream initiation methionine results in the generation of Ii-p35 (which
contains an endoplasmic reticulum–retention motif), and alternative splicing of each of these isoforms leads to the generation of Ii-p41
and Ii-p43 forms of Ii that possess proteinase inhibitor activity. The relative positions of the transmembrane domain (TM) and class
II-associated Ii peptide region of each isoform of Ii are indicated. B: Models of Ii trimer (left panel) and the MHC-II–Ii complex (right
panel). Only two of the three pairs of MHC-II αβ dimers are shown bound to the Ii trimer. The model shows the peptide-binding groove
of MHC-II αβ dimers occupied by the CLIP region of Ii. (From Cresswell P. The biochemistry and cell biology of antigen processing. In:
Paul WE. Fundamental Immunology. 5th ed. Philadelphia, PA: Lippincott Williams & Wilkins; 2003:613–629, reprinted with permission).

these longer Ii isoforms are capable of exiting the ER and recognition of two dileucine sorting motifs present in the cy-
moving to antigen-processing compartments,147,152 demon- tosolic domain of every isoform of Ii.158,159 The machinery that
strating that assembly with MHC-II masks the ER-retention recognizes these motifs in the trans-Golgi network remains
motif on the Ii trimer. to be defined; however, it is clear that these motifs are recog-
Ii association does more for MHC-II than simply prevent nized at the plasma membrane by the AP-2 adaptor protein
premature peptide binding. Assembly with Ii is critical for complex.160,161 This protein complex serves as a linker between
some alleles of MHC-II to fold property and achieve compe- Ii and the clathrin-mediated endocytosis pathway and serves
tence to leave the ER, thus revealing a role for Ii as an assem- to deliver plasma membrane-localized MHC-II-Ii to the early
bly/folding molecular chaperone.153 After leaving the ER, the endosomes. Recognition and internalization of MHC-II-Ii
MHC-II-Ii complex moves through the Golgi apparatus and from the plasma membrane is efficient and rapid,157 and as
is targeted to the endosomal pathway150,154 (Fig. 22.5A). It is a consequence of this, very little MHC-II-Ii is present on the
likely that MHC-II-Ii complexes arrive in the endocytic path- plasma membrane of APCs at any given moment.
way by two distinct mechanisms and the relative contribution
of each pathway may be cell-type specific. Some MHC-II-Ii
complexes target directly from the trans-Golgi network to Removal of Invariant Chain in Antigen-Processing
endosomes,155,156 while other complexes move from the trans- Compartments
Golgi network directly to the plasma membrane and enter Regardless of the mechanism responsible for delivering
endosomes only after endocytosis.157 The movement of MHC- MHC-II-Ii to endosomes, Ii must be removed in order to
II-Ii complexes to the endosomal pathway is facilitated by allow antigenic peptides to bind to nascent MHC-II. Ii is
MHC-II-Ii MHC-II-peptide
complex complexes
antigen
Recycling
Tubules
Endosome
Multivesicular
Antigen Processing
TGN Compartment
Golgi
Apparatus
ER
Nucleus
Terminal
Lysosome
FIG. 22.5. Transport of Major Histocompatibility Complex (MHC)-II in Antigen-Presenting Cells. A: MHC-II
associates with Ii in the endoplasmic reticulum, moves through the Golgi apparatus to the plasma mem-
brane, and enters the endocytic pathway by clathrin-mediated endocytosis. Some fraction of MHC-II
can also access these endocytic compartments directly from the Golgi apparatus/trans-Golgi network.
Antigens access these same endocytic compartments by a number of different endocytosis pathways. The
MHC-II–invariant chain (Ii) complex moves to late endosomal/prelysosomal antigen-processing compart-
ments in which Ii is degraded, the MHC-II–associated class II-associated Ii peptide (CLIP) fragment of Ii is
removed by DM, and peptides derived from foreign antigens then bind to MHC-II (this compartment is high-
lighted in panel B). MHC-II–peptide complexes leave these intracellular antigen-processing compartments
via vesicles or tubules that fuse with the plasma membrane, thereby delivering MHC-II–peptide complexes
on the antigen-presenting cell surface. Once at the surface, MHC-II–peptide complexes can recycle be-
tween the plasma membrane and the endosomal system, potentially exchanging antigenic peptides. Some
fraction of internalized MHC-II is also targeted for degradation in terminal lysosomes. (continued)

MHC-II αβ-peptide
GILT in microdomain
Ag
S-S
αβIi cleaved
MHC-II αβIi
αβ-CLIP
HLA-DO HLA-DM
releases catalyzed
HLA-DM CLIP
removal
Membrane Foreign Cysteine Antigenic

αβ-Ii αβ-CLIP αβ-Peptide HLA-DM HLA-DO
B microdomain antigen proteases peptides
FIG. 22.5. (continued) B: Binding of peptides to MHC-II in antigen-processing compartments. The multi-
vesicular antigen-processing compartment highlighted in panel A is shown here. MHC-II–Ii complexes
are targeted to intraluminal vesicles of late endosomal multivesicular antigen-processing compart-
ments where Ii proteolysis is initiated. This process results in the generation of MHC-II–CLIP com-
plexes that are subject to editing by DM. DO binds to DM during intracellular transport and dissociates
at the low pH of these compartments, thereby liberating free DM that catalyzes the removal of CLIP
from the MHC-II–CLIP complex. DM also stabilizes the structure of peptide-free MHC-II until high-
affinity peptide binding. Disulfide bonds present in protein antigens can be reduced by IFNγ-induced
thiol reductase, thereby enhancing the formation of antigenic peptides generated by proteolysis. In
this cartoon MHC-II, DM, and DO have particular membrane distributions; however, whether these
distributions are maintained in all antigen-presenting cells and whether the reactions highlighted here
occur on the indicated membranes remains to be determined. MHC-II molecules present in antigen-
processing compartments are present in distinct membrane microdomains, and after peptide binding,
microdomain-associated MHC-II is transported to the plasma membrane in vesicles or tubules that
bud from the limiting membrane of the antigen processing compartments (DM, HLA-DM; DO, HLA-DO).
degraded in discrete steps by proteolysis in acidic endo- product, highlighting the importance of this proteinase in Ii
somes (see Fig. 22.5B), with proteolysis initiated at the lumi- degradation. Curiously, MHC-II–expressing epithelial cells
nal C-terminus of Ii in earlier endosomes and Ii degradation in the thymus do not rely on cathepsin S for Ii degradation
completed in more acidic late endosomal/prelysosomal but require cathepsin L for terminal Ii degradation.164
structures. In vitro studies have shown that many different Proteolysis of Ii from both the luminal C-terminus and
proteinases are capable of degrading Ii; however, in vivo, from residues adjacent to the luminal Ii transmembrane
the initial cleavage of Ii is initiated by a proteinase whose domain ultimately leaves a small fragment of Ii bound to
activity in insensitive to the protease inhibitor leupeptin MHC-II. This class II-associated Ii polypeptide, termed CLIP,
while additional cleavage is strictly leupeptin-dependent.162 is bound to the peptide-binding groove of MHC-II and serves
Leupeptin treatment of living APCs results in the accumu- as a “placeholder” to occupy the peptide-binding site until
lation of a 21 kDa Ii degradation intermediate that retains antigenic peptides are available to bind (see Fig. 22.4). While
the nonameric MHC-II-Ii stoichiometry and is unable to some MHC-II alleles have a very low affinity for CLIP and CLIP
bind antigenic peptides.152 After the initial cleavage of Ii, a spontaneously dissociates from MHC-II,165 in most cases CLIP
variety of proteinases can cleave the remaining MHC-II– must be actively removed from the peptide-binding site of
associated Ii molecule; however, the terminal stages of Ii MHC-II prior to antigenic peptide binding. A seminal finding
degradation in B cells and DCs are dependent on the cys- in antigen processing and presentation came in 1994 with the
teine proteinase cathepsin S.163 Inhibition of cathepsin S ac- discovery of HLA-DM (DM).166,167 HLA-DM (known as H-2M
tivity or analysis of DCs isolated from cathepsin S-deficient in mice) is a nonpolymorphic transmembrane heterodimer
mice reveal the accumulation of a 10 kDa Ii degradation whose α- and β-subunits are encoded in the MHC gene locus.

Like MHC-II, DM is expressed exclusively in APCs. An im- foreign (and self-) antigens contain intrachain disulfide
portant function of DM is to facilitate the removal of CLIP bonds, and reduction of these bonds would alter protein
from the MHC-II-CLIP complex.168,169 DM functions enzy- conformation and render a protein more (or less) sensitive
matically, and while the precise mechanism by which DM to proteolysis in antigen-processing compartments. As its
promotes CLIP dissociation from MHC-II remains to be com- name implies, the expression of IFNγ-induced thiol reduc-
pletely elucidated, it is thought that DM disrupts the hydrogen tase, GILT, can be induced by exposure to IFNγ.182 This thiol
bonding network in the peptide binding groove of MHC-II reductase is expressed in antigen-processing compartments
and thereby enhances the kinetics of CLIP dissociation from in APCs, and the activity of GILT is important for the pre-
MHC-II.170 Not only does DM catalyze CLIP removal from sentation of antigenic epitopes that are generated from pro-
MHC-II, but DM also facilitates the dissociation of weakly teins that contain many intrachain disulfide bonds.183
bound antigenic peptides from MHC-II168,170,171; it is for this Just as important as generation of antigenic peptides is
reason that DM has been termed a “peptide editor.” Unlike the the ability of the antigen-processing compartment to limit
MHC-II molecule itself, DM contains a tyrosine-based endo- excessive proteolytic degradation of antigenic peptides.
cytic sorting motif in its cytosolic domain that directly inter- While asparaginyl cysteine endoprotease positively regulates
acts with clathrin-associated adaptor proteins.172 The presence expression tetanus toxoid epitopes, this same proteinase de-
of this signal allows for the entry of DM into the endocytic stroys expression of the immunodominant epitope of myelin
pathway after arrival at the cell surface. Most DM is localized basic protein,184 a protein thought to be involved in the de-
in late endocytic compartments173 that contain MHC-II-CLIP myelinating autoimmune disease multiple sclerosis. Clearly,
complexes,174 and the activity of DM is enhanced by the low the proteinase activity of antigen-processing compartments
pH of these later endocytic compartments.168,169,175 must be sufficient to generate antigenic peptides capable of
DM editing activity in APCs can be regulated by its asso- binding to MHC-II but not so great as to completely destroy
ciation with another MHC-encoded transmembrane protein these same epitopes. Measurement of proteinase activity in
termed HLA-DO (DO; H-2O in mice176). While the precise various APC subtypes has revealed that macrophages have
function of DO in antigen processing and presentation re- high proteinase activity and efficiently degrade proteins,
mains to be defined, in vitro studies have shown that DO but are relatively pool generators of antigenic peptides.185
binds to DM and inhibits the ability of DM to catalyze CLIP By contrast, DCs and B cells possess less proteinase activ-
removal.177,178 Like DM, DO expression is relatively restricted ity, degrade proteins less efficiently, and are superb genera-
to APCs. Unlike MHC-II-Ii or DM, DO does not contain tors of antigen peptides.185,186 This specialization of limiting
any intrinsic endo-/lysosomal sorting motifs. Instead, DO proteinase activity helps DCs function as initiators of naïve
associated with DM in the ER and DM serves to “escort” DO T-cell activation and B cell capacity for immunoglobulin
to antigen-processing compartments.179 It is in the low pH of class switching after encounter with antigen-specific T cells.
these compartments that DO dissociates from DM, thereby It has been proposed that binding to the MHC-II mole-
enhancing DM activity. In this respect, DO acts as an in- cules can serve as a mechanism to preserve the integrity of
hibitor of DM function in much the same way that the CLIP antigenic peptides. The peptide-binding groove of MHC-II
region of Ii acts as an inhibitor of MHC-II peptide binding. is open, and in vitro binding experiments have shown that
MHC-II is capable of binding to unfolded (but otherwise in-
tact) protein antigens.187 The association of distinct regions
Antigen Processing and Peptide Loading onto MHC-II within a full-length (unfolded) protein with the peptide-
While MHC-II in lysosome-like antigen-processing com- binding site of MHC-II could therefore protect these regions
partments must be “prepared” for peptide binding (by from proteolytic degradation by endo-/lysosomal proteinases.
degrading MHC-II–associated Ii and removing CLIP by This model of “bind first trim later” has been championed as
DM), internalized foreign antigens must be similarly pre- a major mechanism of peptide loading onto MHC-II188 and
pared for binding to the nascent MHC-II molecule. This has been supported by experimental studies in B cells189 ; how-
task is initiated by proteolysis by any number of endo- ever, the extent to which “bind first trim later” versus “trim
cytic proteinases. Antigen-processing compartments have first bind later” predominates in distinct APC subsets and for
a pH between 4.5 and 6.5, and most proteinases in these different antigens remains to be determined.
compartments have enzymatic activity optima at this pH. Curiously, the Ii molecule itself also can serve to limit
There is remarkable diversity in the proteinase pool in these proteinase activity of antigen-processing compartments.
compartments and great redundancy in the ability of dis- The additional exon that is present in Ii isoforms that are
tinct proteinases to degrade particular proteins. For this products of alternative splicing (Ii-p41 and Ii-p43) encodes
reason, it has been difficult to identify specific proteinases a domain that functions as an inhibitor of cathepsin L.190,191
whose activity is required to generate specific antigenic pep- In vitro studies showed that coexpression with Ii-p41 in-
tides. There are exceptions to this, however, as asparaginyl hibits the kinetics of Ii-p33 degradation,192 a fi nding that
cysteine endoprotease is required for the generation of an is consistent with the role of Ii-p41 as a proteinase inhibi-
immunodominant epitope to tetanus toxoid180 and both ca- tor. However, no defects in function have been identified
thepsin L and cathepsin S affect quantitative and qualitative using APCs isolated from mice expression only the Ii-p41,193
differences in the repertoire of peptides bound to MHC-II leaving the importance of this domain in APC function an
molecules.181 It is also a known fact that many internalized open issue.

DCs are particularly interesting APCs in that they can major mechanism leading to enhanced antigen processing,
regulate their proteinase activity upon cell activation. peptide generation, and formation of MHC-II–peptide com-
Immature (resting) DCs are highly phagocytic and have a plexes upon DC activation.
remarkable capacity for macropinocytosis. Nevertheless,
these cells have little proteinase activity and are relatively
inefficient generators of antigenic MHC-II–peptide com- Delivery of MHC-II to the Cell Surface
plexes, potentially allowing them to serve as a “depot” for The sites of antigenic protein degradation and subsequent
antigen internalized while the cell is in the quiescent state. binding of peptides onto nascent MHC-II are commonly re-
It has even been shown that internalized proteins can referred to as antigen-processing compartments. At one time,
main intact in immature DCs for days.194,195 Activation of it was thought that APCs possessed a “special” compartment
the immature DC by any of a number of mechanisms (ei- that served as the major antigen-processing compartment199 ;
ther by toll-like receptor signaling or stimulation by acti- however, it is now clear that “antigen-processing compart-
vated T cells) leads to a number of changes in the cell, with ments” refer to a heterogenous collection of intracellular
perhaps the most important being the activation of the (generally acidic) organelles that permit peptide binding
vacuolar proton ATPase.196 Activation of this ATPase leads onto MHC-II.200 These organelles generally have charac-
to the acidification of antigen-processing compartments, teristics of typical late endosomes/prelysosomes found in
enhances the activity of endo-/lysosomal proteinases that all cell types (Fig. 22.6); however, in APCs these organelles
have low pH optima, and leads to the degradation of in- contain the specialized protein machinery required for
ternalized antigens to form antigenic peptides capable of antigen processing and presentation such as MHC-II, DM,
binding to MHC-II. Other changes also take place follow- DO, GILT, and Ii-processing enzymes. A “typical” antigen-
ing activation of immature DCs, including downregulating processing compartment is a late endosomal multivesicu-
expression of the endogenous cysteine proteinase inhibitor lar body (MVB), an organelle that possesses intraluminal
cystatin C197 and reorganizing the internal membrane struc- vesicles encapsulated in a limiting membrane.201 These com-
ture of the antigen processing compartments198 ; however, partments can also have a multilamellar structure, contain-
it is likely that acidification of these compartments is the ing numerous membrane sheets contained within an even
FIG. 22.6. Structure of Multivesicular Antigen-Processing

Compartments. A: Ultrathin cryosections of a human B-cell line
were stained with antibodies recognizing DM (10 nm gold) and
total major histocompatibility complex (MHC)-II (15 nm gold).
The plasma membrane stains almost exclusively for MHC-II
while various forms of multivesicular antigen-processing com-
partments contain both MHC-II and DM (reprinted with permis-
sion Kleijmeer et al., Methods, 1996). B: Ultrathin cryosections
of a mouse immature DC line were stained with antibodies rec-
ognizing MHC-II (10 nm gold). MHC-II staining is observed on
the limiting membrane as well as in the intraluminal vesicles of
a multivesicular body (left ) and the numerous intraluminal mem-
brane sheets in a multilamellar endosome (right ) (reprinted with
permission Kleijmeer et al.198).

more acidic lysosome-like organelles. Curiously, the in DCs receptors. MHC-II–peptide complexes are clustered on
majority of MHC-II is present on the intraluminal vesicles the plasma membrane195,213 and associate with distinct
of these MVBs while the CLIP editor DM and MHC-II– lipid raft-214 and tetraspanin-membrane215 microdomains,
peptide complexes are enriched on the MVB-limiting mem- thereby locally concentrating MHC-II on the cell surface.
brane.198 Although most MHC-II and DM are present on Biochemical studies have revealed that MHC-II associates
distinct MVB membranes, it is unclear on which membranes with these membrane microdomains while still present in
the process of Ii proteolysis, CLIP release, and peptide bind- antigen-processing compartments even before antigenic
ing to MHC-II occurs. For example, while there is in vitro peptide binding.216 This finding is consistent with the
evidence that DM primarily interacts with MHC-II when observation that MVBs contain large amounts of choles-
these proteins are on the same membrane,202,203 whether this terol, a major constituent of lipid raft membrane microdo-
is the intraluminal vesicle or the MVB-limiting membrane mains.217 While lipid raft- and tetraspanin-microdomain
remains to be determined. associations are important for the T-cell stimulatory func-
The presence of MHC-II–peptide complexes on the in- tion of APCs,214,218 the molecular signals leading to mem-
traluminal vesicles of MVB presents a topological problem brane microdomain association of MHC-II remain to be
for the APC, as direct fusion of the MVB with the plasma determined.
membrane would release the intraluminal vesicles from MHC-II–peptide complexes are generated and reside in a
these compartments to the extracellular space. Such a pro- relatively hostile (acidic, proteinase-rich) environment, and
cess would effectively result in the secretion of MHC-II therefore it is not surprising that cell surface MHC-II has a
from the APC. This process does occur in DCs with some relatively long half-life.219,220 While this is true on activated B
frequency, and these secreted intraluminal vesicles are cells and mature DCs, MHC-II on the surface of immature
termed exosomes that are capable of stimulating naïve T DCs has a significantly reduced half-life. This is likely due
cells in vivo.204 While APC-derived exosomes are being uti- to rapid endocytosis and lysosomal degradation of surface
lized as cell-free vaccines,205 their physiologic role remains MHC-II in resting (immature) DCs.221,222 The dichotomy
to be unambiguously determined. The majority of MHC-II between the endocytosis and survival of MHC-II in imma-
in antigen-processing compartments is not released as exo- ture and mature DCs can be explained in part by the fact
somes, but is instead either degraded following fusion of the that internalized MHC-II is ubiquitinated in immature DCs
MVB with lysosomes or inserted into the plasma membrane and targeted for lysosomal degradation while internalized
by “vesicle”-mediated protein transport. While vesicles con- MHC-II is not ubiquitined in mature DCs and efficiently
taining intracellular MHC-II have been observed to fuse recycles back to the plasma membrane.223,224 The E3 ubiqui-
with the plasma membrane,206 in APCs such as DCs it is tin ligase March-I, whose expression appears to be limited
likely that these “vesicles” are actually tubules that bud from to APCs, ubiquitinates the cytosolic domain of the MHC-II
the MVB membrane and subsequently fuse with the plasma β -chain in APCs,118 thereby preventing the recycling of in-
membrane. The tubulation of these compartments is stimu- ternalized MHC-II and targeting ubiquitinated MHC-II for
lated in DCs by APC activation,198,207,208 and the encounter of lysosomal degradation.225 This mechanism ensures that the
DCs with T cells leads to the fusion of these tubules directly kinetics of MHC-II synthesis and degradation in immature
to the plasma membrane region at the point of DC:T-cell DCs are similar and provides a means to halt the degradation
contact,208 thereby delivering MHC-II–peptides generated of relevant MHC-II–peptide complexes upon DC activation.
in DCs directly to the site of T-cell receptor concentration at Simply because MHC-II has a long half-life in B cells
the immunologic synapse. and DCs does not mean that these molecules do not in-
The cellular processes that govern the movement of ternalize. In these APC subtypes, internalized MHC-II ef-
MHC-II to the surface of APCs are similar to those that gov- ficiently recycles from early (recycling) endosomes back to
ern the movement of any late endosomal protein the surface the plasma membrane,221 thereby limiting their delivery to
of any cell. In some cases, the machinery regulating discrete lysosomes and prolonging MHC-II half-life. The ability of
transport steps remains a mystery while other steps are more internalized MHC-II to exchange one peptide for another
well characterized. For example, the mechanism leading to in endosomes has been demonstrated in vitro226,227; how-
the delivery of intraluminal vesicle-bound MHC-II back ever, it remains to be determined whether peptide exchange
to the MVB0limiting membrane (from which tubules are onto recycling MHC-II in vivo is a significant mechanism
formed) is termed back-fusion,209 and the proteins and lip- of peptide loading onto MHC-II. There is convincing evi-
ids that regulate this curious process remain to be identified. dence, however, that certain protein epitopes are primarily
By contrast, it is known that the ability of MHC-II to insert presented by recycling MHC-II.228 Unlike presentation of
into the plasma membrane is regulated by the small GTPase epitopes from “classically processed” antigens, presentation
ARL14/ARF7 that couples the tubulovesicle membrane with of these epitopes does not require neosynthesis of MHC-II
the actin cytoskeleton.210 Like other vesicle-mediated protein and is independent of Ii expression or DM activity. A par-
transport steps, microtubules and dynein/kinesin motors211 ticularly intriguing example of this pathway can be found in
as well as actin and myosin motors212 are important for the the analysis of an immunogenic epitope presentation of the
locomotion of MHC-II into and out of antigen-processing influenza virus hemagglutinin,228 in which presentation of
compartments. one epitope is via the classical (late endosomal) processing
Once at the plasma membrane, MHC-II–peptide com- and presentation pathway while another is presented via a
plexes are poised to interact with antigen-specific T-cell nonclassical (early endosomal) recycling pathway.

Alternate Pathways of Antigen Presentation Interestingly, disparate targeting motifs in their cytoplasmic
In this chapter, we have reviewed what is typically called the tails facilitate the binding of lipid antigens in the early endo-
classical MHC-I and MHC-II antigen-processing and -pre- somes by CD1a, in the late endosome/lysosome by CD1b and
sentation pathways. In addition to these pathways, there are CD1d, and at both sites by CD1c. This diversity of location
also interesting alternative antigen-presentation pathways. of lipid antigen binding by different human CD1 isoforms
Although not the focus of this chapter, these nonclassical allows surveying of different endosomal compartments for
pathways provide clear evidence that antigen presentation microbes. The assembly of CD1 isoforms in the ER uses
is more complex than a simple dichotomy of the two classi- molecular chaperones CXN, CRT, and ERp57 similar to the
cal pathways. The most extensively studied of these alterna- classical MHC-I pathways. As expected, the processing and
tive pathways is cross-presentation, a process by which CD8 loading of lipid antigens presented by CD1 isoforms requires
T cells present exogenous antigen taken up by professional unique players.234 However, the trafficking of CD1 isoforms
APCs such as DCs. Cross-presentation is critical for initi- between the plasma membrane and endosomal compart-
ating CD8 + immune responses to pathogens that do not ments to acquire lipid antigens has parallels with the clas-
infect professional APCs or tumors that do not arise in pro- sical MHC-II pathway. In sum, with the underlying goal of
fessional APCs. Although several details remain unsolved, providing maximal immunity for detection and protection
cell biologic approaches suggest that the loading of cross- against diverse pathogens, mammals have two classical and
presented antigens occurs by several different pathways each other nonclassical pathways of antigen presentation.
with hybrid features of both of the classical pathways.229–231
The presentation of lipid antigens by MHC-I like CD1 ACKNOWLEDGMENTS
molecules also occurs by a nonclassical pathway.232,233 Human The authors thank Drs. Peter Cresswell and Karin Reinisch for
CD1a, CD1b, and CD1c isoforms present bacterial lipids to sharing Protein Data Bank (PDB) files with the model shown
conventional αβ T cells, whereas CD1d presents endog- in Figure 22.3, Dr. William McCoy IV for his rendering of
enous and microbial lipids to invariant natural killer T cells. Figure 22.3, and Dr. Janet Connolly for professional editing.

Induction, Regulation, and Effector

SECTION
VI Functions of the Immune Response
Immunogenicity and Antigen

CHAPTER
23 Structure
Jay A. Berzofsky • Ira J. Berkower
THE NATURE OF ANTIGENIC DETERMINANTS

configurations was reemphasized in later studies measuring
RECOGNIZED BY ANTIBODIES relative affinities of antibodies to maleic and fumaric acid
Haptens conjugates4 (Table 23.1A). Table 23.1B shows the specificity
In the antigen–antibody binding reaction, the antibody- of antibodies prepared against p-azobenzenearsonate cou-
binding site is often unable to accommodate the entire pled to bovine gamma globulin.5 As the hapten is coupled
antigen. The part of the antigen that is the target of anti- through the p-azo group to aromatic amino acids of the car-
body binding is called an antigenic determinant, and there rier, haptens containing bulky substitutions in the para po-
may be one or more antigenic determinants per molecule. sition would most resemble the immunizing antigen. In fact,
To study antibody specificity, we need to have antibodies p-methyl–substituted benzene arsonate has a higher bind-
against single antigenic determinants. Small functional ing affinity than unsubstituted benzene arsonate. However,
groups that correspond to a single antigenic determinant methyl substitution elsewhere in the benzene ring reduces
are called haptens. For example, these may be organic affinity, presumably due to interference with the way hapten
compounds, such as trinitrophenyl or benzene arsonate, fits into the antibody-binding site. Thus, methyl substitu-
a monosaccharide or oligosaccharide such as glucose or tions can have positive or negative effects on binding energy,
lactose, or an oligopeptide such as pentalysine. Although depending on where the substitution occurs. Table 23.1C
these haptens can bind to antibody, immunization with shows the specificity of antilactose antibodies for lactose
them usually will not provoke an antibody response (for versus cellobiose.6 These disaccharides differ only by the
exceptions, see Goodman1). Immunogenicity often can be orientation of the hydroxyl attached to C4 of the first sugar
achieved by covalently attaching haptens to a larger mol- either above or below the hexose ring. The three examples in
ecule, called the carrier. The carrier is immunogenic in this table, as well as many others,1 show the marked specific-
its own right, and immunization with the hapten-carrier ity of antibodies for cis–trans, ortho–meta–para, and ste-
conjugate elicits an antibody response to both hapten and reoisomeric forms of the antigenic determinant.
carrier. However, the antibodies specific for hapten can be Comparative binding studies of haptens have been able
studied by equilibrium dialysis using pure hapten (without to demonstrate antibody specificity despite the marked
carrier) or by immunoprecipitation using hapten coupled heterogeneity of antibodies. Unlike the antibodies against
to a different (and non–cross-reacting) carrier or by inhibi- a multideterminant antigen, the population of antibodies
tion of precipitation with free hapten. specific for a single hapten determinant is a relatively re-
This technique was pioneered by Landsteiner2 and helped stricted population due to the shared structural constraints
to elucidate the exquisite specificity of antibodies for anti- necessary for hapten to fit within the antibody-combining
genic determinants. For instance, the relative binding af- site. However, the specificity of an antiserum depends on the
finity of antibodies prepared against succinic acid–serum collective specificities of the entire population of antibod-
protein conjugates shows marked specificity for the maleic ies, which are determined by the structures of the various
acid analog, which is in the cis configuration, as compared antibody-binding sites. When studying the cross-reactions
to the fumaric acid (trans) form.3 Presumably, the immu- of hapten analogs, some haptens bind all antibodies but with
nogenic form of succinic acid corresponds to the cis form.3 reduced K A. Other hapten analogs reach a plateau of bind-
This ability of antibodies to distinguish cis from trans ing because they fit some antibody-combining sites quite
539

540 | SECTION VI INDUCTION, REGULATION, AND EFFECTOR FUNCTIONS OF THE IMMUNE RESPONSE
This is important for immunoprecipitation by lattice forma-

Exquisite Specificity of Antihapten tion, as discussed in Chapter 5. Several examples illustrating
TABLE 23.1
Antibodies structural studies of oligosaccharide antigens are given later.
The technique used most widely to analyze the anti-
Krel of Antibody
Hapten Structure Specific for genic determinants of polysaccharides is called hapten in-
hibition.8 In this method, the precipitation reaction between
A. Maleic Fumaric antigen and antibody is inhibited by adding short oligosac-
(cis) (trans) charides. These oligosaccharides are large enough to bind
Maleanilate 1.0 < 0.01 with the same affinity and specificity as the polysaccharide,
Fumaranilate < 0.01 1.0
but because they are monomeric, no precipitate forms. As
B. Parasubstituted
benzene arsonate
more inhibitor is added, fewer antibody-combining sites
Benzene arsonate 1.0
remain available for precipitation. Using antiserum spe-
o-Methyl benzene arsonate 0.2 cific for a single antigenic determinant, it is often possible
m-Methyl benzene arsonate 0.8 to block precipitation completely with a short oligosaccha-
p-Methyl benzene arsonate 1.9 ride corresponding to the nonreducing end of the polysac-
C. Lactose charide chain. Besides showing the “immunodominance” of
Lactose β Gal (1 → 4) Glu 1.00 the nonreducing end of the chain, this result also shows that
Cellobiose β Glu (1 → 4) Glu 0.0025 the structure of the antigenic determinant of polysaccha-
rides depends on the sequence of carbohydrates and their
Part A from Pressman and Grossberg,4 part B from Pressman et al.,5 and part C linkage, rather than their conformation. For inhibition by
from Karush,6 with permission.
hapten to be complete, the antigen–antibody system studied
must be made specific for a single antigenic determinant.
For optimal sensitivity, the equivalence point of antigen and
well but not others (see discussion of cross-reactivity in antibody should be used.
Chapter 7). Antibodies raised in different animals may show We illustrate the types of carbohydrate antigens encoun-
different cross-reactivities with related haptens. Even within tered by examining three classic examples in more detail:
a single animal, antibody affinity and specificity are known the salmonella O antigens, the blood group antigens, and
to increase over time following immunization under certain dextrans that bind to myeloma proteins.
conditions.7 Thus any statements about the cross-reactivity
of two haptens reflect both structural differences between
Immunochemistry of Salmonella O Antigens
the haptens that affect antigen–antibody fit and the diversity
of antibody-binding sites present in a given antiserum. The antigenic diversity among numerous salmonella species
resides in the structural differences of the lipopolysaccha-
ride (LPS) component of the outer membrane.9 These mole-
Carbohydrate Antigens cules are the main target for antisalmonella antibodies. The
The antigenic determinants of a number of biologically im- polysaccharide moiety contains the antigenic determinant,
portant substances consist of carbohydrates. These often whereas the lipid moiety is responsible for endotoxin effects.
occur as glycolipids or glycoproteins. Examples of the for- The chemical structure of LPS can be divided into three re-
mer include bacterial cell wall antigens and the major blood gions (Fig. 23.1). Region I contains the antigenic O-specific
group antigens, whereas the latter group includes “minor” polysaccharide, usually made up of repeated oligosaccharide
blood group antigens such as Rh. In addition, the capsular units, which vary widely among different strains. Region II
polysaccharides of bacteria are important for virulence and contains an oligosaccharide “common core” shared among
are often targeted by protective antibodies. A number of many different strains. Failure to synthesize region II oligo-
spontaneously arising myeloma proteins have been found to saccharide or to couple completed region I polysaccharide
show carbohydrate specificity, possibly reflecting the fact that to the growing region II core results in R (rough) mutants,
carbohydrates are common environmental antigens. In the which have “rough” colony morphology and lack the O anti-
days prior to hybridoma technology, these carbohydrate- gen. Region III is the lipid part, called lipid A, which is shared
specific myeloma proteins provided an important model for among all salmonellae and serves to anchor LPS on the outer
studying the reaction of antigen with a monoclonal antibody. membrane. Early immunologic attempts to classify the O
Empirically, the predominant antigenic determinants of antigens of different salmonellae revealed a large number
polysaccharides often consist of short oligosaccharides (one of cross-reactions between different strains. These were de-
to five sugars long) at the nonreducing end of the polymer tected by preparing antiserum to one strain of salmonella
chain.8 This situation is analogous to a hapten consisting of and using it to agglutinate bacteria of a second strain. Each
several sugar residues linked to a large nonantigenic poly- cross-reacting determinant was assigned a number, and
saccharide backbone. The remainder of the polysaccharide each strain was characterized by a series of O antigen de-
is important for immunogenicity, just as the carrier mol- terminants (in aggregate, the “serotype” of the strain) based
ecule was important for haptens. In addition, branch points on its pattern of cross-reactivity. Each strain was classified
in the polysaccharide structure allow for multiple antigenic within a group, based on sharing a strong O determinant.
determinants to be attached to the same macromolecule. For example, group A strains share determinant 2, whereas

CHAPTER 23 IMMUNOGENICITY AND ANTIGEN STRUCTURE | 541
FIG. 23.1. Structure of Salmonella Lipopolysaccharide. Region I contains the unique O-antigen determinants, which consist
of repeating units of oligosaccharides. These are attached to lipid moiety through the core polysaccharide. Three examples
of oligosaccharide units are shown.9 (Part A adapted from Kabat,8 with permission; part B based on Jann and Westphal.9)
group B strains share determinant 4 (Table 23.2). However, cross-reactivity based on sharing of a subset of antigenic de-
within a group, each strain possesses additional O determi- terminants is commonly encountered in complex antigen–
nants, which serve to differentiate it from other members antibody systems. The problem may be simplified by making
of that group. Thus, determinant 2 coexists with determi- antibodies monospecific for individual antigenic determi-
nants 1 and 12 on Salmonella paratyphi A. This problem of nants. To do this, antibodies are absorbed to remove irrel-
evant specificities, or cross-reactive strains are chosen that
share only a single determinant with the immunizing strain.
TABLE 23.2 Salmonella Q Antigen Serotyping The reaction of each determinant with its specific antibody
can be thought of as an antigen–antibody system. Thus, for
Salmonella O Antigenic the strains shown in Table 23.2, antiserum to Salmonella
Strain Serogroup Determinants
typhi (containing anti-9 and anti-12 antibodies) may be
S. paratyphi A A 1, 2, 12 absorbed with S. paratyphi A to remove anti-12, leaving a
S. paratyphi B B 1, 4, 5, 12 reagent specific for antigen 9 (see Table 23.2). Alternatively,
S. typhi D 9, 12 the unabsorbed antiserum may be used to study the system
Single antigen 12–anti-12 by allowing it to agglutinate S. paraty-
Determinant phi B, which shares only antigen 12 with the immunogen.
Antiserum Absorbed Tested on Measured Because the other determinants on S. paratyphi B were ab-
sent from the immunizing strain, the antiserum contains no
Anti-S. typhi S. paratyphi B 12
antibodies to them.
Anti-S. typhi S. paratyphi A S. typhi 9
Once the antigen–antibody reaction is made specific for a
Reprinted with permission from Kabat.8 single determinant, a variety of oligosaccharides can be added

surfaces, they are either glycolipids that are synthesized

Analysis of Salmonella O-Antigen within the cell (AB and H antigens) or glycoproteins taken up
TABLE 23.3
Structure by Hapten Inhibition from serum (Le antigens). In mucinous secretions, such as sa-
liva, they occur as glycoproteins. Milk, ovarian cyst fluid, and
Antigen System
gastric mucosa contain soluble oligosaccharides containing
Maximum Inhibition
by Hapten (%) 1: anti-1 19: anti-19
blood group reactivity. In addition, these antigens occur fre-
quently in other species, including about half of the bacteria
d-Glu — 0 in the normal flora of the gut.10 This widespread occurrence
Me-α-D-Glu 35 10 may account for the ubiquitous anti-AB reactivity of human
α-D-Glu(1 → 6)-D-Gal 80 25 sera, even in people never previously exposed to human blood
Glu.Gal.Man 80 70 group substances through transfusion or pregnancy.
Gtu.Gal.Man.L-Rham > 70 > 70 The immunochemistry of these antigens was simplified
Deduced structure α-D-Glu(1 → 6)- D-Glu-D-Gal-D- greatly by the use of oligosaccharides in hapten inhibition
D-Gal Man-L-Rham
studies. Group A oligosaccharides, for example, would in-
Reprinted with permission from Kabat.8 hibit the agglutination of group A red blood cells by anti-A
antibodies. They could also inhibit the immunoprecipita-
tion of group A–bearing glycoproteins by anti-A antibodies.
to test for hapten inhibition. Because the O antigens contain Because the oligosaccharides are monomeric, their reaction
repeating oligosaccharide units, it is often possible to obtain with antibody does not form a precipitate but does block an
model oligosaccharides by mild chemical or enzymatic deg- antibody-combining site.
radation of the LPS itself. Once the most inhibitory oligo- The inhibitory oligosaccharides from cyst fluid were
saccharide is found, its chemical structure is determined. purified and found to contain D-galactose, L-fucose,
Alternatively, a variety of synthetic monosaccharide, disaccha- N-acetylgalactosamine, and N-acetylglucosamine. The most
ride, trisaccharide and oligosaccharides are tested for hapten inhibitory oligosaccharides for each antigen are indicated in
inhibition of precipitation. For example, as shown in Table Figure 23.2. As can be seen in Figure 23.2, the ABH and Le
23.3, antigen 1–anti-1 antibody precipitation is inhibited by antigens all share a common oligosaccharide core sequence,
methyl-α d-glucoside. Therefore, various disaccharides incor- and the antigens appear to differ from each other by the se-
porating this structure were tested, of which α-d-Glu-(1 → 6)- quential addition of individual sugars at the end or at branch
d-Gal was the most inhibitory. Then various trisaccharides points. Besides hapten inhibition, other biochemical data
incorporating this sequence were tested. The results indicate support this relationship among the different determinants.
the sequence and size of the determinant recognized by anti-1 Enzymatic digestion of A, B, or H antigens yields a common
antibodies to be a disaccharide with the previously discussed core oligosaccharide from each. This product cross-reacts
structure. The test sequences can be guessed by analyzing the with antiserum specific for pneumococcal polysaccha-
oligosaccharide breakdown products of the LPS, which in- ride type XIV, which contains structural elements shared
clude tetramers of d-Glu-d-Gal-d-Man-l-Rham. The results with blood group determinants, as shown at the bottom of
in Table 23.3 also suggest that the difference between deter- Figure 23.2. In addition, this structure, known as precursor
minants 1 and 19 is the length of oligosaccharide recognized substance, has been isolated from ovarian cyst fluid.
by antibodies specific for each determinant. This hypothesis is Starting from precursor substance, the H determinant
supported by the observation that determinant 1 is found in results from the addition of L-fucose to galactose, whereas
some strains with, and in other strains without, determinant Lea determinant results from the addition of L-fucose to
19; whereas determinant 19 is always found with determinant N-acetylglucosamine and Leb from the addition of L-fucose
1. As shown in Table 23.3, determinant 19 requires the full tet- to both sugars. Addition of N-acetylgalactosamine to H sub-
rasaccharide for maximal hapten inhibition, including the se- stance produces the A determinant, whereas addition of ga-
quence coding for determinant 1. Besides identifying the antilactose produces the B determinant, in each case blocking
genic structures, these results indicate that there is variation in reactivity of the H determinant.
the size of different antigenic determinants of polysaccharides. The genetics of ABH and Le antigens is explained by this
sequential addition of sugars via glycosyltransferases. The
allelic nature of the AB antigens is explained by the addition
Blood Group Antigens of N-acetylgalactosamine, galactose, or nothing to the H an-
The major blood group antigens A and B were originally tigen. The rare inherited trait of inability to synthesize the
detected by the ability of serum from individuals lacking H determinants from precursor substance (Bombay pheno-
either determinant to agglutinate red blood cells bearing them type) also blocks the expression of A and B antigens because
(for reviews, see Kabat,8 Springer,10 Marcus,11 and Watkins12). the A and B transferases lack an acceptor substrate. However,
In addition, group O individuals have an H antigenic deter- the appearance of the Lea antigen on red cells is independent
minant that is distinct from A or B types, and individuals in of H antigen synthesis. Its structure, shown in Figure 23.2,
all three groups may have additional determinants such as can be derived directly from precursor substance without
the Lewis (Le) antigens. Although the ABH and Le antigenic going through an H antigen intermediate. Comparing dif-
determinants are found on a carbohydrate moiety, the carbo- ferent individuals, the appearance of Lea antigen on red
hydrate may occur in a variety of biochemical forms. On cell blood cells correlates with its presence in saliva, as the Lea

specificities. Careful studies of these monoclonal antibodies

support the clonal expansion model of antibody diversity:
heterogeneous antisera behave as the sum of many individual
clones of antibody with respect to affinity and specificity. In
the case of the Ig Aκ myeloma proteins W3129 and W3434,
both antibodies were found to be specific for dextrans con-
taining α-glu (1 → 6)glu bonds.12 Hapten inhibition with a
series of monosaccharide or oligosaccharides of increasing
chain length indicated that the percentage of binding energy
derived from the reaction with one glucose was 75%, two
glucoses 95%, three glucoses 95% to 98%, and four gluco-
ses 100%. This suggests that most binding energy between
antidextran antibodies and dextran derives from the ter-
minal monosaccharide, and that oligosaccharides of chain
length four to six commonly fi ll the antibody-combining
site. Human antidextran antisera behaved similarly, with
tetrasaccharides contributing 95% of the binding energy.
These experiments provided the first measure of the size of
an antigenic determinant, four to six residues.13,14 In addi-
tion, as was observed for antisera, binding affi nity of my-
eloma proteins was highly sensitive to modifications of the
terminal sugar and highly specific for α (1 → 6) versus α (1
→ 3) glycosidic bonds. However, modification of the third
or fourth sugar of an oligosaccharide had relatively less ef-
fect on hapten inhibition of either myeloma protein or of
antisera reacting with dextran.
Studies with additional dextran-binding myeloma pro-
teins15 revealed that not all antipolysaccharide monoclonal
antibodies are specific for the nonreducing end, as exempli-
fied by QUPC 52. Competitive inhibition with monosaccha-
ride and oligosaccharides revealed that < 5% of binding en-
ergy derived from monosaccharides or disaccharides, 72%
from trisaccharides, 88% from tetrasaccharides, and 100%
FIG. 23.2. Oligosaccharide Chain Specificity. Structure of the ABH from hexasaccharides, in marked contrast to other myeloma
and Le blood group antigens as determined by hapten inhibition proteins. A second distinctive property of myeloma protein
studies.8,11 There are two variants of each of these determinants. In QUPC 52 was its ability to precipitate unbranched dextran
type 1, the Gal-GNAc linkage is β(1 → 3), whereas in type 2, the Gal- of chain length 200. As the unbranched dextran has only
GNAc linkage is β(1 → 4). In addition, there is heterogeneity in the
A and B antigens with respect to the presence of the Le fucose at- one nonreducing end, and as the myeloma protein has only
tached to the GNAc. In the molecules that contain the extra fucose, one specificity, lattice formation due to cross-linking be-
when the Gal-GNAc linkage is β(1 → 3) (type 1), the fucose must be tween the nonreducing ends is impossible, and precipita-
linked α(1 → 4), whereas the type 2 molecules, with the β(1 → 4) tion must be explained by binding some other determinant.
Gal-GNAc linkage, contain α(1 → 3)-linked fucose. The asterisks Therefore, QUPC 52 appears to be specific for internal oligo-
indicate the sites of this variability in linkage. saccharide units of three to seven chain length. The W3129
is specific for end determinants and will not precipitate un-
branched dextran chains. Antibodies precipitating linear
antigen is not an intrinsic membrane component but must dextran were also detected in six antidextran human sera,
be absorbed from serum glycoproteins, which, in turn, de- comprising 48% to 90% of the total antibodies to branched
pend on secretion. In addition to the independent synthetic chain dextran. Thus, antidextrans can be divided into those
pathway, the secretion of Lea antigen is also independent of specific for terminal oligosaccharides and those specific for
the secretory process for ABH antigens. Therefore, salivary internal oligosaccharides; monoclonal examples of both
nonsecretors of ABH antigens (which occur in 20% of indi- types are available, and both types are present in human im-
viduals) may still secrete Lea antigen if they have the fucosyl mune serum. Cisar et al.15 speculated as to the different to-
transferase encoded by the Le gene. In contrast, salivary se- pology of the binding sites of W3129 or QUPC 52 necessary
cretion of ABH is required for red blood cells to express Leb. for terminal or internal oligosaccharide specificity. Both
terminal and internal oligosaccharides have nearly identi-
Dextran-Binding Myeloma Proteins cal chemical structures, differing at a single C–OH or gly-
Because polysaccharides are common environmental an- coside bond. Perhaps the terminal oligosaccharide specific-
tigens, it is not surprising that randomly induced my- ity of W3129 is due to the shape of the antibody-combining
eloma proteins were frequently found to have carbohydrate site—a cavity into which only the end can fit—whereas the

internal oligosaccharide-binding site of QUPC 52 could be Dextran binding depends on the antigen fitting into the
a surface groove in the antibody, which would allow the rest groove and interacting favorably with the residues form-
of the polymer to protrude out at both ends. A more de- ing the sides and bottom of the groove. The results indicate
finitive answer depends on x-ray crystallographic studies of that divergent variable region sequences, both in and out of
the combining sites of monoclonal antibodies with precisely the complementarity-determining regions, can be folded
defined specificity, performed with antigen occupying the to form similar binding site contours, which result in simi-
binding site. lar immunochemical characteristics. Similar results have
With the advent of hybridoma technology, it became been reported in other antigen–antibody systems, such as
possible to produce monoclonal antibodies of any desired phenyloxazolone.22
specificity. Immunizing mice with nearly linear dextran Additional studies of carbohydrate binding monoclonal
(the preferred antigen of QUPC 52), followed by fusion antibodies have revealed significant information about how
and screening (with linear dextran) for dextran-binding the antibody variable regions can bind a carbohydrate struc-
antibodies, yielded 12 hybridomas,16 all with specificity ture with high affinity and specificity. Several examples are
similar to QUPC 52. First, oligosaccharide inhibition of all now available of crystal structures of carbohydrates bound
12 monoclonals showed considerable increments in affinity to antibodies.
up to hexasaccharides, with little affinity for disaccharides For example, monoclonal antibody Se155-4 is specific for
and only 49% to 77% of binding energy derived from tri- the group B determinant of the salmonella O antigen, which
saccharides.17 Second, all 12 monoclonals had internal α (1 consists of the sugars Gal-Abequose-Man.23,24 The crystal
→ 6) dextran specificity, as they could all precipitate linear structure of antibody bound to the polysaccharide shows
dextran. Third, 9 out of 11 BALB/c monoclonals shared a that one hexose, abequose, fits into the binding pocket,
cross-reactive idiotype with QUPC 52, whereas none shared while the rest of the interactions occur along the surface of
idiotype with W3129.18 These data support the hypothesis the antibody, similar to the groove-type sites described pre-
that different antibodies with similar specificity and similar viously. Binding energy depends on hydrogen bonds formed
groove-type sites may be derived from the same family of between the protein residues and the hydroxyl groups of
germline V H genes bearing the QUPC 52 idiotype.18 the carbohydrate. The protein residues include aromatic
The large number of environmental carbohydrate anti- amines, such as His 32, Trp 91, and Trp 96 of the light chain,
gens and the high degree of specificity of antibodies elicited in as well as His 97 and His 35 of the heavy chain. In addition,
response to each carbohydrate antigen suggest that a tremen- one of the sugars is hydrogen bonded via a water molecule
dous diversity of antibody molecules must be available, from bridge to the amide bonds of the protein backbone. About
which some antibodies can be selected for every possible an- three quarters of all sugar hydroxyl groups are involved in
tigenic structure. Studies of a series of 17 monoclonal anti- hydrogen bonds with the protein. Although each H-bond
α (1 → 6) dextran hybridomas19,20 have investigated whether is relatively weak by itself, the combined effect of eight hy-
the binding sites of closely related antibodies were derived drogen bonds results in high-affinity binding. Antibody
from a small number of variable region genes, for both heavy specificity derives from the fact that the carbohydrate fits
and light chains, or whether antibodies of the same specificity into a binding pocket, where H-bond formation depends on
could derive from variable region genes with highly divergent precise interactions with amino acid residues that are ori-
sequences. Each monoclonal had a groove-type site that could ented about the pocket. Surprisingly, most of these bonds
hold six or seven sugar residues (with one exception), based are formed between sugar hydroxyls and aromatic amino
on inhibition of immunoprecipitation by different length oli- acids that are neither charged nor very polar at neutral pH.
gosaccharides. Thus, unlike monoclonals to haptenated pro- Similarly, monoclonal antibody BR96 and the humanized
teins, the precise epitope could be well characterized and was monoclonal hu3S193 are specific for the Le Y antigen, which
generally quite similar among the entire series. resembles the Le B antigen described in Figure 23.2, except
Studies of the Vκ sequences revealed that only three Vκ that the fucose-N-acetylglucosamine bond is β(1 → 3) in-
groups were used in these hybridomas. Use of each Vκ group stead of β(1 → 4). The Le Y antigen is commonly expressed
correlated with the particular antigen used to immunize the on tumor cells of epithelial origin. The crystal structures
animals, whether linear dextran or short oligosaccharides, have revealed the sources of the binding energy that results
so that 10 of the monoclonals from mice immunized the in affinity and specificity for this carbohydrate antigen.25,26
same way all used the same Vκ. These two monoclonals bind Le Y antigen in a large, deep
In contrast, the 17 V H chains were derived from at least pocket, which accommodates all four hexoses and corre-
five different germline genes from three different V H gene spond to the cavity type binding site predicted by Kabat.15
families.21 The two most frequently used germline V H genes The terminal fucose goes in first while the other three sugars
were found in seven and five monoclonals, respectively, are hydrogen bonded to amino acid side chains lining the
with minor variations explainable by somatic mutations. pocket, including Tyr 33, Tyr 35, and Gln 52 of the heavy
The remarkable finding is that very different V H chains chain and His 27 of the light chain. Once again, hydrogen
(about 50% homologous) can combine with the same Vκ bonds between hydroxyl groups of the sugars and aromatic
to produce antibody-binding sites with nearly the same amines (Trp and Tyr) of the protein play a dominant role
size, shape, antigen specificity, and affinity. Even when dif- in determining affinity and specificity of binding. A smaller
ferent V H sequences combine with different Vκ sequences, number of H-bonds depend on amide groups of the protein
they can produce antibodies with very similar properties. backbone.

A third example is provided by human monoclonal 2G12, to 20 Å × 10 Å (D. R. Davies, personal communication),
which has neutralizing activity against a broad spectrum of these contact residues comprising the antigenic determinant
human immunodeficiency virus (HIV) isolates. This anti- may cover a significant area of protein surface, as measured
body binds the mannose-rich oligosaccharide side chains by x-ray crystallography of antibody–protein antigen com-
that form a protective surface, called a glycoshield, on the plexes.34–37 The size of the combining sites has also been es-
envelope glycoprotein gp120. The crystal structure shows timated using simple synthetic oligopeptides of increasing
that the two terminal mannose sugars of each oligosaccha- length, such as oligolysine. In this case, a series of elegant
ride bind end on into a deep pocket of the antibody, in a studies38–40 suggested that the maximum chain length a
cavity-type site.27,28 Twelve hydrogen bonds form between combining site could accommodate was six to eight residues,
the two terminal mannose residues and the protein, de- corresponding closely to that found earlier for oligosaccha-
pending mainly on the amide groups of the protein back- rides,13,14 as discussed previously.
bone. Additional hydrogen bonds form between the third Several types of interactions contribute to the binding
mannose residue and the side chain of Asp 100 of the heavy energy. Many of the amino acid residues exposed to solvent
chain and between the fourth mannose residue and Tyr 94 on the surface of a protein antigen will be hydrophilic. These
of the light chain and Tyr 56 of the heavy chain. A unique are likely to interact with antibody contact residues via polar
feature of this antibody is the crossover of variable regions interactions. For instance, an anionic glutamic acid car-
between heavy and light chains so that each binding site is boxyl group may bind to a complementary cationic lysine
made up of the V H from one HL pair combining with the V L amino group on the antibody, or vice versa, or a glutamine
chain of the opposite pair. This arrangement allows the an- amide side chain may form a hydrogen bond with the an-
tibody to bind one branch of an oligosaccharide and the tibody. However, hydrophobic interactions can also play a
opposite branch of a nearby oligosaccharide and makes it major role. Proteins cannot exist in aqueous solution as sta-
ideally suited for cross-linking the densely clustered oligo- ble monomers with too many hydrophobic residues on their
saccharides of gp120. surface. Those hydrophobic residues that are on the surface
can contribute to binding to antibody for exactly the same
Immunogenicity of Polysaccharide Conjugates reason. When a hydrophobic residue in a protein antigenic
Capsular polysaccharides are the main target of protective determinant or, similarly, in a carbohydrate determinant8
antibodies against bacterial infection, and, as such, are im- interacts with a corresponding hydrophobic residue in the
portant vaccine antigens. In adults, the chain length of the antibody-combining site, the water molecules previously in
polysaccharide is an important determinant of immunoge- contact with each of them are excluded. The result is a sig-
nicity, and the polysaccharides induce a T-independent re- nificant stabilization of the interaction. A thorough review
sponse that cannot be boosted on repeat exposure. In young of these aspects of the chemistry of antigen–antibody bind-
children, whose maternal antibodies wane by 6 months of ing is in Getzoff et al.41
age and who most need immunity to pathogens such as
Haemophilus influenzae type b and Streptococcus pneumoniae Mapping Epitopes: Conformation versus Sequence
of multiple serotypes, the T-independent response to these The other component that defines a protein antigenic de-
polysaccharides is weak, regardless of chain length. To im- terminant, besides the amino acid residues involved, is the
munize children, the polysaccharides were coupled to a pro- way these residues are arrayed in three dimensions. As the
tein carrier to create a new T-dependent antigen that gained residues are on the surface of a protein, we can also think
immunogenicity from T-cell help and boosted antibody ti- of this component as the topography of the antigenic deter-
ters with each successive dose. This strategy has produced minant. Sela42 divided protein antigenic determinants into
highly successful conjugate vaccines against H. influenzae two categories, sequential and conformational, depending
type b,29 resulting in a markedly reduced incidence of men- on whether the primary sequence or the three-dimensional
ingitis caused by this agent in immunized children30,31 and conformation appeared to contribute the most to binding.
evidence of herd immunity even among unimmunized chil- On the other hand, as the antibody-combining site has a
dren. The same strategy has produced an effective vaccine preferred topography in the native antibody, it would seem a
against invasive disease32 and otitis media33 caused by the priori that some conformations of a particular polypeptide
most prevalent serotypes of S. pneumoniae. sequence would produce a better fit than others and there-
fore would be energetically favored in binding. Thus, con-
formation or topography must always play some role in the
Protein and Polypeptide Antigenic Determinants structure of an antigenic determinant.
Like the proteins themselves, the antigen determinants of Moreover, when one looks at the surface of a protein in
proteins consist of amino acid residues in a particular three- a space-fi lling model, one cannot ascertain the direction of
dimensional array. The residues that make contact with the backbone or the positions of the helices (contrast Figs.
complementary residues in the antibody-combining site are 23.3A and 23.3B).43–47 It is hard to recognize whether two
called contact residues. To make contact, of course, these residues that are side by side on the surface are adjacent on
residues must be exposed on the surface of the protein, not the polypeptide backbone or whether they come from dif-
buried in the hydrophobic core. As the complementarity- ferent parts of the sequence and are brought together by the
determining residues in the hypervariable regions of anti- folding of the molecule. If a protein maintains its native con-
bodies have been found to span as much as 30 to 40 Å × 15 formation when an antibody binds, then it must similarly

FIG. 23.3. A: Artist’s representation of the polypeptide backbone of sperm whale myoglobin in its native
three-dimensional conformation. The α helices are labeled A through H from the amino terminal to the
carboxyl terminal. Side chains are omitted, except for the two histidine rings (F8 and E7) involved with the
heme iron. Methionines at positions 55 and 131 are the sites of cleavage by cyanogen bromide (CNBr), al-
lowing myoglobin to be cleaved into three fragments. Most of the helicity and other features of the native
conformation are lost when the molecule is cleaved. A less drastic change in conformation is produced
by removal of the heme to form apomyoglobin, as the heme interacts with several helices and stabilizes
their positions relative to one another. The other labeled residues (Glu 4, Lys 49, Glu 83, Lys 140, Ala 144,
and Lys 145) are residues that have been found to be involved in antigenic determinants recognized by
monoclonal antibodies.43 Note that cleavage by CNBr separates Lys 79 from Glu 4 and separates Glu 83
from Ala 144 and Lys 145. The “sequential” determinant of Koketsu and Atassi44 (residues 15 to 22) is lo-
cated at the elbow, lower right, from the end of the A helix to the beginning of the B helix. (Adapted from
Dickerson.45) B: Stereoscopic views of a computer-generated space-filling molecular model of sperm
whale myoglobin, based on the Takano46 x-ray diffraction coordinates. This orientation, which corre-
sponds to that in Panel A, is arbitrarily designated the “front view.” The computer method was described
by Feldmann et al.47 The heme and aromatic carbons are shaded darkest, followed by carboxyl oxygens,
then other oxygens, then primary amino groups, then other nitrogens, and finally side chains of aliphatic
residues. The backbone and the side chains of nonaliphatic residues, except for the functional groups,
are shown in white. Note that the direction of the helices is not apparent on the surface, in contrast
to the backbone drawing in Panel A. The residues Glu 4, Lys 79, and His 12 are believed to be part of
a topographic antigenic determinant recognized by a monoclonal antibody to myoglobin.43 This stereo
pair can be viewed in three dimensions using an inexpensive stereoviewer such as the “stereoscopes”
sold by Abrams Instrument Corp. (Lansing, MI) or Hubbard Scientific Co. (Northbrook, IL). (Adapted from
Berzofsky et al.43)

be hard for the antibody to discriminate between residues folds in its native conformation have been called “assembled
that are covalently connected directly and those connected topographic” sites50,51 because they are assembled from dif-
only through a great deal of intervening polypeptide. Thus, ferent parts of the sequence and exist only in the surface to-
the probability that an antigenic determinant on a native pography of the native molecule. By contrast, the sites that
globular protein consists of only a consecutive sequence of consist of only a single continuous segment of protein se-
amino acids in the primary structure is likely to be rather quence have been called “segmental” antigenic sites.50,51
small. Even if most of the determinant were a continuous se- In contrast to T-cell recognition of “processed” frag-
quence, other nearby residues would probably play a role as ments retaining only primary and secondary structures, the
well. Only if the protein were cleaved into fragments before evidence is overwhelming that most antibodies are made
the antibodies were made would there be any reason to favor against the native conformation when the native protein
connected sequences. is used as immunogen. For instance, antibodies to native
This concept was analyzed and confirmed quantitatively staphylococcal nuclease were found to have about a 5000-
by Barlow et al.,48 who examined the atoms lying within fold higher affinity for the native protein than for the cor-
spheres of different radii from a given surface atom on a pro- responding polypeptide on which they were isolated (by
tein. As the radius increases, the probability that all the atoms binding to the peptide attached to Sepharose).52 An even
within the sphere will be from the same continuous segment more dramatic example is that demonstrated by Crumpton53
of protein sequence decreases rapidly. Correspondingly, the for antibodies to native myoglobin or to apomyoglobin.
fraction of surface atoms that would be located at the center Antibodies to native ferric myoglobin produced a brown
of a sphere containing only residues from the same continu- precipitate with myoglobin but did not bind well to apo-
ous segment falls dramatically as the radius of the sphere myoglobin, which, without the heme, has a slightly altered
increases. For instance, for lysozyme, with a radius of 8 Å, conformation. On the other hand, antibodies to the apo-
fewer than 10% of the surface residues would lie in such a myoglobin, when mixed with native (brown) myoglobin,
“continuous patch” of surface. These are primarily in reproduced a white precipitate. These antibodies so strongly
gions that protrude from the surface. With a radius of 10 favored the conformation of apomyoglobin, from which the
Å, almost none of the surface residues fall in the center of heme was excluded, that they trapped those molecules that
a continuous patch. Thus, for a contact area of about 20 Å vibrated toward that conformation and pulled the equilib-
× 25 Å, as found for a lysozyme–antibody complex studied rium state over to the apo form. One could almost say, figu-
by x-ray crystallography, none of the antigenic sites could ratively, that the antibodies squeezed the heme out of the
be completely continuous segmental sites (see following dis- myoglobin. Looked at it thermodynamically, it is clear that
cussion and Fig. 23.4). On the other hand, other analyses the conformational preference of the antibody for the apo
did not find a correlation of epitope residues with surface versus native forms, in terms of free energy, had to be greater
accessibility, suggesting that the situation is more complex.49 than the free energy of binding of the heme to myoglobin.
Antigenic sites consisting of amino acid residues that Thus, in general, antibodies are made that are very specific
are widely separated in the primary protein sequence but for the conformation of the protein used as immunogen.
brought together on the surface of the protein by the way it Other more recent examples also show that antibodies can
FIG. 23.4. Assembled Topographic Sites of

Lysozyme Illustrated by the Footprints of Three
Nonoverlapping Monoclonal Antibodies.
Shown are the α-carbon backbones of lysozyme
in the center and the Fv portions of three anti-
lysozyme monoclonal antibodies D1.3, HyHEL-5,
and HyHEL-10. The footprints of the antibodies
on lysozyme and lysozyme on the antibodies (ie,
their interacting surfaces) are shown by a dotted
representation. Note that the three antibodies
each contact more than one continuous loop of
lysozyme and so define assembled topographic
sites. Reproduced from Davies and Padlan37 with
permission.

enforce structures on disordered or denatured structures in conformation peptide. Moreover, there was no evidence to
proteins such as HIV-1 Tat54 or influenza hemagglutinin.55 exclude the participation of other residues, nearby on the
Synthetic peptides corresponding to segments of the surface of myoglobin but not in this sequence, in the anti-
protein antigen sequence can be used to identify the struc- genic determinant.71–74*
tures bound by antibodies specific for segmental antigenic A good example of the importance of secondary struc-
sites. To identify assembled topographic sites, more com- ture is the case of the loop peptide (residues 64 to 80) of
plex approaches have been necessary. The earliest was the hen egg white lysozyme.75 This loop in the protein sequence
use of natural variants of the protein antigen with known is created by the disulfide linkage between cysteine residues
amino acid substitutions, where such evolutionary vari- 64 and 80 and has been shown to be a major antigenic de-
ants exist.50 Thus, substitution of different amino acids in terminant for antibodies to lysozyme.75 The isolated peptide
proteins in the native conformation can be examined. The 60 to 83, containing the loop, binds antibodies with high
use of this method, which is illustrated later, is limited to affinity, but opening of the loop by cleavage of the disulfide
studying the function of amino acids that vary among ho- bond destroys most of the antigenic activity for antilyso-
mologous proteins, that is, those that are polymorphic. It zyme antibodies.75
may now be extended to other residues by use of site-di- At the other end of the range of conformational re-
rected mutagenesis. A second method is to use the antibody quirements are those determinants involving residues far
that binds to the native protein to protect the antigenic site apart in the primary sequences that are brought close to-
from modification56 or proteolytic degradation.57 A related gether on the surface of the native molecule by its folding
but less sensitive approach makes use of competition with in three dimensions, called assembled topographic deter-
other antibodies.58–60 A third approach, taking advantage minants.50,51 Of six monoclonal antibodies to sperm whale
of the capability of producing thousands of peptides on a myoglobin studied by Berzofsky et al.,43,76 none bound to
solid-phase surface for direct binding assays,61 is to study any of the three cyanogen bromide (CNBr) cleavage frag-
binding of a monoclonal antibody to every possible comments of myoglobin that together span the whole sequence
bination of six amino acids.61 If the assembled topographic of the molecule. Therefore, these monoclonal antibodies
site can be mimicked by a combination of six amino acids (all with affi nities between 2 × 108 and 2 × 109 M −1) were
not corresponding to any continuous segment of the protein all highly specific for the native conformation. These were
sequence but structurally resembling a part of the surface, studied by comparing the relative affi nities for a series
then one can produce a “mimotope” defining the specific- of native myoglobins from different species with known
ity of that antibody.61 Mimotopes have become widely used amino acid sequences. This approach allowed the defi ni-
and can be combined with mutational analysis to map as- tion of some of the residues involved in binding to three
sembled topographic epitopes.62 Mimetics have even been of these antibodies. Two of these three monoclonal anti-
made for quaternary structural epitopes.63 Many mimotope bodies were found to recognize topographic determinants,
approaches use phase display peptide libraries to map epit- as defi ned previously. One recognized a determinant in-
opes of monoclonal antibodies.64–66 However, other studies cluding Glu 4 and Lys 79, which come within about 2 Å
have been less optimistic about the ability to predict assem- of each other to form a salt bridge in the native molecule
bled topographic or discontinuous epitopes from mimotope (see Fig. 23.3A, B). The other antibody recognized a de-
binding67 or random peptide libraries.68 terminant involving Glu 83, Ala 144, and Lys 145 (see Fig.
Myoglobin also serves as a good model protein anti- 23.3A). Again, these are far apart in the primary sequence
gen for studying the range of variation of antigenic deter- but are brought within 12 Å of each other by the folding of
minants from those that are more sequential in nature to the molecule in its native conformation. Similar examples
those that do not even exist without the native conforma- have been reported for monoclonal antibodies to human
tion of the protein (see Fig. 23.3). A good example of the myoglobin77 and to lysozyme37,58 as well as the HIV-1 enve-
first more segmental type of determinant is that consisting lope protein (neutralizing epitopes)78,79 and the prion pro-
of residues 15 to 22 in the amino terminal portion of the tein.80 Other examples of such conformation-dependent
molecule. Crumpton and Wilkinson69 first discovered that antigenic determinants have been suggested using conven-
the chymotrypsin cleavage fragment consisting of residues tional antisera to such proteins as insulin,81 hemoglobin,82
15 to 29 had antigenic activity for antibodies raised to either tobacco mosaic virus,83 and cytochrome c.84 Moreover, the
native or apomyoglobin. Two other groups44,70 then found crystallographic structures of lysozyme-antibody34,36,37 and
that synthetic peptides corresponding to residues 15 to 22 neuraminidase-antibody35 complexes, as well as HIV-1 en-
bind antibodies made to native sperm whale myoglobin, velope antibody complexes,78,79 show clearly that, in both
even though the synthetic peptides were only seven to eight cases, the epitope bound is an assembled topographic site.
residues long. Peptides of this length do not spend much
time (in solution) in a conformation corresponding to that
of the native protein. On the other hand, these synthetic
*This is the only segmental antigenic determinant of myoglobin that has clearly been
peptides had a several hundred-fold lower affi nity for the confi rmed by more than one independent group of investigators. Crumpton and
antibodies than did the native protein. Thus, even if most of Wilkinson69 did measure antigenic activity for a chymotryptic fragment 147 to 153
the determinant was included in the consecutive sequence that overlaps one of the other reported sequential determinants.71 However, two of
the other reported sequential determinants,71 corresponding to residues 56 to 62 and
15 to 22, the antibodies were still much more specific for the 94 to 100, have not been reproducible when tested with other antisera, even raised in
native conformation of this sequence than for the random the same species.72 For related studies, see Hurrell et al.73 and East et al.74

In the case of the three monoclonal antibodies binding to fraction of the peptide molecules are in a native-like con-
nonoverlapping sites of lysozyme (Fig. 23.4), it is clear that formation at any time, assuming that the antibody binds
the footprints of all three antibody-combining sites cover only to the native conformation. Because the concentration
more than one loop of polypeptide chain, and thus, each of peptide molecules in native conformation is lower than
encompasses an assembled topographic site.37 This result the total peptide concentration by a factor that corre-
illustrates the concept that most antibody-combining sites sponds to the conformational equilibrium constant of the
must interact with more than a continuous loop of polypeptide, the apparent affi nity is also lower by this factor.
peptide chain and thus must defi ne assembled topographic This model is analogous to an allosteric model. A third,
sites.48 Another important example is represented by neu- intermediate hypothesis would suggest that initial binding
tralizing antibodies to the HIV envelope protein that simi- of the peptide in a nonnative conformation occurs with
larly bind assembled topographic sites85,86 (see the end of submaximal complementarity and is followed by an intra-
this section). molecular conformational change in the peptide to achieve
How frequent are antibodies specific for topographic energy minimization by assuming a native-like conforma-
determinants compared to those that bind consecutive se- tion. This third hypothesis corresponds to an induced fit
quences when conventional antisera are examined? This model. The loss of affi nity is due to the energy required
question was studied by Lando et al.,87 who passed goat, to change the conformation of the peptide, which in turn
sheep, and rabbit antisera to sperm whale myoglobin over corresponds to the conformational equilibrium constant
columns of myoglobin fragments, together spanning the in the second hypothesis. To some extent, these models
whole sequence. After removal of all antibodies binding to could be distinguished kinetically, as the fi rst hypothesis
the fragments, 30% to 40% of the antibodies remained that predicts a faster “on” rate and a faster “off ” rate than does
still bound to the native myoglobin molecule with high af- the second hypothesis.91 Such kinetic approaches have like-
finity but did not bind to any of the fragments in solution wise been used to support an “encounter-docking” model
by radioimmunoassay. Thus, in four of four antimyoglobin related to this concept.92
sera tested, 60% to 70% of the antibodies could bind pep- Although not the only way to explain the data, the second
tides, and 30% to 40% could bind only native-conformation hypothesis is useful because it provides a method to estimate
intact protein. the conformational equilibria of proteins and peptides.52,93
On the basis of studies such as these, it has been sug- The method assumes the second hypothesis, which can be
gested that much of the surface of a protein molecule may expressed as follows:
be antigenic,50,88 but that the surface can be divided up into
antigenic domains.43,73,74,77 Each of these domains consists
of many overlapping determinants recognized by different
antibodies.
An additional interesting point is that in three published
crystal structures of protein antigen–antibody complexes,
(1)
the contact surfaces were broad, with local complementary
pairs of concave and convex regions in both directions.34–37 where A = antibody, Pn = native peptide, and Pr = random
Thus, the concept of an antigen binding in the groove or conformation peptide so that
pocket of an antibody may be oversimplified, and antibod-
Kapparent = Kconf K native (2)
ies may sometimes bind by extending into pockets on an
antigen. Thus, the ratio of the apparent association constant for
Further information on the subjects discussed in this peptide to the measured association constant for the na-
section is available in the reviews by Sela,42 Crumpton,53 tive molecule should give the conformational equilibrium
Reichlin,89 Kabat,90 Benjamin et al.,50 Berzofsky,51 Getzoff constant of the peptide. Note the implicit assumption that
et al.,41 and Davis and Padlan.37 the total peptide concentration can be approximated by
[Pr]. This will generally be true, as most peptide fragments
Conformational Equilibria of Protein and Peptide of proteins demonstrate little native conformation, that is,
Antigenic Determinants Kconf = [Pn] / [Pr] is much less than one. Also note that if
There are several possible mechanisms to explain why an the first hypothesis (or third) occurs to some extent, this
antibody specific for a native protein will bind a peptide method will overestimate Kconf. On the other hand, if the af-
fragment in random conformation with lower affi nity. Of finity for the peptide is lower because it lacks some of the
course, the peptide may not contain all the contact resi- contact residues of the determinant, this method will under-
dues of the antigenic determinant so that the binding en- estimate Kconf (by assuming that all the affinity difference is
ergy would be lower. However, for cases in which all the due to conformation). To some extent, the two errors may
residues in the determinant are present in the peptide, partially cancel out. When this method was used to deter-
several mechanisms still remain. First, the affi nity may be mine the Kconf for a peptide from staphylococcal nuclease, a
lower because the topography of the residues in the peptide value of 2 × 10−4 (unitless because it is a ratio of two concen-
may not produce as complementary a fit in the antibody- trations) was obtained.52 Similarly, when antibodies raised
combining site as the native conformation would. Second, to a peptide fragment were used, it was possible to estimate
the apparent affi nity may be reduced because only a small the fraction of time the native nuclease spends in nonnative

conformations.93 In this case, the Kconf was found to be about In contrast, the ability of proteins to crystallize (a feature
3000-fold in favor of the native conformation. that allows the study of their structure by x-ray crystallog-
raphy) has long been taken as evidence of protein rigidity.100
Antipeptide Antibodies that Bind to Native Proteins at a In addition, the existence of discrete functional states of
Specific Site allosteric enzymes101 provides additional evidence of stable
In light of the conformational differences between native structural states of a protein. Finally, the fact that antibodies
proteins and peptides and the observed Kconf effects shown by can distinguish native from denatured forms of intact pro-
antibodies to native proteins when tested on the correspond- teins is well known for proteins such as myoglobin.53
ing peptides, it was somewhat surprising to find that anti- However, protein crystals are a somewhat artificial situ-
bodies to synthetic peptides show extensive cross-reactions ation, as the formation of the crystal lattice imposes order
with native proteins.94,95 These two types of cross-reactions on the components, each of which occupies a local energy
can be thought of as working in opposite directions: The minimum at the expense of considerable loss of randomness
binding of antiprotein antibodies to the peptide is inefficient, (entropy). Thus, the crystal structure may have artificial ri-
whereas the binding of antipeptide antibodies to the protein gidity that exceeds the actual rigidity of protein molecules
is quite efficient and commonly observed. This finding is in solution. On the contrary, we may attribute some of the
quite useful, as automated solid-phase peptide synthesis has considerable difficulty in crystallizing proteins to disorder
become readily available. This has been particularly useful within the native conformation. Second, allosterism may be
in three areas: exploitation of protein sequences deduced by explained by two distinct conformations that are discrete
recombinant deoxyribonucleic acid (DNA) methods, prepa- without being particularly rigid. Finally, the ability to gen-
ration of site specific antibodies, and the attempt to focus the erate antiprotein antibodies that are conformation specific
immune response on a single protein site that is biologically does not rule out the existence of antipeptide antibodies that
important but may not be particularly immunogenic. This are not. All antibodies are probably specific for some con-
section focuses on the explanation of the cross-reaction, uses formation of the antigen, but this need not be the crystallo-
of the cross-reaction, and the potential limitations regarding graphic native conformation in order to achieve a significant
immunogenicity. affinity for those proteins or protein segments that have a
The basic assumption is that antibodies raised against “loose” native conformation.
peptides in an unfolded structure will bind the correspond- Antipeptide antibodies have proved to be very powerful
ing site on proteins folded into the native structure.95 This is reagents when combined with recombinant DNA methods
not immediately obvious, as antibody binding to antigen is of gene sequencing.95,102 From the DNA sequence, the pro-
the direct result of the antigen fitting into the binding site. tein sequence is predicted. A synthetic peptide is con-
Affinity is the direct consequence of “goodness of fit” be- structed, coupled to a suitable carrier molecule, and used
tween antibody and antigen, whereas antibody specificity is to immunize animals. The resulting polyclonal antibod-
due to the inability of other antigens to occupy the same site. ies can be detected with a peptide-coated enzyme-linked
How then can the antipeptide antibodies overcome the ef- immunosorbent assay plate (see Chapter 7). They are used
fect of Kconf and still bind native proteins with good affinity to immunoprecipitate the native protein from a 35S-labeled
and specificity? The whole process depends on the antibody- cell lysate and thus confirm expression of the gene product
binding site forming a three-dimensional space and the an- in these cells. The antipeptide antibodies can also be used
tigen fi lling it in an energetically favorable way. to isolate the previously unidentified gene product of a new
Because the peptides are randomly folded, they rarely oc- gene. The site-specific antibodies are also useful in detecting
cupy the native conformation, so they are not likely to elicit posttranslational processing, as they bind all precursors and
antibodies against a conformation they do not maintain. If products that contain the site. In addition, because the an-
the antibodies are specific for a denatured structure, then, tibodies bind only to the site corresponding to the peptide,
like the myoglobin molecules that were denatured to apo- they are useful in probing structure–function relationships.
myoglobin by antibody binding,53 the cross-reaction may They can be used to block the binding of a substrate to an
depend on the native protein’s ability to assume different enzyme or the binding of a virus to its cellular receptor.
conformational states. If the native protein is quite rigid,
then the possibility of it assuming a random conformation is Immunogenicity of Proteins and Peptides
quite small; if it is a flexible three-dimensional spring, then Up to this point, we have considered the ability of antibod-
local unfolding and refolding may occur all the time. Local ies to react with proteins or peptides as antigens. However,
unfolding of protein segments may permit the immunologic immunogenicity refers to the ability of these compounds to
cross-reaction with antipeptide antibodies, as a flexible seg- elicit antibodies following immunization. Several factors
ment could assume many of the same conformations as the limit the immunogenicity of different regions of proteins,
randomly folded peptide.95 On the other hand, peptides with and these have been divided into those that are intrinsic to
more stable conformations may be more likely to elicit an- protein structure itself versus those extrinsic to the antigen
tibodies that bind both the peptide and the native protein.96 that are related to the responder and vary from one animal
To this end, scaffolding has been used to maintain the con- or species to another.51 In addition, we consider the special
formation of peptides or protein fragments to be used as im- case of peptide immunogenicity, as it applies to vaccine de-
munogens/vaccines, such as for respiratory syncytial virus97 velopment. The features of protein structure that have been
or HIV epitopes.98,99 suggested to explain the results include surface accessibility

of the site, hydrophilicity, flexibility, and proximity to a site A third factor suggested to play a role in immunogenic-
recognized by helper T cells. ity of protein epitopes is mobility. Measurement of mobility
When the x-ray crystallographic structure and antigenic in the native protein is largely dependent on the availability
structure are known for the same protein, it is not surprising of a high-resolution crystal structure, so its applicability is
to find that a series of monoclonal antibodies binding to a limited to only a small subset of proteins. Furthermore, it
molecule such as influenza neuraminidase choose an over- has been studied only for antibodies specific for segmental
lapping pattern of sites at the exposed head of the protein.103 antigenic sites; therefore, it may not apply to the large frac-
The stalk of neuraminidase was not immunogenic appar- tion of antibodies to assembled topographic sites. Studies of
ently because it was almost entirely covered by carbohydrate. mobility have taken two directions. The case of antipeptide
Beyond such things as carbohydrate, which may steri- antibodies has already been discussed, in which antibodies
cally interfere with antibody binding to protein, accessibil- made to peptides corresponding to more mobile segments
ity on the surface is clearly a sine qua non for an antigenic of the native protein were more likely to bind to the na-
determinant to be bound by an antibody specific for the tive protein.95,115 This is not considered just a consequence
native conformation, without any requirement for unfold- of the fact that more mobile segments are likely to be those
ing of the structure.51 Several measures of such accessibil- on the surface and therefore more exposed because in the
ity have been suggested. All these require knowledge of the case of myohemerythrin (which was used as a model), two
x-ray crystallographic three-dimensional structure. Some regions of the native protein that were equally exposed but
have measured accessibility to solvent by rolling a sphere less mobile did not bind nearly as well to the correspond-
with the radius of a water molecule over the surface of a pro- ing antipeptide antibodies.116 However, as is clear from the
tein.104,105 Others have suggested that accessibility to water is previous discussion, this result applies to antibodies made
not the best measure of accessibility to antibody and have against short peptides and therefore is not directly relevant
demonstrated a better correlation by rolling a sphere with to immunogenicity of parts of the native protein. Rather, it
the radius of an antibody-combining domain.106 Another concerns the cross-reactivity of antipeptide antibodies with
approach to predicting antigenic sites on the basis of acces- the native protein and therefore is of considerable practical
sibility is to examine the degree of protrusion from the sur- importance for the purposes outlined in the section on an-
face of the protein.107 This was done by modeling the body tipeptide antibodies.
of the protein as an ellipsoid and examining which amino Studies in the other direction—that is, of antibodies
acid residues remain outside ellipsoids of increasing dimen- raised against native proteins—would be by definition more
sions. The most protruding residues were found to be part relevant to the question of immunogenicity of parts of the
of antigenic sites bound by antibodies, but usually, these native protein. Westhof et al.117 used a series of hexapeptides
sites had been identified by using short synthetic peptides to determine the specificity of antibodies raised against na-
and so were segmental in nature. As noted previously, for tive tobacco mosaic virus protein and found that six of the
an antigenic site to be contained completely within a single seven peptides that bound antibodies to native protein corre-
continuous segment of protein sequence, the site is likely sponded to peaks of high mobility in the native protein. The
to have to protrude from the surface, as otherwise residues correlation was better than could be accounted for just by
from other parts of the sequence would fall within the area accessibility because three peptides that corresponded to ex-
contacting the antibody.48 However, inability of such surface posed regions of only average mobility did not bind antibod-
or protrusion information to predict antigenic sites has also ies to the native protein. However, when longer peptides—on
been encountered in some studies.49 the order of 20 amino acid residues—were used as probes, it
Because the three-dimensional structure of most proteins was found that antibodies were present in the same antisera
is not known, other ways of predicting surface exposure have that bound to less mobile regions of the protein.118 They sim-
been proposed for the vast majority of antigens. For exam- ply had not been detected with the short hexapeptides with
ple, hydrophilic sites tend to be found on the water-exposed less conformational stability. Thus, it was not that the more
surface of proteins. Thus, hydrophilicity has been proposed mobile regions were necessarily more immunogenic but
as a second indication of immunogenicity.108–110 This model rather that antibodies to these were more easily detected with
has been used to analyze 12 proteins with known antigenic short peptides as probes. A similar good correlation of anti-
sites: The most hydrophilic site of each protein was indeed genic sites with mobile regions of the native protein in the
one of the antigenic sites. However, among the limitations case of myoglobin117 may also be attributed to the fact that
are the facts that a significant fraction of surface residues seven of the nine sites were defined with short peptides of
can be nonpolar,104,105 and that several important examples six to eight residues.71 Again, this result becomes a statement
of hydrophobic and aromatic amino acids involved in the about cross-reactivity between peptides and native protein
antigenic sites are known.42,83,111,112 Specificity of antibody rather than about the immunogenicity of the native protein.
binding likely depends on the complementarity of surfaces For reviews, see Van Regenmortel119 and Getzoff et al.41
for hydrogen bonding and polar bonding as well as van der To address the role of mobility in immunogenicity, an
Waals contacts, whereas hydrophobic interactions and the attempt was made to quantitate the relative fraction of anti-
exclusion of water from the interacting surfaces of proteins bodies specific for different sites on the antigen myohemery-
may contribute a large but nonspecific component to the en- thrin.120 The premise was that, although the entire surface of
ergy of binding.113 Another study suggested that amino acid the protein may be immunogenic, certain regions may elicit
pairs were better predictors of epitopes.114 significantly more antibodies than others and therefore may

be considered immunodominant or at least more immu- silent residues.50,84,125 In another example, rabbit and dog
nogenic. Because this study was done with short synthetic antibodies to beef myoglobin bound almost equally well
peptides from 6 to 14 residues long based on the protein se- to beef or sheep myoglobin.126 However, sheep antibodies
quence, it was limited to the subset of antibodies specific for bound beef but not sheep myoglobin, even though these two
segmental antigenic sites. Among these, it was clear that the myoglobins differ by just six amino acids. Thus, the sheep
most immunogenic sites were in regions of the surface that immune system was able to screen out those clones that
were most mobile, convex in shape, and often of negative would be autoreactive with sheep myoglobin.
electrostatic potential. Other more recent studies corrobo- Ir genes of the host also play an important role in reg-
rate the greater immunogenicity of more flexible segments ulating the ability of an individual to make antibodies to
of protein structures.121 The role of these parameters has a specific antigen.127 These antigen-specific immune re-
been reviewed.41 sponse genes are among the major histocompatibility com-
These results have important practical and theoretical plex (MHC) genes that code for transplantation antigens.
implications. First, to use peptides to fractionate antiprotein Structural mutations, gene transfer experiments, and bio-
antisera by affinity chromatography, peptides correspond- chemical studies127 all indicate that Ir genes are actually
ing to more mobile segments of the native protein should be the structural genes for MHC antigens. The mechanism of
chosen when possible. If the crystal structure is not known, action of the MHC antigens works through their effect on
it may be possible to use peptides from amino or carboxyl helper T cells (described later in this chapter). There appear
termini or from exon–intron boundaries, as these are more to be constraints on which B and T cells of a given specificity
likely to be mobile.115 Second, these results may explain how can help,128,129 a process called T–B reciprocity.130 Thus, if Ir
a large but finite repertoire of antibody-producing B cells genes control helper T-cell specificity, they will in turn limit
can respond to any antigen in nature or even artificial anti- which B cells are activated and which antibodies are made.
gens never encountered in nature. Protein segments that are The immunogenicity of peptide antigens is also limited
more flexible may be able to bind by induced fit in an anti- by intrinsic and extrinsic factors. With less structure to go
body-combining site that is not perfectly complementary to on, each small peptide must presumably contain some non–
the average native structure.41,51 Indeed, evidence from the self-structural feature in order to overcome self-tolerance.
crystal structure of antigen–antibody complexes122–124 sug- In addition, the same peptide must contain antigenic sites
gests that mobility in the antibody-combining site as well that can be recognized by helper T cells as well as by B cells.
as in the antigen may allow both reactants to adopt more When no T-cell site is present, three approaches may be
complementary conformations on binding to each other, helpful: graft on a T-cell site, couple the peptide to a car-
that is, a two-way induced fit. A nice example comes from rier protein, or overcome T-cell nonresponsiveness to the
the study of antibodies to myohemerythrin,123 in which the available structure with various immunologic agents, such
data suggested that initial binding of exposed side chains of as interleukin 2.
the antigen to the antibody promoted local displacements An example of a biologically relevant but poorly immu-
that allowed exposure and binding of other, previously bur- nogenic peptide is the asparagine-alanine-asparagine-pro-
ied residues that served as contact residues. The only way line (NANP) repeat unit of the circumsporozoite (CS) pro-
this could occur would be for such residues to become ex- tein of malaria sporozoites. A monoclonal antibody to the
posed during the course of an induced fit conformational repeat unit of the CS protein can protect against murine ma-
change in the antigen.41,123 In a second very clear example of laria.131 Thus, it would be desirable to make a malaria vac-
induced fit, the contribution of antibody mobility to pep- cine of the repeat unit of Plasmodium falciparum (NANP) n.
tide binding was demonstrated for a monoclonal antibody However, only mice of one MHC type (H-2b) of all mouse
to peptide 75 to 110 of influenza hemagglutinin, which was strains tested were able to respond to (NANP) n.132,133 One ap-
crystallized with or without peptide in the binding site and proach to overcome this limitation is to couple (NANP) n to a
analyzed by x-ray crystallography for evidence of an induced site recognizable by T cells, perhaps a carrier protein such as
fit.124 Despite flexibility of the peptide, the antibody-binding tetanus toxoid.134 In human trials, this conjugate was weakly
site probably could not accommodate the peptide without a immunogenic and only partially protective. Moreover, as
conformational change in the third complementarity deter- helper T cells produced by this approach are specific for the
mining region of the heavy chain, in which an asparagine unrelated carrier, a secondary or memory response would
residue of the antibody was rotated out of the way to allow not be expected to be elicited by the pathogen itself.
a tyrosine residue of the peptide to fit in the binding pocket Another choice might be to identify a T-cell site on the CS
of the antibody.124 protein itself and couple the two synthetic peptides together
Regarding host-limited factors, immunogenicity is cer- to make one complete immunogen. The result with one such
tainly limited by self-tolerance. Thus, the repertoire of posite, called Th2R, was to increase the range of responding
tential antigenic sites on mammalian protein antigens such mouse MHC types by one, to include H-2k as well as H-2b.135
as myoglobin or cytochrome c can be thought of as greatly This approach has the potential advantage of inducing a
simplified by the sharing of numerous amino acids with state of immunity that could be boosted by natural exposure
the endogenous host protein. For mouse, guanaco, or horse to the sporozoite antigen. As CS-specific T and B cells are
cytochrome c injected into rabbits, each of the differences both elicited by the vaccine, natural exposure to the antigen
between the immunogen and rabbit cytochrome c is seen as could help maintain the level of immunity during the entire
an immunogenic site on a background of immunologically period of exposure.

Another strategy to improve the immunogenicity of Perhaps the reason that neutralizing sites are clustered near
peptide vaccines is to stimulate the T- and B-cell responses receptor-binding sites is that occupation of such sites by
artificially by adding interleukin 2 to the vaccine. Results antibody mimics events normally caused by binding to the
with myoglobin indicate that genetic nonresponsiveness can cellular receptor, causing the virus to prematurely trigger
be overcome by appropriate doses of interleukin 2.136 The its cell entry mechanisms. However, in order to transmit a
same effect was found for peptides derived from malaria physiologic signal, the antibody may need to bind viral cap-
proteins.137,137a sid proteins in the native conformation (especially assembled
One of the most important possible uses of peptide anti- topographic sites), which antipeptide antibodies may fail to
gens is as synthetic vaccines. However, even though it is possi- do. Antibodies of this specificity are similar to the viral re-
ble to elicit with synthetic peptides anti-influenza antibodies ceptors on the cell surface, some of which have been cloned
to nearly every part of the influenza hemagglutinin,94 anti- and expressed without their transmembrane sequences as
bodies that neutralize viral infectivity have not been elicited soluble proteins. The soluble recombinant receptors for po-
by immunization with synthetic peptides. This may reflect liovirus144 and HIV-1145–147 exhibit high-affinity binding to
the fact that antibody binding by itself often does not result the virus and potent neutralizing activity in vitro. The HIV-1
in virus inactivation. Viral inactivation occurs only when an- receptor, cluster of differentiation (CD)4, has been combined
tibody interferes with one of the steps in the life cycle of the with the human Ig heavy chain in a hybrid protein CD4-Ig,148
virus, including binding to its cell surface receptor, internal- which spontaneously assembles into dimers and resembles a
ization, and virus uncoating within the cell. Apparently, an- monoclonal antibody, in which the binding site is the same
tibodies can bind to most of the exposed surface of the virus as the receptor-binding site for HIV-1. In these recombinant
without affecting these functions. Only those antibodies that constructs, high-affinity binding depends on the native con-
bind to certain “neutralizing” sites can inactivate the virus. formation of the viral envelope glycoprotein gp120. Binding
In addition, as in the case of the VP1 coat protein of polio- of CD4 to gp120 elicits a conformational change exposing a
virus, certain neutralizing sites are found only on the native CD4-induced epitope, and fusions of CD4 domains to gp120
protein and not on the heat-denatured protein.138 Thus, not can be used as vaccines to elicit such antibodies.149
only the site but also the conformation that is bound by the For HIV-1, two types of neutralizing antibodies have
antibodies may be important for the antibody to inactivate been identified. The first type binds a continuous or segmen-
the virus. These sites may often be assembled topographic tal determinant, such as the “V3 loop” sequence between
sites not mimicked by peptide segments of the sequence. amino acids 296 and 331 of gp120.150–152 Antipeptide anti-
Perhaps binding of an antibody to such an assembled site bodies against this site can neutralize the virus.150 However,
can alter the relative positions of the component subsites so because this site is located in a highly variable region of the
as to induce an allosteric neutralizing effect. Alternatively, envelope, these antibodies tend to neutralize a narrow range
antibodies to such an assembled site may prevent a confor- of viral variants with nearly the same sequence as the im-
mational change necessary for activity of the viral protein. munogen. Even for this highly variable site, more broadly
One method of mapping neutralizing sites is based on neutralizing antibodies can be obtained that recognize con-
the use of neutralizing monoclonal antibodies. The virus served conformations.153–155 The second type of neutraliz-
is grown in the presence of neutralizing concentrations of ing antibody binds conserved sites on the native structure
the monoclonal antibody, and virus mutants are selected of gp120, allowing them to neutralize a broad spectrum of
for the ability to overcome antibody inhibition. These are HIV-1 isolates. These antibodies are commonly found in
sequenced, revealing the mutation that permits “escape” by the sera of infected patients,156 and a panel of neutralizing
altering the antigenic site for that antibody. This method has monoclonals derived from these subjects has been analyzed.
been used to map the neutralizing sites of influenza hemag- These monoclonals can be divided into three types. One
glutinin139 as well as poliovirus capsid protein VP1.140 The group, possibly the most common ones in human poly-
influenza escaping mutations are clustered to form an as- clonal sera, bind at or near the CD4 receptor–binding site
sembled topographic site, with mutations distant from each of gp120.79,157–161 A second type of monoclonal, called 2G12,
other in the primary sequence of hemagglutinin but brought binds a conformational site on gp120 that also depends on
together by the three-dimensional folding of the native pro- glycosylation, but has no direct effect on CD4 binding.162
tein. At first, it was thought that neutralization was the re- A third type, quite rare in human sera, is represented by
sult of steric hindrance of the hemagglutinin-binding site monoclonal antibody 2F5163 and binds a conserved site on the
for the cell surface receptor of the virus.141 However, simi- transmembrane protein gp41. Although this site is contained
lar work with poliovirus reveals that neutralizing antibod- on a linear peptide ELDKWA, antibodies such as 2F5 cannot
ies that bind to assembled topographic sites may inactivate be elicited by immunizing with the peptide, again suggest-
the virus at less than stoichiometric amounts, when at least ing the conformational aspect of this site.164,165 Indeed, the
half of the sites are unbound by antibody.142 The neutral- binding of antibodies to this membrane-proximal site has
izing antibodies all cause a conformational change in the even been found to involve interaction of the antibody with
virus, which is reflected in a change in the isoelectric point the lipid membrane.166 One might view this intriguing case
of the particles from pH 7 to pH 4.140,143 Antibodies that bind as an example of a discontinuous or assembled topographic
without neutralizing do not cause this shift. Thus, an alter- site created by the proximity of residues of the protein with
native explanation for the mechanism of antibody-mediated structures in the lipid membrane, thus, spanning more than
neutralization is the triggering of the virus to self-destruct. just different parts of the antigenic protein.

These monoclonals neutralize fresh isolates, as well as macrophage, dendritic cell, or B cell. The growth of T cells in
laboratory-adapted strains, and they neutralize viruses tropic culture is measured as the incorporation of [3H]thymidine
for T cells or macrophages,167 regardless of the use of CXCR4 into newly formed DNA. Under appropriate conditions, thy-
or CCR5 as second receptor. These monoclonals, which tar- midine incorporation increases with antigen concentration.
get different sites, act synergistically. A cocktail combining This assay permits the substitution of different APCs and is
all three types of monoclonals can protect monkeys against highly useful in defining the MHC and antigen-processing
iv challenge or vaginal challenge with a simian immunode- requirements of the APCs.
ficiency virus/HIV hybrid virus, indicating the potential for Using primarily this assay, several different approaches
antibodies alone to prevent HIV infection.168,169 Because each have been taken to mapping T-cell epitopes. First, T cells
of the three conserved neutralizing determinants depends on immunized to one protein have been tested for a prolifera-
the native conformation of the protein,170 a prospective gp120 tive response in vitro to the identical protein or to a series of
vaccine (or gp160 vaccine) would need to be in the native naturally occurring variants. By comparing the sequences of
conformation to be able to elicit these antibodies. stimulatory and nonstimulatory variants, it was possible to
identify potential epitopes recognized by T cells. For example,
the T-cell response to myoglobin was analyzed by immunizing
ANTIGENIC DETERMINANTS RECOGNIZED mice with sperm whale or horse myoglobin and testing the re-
BY T CELLS sulting T cells for proliferation in response to a series of myo-
Studies of T-cell specificity for antigen were motivated by globins from different species with known amino acid substi-
the fact that the immune response to protein antigens is tutions.173 Reciprocal patterns were observed in T cells from
regulated at the T-cell level. A hapten, not immunogenic by mice immunized with sperm whale or horse myoglobin. The
itself, will elicit antibodies only when coupled to a protein response to the cross-stimulatory myoglobins was as strong as
that elicits a T-cell response in that animal. This ability of to the myoglobin used to immunize the mice. This suggested
the protein component of the conjugate to confer immuno- that a few shared amino acid residues formed an immuno-
genicity on the hapten has been termed the “carrier effect.” dominant epitope, and that most substitutions had no effect
Recognition of the carrier by specific helper T cells induces on the dominant epitope. A comparison of the sequences re-
the B cells to make antibodies. Thus, the factors contribut- vealed that substitutions at a single residue could explain the
ing to a good T-cell response appear to control the B-cell pattern observed. All myoglobins that cross-stimulated sperm
response as well. whale–immune T cells had Glu at position 109, whereas all
“Nonresponder” animals display an antigen-specific fail- that cross-stimulated horse-immune T cells had Asp at 109.
ure to respond to a protein antigen, both for T cells and an- No member of one group could stimulate T cells from do-
tibody responses. The “high responder” phenotype for each nors immunized with a myoglobin of the other group. This
antigen is a genetically inheritable, usually dominant trait. suggested that an immunodominant epitope recognized by T
Using inbred strains of mice, the genes controlling the im- cells was centered on position 109, regardless of which amino
mune response were found to be tightly linked to the MHC acid was substituted. Usually, this approach has led to correct
genes.127,171 MHC-linked immune responsiveness has been localization of the antigenic site in the protein,173–175 but the
shown to depend on the T-cell recognition of antigen bound possibility of long-range effects on antigen processing must be
within a groove of MHC antigens of the antigen-presenting kept in mind (see the section on “Antigen Processing”). Also,
cell (APC) (discussed herein below and see Chapters 21 this approach using natural variants is limited in that it can
and 22). The recognition of antigen in association with focus on the correct region of the molecule but cannot de-
MHC molecules of the B cell is necessary for carrier-specific fine the boundaries of the site. Site-directed mutagenesis may
T cells to expand and provide helper signals to B cells. therefore expand the capabilities of this approach.
In contrast to the range of antigens recognized by antibod- A second approach is to use short peptide segments of the
ies, the repertoire recognized by helper and cytotoxic T cells protein sequence, taking advantage of the fact that T cells spe-
appears to be limited largely to protein and peptide antigens, cific for soluble protein antigens appear to see only segmental
although exceptions such as the small molecule tyrosine- antigenic sites not assembled topographic ones.127,176–180 These
azobenzene arsonate172 exist. Once the antigenic determinants may be produced by chemical or enzymatic cleavage of the
on proteins recognized by T cells are identified, it may be pos- natural protein,178–186 solid-phase peptide synthesis,185,187–190
sible to better understand immunogenicity and perhaps even or recombinant DNA expression of cloned genes or gene
to manipulate the antibody response to biologically relevant fragments.191 In the case of class I MHC molecule–restricted
antigens by altering the helper T–cell response to the antigen. cytotoxic T cells, viral gene deletion mutants expressing only
part of the gene product have also been used.192–194
In the case of myoglobin-specific T cells, mapping of an
Defining Antigenic Structures epitope to residue Glu 109 was confirmed by use of a synthetic
Polyclonal T-Cell Response peptide 102 to 118, which stimulated the T cells.189,195 The T
Significant progress in understanding T-cell specificity cells elicited by a myoglobin with either Glu or Asp 109 could
was made possible by focusing on T-cell proliferation in readily distinguish between synthetic peptides containing
vitro, mimicking the clonal expansion of antigen specific Glu or Asp at this position. Similar results were obtained
clones in vivo. The proliferative response depends on only with cytochrome c, where the predominant site recognized
two cells: the antigen-specific T cell and an APC, usually a by T cells was localized with sequence variants to the region

around residue 100 at the carboxyl end of cytochrome.174 Monoclonal T cells may be useful in identifying which of
Furthermore, the response to cytochrome c peptide 81 to the many proteins from a pathogen are important for T-cell
104 was as great as the response to the whole molecule. This responses. For instance, Young and Lamb203 have developed
indicated that a 24 amino acid peptide contained an entire a way to screen proteins separated by SDS–polyacrylamide
antigenic site recognized by T cells. The T cells could distin- gel electrophoresis and blotted onto nitrocellulose for stim-
guish between synthetic peptides with Lys or Gln at position ulation of T-cell clones and have used this to identify anti-
99, although both were immunogenic with the same MHC gens of Mycobacterium tuberculosis.204 Mustafa et al.205 have
molecule.196–198 This residue determined T-cell memory even used T-cell clones to screen recombinant DNA expres-
and specificity, and so presumably was interacting with the sion libraries to identify relevant antigens of Mycobacterium
T-cell receptor (TCR). A similar conclusion could be drawn leprae. Use of T cells to map epitopes has also been impor-
for residue 109 of myoglobin. However, this type of analysis tant in defining tumor antigens.206–211
must be used with caution. When multiple substitutions at Precise mapping of antigenic sites recognized by T cells
position 109 were examined for T-cell recognition and MHC was made possible by the fact that T cells would respond to
binding, residue 109 was found to affect both functions.199 peptide fragments of the antigen when they contain a com-
The ultimate use of synthetic peptides to analyze the seg- plete antigenic determinant. A series of overlapping peptides
mental sites of a protein that are recognized by T cells was can be used to walk along the protein sequence and find the
to synthesize a complete set of peptides, each staggered by antigenic site. Then, by truncating the peptide at either end,
just one amino acid from the previous peptide, correspond- the minimum antigenic peptide can be determined. For ex-
ing to the entire sequence of hen egg lysozyme.200 Around ample, in the case of myoglobin, a critical amino acid resi-
each immunodominant site, a cluster of several stimulatory due, such as Glu 109 or Lys 140, was found by comparing
peptides was found. The minimum “core” sequence con- the sequences of stimulatory and nonstimulatory myoglobin
sisted of just those residues shared by all antigenic peptides variants and large CNBr cleavage fragments212 as previously
within a cluster, whereas the full extent of sequences span- discussed, and then a series of truncated peptides containing
ning all stimulatory peptides within the same cluster defined the critical residue was synthesized with different overlap-
the “determinant envelope.” These two ways of defining an ping lengths at either end.185,189 Because solid-phase peptide
antigenic site differ, and one interpretation is that each core synthesis starts from a fi xed carboxyl end and proceeds to-
sequence corresponds to an MHC-binding site, whereas the ward the amino end, it can be stopped at various positions
determinant envelope includes the many ways for T cells to to produce a nested series of peptides that vary in length at
recognize the same peptide bound to the MHC. the amino end. In this way, it was found that two of the Glu
In each case, the polyclonal T-cell response could be 109–specific T-cell clones responded to synthetic peptides
mapped to a single predominant antigenic site. These results 102 to 118 and 106 to 118 but not to peptide 109 to 118.189 One
are consistent with the idea that each protein antigen has clone responded to peptide 108 to 118, whereas the other did
a limited number of immunodominant sites (possibly one) not. Thus, the amino end of the peptide recognized by one
recognized by T cells in association with MHC molecules of clone was Ser 108, whereas the other clone required Phe 106
the high-responder type. If none of the antigenic sites could and/or Ile 107. Similar fine specificity differences have been
associate with MHC molecules on the APCs, then the strain observed with T-cell clones specific for the peptides 52 to 61
would be a low responder, and the antigen would have little and 74 to 96 of hen egg lysozyme,182,213,214 the peptide 323 to
or no immunogenicity. 339 of chicken ovalbumin,183 and the peptide 81 to 104 of
pigeon cytochrome c188 : The epitopes recognized by several
Monoclonal T Cells T-cell clones overlap but are distinct. In addition, nine T-cell
Further progress in mapping T-cell sites depended on the clones recognized a second T-cell determinant in myoglobin
analysis of cloned T-cell lines. These were either antigen- located around Lys 140, and each one responded to the CNBr
specific T-cell lines made by the method of Komoto and cleavage fragment 132 to 153.215 Further studies with a nested
Fathman201 or T-cell hybridomas made by the method of series of synthetic peptides showed that the stimulatory se-
Kappler et al.202 In the former method, T cells are allowed quence is contained in peptide 136 to 145.185
to proliferate in response to antigen and APCs, rested, and These findings can be generalized to characterize a large
then restimulated again. After stimulation, the blasts can number of epitopes recognized by T cells from a number of
be cloned by limiting dilution and grown from a single protein antigens (Table 23.4).212–227 What these studies and
cell in the presence of interleukin 2. In the second method, others demonstrated about epitopes recognized by T cells is
enriched populations of antigen-specific T cells are fused that in each case, the entire site is contained on a short pep-
with a drug-sensitive T-cell tumor, and the fused cells are tide. MHC class I–restricted antigens also follow this rule,228
selected for their ability to grow in the presence of the drug. even when the protein antigen is normally expressed on the
Then the antigen specificity of each fused cell line must be surface of infected cells. This applies to viral glycoproteins,
determined. The key to determining this in a tumor line is such as influenza hemagglutinin, which are recognized by
that antigen-specific stimulation of a T-cell hybridoma re- cytolytic T cells after antigen processing229 (see section on
sults in release of interleukin 2, even though proliferation “Antigen Processing”). These peptides consist of no more
is constitutive. T cells produced by either method are useful than about 12 to 17 amino acid residues for class II MHC
in defining epitopes, measuring their MHC associations and or 8 to 10 residues for class I. Within this size, they must
studying antigen-processing requirements. contain all the information necessary to survive processing

sites tend to be the focus for a polyclonal response of a num-

Examples of Immunodominant T-Cell ber of distinct T-cell clones recognizing overlapping subsites
TABLE 23.4 Epitopes Recognized in Association within the antigenic site or having different sensitivities to sub-
with Class II MHC Molecules stitutions of amino acids within the site.182,183,188,189,200,213,214,237
Immunodominant antigenic sites appear to be qualita-
T-Cell Antigenic
Sites and Amphipathic tively different from other sites. For example, in the case of
Protein Reference Segments myoglobin, when the number of clones responding to dif-
ferent epitopes after immunization with native protein was
Sperm whale myoglobin 69–78148 64–78 quantitated by limiting dilution, it was observed that the bulk
102–118159 99–117 of the response to the whole protein in association with the
132–145155 128–145 high-responder class II MHC molecules was focused on a sin-
Pigeon cytochrome c 93–104158 92–103
gle site within residues 102 to 118195 (Fig. 23.5). When T cells in
Beef cytochrome c 11–25192 9–29
the (high × low responder) F1 hybrid restricted to each MHC
66–80193 58–78
haplotype were compared, there was little difference in the
Influenza 109–119186 97–120
responses to nondominant epitopes, and all the overall dif-
Hemagglutinin 130–140187 —
ference in magnitude of response restricted by the high versus
A/PR/8/34 302–313187,188 291–314
low responder MHC could be attributed to the high response
Pork insulin B 5–16157 4–16
A 4–14189 1–21
to the immunodominant determinant in the former and the
Chicken lysozyme 46–61185 — complete absence of this response in the latter (see Fig. 23.5).
74–86184 72–86 Similar results were found for two different high-responder
81–96175 86–102 and two different low-responder MHC haplotypes.195 Why did
109–119145 — the response to the other sites not compensate for the lack of
Chicken ovalbumin 323–339153 329–346 response to the immunodominant site in the low responders?
Foot and mouth virus VP1 141–160191 148–165
Hepatitis B virus
Pre-S 120–132190 121–135
Major surface antigen 38–52194 36–49
95–109194 —
140–154194 —
λ Repressor protein Cl 12–26195 8–25
Rabies virus spike 32–44196 29–46
glycoprotein
precursor
Adapted with permission from Schwartz et al.188
within the APC, associate with the MHC antigen, and bind
to the TCR, as discussed in the following sections.
Sequential Steps that Focus the T-Cell Response

on Immunodominant Determinants
In contrast to antibodies that bind all over the surface of
a native protein50 (see “Protein and Polypeptide Antigenic FIG. 23.5. Frequency of High- and Low-Responder Major Histo-
Determinants” section), it has been observed that T cells elic- compatibility Complex (MHC)-Restricted T Cells in F1 Hybrid. High
ited by immunization with the native protein tend to be fo- responsiveness may be accounted for by the response to a single
cused on one or a few immunodominant sites.230–232 This is true immunodominant epitope. Lymph node T cells from (low-responder
whether one deals with model mammalian or avian proteins [H-2k] × high-responder [H-2d]) F1 hybrid mice immunized with whole
myoglobin were plated at different limiting dilutions in microtiter
such as cytochrome c,179 myoglobin,178,180 lysozyme,182,214,233,234
wells with either high- or low-responder presenting cells and myo-
insulin,187,219 and ovalbumin,183 or with bacterial, viral, and globin as antigen. The cells growing in each well were tested for
parasitic proteins from pathogens, such as influenza hemag- responsiveness to whole myoglobin and to various peptide epitopes
glutinin217 or nucleoprotein (NP),228 staphylococcal nuclease,235 of myoglobin. The frequency of T cells of each specificity and MHC
or malarial CS protein.135,236 Because the latter category of pro- restriction was calculated from Poisson statistics and is plotted on
teins shares no obvious homology to mammalian proteins, the the ordinate. Most of the difference in T-cell frequency between
immunodominance of a few sites cannot be attributed simply high- and low-responder restriction types (solid bars) can be ac-
counted for by the presence of T cells responding to the immuno-
to tolerance for the rest of the protein because of homologous dominant site at residues 102 to 118, accounting for more than two-
host proteins. Moreover, immunodominance is not simply the thirds of the high-responder myoglobin-specific T cells, in contrast
preemption of the response by a single clone of predominant to the absence of such T cells restricted to the low-responder MHC
T cells because it has been observed that immunodominant type. (Based on the data in Kojima et al.195.)

The greater frequency of T cells specific for the immunodomi- myoglobin, a site of equine myoglobin (residues 102 to 118) that
nant site may in part be attributed to the large number of ways did not elicit a response when H-2k mice were immunized with
this site can be recognized by different T-cell clones, as men- native myoglobin nevertheless was found to be immunogenic
tioned previously, but this only pushes the problem back one when such mice were immunized with the peptide.238 Thus,
level. Why is an immunodominant site the focus for so many the low responsiveness to this site in mice immunized with the
different T-cell clones? Because the answer cannot depend on native myoglobin was not due to either of the classical mecha-
any particular T cell, it must depend on other factors primar- nisms of Ir gene defects—namely, a hole in the T-cell repertoire
ily involved in the steps in antigen processing and presenta- or a failure of the site to interact with MHC molecules of that
tion by MHC molecules. strain. However, the peptide-immune T cells responded only
It has also been observed that some peptides may be im- poorly to native equine myoglobin in vitro. Thus, the peptide
munogenic themselves, but the T-cell response they elicit and the native molecule did not cross-react well in either di-
is specific only for the peptide and does not cross-react rection. The problem was not simply a failure to process the
with the native protein nor do T cells specific for the na- native molecule to produce this epitope because (H-2k × H-2s)
tive protein recognize this site.238–240 These are called cryp- F1-presenting cells could present this epitope to H-2s T cells
tic determinants.240 The reasons for these differences may when given native myoglobin but could not present it to H-2k T
involve the way the native protein is processed to produce cells. Also, because the same results applied to individual T-cell
fragments distinct from, but including or overlapping, the clones, the failure to respond to the native molecule was ap-
synthetic peptides used in experiments and also the com- parently not due to suppressor cells induced by the native mol-
petition among sites within the protein for binding to the ecule. Similar observations were made for the response to the
same MHC molecules, as discussed further in the next sec- peptide 74 to 96 of hen lysozyme in B10.A mice.239 The peptide,
tion. To understand these factors that determine dominance not the native molecule, induced T cells specific for this site,
or crypticity, one must understand the steps through which and these T cells did not cross-react with the native molecule.
an antigen must go before it can stimulate a T-cell response. With these alternative mechanisms excluded, we are left with
Unlike B cells, T-cell recognition of antigen depends on the the conclusion that an appropriate peptide was produced, but
function of another cell, the APC.241 Antigen must pass through it differed from the synthetic peptide in such a way that a hin-
a number of intracellular compartments and survive process- dering site outside the minimal antigenic site interfered with
ing and transport steps before it can be effectively presented to presentation by presenting cells of certain MHC types. Further
T cells. Following antigen synthesis in the cell (as in a virally in- evidence consistent with this mechanism came from the work
fected cell) or antigen uptake via phagocytosis, pinocytosis, or, of Shastri et al.,242 who found that different epitopes within the
in some cases, receptor-mediated endocytosis, the subsequent 74 to 96 region of lysozyme were immunodominant in H-2b
steps include 1) partial degradation (“processing”) into dis- mice when different forms of the immunogen were used.
crete antigenic fragments that can be recognized by T cells, 2) Another line of evidence came from fine specificity studies of
transport of these fragments into a cellular compartment where individual T-cell clones. Shastri et al.243 observed that H-2b T-cell
MHC binding can occur, 3) MHC binding and assembly of a clones specific for hen lysozymes were about 100-fold more sen-
stable peptide-MHC complex, and 4) recognition of that pep- sitive to ring-necked pheasant lysozyme than to hen lysozyme.
tide-MHC complex by the expressed T-cell repertoire. At each Nevertheless, they were equally sensitive to the CNBr cleavage
step, a potential antigenic determinant runs the risk of being fragments containing the antigenic sites from both lysozymes.
lost from the process, for example, by excessive degradation or Thus, regions outside the minimal antigenic site removable by
failure to meet the binding requirements needed for transport CNBr cleavage presumably interfered with processing, presen-
to the next step. Only those peptides that surmount the four se- tation, or recognition of the corresponding site in hen lysozyme.
lective hurdles will prove to be antigenic for T cells. We will now Similarly, it was observed that a T-cell clone specific for sperm
consider each step in detail, for its contribution to the strength whale myoglobin, not equine myoglobin, responded equally
and specificity of the T-cell response to protein antigens. well to the minimal epitope synthetic peptides from the two
species.238 Also, residues outside the actual site must be distin-
Antigen Processing guishing equine from sperm whale myoglobin. Experiments
Influence of Antigen Processing on the Expressed T-Cell using F1-presenting cells that can clearly produce this epitope
Repertoire. Several lines of evidence indicate that antigen for presentation to other T cells proved that the problem was
processing plays a critical role in determining which po- not a failure to produce the appropriate fragment from hen
tential antigenic sites are recognized and, therefore, what lysozyme239 or equine myoglobin.238 Thus, these cases provide
part of the potential T-cell repertoire is expressed upon evidence that a structure outside the minimal site can hinder
immunization with a protein antigen. Because the T cell presentation in association with a particular MHC molecule.
does not see the native antigen but only the products of Such a hindering structure was elegantly identified in a
antigen processing, it is not unreasonable that the nature of study by Grewal et al.244 comparing hen egg lysozyme pep-
these products would at least partly determine which poten- tides presented by strains C57BL/6 and C3H.SW that share
tial epitopes could be recognized by T cells. H-2b but differ in non-MHC genes. After immunization
One line of evidence that processing plays a major role in with whole lysozyme, a strong T-cell response was seen to
T-cell repertoire expression came from comparisons that were peptide 46 to 61 in C3H.SW mice but not at all in C57BL/6
made of the immunogenicity of peptide versus native mol- mice. Because the F1 hybrids of these two strains responded,
ecule in the cases of myoglobin238 or lysozyme.239 In the case of the lack of response in one strain was not due to a hole in

the T-cell repertoire produced by self-tolerance. It was due to competition from Ed, which may preempt by binding
found that peptide 46 to 60 bound directly to the I-Ab class the 108 to 120 site with higher affinity and preventing the 13
II MHC molecule, whereas peptide 46 to 61 did not, indicat- to 35 site from binding to Ad. Competition between different
ing that the C-terminal Arg at position 61 hindered binding. peptides binding to the same MHC molecule could also occur.
Evidently, a non–MHC-linked difference in antigen pro- All these results, taken together, indicate that antigen
cessing allowed this Arg to be cleaved off the 46 to 61 pep- processing not only facilitates interaction of the antigenic
tide in C3H.SW mice, in which the peptide was dominant, site with the MHC molecule and/or the TCR but also influ-
but not in C57Bl/6 mice, in which the peptide was cryptic. ences the specificity of these interactions and, in turn, the
Even a small peptide that does not need processing may specificity of the elicited T-cell repertoire. The molecular
nevertheless be processed, and that processing may affect its in- mechanisms behind such effects are just now being eluci-
teraction with MHC molecules. Fox et al.245 found that substi- dated, as described in the following sections.
tution of a tyrosine for isoleucine at position 95 of cytochrome
c peptide 93 to 103 enhanced presentation with Eβb but di- Processing of Antigen for T Cells Restricted to Class II Major
minished presentation with Eβk when live APCs were used but Histocompatibility Complex Molecules. It has long been
not when the APCs were fixed and could not process antigen. known that T-cell responses such as delayed hypersensitiv-
Therefore, the tyrosine residue was not directly interacting with ity in vivo or T-cell proliferation in vitro to exogenous pro-
the different MHC molecule but was affecting the way the pep- teins can be stimulated not only by the native protein but also
tide was processed, which in turn affected MHC interaction. by denatured protein176 and fragments of native protein.219
Besides the mechanisms suggested previously, Gammon Indeed, this feature, along with the requirement for recog-
et al.239 and Sercarz et al.246 have proposed the possibility of nition in association with class II MHC molecules, distin-
competition between different MHC-binding structures guishes T- from B-cell responses. In a number of cases, the
(“agretopes”) within the same processed fragment. If a par- site recognized by cloned T cells has been located to a discrete
tially unfolded fragment first binds to MHC by one such site synthetic peptide corresponding to a segment of the primary
already exposed, further processing may stop, and other po- sequence of the protein. Examples include insulin,187,219 cyto-
tential binding sites for MHC may never become accessible for chrome c,188 lysozyme,182,213 and myoglobin.178,185,189 In each
binding. Such competition could also occur between different case, the stimulatory peptide must contain all the informa-
MHC molecules on the same presenting cell.239 For instance, tion required for antigen presentation and T-cell stimulation.
BALB/c mice, expressing both Ad and Ed, produce a response The lack of conformational specificity does not indicate a lack
to hen lysozyme specific for 108 to 120, not for 13 to 35,239 of TCR specificity. Rather, it results from antigen processing
and this response is restricted to Ed. However, B10.GD mice into peptide fragments that destroys conformational differ-
that express only Ad respond well to 13 to 35 when immu- ences prior to binding the TCR. One way to accomplish this is
nized with lysozyme. The BALB/c mice clearly express an Ad via antigen processing, which involves the partial degradation
molecule, so the failure to present this 13 to 35 epitope may be of a protein antigen into peptide fragments (Fig. 23.6).
FIG. 23.6. Steps in Antigen Presentation

by Class II Major Histocompatibility
Complex (MHC) Molecules. Soluble anti-
gen enters the presenting cell by phago-
cytosis, pinocytosis, or receptor-mediated
endocytosis. It is partially degraded to
peptide fragments by acid-dependent
proteases in endosomes. Antigenic pep-
tides associate with MHC class II mol-
ecules (I-A or I-E in the mouse) to form
an antigenic complex that is transported
to the cell surface. Before an MHC class
II molecule can bind the peptide, it must
release the class II–associated invari-
ant chain peptide fragment of invariant
chain from the binding groove, which is
catalyzed by human leukocyte antigen
(HLA)-DM. Binding of T-cell receptors to
the peptide-MHC complex triggers T-cell
proliferation, resulting in clonal expansion
of antigen-specific T cells.

Evidence of processing came from the fact that a single

protein antigen could stimulate T cells to different epitopes,
each specific for a different MHC antigen. For example,
when a series of myoglobin-specific T-cell clones were tested
for both antigen specificity and MHC restriction, six clones
were specific for a site centering on amino acid Glu 109,
and all six recognized the antigen in association with I-Ad.
Nine additional T-cell clones were specific for a second epi-
tope centered on Lys 140 and were restricted to a different
MHC antigen, I-Ed. Thus, the antigen behaved as if it was
split up into distinct epitopes, each with its own ability to
bind MHC.215
That T cells recognize processed antigen was demon-
strated by the fact that inhibitors of processing can block
antigen presentation. Early experiments by Ziegler and
Unanue247 showed that processing depends on intracellular
degradative endosomes, as drugs such as chloroquine and
ammonium chloride (NH4 Cl), which raise endosomal pH
and inhibit acid-dependent proteases, could block the pro-
cess. However, prior degradation of proteins into peptide
fragments allows them to trigger T cells even in the presence
of these inhibitors of processing.248 For example, T-cell clone
14.5 recognizes the Lys 140 site of myoglobin equally well on
the antigenic peptide (residues 132 to 153) as on the native
protein (Fig. 23.7). The difference between these two forms
of antigen is brought out by the presence of processing in-
hibitors. Leupeptin, for example, inhibits lysosomal prote-
ases and blocks the T-cell responses to native myoglobin but FIG. 23.7. Inhibition of Antigen Presentation by the Protease Inhibitor
not to peptide 132 to 153. Thus, native myoglobin cannot Leupeptin: Differential Effect on Presentation of the Same Epitope
stimulate T cells without further processing, whereas the of Native Myoglobin or Peptide 132 to 153 to the Same Monoclonal
peptide requires little or no additional processing.249 T-Cell Population. Splenic cells from nonimmunized B10.D2 mice, as
Why is antigen processing necessary? For class II MHC a source of antigen-presenting cells, were incubated with leupeptin
molecules, experiments suggest that antigen processing may at the concentration indicated for 15 minutes prior to and during expo-
sure to 2 mM native myoglobin or 1 mM peptide fragment, washed, irra-
uncover functional sites that are buried in the native protein diated, and cultured at 400,000 cells per well with 10,000 T cells of clone
structure. For example, a form of intact myoglobin that has 14.5; thymidine incorporation was measured after 4 days of culture.249
been partially unfolded through chemical modification can
behave like a myoglobin peptide and can be presented by
APC even in the presence of enough protease inhibitor or unfolding.251 Therefore, large size does not mandate process-
chloroquine to completely block the presentation of native ing, and small size does not necessarily obviate the need for
myoglobin.249 Denatured lysozyme could also be presented processing, at least for class II presentation. The common
without processing to one T-cell clone.184 This result sug- feature throughout the size range seems to be the need for
gests that the requirement for processing may simply be a unfolding. This evidence, taken together with the earlier
steric requirement, that is, to uncover the two sites needed to data on unfolding of myoglobin and lysozyme, strongly
form the trimolecular complex between antigen and MHC supports the conclusion that unfolding, rather than size
and between antigen and TCR. Thus, unfolding may be suf- reduction, is the primary goal of antigen processing, and
ficient without proteolysis, and proteolysis may simply ac- that either antigen presentation by MHC molecules or TCR
complish an unfolding analogous to Alexander’s approach recognition frequently requires exposure of residues not
to the Gordian knot. normally exposed on the surface of the native protein. This
The importance of antigen unfolding for T-cell recogni- conclusion is supported by recent studies of peptides eluted
tion and the ability of unfolding to bypass the need for an- from class II MHC molecules, and the crystal structures of
tigen processing apply to a range of polypeptide sizes from class II MHC-peptide complexes, which show that longer
small peptides to extremely large proteins. At one extreme, peptides can bind with both ends extending beyond the two
Lee et al.250 found that even fibrinogen, of Mr 340,000, does ends of the MHC groove252–254 (see “Antigen Interaction with
not need to be processed if the epitope recognized is on the MHC Molecules” section).
carboxy-terminal portion of the α chain, which is naturally Besides proteolysis, unfolding may require the reduction of
unfolded in the native molecule. At the other extreme, even disulfide bonds between or within protein antigens. A gamma
a small peptide of only 18 amino acid residues, apamin, re- interferon–inducible lysosomal thiol reductase (GILT) is
quires processing unless the two disulfide bonds that hold it expressed in APCs and localizes to the late endosomal and
in the native conformation are cleaved artificially to allow lysosomal compartments where MHC class II peptide loading

occurs.255 Unlike thioredoxin, this enzyme works at the acid with the protease activities arrayed on the inner surface, and
pH of endosomes and uses Cys but not glutathione as a re- unfolded proteins are believed to enter the barrel at one end,
ducing agent. APCs from GILT knockout mice were tested for leaving as peptides at the other end. The different proteases
the ability to present hen egg lysozyme to hen egg lysozyme– cut preferentially after aromatic or branched chain amino
specific T-cell lines.256 For two epitopes, the T-cell response acids (chymotryptic-like activity of the β5 subunit), basic
was insensitive to the GILT defect, even though they involved amino acids (trypsin-like activity of the β2 subunit), or
a disulfide bond in the native protein. But for one epitope, acidic residues (glutamate preferring of the β1 subunit).262,263
located between disulfide bonds, the T-cell response was com- Protease activity is increased against misfolded proteins,
pletely inhibited when the APCs lacked GILT reductase. In such as senescent proteins, which are tagged with ubiquitin
this case, reduction of disulfide bonds was an essential step and directed to the proteasome. In addition, viral proteins
for antigen presentation, presumably needed to generate free produced during infection and proteins synthesized with ar-
peptides for MHC class II binding. tificial amino acids are particularly susceptible to degrada-
tion by proteasomes. The products of protease digestion are
Processing of Antigen for T Cells Restricted to Class I Major peptides ranging from to 3 to 14 amino acids in length, in-
Histocompatibility Complex Molecules. Early studies on cluding 9-mers, of just the right size for MHC binding. The
class I–restricted T cells, such as cytolytic T cells (CTLs) spe- chymotrypsin- and trypsin-like activities may be particu-
cific for virus-infected cells, assumed that they responded larly important for antigenic peptides, as many peptides that
mainly to unprocessed viral glycoproteins expressed on the naturally bind MHC end in hydrophobic or basic residues.264
surface of infected cells. However, since the mid-1980s, it has The proteasome is the major processing machinery
been clear that CTLs, like other T cells, recognize processed of the nonendosomal processing pathway. This is shown
antigens. For example, influenza NP was a major target an- by the effect of proteasome inhibitors on MHC class I as-
tigen for influenza specific CTLs, even though NP remains sembly and antigen presentation and by the effect of Large
in the nucleus of infected cells and none is detectable on the Multifunctional Peptidase (LMP-2) and LMP-7 mutations
cell surface.257 Further support came from the finding that on antigen processing. A family of proteasome inhibitors
target cells that take up synthetic NP peptide 366 to 379 have been described262,263,265 that consist of short peptides,
were lysed by NP-specific CTLs.228 This constitutes evidence three to four amino acids in length, ending in an aldehyde,
that antigen presented in association with class I molecules such as Ac-Leu Leu norLeu-al, carbobenzoxy-Leu Leu nor-
requires processing into antigenic fragments. Also, the dem- Val-al,263 or nonpeptides such as lactacystin.266 Although the
onstration that synthetic peptides could sensitize targets for peptides appear to be directed primarily at the chymotryp-
CTLs introduced a powerful tool for mapping and studying sin-like protease activity, as false substrates, in fact, they in-
CTL epitopes. hibit all three types of protease activity.
Even for influenza hemagglutinin, which is expressed on By inhibiting antigen processing, these inhibitors induce a
the surface of infected cells, surface expression was not re- phenotype of reduced expression of MHC class I and inabil-
quired for antigenicity, implying that it is the processed anti- ity to present antigen to class I–restricted CTL.263 The MHC
gen that stimulates a T-cell response. Target cells expressing class I heavy chains remain in the ER, as shown by failure to
leader-negative hemagglutinin, which is not transported to become resistant to endoglycosidase H,267 which occurs in the
the cell surface but remains in the cytosol, were lysed equally Golgi. They are also unable to form stable complexes with β2-
well as those with surface hemagglutinin.229 Similar conclu- microglobulin due to a lack of peptides. These effects are spe-
sions were drawn from anchor-negative mutants.258 Indeed, cific for the protease function because the inhibitors do not
studies of HIV-1 gp160 genes with or without a leader se- block presentation of synthetic peptides, which also rescue
quence suggest that removal of the leader sequence can in- MHC class I expression, and because inhibition is reversible
crease the amount of protein that is retained in the cytosol when inhibitor is removed. These results suggest that protea-
and is available for processing and presentation through the somes are the primary supplier of antigenic peptides for class
class I MHC processing pathway.259 The explanation may be I, as other pathways are unable to compensate. However, it
that the signal peptide results in cotranslational translocation is also possible that the inhibitors could block other poten-
of the growing peptide chain into the endoplasmic reticulum tial processing enzymes as well. An alternative processing
(ER), whereas proteins without a signal peptide remain in pathway that bypasses the proteasome is provided by signal
the cytosol, where they are accessible to the processing ma- peptidase. As signal peptides are cleaved from proteins enter-
chinery of the class I pathway (see subsequent discussion). ing the ER, these hydrophobic peptides can bind MHC class
This cytosolic protein processing machinery consists pri- I.268 Particularly for MHC molecules such human leukocyte
marily of the 26S proteasomes.260,261 The specificity of such antigen (HLA)-A2, which prefer hydrophobic sequences, this
proteasomes to cleave at certain positions in a protein se- peptidase can be an alternative source of antigenic peptides
quence thus provides the first hurdle that a potential epitope that are independent of proteasomes and transporter associ-
must surmount to be presented by class I MHC molecules ated with antigen presentation (TAP)-1/2 transport (see sec-
to be cut out correctly but not destroyed by the proteasome. tion on “Transport”)because they are formed inside the ER.
In the standard proteasome, 14 distinct subunits assem- The proteins destined for proteasomal processing include
ble to form a high-molecular weight complex of about 580 some normally short-lived proteins with a half-life of about
kD with three distinct protease activities located on differ- 10 minutes, which constitute about 25% of the proteins in the
ent subunits. The proteasome is a barrel-shaped structure, cell. The rest arise from long-lived proteins, with a half-life

of about 1 day, which may be synthesized incorrectly. These increase production of epitopes from adenovirus277 and lym-
defective ribosomal products are ubiquitinated and marked phocytic choriomeningitis virus.278 Similarly, certain epitopes
for rapid degradation in proteasomes.269,270 In a normal cell, of latent membrane protein 2 of Epstein-Barr virus depended
these can arise from errors in ribonucleic acid transcription, on immunoproteasomes.279 In that case, the requirement for
protein translation, assembly, folding, or targeting. But in a immunoproteasomes depended on the context of the peptide
virally infected cell, misfolded viral proteins provide a ready within the native protein. Incomplete protein synthesis due
supply of antigenic peptides for antigen presentation and to puromycin or expression of the epitope surrounded by
T-cell recognition almost as soon as the virus starts to pro- protein fragments with fewer membrane spanning domains
duce new viral proteins. Similarly, incorporation of amino allowed epitope generation by constitutive proteasomes.
acid analogues, such as canavavanine in place of arginine, On the other hand, some epitopes are generated more
creates misfolded proteins that are rapidly processed and effectively by the constitutive proteasome than the immu-
more efficiently presented via the proteasomal pathway.271 noproteasome.275,280 When the repertoire of seven defined
Interestingly, the MHC itself encodes, near the class II re- class I MHC-restricted epitopes was compared in an elegant
gion, three proteins, known as LMP-2 and LMP-7 for “low- quantitative study in LMP-2–deficient or wild-type C57BL/6
molecular-weight protein,” and MECL-1, for “multicatalytic mice, it was found that responses to the two epitopes that are
endopeptidase complex-like 1,” which contribute to the pro- immunodominant in wild-type mice were greatly reduced
teasome structure. The LMP-2, MECL-1, and LMP-7 subunits in the LMP-2–deficient mice, which lack immunoprotea-
are upregulated by interferon-γ, and substitute for the subunits somes, and two normally subdominant epitopes became
β1, β2, and β5, respectively, forming what has been dubbed dominant.281 However, from adoptive transfer experiments
an “immunoproteasome,” present in professional APCs. All in both directions, it was found that the reduced response to
complexes with LMPs contain proteasome proteins, but only one normally dominant epitope was due to decreased pro-
5% to 10% of proteasomes contain LMP-2 and LMP-7. The duction without immunoproteasomes, but that to the other
ones without LMPs are called constitutive proteasomes. was due to an altered T-cell repertoire in the LMP-2–defi-
These MHC-encoded subunits of the immunoproteasome cient mice, presumably due to alterations in the peptides
shift the preference of proteasomes for cleaving after certain presented in the thymus. Further, the increased response
sequences, resulting in the production of different peptide to one of the subdominant determinants was related to in-
fragments.272–274 Proteasomes lacking LMP-2 through muta- creased production of this peptide by the constitutive pro-
tion or gene knockout have the same affinity, but decreased teasomes compared to the immunoproteasomes. Thus, the
cleavage rate, for sequences ending in hydrophobic or basic immunoproteasome specificity plays a significant role in
amino acids. The effect is specific for these proteolytic sites, determining the repertoire of epitopes presented, and in se-
as the activity against sequences containing acidic amino lecting those that are immunodominant, as well as regulat-
acids actually increased.273 Despite the shift in specific pep- ing the CD8 + T-cell repertoire generated in the thymus.
tides released, the overall level of MHC class I expression was Another protein associated with proteasomes is the prote-
reduced only slightly in LMP-7 knockouts274 and not at all in asome activator PA28, which assembles into 11s structures.282
the LMP-2 knockouts. However, presentation of specific epit- Like LMP-2 and LMP-7,272 PA28 is inducible by interferon-γ,
opes of the male H-Y antigen or of influenza NP was reduced and its induction causes a shift in proteasome function that
by three-fold to five-fold in these knockouts. Toes et al.275 may lead to the production of greater amounts of and differ-
quantitatively compared the cleavage fragments produced by ent repertoires of antigenic peptides. For example, synthetic
standard proteasomes and immunoproteasomes and defined substrates were designed to test the ability of proteasomes to
the prevalence of different amino acids on each side of the generate authentic MHC-binding peptides. These substrates
cleavage site. Consistent with earlier studies, there is a strong contained the MHC-binding ligand flanked by the natural
preference for both to cleave after leucine and also to a lesser sequence as found in the original protein.
extent after other hydrophobic residues, both aliphatic and To generate the MHC-binding ligand, the proteasome
aromatic. However, the immunoproteasomes have a stron- would have to cleave the substrate twice.283 By itself, the 20s
ger tendency to cleave after such hydrophobic residues and proteasome was able to produce singly cleaved fragments,
a much reduced cleavage frequency after acidic residues, Asp but with added PA28, doubly cut peptides were generated
and Glu, than standard or constitutive proteasomes. This preferentially. Thus, PA28 favored the production of anti-
shift in specificity is concordant with the observation that genic peptides, possibly by keeping the peptide in the pro-
class I MHC molecules tend to bind peptides with C-terminal teasome until processing was complete. Alternatively, PA28
hydrophobic or basic residues, not acidic ones. Thus, the im- may coordinate the proteolytic activity of two adjacent sites
munoproteasomes in professional APCs may be more effec- to generate doubly cut peptides of just the right length (8- to
tive at generating antigenic peptides that can be presented by 9-mers) to fit in the MHC groove. The distance between
MHC molecules.272,275 Protein degradation by proteasomes is these nearby sites would determine the size of the peptides
processive, so the peptides released after 5% digestion are the produced. The PA28 has been shown to increase generation
same as the fragments released after 90% digestion. This sug- of a dominant lymphocytic choriomeningitis virus epitope
gests that intact proteins may enter the barrel, but they are independently of the presence of the other interferon-γ –
not released at the other end until processing is complete. inducible components LMP-2, LMP-7, and MECL-1.284
Immunoproteasomes were shown to be essential for pro- The specificity of this proteasomal processing system
duction of a hepatitis B virus core antigenic epitope276 and to determines the first step in winnowing the number of

protein segments that can become CTL epitopes, by selec- Recent studies have revealed that most peptides are
tively producing some peptide fragments in abundance and released from the proteasome when they still need fur-
destroying others. Thus, it is probably not just coincidental ther processing to become antigenic peptides.287,288 In the
that the C-terminal residues produced by proteasomal cleav- cytoplasm, the major proteolytic activity comes from an
age often serve as anchor residues for binding class I MHC enzyme called tripeptidyl amino peptidase II (TPPII), which
molecules, or that the lengths of peptides produced are op- is located on a large particle. It has amino peptidase activity,
timal for class I MHC binding.272,285 Better understanding which can remove 1 to 3 amino acids at a time and is useful
of the specificity of proteasomes will contribute to the new for peptides of 15 amino acids or less. It also has endopep-
methods to predict dominant CD8 + T-cell epitopes.232 tidase activity that can cut in the middle of peptides larger
An analysis of proteasomal target sites was based on the than 15 amino acids, with the release of fragments of at least
proteasomal degradation pattern of three proteins: beta-casein, 9 amino acids. These are a significant source of 9-mers with
enolase, and prions.286 The resulting peptides were analyzed by new carboxyl ends, and this may be the only way to generate
mass spectrometry, and the cleavage sites identified, including carboxyl ends other than the proteasome itself.
four amino acids on the amino end of each cleavage site and two The importance of TPPII activity for antigen processing
amino acids on the carboxy end. The results are summarized in is shown by the fact that a specific inhibitor, butabindide,
Figure 23.8. The three proteolytic activities of the proteasome can prevent peptide loading of MHC, resulting in reduced
were represented by favorable amino acids: Arg (trypsin-like), surface expression. The most likely path for most protein
Tyr and Phe (chymotrypsin-like), and peptidylglutamylpeptide antigens is to enter the proteasome and emerge as peptides
(Asp and Glu) at position P1. At this site, Pro was unfavored, of 15 amino acids or larger. These are then trimmed by
as were Asn, Lys, and Ser. Additional negative effects were ob- TPPII, resulting in peptides ready for TAP transport. Longer
served for Asp, Pro, and Ile at position P2. On the carboxy side peptides are trimmed internally by TPPII endoprotease,
of the cleavage site, at position P’1, the positive effect of Tyr was which may generate carboxyl ends needed for TAP binding.
noted as well as negative effects of Ile, Phe, and Val. The resulting peptides are then transported by TAP into the
Additional downstream processing steps, after the pro- ER, where they may be trimmed further to prepare them for
teasome, are known to be important for generation of anti- MHC binding. An additional role of TPPII is shown by the
genic peptides. One of these occurs in the cytoplasm, prior generation of unique epitopes that are not produced by pro-
to TAP transport, and others occur after transport into the teasomes. For example, an important T-cell epitope of the
ER. The proteasome creates a first draft of the peptides, Nef protein of HIV requires TPPII processing. Proteasomal
which are then selected and trimmed to produce the fi nal inhibitors have no effect on it, but butabindide prevents its
pool of antigenic peptides of optimal size and sequence for processing and presentation to HLA-A3– or A11-restricted T
MHC binding while protecting the nascent peptides from cells.289 This pathway seems particularly important for gen-
degradation before they can bind MHC. erating epitopes ending in lysine groups, which bind these
two MHC types but are rarely generated by proteasomes
alone. TPPII can act in parallel with proteasomes or in tan-
dem with them to release this epitope from intact protein or
P4 P3 P2 P1 P′1 P′2 its partially degraded fragments.
Ala (A)
Arg (R) Transport into a Cellular Compartment Where Major
Asn (N) Histocompatibility Complex Binding Can Occur
Asp (D) > 1.0 The second hurdle a potential epitope must surmount is to
Gln (Q) be transported into the cellular compartment for loading
Glu (E)
onto MHC molecules. These compartments are different for
Gly (G)
class I and II molecules, as noted previously.
His (H)
Ile (I)
0 Transport Pathways Leading to Major Histocompatibility
Leu (L)
Lys (K) Complex Class I Presentation. The second hurdle for pep-
Phe (F) tide presentation by class I MHC molecules is to get from the
Pro (P) cytosol, where the peptides are produced, to the ER, where
Ser (S) the newly synthesized class I MHC molecules are assembled
Thr (T) < −1.0 and loaded with peptide. The discovery of a specific active
Tyr (Y) transporter suggested that specificity of transport could
Val (V) further restrict the repertoire of peptides available to load
onto class I MHC molecules. Genetic analysis of mutant cell
lines that failed to load endogenous peptides onto class I
MHC molecules revealed homozygous deletions of part of
FIG. 23.8. The Effect of Specific Amino Acids on Proteasomal
Cleavage between Positions P1 and P’1. Amino acids that favor
the MHC class II region near the DR locus. Molecular clon-
cleavage are shown in black, whereas those that inhibit cleavage ing of DNA from this region revealed at least 12 new genes,
are shown in white. Reproduced from Donnes and Kohlbacher286 of which 2, called TAP-1 and TAP-2 showed a typical se-
with permission. quence for ABC transporter proteins.290–292 Their function is

to transport processed peptides from the cytosol to the ER. the observation that two rat strains with the same MHC type
Once in this compartment, peptides are handed off by TAP (RT1.Aa) were nevertheless not histocompatible, and CTL
to newly formed MHC class I molecules and stabilize a tri- could recognize the difference between them.297 The differ-
molecular complex with β2-microglobulin. This complex is ence, called a cim effect, for class I modification, occurred
then transported to the cell surface, where antigen presenta- because different peptides were binding the same MHC in the
tion occurs. Without the peptide transporters, empty dimers two strains.298,299 The rat has two alleles for a peptide trans-
of MHC class I with β2-microglobulin form, but these are porter supplying peptides to MHC. The one called TAP2A
unstable. Excess free peptide would rescue MHC class I by has peptide specificity matching that of RT1.Aa and delivers a
stabilizing the few short-lived empty complexes that reach broad set of peptides for MHC binding. The other transporter
the surface, as shown by Townsend et al.293 and Schumacher allele, called TAP-2B, supplies a different set of peptides that
et al.294 Thus, MHC-linked genes coding for proteolysis, are discordant with RT1.Aa, so fewer types of peptides are
peptide transport, and presentation at the cell surface have bound. Although RT1.Aa would prefer to bind peptides with
been identified. In effect, the MHC now appears to encode Arg at position 9, it has to settle for peptides with hydrophobic
a complex system of multiple elements devoted to the rapid termini as provided by TAP-2B,300 thereby accounting for the
display of foreign protein determinants on the surface of an apparent histocompatibility difference. Thus, the specificity
infected cell. By continuously sampling the output of the of TAP transport was shown to provide a selective step in nar-
protein synthesizing machinery, this system permits rapid rowing the potential repertoire of CTL epitopes.
identification and destruction of infected cells by CTL before To measure TAP specificity in other species, a trans-
infectious virus can be released. portable peptide bearing an N-linked glycosylation site was
In an infected cell, as soon as viral proteins are made, pep- added to cells permeabilized by treatment with streptoly-
tide fragments generated by the proteasome become avail- sin. If the peptide was transported by TAP, it would enter
able to the TAP-1 and TAP-2 transporter proteins (Fig. 23.9). the ER and cis Golgi, where it would be glycosylated.301–303
These transport the peptide fragments into the ER for as- The extent of glycosylation served as a measure of TAP
sociation with newly formed MHC class I molecules, which function. When competitor peptides were added as well,
would carry them to the cell surface for antigen presenta- TAP-mediated transport of the reporter peptide decreased,
tion, all within 30 minutes. Indeed, the finding of a physi- indicating saturation of peptide-binding sites. In this way,
cal association between TAP and the nascent class I heavy a series of related peptides could be tested for the ability to
chain/β2-microglobulin complex suggests that the peptide compete for TAP binding and transport in order to identify
may be directly handed off from TAP to the new MHC mol- the requirements for TAP binding and transport.
ecule without being free in solution.295,296 If TAP transport TAP binding and transport depended strongly on pep-
is highly selective, then some cytosolic peptides may fail to tide length.303,304 Mouse TAP was shown to have a strong
enter the ER for presentation with class I, but if it is promis- preference for peptides of nine residues or longer.304 For
cuous, then some peptides may be transported that were bet- human TAP, peptides shorter than seven amino acids long
ter off not presented, such as those leading to autoimmunity. were not transported, regardless of sequence.303 Peptides 8
The idea that other proteins may control accessibility of to 11 amino acids long were almost all transported, with
MHC class I–binding sites for peptides originally came from some variation in binding affinities depending on sequence.
FIG. 23.9. Cytoplasmic Antigen-Processing

Pathway leading to Major Histocom-
patibility Complex (MHC) Class I
Presentation. Cytoplasmic antigen is de-
graded in the proteasome to fragments,
which are transported into the endo-
plasmic reticulum by the transporter as-
sociated with antigen presentation (TAP)
transporter. Peptide supplied by TAP forms
a stable complex with MHC class I heavy
chains and β2-microglobulin, and the
complex is transported through the Golgi,
where it achieves mature glycosylation,
and out to the cell surface for presenta-
tion to class I–restricted T cells. Additional
processing options by tripeptidyl amino
peptidase II in the cytoplasm and endo-
plasmic reticulum aminopeptidase I in the
endoplasmic reticulum, before and after
TAP transport, allow greater flexibility in
meeting the requirements of binding both
TAP and MHC molecules.

Peptides 14 to 21 amino acids in length were transported se- human TAP was also sensitive to negative effects of amino
lectively, whereas those longer than 24 amino acids were al- acids Asp, Glu, and Gly at these positions. In both cases,
most never transported intact. Thus, human TAP transport TAP binding was insensitive to amino acid substitutions at
selected against peptides < 7 or > 24 amino acids in length, intermediate P4 through P8.
regardless of sequence. Unlike the rat, human and mouse TAP preferences such as these would selectively transport
TAP do not show allelic differences in peptide transport. some peptides more than others from cytoplasm to ER. Use of
Although TAP can and must transport a wide variety combinatorial peptide libraries independently confirmed that
of peptides, it may still have preferences for which peptides the critical residues influencing TAP transport were the first
are transported most efficiently and which MHC types are three N-terminal residues and the last C-terminal residue.310
provided with the peptides they need. For example, a self- Interestingly, these preferred residues are many of the same
peptide that naturally binds HLA-B27 was modified slightly ones forming the MHC class I–binding motifs (P2 and P9).
to produce an N-linked glycosylation site, resulting in the However, as the MHC-binding motifs differ from each other,
sequence RRYQNSTEL.303 Using the glycosylation of this it is not possible for TAP preferences to match them all. For
peptide to measure transport, saturation of TAP by homolo- example, the TAP preference for Arg at P2 and Phe, Tyr, Leu,
gous peptides occurred with a 50% inhibitory concentra- Arg, or Lys at P9 overlaps with the binding motif of HLA-B27
tion of < 1 m M. Other peptides with unrelated sequences and may favor the transport of peptide ligands for this MHC
also inhibited, often with equally high affi nity. Not only did type. Remarkably, the variant B*2709, which does not prefer
natural HLA-binding peptides compete but also did peptide Tyr or Phe at P9, is not associated with autoimmune disease
variants lacking the MHC-binding motif at positions 2 and 9 as in the more common form of HLA-B27. In contrast, HLA-
(see subsequent discussion). Clearly, peptides binding differ- B7 requires a Pro at P2, which greatly decreases TAP binding.
ent MHC types were transported by the same TAP protein, Similarly, some peptides binding HLA-A2 have hydrophobic
and even peptides that bound mouse MHC were transported residues unfavorable for TAP binding, suggesting suboptimal
by human TAP. In another example, using rat TAP proteins, compatibility between TAP and the most common HLA class
peptides with Pro at position 2, 6, or 9 were found to be poor I allele. Measurements with a series of naturally presented
competitors for transport of a reference peptide.302 peptides from HLA-A2 and HLA-B27 indicated a mean 300-
In a different approach, using a baculovirus system over- fold higher affinity of TAP for the HLA-B27 peptides than for
expressing TAP proteins in microsomes, the affinity of TAP those from HLA-A2, and some of the HLA-A2 peptides did
binding was determined for a wide variety of synthetic pep- not bind TAP at all.308 How are these low-affinity peptides
tides, allowing mapping of the important residues.305,306 delivered to MHC? One suggestion is that peptide ligands for
Binding, rather than transport, appears to be the major HLA-A2 and HLA-B7 may be transported as a larger precur-
step determining TAP peptide selectivity.307 Indeed, artifi- sor peptide containing the correct amino acids, which are
cial neural networks have been developed to predict peptide then trimmed off to fit the MHC groove.
binding to human TAP.308 Using this scheme, it was found A series of studies support this mechanism, showing that
that peptides eluted from three different human class I mol- longer peptide precursors are trimmed at the N-terminus
ecules had higher predicted affi nities for TAP than a con- by aminopeptidases in the ER to form HLA-binding pep-
trol set of peptides with equal binding to those class I MHC tides.311 The ability to transport larger peptides, followed by
molecules, supporting the hypothesis that TAP specificity trimming, could increase the range of permissible antigenic
contributes to the selection of the subset of peptides able to peptides, by facilitating transport of nonantigenic peptides,
bind a class I molecule that actually bind in vivo.308 Unlike followed by trimming to the size and specificity needed for
MHC class I, there were no anchor positions at which a spe- MHC binding. For example, following hepatitis B virus in-
cific amino acid was required. However, there were several fection, certain peptide epitopes that are frequently recog-
positions where substituting the wrong amino acid caused a nized by CTL are nevertheless not bound or transported by
marked reduction in TAP binding. In a typical MHC class TAP. By extending these peptides by one or two amino acids
I binding 9-mer, the strongest substitution effects were ob- of the natural sequence at the amino end, their TAP binding
served at position 9 (P9), followed by substitutions at P2 was greatly enhanced, but at the expense of reduced MHC
and P3, followed by P1. At the carboxy-terminal P9, the pre- binding. When tested on permeabilized cells, the overall
ferred residues were Tyr and Phe (as well as Arg and Lys), effect of the extended peptides was to increase MHC bind-
whereas Glu was worst, causing a 3 log reduction in binding. ing. Apparently, by improving TAP transport, these peptides
Similarly, substituting Pro at P2 caused a 1.5 to 2 log reduc- were able to increase peptide delivery into the ER, where
tion in binding, as compared to preferred residues Arg, Val, they were trimmed to produce peptides compatible with
and Ile. MHC binding.312
These binding studies were extended to 231 peptides, Similarly, it is known that peptides with a Pro at P2, needed
which were evaluated for binding to both mouse and human to bind to certain MHC molecules but poorly transported,
TAP proteins.309 Both TAP proteins showed strong positive are produced from longer precursors that are transported by
selection favoring certain amino acids (Tyr and Phe) at po- N-terminal trimming in the ER by aminopeptidases.313 In
sition P9 and strong negative selection against others (Glu, fact, the inability of the aminopeptidase to cleave beyond a
Asp, and Ser) at P9. However, mouse TAP showed greater residue preceding a Pro naturally leads to trimming of pep-
negative effects from basic amino acids (Arg or Lys) at tides to produce ones with a Pro at P2. Alternatively, some
P9. Both TAPs were sensitive to Pro at P1, P2, and P3, but of these peptides may derive from signal peptides and enter

the ER in a TAP-independent manner. As HLA-A2 tends to Transport Pathways Leading to Major Histocompatibility
bind hydrophobic peptides, signal peptides, which are usu- Complex Class II Presentation. Unlike the class I pathway,
ally hydrophobic, may account for some HLA-A2-binding which delivers peptides to MHC, the MHC class II pathway
epitopes. The ER trimming enzyme has been identified as transports MHC molecules to the endosomal compartment,
endoplasmic reticulum aminopeptidase I (ERAP I).314–316 where antigenic peptides are produced. During transport,
This enzyme is a 106 kD zinc metalloproteinase, and it is the the peptide-binding groove must be kept free of endogenous
major ER protease with broad specificity. It is inducible by peptides. The cell uses one protein, called invariant chain
interferon-γ and inhibited by the aminopeptidase inhibitor (and its processed fragment CLIP), to block the binding site
leucinethiol. Downregulation of ERAP I with small interfer- until needed, and another protein, HLA-DM, to facilitate
ing ribonucleic acid in decreased expression of MHC class I, release of CLIP peptides and their exchange for antigenic
indicating that ERAP I contributes to the supply of peptides. peptides as they become available.
ERAP I prefers peptides of 9 or 10 amino acids or longer and MHC class II molecules assemble in the ER, where α
ignores peptides of 8 amino acids or fewer, which may allow and β chains form a complex with invariant chain.324–326
it to generate antigenic peptides for MHC binding without Invariant chain binds MHC and blocks the peptide-binding
degrading them beyond recognition. groove, so endogenous peptides transported into the ER, for
The aminopeptidase provides a way to relax the require- example by TAP, cannot bind.325,327–332 The complex of α, β,
ment for proteasomes to generate ends that can simul- and invariant chains, consisting of nine polypeptide chains
taneously bind TAP and MHC. In contrast, the fact that a in all,333 is transported via the Golgi and directed by signals
comparable carboxypeptidase has not been identified sug- on invariant chain into endosome-/lysosome-like vesicles
gests that carboxyl ends generated by proteasomal processing called MHC class II compartments. The compartments con-
must be suitable for TAP transport and compatible with MHC tain acid-activated proteases capable of digesting foreign
binding. This requirement is quite stringent, as the carboxyl- proteins into antigenic peptides. In addition, they degrade
terminal residue is the most important for TAP binding, and invariant chain to a fragment called CLIP, corresponding to
it is frequently an anchor residue for MHC as well. amino acids 80 to 103. As long as CLIP remains in the bind-
The significance of selective peptide transport may be to ing groove, antigenic peptides cannot bind, so the rate of
limit immunity to self-peptides. If the match between HLA- CLIP release limits the capacity of MHC to take up antigenic
B27 and TAP specificity is too good, it may contribute to the peptides.
increased incidence of autoimmune disease associated with Peptide loading can be measured by its effect on MHC
HLA-B27.305 An effect of human TAP specificity in loading structure. When an MHC class II molecule binds a pep-
of peptides in viral infection has confirmed the biologic sig- tide, it changes conformation, and certain monoclonal
nificance of TAP specificity.312 TAP-binding specificity also antibodies are specific for the peptide-bound conforma-
limited the repertoire of alloantigenic peptides presented by tion.334 Also, the α-β complex becomes more stable after
HLA-B27.317 Tapasin, which is involved in the TAP-binding peptide binding, which can be detected by running the
process, may also contribute to selective MHC loading con- MHC on an SDS gel without boiling. The peptide bound
tributing to immunodominance by favoring binding of pep- form runs on gels as a large α-β dimer while MHC without
tides with slow dissociation rates from MHC molecules.318 peptides (but still bound to CLIP) is unstable under these
By combining the selectivity of proteasomal processing, conditions and falls apart to give α and β chain monomers
TAP transport and MHC binding, it has been possible to on SDS gels.335
generate models that correlate with known antigenic pep- Mutant cell lines have been generated with a deletion
tides.319,320 These can be used to analyze the sequence of any between HLA-DP and HLA-DQ genes on chromosome
given protein and to predict epitopes that may be recognized 6.334,336–338 These cells express normal levels of MHC class
by T cells restricted to MHC class I. These may be important II structural proteins, HLA-DQ and HLA-DR, but fail to
for analyzing the T-cell response to viral or neoplastic anti- present protein antigens.338 Some of their class II MHC pro-
gens and for generating synthetic vaccines capable of elicit- teins appear on the cell surface but more are retained in the
ing a T-cell response to these antigens. MHC class II compartments. Biochemically, they still con-
The importance of TAP proteins to antiviral immunity tain CLIP peptides,339 rather than peptide antigens, and they
is shown by the fact that herpesviruses have targeted TAP-1 have not achieved the conformation334 or SDS stability of
function as a way to interfere with antigen presentation to peptide-binding MHC class II complexes.340 The defect was
CD8 + CTL. A herpes simplex virus immediate early viral pro- discovered to be due to loss of either of the two chains of a
tein called ICP47 binds to TAP and inhibits its function, caus- class II molecule, HLA-DM, and the phenotype can be cor-
ing reduced expression of new MHC class I molecules on the rected by adding back the missing gene.341 In the presence of
cell surface and inability to present viral or other antigens with normal HLA-DM, MHC releases CLIP and binds antigenic
MHC class I.321–323 As a way to evade immune surveillance, this peptides for presentation to T cells.
strategy could contribute to viral persistence in chronic infec- The importance of HLA-DM function for T-cell help in
tion and viral activation in recurrent disease as frequently oc- vivo was studied in H2-DM knockout mice.342 These mice
curs with herpes simplex virus-1 and herpes simplex virus-2. have reduced numbers of T cells, their class II MHC mol-
These findings also raise the possibility of making a live-atten- ecules reach the cell surface bearing high levels of CLIP
uated ICP47-defective herpes simplex virus vaccine that would peptide, and their B cells are unable to present certain anti-
be more immunogenic than natural infection. gens, such as ovalbumin, to T cells. When H2-DM knockout

mice were immunized with 4-hydroxy-5-nitrophenyl acetyl 307 to 319, was not affected at all. The differential effect on
ovalbumin, specific IgG antibodies were reduced 20-fold, as these antigenic peptides suggests that HLA-DM can serve a
compared to wild type. Germinal center formation and class potential role in editing which peptides stay on MHC long
switching were greatly reduced, and affinity maturation was enough to be presented and which are removed.343 By releas-
not observed. The phenotype was more pronounced for ing myelin basic protein preferentially and not the hemagglu-
some MHC types, such as I-Ab, than for others, such as I-A k. tinin peptide, HLA-DM would favor the stable MHC bind-
Due to tighter binding of CLIP peptides, these MHC types ing and presentation of hemagglutinin peptides over myelin
may be more dependent on H2-DM to maintain empty class basic protein peptides. The affinity of each peptide is deter-
II molecules in a peptide receptive state. mined by the fit between peptide and MHC groove, not by
In vitro studies with purified MHC class II molecules and HLA-DM. However, DM can amplify the impact of the differ-
biotin-labeled peptides have shown that HLA-DM can ac- ence in affinity (ie, signal to noise ratio) by facilitating release
celerate loading of exogenous peptides into HLA-DR–bind- of low-affinity peptides and allowing the high-affinity ones to
ing sites.343,344 For example, loading of myelin basic protein remain. This editing function could have an important effect
fragment 90 to 102 was accomplished in 9 minutes with on which peptides get presented and elicit a T-cell response.
HLA-DM versus 60 minutes without it (Table 23.5). Other HLA-DM could contribute to immunodominance of a pep-
peptides were also loaded at the same rate, suggesting that tide binding MHC with high affinity by releasing its lower
the rate limiting step was the same for each: removal of CLIP affinity competitors. Alternatively, HLA-DM could contrib-
peptides to expose the peptide binding sites on HLA-DR. ute to self-tolerance by releasing self-peptides of low affinity
The kinetic effect was optimal between pH 4.5 and 5.8, before they could stimulate self-reactive T cells either at the
which is typical of the endosomal/lysosomal compartment time of positive selection (so they would fail to be positively
where HLA-DM operates. HLA-DM did not affect the af- selected) or in the periphery (so they would fail to be acti-
finity, as measured by half-maximal binding, but it had a vated). When HLA-DM was engineered to be expressed on
marked effect on the kinetics of binding. the cell surface, it also facilitated loading of exogenous pep-
Conversely, when biotinylated peptides were allowed to tides onto class II MHC molecules and affected the activation
saturate HLA-DR–binding sites overnight, and then free of T cells of different fine specificity for the same epitope.345
peptides were removed, the off rate could be measured over
time.343,344 As shown in Table 23.5 (adapted from Sloan Major Histocompatibility Complex Binding and Assembly
et al.343), the off rate for different peptides could be compared of a Stable Peptide-Major Histocompatibility Complex
in the absence or presence of HLA-DM. The half-life for CLIP Antigen Interaction with Major Histocompatibility Complex
peptides was reduced from 11 hours to 20 minutes by add- Molecules. Perhaps the most selective step a potential anti-
ing HLA-DM. This could explain the enhanced loading of all genic site must pass is to bind with sufficiently high affinity to
other peptides, as they must wait for CLIP to come off. In an appropriate MHC molecule.
the case of antigenic peptides, myelin basic protein 90 to 112 The response specificity of T cells to antigens on APCs
was released 80-fold faster in the presence of DM than in its or target cells provides a number of hints that antigen inter-
absence. However, another peptide, influenza hemagglutinin acts directly with MHC molecules of the APCs to form an
antigenic complex recognized by T cells. First, genes coding
for immune responsiveness (Ir genes) to a specific antigen
Effect of Human Leukocyte are tightly linked to genes for MHC-encoded cell surface
Antigen-DM on Peptide on Rates molecules.127,171 Second, it became apparent that T-cell rec-
TABLE 23.5 ognition of antigen is the step at which MHC restriction
and Off Rates for Binding to
Human Leukocyte Antigen-DR1 occurs.127,179,219,346 For example, in vitro T-cell responses to
small protein and polypeptide antigens were found to paral-
Half-Time Half-life lel in vivo responses controlled by Ir genes, and T cells were
Peptide HLA-DM for Binding for Release exquisitely sensitive to differences in MHC antigens of the
CLIP (80–103) − 60 min 11 hr APCs in all their antigen recognition functions. This obser-
+ 9 min 0.3 hr vation in vitro made it possible to separate the MHC of the
MBP (90–102) − 62 min 86 hr T cell from that of the APC. The T-cell response to antigenic
+ 9 min 1 hr determinants on each chain of insulin depended on the
HA (307–319) − 67 min 144 hr MHC antigens of the APC. This was particularly apparent
+ 10 min 144 hr when T cells from an (A × B)F1 animal responded to anti-
gen presented by APCs of either the A or B parental MHC
CLIP, class II–associated invariant chain peptide; HA, hemagglutinin; HLA, human type.219,347 Neither parental APC stimulated an allogeneic
leukocyte antigen; MBP, myelin basic protein. response from (A × B)F1 T cells, and the response to anti-
The on (association) and off (dissociation) rates of biotinylated peptide from purified
soluble HLA-DR1 were measured by fluorescence assay, in the presence or gen was now limited by the MHC of the APCs. This ability
absence of HLA-DM. The on rates of all three peptides are increased similarly of the APCs to limit what could be presented to the T cells
in the presence of HLA-DM and probably reflect the rate-limiting dissociation of
the bound CLIP fragment of the invariant chain. In contrast, once the peptides was termed “determinant selection.”219,347 It became obvious
are bound, the off rates differ as a result of differences in affinity. Thus, HLA-DM that even in a single (A × B)F1 animal, distinct sets of
catalyzes release of more weakly binding peptides and allows stable binding of
higher affinity peptides. In effect, this is an editing function of HLA-DM.
antigen-specific T cells exist that respond to each antigenic
Adapted from the data of Sloan et al.298 determinant only in association with MHC type A or B.348

Experiments on the fine specificity of antigen-specific class II molecules) into I-A–expressing cells. These cells
T-cell clones suggested that the MHC of the APCs could were able to present several antigens to I-A–restricted T-cell
influence the T-cell response in more subtle ways than just clones and hybridomas351 and similar I-E transfectants pre-
allowing or inhibiting it. Determinant selection implied sented to I-E–restricted T cells.215 Thus, whatever process-
that a given processed peptide should contain both a site ing enzymes are required are already present in fibroblasts,
for MHC interaction and a distinct functional site for TCR and the only additional requirement for antigen-presenting
binding. Thus, a protein with multiple determinants could function is the expression of I-A or I-E antigens.
be processed into different peptides, each with a different The planar membrane technique has been applied to
MHC restriction, consistent with the independent Ir gene determine the minimum number of MHC-antigen com-
control of the response to each antigenic determinant on plexes per APC necessary to induce T-cell activation.352
the same protein.181 For example, T-cell clones specific for After pulsing the presenting cells with antigen, the cells were
myoglobin responded to different antigenic determinants studied for antigen-presenting activity, and some of the cells
on different peptide fragments of myoglobin215 : Those spe- were lysed to produce a purified fraction containing MHC
cific for one of the epitopes were always restricted to I-A, charged with antigenic peptides. These MHC-peptide com-
whereas those specific for the other were always restricted plexes were used to reconstitute planar membranes, and
to I-E. The simplest interpretation was that each antigenic their potency was compared to a reference MHC prepara-
peptide contained an MHC association site for interacting tion pulsed with a high peptide concentration in vitro and
with I-A or I-E. At the level of Ir genes, mouse strains lacking presumed to be fully loaded. In this way, the relative peptide
a functional I-E molecule could respond to one of the sites occupancy of MHC-binding sites corresponding to any level
only, and those with neither I-A nor I-E molecules capable of of antigen presentation could be determined. For B cells and
binding to any myoglobin peptide would be low responders macrophages, the threshold of antigen loading necessary for
to myoglobin. triggering T cells was 0.2% of I-Ed molecules occupied by
Evidence for a discrete MHC association site on peptide peptide, corresponding to about 200 MHC-peptide com-
antigens came from studies with pigeon cytochrome c. The plexes per presenting cell. For artificial presenting cells, such
murine T-cell response to pigeon cytochrome c and its car- as L cells transfected with I-Ed, the threshold was 23 times
boxy-terminal peptide (81 to 104) depends on the I-E mol- greater or 4.6% of MHC occupied by peptide. Similarly,
ecules of the APCs.179 However, distinct structural sites on when MHC-peptide binding was measured directly, using
the synthetic peptide antigen appear to constitute two func- radiolabeled peptide to determine the minimum level of
tional sites: an epitope site for binding to the TCR and an MHC-peptide complexes required for T-cell triggering,
“agretope” (for “antigen restriction tope”) site for interact- B cells were capable of presenting antigen with as few as
ing with the MHC molecule of the APC.179,196–198 Amino acid 200 to 300 MHC-peptide complexes per cell.353 A similar
substitutions for Lys at position 99 on the peptide destroyed number of peptide-MHC class I molecule complexes was
the ability to stimulate T-cell clones specific for the peptide, reported to be required on a cell for recognition by CD8 +
whereas the difference between Ala and a deletion at posi- cytotoxic T cells.354 These results explain how newly gener-
tion 103 determined T-cell stimulation in association with ated peptide antigens can bind enough MHC molecules to
some MHC antigens but not others, independent of the stimulate a T-cell response, even in the presence of compet-
T-cell fine specificity. In addition, immunizing with the pep- ing cellular antigens, as a low level of MHC occupancy is
tides substituted at position 99 elicited new T-cell clones that sufficient. In addition, this threshold of presentation may
responded to the substituted peptide but not the original and explain how multivalent protein antigens, such as viral
showed the same pattern of genetic restriction, correlated particles, with 100 to 200 protein copies each, can be over
with the residue at position 103, as the clones specific for the 103-fold more immunogenic than the same weight of protein
original peptide. These results implied that the substitutions monomers.355–357 Studies on the number of TCRs needed
at position 99 had not affected the MHC association site but for triggering, based on titrating peptide and recombinant
independently altered the epitope site that interacts directly soluble class I MHC molecules on plastic, suggested that
with the TCR. In contrast, position 103 was a likely subsite interaction of three to five TCRs with MHC-peptide com-
for MHC interaction, without altering the TCR-binding site. plexes was sufficient, consistent with several T cells interact-
It remained to be shown that MHC molecules without ing with one APC.358,359 Biochemical evidence for the direct
any other cell surface protein were sufficient for presenta- association between processed peptide and MHC mol-
tion of processed peptide antigens. This was demonstrated ecules was demonstrated by competition between peptides
by Watts et al.,349 who showed that glass slides coated with for antigen presentation225,360–363 and then more directly by
lipid containing purified I-A molecules could present an equilibrium dialysis,364 molecular sieve chromatography,365
ovalbumin peptide to an ovalbumin-specific T-cell hybrid- or affinity labeling.366 Equilibrium dialysis (see Chapter 7)
oma. This result meant that no other special steps were was performed by incubating detergent-solubilized class II
required other than antigen processing and MHC associa- molecules with fluoresceinated or radioactive antigenic pep-
tion. Likewise, Walden et al.350 specifically stimulated T-cell tides, followed by dialysis against a large volume of buffer.
hybridomas with liposomes containing nothing but antigen Peptide can pass in or out of the dialysis bag, but the class
and MHC molecules. Also, Norcross et al.351 transformed II molecules are trapped inside. In the absence of binding
mouse L cells with the genes for the I-A α and β chains and by class II molecules, the labeled peptide would distribute
converted the fibroblasts (which do not express their own itself equally between the inside and outside of the dialysis

only two amino acids (Val 327 and Ala 332) were essential for
Correlation between Major MHC interaction, and changes at either of these resulted in
Histocompatibility Complex a loss of antigenicity for the clone. Seven other amino acids
TABLE 23.6 Restriction and Binding to were critical for T-cell stimulation but did not affect MHC
Major Histocompatibility binding. Thus, these must be part of the functional epitope.
Complex Molecules Interestingly, certain substitutions for His 328, Ala 330, and
Glu 333 had effects on MHC binding, whereas others had
Ova Myo HEL Cyto
effects on T-cell stimulation without affecting MHC bind-
Competitor Peptide + Ad + Ed + Ak + Ek
ing. These amino acids might participate in both agretope
Ova 323–339 ++++ − ++ + and epitope functional sites, or, alternatively, the substitu-
Myo 106–118 ++++ − ++ +/− tions may affect the conformation of the peptide as it binds,
Myo 132–153 − ++++ − ++ thus indirectly affecting T-cell recognition371 (see following
HEL 46–61 + + ++++ + discussion). The fact that substitutions at 9 of 11 amino acids
Cytochrome c 88–104 ++ +/− ++ ++++ could be tolerated without affecting MHC binding is consis-
tent with the determinant selection hypothesis in that mul-
HEL, hen egg lysozyme. tiple antigenic peptides are capable of interacting with the
Data from Früh et al.322
same antigen-binding site on the MHC molecule.
Similarly, by using a T-cell clone specific for peptide 52
chamber. However, when the appropriate class II molecules to 61 of hen egg lysozyme, substitutions at each amino acid
were added to the chamber, extra peptide molecules were were analyzed for the ability to bind to I-A k and stimulate
retained inside it due to formation of a complex with MHC the clone.372 Four of eleven amino acid residues were silent,
class II. In this way, direct binding of antigen and MHC was whereas substitutions at three positions resulted in reduced
shown, and an affinity constant was determined.364,365 binding to I-A k. Substitutions at the remaining three
A second approach was to form the antigen-MHC complex positions resulted in decreased T-cell stimulation without
over 48 hours, followed by rapid passage over a Sephadex G50 affecting MHC association. The epitope was very sensitive to
sizing column (Sigma-Aldrich). The bound peptide came off substitutions, even conservative ones such as changing Leu
the column early because it is the size of class II molecules 56 to Ile, norLeu, or Val. The results in both of these studies
(about 58 kD), while free peptide was usually included in the confirmed by competitive binding that the MHC molecule
column and eluted later, as it is only approximately 2 kD.365 contains a single saturable site for peptide binding. This
Peptide bound to specific and saturable sites on MHC. site must be capable of binding a broad range of antigenic
Competitive binding showed that different peptide antigens peptides. In binding the MHC groove, antigenic peptides
with the same MHC restriction bind to the same site on the assume the extended conformation that exposes the epitope
MHC class II molecule.367,368 For example, Table 23.6 shows the for recognition by the TCR.
results with peptide antigens that are known to be presented Although a full set of general principles explaining
with I-A or I-E antigens of the D or K haplotype. We observe the specificity of antigen presentation and T-cell recog-
that Ova peptide 323 to 329, which is presented with I-Ad, also nition has not yet emerged, it is studies such as these,
binds well to purified I-Ad, whereas nonradioactive peptide combined with complementary structural studies char-
competes for the peptide-binding sites of the I-Ad molecule. acterizing the antigen-interacting portions of MHC mol-
Similarly, the other I-Ad–restricted peptide, myoglobin 106 to ecules214,253,254,369,373–380 (see Chapters 21 and 22) and of
118, competes with Ova 323 to 339 for the same site. However, TCRs381–387 (see Chapter 11) that will ultimately lead to an
myoglobin 132 to 153, which is not restricted to I-Ad, does not understanding of these principles.
compete for it but does compete for its own restriction ele- One observation that came out of this type of structure-
ment, I-Ed. Similarly, pigeon cytochrome c competes best for function study was that a single peptide can bind to a class
its restriction element I-Ek rather than I-A k or I-Ed, which do II MHC molecule in more than one way, and thus be seen
not present cytochrome. Conversely, recombinant Eβ genes by different T cells in different orientations or conforma-
have been used to map separate sites on a class II MHC mol- tions.371,388 The same conclusion can be reached from an
ecule for binding to peptide antigen and to the TCR.369 entirely different type of study, in which mutations are
Using these two biochemical methods, it has been pos- introduced into the MHC molecule. Mutations in the floor
sible to explain major losses of peptide antigenicity result- of the peptide-binding groove, which cannot directly inter-
ing from amino acid substitutions in terms of their adverse act with the TCR, can differentially affect recognition of a
effect on epitope or agretope function. For example, the peptide by one clone and not another.389–391 In a particularly
response of each of two ovalbumin-specific T-cell clones was thoroughly studied case, it was clear that the quantitative
mapped to peptide 325 to 335 by using a nested set of syn- level of peptide binding was not affected by the mutation,
thetic peptides. Five substitutions were made for each amino but rather the change in the floor of the groove imposed
acid in the peptide, and the resulting 55 different peptides an altered conformation on the peptide that differentially
were each tested for the ability to stimulate the clone.370 affected recognition by different T cells.391 If indeed the TCR
Presumably, those peptides that failed to stimulate could be cannot detect the mutation in the MHC molecule except
defective at an epitope or an agretope functional site. In fact, indirectly by its effect on the peptide conformation, then

one is forced to conclude that different T cells have prefer-

ences for different conformations of the same peptide bound
to (what appears to the T cell as) the same MHC molecule.
Another general observation to come from this type
of study is that substitution of amino acids often affects
presentation by MHC and recognition by T cells through
introduction of dominant negative interactions or interfer-
ing groups, whereas only a few residues are actually essential
for peptide binding.392 Both for class II binding392–395 and for
class I MHC binding,396 most residues can be replaced with
Ala or sometimes Pro without losing MHC binding, as long
as a few critical residues are retained. Of course, T-cell rec-
ognition may require retention of other residues. If many
of the amino acid side chains are not necessary for binding
to the MHC molecule, then one might expect side chains of
noncritical amino acids to occasionally interfere with bind-
ing, either directly or through an effect on conformation.
That is exactly what was observed for a helper epitope from
the HIV-1 envelope protein when a heteroclitic peptide—
that is, one that stimulated the T cells at much lower con-
centrations than did the wild-type peptide—was obtained
by replacing a negatively charged Glu with Ala or with Gln, FIG. 23.10. Enhancement of Immunogenicity of a Peptide Vaccine
which has the same size but no charge.392 An Asp, negatively for Induction of Class I Major Histocompatibility Complex (MHC)-
charged but smaller, behaved like the Glu. Thus, this residue Restricted Cytotoxic T-Lymphocytes by Modification of the Class II
was not necessary for binding to the class II MHC molecule, MHC-Binding Portion to Increase Cluster of Differentiation (CD)4+
but a negatively charged side chain interfered with binding T-Cell Help. Peptide vaccine PCLUS3-18 IIIB contains a class II MHC
to the MHC molecule as measured by competition studies. binding helper region, consisting of a cluster of overlapping deter-
Information about residues that interfere with binding has minants from the human immunodeficiency virsu-1 envelope protein
gp160, and a class I MHC-binding cytolytic T cell (CTL) epitope,
allowed the refinement of sequence motifs for peptides P18 IIIB. Modification of the helper epitope to remove an adverse
binding to MHC molecules to permit more reliable predic- negative charge by replacement of a Glu with an Ala residue was
tion of binding397 (see following discussion). shown to increase binding to the class II MHC molecule.392 Here,
This observation also provides a novel approach to introduction of the same modification of the helper epitope, to pro-
make more potent vaccines by “epitope enhancement,” the duce PCLUS3(A)-18IIIB, is shown to greatly increase immunogenic-
process of modifying the internal sequence of epitopes to ity in vivo for induction of CTL to the class I-binding P18IIIB portion.
Immunization of A.AL mice with 5 nmol of either vaccine construct
make them more potent, for example, by increasing affin-
subcutaneously in montanide ISA 51 adjuvant, and stimulation of
ity for an MHC molecule or TCR, or able to induce more resulting spleen cells with P18IIIB for 1 week in culture, resulted in
broadly cross-reactive T cells specific for multiple strains of 33-fold more lytic units for lysing targets coated with P18IIIB when
a virus.398–401 Proof of principle that this approach can make mice were immunized with the modified second-generation vac-
more potent peptide vaccines has recently been obtained.401 cine than when they were immunized with the original construct
The modified “enhanced” helper T-cell epitope from the with the natural sequence. Thus, class II–restricted CD4+ T-cell help
HIV-1 envelope protein described previously,392 with Ala has a major impact on induction of class I–restricted CTL, and this
process of “epitope enhancement” can be used to make vaccines
substituted for Glu, was shown to be immunogenic at 10- more potent than the natural viral antigens. Targets: BALB/c 3T3
to 100-fold lower doses for in vivo immunization than the fibroblasts with P18IIIB (solid lines) or no peptide (dashed lines).
wild-type HIV-1 peptide to induce a T-cell proliferative (Modified from Ahlers et al.401 with permission.)
response specific for the wild-type peptide. Further, when a
peptide vaccine construct using this helper epitope coupled
to a CTL epitope402,403 was modified with the same Glu-to- struction of more potent vaccines, providing greater protec-
Ala substitution, it was more potent at inducing CD8 + CTL tion against viral infection.404 Further, the improved help
specific for the CTL epitope than was the original vaccine was found to be qualitatively, not just quantitatively, differ-
construct, even though the CTL epitope was unchanged ent, skewed more toward Th1 cytokines.404 The mechanism
(Fig. 23.10).401 The increased potency of the vaccine con- was found to involve greater induction of CD40 ligand on
struct was shown to be due to improved class II MHC– the helper T cells, resulting in greater interleukin 12 pro-
restricted help by genetic mapping using congenic strains of duction by the antigen-presenting dendritic cells, which in
mice expressing the same class I MHC molecule to present turn polarized the helper cells toward Th1 phenotype.404
the CTL epitope and the same background genes but differ- The dendritic cells conditioned with the helper T cells and
ing in class II MHC molecules.401 Thus, class II–restricted the higher affinity peptide and then purified were also more
help makes an enormous difference in induction of class effective at activating CD8 + CTL precursors in the absence
I–restricted CTL, and epitope enhancement can allow con- of helper cells, supporting a mechanism of help mediated

through activation of dendritic cells.405–407 This study showed the most T cells specific for the wild-type epitope are the
also that such help was mediated primarily through upreg- most effective at clearing tumors.420 Thus, it is important in
ulation of interleukin 12 production and CD80 and CD86 using epitope-enhanced peptides that the peptides still acti-
expression on the dendritic cell.404 A similar or overlapping vate a large number of T cells that recognize the wild-type
HIV envelope epitope recognized by human CD4 + T cells sequence. These results emphasize the importance of affin-
when presented by HLA-DR13 has been mapped and has ity for MHC molecules vaccine efficacy421 and suggest that
undergone epitope enhancement as well.408 Understanding rational design of vaccines with higher affinity epitopes may
this mechanism of epitope enhancement may contribute to produce more effective second-generation vaccines.422,423 To
the design of improved vaccines. that end, the discovery of sequence motifs predicting MHC
Similar epitope enhancement has been carried out for class molecule binding and the development of bioinformatics
I MHC–binding viral or tumor peptides as well,409–414 and in strategies to predict peptide affinity have proven a great
one case has been found to result in greater clinical efficacy impetus to the field.424
of a melanoma vaccine used for human immunotherapy.415 In the case of class I MHC molecules, results defining
Cautions have also been raised about the use of enhanced sequence binding motifs generalize the conclusion that only
epitopes (sometimes also called agonistic or heteroclitic epi- a few critical “anchor” residues determine the specificity of
topes), in that the amino acid sequence modifications that binding to the MHC molecule (Table 23.7).264,396,425–429 These
affect MHC binding can also alter the conformation of the motifs were defined by a detailed study of one peptide-MHC
MHC-bound peptide and affect the specificity of T cells system427 by sequencing the mixture of natural peptides
elicited.416–419 The altered or enhanced peptides that elicit eluted from a class I MHC molecule and finding that at
Examples of Motifs for Peptides Binding to Class I and II Major Histocompatibility

TABLE 23.7
Complex Molecules
Residue Number
MHC Molecule 1 2 3 4 5 6 7 8 9
Class I
H-2Kd Y I, L, V
H-2Db N M, I
H-2Kb F, Y L, M
H-2Ld P (hydrophilic K, R) M, L, F
H-2Dd G P K, R L
H-2Kk E I
HLA-A2.1 L, M V
HLA-A3 L (F) Y, K
HLA-B27 K, R R I, Y, K, R
G F, W
Class II
DRB1*0101 Y, V, L, A A, G L, A
L, F,
I, A
DRB1*0301 L, I D K, R Y, L
F, M E, Q F
V
DRB1*0401 F, Y no N, S Polar Polar
(DR4Dw4) W R, K T, Q Charged Aliphatic
Aliphatic K
DRB1*0402 V, I, no N, Q, R, K Polar
(DR4Dw10) L, M D, E S, T H, N Aliphatic
K Q, P H
DRB1*1501 L, V, F, Y I, L
(DR2b) I I V, M
F
DQA1*0501 F, Y, V, L Y, F
DQB1*0301 I, M I, M M, L,
L, V Y V, I
HLA, human leukocyte antigen; MHC, major histocompatibility complex.
Data from Goodman and Sercarz,234 Walden et al.,350 Sette et al.,370 Kurata and Berzofsky,371 Allen et al.,372 Brown et al.,373 and Bjorkman et al.374

certain positions in the sequence, a single residue predomi- In contrast, when natural self-peptides were eluted from
nated within the pool of peptides,425 and by separating class II MHC molecules,252,432 the lengths were much more
and sequencing individual natural peptides eluted from a variable, ranging from 13 to 18 residues, and several variants
class I molecule and finding a conserved residue at certain of the same peptide were found with different lengths of
positions.426 The latter two studies also made the important extra sequence at one end or the other (“ragged ends”). This
observation that the natural peptides eluted from class I finding suggested that both ends of the peptide-binding
MHC molecules were all about the same length, eight or groove of class II MHC molecules are open, in contrast
nine residues, and this was confi rmed for a much larger col- to class I, so that additional lengths of peptide can hang
lection of peptides eluted from HLA-A2 and analyzed by out either side, and trimming does not have to be precise.
tandem mass spectrometry.428 This finding was consistent However, a corollary is that the peptides eluted from class II
with other studies demonstrating that a minimal nonapep- molecules would not be aligned in a motif starting from the
tide was many orders of magnitude more potent than longer exact amino terminus and that was indeed what was found.
peptides in presentation by class I molecules to T cells.430,431 Although a moderately conserved motif was found in some
This conservation of length was critical to the success of the of the peptides eluted from the murine class II molecule I-Ad,
approach of sequencing mixtures of peptides eluted from a consistent with the motif defined based on known antigenic
class I molecule425 because such a method requires that the peptides binding to I-Ad,433 the motif was neither so clearly
conserved anchor residues all be at the same distance from defined nor so highly conserved as in the class I case and
the N-terminus. The fact that Falk et al.425 could find a single required aligning of sequences to identify a core motif of
amino acid at certain positions, such as a Tyr at position 2 about nine amino acid residues.432 Subsequently, a num-
in peptides eluted from Kd, implies not only that most or all ber of motifs for peptides binding to human class II MHC
of the peptides bound to Kd had a Tyr that could be aligned molecules have been defined.264,434–439 Unlike peptides eluted
but also that the peptides were already aligned as bound to from MHC class I grooves, these class II–binding peptides
the MHC molecule, with each one having just one residue may locate the core-binding motif at various distances from
N-terminal to the Tyr. This result implies that the position the amino or carboxyl end of the peptide.
of the N-terminal residue is fi xed in the MHC molecule. It is The crystal structure of a peptide bound to a human class
this fact that has made the identification of motifs for bind- II MHC molecule, DR1, revealed that indeed the ends of
ing to class I molecules much more straightforward than the groove are open, and the peptide can extend beyond the
finding motifs for binding class II molecules. groove in either direction.254,440 In addition, the more broadly
This conclusion has not only been confirmed but also defined class II motifs in Table 23.7 can be explained by less
explained by x-ray crystallographic data on class I peptide– stringent requirements for amino acid side chains to interact
MHC complexes.376–378 It appears that both the N-terminal with binding pockets in class II. In general, the MHC class II
α-amino group and the C-terminal carboxyl group are fi xed binding pockets are shallower than for class I, and a selected
in pockets at either end of the MHC groove, independent of peptide derives less binding energy from each pocket. In fact,
what amino acids are occupying those positions, and that they form fewer H-bonds with the peptide side chains, and
the rest of the peptide spans these fi xed points in a more or more H-bonds are directed at the peptide backbone, allowing
less extended conformation. The minimum length that can a variety of different peptides to bind. Rather than requir-
span the distance between these pockets is 8 residues, but ing a specific amino acid at each position, the shallow bind-
9 or 10 residues can be accommodated with a slight bulge ing pockets of MHC class II tend to exclude peptides based
or β turn in the middle of the peptide, explaining the nar- on unfavorable interactions, such as side chains too large to
row restriction on length. Between these ends, one or two fit the binding pocket. Even one amino acid side chain that
pockets in the groove can accommodate the side chain of binds strongly to an MHC pocket is sufficient to anchor the
an amino acid, usually either at P2 binding in the B pocket peptide to MHC class II and set the frame for the interaction
or at P5 binding in the C pocket, depending on the par- of the rest of the peptide with the MHC groove.
ticular MHC molecule. Additionally, the side chain of the For example, binding of three peptides to the class II
C-terminal residue serves as an anchor in the F pocket at molecules I-A in mice or HLA-DQ in humans are shown in
the end of the groove. These residues that fit into pockets Figure 23.11. The first residue of the peptide motif is desig-
correspond exactly to the “anchor” residues, at P2 or P5, and nated P1, the next is P2, and so on. The α-helical walls and
P8, P9, or P10, defined by the sequence motifs, and appear β-sheet floor of the MHC class II groove (see Chapter 21) are
to be the primary determinants of specificity for peptide peeled away to reveal the peptide backbone and side chains
binding because the rest of the interactions are largely with in relation to MHC-binding pockets. For the ovalbumin
peptide backbone atoms, including the α-amino and car- peptide Ova323-329 binding to I-Ad, residues P1, P4, and P9
boxyl groups, and therefore do not contribute to sequence all point down into the binding pockets.441 The best fit is
specificity. This finding can explain both the breadth of between Val 327 and the P4 pocket, which creates mainly
peptides that can bind to a single MHC molecule because hydrophobic interactions with MHC and serves as the an-
most of the binding involves only backbone atoms common chor residue. Residues P5 (His 328) and P8 (His 331) project
to all peptides and also the exquisite specificity of binding upward for binding to the TCR. The shallow P4 pocket can
is determined by the anchor residues that account for the Ir tolerate only small hydrophobic side chains, such as Val, so
gene control of responsiveness. it dictates which peptides can bind here. The other MHC

Based on affinity for MHC, these interactions explain

the peptide-binding motifs for MHC class II that select
which peptides can be presented to T cells. In addition, these
interactions orient the peptide in the MHC groove and deter-
mine which residues are accessible for recognition by the TCR.
T-Cell Receptor Recognition

The last hurdle that a potential antigenic determinant must
surmount is recognition by a TCR within the repertoire
of the individual responding. This repertoire may be lim-
ited by the availability of combinations of V, D, and J genes
(see Chapter 11) in the genome that can combine to form
an appropriate receptor, given the lack of somatic hypermu-
tation in TCRs in contrast to antibodies,383,444 and then by
self-tolerance, as mediated by thymic or peripheral nega-
tive selection, or by limits on the repertoire that is posi-
tively selected in the thymus on existing self-peptide–MHC
complexes. The available repertoire may also be influenced
by prior exposure to cross-reactive antigens. In general,
however, it has been hard to find holes in the repertoire.445
Indeed, studies examining large panels of peptides bind-
ing to particular class I molecules for correlations between
peptide affinity for MHC and T-cell responses have failed
to find high-affinity peptides for which no T-cell response
can be raised, whereas not all lower affinity peptides elicit a
response.446,447 These results suggest that if the peptide can
bind well enough to the MHC molecule, there is virtually
always a T cell that can see the complex. However, the array
FIG. 23.11. Interaction of Peptides with the Binding Groove of of other MHC molecules present can influence the breadth
Major Histocompatibility Complex Class II, as Determined by X-ray of the T-cell repertoire.448 Furthermore, when TCR reper-
Crystallography. Anchor residue side chains fill binding pockets to toires of mice and humans were compared for peptides pre-
a greater or lesser extent, supplemented by H-bonds to the pep- sented by HLA-A2.1, they seemed to be capable of seeing the
tide backbone. Examples include I-Ad with ovalbumin peptide 323 same spectrum of peptides.449 Eleven peptides from hepa-
to 334 (Top),441 I-Ak with hen egg lysozyme 52 to 60 (Middle),442 and
HLA-DR3 with invariant chain class II–associated invariant chain titis C virus proteins, each of which had a motif for bind-
peptide (Bottom).443 ing to HLA-A2.1, were tested for recognition by CTL from
HLA-A2.1 transgenic mice and human HLA-A2.1 positive
patients infected with hepatitis C virus. The same four pep-
pockets, P1 and P9, accommodate many different residues, tides that were recognized by the T cells from the mice were
so they have little effect on which peptides can be presented the ones recognized by the human T cells, whereas the oth-
by I-Ad. ers were not recognized well by either murine or human
For hen egg lysozyme peptide50-62 binding to I-A k, inter- T cells. The selection of which peptides were recognized
actions with MHC are observed for P1, P4, P6, and P9.442 The seemed to be determined by binding to the HLA-A2.1 mol-
P1 interaction is very different from I-Ad, as the P1 pocket is ecule rather than by the availability of T cells. Thus, despite
a perfect fit for Asp and has an arginine at the end of the tun- the differences in TCR genes in mice and humans, the rep-
nel to neutralize charge. This structure explains why nearly ertoires are plastic enough that if a peptide passes the other
all peptides presented by I-A k must have Asp at this position. three hurdles of processing, transport, and binding to MHC
In contrast, the P4 and P9 pockets are partially fi lled and will molecules, T cells can be elicited to respond to it in either
tolerate a number of different side chains at these positions. species.449 Likewise, no hole in the helper T-cell repertoire
The P6 pocket requires a Glu or Gln here, even though the could be found to explain low responders to the recombi-
MHC residues deep in the pocket are acidic. It is presumed nant hepatitis B vaccine.450
that one of the Glu residues must be protonated to allow Glu On the other hand, evidence exists that MHC binding
binding at this position. This arrangement of the peptide is not the whole story. Schaeffer et al.451 examined 14 over-
leaves P2, P5, and P8 exposed to solvent in the crystal struc- lapping peptides covering the sequence of staphylococcal
ture and to the TCR during antigen presentation. nuclease with different class II MHC molecules, constitut-
For the CLIP peptide binding to HLA-DR3, deep pockets ing 54 different peptide-MHC combinations. Clearly, MHC
at P1 and P9 are more fully occupied by the peptide side binding plays a major role because 12 out of 13 immuno-
chains.443 The pH-dependent binding is important because genic peptides were high or intermediate binders to MHC
CLIP must be stable at neutral pH and unstable at acid pH molecules, whereas only 1 of 37 poor binders were immu-
in the presence of HLA-DM in order to perform its function. nogenic. Of high-affinity binders, five out of five peptides

were immunogenic. However, for intermediate affinity in LMP-2, and so unable to make immunoproteasomes.
MHC-binding peptides, only 7 out of 12 were immunogenic. Reduced response to one normally immunodominant
Similar results were found for the class I MHC molecule influenza epitope was found by T-cell adoptive transfer
HLA-B*0702, in which it was found that 7 of 7 high-affinity studies to be due to an alteration in the T-cell repertoire,
binders but only 9 out of 12 intermediate affinity binders, presumably because of altered processing of self-peptides in
were immunogenic.447 Thus, MHC binding alone is not the thymus.281
sufficient to ensure immunogenicity. The T-cell repertoire A recent study found another mechanism by which the
was one factor suggested that might limit the spectrum of T-cell repertoire contributes to immunodominance. It was
immunogenic peptides. found that the relative immunodominance of a dominant
Indeed, examples for selection at the level of the TCR determinant of the HIV envelope protein among different
repertoire exist. A particularly elegant example described by strains of H-2D mice all presenting the same peptide–H-2Dd
Moudgil et al.452 is one in which a peptide (46 to 61) of mouse complex correlated with the avidity of the T cells respond-
lysozyme presented by I-A k is recognized T cells from CBA/J ing in those strains.459 The mechanism found was that high-
and B10.A mice, expressing I-A k and I-Ek, but not by T cells avidity T cells proliferate faster than low-avidity ones when
from B10.A(4R) mice expressing only I-A k, even though the exposed to antigen and therefore dominate the response.
APC from B10.A(4R) mice can present the peptide to T cells This mechanism differs from those at earlier steps that pro-
from the other strains. T cells from the B10.A(4R) mice can gressively constrain the response, and thereby narrow the
respond to variant 46 to 61 peptides in which the C-terminal repertoire to dominant epitopes, whereas this mechanism
Arg is replaced by Ala, Leu, Phe, Asn, or Lys, indicating that selectively expands the dominant repertoire. This mecha-
the C-terminal Arg is hindering recognition, but not bind- nism may also explain other recent findings such as the
ing by I-A k, and in this case, not processing as the B10.A(4R) higher affinity of dominant compared to subdominant
APC can present the peptide. It appears that the hindrance TCRs specific for human cytomegalovirus.460 Because high-
interferes with recognition by TCRs available in B10.A(4R) avidity T cells are better able to clear virus infection,461,462 a
mice, but not TCRs available in B10.A or CBA/J mice, or in recent study even suggests that the driving force for MHC
(B10.A(4R) × CBA/J)F1 mice. Because the B10.A mice are polymorphism is to create a large enough repertoire to select
congenic with the B10.A(4R) mice, the difference is not one for high-avidity T cells.463
of non–MHC-linked genes such as TCR structural genes or Several critical parameters have been reported to be
non-MHC self-antigens producing self-tolerance. Further, involved in more effective T-cell activation during TCR-
because the F1 mice respond, the difference is not due to a hole peptide-MHC interaction. In some situations, TCR avidity
in the repertoire produced by a self-antigen of the B10.A(4R) for the peptide-MHC complex appears to play a more criti-
mice. It was concluded that the CBA/J and B10.A mice con- cal role than dissociation rate464 and a threshold for maximal
tain an additional repertoire, positively selected on I-Ek or activity may even be defined,465 whereas in other situations,
possibly an H-2D/L class I molecule, in which these strains the half-life of the TCR-peptide-MHC interaction seems
differ, which can recognize the 46 to 61 peptide despite the most critical.466–468 Moreover, in situ in two-dimensional
hindering Arg at the C-terminus. An alternative related ex- surface interactions, a rapid on-rate may compensate for a
planation is that strains that express I-Ek or Dk or Dd /Ld have faster off-rate (shorter half-life) and allow effective TCR-
an additional repertoire of TCRs positively selected on I-A k– peptide-MHC engagement, suggesting the total cumulative
presenting self-peptides from processing of these other MHC duration time (including multiple repeat engagements) may
molecules in the thymus. This example illustrates a case in be most critical.469–472
point that subtle differences in TCRs, presumably caused in As more is understood about the molecular basis of
this case by positive selection, can lead to responsiveness or TCR recognition, with crystallographic data now avail-
nonresponsiveness to a determinant that has already passed able,386,387,473 it becomes possible to apply epitope enhance-
all of the three earlier hurdles, processing, transport, and ment in a rational way to the affinity of the peptide-MHC
MHC binding. Of course, there are some holes related to self- complex to the TCR, as was described for the peptide
tolerance, primarily related to the loss of response to domi- affinity for MHC molecules. Sequence modifications in the
nant determinants453 and especially loss of high-avidity T peptide that increase the affinity for the TCR were shown
cells.454–457 Interestingly, not only is there no loss of response to be more effective at expanding in vivo the T cells specific
to cryptic determinants but also T cells recognizing cryptic for tumor antigens.474–476 Most of these modifications were
determinants of mouse lysozyme can be positively selected found empirically, but a systematic study of substitutions
on other nonmouse lysozyme self-ligands in a lysozyme throughout a number of peptides revealed a pattern in
knockout mouse.453 The selective loss of high-avidity T cells which peptides with conservative substitutions at P3, P5,
to immunodominant determinants has suggested a strategy or P7 were most likely to yield increased TCR affinity, nar-
to apply epitope enhancement to modify subdominant lower rowing the candidate list of peptides that require empirical
affinity tumor antigen peptides to increase their affinity for screening.477 This strategy provides a second type of epitope
MHC molecules, and thereby make them more immuno- enhancement, derived from basic immunologic principles,
genic in order by to take advantage of a repertoire not already to produce more effective vaccines.
crippled by loss of the high-avidity clones.458 The advent of DNA vaccines has made it possible to link a
Another elegant example of T-cell repertoire limitations series T-cell epitopes in tandem and express them in the form
on immunodominance comes from a study of mice deficient of a multiepitope vaccine. These chimeric proteins have the

potential advantages of focusing the immune response on was enhanced by four logs when Lys was substituted for Phe
biologically important epitopes, combining T helper epit- at this position. These amino acid preferences may reflect
opes with CTL epitopes, using epitopes with high affinity for the major protease activities of proteasomes: The most tar-
MHC, and linking the epitopes in any desired order to opti- geted amino acids could improve the yield of intact epitopes.
mize immunogenicity. Ideally, antigen processing through- This strategy was used to construct a multiepitope vaccine
out the construct would produce multiple epitopes that from HIV antigens.482 Twenty-one epitopes were identified,
could be presented independently. In the case of HIV, where based on the T-cell response of infected humans, and these
antigenic variation is so common, these vaccines could be were linked together in a multiepitope DNA vaccine. Spacers
designed to overcome variation by including epitopes that of one to four amino acids were inserted between epitopes to
are conserved within each geographical region (clades). optimize antigen processing. The DNA vaccine was used to
For example, Létourneau et al.478 expressed a string of immunize HLA transgenic mice, and the response to each
HIV sequence fragments from the gag, pol, vif, and env epitope was compared to peptide vaccinated mice. For two
genes that were highly conserved within their own clades. HLA types, A2 and A3, the peptide and multiepitope vac-
As they progressed from one fragment to the next, they al- cines were nearly equal, suggesting efficient processing. In
ternated sequences from different HIV clades: Immunity addition, for HLA-A2, the potency of the T-cell response to
to multiple epitopes would produce broad based immunity different epitopes was largely independent of their position
across clades. Fourteen fragments were linked in tandem. in the vaccine sequence, suggesting efficient antigen process-
Each fragment contained multiple T-cell epitopes, for a total ing at various points throughout the chimeric protein. But
of 270 known epitopes recognized with human MHC. In for a third HLA type, B7, despite a normal peptide response,
addition, T cells from most patients infected with HIV re- the multiepitope vaccine failed to elicit immunity to four
sponded to peptide pools from the vaccine, indicating that out of seven epitopes tested, indicating that processing was
the epitopes were processed and presented to T cells. But limiting the response to these residues.
when the multiepitope vaccine was given as a DNA vaccine Design of an effective multiepitope vaccine depends on
followed by two viral vector boosts, it elicited relatively weak finding the optimal balance between antigen processing and
T-cell responses to two out of six peptide pools in HLA-A2 epitope survival. Antigen processing between epitopes gen-
transgenic mice. erates antigenic peptides, but processing within the epitope
For these constructs to function as multiepitope vac- would destroy its antigenicity. Delamarre et al.483 have com-
cines, antigen processing between epitopes is required. In pared pairs of proteins with the same sequence but different
any given string, all of the epitopes are expressed equally, conformations, resulting in different susceptibilities to lyso-
but different locations might be processed differently. In somal processing. Ribonuclease (RNAse) A was relatively sta-
order to elicit T-cell immunity to multiple epitopes at the ble to lysosomal extracts of dendritic cells, whereas RNAse S
same time, there must be efficient processing and present- was degraded quickly. Of these two forms of the same protein,
ing of epitopes in all parts of the string. For example, Depla RNAse A was a far more potent immunogen, as measured by
et al.479 created a vaccine with 30 epitopes from hepatitis B T-cell proliferation or Delayed Type Hypersensitivity (DTH).
virus that were restricted to five different MHC class I types. Similarly, horseradish peroxidase is stable to lysosomal pro-
The immune response to the combined antigen was com- teases in the native form, but is quickly degraded in the apo
pared to individual peptides as an indication of the effect form. Native horse radish peroxidase (HRP) was a much
of antigen processing on presentation of internal epitopes. more potent immunogen than the unstable apo form. These
For two MHC types, the same 5 out of 12 epitopes were results indicate the harmful effect of excessive processing on
immunogenic, regardless of whether they were presented as T-cell stimulation and immunogenicity.
a string of epitopes or as free peptides. However, for three Multiepitope immunogens have been designed to elicit
other MHC types, whereas 14 out of 18 epitopes were im- CTLs by stimulating T helper cells and CTLs simultane-
munogenic as peptides, only 3 out of 18 were immunogenic ously. These immunogens have combined T helper and CTL
in the multiepitope vaccine. This difference suggests inef- epitopes in a variety of ways. Some vaccines contain at least
ficient antigen processing at multiple sites, either because of one T helper epitope and one CTL epitope on each poly-
failure to process the epitopes correctly or due to excessive peptide chain.484 Others have expressed them separately:
degradation. A polypeptide consisting of multiple T helper epitopes was
One way to control processing was suggested by Livingston coadministered with a DNA plasmid coding for multiple
et al.480,481 These authors found that the first amino acid fol- CTL epitopes.485 A third way would be to express alternating
lowing each epitope is an important marker for antigen T helper and CTL epitopes in tandem in the same multi-
processing. They demonstrated that poor presentation of epitope vaccine, either as DNA or polypeptide. The T-cell
an internal epitope can be corrected by adding a spacer of response to paired T helper epitopes was generally weak,484
as little as one amino acid between epitopes to improve the whereas the response to a string of multiple T helper epi-
yield of processed peptides. The most favorable amino acid topes was reportedly stronger, although it did not confer
spacers were Lys, Arg, Asn, and Gln, followed by Cys, Gly, any improvement on the immune response to CTL epitopes
Ala, Thr, or Phe. The least active spacer residues were Asp, expressed separately.485
Val, Met, or Leu.480,481 For example, an epitope from hepa- As with T-cell help for B cells, a multiepitope vaccine
titis B virus core antigen was poorly immunogenic when it for CTLs should combine T helper and CTL epitopes in
was located next to a Phe residue, but its immunogenicity the same construct.402 It would deliver both epitopes into

the same APC, to facilitate helper T cell effects on CTL in- chain. Similarly, in systemic lupus erythematosus, human T
duction. But the epitopes should target different cellular cells specific for the Sm antigen are narrowly restricted to a
compartments, leading to the main processing pathways: T few epitopes that are found on a small group of proteins.489
helper epitopes should travel to endosomes for processing On the Sm-B antigen, three epitopes were recognized. On
and presentation with MHC class II. CTL epitopes should Sm-D antigen, there were two. In each case, alanine scans
traffic to the cytoplasm for proteasomal processing and pre- showed that the same epitopes were recognized by distinct
sentation with MHC class I. T-cell clones. These results are consistent with the hypoth-
eses that the autoimmune response to insulin or to Sm an-
tigen may be induced by abnormal exposure of a very few
Defining the Role of Individual Amino Acids and cryptic epitopes, or they may depend on selective loss of tol-
Effects of Altered Peptide Ligands erance for a limited number of epitopes shared by a small
Once an antigenic peptide is identified, the next step is to subset of self-proteins.
map key amino acid residues by making a series of variant TCRs may distinguish different chemical classes of
peptides, each of which differs from the native sequence by amino acid side chains. An example of structural differ-
a single amino acid substitution, as previously described in ences between amino acid side chains recognized by the
the section on MHC binding. One approach, called an ala- TCR comes from an analysis of non–cross-reactive CTL
nine scan, substitutes Ala for the natural amino acid at each that distinguishes homologous peptides from the V3 loop
position in the peptide or uses Ser or Gly to replace natu- of different strains of HIV-1 envelope protein. The residue
rally occurring Ala. Ala is used because the side chain is at P8 in the minimal determinant was identified as a key
only a methyl group, so it replaces whatever functional side
chain is present with the smallest one other than that of Gly,
which is not used because of its effects on conformation.
Thus, one can ask whether loss of the naturally occurring
side chain affects function, without the introduction of a
new side chain that might itself affect function. Generally,
each peptide will have several amino acids where Ala sub-
stitution destroys antigenicity. Some of these correspond
to contact residues for the TCR, whereas others are contact
residues for MHC. In many cases, the MHC-binding resi-
dues can be determined by testing the substituted peptides
in a competitive MHC-binding assay (discussed previ-
ously). The amino acid substitutions that knockout T-cell
proliferation but not MHC binding are presumed to be
in the epitope recognized by the TCR directly, and these
can be studied with additional substitutions. For example,
this technique was used to compare the residues interact-
ing with the MHC molecule or TCR when the same HIV-1
V3 loop peptide P18 (residues 308 to 322) was presented by
three different MHC molecules, a human class I molecule,
a mouse class I molecule, and a mouse class II molecule
(Fig. 23.12).486,487 Interestingly, there was a striking concor-
dance of function of several of the residues as presented by
FIG. 23.12. Comparison of the Major Histocompatibility Complex
all three MHC molecules (see Fig. 23.12). For example, Pro (MHC)-Interacting (“Agretopic”) and T-Cell Receptor (TCR)-Interacting
and Phe interacted with the MHC in all three cases, and the (“Epitopic”) Residues of the Same Human Immunodeficiency Virus
same Val interacted with the TCR in all three cases. Also, (HIV)-1 Envelope V3 Loop Peptide as It Is Presented by Human Class
the same Arg in the middle of the peptide interacted with I, Murine Class I, and Murine Class II MHC Molecules to Cluster of
both the mouse class I and II molecules, and the C-terminal Differentiation (CD)8+ Cytolytic T Cell and CD4+ Helper T Cells. The
Ile was an anchor residue for both human and murine sequence of the optimal binding portion of peptide P18 IIIB from the
HIV-1 envelope protein V3 loop for each MHC molecule, in single
class I molecules.486,487
letter amino acid code, is shown. Arrows pointing up indicate residues
In the case of autoimmune T cells, these techniques determined to interact with the TCR, and arrows pointing down indi-
have been used to study the number and variety of epit- cate residues determined to interact with the MHC molecule. Mapping
opes recognized by self-reactive T cells. In the nonobese of residue function for binding to the human class I MHC molecule
diabetic mouse, the β chain of insulin is a major target of HLA-A2.1 was described in Alexander-Miller et al.,487 and binding to
T cells recovered from pancreatic islets.488 Alanine scanning the murine class I molecule H-2Dd and the murine class II molecule
of β chain peptide 9 to 23 revealed two patterns of T-cell I-Ad was described in Takeshita et al.486 Note the common use of the
Pro and Phe for binding all three MHC molecules and the use of the key
recognition for the same peptide. Some T cells recognize Val residue for binding all the TCRs. Also, the murine MHC molecules
peptide 9 to 16, whereas others respond to peptide 13 to 23. both use the central Arg residue as a contact residue, whereas both
Each epitope appears to have distinct sites for MHC and class I molecules use the C-terminal Ile residue as an anchor residue.
TCR binding, even though they come from the same peptide Thus, there is a surprising degree of concordance.

“epitopic” TCR contact residue in both strain IIIB, which A

has a Val at this position, and strain MN, which has a Tyr at
this position.486,490,491 CTLs specific for strain IIIB do not rec-
ognize the MN sequence but will recognize peptides identi-
cal to MN except for the substitution of any aliphatic amino
acid at that position, such as Val, Leu, or Ile.400 In contrast,
CTL specific for the MN strain do not recognize the IIIB
sequence but will recognize the IIIB peptide if the Val at this
position is replaced by a Tyr.490 Moreover, they will see any
MN variant in which the Tyr is replaced by another aromatic
amino acid, such as Phe, Trp, or His.400 Thus, the two non–
cross-reactive TCRs see similar peptides but discriminate
strongly between peptides with amino acids with aliphatic
versus aromatic side chains. On the other hand, they do not B
distinguish strongly among different aliphatic residues or
among different aromatic residues. Interestingly, however,
in each category, the least active is the bulkiest member of
the category, Ile and Trp, respectively, suggesting that these
residues must fit into a pocket of limited size in the TCR.
The interaction of peptide ligand with TCR can be
studied by introducing single substitutions of conservative
amino acids at these contact residues, such as Glu for Asp,
Ser for Thr, or Gln for Asn. The TCR readily distinguishes
among peptides with these minor differences at a single
residue, and the results have been revealing. Depending on
affinity for the TCR, closely related (altered) peptides can
elicit very different responses in T cells. Thus, although a
substituted peptide may be very weak or nonstimulatory by FIG. 23.13. Differential Effect of Altered Peptide Ligands on the
itself, it may still act as a partial agonist or even a strong Response to Peptide 64 to 76 from Hemoglobin. A: Proliferative
antagonist of an ongoing T-cell response. Antagonistic response of a T-cell line incubated with antigen-presenting cells
and the natural Hb (64-76) peptide or with peptides substituted at
peptides can be demonstrated by pulsing APCs with native positions 72, 73, or 76. B: Interleukin 4 release by the T-cell line
peptide antigen first so that one is not measuring compe- under the same conditions. The peptide substituted with Asp for
tition for binding to MHC molecules, followed by pulsing Glu at position 73 is unable to induce T-cell proliferation, but it can
with a 10-fold or greater excess of the antagonist before add- still induce production of interleukin 4, so it is a partial agonist.
ing T cells. In the case of influenza hemagglutinin peptide In contrast, substitution of Gln for Asn at position 72 knocks out
307 to 319 presented with HLA-DR1, peptide analogues such both responses equally. Modified from Evavold and Allen,495 with
permission.
as Gln substituted for Asn 313 inhibited the proliferation of
a human T-cell clone, even though they did not stimulate
the clone. Anergy was not induced, and the antagonist pep- binding in a competitive-binding assay.496 Similar alteration
tide had to be present throughout the culture to inhibit the of cytokine profi le by altering the peptide ligand can be seen
response.492,493 Thus, lack of antagonist activity is another in other systems.493,497,498
feature of the interaction between peptide (in complex with For one of the Hb 64 to 76 specific T-cell clones, PL.17,
MHC molecule) and TCR that is required for the peptide to substitutions at amino acids 70, 72, 73, or 76 reduced anti-
be a stimulatory antigenic determinant. genic potency by 1000-fold or more, even though conserva-
Partial agonists were first demonstrated using T-cell tive amino acids were substituted. Although substitution of
clones specific for an allelic form of mouse hemoglobin. Ser for Ala 70 prevented T-cell stimulation in both assays,
These T cells were from CE/J mice, which express the Hbs there was clearly some response to this peptide, as it induced
allele of mouse hemoglobin, after immunization with the expression of the interleukin 2 receptor.499 In addition, once
Hbd allele. The minimum antigenic peptide corresponds to T cells were exposed to the Ser 70 peptide, they became
amino acids 67 to 76 of the Hbd sequence and differs from unresponsive to subsequent exposure to the natural Hbd
Hbs at P72, P73, and P76.494 Peptides substituted at each peptide. This phenomenon closely resembled T-cell anergy
residue from amino acid 69 to 76 were tested for T-cell pro- and persisted for a week or more. The Ser 70 substitution
liferation and cytokine release. Some substitutions, such as alters a contact residue of the peptide for the TCR of clone
Gln for Asn at P72, blocked T-cell stimulation completely PL.17 and affects its affinity. Other T-cell clones, however,
in both assays. Other substituted peptides, such as Asp can respond to this peptide presented on the same MHC
for Glu at P73, lost T-cell proliferation but still stimulated molecule (I-Ek). Other Hbd peptides substituted at this posi-
interleukin 4 release, and these are considered partial ago- tion, such as Met and Gly 70, also induced anergy but not
nists (Fig. 23.13).495 Lack of stimulation was not due to failure proliferation, whereas nonconservative substitutions such as
to bind MHC, as both substituted peptides gave reasonable Phe, Asn, Asp, and His 70 induced neither.500

Another well-studied example is influenza hemaggluti- occur. This abnormal pattern occurred regardless of the
nin peptide 306 to 318 as presented on human HLA-DR1. method of anergy induction.
Based on the known crystal structure of the peptide-MHC Partial signaling may be important for T-cell survival
complex,253 amino acid substitutions could be targeted to during negative selection in the thymus or in maintaining
contact residues for the TCR at P307, P309, P310, P312, P315, peripheral tolerance. By responding to self-antigens as if
and P318.501 At each position, nonconservative substitutions they were altered ligands presented in the thymus, T cells
often rendered the peptide inactive, whereas conservative could use anergy induction as a successful strategy for
substitutions at several sites either gave full antigenicity or avoiding clonal deletion. Similarly, peripheral tolerance may
gave progressively lower stimulatory activity, down to 1000- be an important mechanism for preventing autoimmune
fold less than native peptide while retaining the ability to disease. Immunotherapy with altered peptide ligands could
induce anergy. For example, substituting His or Gly for Lys be envisioned as a way to block an ongoing response or
307 gave 1000-fold reduced stimulation of T-cell prolifera- induce tolerance to a specific antigen, such as the synovium
tion but full ability to induce tolerance. Similarly, substi- in arthritis, or foreign MHC antigens in allograft rejection.
tuting His for Lys 315 gave complete loss of stimulation but However, a potential pitfall is that different T cells recognize
nearly full anergy inducing activity. As before, induction of the same peptide differently, so a peptide that is seen as an
the interleukin 2 receptor (CD25) was a sign of T-cell activa- altered peptide ligand by some T-cell clones may be seen as
tion by these altered peptide ligands, even when they did not a complete antigen by others. In addition, the choice of pep-
induce proliferation. Unlike these peptides, the antagonists tide would vary with MHC type. To be effective, an altered
do not induce interleukin receptors or secretion, and they peptide ligand should antagonize or anergize polyclonal
do not cause long-lasting tolerance. Overall, a number of T cells and should work with each patient’s MHC type.
altered peptide ligands have now been identified that, in ap- A similar mechanism may be invoked to explain the gen-
propriate complexes with MHC molecules, induce anergy or erally weak immunogenicity of tumor antigens. According
act as antagonists of the TCR and block activation by agonist to this hypothesis, the only T cells capable of responding to
ligands by delivering an abortive signal.493 self-antigens on tumors may have low-affinity receptors for
Several methods have been found to anergize T cells to a them. In effect, the natural sequence is the altered ligand
specific antigen for up to a week, and all have the common that induces tolerance. In some cases, this anergy can be
theme of delivering a partial signal via the TCR, resulting overcome with modified peptides that have greater affinity
in tolerance rather than stimulation. The first method was for the TCR and induce a full stimulatory signal, resulting
to expose the T cells to peptide plus APCs treated with the in an effective immune response to the tumor antigens,474 as
carbodiimide cross-linker ECDI.502 This treatment may previously described under epitope enhancement.
prevent accessory molecules on the presenting cell from
interacting with the TCR complex or costimulatory signals Prediction of T-Cell Epitopes
from contributing to T-cell activation. The second method The fact that T cells recognize processed fragments of
was to present peptide on presenting cells with mutated I-E antigens presented by MHC molecules leads to the ironic
molecules.503,504 The third method was to use altered pep- situation that T-cell recognition of antigen, which is more
tide ligands that act as TCR antagonists as described previ- complex than antibody recognition due to the ternary com-
ously.493,500,501 The final method was to block CD4 function plex needed between TCR, antigen, and MHC molecule,
with a monoclonal antibody, which would delay the recruit- may actually be focused on simpler structures of the anti-
ment of CD4 to the engaged TCR.505 Because generation of a gen than those seen by most antibodies specific for native
complete stimulatory signal requires the interaction of TCR protein antigens. In contrast to the assembled topographic
and accessory molecules, modifications that affect either antigenic sites seen by many antibodies,50,51 T cells specific
component can block signaling. An altered peptide ligand, for processed antigens are limited to seeing short segments
with decreased affinity for the TCR, may form an unstable of continuous sequence.177,233 Therefore, the tertiary struc-
complex, which cannot stay together long enough to recruit ture of the protein plays little if any role in the structure of
accessory molecules and generate a complete signal.505,506 the epitope recognized by T cells, except as it may influence
Altered peptide ligands with low affinity for the TCR can processing. However, the structure of the T-cell antigenic site
also act as partial agonists that can compete with optimal itself must be limited to primary (sequence) and secondary
agonists and reduce T-cell stimulation based on a similar structure, the latter depending only on local rather than
mechanism (short dwell time of peptide-MHC complex on long-range interactions. This limitation greatly simplifies
the TCR).199 the problem of identifying structural properties important
Abnormal TCR signaling can be demonstrated by follow- to T-cell recognition because one can deal with sequence in-
ing the activity of protein kinases. Normal signaling pro- formation, which can now be obtained from DNA without
duces phosphorylation of TCR subunits, such as ζ chain, as having a purified protein, and with the secondary structure
well as phosphorylation and activation of receptor-associated implicit therein without having to obtain an x-ray crystallo-
tyrosine kinases, such as ZAP-70. These kinases generate the graphic three-dimensional structure of the native protein—
downstream signal needed for T-cell activation. However, a much more difficult task.
in each case studied, partial antigen signaling resulted in Because the key feature necessary for a peptide to be
ζ chain phosphorylation without phosphorylation or acti- recognized by T cells is its ability to bind to an MHC molecule,
vation of ZAP-70,500,504,505 so downstream activation did not most approaches for predicting T-cell epitopes are based on

predicting binding to MHC molecules. These approaches, an expanded database (p < .001)227 (see Table 23.4). Indeed,
which have been reviewed,424,507–510 can be divided into those when the database was expanded to twice and then four
that focus on specific individual MHC molecules one at a times its original size, the correlation remained highly sig-
time, such as motif-based methods, and those that look for nificant, and the fraction of sites predicted remained rela-
general structural properties of peptide sequences. We shall tively stable (34/48 sites = 71%, p < .003; 61/92 sites = 66%,
discuss first the methods based on general properties and p < .001).517,518 A similar correlation was found for 65% of
then those directed to individual MHC molecules. peptides presented by class I MHC molecules.518 A primary
The first structural feature of amino acid sequences sequence pattern found in a substantial number of T-cell
found associated with T-cell epitopes that remains in use epitopes by Rothbard and Taylor519 was consistent with one
today is helical amphipathicity,227,511–514 which is statistically turn of an amphipathic helix. Also, another approach, called
significant independent of the tendency to form a helix per the “strip-of-the-helix” algorithm, which searches for heli-
se.512 Because the x-ray crystallographic structures of both ces with a hydrophobic strip down one face, found a cor-
MHC class I376–378,515 and II molecules253,443 have consistently relation between amphipathic helices and determinants pre-
shown peptides to be bound in extended, not α-helical, sented by both class II and I MHC molecules.513,520
conformation, helicity per se has been abandoned as an Newer data suggest at least two explanations, not
associated structural feature of T-cell epitopes. However, as mutually exclusive, for this correlation in the absence of
discussed in the following text, other explanations of am- helical structure found in the peptides bound to MHC mol-
phipathic structures have been discovered that do not re- ecules.521 First, crystal structures of peptides bound to class
quire the peptide to be bound to the MHC molecule as an α II MHC molecules have found that the peptides are bound
helix. Amphipathicity is the property of having hydrophobic in an extended conformation, but with a −130-degree twist
and hydrophilic regions separated in space. It was observed like that of a type II polyproline helix, 253,443 and can be
that the immunodominant T-cell epitopes of myoglobin and quite amphipathic because of this twist.253 Although the
cytochrome c corresponded to amphipathic helices.185,189,516 −130-degree twist is distinct from that of an α helix, it
DeLisi and Berzofsky511 developed an algorithm to search for gives a periodicity similar enough to be detected. Second,
segments of protein sequence that could fold as amphipa- it was observed that spacing of the anchor residues in the
thic helices based on the idea that the hydrophobicity of motifs for peptides binding to class I and II MHC mole-
the amino acids in the sequence must oscillate as one goes cules was consistent with the spacing of turns of an α helix,
around an amphipathic helix. For the hydrophobic residues for example, at P2 and P9 (seven residues apart like two
to line up on one side and the hydrophilic residues on the turns of a helix) or at P5 and P9 (spaced like one turn of a
other, the periodicity of this oscillation must be approxi- helix).521 Because the anchor residues are most often hydro-
mately the same as the structural periodicity of the helix, phobic, this pattern resulted in an amphipathic periodic-
about 100 degrees per residue (360 degrees per 3.6 residues ity pattern like that of an amphipathic α helix for just the
per turn) (Fig. 23.14). anchor residues alone, seen in most motifs.521 Thus, if the
A microcomputer program implementing this analysis other residues have a random pattern, the anchor residue
was published.514 Margalit et al.227 optimized the original spacing alone, which is enforced by the spacing of the pock-
approach,511 correctly identifying 18 of the 23 immunodom- ets in the MHC molecules that bind these anchor residues,
inant helper T cell antigenic sites from the 12 proteins in will produce the amphipathic helical signal, even though
the peptide is bound in an extended conformation. This
amphipathic helical periodicity has held up as a correlate
for peptides defi ned as T-cell epitopes521 and has continued
to be a useful predictive tool for identifying potential epit-
opes, successful in a number of studies, when one does not
want to focus on individual MHC alleles or wants to fi nd
regions of high epitope density. Other structural properties
such as coil content and exposure have been used as general
predictors of MHC class II binding.522
Other approaches to predicting T-cell epitopes are gen-
erally based on sequences found to bind to specific MHC
molecules.424,507–510,523–525 The simplest approach is to apply
standard sequence search algorithms to known protein
sequences to locate motifs for peptides binding to particular
MHC molecules, using collections of motifs identified in the
literature.264 This approach showed early success for epitopes
in proteins from Listeria monocytogenes526 and malaria,527
FIG 23.14. Plot of hydrophobicity of each amino acid in sperm whale
myoglobin 102 to 118, according to the scale of Fauchère and Pliska,604 but it also became apparent that only about 30% of se-
as a function of amino acid sequence, showing least-squares fit of a quences bearing motifs actually bound to the corresponding
sinusoidal function to the sequence of hydrophobicities from 107 to MHC molecules.526,528,529 This discrepancy may relate to ad-
117. (From Berzofsky et al.,180 with permission.) verse interactions created by nonanchor residues392,397,530 and

could be overcome to some extent by generating extended ecules for which more data exists. This matrix approach has
motifs taking into account the role of each residue in the been used for both class I and II MHC molecules. Current
sequence.397,530 predictive matrix algorithms, compared to experimental
To determine whether one could locate regions of pro- screening, have been found to detect the vast majority of
teins with high densities of motifs for binding multiple CD8 + T-cell epitopes detected experimentally, for example,
MHC molecules, Meister et al.531 developed the algorithm in vaccinia virus, in which the top-ranked 300 peptides
Epimer, which determined the density of motifs per length using four different algorithms predicted 40 per 49 epitopes
of sequence. A surprising result was that the motifs were not found.546
uniformly distributed but clustered. This clustering may re- A potentially very useful observation is the finding that
flect the fact that many motifs are related, and that the same HLA class I molecules can be grouped into families (HLA
anchor residues are shared by several motifs, perhaps because supertypes) that share similar binding motifs.438,536,547,548
MHC molecules are also related and their variable segments The broader motifs that encompass several MHC molecules
that define some of the binding pockets are sometimes ex- have been called supermotifs. For example, HLA-A∗ 0301,
changed by gene conversion events.532 This hypothesis has A∗1101, A∗3101, A∗3401, A∗ 6601, A∗ 6801, and A∗7401 all fall
now been confirmed and extended by studies showing that into the HLA-A3 superfamily.547 A peptide that carried this
each anchor pocket can be grouped into families of MHC supermotif should be active in a broader range of individu-
molecules sharing similar pockets and therefore anchor als than one which was presented by a single HLA molecule.
residues, but the families for the B, C, and F pockets do not Moreover, as several HLA supertypes have been defined, it
coincide, so there is a reassortment between pockets.533,534 should be possible to design a vaccine effective in a large
These observations allow prediction of motifs for additional fraction of the population with only a limited number of
MHC molecules. In the case of HIV, the densities of motifs well-selected antigenic determinants.549
for class I MHC binding were anomalous at both the low and Another approach for predicting peptides that bind to
high ends of the spectrum.535 Clustering at the high end may MHC molecules is based on free energy calculations of pep-
be due to anchor sharing and showed no correlation with tides docked into the groove of a known MHC structure,
conserved or variable regions of the sequence. However, at for which the crystallographic coordinates are known, or
the low end, long stretches with low motif density occurred on structural modeling of the MHC molecule by homolo-
preferentially in variable regions, suggesting that the virus gous extension from another MHC molecule, when the
was mutating to escape the CTL immune system.535 This crystal structure is not known, followed by peptide docking
clustering may be useful in vaccine development because calculations.550–552 It is important to use free energy rather
identification of sequences containing overlapping motifs than energy, as the latter alone cannot find the most stable
for multiple MHC molecules may define promiscuously pre- orientation of a side chain and cannot correctly rank order
sented peptides that would elicit responses in a broad seg- different side chains at the same position. This approach cor-
ment of the population.531 rectly predicts the structure of several known peptide-MHC
Another type of MHC allele–specific approach is the use complexes when starting with the crystal structure of a dif-
of matrices defining the positive or negative contribution of ferent complex, in each case to within 1.2 to 1.6 Å all-atom
each amino acid possible at each position in the sequence root-mean square deviation.551 Using this structural model-
toward binding to an MHC molecule. A positive or negative ing can allow one to extend motifs to nonanchor positions
value is assigned to each of the 20 possible amino acids that for cases where only anchor residue motifs are known and
can occur at each position in a peptide sequence, and these can allow one to predict new motifs for MHC molecules
are summed to give the estimated potential of that peptide whose motifs have not yet been determined.
for binding. The values in the matrix are derived from ei- Another approach to predicting MHC-binding sequences
ther experimental binding studies using peptide panels uses a technique called threading that has been developed
with single positions substituted with each possible amino for predicting peptide secondary structure, based on thread-
acid,439,536–539 or from comparisons of peptides known to ing a sequence through a series of known secondary struc-
bind in a compilation of the literature, if the number known tures, and calculating the energies of each structure. Altuvia
is sufficiently large.507,540 One improvement described is an et al.553,554 showed that threading could be applied to pep-
amino acid similarity matrix that disfavors substitutions tides in the groove of MHC molecules, because when several
with opposite charge.541 Davenport et al.542,543 also developed peptides that bind to the same MHC molecule are compared
a motif method based on Edman degradation sequencing of crystallographically, the conformation of the peptides is
pooled peptides eluted from MHC molecules. All of these fairly similar, as for example in several peptides cyrstallized
methods have had some success in predicting peptides bind- bound to HLA-A2.1.515 In testing the threading approach,
ing to particular MHC molecules,507 but they all require the Altuvia et al.553,554 showed that known antigenic peptides
assumption that each position in a peptide must be acting are highly ranked among all peptides in a given protein
independently of its neighbors, which is a reasonable first sequence, and the rank order of peptides in competitive
approximation, but exceptions are known.544,545 The more binding studies could be correctly predicted. The advantage
experimental data that goes into generating the matrix, the of this approach is that it is independent of known binding
more reliable the predictions. Therefore, the predictive suc- motifs and can identify peptides that bind despite lack of the
cess may be greater for some of the more common HLA mol- common motif for the MHC molecule in question. It can

also be used to rank a set of peptides all containing a known between helper T cell specificity and B cell specificity for the
motif. For both class I and II MHC–binding peptides, a same protein antigen.
number of structure-based predictive approaches have been Early evidence that helper T cells might influence the
developed as well.522,555–558 specificity of the antibodies produced came from a num-
Finally, artificial neural networks can be trained on a set ber of studies showing that Ir genes, which appeared to act
of peptides that bind to a given MHC molecule to recognize through effects of T-cell help, could influence the speci-
patterns present in binding peptides.559,560 When the predic- ficity of antibodies produced to a given antigen.181,576–583 It
tions of the artificial neural networks are tested, the results was hard to imagine how MHC-encoded Ir genes could
can be used to further train the artificial neural networks determine which epitopes of a protein elicit antibodies,
to improve the predictive capability in an iterative fashion. when such antibodies are generally not MHC restricted.
Recently, predictive algorithms for proteasomal cleavage One explanation suggested was that the Ir genes first select
sites561,562 and TAP-transported peptides319,563 have been which helper T cells are activated, and these in turn influ-
developed. Combining these into approaches to predict ence which B cells, specific for particular epitopes, could
epitopes based on all the steps a peptide must pass through, be activated.130 Because, for cognate help, the B cell has to
cleavage, transport, and MHC binding (except for TCR present the antigen in association with an MHC molecule to
binding), has led to the most recent comprehensive algo- the helper T cell, the Ir gene control of antibody specificity
rithms for epitope prediction that can achieve up to 72% must operate at least partly at this step by selecting which
sensitivity.286,320,508,564–568 helper T cell can be activated by and help a given B cell.
As all these methods are further developed and refined, Conversely, if the helper T cell selects a subset of B cells to
they promise to allow accurate prediction of peptides that be activated on the basis of their antibody specificity, then
will bind to different MHC molecules and thus allow the there is a reciprocal interaction between T and B cells influ-
design of vaccines without empirical binding studies until encing each other’s specificity. Therefore this hypothesis was
the end of the process. In addition, empirical high through- called “T–B reciprocity.”130 Steric constraints on the epitopes
put methods for epitope mapping have been developed to that could be used by helper T cells to help a B cell specific
speed the latter.569 Further, localization of clusters of adja- for another particular epitope of the same protein were also
cent or overlapping binding sequences in a short segment of proposed by Sercarz et al.584
protein sequence can also be useful for selecting sequences The concept was first tested by limiting the fine specificity
that will be broadly recognized. Predictive algorithms for of helper T cells to one or a few epitopes and then deter-
locating T-cell epitopes also have the potential use to reduce mining the effect on the specificities of antibodies produced
immunogenicity of therapeutic proteins (such as hormones, in response to the whole molecule. This was accomplished
cytokines, and monoclonal antibodies) against which one by inducing T-cell tolerance to certain epitopes585 or using
would not want an immune response to develop.570,571 For T cells from animals immune to peptide fragments of the
this purpose and others, it is useful to be able to predict all protein.128,586,587 In each case, the limitation on the helper
the epitopes likely to be recognized not just with a single T-cell specificity repertoire influenced the repertoire of
MHC allele but with all of a person’s MHC molecules or antibodies produced.
even those of a larger population; such methods have also One purpose of the B-cell surface Ig is to take up the
been developed.572 specific antigen with high affinity, which is then internal-
ized by receptor-mediated endocytosis and processed like
any other antigen.588–595 Therefore, the explanation was
RELATIONSHIP BETWEEN HELPER T-CELL proposed that the surface Ig, which acts as the receptor to
EPITOPES AND B-CELL EPITOPES ON mediate endocytosis, sterically influences the rate at which
A COMPLEX PROTEIN ANTIGEN different parts of the antigen are processed because what
As we have seen, the factors that determine the location of the B cell is processing is not free antigen but a monoclonal
antigenic sites for T and B cells, with the possible excep- antibody–antigen immune complex.130 This concept presup-
tion of self-tolerance, are largely different. Indeed, if B cells poses that many antibody–antigen complexes are stable near
(with their surface antibody) bind sites that tend to be espe- pH 6 in the endosome, and that what matters is the kinetics
cially exposed or protruding—sites that are also more ac- of production of large fragments, rather than the products
cessible and susceptible to proteolytic enzymes—then there of complete digestion, when both the antigen and the anti-
is reason to think that T cells may have a lower probability body may be degraded to single amino acids. Such protec-
of being able to recognize these same sites, which may be tion from proteolysis of antigen epitopes by bound antibody
more likely to be destroyed during processing. Certainly, as- can be demonstrated at least in vitro.57 More recently, the
sembled topographic sites will be destroyed during process- effect of antigen-specific B-cell surface Ig on the fragments
ing. On the other hand, there are examples in which T cells produced by proteolytic processing of antigen was elegantly
and antibodies seem to see the same, or very closely overlap- demonstrated by Davidson and Watts.596 They demonstrated
ping, sites on a protein,43,185,215,573–575 although fine specificity that the pattern of fragmentation of tetanus toxoid, as mea-
analysis usually indicates that the antibody and T cell fine sured by SDS–polyacrylamide gel electrophoresis, produced
specificities are not identical. The question dealt with here during processing by B-lymphoblastoid clones specific for
is whether there are any functional or regulatory factors in tetanus toxoid varied among B-cell clones depending on
T cell–B cell cooperation that would produce a relationship their specificity for different epitopes within the antigen.

Binding to the antibody may also influence which frag- The issue of whether bound antibody enhanced or sup-
ments are shuttled to the surface and which are shunted into pressed presentation of specific determinants to T cells was
true lysosomes for total degradation. Thus, different B cells explored further by Watts et al.600 and Simitsek et al.601
bearing different surface Ig would preferentially process the They first found that a particular tetanus toxoid–specific
antigen differently to put more of some potential fragments Epstein-Barr virus–transformed human B-cell clone 11.3
than others on their surface in contrast to nonspecific pre- failed to present the tetanus toxoid epitope 1174 to 1189 to
senting cells that would process the antigen indifferently. By specific T cells, whereas it presented another epitope as well
this mechanism, it is proposed that B-cell specificity leads as did other B cells, and another B-cell clone presented the
to selective antigen presentation to helper T cells and, there- 1174 to 1189 epitope well. Moreover, the free 11.3 antibody
fore, to selective help from T cells specific for some epitopes also inhibited presentation of this epitope to T cells at the
more than from T cells specific for others.130 same time that it enhanced presentation of other epitopes
To test this hypothesis, Ozaki and Berzofsky129 made by Fc receptor facilitated uptake.600 They subsequently
populations of B cells effectively monoclonal for purposes found that the same 11.3 B cell and antibody actually en-
of antigen presentation by coating polyclonal B cells with hanced presentation of another epitope of tetanus toxoid,
a conjugate of monoclonal antimyoglobin coupled to anti- 1273 to 1284, by about 10-fold, even though both epitopes
IgM antibodies. B cells coated with one such conjugate pre- were within the footprint of the antibody as determined by
sented myoglobin less well to one myoglobin-specific T-cell protection from proteolytic digestion.601 The enhancement
clone than to others. B cells coated with other conjugates could be mediated also by free antibody as well as Fab frag-
presented myoglobin to this clone equally well as to other ments thereof, indicating that the mechanism did not in-
clones. Therefore, the limitation on myoglobin presentation volve Fc receptor facilitated uptake. Furthermore, the 11.3
by this B cell to this T-cell clone depended on the specificity antibody had no effect on presentation of another deter-
of both the monoclonal antibody coating the B cell and the minant in the same tetanus toxoid C fragment, 947 to 967,
receptor of the T-cell clone. It happened in this case that both which was not within the footprint of the antibody, and
the monoclonal antibody and the T-cell clone were specific another antibody to the C fragment did not enhance pre-
for the same or closely overlapping epitopes. Therefore, it ap- sentation of 1273 to 1284. The authors concluded that the
pears that the site bound by the B-cell surface Ig is less well same antibody or surface Ig can protect two determinants
presented to T cells. This finding is also consistent with a from proteolysis but sterically hinder the binding of one to
recent study of chimeric proteins in which one or more cop- class II MHC molecules while facilitating the binding of the
ies of an ovalbumin helper T-cell determinant were inserted other.601 The facilitation may involve protection from deg-
in different positions.597 Although the position of the oval- radation. This antibody-mediated enhancement of presen-
bumin determinants did not affect the antibody response to tation of selected epitopes to helper T cells can greatly lower
one epitope, the position did matter for antibody produc- the threshold for induction of a T-cell response and may
tion to an epitope of the chimeric protein derived from insu- thereby elicit responses to otherwise subdominant epitopes.
lin-like growth factor I. An ovalbumin determinant inserted It can also contribute to epitope spreading, for example, in
distal to this epitope was much more effective in providing autoimmune disease, in which an initial response to one
help than one inserted adjacent to the same epitope, when dominant determinant leads to a subsequent response to
both constructs were used as immunogens, even though other subdominant determinants, perhaps by helping for
both constructs elicited similar levels of ovalbumin-specific antibody production, which in turn facilitates presentation
T-cell proliferation in the presence of nonspecific presenting of the other determinants.
cells in vitro as a control for nonspecific effects of flanking Taken together, these results support the concept of T–B
residues on processing and presentation of the helper T-cell reciprocity in which helper T cells and B cells each influence
determinants. However, circumstantial evidence from the the specificity of the other’s expressed repertoire.130 This
Ir gene studies mentioned previously suggests that T cells mechanism may also provide an explanation for some of the
may preferentially help B cells that bind with some degree cases in which Ir genes have been found to control antibody
of proximity to the T-cell epitope, as there was a correlation idiotype.602,603 These relationships probably play a significant
between T cell and antibody specificity for large fragments role in regulating the fine specificity of immune response of
of protein antigens under Ir gene control.127,130,181,579,580,583 both arms of the immune system. Therefore, they will also
Therefore, antibodies may have both positive and negative be of importance in the design of synthetic or recombinant
selective effects on processing. Further studies on presen- fragment vaccines that incorporate both T- and B-cell
tation of β -galactosidase–monoclonal antibody complexes epitopes to elicit an antibody response.
by nonspecific APCs suggest similar conclusions.598,599
Presumably, the conjugates are taken up via Ig fragment c
(Fc) receptors on the presenting cells and processed differ- CONCLUSION
entially according to the site bound by the antibody so that Overall, antibodies and T cells recognize different
they are presented differentially to different T-cell clones. structural features in different contexts or environments,
Thus, non–B-presenting cells can be made to mimic specific and thus complement each other to detect the spectrum
B-presenting cells. This also suggests that circulating anti- of foreign (or self-) antigens encountered. Antibodies rec-
body may have a role in the selection of which T cells are ognize three-dimensional structures on the exposed sur-
activated in a subsequent exposure to antigen. face of molecules either in solution or on a cell surface.

Therefore, they are dependent on the conformation of the TCR. These hurdles limit the number of amino acid se-
antigen and can recognize structures that are assembled on quences that can be recognized and account in part for
the surface of the antigen molecule by the way it folds but immunodominance. Furthermore, the two major subsets
that are not contiguous in the primary sequence. However, of T cells also complement each other by recognizing exog-
they do not generally recognize structures buried within enous and endogenous proteins processed and transported
protein molecules or inside cells. In contrast, T cells are de- through different pathways and presented by difference
signed to recognize short segments of primary amino acid classes of MHC molecules. Thus, the two major classes of
sequence of protein antigens, and thus are not dependent T cells complement the types of structures and the locales
on the conformation of the original protein unless it af- surveyed by antibodies. For example, in the case of viruses,
fects processing. Furthermore, T cells particularly provide antibodies can detect intact virions and shed viral proteins
internal surveillance of proteins inside cells, recognizing in solution as well as viral proteins expressed on the surface
peptide fragments of these presented by MHC molecules of infected cells; CD4 + T cells can recognize viral proteins
that carry these fragments to the cell surface. This surveil- taken up by APCs and processed in an endosomal pathway
lance of intracellular proteins requires a number of hurdles and CD8 + T cells can detect proteins synthesized within
including proteolytic processing of the antigen into frag- the infected cell, whether or not they are ever expressed
ments, transport of these into the compartments where intact on the cell surface or secreted. Together, they pro-
they are loaded onto MHC molecules, binding with ap- vide the immune system with a strategy to detect all forms
propriate specificity to the combining site of specific MHC of foreign invaders as well as to protect the host through
molecules, and then fi nally recognition by an appropriate different effector mechanisms.

Fc Receptors and Their Role in

CHAPTER
24 Immune Regulation and Inflammation
Jeffrey V. Ravetch • Falk Nimmerjahn
HISTORICAL BACKGROUND link between IC diseases and allergic reactions, a prediction

Cellular receptors for immunoglobulins (Igs) were antici- that was confirmed through mouse knockout studies of IgG
pated by the description of cytophilic antibodies of the IgG FcRs.10,11 The FcRs, through their dependence on the immu-
class, identified by Boyden and Sorkin in 1960.1 These anti- noreceptor tyrosine-based activation motif (ITAM) pathway
bodies conferred upon normal cells, like macrophages, the of cellular activation, belonged to the family of immunore-
capacity to specifically absorb antigen. Using sheep red blood ceptors that included the antigen receptors on B and T cells.
cells (RBCs) as the antigen resulted in rosette formation be- Three-dimensional crystal structures have been solved for
tween the cytophilic antisheep RBC antibodies and macro- the low-affinity IgG FcRs12,13 and the high-affinity IgE FcR,14
phages and provided a convenient means of visualization alone and in complex with their Ig ligands,15,16 further estab-
of the binding of cytophilic antibodies with normal cells. lishing the close structural link between these Ig receptors.
Subsequent studies by Berken and Benacerraf2 suggested The functional roles of IgG FcRs were suggested by the
that the crystallized fragment (Fc) of the cytophilic anti- distribution of these receptors on both lymphoid and my-
body interacted with a cell surface receptor on macrophages. eloid cells.17 On myeloid cells, they were presumed to me-
Similar studies on B-lymphocytes extended the generality diate effector cell activation, resulting in phagocytosis,
of these receptors and led to the term Fc receptor (FcR) to antibody-dependent cellular cytotoxicity (ADCC), and re-
denote the surface molecules on lymphoid and myeloid cells lease of inflammatory mediators. However, the well-known
that are capable of interacting with the Fc of immunoglobu- ability of the classical pathway of complement to generate
lin molecules.3 Studies on IgE, IgM, and IgA demonstrated activated fragments in response to ICs capable of induc-
the existence of distinct receptors for those isotypes as well ing inflammatory responses by myeloid cells complicated
on various immune cell types. Detailed biochemical charac- the interpretation of the physiological role of IgG FcRs.
terization of Fc receptors was inaugurated by the studies of Thus, the contribution of IgG FcRs to the mechanism of
Kulczycki et al.4 on the high-affinity IgE FcR of mast cells, IC-mediated inflammation, as distinct from the role of
revealing a hetero-oligomeric abg2 subunit structure. A dis- complement, remained uncertain. Insight into this distinc-
tinction between FcRs for the IgE and IgG isotypes emerged tion was gained through the generation of mouse strains
with the observation of the very high (1010 M−1) binding specifically deficient or blocked in either FcRs10,18–20 or com-
affinity of IgE for its receptor in comparison with the low ponents of the classical complement pathway.21 Studies on
binding (106 M−1) of IgG1 to its receptor. This distinction IC-mediated inflammatory responses in these animals, such
led to the realization that the functional IgG1 ligand was ex- as the Arthus reaction, led to the realization that IgG FcRs
clusively in the form of an immune complex (IC), whereas and not the classical pathway of complement activation
IgE binding occurred through monomer interaction with its were the functional mediators of inflammatory responses
receptor. This difference in binding affinity had significant triggered by ICs.20,22,23 The situation for IgE was less con-
functional implications for the structures of these receptors founding, and the identification and characterization of a
and mechanisms by which each isotype activated its target high-affinity receptor for this isotype on mast cells offered a
cell. Determination of the structure of these receptors was plausible explanation for many of the inflammatory features
facilitated by their molecular cloning, beginning with the of allergic reactions,24 validated later by mouse knockouts
IgG FcRs5,6 and followed by the IgE FcR.7 Two distinct types of this receptor. IgG ICs had also been observed to medi-
of IgG receptors, differing in their transmembrane and cyto- ate suppression of B-cell responses; thus, the presence of an
plasmic sequences, were identified that offered a molecular IgG FcR activity on B cells provided a possible, but unchar-
explanation for the apparent contradictory activation and in- acterized, mechanism for this inhibitory activity. Molecular
hibitory activities attributed to IgG FcRs. The primary struc- characterization of this inhibitory activity for the B-cell
ture of the subunits of the high-affinity IgE FcR revealed FcR, FcRIIB, resulted in the first detailed description of an
homology in the ligand binding a subunit to its IgG counter- inhibitory motif, now termed the immunoreceptor tyro-
parts. However, the extent of similarity between these recep- sine-based inhibitory motif (ITIM),25,26 and the signaling
tors became apparent with the observation that the g chain pathway by which it abrogates ITAM-triggered activation.
subunit was common to both IgG and IgE FcRs, providing The ITIM mechanism is now recognized as ubiquitous and
both assembly and signaling functions to these activation has resulted in the recognition of a large family of inhibitory
receptors.8,9 This common structure suggested a functional receptors on immune cells that function to maintain proper
583

thresholds for activation and abrogate activation responses FcμR, Fca/ μR, the poly-Ig receptor, and the sialylated IgG
to terminate an immune reaction.27 Fc receptors SIGN-R1 and DC-SIGN, contain neither ITAM
FcRs are now recognized as central mediators of anti- nor ITIM motifs.
body-triggered responses, coupling the innate and adaptive
immune responses in effector cell activation.28 In addition Subunit Composition
to these specialized roles, the IgG FcRs have served as an Canonical FcRs are typically type I integral membrane
example of the emerging class of balanced immunorecep- glycoproteins consisting of, at the least, a ligand recogni-
tors, in which activation and inhibition are tightly coupled tion a subunit that confers isotype specificity for the re-
in response to ligand binding. Perturbations in either arm of ceptor. a subunits for IgG, IgE, IgM, and IgA have been
the response have been shown to lead to pathological con- described.17,33–35 These subunits typically consist of two
sequences and have been taken as a paradigm of how these extracellular domains of the IgV type superfamily, a single
systems are likely to work for those paired immunorecep- transmembrane domain and a relatively short intracytoplas-
tors with unknown ligand-binding functions. The newly de- mic domain. In activation FcRs, a signaling subunit of the
scribed roles for FcRs in maintaining peripheral tolerance, g family is often found, resulting in an ag2 complex. The
shaping the antibody repertoire, regulating antigen-present- inhibitory Fc γRIIB molecule, in contrast, is expressed as a
ing cell (APC) maturation, and promoting mast cell survival single-chain receptor. The a subunits have apparent mo-
indicate the diversity of functions that these receptors pos- lecular weights of between 40 and 75 kDa, and share signifi-
sess and their central role in modulating both afferent and cant amino acid sequence homology in their extracellular
efferent responses in the immune response. domains. Alternatively, spliced forms of Fcg RIIB modify
This chapter focuses primarily on the IgG and IgE FcRs, the intracytoplasmic domain of this molecule. For example,
for which substantial data on their structure, function, reg- the B2 form lacks sequences that inhibit internalization
ulation, and role in a variety of physiological and pathologi- and thus demonstrates enhanced internalization of ICs, in
cal conditions are now available. The similarity in structure comparison with Fcg RIIB1. However, all the splice variants
and signaling between those receptors and other members contain the ITIM motif, a necessary and sufficient domain
of this family, such as the IgA FcR and the recently discov- for mediating inhibitory signaling. The conservation of this
ered Fc-receptor for mouse and human IgM (FcμR), is also sequence in mice and humans, its presence in all splice vari-
discussed. These FcRs will be referred to as “canonical Fc ants, and the hyperresponsive phenotypes generated in mice
receptors,” reflecting their shared structural and functional deficient in this receptor all support inhibition as the central
properties. In contrast, lectins such as SIGN-R1 and DC- function of Fcg RIIB. The specific structures of the a sub-
SIGN, which bind IgG Fc when sialylated, represent a novel units of the canonical FcRs are shown in Figure 24.1. The
class of FcRs with distinct structures and functions and will notable exceptions to the general structure just outlined are
be discussed in detail in the following. Other Ig receptors seen for the high-affinity Fcg RIa subunit, which has three
with specialized functions in the transport of Igs, such as extracellular domains; the activation Fcg RIIAa subunit,
the FcRn29 and the poly-Ig FcR,30 will not be discussed here. which does not require additional subunits for assembly
or signaling; and the glycosylphosphatidyl-inositol (GPI)-
linked Fcg RIIIB, which attaches to the cell surface through a
STRUCTURE AND EXPRESSION GPI linkage, rather than through a transmembrane domain.
Molecular Genetics The g subunit is found associated with activation IgG,
Two general classes of canonical FcRs are now recognized: IgE, and IgA FcRs, as well as with non-FcR molecules, such as
the activation receptors, characterized by the presence of a paired Ig-like receptor A (PIR-A) and natural killer (NK) cell
cytoplasmic ITAM sequence associated with the receptor, cytotoxicity receptors, but not with IgM receptors, nor the
and the inhibitory receptor, characterized by the presence of lectins SIGN-R1 and DC-SIGN.36–38 It is required for assem-
an ITIM sequence. These two classes of receptors function bly of the a subunits of these receptors by protecting these
in concert and are usually found coexpressed on the cell subunits from degradation in the endoplasmic reticulum.
surface. Thus coengagement of both signaling pathways is The g chain is found as a disulfide-linked homodimer, with a
the rule, setting thresholds for and ultimately determining short extracellular domain containing the cysteine involved
the physiological outcome of effector cell responses. Among in dimerization, a transmembrane domain, and an intracy-
the factors that determine this threshold level are the ac- toplasmic domain containing the ITAM. An aspartic acid
tual affinities of the individual activating and inhibitory residue found in the transmembrane domain of the g chain is
receptors for a specific IgG ligand and the expression level often associated with a basic amino acid residue in the trans-
of the receptor pairs on immune effector cells. Importantly, membrane domain of the a subunit. The g subunit belongs
the affi nity of different antibody isotypes and subclasses to a gene family that includes the T-cell receptor–associated z
for their respective activating and inhibitory FcRs varies chain and the NK receptor DAP-10– and DAP-12–associated
significantly, thus explaining the differential activity of molecules.39 Fcg RIIIA can associate with the z chain, result-
antibody isotypes in vivo.20,31,32 In addition, alleles of spe- ing in the az2 complex found in human NK cells.
cific FcRs have been described that alter their affi nity for A third subunit is found associated with the activation
individual subclasses thus accounting for the variable re- FcRs Fc e RI and Fcg RIII, the b subunit. This 33-kDa subunit
sponses seen in a population to an antibody (see subsequent has four transmembrane-spanning domains and amino and
discussion). Noncanonical FcRs such as the IgM receptors carboxy intracytoplasmic domains, belonging to the cluster

CHAPTER 24 Fc RECEPTORS AND THEIR ROLE IN IMMUNE REGULATION AND INFLAMMATION | 585
FcγRI FcγRIIA FcγRIIB FcγRIIIA FcγRIIIB FcεRI FcαRI

CD64 CD32 CD32 CD16 CD16 CD89
Structure
ITIM
Subunit
composition γ2 α α α γ2 α β γ2 α α -GPI γ2 α β γ2 α γ2 α
Ka 108 M-1 2x106 M-1 2x106 M-1 5x105 M-1 5x105 M-1 2x105 M-1 1010 M-1 1010 M-1 5x107 M-1
Binding 1. IgG1=IgG3 1. IgG1 1. IgG1 1. IgG1=IgG3 1. IgG1=IgG3 1. IgG1=IgG3 IgE IgE IgA2=IgA2
2. IgG4 2. IgG2=IgG3 2. IgG2=IgG3
Specificity
3. IgG2 3. IgG4 3. IgG4
Macrophages Macrophages Macrophages Mast cells Macrophages Neutrophils Mast cells Mast cells Macrophages
Neutrophils Neutrophils Neutrophils Basophils Mast cells Basophils Basophils Neutrophils
Expression Eosinophils Mast cells Mast cells Basophils Eosinophils Eosinophils
Dendritic Cells Eosinophils Eosinophils NK cells Platelets
Platelets Dendritic Cells Dendritic Cells Dendritic Cells
Dendritic Cells FDC
B cells
Class Activation Activation Inhibition Activation Decoy Activation Activation Activation
-Inducible by -Effector cell -Set threshold -Dominant pathway for effector -Sink for IC -Degranulation -Degranulation -IgA binding
Function
inflammatory activation for effector cell activation by IgG -Focus IC to -Allergic reactions -Allergy -IgA activation
cytokines by IC’s, activation by Fc -In vivo ADCC PMN (Type I) -Antigen of effector cells
-Enhance effector cytotoxic Ab -B cell -Arthus reaction -Synergize caption by DC
responses at repression -IC capture by DC with Fcγ RIIA
inflammatory sites -Maintain -IC mediated DC maturation
-IC capturing by tolerance
DC
FIG. 24.1. Summary of Fc Receptor Structures, Expression Patterns, and in Vivo Functions. The immunoreceptor tyrosine-based activation
motif signaling motif is indicated by the green rectangle; the immunoreceptor tyrosine-based inhibitor motif is indicated as a red rectangle.
Alleles of FcγRIIA and FcγRIIIB and their binding properties are discussed in the text.
of differentiation (CD)20 family of tetraspan molecules.24 IgA receptor is found on chromosome 19 as is the lectin DC-
An ITAM sequence is found in the intracytoplasmic carboxy SIGN (CD209) and the b subunit is on chromosome 11. The
domain. In mast cells and basophils, the b chain assembles Fcg RII–Fcg RIII locus on chromosome 1 is further linked
into an abg2 complex with the a chain belonging to either to a variety of lupus susceptibility genes found in that re-
Fcg RIII or Fc εRI. Its presence is required for assembly of gion, including the Sle1 cluster.45 A locus linked to atopy has
Fce RI in rodents. In humans, however, ag2 complexes of been identified at 11q12–13 and further delineated polymor-
FceRI are found in monocytes, Langerhans cells, and den- phisms of the b chain (I181V and V183L) that are associated
dritic cells (DCs), in addition to the abg2 complexes found with a heightened risk of atopy. However, a direct functional
in mast cells and basophils. The ITAM motif found in the association of these polymorphisms with the known biolog-
b subunit is not an autonomous activation sequence but ical activities of the b chain has not been found.46
functions as a signaling amplifier of the ITAM found in the Polymorphisms in the α chains of the Fc γRs have been de-
g subunits.40 scribed, most notably in Fc γRIIA and Fc γRIIIA; these poly-
morphisms result in differences in binding affinity to specific
Gene Organization, Linkage, and Polymorphisms IgG subclasses.47 For example, a histidine at position 131 in
All a subunits of canonical FcRs share a common gene or- Fc γRIIA results in higher affinity binding to IgG2 and IgG3
ganization, which indicates that the evolution of this family than does an arginine at that position. Similarly, Fcg RIIIA
of receptors resulted from gene duplication from a common with valine at position 158 of the α chain has a higher bind-
ancestor.41 Sequence divergence then resulted in the acquisi- ing affinity for IgG1 and 3 than does the polymorphic form
tion of distinctive specificities for these related sequences. with phenylalanine at that position. These polymorphisms
Most of the genes belonging to the expanded FcR family, translate into a more robust ADCC response for the Fc γRIIIA
including the recently identified Fcμ-receptor, the Fcα /μR, val/val and the FcγRIIA his/his haplotype in vitro and has
the poly-Ig receptor, the group of FcR-homologous or -like been positively correlated with a better clinical response to
(FCRL) proteins, the a chains of Fcg RI, Fcg RII, Fcg RIII, antitumor antibodies inclunding the CD20-specific antibody
and FceRI including the common FcR- g chain (Fig. 24.2), rituximab, the Her2/neu-specific antibody trastuzumab, and
are found on the long arm of chromosome 1.34,35,42–44 This the epidermal growth factor receptor–specific antibody ce-
region is syntenic with a comparable region on mouse chro- tuximab in human lymphoma, metastatic breast cancer, and
mosome 1; however, Fcg RIa and several FCRLs are found colorectal cancer patient cohorts, respectively.48–53 This data
on mouse chromosome 3. In humans, the a subunit of the is consistent with previous results obtained with rituximab

A Mouse: Chromosome 1
Cen
Fcer1g Fcgr3 Fcgr4 Fcgr2B FcrX
Rat: Chromosome 13
Cen
Dog: Chromosome 38
Cen
Chimpanzee: Chromosome 1
Cen
Fcer1g Fcgr2A Fcgr2B FcrX
Fcgr3A
Fcgr2C
Fcgr3B
Human: Chromosome 1
Cen
Fcer1g Fcgr2A Fcgr2B FcrX
Fcgr3A
Fcgr2C
Fcgr3B
B
Chromosome 1
Cen
Fc γRI
FCRL5
FCRL4
FCRL3
FCRL2
FCRL1
Fc εRI α
FCRL6
Fc εRI γ
Fc γRIIA
Fc γRIIIA
Fc γRIIC
FcγRIIIB
FcγRIIB
FCRLA
FCRLB
CD1 cluster
C hFc γRIIA
macFc γRIIA
hFc γRIIB
Fc γRII macFc γRIIB
mFc γ RIIB
ratFc γRIIB
hFc γRI
Fc γRI
mFc γRI
hFc γRIIIA
macFc γRIV
Fc γRIV
mFc γRIV
ratFc γRIV
mFc γRIII
Fc γRIII
ratFc γRIII
FIG. 24.2. Chromosomal Organization of Fc Receptor (FcR) and FcR-Like Genes. A: Localization of FcR genes in different species. B: The
human FcR locus on chromosome 1. Classical FcR genes are shown in dark grey, FCRL genes as light grey, and the CD1 gene cluster on
chromosome 1 as white boxes. Pseudogenes are indicated as hatched boxes. (adapted from Davis et al.44). C: The cladogramm shows
the alignment of selective classical FcRs of humans (h), macaques (mac), mice (m), and rats.
and trastuzumab in mouse models, demonstrating an impor- Several studies have attempted to link specific FcR polymor-
tant role of cellular FcγRs for the activity of these therapeu- phisms or copy number variations to autoimmune diseases,
tic antibodies in vivo.54,55 Four amino acids are polymorphic specifically to systemic lupus erythematosus (SLE).56 Recent
for Fc γRIIIB at positions 18, 47, 64, and 88, which contribute studies have reported associations in susceptibility to SLE in
to the neutrophil antigen polymorphisms for this receptor. both murine and human populations with levels of FcγRIIB

canonical Fc γ Rs non-canonical Fc γ Rs
MHC-I C-type lectin
FcγRI FcγRIIB Fc γRIII Fcγ RIV FcRn SignR1

Mouse
ITIM ITAM
γ2 α α γ2 α γ2 α
High affinity Low affinity

MHC-I C-type lectin
FcγRI FcγRIIB FcγRIIA FcγRIIC FcγRIIIA FcγRIIIB FcRn DC-SIGN

Human
ITIM ITAM ITAM α-GPI

γ2 α α α α γ2 α
FIG. 24.3. Comparison of the Human and Mouse Canonical and Noncanonical Fcγ-Receptor Protein Family. In both species, canoni-
cal FcγRs can be distinguished by their affinity for the antibody Fc-portion (high or low affinity) and by the signaling pathways they
trigger (activating versus inhibitory). The noncanonical neonatal FcR belongs to the family of major histocompatibility class I proteins
and regulates immunoglobulin (Ig)G half life. Mouse SIGNR1 or human DC-SIGN are C-type lectin proteins that can only bind to IgG
glycoforms rich in terminal sialic acid residues and are involved in the anti-inflammatory activity of IgG.
expression or alleles of FcγRIIB.57 Reduced expression of the mouse, in comparison with three genes in the human.43
FcγRIIB on activated B cells, such as memory cells, has been Two genes for activation FcRs, Fc γRIIA and C, are found
seen in patients with SLE and chronic inflammatory demy- in the human and not rodents, which is notable because of
elinating polyneuropathy (CIDP), and is associated with a their unusual single-chain activation structure.65 Both mice
promoter polymorphism.58,59 In mouse strains that develop a and human encode a gene referred to as Fc γRIIIA, although
spontaneous, lupus-like disease, restoring the level of FcγRIIB recent studies have identified a novel mouse FcR with higher
expression on their B cells to a wild-type level reverses dis- homology to the human Fc γRIIIA called Fc γRIV.31,64,66 As
ease.60,61 A higher incidence of an allele of FcγRIIB has been mentioned previously, Fc γRIIIB is unique among FcRs in
reported in several populations with autoimmune disease. being expressed as a GPI-anchored protein (see Fig. 24.3).
This allele, found in the transmembrane domain of the re- Its expression is limited to human neutrophils, in compari-
ceptor, is suggested to result in a hypomorphic phenotype, son with Fc γRIIIA, which is expressed widely on cells of the
similar to the reduced expression observed.57 Of note, despite myeloid lineage, such as macrophages, NK cells, mast cells,
increasing the risk for development of SLE, this FcγRIIB allele and DCs. Finally, both mice and humans have only a single
seems to decrease the likelihood of malaria infections and the gene encoding the inhibitory Fc γRIIB molecule. Among the
severity of disease and can be found at increased frequencies noncanonical FcRs, the lectin DC-SIGN is the functional
in areas of the world where malaria is endemic.62,63 homologue of the murine SIGN-R1 molecule with respect
to its ability to bind sialylated IgG Fc and mediate an anti-
Species Comparisons inflammatory response.67,68 While the overall structure of
Detailed comparisons between the canonical FcRs in mice these proteins are similar (see Fig. 24.3), their patterns of ex-
and humans have revealed several notable differences in pression are quite distinct. SIGN-R1 expression is restricted
both structure and expression of these molecules (Figs. 24.1 to marginal zone macrophages of the spleen and lymph
to 24.3). Whereas IgG, IgM, and IgE FcRs are conserved in node, whereas DC-SIGN expression is seen on myeloid DCs
these species, IgA FcRs are not. To date, a murine homolog for and macrophages in a variety of tissues, including spleen,
the IgA FcR has not been identified. In general, murine and bone marrow, and lymph node.69
human IgG FcRs display comparable degrees of heterogene- The genes for the IgE FcR are conserved in mice and
ity and complexity.41,64 Specific differences, however, have humans. The difference that is observed relates to the
been noted. For example, Fc γRI is encoded by a single gene in requirement for the β chain to achieve surface expression

in mice, precluding the expression of the αγ2 complex.24 In interferon-g, interleukin (IL)-4, and transforming growth
humans, this form of the receptor is widely expressed on factor b.73,74 In addition, complement component C5a bind-
monocytes, Langerhans cells, and DCs, and is likely to be ing to its receptor, C5aR, results in the induction of expres-
found on mast cells and basophils as well. This difference in sion of activation Fc γRs.75 Induction of Fcg RI, Fcg RIIA, and
Fc e RI subunit composition is likely to result in functional Fcg RIIIA a and g chains in myeloid cells occurs upon inter-
differences as well. Although these specific interspecies dif- feron-g treatment: IL-4 generally inhibits expression of these
ferences are important, the fundamental organization of the activation receptors but induces expression of the inhibitory
canonical FcR system, with activation and inhibitory signal- Fcg RIIB. Administration of intravenous gamma globulin, a
ing through a shared ligand specificity coupled to opposing widely used treatment for inflammatory diseases, has been
signaling pathways, is well conserved, as is the role of the shown to induce expression of the inhibitory FcγRIIB on ef-
noncanonical FcRs in mediating the immunomodulatory fector macrophages and B cells in mice and humans.59,76–78
functions of sialylated IgG. Thus, conclusions regarding the This induction is not direct but mediated through other,
function of this expanded FcR system in immunity by the noncanonical FcR pathways including molecules such as
analysis of murine models are relevant to an understanding SIGNR1 and its human orthologe DC-SIGN (see Fig. 24.7).
of the role of these receptors to human immunity as well. Engagement of these receptors by sialylated IgG results in
an intrinsic Th2 pathway, ultimately leading to the expres-
sion of IL-4, which induces FcRIIB expression on inflam-
Expression
matory macrophages.67,68,79,80 The situation in B cells is likely
Canonical FcRs are expressed widely on cells of the myeloid to be more complex, whereby regulation of Fcg RIIB is criti-
lineage, including monocytes, macrophages, DCs, mast cells, cal for the maintenance of peripheral tolerance. Germinal
basophils, neutrophils, eosinophils, and NK cells.17,33 In ad- center B cells downregulate Fcg RIIB, perhaps in response to
dition, B cells and follicular DCs (FDCs) express the inhibi- IL-4 production by T cells. Regulation of FcR expression has
tory Fc γRIIB receptor, whereas T cells are generally nega- also been documented to occur upon binding of ligand. IgE
tive for Fc γR expression, but may express the Fc-receptor regulates the expression of Fc e RI by stabilizing the intra-
for IgM. Interestingly, the majority of FCRL proteins are cellular pool of receptor upon receptor engagement.81 Thus,
expressed during varying stages of B-cell development. high IgE levels result in the induction of surface expression
Despite their homology to the canonical FcRs, however, the of Fc e RI. However, this same mechanism of regulation is
FCRL proteins seem not to bind to Igs, rendering Fc γRIIB not seen for Fcg Rs: mice deficient in IgG have Fcg R levels
the only IgG binding FcR on B cells. The specific expression comparable with those of wild-type animals. Competition
pattern for each FcR varies, and these patterns are summa- for limiting subunits also contributes to regulation of re-
rized in Figure 24.1. Because canonical FcRs represent a bal- ceptor expression. In mast cells, it appears that the level of
anced system of activation and inhibition, the general rule g chain is limiting. Competition between a chains for the
of coexpression of FcRs of these classes is maintained. B cells limiting concentration of g chain has been documented in
use the B-cell antigen receptor as the activation coreceptor mouse knockouts, whereby levels of one receptor increase
for Fc γRIIB, whereas NK cells appear to utilize NK inhibi- if the a chain of the other receptor is reduced.82 This type
tory receptors to modulate Fcg RIIIA activation. The decoy of reciprocal regulation has recently also been observed for
Fcg R, Fcg RIIIB, is expressed exclusively on human neutro- Fc γRs in neutrophils and is likely to be significant in the
phils, on which it functions to concentrate and focus ICs cross-regulation of FcRs by different isotypes of Ig.83
without directly triggering cell activation, perhaps also play-
ing role in neutrophil recruitment.70 The Fcg RIIA–Fcg RIIB
pair functions on neutrophils to modulate IC activation. Three-Dimensional Structure
Fc e RI can be modulated by Fcg RIIB, as demonstrated both The crystal structures of Fcg RIIA, Fcg RIIB, Fcg RIIIA, and
in vitro and in vivo; mice deficient in Fcg RIIB display en- Fc e RI have been solved, as have the cocrystals of Fcg RIIIA–
hanced IgE-triggered anaphylaxis71 by virtue of the ability of IgG1 Fc and Fc e RI–IgE Fc84 (Fig. 24.4). These studies dem-
IgE to bind with high affinity to Fc e RI and with low affin- onstrate that the receptors have a common structure in
ity for Fcg RIIB. Other mast cell inhibitory receptors, such which the two extracellular Ig domains fold in a strongly
as glycoprotein 49B1, modulate mast cell sensitivity to IgE: bent overall structure, arranged into a heart-shaped do-
mice deficient in this molecule display enhanced anaphylac- main structure. A 1:1 stoichiometry between the receptor
tic responses to IgE stimulation.72 Expression of the com- and ligand is observed, with the receptor inserted into the
mon g chain is broad: it has been found on all myeloid and cleft formed by the two chains of the Fc fragment (Cg 2 or
lymphoid cells examined to date. In contrast, the b chain Ce3). The binding region of the FcR to Fc fragments con-
appears to be quite restricted in its expression: it has been sists mainly of rather flexible loops that rearrange upon
found only on mast cells and basophils. complex formation. Only domain 2 and the linker region
Regulation of canonical FcR expression can occur at connecting domains 1 and 2 interact in the complex with
several levels. In general, cytokines involved in activation different regions of both chains of the Fc. Conserved tryp-
of inflammatory responses induce expression of activation tophans located on the FcRs interact with proline to form a
Fcg Rs, whereas inhibitory cytokines downregulate these ac- “proline sandwich.” A solvent-exposed hydrophobic residue
tivation receptors. Transcriptional regulation of a chain lev- at position 155 is conserved among all FcRs and represents
els has been documented for a variety of cytokines, including a binding site for the important IgG1 residue Leu 235 (not

FcγRs produced in bacteria and therefore in an aglycosylated

form. In general glycosylation at Asn 297 is critical for FcγR
binding; modulation of FcγR binding by glycan modification
has been observed for several classes of glycan modifications,
including sialylation, fucosylation, and branching N-acetyl
glucosamine.85,86 The first IgG-FcγR cocrystal structure with
differentially glycosylated IgG variants demonstrated that
carbohydrate-carbohydrate interactions between IgG and FcγR
are essential for high-affinity recognition of afucosylated IgG
glycovariants by FcγRIIIA.87 In contrast, low-affinity binding
of FcγRIIIA to fucosylated IgG is independent of these carbo-
hydrate interactions. These novel insights may offer new ways
of optimizing IgG-FcγR interactions for enhanced therapeutic
activity of IgG.
IN VITRO ACTIVITY
Binding Properties
As outlined in Figures 24.1 and 24.3, Ig binding to canonical
FcRs falls into either high- or low-affinity binding classes.
The high-affinity binding class is typified by Fc e RI, with a
binding affinity of 1010 M−1 for IgE, which ensures a mono-
meric interaction between IgE and its receptor. Fcg RI binds
with relatively high affinity for IgG1 and IgG3 (human) and
IgG2a (mouse) with an affinity constant of 108 M−1. In con-
trast to these high-affinity FcRs, the low-affinity receptors,
such as the human Fcg RIIA, Fcg RIIB, Fcg RIIIA, Fcg RIIIB,
and Fc aRI, and the mouse Fc γRIIB, Fc γRIII, and Fc γRIV
bind with affinities ranging from 5 × 105 M−1 (Fcg RIII) to
5 × 107 M−1. This low-affinity binding ensures that these
receptors interact with ICs and not monomeric ligands. As
described subsequently, this dependence on high-avidity
FIG. 24.4. Ribbon Diagram of Crystallizable Fragment (Fc) Receptor and low-affinity interactions ensures that these receptors
III–Immunoglobulin (IG)G1 Fc Structure. The carbohydrate moiety
attached to both CH2 domains of the IgG Fc-fragment is shown as a are activated only by physiologically relevant ICs and not by
stick and ball model. The extracellular domains of Fc receptor III are circulating monomeric Ig, thus avoiding inappropriate acti-
shown, together with the Fc fragment of IgG1. See text for details. vation of effector responses. In general, low-affi nity Fcg Rs
Adapted from Sondermann et al.,15 with permission. bind IgG1 and IgG3 preferentially; their binding to IgG2
and IgG4 is of even lower affinity. As mentioned previously,
polymorphisms in Fcg RIIA and Fcg RIIIA affect binding to
IgG2 and IgG1, respectively, which may have significance in
found in IgE). Specificity is generated among the receptor– vivo in predicting responses to specific cytotoxic antibod-
ligand pairs in a variable region connecting the two extra- ies (Table 24.1). The isolated consideration of the affinity of
cellular domains that is in contact with the lower hinge re- antibody isotypes to their activating FcγRs is not sufficient,
gion of the Fc fragment (residues 234 to 238), a region not however, to explain the differences in in vivo activity. More
conserved among the IgGs and IgE. The binding region of importantly, the ratio of the affinities of an antibody isotype
FcRs to their Ig ligands does not overlap with other Fc bind- to the activating Fc γRs compared to the inhibitory Fc γRIIB
ing molecules such as protein A, protein G, and FcRn. (termed A/I-ratio) predicts antibody activity more consis-
The structure of the FcR bound to its ligand reveal that the tently.20,64 Thus, antibody isotypes with a high A/I-ratio, such
antigen-binding fragment (Fab) arms are quite sharply bent as mouse IgG2a and IgG2b, will show a greater activity than
and may adopt a perpendicular orientation toward the Fc. This counterparts with a lower ratio (see Table 24.1). The sugar
arrangement would give the Fab arms maximal flexibility to moiety of the antibody Fc-portion is essential for FcR bind-
bind antigen when the Fc fragment is oriented parallel to the ing and the presence or absence of certain sugar residues can
membrane of the FcR-expressing cell. The asymmetrical inter- significantly impact on antibody-FcR binding.20,80,88,89 The
action of the two Fc chains with a single FcR prevents a single absence of fucose, for example, will selectively increase the
antibody molecule from triggering dimerization of receptors affinity of human IgG1 or mouse IgG2b for human Fc γRIIIa
and initiating signaling. Instead, dimerization is initiated or mouse Fc γRIV, respectively. In contrast, the presence of
by the interaction of antigen with the Fab arms, thus link- terminal sialic acid on the N-linked Fc glycan reduces affini-
ing adaptive responses to effector cell triggering. Of note, the ties for FcRs by an order of magnitude with a concomitant
majority of the IgG-FcγR cocrystals have been generated with reduction in in vivo activity.80 Subunit interactions have also

TABLE 24.1 Affinities of Mouse and Human Fc Receptors to Different Antibody Isotypes
Mouse Soluble Fcf R (KA in M -1) Human Soluble Fcf R (KA in M -1)
FcγRI FcγRIIB FcγRIII FcγRIV A/I FcγRI FcγRIIA131R FcγRIIA131H FcγRIIB FcγRIIIA158F FcγRIIIA158V A/Ia
mIgG1 n.b. 3.3 × 106 3.1 × 105 n.b. 0.1 n.d. 2.5 × 105 0.4 × 105 1 × 105 < 104b n.d.
mIgG2a 1.8 × 108 0.42 × 106 6.8 × 105 2.9 × 107 69 n.d. 3.2 × 105 1.7 × 105 1.6 × 105 1.0 × 105 n.d.
mIgG2b n.b. 2.2 × 106 6.4 × 105 1.7 × 107 7 n.d. 9.1 × 104 1.2 × 105 1.2 × 105 0.1 × 105 n.d.
mIgG3 n.b. n.b. n.b. n.b. — n.d. < 104b < 104b < 104b < 104b n.d.
hIgG1 3.8 × 106 2 × 105 3.5 × 104 2.2 × 106 9.1 × 108d 3.5 × 105e 5.2 × 105e 1–3.8 × 105c,e 0.4–1.1 × 106e 1.9–4.8 × 106c,e 1/13
hIgG2 n.b. < 104b < 104b n.b n.b. 1.0 × 105e 4.5 × 105e 0.2 × 105e 0.3 × 105e 0.7 × 105e 1.5/3.5
hIgG3 1.2 × 106 8.3 × 104 n.b. < 104b 6.1 × 107e 9.1 × 105e 8.9 × 105e 1.7 × 105e 7.7 × 106e 9.8 × 106e 45/57
hIgG4 7.2 × 104 < 104b < 104b < 104b 3.4 × 107e 2.1 × 105e 1.7 × 105e 2 × 105e 2.0 × 105e 2.5 × 105e 1
Binding constants were obtained by surface plasmon resonance analysis with immobilized antibodies (FITC-isotype switch variants) and soluble Fc receptors (FcRs) produced by
transient transfection in 293T cells. Shown are the association constants (KA in M−1) of soluble FcR binding to the indicated antibody isotypes as determined by surface plasmon
resonance analysis.31,174–176 IG, immunoglobulin; n.b., no detectable binding or binding that is too low to be evaluated; n.d., that no surface plasmon resonance or other quantita-
tive data are available.
a
The two numbers indicate the A/I ratios for the low and high affinity allele of FcγRIII
b
Indicates detectable but very low binding that did not allow to determine exact binding constants by surface plasmon resonance.
c*
Maenaka et al.174; Okazaki et al.175
d#
Paetz et al.176
e
§Bruhns et al.177
been reported to influence affinity for ligand, as demon- The implications of these structural studies are that the
strated for the common g chain associating with Fcg RIIIA.90 Fc domain of IgG may be selectively mutated to direct its
Its affinity for IgG1 is higher than the GPI-anchored form of binding to specific Fcg Rs. Fc mutants that selectively engage
this receptor, Fcg RIIIB. activation Fc γRs (IIIA and IIA) while minimally interact-
The crystal structures of IgG1-Fcg RIIIA and IgE-Fc eRI ing with inhibitory and decoy Fc γRs (IIB and IIIB) would
reveal similarities in the binding properties of these two confer optimal cytotoxic potential for tumoricidal applica-
complexes. Of significance is the 1:1 stoichiometry of the tions. Indications that such Fc engineering is possible are
complexes, which ensures that a single receptor binds to a suggested by IgG mutants with selective binding to Fc γRIII
single immunoglobulin molecule.15,16 This property in turn or Fc γRII.92–94
ensures that activation occurs upon cross-linking of receptor
complexes by multivalent ligands. Two binding sites on the
receptor interact asymmetrically with two sites on the Fc mol- Effector Cell Activation
ecule. The FcR inserts into the cleft formed by the two chains The critical step in triggering effector cell response by ca-
of the Fc molecule, burying a binding surface of 895 Å for nonical FcRs is mediated by the cross-linking of these re-
each binding site. Alterations in the Fc structure that reduce ceptors by Ig. This can occur either by interactions of
the cleft, such as deglycosylation of IgG, inhibit FcR binding. low-affinity, high-avidity IgG ICs or of IgG opsonized cells
Four distinct regions have been defined in the Fc domains with activation Fcg Rs or by the cross-linking of monomeric
involved in FcR interactions. For IgE, this includes residues IgG or IgE bound to Fcg RI or Fc eR, respectively, by multi-
334 to 336, 362 to 365, 393 to 396, and 424. The homologous valent antigens binding to the Fab of the antibody. Cross-
regions for IgG are residues 234 to 239, 265 to 269, 297 to 299, linking of ITAM-bearing FcRs results in common cellular
and 327 to 332. Interactions of these residues occur with the responses, determined by the cell type, rather than the FcR.
carboxy-terminal domain 2 of the respective FcRs. In view of Thus, for example, Fc eRI or Fcg RIII cross-linking of mast
the similarities of these complexes and the homologies among cells results in degranulation of these cells, whereas cross-
the receptors and their ligands, an obvious question that linking of macrophage expressed Fc aRI or Fcg RIII by opso-
arises concerns the molecular basis for specificity. Attempts nized cells triggers phagocytosis. These functions underlie
to resolve that question have relied on mutagenesis studies the functional similarity of activation FcRs in which cross-
of the ligands and domain exchanges between receptors. For linking mediates cellular responses by ITAM-mediated ty-
example, exchange of the FG loop in domain 2 of Fc e to Fcg rosine kinase cascades. In addition to degranulation and
receptors confers detectable IgE binding; similarly, variation phagocytosis, activation FcR cross-linking has been demon-
in this loop in Fcg Rs may provide interactions that determine strated to induce ADCC, the oxidative burst, and the release
IgG specificity for these receptors. Mutagenesis of IgG1 re- of cytokines and other inflammatory cell mediators. A sus-
vealed that a common set of residues is involved in binding tained calcium influx is associated with these functions, as
to all Fcg Rs, but Fcg RII and Fcg RIII also utilize distinct resi- is transcription of genes associated with the activated state.
dues.91 Several IgG1 residues not found at the IgG–FcR inter- Cellular activation initiated by ITAM-bearing activa-
face by crystallographic determination had a profound effect tion FcRs can be enhanced by coengagement with inte-
on binding, which indicates the greater complexity of these grin and complement receptors. Although the ability of
interactions in solution. these receptors to mediate phagocytosis, for example, are

modest, synergistic interactions with FcRs result in sus- in a fraction of plasma cells thereby creating niches for newly
tained activation and enhancement. Synergistic interactions generated plasma cells during exposure to a new antigenic
between activation FcRs and toll receptors, mannose re- stimulus.99
ceptors, and other pattern-recognition molecules have also
been reported in vitro and suggest that interplay between
the innate and adaptive effector mechanisms of an immune SIGNALING
response are involved in mediating efficient protection from
microbial pathogens.
Immunoreceptor Tyrosine-Based Activation
In contrast to the activation of effector cell responses Motif Pathways
triggered by cross-linking of ITAM-bearing FcRs in vitro, The general features of signal transduction through ITAM
cross-linking of an ITIM-bearing inhibitory receptor to an receptors are conserved among all members of this fam-
ITAM-bearing receptor results in the arrest of these effec- ily, including T-cell receptors, BCRs, and various FcRs. The
tor responses. Homoaggregation of Fc γRIIB by its cross- 19 amino acid–conserved ITAM is necessary and sufficient
linking on effector cells by ICs does not result in cellular to generate an activation response, as demonstrated by the
responses; rather, it is the coengagement of ITAM- and analysis of chimeric receptors. With a single exception, FcRs
ITIM-bearing receptors that results in the functional gen- associate with accessory subunits that contain these signal-
eration of an inhibitory signal. In vitro, it is possible to ligate ing motifs. As described previously, the common g chain con-
any ITAM-bearing receptor to any ITIM-bearing receptor tains an ITAM and is associated with Fc eRI, Fcg RI, Fcg RIII,
with a resulting inhibitory response. This activity is used Fcg RIV, and FcaRI. In addition, both Fc eRI and Fcg RIII may
functionally to define putative ITIMs and has proved to be associate with the b subunit in mast cells. The ITAM found in
a useful device in dissecting the signaling pathways induced the b chain does not function as an autonomous activation
by ITAM-ITIM coligation. cassette, as has been found for most other ITAMs. Rather, it
functions to amplify the activation response generated by the
g chain ITAM by increasing the local concentration of Lyn
B-Lymphocyte Suppression available for activation upon aggregation of the receptor.40
B-cell stimulation through the B-cell antigen receptor can be Fcg RIIA contains an ITAM in the cytoplasmic domain of its
arrested by the coligation of Fcg RIIB to the B-cell receptor ligand recognition a subunit and is thus able to activate in the
(BCR). This occurs naturally when ICs, retained on FDCs in absence of any associated subunit.
the germinal center, interact with both the BCR and Fcg RIIB Upon sustained receptor aggregation, Src family kinases
during the affinity maturation of an antibody response. In that may be associated with the receptor in an inactive form
vitro suppression of B-cell activation has been demonstrated become activated and rapidly tyrosine-phosphorylate the
by coligation of BCR and Fcg RIIB, resulting in arrest of ITAM sequences, creating SH2 sites for the docking and
calcium influx and proliferative responses triggered by the subsequent activation of Syk kinases. Ligands that rapidly
BCR,25,95 the result of recruitment of the SH2-containing ino- dissociate from their receptors result in nonproductive sig-
sitol 5′-phosphatase (SHIP)-1.96 Calcium release from the en- naling complexes that fail to couple to downstream events
doplasmic reticulum is not affected, and there is thus an initial and behave as antagonists.100 The specific Src kinase in-
rise in intracellular calcium; however, this calcium flux is not volved for each FcR depends on the receptor and cell type
sustained, because SHIP recruitment blocks calcium influx in which it is studied. Thus, Lyn is associated with the Fce RI
by uncoupling of the capacitance channel. Homoaggregation pathway in mast cells, Lck is associated with Fcg RIIIA in
of Fcg RIIB by ICs can trigger apoptosis in B cells, as dem- NK cells, and both of these kinases as well as Hck are as-
onstrated in the DT40 B-cell line and in murine splenocyte sociated with Fcg RI and Fcg RIIA in macrophages. After ac-
preparations.97 This activity is retained in ITIM mutants and tivation of the Src kinase, tyrosine phosphorylation of the
is dependent on the transmembrane sequence of Fcg RIIB. ITAM motif rapidly ensues, leading to the recruitment and
Consistent with this, other studies showed that a SHIP- activation of Syk kinases. This two-step process is absolutely
independent but BTK-, JNK1-, and cABL-dependent pathway necessary to transduce the aggregation signal to a sustain-
is involved in induction of apoptosis upon FcγRIIB homoag- able intracellular response. Once activated, Syk kinases lead
gregation.98 The potent inhibitory effect of FcγRIIB on B-cell to the phosphorylation or recruitment of a variety of intra-
activation has recently been used therapeutically to suppress cellular substrates, including PI3K, Btk and other Tec fam-
autoantibody production by B cells. By engineering the Fc- ily kinases, phospholipase C- g (PLCg ), and adaptor proteins
fragment of CD19-specific antibodies to have increased affin- such as SLP-76 and BLNK. The Ras pathway is also activated
ity to FcγRIIB, resulting in an enhanced coengagment of the through Sos bound to Grb2 that is recruited upon phosphor-
inhibitory FcγRIIB with the BCR signaling complex, it was ylation of Shc. Ras phosphorylates Raf, which in turn leads
possible to suppress humoral immunity in mice and to shut to MEK kinase and MAP kinase activation. A summary of
down autoantibody production by B cells.94 In contrast to ma- these intracellular pathways is shown in Figure 24.5A. A
ture B cells, which coexpress the BCR and FcγRIIB, plasma crucial step in this sequential activation cascade occurs with
cells responsible for high-level antibody production down- the activation of PI3K by Syk. By generating phosphatidyl
regulate BCR expression but maintain FcγRIIB expression. It inositol polyphosphates, such as PIP3, PI3K leads to the re-
was suggested that crossliniking of FcγRIIB by ICs on bone cruitment of pleckstrin homology (PH) domain–expressing
marrow plasma cells might be involved in inducing apoptosis proteins such as Btk and PLCg, which in turn leads to the

A B
Fcγ RIII Ca++ FcγRIII FcγRIIB
PiP3
PI3K Btk PLCγ PI3K PI(3,4,5)P3
ITIM
ITAM
ITAM
ITAM
ITAM
ITAM
ITAM
syk P PI(3,4)P2
P SHIP
P DAG
Shc Dok
IP3
PLCγ
Lyn/Lck PKC Lyn
Ca++
SOS Btk
Ras Rac Ras
ER
downstream signaling pathways (ERK, p38, JNK) Inhibition of calcium flux and proliferation
FIG. 24.5. Signaling by Activation and Inhibitory Fc Receptors (FcRs). A: FcγRIII signaling in natural killer cells is shown as an example of the
activation class of FcRs. Cross-linking by an immune complex initiates the signaling cascade. The specific Src family kinase varies, depend-
ing on the cell type. B: FcR signaling in macrophages is shown as an example of simultaneous triggering of activation and inhibitory FcRs.
generation of inositol triphosphate and diacylglycerol, inter- different signaling pathways; calcium inhibition requires
mediates crucial to the mobilization of intracellular calcium the phosphatase activity of SHIP to hydrolyze PIP3 and the
and activation of protein kinase C, respectively. ensuing dissociation of PH domain–containing proteins
such as Btk and PLCg.104 The net effect is to block calcium
influx and prevent sustained calcium signaling. Calcium-
Immunoreceptor Tyrosine-Based Inhibitory dependent processes such as degranulation, phagocytosis,
Motif Pathways ADCC, cytokine release, and proinflammatory activation
The inhibitory motif, embedded in the cytoplasmic do- are all blocked. Arrest of proliferation in B cells is also de-
main of the single-chain Fcg RIIB molecule, was defined as pendent on the ITIM pathway, through the activation of the
a 13 amino acid sequence AENTITYSLLKHP, shown to be adaptor protein Dok and subsequent inactivation of MAP
both necessary and sufficient to mediate the inhibition of kinases.105,106 The role of SHIP in this process has not been
BCR-generated calcium mobilization and cellular prolif- fully defined, inasmuch as it can affect proliferation in sev-
eration.25,26 Significantly, phosphorylation of the tyrosine of eral ways. SHIP, through its catalytic phosphatase domain,
this motif was shown to occur upon BCR coligation and was can prevent activation of the PH domain survival factor
required for its inhibitory activity. This modification gen- Akt by hydrolysis of PIP3.107,108 SHIP also contains phospho-
erated an SH2 recognition domain that is the binding site tyrosine-binding domains that could act to recruit Dok to
for the inhibitory signaling molecule SHIP.96,101 In addition the membrane and provide access to the Lyn kinase that is
to its expression on B cells, where it is the only IgG FcR, involved in its activation. Dok-deficient B cells are unable
Fcg RIIB is widely expressed on macrophages, neutrophils, to mediate Fcg RIIB–triggered arrest of BCR-induced pro-
mast cells, DCs, and FDCs, absent only from T and NK cells. liferation, while retaining their ability to inhibit a calcium
Studies on Fcg RIIB provided the impetus to identify similar influx, which demonstrates the dissociation of these two
sequences in other surface molecules that mediated cellu- ITIM-dependent pathways.
lar inhibition and resulted in the description of the ITIM, a Another inhibitory activity displayed by Fcg RIIB is in-
general feature of inhibitory receptors. dependent of the ITIM sequence and is displayed upon
Fcg RIIB displays multiple inhibitory activities. homoaggregation of the receptor. Under these conditions
Coengagement of Fcg RIIB to an ITAM-containing receptor of Fcg RIIB clustering, a proapoptotic signal is generated
leads to tyrosine phosphorylation of the ITIM by the Lyn through the transmembrane sequence. This proapoptotic
kinase, recruitment of SHIP, and the inhibition of ITAM- signal is blocked by recruitment of SHIP, which occurs upon
triggered calcium mobilization and cellular proliferation coligation of Fcg RIIB to the BCR, because of the Btk re-
(see Fig. 24.5B).96,102,103 These two activities result from quirement for this apoptotic pathway.97 This novel activity

has been reported only in B cells and has been proposed to Moreover, if Fcg RIIB indeed functions in vivo to main-
act as a means of maintaining peripheral tolerance for B tain peripheral tolerance, then its loss should allow for
cells that have undergone somatic hypermutation. The in the emergence of autoantibodies when otherwise resistant
vivo relevance of this pathway remains to be proven. animals are challenged with potentially crossreactive an-
tigens. This hypothesis has been validated in models of
collagen-induced arthritis and Goodpasture syndrome.
IN VIVO FUNCTIONS
Fcg RIIB-deficient mice, with the nonpermissive H-2b hap-
Fcf Receptors in the Afferent Response lotype, develop arthritis when immunized with bovine
The ability of IgG ICs to influence the afferent response type II collagen.115 The loss of Fcg RIIB thus bypasses the
has been known since the 1950s and can be either enhanc- requirement for the specific H-2q and H-2r alleles previously
ing or suppressive, depending on the precise combination demonstrated to be necessary in this model by allowing
of antibody and antigen and the mode of administration.109 Fcg RIIB-deficient autoreactive B-cell clones to expand and
Investigators have attempted to define the molecular mecha- produce pathogenic autoantibodies. When the permissive
nisms behind these activities with the availability of defined DBA/1 strain (H-2q) is made deficient in Fcg RIIB, autoan-
mouse strains with mutations in activation or inhibitory tibody development is augmented and disease is greatly en-
FcRs and through the use of specific blocking antibodies to hanced. In a similar manner, immunization of H-2b mice
individual receptors. Direct effects on B cells stem from the deficient in Fcg RIIB with bovine type IV collagen results
ability of the inhibitory Fcg RIIB molecule to influence the in crossreactive autoantibodies to murine type IV collagen,
state of B-cell activation and survival. Because antigen is re- with dramatic pathogenic effects.115,116 These mice develop
tained in the form of ICs on FDCs, it can interact with B cells hemorrhagic lung disease and glomerulonephritis with a
by coengaging Fcg RIIB with BCR, modulating the activation “ribbon deposition” pattern of ICs in the glomeruli. These
state of the cell. Mice deficient in this inhibitory receptor de- characteristics are indicative of Goodpasture syndrome, a
velop anti-deoxyribonucleic acid (DNA) and antichromatin human disease not previously modeled in an animal species.
antibodies and die of a fatal, autoimmune glomerulonephri- Expression of the inhibitory Fcg RIIB on B cells thus
tis at 8 months of age. The phenotype is strain dependent and provides a mechanism for the suppressive effects of ICs on
is not seen in BALB/c or 129 strains of mice.110 Combining antibody production. Although FcγRIIB is expressed through-
Fcg RIIB deficiency with defects in other inhibitory recep- out peripheral B-cell development, recent data suggest that
tor pathways, such as PD-1, results in autoantibodies with it represents a late checkpoint controlling the expansion of
different specifities and distinct pathological presentation. autoreactive IgG-positive plasma cells. In contrast, deficiency
Thus, PD–1–deficient Balb/c mice develop cardiomyopathy of the inhibitory receptor did not impact the generation of au-
resulting from anticardiac myosin antibodies. PD-1/FcγRIIB toreactive IgM antibodies.112 Taking the considerably higher
double deficient BALB/c mice develop antiuroepithelial an- pathogenic potential of IgG compared to IgM antibodies into
tibodies and hydronephrosis.111 FcγRIIB thus acts as a modi- account, this late stage of FcγRIIB-mediated regulation seems
fier of autoimmune disease, a conclusion further supported to be sufficient to prevent severe autoreactive processes.
by studies that determined the contribution of the C57BL/6 The enhancing property of ICs on the afferent response
background to the spontaneous lupus-like disease observed is likely to arise from the expression of FcRs on APCs, such
in those mice. The B6 background, by virtue of an incom- as DCs.117–119 DCs express all three classes of IgG FcRs as well
plete light chain editing pathway, provides a source of auto- as Fc eRI. Although in vitro studies have suggested that trig-
reactive B cells in the periphery, which, when combined with gering of activation FcRs can induce DC maturation, the in
defects in the inhibitory Fcg RIIB pathway, leads to the ac- vivo significance of this pathway has not been established.120
cumulation of autoantibodies, pathogenic ICs, and disease.112 The ability of FcRs, particularly Fcg RI, to internalize ICs
Further support for this conclusion is provided by the obser- could provide a mechanism for enhanced presentation and
vations that autoimmune disease–prone strains of mice, such augmented antibody responses, whereas the presence of the
as New Zealand black (NZB), BXSB, SB/Le, MRL, and non- inhibitory Fcg RIIB molecule appears to reduce the enhanc-
obese diabetic, have reduced surface expression of Fcg RIIB ing effect. Mice deficient in Fcg RIIB display enhanced anti-
on activated B cells, attributed to DNA polymorphisms in body responses to soluble antibody–antigen complexes, in
the promoter region of the gene encoding this receptor.113,114 some cases dramatically so, which is likely to result from
This reduced expression of Fcg RIIB is thus suggested to con- enhanced presentation.121,122 In addition, in vitro studies
tribute to the increased susceptibility of these animals to the suggest that internalization through specific FcRs on APCs
development of autoantibodies and autoimmune disease. may influence the epitopes presented and T-cell response
Direct evidence that this is indeed the case comes from stud- generated as a result. At present, a growing body of data sug-
ies where FcγRIIB expression levels have been restored by ret- gests that FcRs are indeed involved in enhancement of the
rovirus-mediated gene transfer in BXSB, NZM, and FcγRIIB afferent response, by influencing antigen presentation and
knockout animals. These mice had dramatically reduced cognate T-cell interactions. Fc γRIIB-deficient DCs pulsed ex
levels of autoreactive antibodies and were protected from vivo with antigen in the form of ICs induce a strong and pro-
the development of fatal autoimmune disease.60 Consistent tective cytotoxic immune response after transfer into naïve
with these results obtained in mouse model systems, human mice.123 In contrast, wild-type DCs induce a much smaller
patients with SLE and CIDP were demonstrated to have a and nonprotective response, indicating that the threshold
reduced FcγRIIB expression level on B cells.58,59 set by co-crosslinking of the inhibitory and activating FcRs

prevents complete DC maturation. Moreover, blocking of IC A

binding to Fc γRIIB on human DCs with an FcγRIIB-specific IgG1 IgG2a IgG2b IgG3
antibody resulted in spontaneous maturation of the cells by 200
ICs present in low amounts in human serum.124,125 Taken to-
platelet count (×104/ul)

gether, these data indicate that the inhibitory receptor is an 150
important regulator of DC activation. As the DC maturation
state will determine whether an activating or a tolerogenic 100
signal will be delivered to T cells, FcRs might be important
factors for the maintenance of peripheral tolerance in the 50
cellular immune system. Transiently blocking Fc γRIIB ac-
tivity with monoclonal antibodies in vivo might thus be an 0
interesting strategy to optimize immunotherapeutic and 0 10 20 30 40
vaccination approaches. Further defi ning the precise role of hours after 6A6 isotype injection
each FcR expressed on APCs will require conditional knock-
outs of these molecules on specific DC populations to re- B
solve the contribution of these systems to the generation of
an appropriate antibody response. TA99-IgG2a
C57Bl/6
Fcf Receptors in the Efferent Response
The first canonical FcR knockout to be described was for TA99-IgG1
the common activation subunit, the g chain, which resulted
in the loss of surface assembly and signaling of Fcg RI,
FcγRIIB-/-
Fcg RIII and Fc γRIV as well as Fc e RI (10). Mice deficient in
TA99-IgG1
the common g chain were systematically studied in diverse
models of inflammation and found to be unable to medi-
ate IgG-triggered inflammatory responses for cytotoxic or
C
IC reaction; attributed to low-affi nity activation receptors, A/I ratio: 7 20
the high-affi nity Fcg RI played a minimal role in the in vivo
inflammatory response triggered by IgG.22,126–128 The results
were further confirmed by comparisons of mice deficient or
blocked for either Fcg RI, Fcg RIII, or Fc γRIV.19,20,32,129 The
TA99-IgG2b
TA99-IgG2b
loss of Fc e RI ablated IgE-mediated anaphylaxis; this was (+fucose)
(-fucose)
Mock
demonstrated independently by gene disruption in the a

subunit of that receptor.11 Subsequent studies on mice de-
ficient in the inhibitory Fcg RIIB molecule established the
opposing action of this receptor, in which mice deficient in
FIG. 24.6. Antibody Activity is Determined by Activating and
that receptor displayed enhanced B-cell responses, autoim-
Inhibitory Fc Receptors (FcRs). A: Mice were injected with the same
munity, and augmented IgG-mediated inflammation in a amount of different 6A6 antibody isotypes, which recognize an in-
subclass and effector cell–dependent manner.18,20,32,55,128 The tegrin on mouse platelets and lead to FcR-dependent clearance of
general fi nding, which is discussed in detail subsequently, platelets from the blood as observed by the drop in platelet count
illustrates that IgGs initiate their effector responses in vivo after antibody injection. B: Wild-type or FcγRIIB knockout mice were
through coengagement of activating and inhibitory FcRs. injected intravenously with B16 melanoma cells on day 0 and with
The physiological response is thus the net of the opposing IgG1 and IgG2a isotype switch variants of the monoclonal antibody
TA99 on alternate days. C: Mice were injected with antitumor an-
activation and inhibitory signaling pathways that each re- tibodies containing or lacking branching fucose residues, thereby
ceptor triggers and is determined by the level of expression enhancing binding to activating FcRs and changing the A/I ratio
of each receptor and the selective avidity of the IgG ligand (see text for details). Lungs were harvested on day 14. Adapted from
(see Table 24.1). This also explains the longstanding obser- Nimmerjahn and Ravetch,20 with permission.
vation that different IgG isotypes have a differential activ-
ity in vivo (Fig. 24.6A). The absence of a murine homolog
for Fc aRI has precluded similar studies for that recep-
tor. Studies on mice bearing a human transgene of Fc aRI with the observations obtained in Fce RI-deficient mice
suggest that this receptor is involved in IgA nephropathy and confi rming the role of the high-affinity IgE receptor
(Berger disease).130 in mediating IgE-induced anaphylactic responses.10,11,40,131
Fcg RIIB-deficient mice, challenged in this model, displayed
Type I: Immediate Hypersensitivity an unexpected enhancement of IgE-mediated anaphylaxis,
Both cutaneous and systemic models of passive anaphy- which suggests a physiological interaction between this in-
laxis, induced by IgE, were studied in FcRg chain–deficient hibitory receptor and Fc e RI.71 The molecular basis for this
mice and were found to be absent, a fi nding fully consistent modulation of Fc e RI signaling by Fcg RIIB has not been

determined, although previous studies have indicated that antibodies to bind as monomers to Fcg RI. These and other
IgE can bind with low affi nity to Fcg RII/Fcg III, which sug- studies suggest that the role of the high-affinity Fcg RI in
gests that there exists a mechanism for coengagement of IgG-mediated inflammation is likely to be restricted to aug-
these receptors. Deletion of the mast cell inhibitory recep- menting the effector response (determined by Fcg RIII and
tor glycoprotein 49B1 also results in enhanced IgE-induced IV) in situations that involve high concentrations of murine
anaphylaxis.72 In addition to Fc e RI, mast cells also express IgG2a antibodies which are found at localized inflamma-
the IgG Fc γ RIIB and Fc γ RIII, but not Fc γ RIV. Passive sys- tory sites where Fcg RI expression is induced on recruited
temic anaphylaxis induced by IgG was attenuated in FcRg macrophages.
chain–deficient and Fcg RIII-deficient mice, which in- Experimental models of immune thrombocytopenic
dicates the capacity of IgG and Fcg RIII to mediate mast purpura (ITP) in which murine IgG1 antiplatelet antibod-
cell activation in vivo. Fcg RIIB-deficient mice displayed ies trigger thrombocytopenia and yielded results similar to
enhanced IgG-induced anaphylaxis. Active anaphylaxis, those of the anti-RBC studies cited previously. The specific
induced by immunization with antigen in alum, was en- Fc γR involved depended on the subclass of antibody used.
hanced in Fc e RI-, Fcg RIIB-, and glycoprotein 49B1–defi- IgG1 antibodies mediated their activity exclusively through
cient mice and attenuated in FcRg- and Fcg RIII-deficient Fcg RIII, while IgG2a and 2b were FcγRIV dependent. In
mice. All these animals displayed antigen-specific antibod- contrast, Fcg RI- or C3-deficient mice were fully suscep-
ies for IgE and IgGs, which indicates that the active ana- tible to antibody-induced thrombocytopenia.20,31 Fcg RIIB-
phylaxis seen was attributed primarily to IgG antibodies. deficient mice showed an isotype-specific enhancement of
The reason for the enhancement of anaphylactic responses antibody-mediated platelet depletion, with the strongest
in Fc eRI-deficient animals resulted from the increased ex- impact on IgG1 and much smaller increases for IgG2a and
pression of Fcg RIII on mast cells in these mice, normally IgG2b isotypes. This is consistent with the affi nities of these
limited by competition of a chains for the available pool isotypes for their specific activating and the inhibitory re-
of the common g chain.82 In the absence of Fc e RI a chain, ceptor, which will determine antibody activity in vivo (see
FcRg chain is available to associate with Fcg RIII a chain Table 24.1). In a passive protection model of Cryptococcus
and assemble on the cell surface as a functional signaling neoformans –induced disease, passive immunization with
receptor. A similar type of regulation was also described for mouse IgG1, IgG2a, and IgG2b antibodies resulted in pro-
expression of Fc γ RIII and Fc γ RIV on mouse neutrophils, tection in wild-type animals but not in FcRg chain–deficient
where deletion of one receptor resulted in enhanced ex- animals.134
pression of the other γ -chain– dependent molecule.83 These IgG antibodies raised to murine glomerular basement
studies indicated the importance of the g chain in regulat- membrane preparations induce acute glomerulonephri-
ing the level of surface expression of Fc e RI and Fcg RIII. tis in a model of Goodpasture disease in wild-type but not
Because g chain is also associated with other members of FcRg- or Fcg RIV-deficient animals.77,135,136 Fcg RIIB-deficient
the activation/inhibition paired receptors expressed on animals displayed enhanced disease in this model, which
mast cells, such as PIR-A/PIR-B, the intracellular compe- indicates that the effector cells involved were constitutively
tition between these diverse a subunits and the common expressing significant levels of Fcg RIIB. Similar results were
g chain determines the level of surface expression of indi- obtained when DBA/1 animals were immunized with bo-
vidual receptors and thus their ability to respond to specific vine type II collagen to induce arthritis. Deficiency of FcRg
biological stimuli. The absolute level of surface expression chain protected these mice from the pathogenic effects of
of FcRs on mast cells is clearly of therapeutic significance the anticollagen antibodies that were generated.137 As men-
in both IgE- and IgG-mediated inflammatory responses; tioned previously, deficiency of Fcg RIIB in the DBA/1 colla-
modulation of g chain expression could thus represent a gen-induced arthritis model resulted in enhanced disease,
new therapeutic avenue for intervention in diseases such as through increased autoantibody production and elevated
anaphylaxis and asthma. effector responses.
A dramatic example of the importance of these pathways
Type II Inflammation: Cytotoxic Immunoglobulin G in determining the in vivo activity of cytotoxic antibodies
Cytotoxic IgGs are found in a variety of autoimmune dis- was obtained in models of antitumor antibody response. In
orders and have been developed for therapeutic indications a syngenic murine model of metastatic melanoma, a murine
in the treatment of infectious and neoplastic diseases. The IgG2a antimelanocyte antibody was able to reduce tumor
mechanisms by which these antibodies trigger cytotoxic- metastasis in wild-type animals but was ineffective in FcRg-
ity in vivo have been investigated in FcR knockout mice. or Fc γRIV-deficient mice.20,83,138 In the absence of Fcg RIIB,
Anti-RBC antibodies trigger erythrophagocytosis of IgG- the activity of an IgG1 antibody, matched in its antigen bind-
opsonized RBCs in an FcR-dependent manner; g chain– ing domain, was enhanced, which indicates that the in vivo
deficient mice were protected from the pathogenic effect of cytotoxic activity of the antibody was the net of activation
these antibodies, whereas complement C3–deficient mice and inhibitory receptor engagement20,55 (see Fig. 24.6B).
were indistinguishable from wild-type animals in their abil- These studies, together with similar studies performed with
ity to clear the targeted RBCs.132,133 Fcg RIII plays the exclu- antiplatelet antibodies or defucosylated antibodies, dem-
sive role in this process for the mouse IgG1 isotype. Murine onstrated that the in vivo activity of a cytotoxic antibody
IgG2a anti-RBC antibodies utilize primarily the Fcg RIV re- could be predicted by a simple equilibrium binding model
ceptor pathway despite the singular ability of murine IgG2a in which the ratio of the monomeric affinity constants for

the activation and inhibitory Fc receptors (A/I ratio) are the requirement for Fcg R activation.22 Fcg RIIB modulated the
dominant parameters, as demonstrated for T-cell receptor– magnitude of the response, with enhanced Arthus reactions
major histocompability complex interactions.139 An example observed in Fcg RIIB-deficient strains.18 The effector cell in
for the enhanced cytotoxic activity of an antitumor anti- the cutaneous reaction was determined to be the mast cell,
body due to increased affinity for activating FcRs is shown as demonstrated by the use of mast cell–deficient strains and
in Figure 24.6C. The generation of an afucosylated antitu- by mast cell reconstitution studies.131 The generality of this
mor antibody resulted in an increase of the A/I ratio of 7 to result was demonstrated in similar reactions performed in
20 for this IgG subclass, resulting in enhanced antitumor the lung, illustrating the FcR dependence and relative com-
activity in vivo. Xenograft models of human tumors trans- plement independence of this response.116 Thus, all studies
plated into nude mice demonstrated, for a variety of tumors have demonstrated an absolute dependence on FcR expres-
and cytotoxic antibodies, the requirement for FcR effector sion in the Arthus reaction. One model for IC-induced ar-
activity. For example, human breast carcinoma or lym- thritis, the KRN/nonobese diabetic model, in which IgG1
phoma lines transplanted into nude mice and treated with anti-GPI antibodies are responsible for IC deposition in the
either the humanized IgG1 or chimerized IgG1 antibodies synovium,142 has been shown to depend on both Fc γRIII
(trastuzumab and rituximab), respectively, revealed that the and C3 but not on components of the classical complement
ability of these antibodies to modulate tumor growth was pathway, such as C1q and C4; transfer of serum to animals
abrogated in FcRγ chain–deficient mice. A point mutation deleted for Fc γRIII or C3 prevented the development of dis-
that eliminated FcR binding of the anti-Her2/neu murine ease.143,144 Deficiency in the late components of complement,
IgG1 antibody 4D5 abolished the in vivo cytotoxic activity such as C5a or its receptor, have also been reported to result
of the antibody against a human xenograft but did not affect in a partial reduction in the magnitude of the response in
the in vitro growth inhibitory activity; this again illustrates IC-induced lung inflammation145 and to result in a complete
the difference between in vivo and in vitro mechanisms. block in the KRN/nonobese diabetic arthritis model. The
Similar results were obtained for T-cell lymphoma xeno- likely mechanism by which C5a exerts its effects is through
graft models and anti-CD2 antibodies, among others.140,141 upregulation of activating Fc γRs, resulting in an amplifica-
The general conclusions that can be drawn from these stud- tion loop.75 C5a is generated in this system as a result of FcR
ies support a dominant role for the low-affinity activating activation of effector cells and is independent of the classical,
Fcg Rs in mediating cytotoxicity by IgG antibodies. Fcg RIIB alternative, and mannan-binding protein pathways. Binding
restricts the effector response for those antibodies with low of C5a to the C5aR results in upregulation of activation re-
A/I ratios and in situations in which the effector cell ex- ceptors on these effector macrophages, thus augmenting
presses this inhibitory molecule. the inflammatory response triggered by FcRs. These studies
The relevance of these murine in vivo studies to the treat- have led to a revision of the hypotheses about the mecha-
ment of human populations with antitumor cytotoxic anti- nism of IC-mediated inflammation, typified by the Arthus
bodies has been demonstrated in two studies that investigated reaction, in which there is an absolute requirement for acti-
the differential responses of patients treated with anti-CD20 vating Fcg Rs in initiating mast cell activation by ICs. Fcg R
(rituximab) for lymphoma.48,52 Both studies demonstrated a activation is, in turn, modulated by the inhibitory receptor
highly significant correlation between patient response, as Fcg RIIB. The A/I value of a specific IgG antibody and the
measured by the time to relapse, and alleles of Fc γRIIIA. In densities of these opposing signaling receptors determines
patients with an allele of this low-affi nity activation receptor the concentration threshold for IC activation and the mag-
(158V) that confers higher binding affinity for human IgG1 nitude of the effector response that can be obtained. The
Fc, improved outcome was observed, as compared to those classical pathway of complement activation is not required;
with a lower binding allele of this receptor (158F). In addi- however, C5a activation, through the Fc γR pathway, may
tion, lymphoma patients with the high affinity allele showed enhance the response under some circumstances through
a significantly better clinical response after receiving anti- an amplifying loop. The release of inflammatory mediators
idiotype vaccination.51 such as vasoactive amines, chemokines, and cytokines leads
to the hallmarks of this reaction: edema, hemorrhage, and
Type III Responses: Immune Complex–Mediated neutrophil infi ltration at the site of IC deposition.
Inflammation The significance of the FcR pathway in initiating IC in-
The classic example of this reaction, the Arthus reaction, has flammation in autoimmune disease was further established
been studied in a variety of FcR- and complement-deficient by investigating a spontaneous murine model of lupus, the
animals. The initial studies were performed by using the cu- B/W F1 mouse. The Arthus reaction results predicted the
taneous reverse passive Arthus reaction, in which antibody absolute requirement of activation Fcg R in initiating in-
was injected intradermally and antigen was given intrave- flammation and tissue damage in IC diseases such as lupus.
nously. An inflammatory response, characterized by edema, The FcRg chain deletion was backcrossed onto the NZB and
hemorrhage, and neutrophil infi ltration, developed within New Zealand white strains for eight generations, and the in-
2 hours. This reaction was elicited in a variety of comple- tercrossed progeny were segregated into B/W FcRg–/– and
ment- and FcR-deficient animals. The results from several FcRg +/–. Anti-DNA antibodies and circulating ICs developed
independent studies confirmed the initial observations: that in all animals; IC and complement C3 deposition was simi-
IgG ICs triggered cutaneous inflammatory reactions even larly observed in all animals. However, mice deficient in the
in the absence of complement but displayed an absolute common g chain showed no evidence of glomerulonephritis

and had normal life expectancy, despite comparable levels of in vivo activity was demonstrated for agonistic anti-Fas an-
circulating ICs and glomerular deposition of these complexes tibodies.152 A variety of CD40-specific agonistic antibodies
along with complement C3. Mice heterozygous for the g chain have been used to deliver costimulatory signals to augment
mutation were indistinguishable from B/W F1 animals with and sustain T-cell responses to result in tumor clearance.151
wild-type g chains in developing glomerulonephritis and dis- Recent studies showed that the inhibitory Fc γRIIB was es-
playing reduced viability.127 This spontaneous model supports sential for this agonistic activity and could not be replaced
the conclusions stated previously about the absolute require- by activation Fc γRs.153,154 Although the exact mechanism of
ment for FcγRIII in the activation of inflammatory disease these FcR-dependent pathways remain to be established,
by ICs: in the absence of this receptor, deposited ICs and C3 these findings open the path toward generating improved
are not sufficient to trigger effector cell activation, which in- agonistic anti-tumor necrosis factor receptor antibodies by
dicates that it is possible to uncouple pathogenic ICs from Fc engineering. Consistent with this notion, a recent study
inflammatory disease by removing activating FcR engage- showed that generating anti-CD40 antibodies with en-
ment. Similar conclusions were reached in a murine model hanced Fc γRIIB binding considerably augmented their in
of Goodpasture disease where IC deposition, composed of vivo adjuvant activity.154
mouse IgG2b antibodies, resulted in a fulminant glomeru-
lonephritis.77 Blocking the relevant activation FcR, FcγRIV,
with a monoclonal antibody, protected the animals from fatal
DISEASE ASSOCIATIONS
disease. These results further indicate that intervention in the Autoimmunity and Tolerance
effector stage of IC diseases, such as lupus and rheumatoid ar- In view of their functional capacity to link autoantibod-
thritis, would be accomplished by blocking activation Fcg Rs ies to effector cells, FcRs have naturally been considered to
to prevent initiation of effector cell responses. have a pathogenic role in the development of autoimmune
diseases. Several studies have attempted to correlate specific
Immunoglobulin G–Dependent Neutralization and polymorphisms in Fc γRIIA, Fc γRIIIA, or Fc γRIIIB with
Induction of Agonistic Signaling incidence or severity of lupus or rheumatoid arthritis.155
In contrast to these FcR-dependent activities, IgG mol- In view of the heterogeneity of these diseases, it is perhaps
ecules, by virtue of their exquisite specifity for discrete an- not surprising that inconsistent results have been obtained.
tigens, are expected to mediate Fc γR-independent effects, Alleles that increase the ability of Fc γRIIA to bind IgG2 or
such as neutralization of bacterial toxins or viruses. Here, Fc γRIIIA to bind IgG1 might be expected to correlate with
IgG molecules would prevent the toxin or microorganism disease severity in some populations. Indeed, these types of
from binding to its cellular receptor, thereby protecting the associations have been reported in some studies but not in
host from the pathogenic effects of the infecting microbe. others.156 These variable results have often been explained as
This picture changed, however, as it was demonstrated that an indication that other genes may be in linkage disequilib-
both IgG-mediated toxin and virus neutralization can be rium with the FcR alleles under investigation. This is a plau-
dependent on cellular FcRs in vivo. Thus, neutralization of sible explanation when viewed in light of the autoimmunity
the anthrax toxin (called protective antigen component) by susceptibility genes mapping in or near the region of the FcR
anti–protective antigen antibodies required FcR expression genes, chromosome 1q21–24.157 This region of chromosome
in vivo and anti–protective antigen antibodies that differed 1 has been implicated in a variety of human and murine
only by their Fc subclasses showed a hierarchy of protection linkage studies. For example, the Sle1 alleles derived from
with IgG2a and IgG2b antibodies affording a better protec- NZB flank the Fc γRIIB gene and form a linkage group with
tion than IgG1 antibodies, consistent with previous observa- the ability to break tolerance to nuclear antigens, resulting
tion for cytotoxic IgG.146,147 Similar results were obtained for in production of antichromatin antibodies. Epistatic inter-
the mechanism of activity of neutralizing antibodies pro- actions between Fc γRIIB and lupus susceptibility genes have
tecting from influenza or human immunodeficiency virus been demonstrated in the murine lupus model of B6.RIIB.
infection.148,149 These results suggest that current in vitro Crossing the yaa gene to this strain accelerates the develop-
assays using only the pathogen-specific antibody, a target ment of disease; 50% survival is decreased from 8 months
cell line might miss an important component required for to 4 months, with 100% fatality by 8 months. This increase
antibody activity in vivo, and that novel assays including the in severity correlates with a change in the specificity of the
FcR-dependent component might provide a better predict- autoantibodies, from diffuse antinuclear antibodies to an-
ability of IgG activity.150 tibodies that stain with a punctate, nucleolar pattern on
Beyond the central function of IgG in host defense, ago- antinuclear antibody staining. Recently, it has been shown
nistic antibodies triggering signaling pathways in target that the yaa susceptibility locus contains a duplication of the
cells are being evaluated for their therapeutic potential. toll-like receptor 7 gene in the pseudoautosomal region of
For example, agonistic antibodies for tumor necrosis fac- the y-chromosome (158). Moreover, toll-like receptor 7 and
tor receptor family members such as anti-CD40 or anti- toll-like receptor 9 with their ability to recognize potential
DR5 (drozitumab) are under evaluation for their ability to self-antigens such as ribonucleic acid or DNA, respectively,
enhance tumor cell clearance by both direct and indirect were shown to be important components in the generation
mechanisms. Interestingly, the apoptosis inducing activity of pathogenic autoreactive antibodies.159,160 Compared to
of anti-DR5 agonistic antibodies was found to be depen- other Yaa-independent SLE susceptibility loci, FcγRIIB may
dent on FcRs in vivo.151 A similar dependence on FcRs for be a dominant factor for the development of autoantibodies

and the systemic activation of the immune system as dem- antibodies containing an engineered Fc-fragment with
onstrated by a study using NZB/BXSB F1 animals congenic enhanced binding to Fc γRIIB has been successfully used to
for the wild-type fcgr2b gene. In these mice, a dramatic re- shut down autoantibody production by human B cells in a
duction in the production of autoantibodies and immune xenograft mouse model system.168 The ability to exploit the
cell activation was noted compared to mice with impaired inhibitory pathway to reduce the activity of activation Fcg Rs
Fc γRIIB expression.161 These data are consistent with stud- has been demonstrated to account for some of the anti-
ies that overexpressed Fc γRIIB in autoimmune-prone inflammatory activity associated with high-dose intrave-
mouse strains resulting in suppression of autoantibody nous gamma globulin (IVIG), which consists of the pooled
production.60,61 IgG fraction of serum from thousands of donors.76 The use
Two types of polymorphisms in the Fc γRIIB gene have of IVIG for the treatment of ITP and other autoimmune dis-
been associated with SLE. Promoter polymorphisms have eases is well established, although the mechanism of action
been described resulting in reduced expression of Fc γRIIB has been elusive. In murine models of ITP, arthritis, and ne-
on activated B cells in both mouse and human SLE and phritis, it has been demonstrated that protection by IVIG is
CIDP populations.59,113,162 In addition, a polymorphism in dependent on the presence of FcγRIIB; deletion of Fc γRIIB
the transmembrane domain of Fc γRIIB has been identified or blocking Fc γRIIB by a monoclonal antibody eliminates
that is associated with susceptibility to SLE in Japanese pop- the ability of IVIG to protect the animal against an inflam-
ulations. This polymorphism results in an Fc γRIIB protein matory response.76,78,80 IVIG was demonstrated to lead to
with reduced ability to enter lipid rafts and thus behaves as a the in vivo induction of Fc γRIIB on splenic effector macro-
hypomorphic allele for inhibitory function.163 phages, which would raise the threshold required for platelet
Together, these studies point to Fc γRIIB as a susceptibility clearance by activating FcγRs on these cells.20,31,78,83 A similar
factor in the development of autoimmunity with the ability upregulation of the human inhibitory Fc γRIIB was demon-
to interact with other susceptibility factors to modify both strated in patients with CIDP receiving IVIG treatment as a
the afferent and efferent limbs of the autoimmune response. first-line therapy.59 Recently, it has become clear that the si-
alic acid–rich fraction of IgG antibodies in the IVIG prepa-
ration is responsible for the anti-inflammatory activity and
Inflammation Fc γRIIB upregulation, enabling for the first time the genera-
Antibody-mediated inflammatory diseases have been tion of a recombinant IVIG replacement consisting of a si-
clearly demonstrated to involve the coupling of patho- alic acid–rich IgG Fc fragment.79,80 Human and mouse sialic
genic autoantibodies or ICs to cellular FcRs. Therapeutics acid–rich IgG has a reduced affinity for classical FcRs and
targeted to disrupt these interactions are in development, acquires the capacity to bind to SIGNR1 strongly expressed
beginning with a monoclonal antibody to human IgE that on mouse splenic marginal zone macrophages.68,80 By using
functions to reduce IgE binding to its high-affi nity receptor SIGNR1-deficient mice expressing human DC-SIGN, it was
and thereby prevent allergic and anaphylactic reactions.164 demonstrated that human DC-SIGN expressed on macro-
Because IgE is required for the survival of mast cells as well phages or DCs can substitute for mouse SIGNR1 affording
as in the regulation of Fc e RI expression, reduction in IgE a candidate molecule essential for the anti-inflammatory
has synergistic effects on the ligand, receptor, and effector activity of IVIG in humans.67 Both molecules belong to the
cell. The success of this approach will undoubtedly lead to C-type lectin family and have been implicated in recogni-
other approaches that target the receptor or its signaling tion of pathogenic microorganisms including human im-
pathway. Blocking Fcg RIIIA or IIA is expected to mimic munodeficiency virus and Mycobacterium tuberculosis.169
the phenotype of FcRgamma-deficient animals in models Therefore, SIGNR1 and human DC-SIGN can be considered
of IgG-induced disease. Early attempts to use this approach as noncanonical FcRs recognizing specific IgG glycovariants
in ITP were promising but limited by the crossreactivity and, together with the family of toll-like receptors, provide
to receptors on neutrophils, which led to neutropenia and a prime example for the dual usage of one molecule in the
the development of immune response to the murine anti- recognition of self- and non–self-ligands. Moreover, infu-
body.165 Development of second-generation anti-Fc γRIIIA sion of sialic acid–rich IgG resulted in an IL33-dependent
antibodies with greater specificity and reduced toxicity now expansion of basophils and IL4 production. IL4 is one of
appears to be a viable approach for the treatment of autoim- the interleukins known to upregulate Fc γRIIB on innate
mune diseases. Moreover, using small molecules to inhibit immune effector cells, consistent with the upregulation
signaling molecules such as Syk seems to have promising ef- of Fc γRIIB noted in mouse model systems and humans.
fects in preventing autoantibody dependent platelet deple- Together with previous data on the dependence of Fc γRIIB
tion in mice and humans.166 upregulation on colony-stimulating factor-1–dependent
An alternative approach to limiting the activation of FcRs macrophages, so called sensor macrophages, this suggests
is to utilize the endogenous inhibitory pathway to abrogate the following model for the anti-inflammatory activity of
IgE or IgG activation of their cognate receptors through co- IVIG (Fig. 24.7). In this model, sialic acid–rich IVIG binds to
ligation to Fcg RIIB. This mechanism has been proposed to mouse SIGNR1 or human DC-SIGN on sensor macrophages
explain the ability to induce desensitization for the treat- or DCs, which results in the production of IL33 that in turn
ment of allergic diseases.33,167 Inducing production of IgG leads to IL4 production by basophils, inducing the upregu-
antibodies to an allergen may facilitate cross-linking of lation of Fc γRIIB on effector macrophages, thereby raising
Fcg RIIB to Fc e RI. A similar approach involving anti-CD19 the threshold for activation.67 There is evidence that this

Human:
DC-SIGN+ Mφ or
dendritic cell
(lymph node, spleen)
Joint Space
IL-33 IL-4
NO INFLAMMATION
Inhibitory FcR
up-regulation
-
Activating FcR
FIG. 24.7. Model for the Anti-inflammatory down-regulation
Activity of Intravenous Gamma Globulin
(IVIG) in Mice and Humans. Sialic acid–rich IL-33 IL-4
antibodies in the IVIG preparation lose affinity Effector Mφ
for classical FcγRs and acquire the capacity IL4/IL13R+
to bind to molecules such as mouse SIGNR1 Innate leukocyte
or human DC-SIGN on the surface of colony- (e.g. basophil)
stimulating factor-1–dependent sensor mac-
rophages or dendritic cells. This results in Figure legend
the production of IL33 and the expansion of
Mouse:
innate leukocytes such as basophils followed IVIG immune
CSF-1 dependent SignR1
by IL4 release resulting in the upregulation of IVIG-SA-rich complex
SignR1+ Mφ
the inhibitory FcγRIIB and downmodulation of (spleen)
activating FcγRs on effector macrophages in- autoantigen
activating FcR inhibitory FcR
hibiting the release of inflammatory mediators autoantibody
and tissue destruction.
anti-inflammatory pathway triggered by IgG glycovariants mouse IgA and the human Fc aRI receptor to release soluble
rich in terminal sialic acid residues is of general importance receptor–IgA complexes, which leads to deposition in the
for maintaining immune homeostasis. Thus, patients with mesangium and the sequelae of IgA neuropathy.
rheumatoid arthritis and autoimmune-prone mouse strains
have reduced levels of serum IgG–rich containing terminal
sialic acid and galactose residues during active phases of the CONCLUSION
disease.170,171 A similar reduction of this anti-inflammatory Receptors for the Fc of immunoglobulins provide an es-
IgG variant is observed during aging, consistent with an sential link between the humoral and adaptive response,
increased probability to develop autoimmune disease.172 translating the specificity of antibody diversity into cellular
This downmodulation of sialic acid–rich IgG was also ob- responses. The canonical FcRs mediate their biological re-
served during normal immunization conditions in mice, sponses through the coupling of Fc recognition to ITAM-/
suggesting the general importance of this phenomenon ITIM-based signaling motifs. A diverse array of biological
during proinflammatory conditions.80 The opposite type responses depends on the FcR system, influencing both the
of regulation is seen in women with rheumatoid arthritis afferent and efferent limbs of the immune response. Detailed
during pregnancy.173 Parallelling the decreased frequency of biochemical, structural, and molecular biological data have
arthritic flares an increased level of IgG glycoforms rich in provided a detailed understanding of how these receptors
terminal galactose and sialic acid residues was observed. In are regulated, are assembled, bind their ligand, and trans-
addition, inducing expression of Fc γRIIB might be a clini- duce specific cellular signals. FcRs play a significant role in
cally feasible approach and could be effective at modulat- vivo in maintaining peripheral tolerance by limiting the ac-
ing pathogenic autoantibodies from activation effector cell cumulation of autoreactive B cells that escape central toler-
responses through activating FcRs. ance checkpoints, such as light chain editing or potentially
Studies on the Fc aRI receptor have demonstrated a role arise during somatic hypermutation in germinal centers, in
for this molecule in the pathogenesis of IgA nephropathy, in modulating T-cell responses by regulating both antigen pre-
which circulating macromolecular complexes are deposited sentation and maturation by DCs, and in mediating the cou-
in the mesangium, resulting in hematuria and eventually pling of antigen recognition to effector-cell activation. They
leading to renal failure. Soluble Fc aRI is found in the circu- are the primary pathways by which pathogenic IgG and IgE
lating IgA complexes, which suggests a role for the receptor antibodies trigger inflammatory responses in vivo. Allergic
in the formation of these pathogenic complexes. A trans- reactions, cytotoxic IgG responses, and IC-mediated inflam-
genic mouse expressing Fc aRI spontaneously develops IgA mation are all critically dependent on FcR cross-linking and
nephropathy resulting from the interaction of polymeric have resulted in a fundamental revision of the mechanisms

underlying such classic immunological responses as the importance for maintaining immune homeostasis, suggest-
Arthus reaction. Blocking of these receptors uncouples the ing that IgG glycoforms rich in terminal sialic acid residues
pathogenic potential of autoantibodies and represents an might be considered as a molecular switch either keeping
important new therapeutic target for the development of the immune system in a resting state or, during infection
anti-inflammatory therapeutic agents. Central to the cor- or autoimmune responses, loosening the brakes thereby
rect functioning of these responses is the balance that is allowing full blown inflammation. In this pathway, nonca-
maintained through the pairing of activation and inhibi- nonical FcRs including SignR1 and human DC-SIGN rec-
tory receptors that coengage the IgG ligand; perturbations ognizing specific IgG glycoforms are of central importance.
in either component can result in pathological responses. Conversely, engineering of therapeutic antibodies targeted
The study of FcRs defined the ubiquitous inhibitory motif, to eliminate infectious or neoplastic disease will probably
the ITIM, and has provided a paradigm for how these path- benefit from optimization of their Fc domains for interac-
ways modulate ITAM-based activation responses. Studies in tion with specific canonical FcRs.
mice deficient in individual FcRs have provided the neces-
sary insights for defining comparable activities in human
autoimmune diseases and suggest ways in which manipula- ACKNOWLEDGMENTS
tion of the IgG–FcR interaction may lead to new classes of The Laboratory of Molecular Genetics and Immunology at
therapeutics for the treatment of these diseases. Modulation the Rockefeller University is supported by grants from the
of the inhibitory response, a novel activity associated with National Institutes of Health, the Dana Foundation, the
IVIG to account for some of its anti-inflammatory activity Juvenile Diabetes Research Foundation, the Cancer Research
in vivo, represents a novel approach to the regulation of Ig- Institute, and by Theresa and Eugene Lang. The Laboratory
mediated inflammation and suggests that therapeutic agents of Genetics at the University of Erlangen-Nuernberg is
based on those pathways are likely to be effective. It seems supported by the German Research Foundation and the
clear that this anti-inflammatory pathway is of general Bavarian Genome Research Network.

Type I Cytokines and Interferons,

CHAPTER
25 and Their Receptors
Warren J. Leonard
OVERVIEW AND ISSUES OF NOMENCLATURE Lymphopoietin), even though both can be produced by stro-
Cytokines are proteins that are secreted by cells and trans- mal and epithelial cells and share a receptor component as
duce signals via specific cell surface receptors on either the well as select actions on lymphocytes. The more descriptive
cytokine-producing cell (autocrine actions) or on other name of TSLP correctly describes its production by thymic
target cells (paracrine actions). From this operational type stroma but obscures its production by skin epithelial cells,
of description, it is clear that the distinction between cyto- and that major target cells include dendritic cells (DCs) and
kines, growth factors, and hormones may be imprecise. In cluster of differentiation (CD)4 + T cells.
general, cytokines and growth factors are similar, except that Among the many classes of cytokines, the “type I” cy-
the term cytokine most often refers to molecules involved tokines are distinctive in their sharing a similar four α-
in host defense that have actions on leukocytes, whereas the helical bundle structure, as detailed in the section on Type
term growth factor more often refers to molecules acting on I Cytokines—Structural Considerations, and correspond-
other somatic cell types. Cytokines most often act locally. ingly, their receptors share characteristic features that have
For example, in the interaction between a T cell and an led to their description as the cytokine receptor super-
antigen-presenting cell, cytokines are produced and usually family, hematopoietin receptors, or type I cytokine recep-
exert potent actions locally, with rather limited biologic half- tors.5–8 Although many interleukins are type I cytokines,
lives in the circulation. In contrast, hormones are released some are not. For example, two of the major “proinflam-
and then disseminated via the bloodstream throughout the matory cytokines,” IL-1 and IL-6, are interleukins, but IL-6
body, with actions on distal target organs. Nevertheless, this is a type I cytokine, whereas IL-1 is not. One interleukin,
distinction between cytokines and hormones is not absolute, IL-8, is a CXC family chemokine, an entirely different type
with certain cytokines acting at longer distances as well. of molecule involved in chemotaxis. Moreover, IL-10, IL-
In the immune system, terms such as monokines and 19, IL-20, IL-22, IL-24, IL-26, IL-28A, IL-28B, and IL-29 are
lymphokines were originally devised to identify the cellular more similar to interferons (IFNs) and are denoted as type
source for cytokines.1 Specifically, monokines included mol- II cytokines.
ecules such as interleukin (IL)-1, which was first recognized In summary, the term interleuk in indicates a relationship
to be produced by monocytes, and lymphokines included to leukocytes, whereas the identification of a cytokine as a
molecules such as IL-2, which was first described as a T-cell type I or type II cytokine indicates general properties of its
growth factor produced by T lymphocytes. The monokine/ three-dimensional structure. Knowing that a molecule is a
lymphokine nomenclature can be problematic; however, type I cytokine is instructive as it indicates a likely general
when a cytokine is synthesized by more than one type of structure for the cytokine receptor as well as the mechanism
cell. This resulted in the adoption of the term cytokine, as of signal transduction. In contrast, the identification of a
proposed by Stanley Cohen in 1974.2,3 The term cytokine re- molecule as an interleukin provides little information other
fers to a protein made by a cell (“cyto”) that acts (“kine”) on than that it often, but not always, is a type I or type II cyto-
target cells. Cytokines can have very broad ranges of actions, kine of immunologic interest.
including on cell development, differentiation, growth, cy- In addition to molecules of primarily immunologic in-
tolytic activity of effector cells, survival, apoptosis, and terest, other important proteins, such as growth hormone,
chemotaxis. prolactin, erythropoietin (Epo), thrombopoietin, and leptin,
Many cytokines are referred to as interleukins to indicate are type I cytokines and their receptors are type I cytokine
molecules that are produced by one leukocyte and act on receptors. Despite having their major actions outside the im-
another.4 However, this term is also problematic as some in- mune system, these cytokines nevertheless share important
terleukins (eg, IL-1 and IL-6) are additionally produced by signal transduction pathways with type I cytokines of im-
cells other than leukocytes and/or exert actions on cell types munologic interest. By focusing on type I cytokines and IFNs
beyond the immune system. For example, IL-7 is produced and their receptors, this chapter necessarily focuses on cyto-
by stromal and epithelial cells rather than by typical leuko- kines that are evolutionarily related and share common sig-
cytes. Furthermore, nomenclature can be inconsistent. For naling pathways, instead of focusing on common functions
example, IL-7 is highly related to another cytokine that is de- per se. For example, although IL-6 has overlapping actions
noted as thymic stromal lymphopoietin (TSLP) rather than with IL-1 and tumor necrosis factor (TNF)-α , these latter
as an interleukin (see section on IL-7 and Thymic Stromal proinflammatory cytokines are not discussed in this chapter
601

because they are not type I cytokines, and the signaling path- by Th2 cells), where IFNγ is a type II structure cytokine pro-
ways they use are distinct from those used by IL-6. This il- duced by Th1 cells.
lustrates the important concept that similar end functions
can be mediated via more than one type of signaling path-
way. This is not to minimize the observation that many type
TYPE I CYTOKINES AND THEIR RECEPTORS
I cytokines in fact do have similar/overlapping functions, as Type I Cytokines—Structural Considerations
detailed in the section on “cytokine redundancy.” Type I cytokines typically share only limited amino acid
The field of IFN research is older than the cytokine field, sequence identity, but strikingly, all type I cytokines whose
but both fields more recently have developed in parallel. structures have been solved by nuclear magnetic resonance
In fact, one can consider the IFNs to be the first cytokines and/or x-ray crystallographic methods achieve similar
that were identified. IFN was discovered as an antiviral ac- three-dimensional structures,5–9 and those whose structures
tivity in 1957. This turned out to be type I IFN (IFNα / β). have not yet been solved are believed to share similar three-
Type II IFN (IFNγ) was discovered in 1965. Over time, it was dimensional structures. Type I cytokines contain four α heli-
recognized that type I cytokines and IFNs/type II cytokines ces and thus are designated as four α-helical bundle cytokines
share a number of common features, including signaling (Fig. 25.1). Within their structures, the first two and last two
pathways. α helices are each connected by long overhand loops, result-
An unfortunate nomenclature issue exists that should ing in an “up-up-down-down” topologic structure, as the
be noted. There are specialized popuplations of T cells that first two helices (A and B) can be oriented in an “up” orien-
include T helper 1 (Th1) and T helper 2 (Th2) cells (see fol- tation and the last two helices (C and D) can be oriented in
lowing discussion), which produced specialized sets of cyto- a “down” orientation, as viewed from the N- to C-terminal
kines, such as IFNγ by Th1 and IL-4 by Th2 cells. These are direction. As shown in Figure 25.1, the N- and C-termini of
sometimes called type 1 and type 2 cytokines (for Th1 and the cytokines are positioned on the same part of the molecule.
Th2 cytokines); unfortunately, this jargon can result in con- Type I cytokines are either “short chain” or “long chain”
fusion as IL-4 is a type I (ie, four α-helical bundle) cytokine four α-helical bundle cytokines based on the lengths of the
that is functionally a type 2 cytokine (in that it is produced α helices.8 Some of the short-chain cytokines include IL-2,
FIG. 25.1. Schematic of Four a-Helical Bundle Cytokines. Schematic drawing showing typical short-chain and long-chain four helical bundle
cytokines. Although these both exhibit an “up-up-down-down” topology to their four α helices, note that in the short-chain cytokines, the
AB loop is in behind the CD loop, whereas in the long-chain cytokines the situation is reversed. See text. The figure was provided by Dr. Alex
Wlodawer, National Cancer Institute.

CHAPTER 25 TYPE I CYTOKINES AND INTERFERONS, AND THEIR RECEPTORS | 603
connects helix B and loop CD, and the third disulfide bond
TABLE 25.1 Four Helical Bundle Cytokines (between Cys 3 and Cys 127) connects the residue preceding
helix B with helix D. In GM-CSF, the N-terminus of helix
Short-Chain Cytokines Long-Chain Cytokines
B and the N-terminus of β strand CD are connected by one
IL-2 IL-6 disulfide bond, whereas the other disulfide bond connects
IL-4 IL-11 the C-terminus of helix C and a strand following helix D. In
IL-7 Oncostatin M IL-2, a single essential disulfide bond between Cys 58 and Cys
IL-9 Leukemia inhibitory factor 105 connects helix B to strand CD. Thus, each cytokine has
IL-13 CNTF evolved distinctive disulfide bonds to stabilize its structure,
IL-15 Cardiotropin-1 although it is typical that helix B is connected to the loop be-
IL-21 NNT-1/BSF-3 tween helices C and D. The structures formed by helices A
TSLP and D are more conserved than those formed by helices B and
IL-3 Growth hormone C, primarily due to the interhelical angles; helix D and the
IL-5a Prolactin connecting region are the most highly conserved elements
GM-CSF Erythropoietin among the three cytokines.5 This is of particular interest, as
Thrombopoietin the regions of type I cytokines that are most important for cy-
M-CSFa,b Leptin tokine receptor interactions (based on analogy to the growth
SCFb G-CSF hormone receptor structure, see subsequent discussion) in-
BSF-3, B cell-stimulating factor-3; CNTF, ciliary neurotrophic factor; G-CSF, granulocyte-
clude helices A and D and residues in the AB and CD loops,
colony stimulating factor; GM-CSF, granulocyte macrophage-colony stimulating whereas helices B and C do not form direct contacts.5
factor; IL, interleukin; M-CSF, macrophage-colony stimulating factor; NNT-1, novel
neurotrophin-1; SCF, stem cell factor; TSLP, thymic stromal lymphopoietin.
Variations on these typical four α-helical bundle struc-
a
Dimers. tures can occur. For example, IL-5 is unusual in that it is a
b
Different from the other four helical bundle cytokines in that the M-CSF and SCF dimer, wherein the ends containing the N- and C-termini are
receptors (CSF-1R and c-kit, respectively) have intrinsic tyrosine kinase activity
and are not type I cytokine receptors. juxtaposed, and helix D is “exchanged” between the two co-
valently attached monomers so that helix D of each molecule
actually forms part of the four helix bundle of the other.13
M-CSF is also a dimer, but no exchange of helix D occurs.12
IL-3, IL-4, IL-5, granulocyte macrophage-colony stimulating The IFNs form related albeit distinctive structures from
factor (GM-CSF), IL-7, IL-9, IL-13, IL-15, IL-21, macrophage- type I cytokines and are designated as type II cytokines.8
colony stimulating factor (M-CSF), stem cell factor (SCF), IFNβ has an extra helix that is positioned in place of the
and TSLP, whereas long-chain cytokines include growth CD strand.14 IFNγ is a dimer, each of which consists of six
hormone, prolactin, Epo, thrombopoietin, leptin, IL-6, IL- α helices15 (Fig. 25.2). Two of these helices are interchanged,
11, leukemia inhibitory factor (LIF), oncostatin M (OSM), including one from each four α-helical bundle.12,15 IL-10,
ciliary neurotrophic factor (CNTF), cardiotrophin-1 (CT-1), which is closely related to IFNγ, has a similar structure,16 as
novel neurotrophin-1 (NNT-1)/B cell-stimulating factor-3
(BSF-3)/cardiotrophin-like factor (CLC), and granulocyte-
colony stimulating factor (G-CSF) (Table 25.1).8,10,11 The α
helices are approximately 15 amino acids long in short-chain
helical cytokines and 25 amino acids long in long-chain cy-
tokines. Additional differences include differences in the an-
gles between the pairs of helices, and the AB loop is “under”
the CD loop in the short cytokines but “over” the CD loop in
the long cytokines (see Fig. 25.1).7,8,12 Moreover, short-chain
cytokines have β structures in the AB and CD loops, whereas
long-chain cytokines do not.
The division of type I cytokines into short-chain and long-
chain cytokines has evolutionary considerations and cor-
relates with grouping of their receptor chains for these two
subfamilies of type I cytokines. An analysis of short-chain
cytokines has revealed that 61 residues comprise the family
framework, including most of the 31 residues that contrib-
ute to the buried inner core. The similarities and differences
in the structures of IL-2, IL-4, and GM-CSF have been ana-
lyzed.5 Among these cytokines, there is considerable variation FIG. 25.2. Structure of the Interferon g Receptor (IFNgR). Shown are
in the intrachain disulfide bonds that stabilize the structures. ribbon diagrams of the structures of the IFNγ/IFNγR complex as an
example of a type II cytokine/cytokine receptor. In the IFNγR, only
For example, IL-4 has three intrachain disulfide bonds, GM- IFNGR-1 complexed to the IFNγ dimer is shown, as the full structure
CSF has two, and IL-2 has only one. In IL-4, the first disul- with IFNGR-2 is not available. See text for discussion of the struc-
fide bond (between Cys 24 and Cys 65) connects loop AB to ture. The IFNγ-IFNGR-1 structure is from Walter et al.401 The figure
BC, the second disulfide bond (between Cys 46 and Cys 99) was provided by Dr. Alex Wlodawer, National Cancer Institute.

presumably do the more recently identified IL-10–like mol- WSXWS motif is typically just 5′ to the exon encoding the
ecules, including IL-19, IL-20, IL-22, IL-24, IL-26, IL-28A, transmembrane domain. Although serines can be encoded
IL-28B, and IL-29. Interestingly, although not universally by six different codons (ie, sixfold degeneracy in codon
the case, most four α-helical bundle cytokines have four usage), only two of these (AGC and AGT) dominate as the
exons, with helix A in exon 1, helices B and C in exon 3, codons used for the serines in WSXWS. All of these features
and helix D in exon 4.7 A related organization is found for indicate a common ancestral type I cytokine receptor.
IFNγ, as well as for the long-chain helical cytokines, growth Overall, analogous to limited sequence identity between
hormone, and G-CSF. However, there are exceptions, for ex- type I cytokines, there is only limited sequence identity
ample, IL-15 is divided into nine exons, whereas the IFNα s among type I receptor molecules. Nevertheless, they appear
and IFNβ are encoded by single exons. to form similar structures, based on the known structures
for the receptors for growth hormone, prolactin, erythropoi-
etin, IL-4, IL-13, IL-6, and IL-29,19–25 as well as the modeling
Receptors for Type I Cytokines
of other cytokine receptor molecules based on the known
The first report suggesting that type I cytokines interacted structures. The cytokines and their receptors have presum-
with receptors with similar features identified similarities in ably coevolved, with the differences in amino acid sequences
the sequences of the erythropoietin receptor and the IL-2 re- between different cytokines allowing for their distinctive in-
ceptor β chain,17 and subsequent analysis of a large number teractions with their cognate receptor chains. Despite amino
of type I cytokine receptors established these receptors as a acid differences, there are also several sets of cytokines that
superfamily.18 Type I cytokine receptors are generally type I coevolved to interact with shared receptor chains, which
membrane–spanning glycoproteins (N-terminus extracellu- form type I cytokine and cytokine receptor subfamilies.8,11
lar, C-terminus intracellular). The only exceptions are pro- In addition to these noted similarities in the extracellular
teins such as the CNTF receptor α chain (see the following domains, there are sequence similarities that are conserved
discussion), which lacks a cytoplasmic domain and instead in the cytoplasmic domain of cytokine receptors. In par-
has a glycosylphosphatidylinositol (GPI) anchor; however, ticular, membrane proximal “Box 1/Box 2” regions are con-
the orientation of this protein is otherwise similar to that of served (see Table 25.2), with the proline-rich Box 1 region
a type I membrane protein. In their extracellular domains, being the most conserved.26 This will be discussed in greater
a number of conserved features have been noted (Table detail related to its role in the binding of JAK kinases.
25.2). These include four conserved cysteine residues that
are involved in intrachain disulfide bonding, and a trypto-
phan residue, located two amino acids C-terminal to the sec- Type I Cytokine Receptors Are Homodimers,
ond conserved cysteine. In addition, a membrane proximal Heterodimers, or Higher Order Receptor Oligomers
WSXWS (Trp-Ser-X-Trp-Ser) motif is generally conserved, The first cytokine receptor structure solved was that for
although again exceptions exist, for example, in the growth growth hormone (Fig. 25.3).19 Prior to x-ray crystallographic
hormone receptor, the motif is a substantially different analysis, it was believed that growth hormone bound to its
YGEFS (Tyr-Gly-Glu-Phe-Ser) sequence, and in the IL-23R,
it is WQPWS (Trp-Gln-Pro-Trp-Ser). In some cases, such as
the common cytokine receptor β chain, βc, shared by the IL-3,
IL-5, and GM-CSF receptors (see subsequent discussion), the
extracellular domain is extended, with a duplication of the
domains containing the four conserved cysteines and the
WSXWS motif. Another shared feature of type I cytokine re-
ceptors is the presence of fibronectin type III domains.
The two pairs of conserved cysteine residues are typically
encoded in two adjacent exons, and the exon containing the
Features Common to Type I

TABLE 25.2
Cytokine Receptors
Extracellular domain
1. Four conserved cysteine residues, involved in intrachain
disulfide bonds
2. WSXWS motif
FIG. 25.3. Structure of the Growth Hormone Receptor. Shown are
3. Fibronectin type III modules
ribbon diagrams of the structure of the growth hormone receptor
Cytoplasmic domain as an example of a type I cytokine receptor. For growth hormone,
1. Box 1/Box 2 regions—The Box 1 region is a proline-rich both growth hormone receptor monomers are shown. See text for
region that is involved in the interaction of Janus family discussion of the structures. The growth hormone/growth hormone
tyrosine kinases. receptor structure is from de Vos et al.19 The figure was provided by
Dr. Alex Wlodawer, National Cancer Institute.

receptor with a stoichiometry of 1:1, but x-ray crystal solu- with heterodimeric receptors, and that the coordination of
tion structure revealed that a single growth hormone mole- two different receptor chains in heterodimeric receptors
cule interacts with a dimer of the growth hormone receptor, would have evolved in order to allow higher levels of special-
in which each receptor monomer contributes a total of seven ization. In this regard, it is interesting that growth hormone
β strands. Perhaps the most striking finding is that totally and Epo, whose actions are vital for growth and erythro-
different parts of growth hormone interact with the same poiesis, bind to receptor homodimers, whereas the heterodi-
general region of each receptor monomer. The three-dimen- meric structures that typify the immune system are perhaps
sional structure for the growth hormone/growth hormone more “specialized” functions that arose later in evolution.
receptor complex is shown in Figure 25.3. Solving the struc- Interestingly, all type I cytokines known to interact
ture also clarified the basis for the assembly of the growth with homodimers (growth hormone, prolactin, Epo, and
hormone receptor complex.19 Growth hormone first inter- G-CSF) are long-chain helical cytokines, although other
acts with one receptor monomer via a relatively large and long-chain helical cytokines (eg, cytokines whose recep-
high-affinity interaction surface (site I), spanning ∼1230 Å 2. tors contain gp130; see following discussion) interact with
A second receptor monomer then interacts with the growth heteromeric receptors. The short-chain cytokines that
hormone/growth hormone receptor complex via two con- signal through homodimers are SCF and M-CSF, but in
tact points: one on growth hormone (spanning ∼900 Å 2) these cases, the receptors (c-kit and CSF-1R, respectively)
(site II) and the other on the first receptor monomer (span- are different from type I cytokine receptors in that they
ning ∼500 Å 2) (site III), located much more proximal to the contain intrinsic tyrosine kinase domains. Thus, SCF
cell membrane. Thus, three extracellular interactions are re- and M-CSF are not typical type I cytokines and all other
sponsible for the formation and stabilization of the growth short-chain cytokines signal through heterodimers or
hormone/growth hormone receptor complex. Mutations in more complex receptor structures (eg, IL-2 and IL-15 re-
critical residues in site I should prevent growth hormone ceptors have three components).
binding to its receptor, whereas inactivating mutations in Heterodimeric receptors are involved when site II on a
site II can be predicted to prevent dimerization and signal cytokine has evolved to a point where it interacts with a dif-
transduction, suggesting a basis by which different classes of ferent receptor molecule than site I does. This latter situa-
antagonists might be identified. tion is the case for many cytokines, including all short-chain
The growth hormone/growth hormone receptor struc- type I cytokines except for SCF and M-CSF. Overall, several
ture revealed that the growth hormone receptor extracel- sets of type I cytokines fall into distinct groups, wherein
lular domain is composed of two fibronectin type III mod- each group shares at least one common receptor component.
ules, each of which is approximately 100 amino acids long This phenomenon is observed for certain sets of both short-
and contains seven β strands, resulting in the formation of chain and long-chain four α-helical bundle cytokines, and
an immunoglobulin (Ig)-like structure. The contact sur- depending on the set of cytokines, the shared chain interacts
face between ligand and receptor occurs in the hinge re- either with site I or site II.
gion that separates these two fibronectin type III modules. The structures of the low- and high-affinity forms of
Analysis of a growth hormone–prolactin receptor complex the IL-2 receptor have now been solved9,23,28 ; these are the
revealed a similar structure for the prolactin receptor.20 first complete structures for a short-chain cytokine/recep-
The growth hormone/growth hormone receptor struc- tor complex.29 Moreover, they are of added interest in that
ture served as a paradigm for the structures of other type the low-affinity receptor involves the interaction of IL-2 with
I cytokine receptors. As a receptor homodimer, it immedi- IL-2Rα , which is not a type I cytokine receptor protein but
ately served as a model for other homodimers, such as the instead is a distinctive sushi-domain containing protein,
Epo receptor, whose structure was solved 21 using a small whereas the high-affinity receptor includes not only IL-2Rα
protein mimetic (20 amino acid long peptide) of Epo.27 The but also IL-2Rβ and γc, which are both type I cytokine re-
Epo receptor structure is similar to that of the growth hor- ceptor proteins that are shared either with IL-15 (IL-2Rβ)
mone receptor, although the “site III” stem region interac- or with IL-4, IL-7, IL-9, IL-15, and IL-21 (γc),30 as discussed
tion surface in the Epo receptor is much smaller than that in the following text. In the structure of IL-2 bound to its
in the growth hormone receptor, comprising only 75 Å 2.21 high-affinity receptor (Fig. 25.4), there is a long peptide
In addition to the structural similarities for the growth connecting the IL-2Rα globular head and transmembrane
hormone and Epo receptors and perhaps other homodi- segment, allowing the binding site on this protein to extend
meric type I cytokine receptors, a similar structure was relatively far from the cell surface in order to bind the dor-
achieved by the heterodimeric growth hormone/growth sal surface of IL-2. Both the IL-2/IL-2Rα and IL-2/IL-2Rβ
hormone receptor/prolactin receptor structure,20 in which contacts are independent, and IL-2Rα does not appear to
one of the growth hormone receptor monomers is replaced contact either IL-2Rβ or γc ; however, IL-2/IL-2Rβ forms a
by a prolactin receptor molecule. Thus, the two surfaces of composite surface with γc, somewhat analogous to the com-
growth hormone interact either with two identical mono- posite surface formed by growth hormone and one growth
mers of growth hormone receptor or with two nonidentical hormone receptor monomer for binding to a second mono-
monomers in the case of the growth hormone receptor/pro- mer. As anticipated, the surface interaction between IL-2
lactin receptor heterodimer. and γc is smaller than that between IL-2 and either of the
It seems reasonable that cytokine-receptor systems with other chains. In addition to the heterodimeric IL-2 recep-
a homodimeric receptor are evolutionarily older than those tor structure,23 the IL-631,32 and LIF33 receptor complexes,

FIG. 25.4. Structure of the High-Affinity Interleukin (IL)-2 Receptor. This is the first structure for a short-chain type I cytokine complexed to its
complete receptor. It is particularly interesting in that it includes the sushi-domain containing IL-2Rα chain as well. Reprinted from Wang et al.,23
with permission of Dr. Garcia and Science magazine.
the IL-4/IL-13 receptor34 complexes, and GM-CSF receptor Mature IL-2 is a 133 amino acid long peptide that can
structure35 have been solved. act as a major T-cell growth factor, in keeping with its origi-
nal discovery as a T-cell growth factor.51 Although IL-2 is
TYPE I CYTOKINE RECEPTOR FAMILIES AND THEIR not produced by resting T cells, it is rapidly and potently
RELATIONS induced following antigen encounter with resting CD4 + T
cells, and transcription and synthesis of IL-2 are often used
Cytokines that Share the Common Cytokine Receptor as indicators of T-cell receptor (TCR)-mediated cellular ac-
g Chain (Interleukin-2, Interleukin-4, Interleukin-7, tivation. Although the antigen determines the specificity of
Interleukin-9, Interleukin-15, and Interleukin-21) the T-cell immune response, the interaction of IL-2 with
The receptors for six different immunologically important high-affinity IL-2 receptors regulates the magnitude and
cytokines, IL-2, IL-4, IL-7, IL-9, IL-15, and IL-21, share the duration of the subsequent response, based on the amount
common cytokine receptor γ chain, γc (CD132).11,30,36–45 of IL-2 produced, the levels of high-affinity receptors ex-
These cytokines are all short-chain four α-helical bundle pressed, and the duration of IL-2 production and receptor
cytokines; basic features of these cytokines are summarized expression. IL-2 can act in either an autocrine or paracrine
in Table 25.3. The properties of these cytokines and their fashion, depending on whether the producing cell is also
distinctive receptor chains are summarized in the following the responding cell or whether the responding cell is a non-
text, followed by a discussion of the discovery that they share producing cell. The gene encoding IL-2 is located on chro-
a common receptor component and the implications thereof. mosome 4,52 and like many other helical cytokines, its gene
IL-2 is important not only for its function but also his- consists of four exons.7 IL-2 binds to three different classes
torically, as it was the first type I cytokine that was cloned,46 of receptors. These are formed by different combinations of
the first type I cytokine for which a receptor component three different chains, IL-2Rα ,47,48,53 IL-2Rβ,54–57 and a pro-
was cloned,47,48 and the first short-chain type I cytokine tein initially called IL-2Rγ36 but now known as the common
whose receptor structure was solved.23 Many general prin- cytokine receptor γ chain, γc.11,30,37 These different classes of
ciples have derived from studies of this cytokine, including IL-2 receptors are discussed in the following text.
its being the first cytokine that was demonstrated to act in In addition to its being a T-cell growth factor, IL-2 has
a growth factor–like fashion through specific high-affinity other important actions as well (see Table 25.3).29 For ex-
receptors, analogous to the growth factors being studied by ample, it can increase Ig synthesis and J chain transcription
endocrinologists and biochemists.49,50 in B cells,50,58,59 potently augment the cytolytic activity of

TABLE 25.3 Features of Cytokines Whose Receptors Share γc
Chromosome
Cytokine Major Source Sizea Actions Location (h/m) Genomic Org
IL-2 Activated T cells h153 aa/20aa T-cell growth factor 4q26–27/3 Four exons
(Th1 cells) m169aa/20aa B-cell growth, Ig production, J chain expression
15.5 kDa Induces LAK activity
Induces tumor infiltrating lymphocyte activity
Augments NK activity
Critical roles in antigen-induced cell death
Stimulates macrophage/monocyte
Antitumor effects
Promotes Treg development
Promotes Th1 and Th2 differentiation
Inhibits Th17 differentiation
IL-4 Activated T cells h153/24 aa B-cell proliferation 5q31.1/11 Four exons
(Th2 cells) m140 aa/20 aa Ig class switch: IgG1, IgE production
CD4+NK1.1+ natural T cells 18 kDa Augment MHC II, Fcε receptors, IL-4Rα, and IL-2Rβ expression
Th2 cell differentiation
Antitumor effects
CHAPTER 25
IL-7 Stromal cells h177 aa/25aa Thymocyte growth 8q12–13/3 Six exons
m154aa/25aa T-cell growth
17–25 kDa Pre–B-cell growth in mice but not human
Survival and growth of peripheral T cells
CD4+ and CD8+ T-cell homeostasis
IL-9 Activated Th cells h144aa/18aa Th helper clones 5q31–35/13 Five exons
m144aa/18aa Erythroid progenitors
14 kDa B cells
Mast cells/allergic responses
Fetal thymocytes
IL-15 Monocytes and many cells h162aa/48aa Mast cell growth 4q31/8 Nine exons
outside the immune m162aa/48aa NK cell development and activity
systemb 14–15 kDa T-cell proliferation
CD8+ T-cell homeostasis

IL-21 Activated CD4+ T cells h162aa/31 aa Comitogen for T-cell proliferation 4q67–27/3 Five exons
m146aa/24 aa Inhibits B-cell proliferation to anti-IgM + IL-4
Augments B-cell proliferation to anti-CD40
Conflicting reports related to NK cells
Cooperates with IL-7 and IL-15 to expand CD8 cells
Antitumor effects
Drives terminal B-cell differentiation to plasma cells
Proapoptotic for B and NK cells
TYPE I CYTOKINES AND INTERFERONS, AND THEIR RECEPTORS
Promotes Tfh differentiation

Promotes Th17 differentiation
|
CD, cluster of differentiation; Ig, immunoglobulin; IL, interleukin; LAK, lymphokine activated killer; MHC, major histocompatibility complex; NK, natural killer; Tfh, T follicular helper; Th, T helper; Treg, regulatory T.
a
h and m refer to human and mouse, respectively. The number of amino acids refers to the length of the open reading frame/length of signal peptide. The number of amino acids in the mature protein is therefore the difference between
607
these numbers. Note that for IL-15, residues 1 to 29 have been identified as a signal peptide and 30 to 48 as a propeptide.
b
More IL-15 messenger ribonucleic acid is produced in skeletal muscle, kidney, placenta, and lung than in thymus or spleen. It is important to note, however, that IL-15 messenger ribonucleic acid is widely expressed without concomitant
production of IL-15 protein so that the source of biologically meaningful IL-15 may be more limited.
9/17/12 5:56 AM
natural killer (NK) cells,60–62 induce the cytolytic activity of and secretion of mouse IgG1 (human IgG4) and being es-
lymphokine activated killer cells, promote the elimination sential for the production of IgE.88 IL-4 is involved in the
of autoreactive cells in a process known as antigen-induced physiologic response to parasites, including helminths, and
(or activation-induced) cell death,29,50,63 and promote the for allergen sensitization. IL-4 induces expression of class II
differentiation of regulatory T (Treg) cells,64,65 cells that sup- major histocompatibility complex molecules and increases
press inappropriate responses and are important for immu- cell surface expression of the CD23 (the low-affinity IgE
nologic tolerance. Low-dose IL-2 is sufficient to promote receptor) on B cells. In addition to its actions on B cells,
Treg-cell survival, and thereby, for example, can protect mice IL-4 can also act as a T-cell growth factor, inducing prolif-
from developing autoimmune diabetes.66 Interestingly, IL-2 eration in both human and mouse T cells, and is critical for
can also prime CD8 + T cells during a primary response to normal differentiation of Th2 cells.89 Moreover, when IL-4
undergo enhanced proliferation in vivo during a secondary and transforming growth factor (TGF)-β are combined,
response,67,68 and autocrine IL-2 is believed to be required cells that produce IL-9 (Th9) cells are induced (see follow-
for secondary expansion of CD8 + memory T cells.69 There ing discussion). When combined with phorbol 2-myristate
appears to be a complex interplay between IL-2 and inflam- 3-acetate, IL-4 is also a potent comitogen for thymocytes.
matory signals to regulate effector and memory cytolytic Importantly, IL-4 can inhibit certain responses of cells to
T-lymphocytes generation in lymphocytic choriomenin- IL-2.91 Moreover, IL-4 can exert actions on macrophages, he-
gitis virus infection, with persistent IL-2 promoting effec- matopoietic precursor cells, stromal cells, and fibroblasts.92
tor rather than memory cytotoxic T-lymphocyte develop- The gene encoding IL-4 is located on human chromosome
ment.70,71 In Listeria monocytogenesis infection, Th1 effector 5 (5q23.3–31.2) and mouse chromosome 11,92,93 in the same
memory cells highly express the transcription factor T-Bet region as IL-3, IL-5, IL-13, and GM-CSF. The type I IL-4
and IL-2Rα , and IL-2Rα appears to be critical for the de- receptor is expressed on T cells and other hematopoietic
velopment of these cells.72 Presumably because of its criti- cells consists of the 140 kDa IL-4Rα protein88,94–96 and γc.38,39
cal role in Treg development, the absence of IL-2, IL-2Rα , or Expression of IL-4Rα tends to be quite low, and cells that
IL-2Rβ leads to autoimmunity. Interestingly, the production potently respond to IL-4 often express only a few hundred
of IL-2 by mast cells has been reported to contribute to the receptors per cell. In addition to the type I IL-4 receptor,
suppression of chronic allergic dermatitis by its increasing an alternate form of the receptor (the type II IL-4 receptor),
the relative number of Treg cells at the site of inflammation.73 containing IL-4Rα and IL-13Rα1, although not expressed
IL-2 is also important in inducing or inhibiting Th differ- on mature T cells, is expressed on many other cell types
entiation. It is required for efficient Th174 and Th2 differen- and can transduce IL-4 signals into these cells. For example,
tiation,75–78 which are populations of T helper cells fi rst de- IL-13Rα1 is expressed on neonatal Th1 cells and the type II
scribed based on patterns of cytokine production by mouse IL-4 receptor has been implicated as mediating apoptosis of
T cells.79 This theme was then extended to human cells as these cells.97
well,80–84 although there are some variations in humans IL-7 is not produced by lymphocytes but instead is a 152
and mice in terms of the degree of how tightly restricted amino acid long cytokine that is produced by stromal cells
cytokine expression is. IFNγ is the cytokine most reliably and certain other cells.98–100 Based on the analysis of patients
produced by Th1 cells; IL-2 is also produced by these cells, with X-linked severe combined immunodeficiency (SCID),
although without as rigorous an association. In contrast, JAK3-deficient SCID, and IL-7Rα-deficient SCID, IL-7
IL-4, IL-5, IL-6, IL-9, IL-10, and IL-13 are produced by Th2 receptor-dependent signaling is essential for T-cell develop-
cells. In both species, certain type I cytokines (eg, IL-3 and ment in humans (discussed subsequently). The major role
GM-CSF) are produced by both Th1 and Th2 cells. IL-12 of IL-7 is to enhance thymocyte survival, growth, and dif-
and the transcription factors signal transducer and activa- ferentiation101–105 as well as low-affinity peptide-induced pro-
tor of transcription (STAT)4 and T-Bet are the major tran- liferation, and thus it promotes homeostatic proliferation of
scription factors promoting Th1 differentiation, whereas naive and memory CD8 + T cells.100,105–107 Additionally, IL-7
IL-4 and the transcription factor STAT6 drive Th2 differ- can regulate the homeostasis of CD4 + memory T cells108,109
entiation. For Th1 cells, IL-2 acts by STAT5-dependent in- and also can stimulate the growth of mature T cells.30,105,110,111
duction and expression of IL-12Rβ1/IL-12Rβ2 and T-Bet,74 Although IL-7, as noted previously, is believed to primar-
whereas for Th2 differentiation, IL-2 via STAT5 promotes ily be a stromal factor, IL-7 has also been noted to be pro-
IL-475,76 and IL-4Rα77 expression. In addition, IL-2 also af- duced by the liver and kidney but not spleen or lymph nodes
fects chromatin remodeling at the Ifng locus during Th1 after toll-like receptor (TLR) signaling. Although basal liver
differentiation.85 IL-2 also inhibits Th17 differentiation,86 production is low, after TLR signaling, hepatic IL-7 can pro-
in part via downregulation of IL-6Rα /gp130 expression74 mote CD4 and CD8 T-cell survival and promotes antigen-
and by inducing T-bet,74,87 which prevents Runx1-mediated specific T-cell responses.112 In addition, in the mouse, IL-7
activation of RORγ t, a transcription factor that drives Th17 is vital for the growth of mouse pre-B cells,96,98,103,104,111 and
differentiation.87 transient IL-7 signaling can inhibit Ig heavy chain gene rear-
Like IL-2, IL-4 is produced primarily by activated CD4 + rangements.113 IL-7 via STAT5 has also been shown to inhibit
T cells.88,89 IL-4 is also produced by CD4 + NK1.1+ “natural” Ig-κ recombination in pro-B cells.114 In contrast to its require-
T cells, denoted as natural killer T (NKT) cells,90 and by mast ment for B-cell development in the mouse, there is normal
cells and basophils.88 IL-4 is the major B-cell growth factor, B-cell development in patients with defective IL-7 signaling,
and it promotes Ig class switch, enhancing the production as is found in patients with X-linked SCID, JAK3-deficient

SCID, and IL-7Rα-deficient SCID (see following discussion). which can be induced by IL-4 plus TGF-β in a fashion that is
Thus, human B cells can develop normally in the absence dependent on PU.1 and interferon regulatory factor (IRF)-
of IL-7 responsiveness, demonstrating that in humans, IL-7 4.137 Interestingly, the addition of IL-25 further enhances the
is not vital for the growth of human pre-B cells,30,115 and it production of IL-9.138 IL-9–producing cells are typically IL-
remains unknown whether IL-7 plays important roles in 9 + IL-10 + Foxp3– effector T cells and can be derived from
human B-cell biology. Interestingly, IL-7 also acts on DCs, Th2 cells.139,140 Overall, IL-9 is believed to play a key role
and this appears to limit the homeostatic proliferation of in allergic responses; additionally, given the ability of both
T cells under lymphopenic conditions.100,116 Clinically, IL-7 Th17 and Treg cells to produce IL-9 during autoimmunity
has been of interest in terms of its ability to induce T-cell and transplantation, it will be interesting to further clarify
proliferation and to thereby increase T-cell numbers, with a the role of IL-9 in autoimmunity.141 Currently, there is some
greater effect on CD8+ than on CD4 + T cells, as well as to confusion, as one study showed that antibody blockade of
augment the diversity of the TCR repertoire117; IL-7 also can IL-9 or IL-9Rα blocks experimental autoimmune encepha-
increase antigen-specific responses following vaccination100 lomyelitis (EAE) disease progression,142 whereas another
and can promote antiviral immunity, apparently in part study showed that IL-9Rα knockout mice develop more
due to its induction of IL-22 and repression of SOCS3.118 The severe EAE.138 While mouse IL-9 is active on human cells,
gene encoding IL-7 is located on human chromosome 8q12 human IL-9 is not biologically active on mouse cells (the op-
to 8q13119 and mouse chromosome 3. The functional IL-7 re- posite situation from that for IL-2). Human IL-9 is located
ceptor contains the 75 kDa IL-7Rα120 and γc.37,40 Interestingly, on chromosome 5q31 to 5q35,143 which is also the location
signaling via the TCR, IL-2 or IL-7 can downregulate IL-7Rα for the genes encoding IL-3, IL-4, IL-5, IL-13, and GM-CSF.
expression,121,122 with PI 3-kinase/AKT and GFI1B being im- In contrast, mouse IL-9 is “isolated” on chromosome 13,
plicated in its downregulation in T cells.122–124 The downreg- while IL-3, IL-4, IL-5, IL-13, and GM-CSF are clustered on
ulation of IL-7Rα not only can both decrease responsiveness chromosome 11. IL-9 binds to the 64 kDa IL-9Rα binding
of cells but can also increase the availability of the cytokine protein, which is similar in size to γc,144 and the functional
for other cells that are poised to respond.11 Induction of IL- IL-9 receptor consists of IL-9Rα plus γc.30,41,42
7Rα requires PU.1 in B cells125 and another Ets family pro- IL-15 was identified as a T-cell growth factor that also
tein, GABP, in T cells.123 unexpectedly was expressed in the supernatant of a human
IL-9 was originally described as a mouse T-cell growth T-lymphotropic virus type I (HTLV-I)–transformed T-cell
factor126 that is produced by activated T cells and can support line.145,146 Although IL-15 messenger ribonucleic acid (RNA)
the growth of T-helper clones but not of cytolytic clones.127 is produced by a range of nonlymphocytic cell types, it is
In contrast to IL-2, its production is delayed, suggesting difficult to detect physiologic levels of IL-15 protein.147 Its
its involvement in later, perhaps secondary signals. In the main site of synthesis appears to be DCs and monocytes,
mouse, IL-9 can exert proliferative effects on erythroid pro- and unlike IL-2, IL-15 is not produced by activated T cells.148
genitors, B cells, B-1 cells, mast cells, and fetal thymocytes. IL-15 receptors are widely expressed; although IL-15 is per-
IL-9 is identical to mast cell growth–enhancing activity, a haps most important for the development of NK cells149–151
factor present in conditioned medium from splenocytes,128 and CD8 + memory T cells,150,151 it also has both paracrine
and synergizes with IL-3 for maximal mast cell prolifera- and autocrine actions on DCs, including promoting the sur-
tion. IL-9 is highly expressed in the lung of patients with vival of these cells.11 Interestingly, it also regulates the TCR
asthma,129 and overexpression leads to airway inflammation repertoire of γδ intraepithelial lymphocytes152 and cross talk
and Th2 cytokine production.130 Interestingly, IL-9 is sub- between different types of DCs.153 The receptor for IL-15 on
stantially made in the lung by innate lymphoid cells, and in T cells contains IL-2Rβ,43,147,154 γc,43 and an IL-15–specific
these cells neutralizing antibodies to IL-9 lowered expres- protein, IL-15Rα. IL-15Rα shares a number of structural
sion of IL-13 and IL-5, supporting a link between IL-9 and similarities with IL-2Rα , including that it is a sushi domain-
the regulation of the Th2 response.131 Consistent with the ac- containing protein (IL-2Rα has two sushi domains, whereas
tion of IL-9 on thymocytes in vitro, IL-9 transgenic mice de- IL-15Rα has one),155 and the IL2RA and IL15RA genes are
velop thymic lymphomas, and IL-9 is a major antiapoptotic closely positioned on human chromosome 10p14.156 A dis-
factor for such tumors.132 Nevertheless, IL-9 knockout mice tinctive feature of IL-15 signaling is that it signals substan-
have normal T-cell development133 ; instead, they exhibit a tially by a process called transpresentation, wherein IL-15Rα
defect in pulmonary goblet cell hyperplasia and mastocy- on the surface of dendritic cells or monocytes will transpre-
tosis following challenge with Schistosoma mansoni eggs, sent IL-15 to responding cells such as CD8 + T cells or NK
a synchronous pulmonary granuloma formation model. cells that express IL-2Rβ + γc.148,157 IL-15Rα has a longer cy-
However, there was no defect in eosinophilia or granuloma toplasmic domain than IL-2Rα and appears to be critical for
formation.133 Mice expressing an IL-9 transgene in the lung IL-15Rα function, although potentially not for transpresen-
exhibit airway inflammation and bronchial hyperrespon- tation.158 Interestingly, transpresentation can be utilized by
siveness; nevertheless, IL-9 − / − mice exhibit normal eosino- IL-2 as well.159 The very high affinity of IL-15Rα for IL-15 is
philia and airway hyperreactivity in an ovalbumin-induced explained by a large number of ionic interactions mediated
inflammatory model.134–136 Thus, although IL-9 can contrib- by the sushi domain.160 Note that these responding cells can
ute to allergic/pulmonary responses, there are compensa- also express IL-15Rα. In contrast to IL-2, which is a growth
tory cytokines that substitute for IL-9 in at least certain set- factor as well as a mediator of antigen-induced (or activation-
tings. IL-9 can be produced by the Th9 populations of cells, induced) cell death and promoter of Treg differentiation, the

role of IL-15 appears to be more focused on growth of CD8 + transferred into tumor-bearing mice and animals are vac-
T cells,161 maintaining long-lasting, high-avidity T-cell re- cinated with the cognate antigen, the addition of IL-21 or
sponses to foreign pathogens (ie, CD8 + T-cell memory).148,162 IL-21 plus IL-15 has potent antitumor effects.167 Moreover,
This ability has suggested it could have potential as a vaccine treatment of cells with IL-21 in vitro prior to adoptive trans-
adjuvant.163 Overall, the fact that IL-15 promotes the prolif- fer results in markedly enhanced antitumor effects in vivo,
eration and differentiation of B, T, and NK cells; the cyto- conferring a distinctive differentiation program from that
lytic activity of CD8 T cells; and the maturation of dendritic conferred by IL-2.185 IL-21 is now in phase II clinical trials
cells, yet, unlike IL-2, fails to stimulate immunosuppressive for cancer. In addition to its anticancer activity, IL-21 can
Treg cells indicates that it has an array of properties that make also promote autoimmunity. Elevated IL-21 levels have been
it potentially promising as an anticancer therapeutic agent; reported in the BXSB-Yaa mouse model of systemic lupus
as a result, a number of ongoing clinical trials are now in erythematosus,178 and elevated IL-21 levels have been found
progress.164 in a subset of humans with systemic lupus erythematosus
IL-21 is the most recently identified member of the IL-2 (P. Lipsky, personal communication). Elevated IL-21 has
family of cytokines. IL-21 can bind to specific receptors and also been associated with other autoimmune processes, in-
exert actions on T, B, NK cells, DCs, and macrophages.165 It cluding in the non-obese diabetic (NOD) mouse.181 More
augments T-cell proliferation as a comitogen,166 can cooperate importantly, in mouse models of type 1 diabetes, lupus, and
with IL-7 or IL-15 to drive the expansion of freshly isolated experimental allergic uveitis, no autoimmune disease devel-
mouse CD8 + T cells, and it can augment the antitumor activ- ops on an IL-21R knockout background.186–188 IL-21 unex-
ity of CD8 + T cells.167 IL-21 promotes the differentiation of pectedly can also be immunosuppressive via its induction of
Th17 cells168,169 as well as the generation of T follicular helper IL-10.189,190 Consistent with this, an IL-21/IL-10/STAT3 path-
cells.170 In the case of Th17 differentiation, IL-21 is part of an way is required for normal development of memory CD8+
IL-6 to IL-21 to IL-23 signaling cascade169,171,172 to drive the dif- T cells after lymphocytic choriomeningitis virus (LCMV)
ferentiation of these cells in an RORγ-dependent manner.173 infection191; similarly, STAT3, presumably by an overlap-
Polarization of Th17 cells is dependent on TCR stimulation, ping mechanism is important for human T-cell memory
TGF-β, and IL-6 but is independent of IL-23, which instead development, as evidenced by studies in patients with au-
may be required for maintaining/expanding these cells.174–177 tosomal dominant hyper-IgE syndrome, which is caused by
The actions of IL-21 on B cells are particularly com- dominant negative mutations in STAT3.192 IL-21 has com-
plex. It augments B-cell proliferation when combined with plex roles in viral responses. For LCMV, IL-21 is required
anti-CD40 or lipopolysaccharide (LPS) but inhibits prolif- to control chronic infection,193 whereas IL-21 mediates the
eration in response to anti-IgM + IL-4166 ; this inhibition is pathogenic response after infection with pneumonia virus
reversed if anti-CD40 is additionally provided. It induces of mice.194 Interestingly, for hepatitis B virus, IL-21 appears
apoptosis of incompletely activated B cells, perhaps serving to be critical in determining the age-dependent effectiveness
a role analogous to that of IL-2 in activation-induced cell of immune responses, wherein decreased IL-21 production
death of T cells to eliminate incompletely activated cells.165 in younger individuals hinders the generation of critical
In contrast, IL-21 drives terminal differentiation to plasma lymphoid responses to the virus, whereas more robust IL-21
cells of more fully activated cells. Strikingly, IL-21 can drive production in adults is associated with viral clearance that
plasma cell differentiation of both peripheral memory B occurs in 95% of patients.195
cells and cord blood B cells, at least in part explained by The receptor for IL-21 consists of IL-21R plus γc.30,44
its ability to induce expression of BLIMP1.178,179 Strikingly, IL-21R is most related to IL-2Rβ, and like IL-2Rβ, its ex-
IL-21 regulates not only BLIMP-1 but also a broad range pression is induced following cellular stimulation with anti-
of genes via a functional cooperation between STAT3 and CD3 or phytohemagglutinin, and in addition, its expres-
IRF4.180 In IL-21R knockout mice, following immunization, sion is augmented in T cells following transformation with
IgG1 is diminished related to a role for IL-21 in class switch- HTLV-I.196 Both human and mouse IL-21 can act on cells of
ing to IgG1 and IgG3, whereas IgE is elevated, related to the the other species. IL-21 is on human chromosome 4q26 to
ability of IL-21 to inhibit C ε transcription181 and/or to its 4q27, while its receptor is on chromosome 16p11, immedi-
ability to augment the apoptosis of IgE-producing B cells. ately downstream of the IL4R gene.
Analysis of IL-21R/IL-4 double knockout mice has revealed Thus, IL-2, IL-4, IL-7, IL-9, IL-15, and IL-21 collectively
that IL-21 cooperates with IL-4 to globally regulate Ig pro- exhibit partially overlapping roles related to T cells, NK
duction in that these mice exhibit a panhypogammaglobu- cells, B cells, and mast cells, and together would be expected
linemia, mimicking the T-cell phenotype in humans with to play vital roles for normal development and/or function
X-linked SCID.182 IL-21 can also cooperate with IL-15 and of these cellular lineages. As discussed in the following, IL-2,
Flt-3 ligand to increase development of NK cells166 and aug- IL-7, IL-9, and IL-15 utilize primarily STAT5A and STAT5B;
ment antitumor activity183 ; however, it was also reported to IL-4 activates primarily STAT6; and IL-21 activates STAT1,
oppose the actions of IL-15184 and can also direct NK cell STAT3, STAT5A, and STAT5B, with STAT3 being the domi-
apoptosis. Interestingly, IL-21 exerts potent antitumor ef- nant STAT for this cytokine.11 The fact that these six cyto-
fects, including against large established solid tumors.11,165 In kines share γc is of particular interest, especially as the gene
the pMEL-1 CD8 TCR transgenic system, the effector cells encoding γc is mutated in patients with the most common
recognize a melanoma antigen. When cells are adoptively form of severe combined immunodeficiency in humans.

X-Linked Severe Combined Immunodeficiency Disease

Results from Mutations in the Gene Encoding gc
The γ chain was originally identified as a third component
of the IL-2 receptor36 after it became clear that IL-2 recep-
tor α and β chains alone were not sufficient to transduce
an IL-2 signal. The hypothesis that the γ chain was a shared
component of receptors for cytokines in addition to IL-2 was
motivated from a comparison of the clinical phenotypes in
humans that result from defective expression of IL-2 versus.
the γ chain. In 1993, it was discovered that mutations in the
gene encoding the γ chain resulted in X-linked SCID (the
disease is also designated as SCIDX1).30,197 X-linked SCID
is characterized by profoundly diminished numbers of T
cells and NK cells30,45,197–201 (Table 25.4). Although the B cells
are normal in number, they are nonfunctional, apparently
due to a lack of T-cell help as well as an intrinsic B-cell de-
fect.45,200,202 In contrast to the profoundly decreased number
of T cells in patients with X-linked SCID, IL-2–deficient pa-
tients203,204 and mice205 have normal numbers of T cells (the
phenotypes of mice deficient in type I and type II cytokines
and their receptor, JAK kinases, and STAT proteins are sum-
marized in Table 25.15). This observation indicated that
defective IL-2 signaling was unlikely to be responsible for
X-linked SCID, making the finding that the gene encoding FIG. 25.5. Schematic of the Receptors for Interleukin (IL)-2, IL-4, IL-7,
the γ chain was mutated in X-linked SCID all the more un- IL-9, IL-15, and IL-21, Showing Interactions with JAK1 and JAK3.
expected. Thus, the conundrum was why a defect in a com- The figure shows that IL-2, IL-4, IL-7, IL-9, IL-15, and IL-21 all share
ponent of a receptor would cause a more severe than a defect γc. IL-2Rα and IL-15Rα are not shown. Whereas the distinctive chains
in the corresponding cytokine. This led to the hypothesis associate with JAK1, γc associates with JAK3. Mutations in γc cause
that the γ chain was critical for other cytokine receptors as X-linked severe combined immunodeficiency or more moderate forms
of X-linked immunodeficiency. Mutations in JAK3 cause an autoso-
well.197 In this model, defective IL-2 signaling either did not mal recessive form of severe combined immunodeficiency (see text).
contribute to the defects in X-linked SCID or these defects
were explained by the simultaneous inactivation of multiple
signaling pathways.45,197 Indeed, it was found that the γ chain on the dramatically diminished T-cell development not
was also an essential component of both the IL-4 and IL-7 only in IL-7–deficient104 or IL-7Rα –deficient103 mice but
receptors on T cells,37–40 leading to it being renamed as the also in IL-7R-deficient humans with T–B + NK+ SCID,206,207
common cytokine receptor γ chain, γc.37,38 IL-9, IL-15, and yet normal T-cell development in mice deficient in IL-2,205
IL-21 were subsequently also shown to share γc30 (Fig. 25.5). IL-4,208,209 both IL-2 and IL-4,210 IL-9,133 IL-15,151 IL-15Rα ,150
The sharing of γc by six different cytokine receptors re- IL-21R,182,184 and most if not all of the defect in T-cell devel-
vealed that X-linked SCID is a disease of defective cytokine opment in patients with X-linked SCID can be attributed to
signaling. The major deficiencies in X-linked SCID can be defective IL-7 signaling.30 In addition to profoundly dimin-
attributed to defects related to different cytokines. Based ished numbers of T cells, humans with X-linked SCID lack
NK cells. As discussed previously, NK-cell development is
defective in IL-15– and IL-15Rα –deficient mice, indicating
that it is defective IL-15 signaling that is responsible for the
Features of X-Linked Severe defective NK-cell development in X-linked SCID.30
TABLE 25.4
Combined Immunodeficiency In contrast to the greatly diminished number of T cells
and absent NK cells in patients with X-linked SCID, B-cell
1. Absent or profoundly diminished numbers of T cells and
numbers are normal. This is in contrast to the greatly di-
mitogen responses
2. Absence of NK cells
minished numbers of B cells in γc-deficient mice211,212 as well
3. Normal numbers of B cells, but defective B-cell responses as mice deficient in either IL-7 or IL-7Rα , indicating that
4. IgM can be normal but greatly diminished Igs of other IL-7 is not required for pre-B-cell development in humans
classes and underscoring a major difference for IL-7 in human ver-
5. XSCID carrier females exhibit nonrandom X-inactivation sus mouse biology. Indeed, patients with IL-7R-deficient
patterns in their T cells and NK cells; the X-inactivation SCID have a T–B + NK+ form of SCID.206,207 Although B cells
pattern is random in surface IgM-positive B cells but develop in patients with X-linked SCID, they are nonfunc-
nonrandom in more terminally differentiated B cells. tional. This is due in part to a lack of T-cell help (given the
Ig, immunoglobulin; NK, natural killer; XSCID, X-linked severe combined
near absence of T cells in X-linked SCID), but a variety of
immunodeficiency. data indicate an intrinsic B-cell defect as well.30 As discussed

previously, analysis of IL-4/IL-21R double knockout mice different cytokine pathways, mice deficient in IL-2, IL-2Rα ,
indicate that defective signaling by IL-4 and IL-21 appear to and IL-2Rβ appear to be less healthy than the γc-knockout
explain the intrinsic B-cell defect in X-linked SCID.182 mice with more markedly activated T cells and autoimmu-
nity that is not evident in γc-deficient mice. This is at least
Rationale for the Sharing of gc in part explained by the fact that IL-2– deficient mice lack
Why should there have been evolutionary pressure to main- Treg cells and thus develop autoimmune disease,65 whereas
tain the sharing of γc, given the obvious increased risk associ- γc-deficient mice lack T cells due to defective IL-7 signaling
ated with sharing a receptor component when it is mutated? and thus lack effector T cells. These and other data indicate
There are at least two different types of models.45,213 First, that γc plays a major role in regulating lymphoid homeosta-
the sharing of γc could be a basis for shared actions. For ex- sis, as originally indicated.219
ample, given that IL-2, IL-4, IL-7, IL-9, IL-15, and IL-21 can
each act as T-cell growth factors, at least in vitro, it is pos-
sible that γc might couple to signal transducing molecule(s) Cytokines whose Receptors Share the Common
that promote T-cell growth. Second, the sharing of γc might b Chain, bc (Interleukin-3, Interleukin-5, and
represent a means by which one cytokine using γc can modu- Granulocyte Macrophage-Colony Stimulating Factor)
late the signals of the others. To understand this model, it is The hematopoietic cytokines, IL-3, IL-5, and GM-CSF
important to recognize that, in contrast to antigen receptor (Table 25.5) are all synthesized by T cells and exert effects on
complexes, cytokine receptor components individually are cells of hematopoietic lineage.220–222 These cytokines are vital
targeted to the cell surface, and the formation/stability of for proliferation as well as differentiation of myeloid precur-
receptor complexes is promoted or stabilized by ligand bind- sor cells. Of these three cytokines, IL-3 is the most pluripo-
ing, as noted in the discussion of growth hormone, wherein tent221 and historically was also called multi-CSF, reflecting
the second receptor monomer recognizes the combined the large number of lineages on which it can act. It can act
surface of growth hormone and the first growth hormone to promote proliferation, survival, and development of mul-
receptor monomer19 or for the IL-2 receptor where γc “sees” tipotent hematopoietic progenitor cells and of cells that have
a combined IL-2/IL-2Rβ surface.23 This latter finding is con- become dedicated to a range of different lineages, including
sistent with γc originally having been coprecipitated with IL- granulocyte, macrophage, eosinophil, mast cell, basophil,
2Rβ in the presence but not in the absence of IL-2,214 and megakaryocyte, and erythroid lineages. It induces a range of
dimerization of IL-2Rβ and γc is known to be required for effects, including, for example, inducing the production off
efficient signaling.215,216 Thus, receptor heterodimerization IL-4 by basophils.223 IL-3 also can exert end-function effects,
at a minimum is stabilized by the cytokine and physiologi- such as enhancing phagocytosis and cytotoxicity. GM-CSF
cally may be absolutely dependent on the presence of the cy- is mainly restricted to the granulocyte and monocyte/mac-
tokine. In the absence of stable preformed cytokine receptor rophage lineages, but its actions are nevertheless still quite
complexes between γc and the other receptor chains, one can broad.220 It is both a growth and survival factor. In addition,
envision that γc might be differentially recruited to different it can expand the number of antigen-presenting cells, such as
receptors based on the relative amount of a cytokine or its DCs, and thereby may greatly expand the ability of the host to
binding efficiency. In a situation where γc is limiting, a cy- respond to antigen. IL-23 and RORγ t can augment GM-CSF
tokine might then not only induce its own action but could production in Th cells, whereas IL-12, IFNγ, and IL-27 nega-
also simultaneously inhibit the action of another cytokine tively regulate its expression, and GM-CSF is now known to
that was less efficient at recruiting γc to its cognate receptor be required for the initiation of autoimmune neuroinflamma-
complex. tion in experimental autoimmune encephalitis. Collectively,
An analysis of mice deficient in IL-2,205 IL-2Rα ,217 IL- these results suggest that GM-CSF is important for encephali-
2Rβ,218 and γc211,212 provides the interesting observation that togenic actions of both Th1 and Th17 cells.224,225 GM-CSF me-
although the mice lacking γc have defective signaling in six diates autoimmune effects at least in part by enhancing IL-6–
TABLE 25.5 Features of Cytokines whose Receptors Share the Common β Chain, βc
Cytokine Major Source Size Cellular Targets Chromosome Location (h/m) Exons
IL-3 T cells h152/19aa Multiple lineages 5q31.1/11 5
m166/26 aa
22–34 kDa
IL-5 T cells h139/22aa Eosinophils 5 q31.1/11 4
m133/21 B cells (?)
45 kDa dimer
GM-CSF T cells h144/17aa Granulocytes 5q31.1/11 4
m141/17aa Macrophages
23 kDa
GM-CSF, granulocyte macrophage-colony stimulating factor; IL, interleukin.

dependent Th17 cell development and survival.226 Whereas of function so that knockout mice that lack the ability to re-
IL-3 and GM-CSF can act on eosinophils, they act at much spond to all three cytokines (GM-CSF, IL-3, and IL-5) due
earlier stages than IL-5, presumably expanding the number to deletion of βc as well as the mouse IL-3–specific βc-like
of eosinophil-committed precursors cells. IL-5 stimulates the protein (discussed in the following text) nevertheless exhibit
eosinophilic lineage and eosinophil release from the bone relatively normal hematopoiesis. These observations do not
marrow and is essential for expanding eosinophils after hel- minimize the potency of these particular cytokines but in-
minth infections,227,228 and can mediate the killing of S. man- stead underscore a substantial redundancy for a particularly
soni. IL-5 can also induce Ig production in B cells activated important set of functions.221,222,243 It is also noteworthy that
by contact with activated Th cells in mouse systems, and IL-5 βc-deficient mice exhibit defective host responses to infec-
and IL-5Rα knockout mice have diminished CD5 + B1 cells tious challenge, suggesting that these hematopoietic cyto-
and decreased thymocytes until approximately 6 weeks of kines play a vital role in promoting immune function.
age.227,228 Interestingly, the production of IL-5 by memory Th2
cells is driven by GATA3 but downregulated by expression of
eomesodermin, which interacts with GATA3. Most memory Cytokines whose Receptors Share gp130 (Interleukin-6,
Th2 cells produce the transcription factor Eomes, but the Interleukin-11, Oncostatin M, Ciliary Neurotropic
CD62loCXCR3lo population has decreased Eomes and is the Factor, Leukemia Inhibitory Factor, Cardiotrophin-1,
population of cells that produces IL-5.229 Novel Neurotrophin-1/B Cell-Stimulating Factor-3/
On cells that express receptors for more than one of these Cardiotrophin-Like Factor, and Interleukin-27)
cytokines, such as eosinophilic progenitors that express re- There are now eight cytokines that are known to utilize gp130
ceptors for IL-3, IL-5, and GM-CSF, or on mouse pre-B cells, as a signal transducing molecule.244–255 Some of the properties
which express receptors for IL-3 and IL-5, the signals induced of these cytokines are summarized in Table 25.6. This family
are indistinguishable.221,222 Thus, the differential lineage is often referred to as the IL-6 family of cytokines and includes
specificities of these cytokines are determined by the cellu- IL-6, IL-11, OSM, LIF, CNTF, CT-1, NNT/BSF-3/CLC, and
lar distribution of their receptors rather than by fundamental IL-27. This group of cytokines comprises molecules with a di-
differences in the signals that are induced by each cytokine. verse range of actions, ranging beyond the hematopoietic and
These observations are explained by studies demonstrating
that each of these three cytokines has its own unique 60 to 80
kDa α chain (ie, IL-3Rα, IL-5Rα, and GM-CSFRα),230–236 but
that they share a common 120 to 130 kDa β chain, βc.222,237–239 Cytokines whose Receptors
TABLE 25.6
The α chains are the principal binding proteins for the cyto- Share gp130a
kines, whereas the shared βc subunit augments binding affin-
ity but does not exhibit binding activity in the absence of the Cytokine Chromosome Location (h/m)
proper α chain. The α chains have relatively short cytoplas- IL-6 7p21/5
mic domains (approximately 55 amino acids long for IL-3Rα, IL-11 19q13.3–13.4/7
IL-5Rα, and GM-CSFRα) and are not believed to play major LIF 22q12.1–12.2/11
roles in signaling function, whereas βc, with its cytoplasmic OSM 22q12.1–12.2/11
domain of 432 amino acids, is the primary determinant of the CNTF 11q12.2/19
signal. As a result, there is a relative compartmentalization of CT-1 16p11.1–11.2/7
binding and signaling function for these cytokines, although NNT-1/BSF-3/CLC 11q13/19
the cytoplasmic domains of the GM-CSFRα and IL-5Rα IL-27
chains (and by analogy, perhaps the IL-3Rα chain), as well
as βc, appear to be capable of at least modulating the growth Overlapping actions of several gp130 cytokines
signals in transfected cells.222,240–242 In any case, the sharing IL-6 IL-11 LIF OSM CNTF CT-1
of βc helps to explain why the signals induced by IL-3, IL-5, Growth of myeloma cells + − + + + ?
and GM-CSF are similar on cells that can respond to more Maintenance of embryonic − − + + + +
than one of these cytokines. The situation for the βc family stem cell pluripotency
Induction of hepatic acute + + + + + +
of cytokines is therefore quite different from the receptors for
phase proteins
γc family cytokines, wherein the chains with the largest cyto-
Induction of cardiac − + + + +/− +
plasmic domains (IL-2Rβ, IL-4Rα, IL-7Rα, IL-9Rα, and IL- hypertrophy
21R) not only contribute most to signaling specificity but also Induction of osteoclast − + + + ? ?
are the proteins principally involved in ligand binding (note formation
that for IL-2 and IL-15, IL-2Rα and IL-15Rα, respectively, Enhanced neuronal sur- + + + + + +
serve important roles as well). The shared chain, γc, serves a vival/differentiaion
vital accessory function (hence the development of X-linked Inhibit adipogenesis ? + + ? ? ?
SCID when the IL2RG gene is mutated), but it contributes less
BSF-3, B cell-stimulating factor-3; CLC, cardiotrophin-like factor; CNTF, ciliary neuro-
to cytokine binding and does not provide an obvious basis for trophic factor; CT-1, cardiotrophin-1; IL, interleukin; LIF, leukemia inhibitory factor;
signaling specificity (see subsequent discussion). a
NNT-1, novel neurotrophin-1; OSM, oncostatin M.
Most of the data in this table are derived from Yin et al.248
An interesting feature of the βc family of hematopoietic Note that NNT-1 can support survival of chicken embryonic and sympathetic neurons,
cytokines is that there appears to be considerable redundancy can induce amyloid A, and analogous to IL-6 can induce B-cell hyperplasia.

immune systems to the central nervous and cardiovascular IL-11. IL-11 is also produced by lung eosinophils and various
systems, making them even more “multifunctional” than the structural cells in the lung and is expressed in patients with
γc and βc families of cytokines, whose actions are more re- modest-to-severe asthma.272 IL-11 signals via a receptor com-
stricted to the lymphoid and hematopoietic systems. plex containing both IL-11Rα273 and gp130.244 Interestingly,
IL-6 was originally identified and then cloned as a B-cell IL-11Rα messenger RNA can be alternatively spliced to yield
differentiation factor that stimulated terminal differentia- a form lacking the cytoplasmic domain, and like IL-6Rα, a
tion/maturation of B cells into antibody-producing plasma soluble form of IL-11Rα can coordinate with IL-11 to trans-
cells.256–258 However, IL-6 also can exert effects for T-cell signal in cells expressing gp130.274 Studies on the stoichiom-
growth and differentiation (and thus is a thymocyte “comi- etry of the IL-11 receptor complex failed to reveal dimeriza-
togen”), induce myeloid differentiation into macrophages, tion of gp130 to itself or LIFRβ. Thus, assuming that it forms
induce acute-phase protein synthesis of hepatocytes, and a hexameric receptor complex, only five of the members are
exert actions on keratinocytes, mesangial cells, hematopoi- known: two molecules of IL-11 and IL-11Rα, respectively,
etic stem cells, the development of osteoclasts, and neural and one of gp130, suggesting that another component may
differentiation of PC12 cells.250 IL-6 is also a mediator of still be found.275 IL-11 and STAT3 are expressed in gastroin-
amplified lymphocyte trafficking during febrile inflamma- testinal cancers associated with inflammation, suggesting a
tory responses, which is dependent on IL-6–mediated acti- possible link to cancer.269
vation of L-selectin.259 IL-6 binds to an 80 kDa IL-6 binding LIF is another multifunctional cytokine originally cloned
protein, denoted IL-6Rα , which has a comparatively short based on the activity associated with its name.276 LIF can sup-
82 amino acid long cytoplasmic domain.257,260 This IL-6-IL- press the differentiation of pluripotent embryonic stem cells,
6Rα complex then interacts with and recruits the 130 kDa inhibit adipogenesis (like IL-11), and induce monocyte dif-
signal transducing molecule, gp130, which together with IL- ferentiation of the M1 murine leukemia cell line, thus mim-
6Rα can form a functional IL-6 receptor.250 From a struc- icking a number of the actions of IL-6.277 In addition, it exerts
tural perspective, gp130 contains a total of six fibronectin a number of actions in the central nervous system, and LIF
type III modules, with the four conserved cysteine residues was shown to be identical to cholinergic neural differentia-
and the WSXWS motif being located in the second and third tion factor.278 and can induce acetylcholine synthesis while si-
of these modules, starting from the N-terminus. As such, multaneously suppressing catecholamine production, thereby
these regions are topologically positioned at greater distance inducing cholinergic function while suppressing noradrener-
external to the cell membrane than is the case for the other gic function.278 LIF has been shown to be essential for em-
type I cytokine receptors discussed previously. bryo implantation.279 LIF binds to a receptor (LIFRβ) that is
The IL-6 system illustrates a novel twist related to the prop- structurally related to gp130,280 but the functional LIF recep-
erties of their principal binding proteins: The cytoplasmic do- tor (LIFR) receptor requires the heterodimerization of LIFRβ
main of IL-6Rα is superfluous for signaling; a soluble form of and gp130 as well.246 Interestingly, whereas IL-6 promotes
the IL-6Rα extracellular domain is in fact sufficient for ligand Th17 differentiation, this is inhibited by LIF, apparently
binding and coordination with gp130, a process termed IL-6 based on LIF’s ability to activate ERKs and to augment the
transsignaling.261 Thus, in the presence of soluble IL-6Rα and expression of SOCS-3, which can inhibit STAT3 activation.281
IL-6, many cell types that express gp130 but not IL-6Rα are Interestingly, LIF can also promote the differentiation of Treg
capable of signaling in response to IL-6. It was observed that cells and may exhibit therapeutic potential in multiple sclero-
IL-6 signaling requires the dimerization of gp130 and that the sis.282 LIF maintains mouse embryonic stem cells, but LIF is
overall complex containing two molecules each of IL-6, IL- not capable of maintaining human embryonic stem cells or
6Rα, and gp130 (a dimer of a trimer or a hexamer),32,262–265 mouse epiblast stem cells in the pluripotent state.283
providing a possible paradigm for the stoichiometry of sub- CNTF was discovered based on its ability to promote
units for other members of the IL-6 family of cytokines.24 neuronal survival.284,285 CNTF signals through a receptor
IL-11 was identified as a factor produced by a stromal cell comprising LIFRβ and gp130 but requires a specific bind-
line in response to stimulation with IL-1.266,267 It has a number ing protein, 286,287 now denoted CNTF receptor (CNTFR) α .
of effects on hematopoiesis, particularly in combination with Interestingly, CNTFRα lacks transmembrane and cyto-
IL-3 and SCF. Because IL-11 exhibited “IL-6–like activities,” plasmic domains and instead is a GPI-linked receptor mol-
a complementary deoxyribonucleic acid (DNA) was isolated ecule. CNTFRα appears to provide a receptor-cytokine
based on the presence of IL-6–like activity in the presence of surface with which gp130 and LIFRβ can interact. Thus,
antibodies to IL-6.268 Other actions of IL-11 include the ability CNTF is like IL-6 in that each requires initial binding to
to stimulate the proliferation of lymphoid and hematopoietic a receptor component (CNTFRα or IL-6Rα) that does not
progenitor cells, stimulate megakaryocytic progenitors and require its own cytoplasmic domain for signaling. Whereas
megakaryocyte maturation, and stimulate erythroid progeni- IL-6 signaling involves homodimerization of gp130, CNTF
tors (an action not shared by IL-6).266,267 Overall, it acts on signaling involves the heterodimerization of LIFRβ and
T cells, B cells, hematopoietic cells, epithelium, endothelial gp130. In fact, the functional CNTFR appears to be a hexa-
cells, and osteoclasts.269 Like IL-6, IL-11 induces acute-phase meric structure containing two molecules of CNTF, two of
proteins and augments antigen-specific B-cell responses, CNTFRα , and one each of gp130 and LIFRβ.288 The recep-
but it does not stimulate human myeloma cells.266–268,270 tor is expressed largely within the nervous system and in
Subsequently, adipogenesis inhibitory factor was cloned and skeletal muscle, accounting for largely restricted actions of
found to be identical to IL-11,271 revealing another action of CNTF.286

OSM is a growth regulator that was originally identified

based on its ability to inhibit the growth of A375 human Composition of Receptors for the
TABLE 25.7
melanoma cells.289,290 OSM is a potent growth factor for Interleukin-6 Family of Cytokines
Kaposi sarcoma in patients with acquired immunodeficiency Cytokines whose receptors do not contain LIFRb
syndrome.291,292 Not only OSM can bind directly to gp130 IL-6 IL-6Rα + gp130
and signals through a receptor combination of gp130 and IL-11 IL-11Rα + gp130
LIFRβ246 but also has an alternative receptor comprising a OSM OSMRβ + gp130
specific OSM receptor subunit (OSMRβ) and gp130.293 These
are now known as the type I and type II OSM receptors Cytokines whose receptors contain LIFRb
(OSMRs), respectively. OSM can enhance the development LIF LIFRβ + gp130
of both endothelial cells and hematopoietic cells, possibly by OSM LIFRβ + gp130
increasing hamangioblasts, a common precursor for endo- CNTF CNTFRα + LIFRβ + gp130
thelial and hematopoietic cells.251 CT-1 LIFRβ + gp130 + ?CT1Rα
CT-1 was initially isolated based on its actions on cardiac NNT-1/BSF-3 /CLC + CLF-1 CNTFRα + LIFRβ + gp130
muscle cells.294 However, it is now clear that it is a multifunc- CLC + soluble CNTFRα LIFRβ + gp130
tional cytokine with hematopoietic, neuronal, and develop- Cytokines whose receptors contain OSMRb
mental effects, in addition to its effects on cardiac development OSM OSMRβ + gp130
and hypertrophy.295,296 Like OSM and LIF, CT-1 can also signal IL-31 IL-31R + OSMRβ
through a heterodimer of LIFRβ and gp130.249 Interestingly,
the CT-1 receptor on motor neurons may involve a third re- BSF-3, B cell-stimulating factor-3; CLC, cardiotrophin-like factor; CLF-1, cytokine-
like factor-1; CNTF, ciliary neurotrophic factor; CNTFR, CNTF receptor; CT-1,
ceptor component, possibly GPI linked.297,298 cardiotrophin-1; IL, interleukin; LIF, leukemia inhibitory factor; NNT-1, novel
NNT-1/BSF-3, like CNTF, can also support the survival of neurotrophin-1; OSM, oncostatin M.
Note that the sharing of CNTFRα by CNTF and the dimeric NNT-1/BSF-3/CLC – CLF-1
chicken embryonic sympathetic and motor neurons.253 ligand helps to explain why the phenotype in CNTF−/− mice is less severe than
Interestingly, in mice, NNT-1/BSF-3 can augment the effects that found in CNTFRα−/− mice.
of IL-1 and IL-6 and is a B-cell–stimulating factor (hence
the term BSF-3). The NNT-1 receptor contains LIFRβ and
but only one of which contains LIFRβ. When cytokines share
gp130.253 NNT-1 is also known as CLC, and this forms a
essentially the same receptor, one can hypothesize that two
complex with a soluble receptor protein known as cytokine-
cytokines might exert identical actions on cells that can re-
like factor-1. Together, this complex is a second ligand for
spond to both cytokines. It is clear that the presence of IL-
CNTFR.299 It appears that CLC can also interact directly
6Rα , IL-11Rα , and CNTFRα (either on the cell surface or as
with soluble CNTFR to form a related cytokine.254
a soluble receptor form, discussed subsequently) determines
IL-27 is an IL-6–related cytokine300 that represents a dimer
whether a cell can respond to IL-6, IL-11, and CNTF. This
of the p28 protein (also known as IL-30) and EB13 (Epstein-
raises the interesting question as to whether functional ho-
Barr virus–induced gene 3) (also known as IL-27B), which
mologues of these proteins will also exist for LIF, OSM, CT-1,
can induce proliferation of naive CD4 + T cells. p28 can exist
NNT-1/BSF-3, and IL-27.
by itself independently of EB13 and is capable of antagonizing
gp130-mediated signaling by IL-6, and indeed, overexpres-
sion of p28 in transgenic mice prevents normal development Significance of the Sharing of Receptor Chains
of germinal centers and antibody production.301 Together Interestingly, γc, βc, and gp130 all contribute to signaling, but
with IL-12, IL-23, CLC/cytokine-like factor-1, and CLC/sol- none of these shared cytokine receptor proteins has primary
uble CNTFR, this is one of five cytokines that represent di- binding activity for any known cytokine. Instead, they each
mers including a type I cytokine and a soluble receptor-like increase binding affinity in the context of the primary binding
protein. IL-27 signals via gp130 and the WSX-1/TCCR recep- protein for each cytokine. Consequently, the capacity of a cell
tor,300 which is discussed in the section on diseases of cytokine to respond to a given cytokine is determined by the unique
receptors and related molecules. IL-27 has proinflammatory binding chain(s), but signaling pathways can be shared.
and anti-inflammatory effects. Although it was suggested to
promote Th1 responses, it is clear that it can also antagonize
such responses, and IL-27 can promote effector responses of Other Receptors with Similarities to gp130
both CD4 + and CD8 + T cells, of NK cells, and it additionally (Granulocyte-Colony Stimulating Factor Receptor,
stimulates mast cells. The anti-inflammatory effects of IL-27 OB-R, Interleukin-12Rb1, Interleukin-12Rb2, and
are indicated by the absence of WSX-1 resulting in increased Interleukin-31R)
mast cell and macrophage responses.302,303 As noted previously, LIFRβ and OSMRβ bear some simi-
Thus, eight cytokines (IL-6, IL-11, LIF, CNTF, OSM, CT-1, larities to gp130.293 In addition, the G-CSF receptor, the
NNT-1/BSF-3, and IL-27) all have receptors that are depen- leptin receptor (also denoted OB-R, for obesity receptor),
dent on gp130.304 These can be divided into two sets of cyto- IL-12/IL-23 receptor components, and IL-31R all resemble
kines: those known to not require LIFRβ (IL-6, IL-11, and gp130. The amino acid identity among these different re-
IL-27) and those that use both gp130 and LIFRβ (LIF, CT-1, ceptors, compared pairwise, ranges from 18% to 32%,
OSM, CNTF, and NNT-1/BSF-3) (Table 25.7), with OSM with LIFRβ and OSMRβ being the most similar. Although
having two forms of receptors, each of which contains gp130 G-CSF is granulocyte-colony stimulating factor, it may have

a broader role. For example, G-CSF has been shown to stim- receptor and bears some of the features typical of type I
ulate myoblast proliferation and thereby to affect skeletal receptors, including four conserved cysteines, a conserved
muscle formation,305 indicating broad pleiotropic actions by tryptophan, and a WSEWAS motif, which has similarity to
this cytokine. the typical WSXWS motif.311 Moreover, as both IL-12 recep-
tor (IL-12Rβ1 and IL-12Rβ2) chains bear some similarity to
Leptin gp130,312,316,317 one can think of p40 as a functional homo-
Leptin is the product of the obesity (ob) gene, an adi- logue of the soluble p80 IL-6Rα chain. Thus, for this cyto-
pose tissue–derived cytokine that plays a role in body kine, part of the “receptor” has become part of the cytokine.
weight homeostasis.306–308 Additionally, leptin affects thy- Interestingly, all cells that produce IL-12 synthesize much
mic homeostasis, increasing thymocyte number and hav- more p40 than p35, suggesting that the careful control of
ing antiapoptotic effects in the thymus and periphery. It signaling is at the level of the “primordial” p35 cytokine part
is proinflammatory, promoting Th1 differentiation, and of IL-12. p40 is on human chromosome 5q31 to 5q33 while
augmenting the production of TNF- α , IL-1, and IL-6 by p35 is on 3p12 tp 3p13.2.311 As noted previously, it is interest-
monocytes/macrophages.308 The leptin receptor, OB-R, was ing that IL-2 induces expression of IL-12Rβ1 and IL-12Rβ2.
cloned and found to be most closely related to the gp130 sig- IL-23 is similar to IL-12 in that it also contains p40.318
nal transducer, G-CSF receptor, and LIFRβ.309 Interestingly, However, rather than also containing p35, IL-23 is a dimer
this receptor is encoded by the “diabetes gene,” which is of a p40 and p19 (IL-23A). IL-23 signals via a receptor that
mutated in db/db mice.310 contains IL-12Rβ1 but not IL-12Rβ2.319 Instead, another
receptor chain, denoted IL-23R, is the second component
Interleukin-12, IL-23, and IL-35 of the IL-23R 320 ; both IL-12Rβ2 and IL-23R are located on
As noted previously, IL-12 is the major inducer of Th1 cells. chromosome 1 within 150 kb of each other.320 Both IL-12
IL-12 is primarily produced by phagocytic cells in response and IL-23 activate JAK2 and TYK2. IL-12 and IL-23 can
to bacterial and intracellular parasites, such as Toxoplasma activate STAT1, STAT3, STAT4, and STAT5. STAT4 is the
gondii, but it is also produced by other antigen-presenting dominant STAT protein activated by IL-12.320 Interestingly,
cells, such as B cells.311 IL-12 potently induces the production as compared to human IL-23R, mouse IL-23R contains a 20
of IFNγ by NK cells and T cells and is also a growth factor for amino acid duplicated region that spans the WQPWS motif.
preactivated but not resting NK and T cells. IL-12 was origi- IL-23 appears to play a vital role in the maintenance, rather
nally discovered as NK cell stimulatory factor.312 IL-12 can than in the initial differentiation of Th17 cells (IL-17–pro-
also induce the production of IL-2, IL-3, GM-CSF, IL-9, TNF- ducing cells), and IL-23 is essential for resistance to certain
α, and M-CSF, although inducing IFNγ is perhaps its most diseases, such as EAE, which were originally believed to be
important known action.311,313,314 As is discussed in the sec- Th1-mediated diseases, as elimination of IL-23 by eliminat-
tion on immunodeficiency diseases, IL-12 is essential for the ing either the p40 subunit shared by IL-12 and IL-23 or by
proper clearing of mycobacterial infections. It is interesting eliminating p19 confers resistance to EAE.175,321 Whereas
that Th2 cells do not respond to IL-12; the lack of responsive- IL-23 promotes Th17 cell numbers, IL-27 can negatively reg-
ness of these cells to IL-12 results from their loss of expression ulate their development.321 In the CD40-mediated model of
of the IL-12Rβ2 subunit of the IL-12 receptor.315 Apparently, inflammatory bowel disease, IL-12 p35 secretion controlled
IL-4 inhibits IL-12Rβ2 expression, whereas IL-274 and IFNγ315 wasting disease and serum cytokine production but did
induce its expression. The abilities of mice to survive infec- not affect mucosal pathology, which instead was associated
tions is critically linked to the Th patterns of cytokines. For with IL-23/p19.322 Interestingly, IL-23R has been identified
example, the ability to survive T. gondii infection is depen- as an inflammatory bowel disease gene,323 and IL-23 can
dent on IFNγ /IL-12 production (a Th1 pattern).83 drive intestinal inflammation via direct actions on T cells.324
IL-12 can be thought of as having vital roles in both in- Separate from Th17 cells, a population of innate lymphoid
nate immunity and acquired immune responses. It is rapidly cells that produce IL-17, IL-22, and IFNγ in response to
produced by NK cells and then T cells in response to an- IL-23 have been identified as the mediators of inflammatory
tigens or foreign pathogens. This rapid response facilitates bowel disease.325 IL-23 and its receptor are nevertheless criti-
the activation of first-line defense against infections. In ad- cal for the terminal differentiation of Th17 cells.169,326 IL-27
dition, however, IL-12 is also required for the subsequent is a related cytokine (discussed previously), but it can act
differentiation of specialized T-cell populations, includ- as an inhibitor of Th1 responses associated with intracel-
ing its STAT4-dependent priming of Th1 cells for optimal lular infections such as T. gondii. Whereas host defense to
production of IFNγ and IL-2. IL-12 can act synergistically T. gondii is dependent on IL-12 and IFNγ, IL-27–deficient
with hematopoietic growth factors, such as IL-3 and SCF, to mice control the parasite replication but develop a lethal
support the proliferation and survival of hematpoietic stem inflammatory disease that is dependent on CD4 + T cells.
cells.311 Structurally, IL-12 is a covalently linked dimer of 35 Additionally, IL-27 can inhibit Th2 actions.303
(IL-12A) and 40 (IL-12B) kDa peptides311; thus, successful IL-35 is produced by Treg cells and plays a role in im-
production of IL-12 requires that a cell transcribe both the munosuppression. It is a heterodimer of IL-12A (p35) and
p35 and p40 genes.316 Interestingly, whereas p35 bears se- IL-27B (EBI3).327 IL-35 treatment of naive mouse or human
quence similarity to IL-6 and G-CSF, p40 is homologous to T cells induces regulatory cells that have been called iTR35
the extracellular domains of IL-6Rα , CNTFRα , and G-CSF cells328 and promote tolerance and tumor progression.

Interleukin-31 Two Types of Interleukin-4 Receptors, One of which also

IL-31 is a four α-helical bundle cytokine that fits into the Responds to Interleukin-13
greater IL-6 extended family in that its receptor consists of As detailed previously, on T cells, IL-4 acts through a recep-
IL-31RA and the OSMRβ.329 IL-31R is a gp130-like type 1 re- tor comprising IL-4Rα and γc (the type I IL-4 receptor)38,39;
ceptor, with four splice variants. IL-31 is produced primarily however, IL-4 can also signal through type II IL-4 receptors
by activated T cells and when overexpressed results in pru- comprising IL-4Rα and IL-13Rα1 but not γc.356 IL-13 is another
ritis, alopecia, and skin lesions, indicating a potential role in cytokine that shares some actions with IL-4 and was originally
dermatitis.329,330 described as a T-cell–derived cytokine capable of inhibiting in-
flammation,357 although it is now clear that other cells such as
NK cells and mast cells can also produce IL-13. IL-13 can induce
Other Examples of Shared Receptor Molecules identical signals to IL-4 on non-T cells that respond to IL-4 but
Interleukin-7 and Thymic Stromal Lymphopoietin Share lacks effects on mature T cells, as these cells do not bind IL-
Interleukin-7 Receptor α Chain 13.358,359 The shared actions of IL-4 and IL-13 include the ability
In addition to IL-7, a second stromal factor, TSLP has been to 1) decrease expression of inflammatory cytokines, 2) induce
identified that shares at least some actions with IL-7.11,331–333 major histocompatibility complex class II expression, 3) induce
The TSLP receptor is a heterodimer of TSLPR and IL-7Rα.334,335 CD23 expression and IgE production by B cells in humans, 4)
Interestingly, TSLPR is 24% identical to γc, making it the inhibit IL-2–induced proliferation of chronic lymphocytic leu-
cytokine receptor most like γc.334 Human and mouse TSLP kemia cells of B-cell origin, and 5) costimulate with anti-CD40
share only 43% amino acid identity, and human and mouse antibodies. This is explained by the observation that the type II
TSLP receptors share only 39% identity.336 This is very low IL-4 receptor consists of IL-4Rα plus IL-13Rα1,360,361 with both
for human and mouse orthologues of cytokine receptors, for IL-4 and IL-13 inducing indistinguishable signals on cells ex-
example, human and mouse γc are 70% identical.337 In addi- pressing this receptor. Interestingly, IL-4 binds primarily to IL-
tion to their wide sequence divergence, mouse and human 4Rα and IL-13 binds primarily to IL-13Rα1. This situation may
TSLP were originally suggested to substantially differ in be analogous to the situation for LIF, CT-1, and OSM, which
their actions. Mouse TSLP was first reported to be a B-cell can all act through receptors containing LIFRβ and gp130 but
differentiation factor that is important for the development differ in their abilities to directly interact with each of these
of IgM + immature B cells from pre-B cells and to also be a receptor proteins. An additional IL-13 binding protein, IL-
weak thymic comitogen. In contrast, human TSLP appeared 13Rα2, has much higher binding affinity for IL-13 than does
not to exert effects on these lineages but was found to be an IL-13Rα1.362 IL-13Rα2 is nonfunctional in terms of signaling
epithelial-derived cytokine that potently activated DCs re- and instead acts as a “decoy” receptor.
lated to Th2 allergic responses, an action not then known to Although IL-13 was originally believed to be substantially
be shared by mouse TSLP.338–340 However, it is now clear that redundant with IL-4, it has important distinctive actions
mouse TSLP can also activate DCs and that it plays a critical as well. IL-13 may be vital in asthma, as blocking IL-13 can
role in both atopic disease and asthma in mouse models, and inhibit pathophysiologic changes of asthma.363,364 It also is
that these effects of TSLP are mediated via actions on both clear that IL-13 regulates eosinophilic infi ltration, airway
DCs and CD4 + T cells.341–347 TSLP can also promote immu- hyperresponsiveness, and mucus secretion, and that it is a
nity to helminth parasites,333 and TSLP is also important potent mediator of fibrosis.357 Moreover, IL-13 that is pro-
for mucosal healing in the dextran sulfate sodium–induced duced by innate natural helper cells in the lung contributes
colitis model.348 Interestingly, human TSLP is expressed by to asthma.365 The phenotype of IL-13 knockout mice sug-
Hassall corpuscles349 and was suggested to promote selection gested a role for IL-13 the ability to expel helminths,366 and
of Treg cells in human thymus, thus contributing to central it is clear that IL-13 specifically can modulate resistance to
tolerance.350 It is striking, however, that mice lacking TSLP intracellular organisms, including Leishmania major and
receptor have normal numbers of Treg cells, indicating that Listeria monocytogenes.357 IL-13 has also been identified as a
TSLP is not absolutely essential for this function, and that factor that is secreted by and can stimulate the growth of
other molecules can at least substitute for TSLP. Indeed, Hodgkin and Reed-Sternberg cells.367
IL-7 is such a molecule.350a TSLP is now known to be pro- Although IL-4 and IL-13 are both critical for host defense
duced by a range of cell types, including epithelial cells, fi- to helminths, the use of reporter mice revealed that Th2
broblasts, keratinocytes, stromal cells, basophils, mast cells, cells produce both cytokines. Tfh cells in lymph nodes as
and DCs,11 and to in turn act on multiple lineages, including well as basophils make IL-4, whereas innate helper type 2
DCs, CD4 + and CD8 + T cells, mast cells, NKT cells, eosino- cells secrete IL-13. Thus, although both are produced during
phils, and B cells.11,333,351 Interestingly, TSLP promotes Th2 helminth infection, there are distinctive sites of production,
differentiation and has been shown to promote metastasis in correlating with distinctive roles.368
breast cancer and pancreatic cancer.352–354 From a signaling
perspective, whereas IL-7 activates JAK1 and JAK3, TSLP ac-
tivates JAK1 and JAK2, making it the only type 1 cytokine An Example of Multiple Affinities of Binding for a Single
to use this combination of JAK kinases to activate STAT5.347 Cytokine: Three Classes of Interleukin-2 Receptors
Although IFNγ also uses JAK1 and JAK2, it instead activates Although cytokines typically signal via a single class of high-
STAT1.355 affinity cell surface receptor, more complex situations exist.

TABLE 25.8 Classes of Interleukin-2 Receptors

Affinity Kd Where Expressed Composition Functional
Low 10−8 M Activated cells IL-2Rα No
Intermediate 10−9 M Resting cells IL-2Rβ and γc Yes
High 10−11M Activated cells IL-2Rα, IL-2Rβ, and γc Yes
IL, interleukin
The IL-2 system provides the very interesting illustration of a subpopulation of double negative thymocytes, but it remains
system with three classes of affinities of receptors (Table 25.8). unclear if IL-2 physiologically acts on these cells. Each of the
In addition to the high-affinity receptor (IL-2Rα + IL-2Rβ + components of the IL-2 receptor is located on a different chro-
γc, Kd ≈ 10−11 M), there are both low-affinity (IL-2Rα alone, Kd mosome: Human IL-2Rα is located on chromosome 10p14 to
≈ 10−8 M) and intermediate-affinity (IL-2Rβ + γc, Kd ≈ 10−9 M) 10p15,372 IL-2Rβ is at 22q,373,374 and γc is at Xq13.1,197 while the
receptors.29,50 Low- and high-affinity receptors are expressed mouse homologues are located at chromosomes 2, 15, and X.
on activated lymphocytes, whereas intermediate-affinity re-
ceptors are found on resting lymphocytes, particularly on
NK cells. Both intermediate- and high-affinity receptors Erythropoietin, Thrombopoietin, and Stem Cell Factor
can signal, suggesting that IL-2Rβ and γc are necessary and Epo was the first cytokine that was biochemically purified
sufficient for signaling, in keeping with the theme of dimer- and is vital for erythropoiesis, whereas thrombopoietin is
ization indicated previously. Given that the intermediate- critical for thrombopoiesis. These cytokines each bind to re-
affinity form is functional, what then is the rationale for ceptors that are homodimers.21,375 Interestingly, Epo signaling
having a high-affinity IL-2 receptor that also contains IL- may depend in part on the functional cooperation of the Epo
2Rα ? This is a particularly important question in view of receptor and c-kit, the receptor for SCF.376 This latter receptor
the fact IL-2Rα has an extremely short cytoplasmic domain has intrinsic tyrosine kinase activity and is not a type I cyto-
that does not appear to play a role in signaling. The impor- kine receptor. Epo was the first cytokine with demonstrated
tance of IL-2Rα is clearly demonstrated by the severely ab- clinical efficacy, for example, in renal insufficiency. Although
normal phenotype of IL-2Rα –deficient mice, which exhibit it is dominantly associated with erythrocyte production, Epo
autoimmunity, inflammatory bowel disease, and premature receptors have been found more broadly on immune cells.
death,217 and by the observation that IL-2Rα mutations can Recently, Epo has been shown to have effects on a number
cause an autoimmune syndrome in humans as well.369 A clue of cell types, blocking NF-κB p65 and resulting in inhibi-
to the importance of IL-2Rα comes from the kinetics of asso- tion of the induction of TNF-α and inducible nitric oxide
ciation of IL-2 with each chain. Although the IL-2Rα appears synthase in activated macrophages, with diminished control
to lack a direct signaling function, it has a very fast on-rate of Salmonella infection, whereas in a model of chemical-in-
for IL-2 binding.370 Thus, the combination of this rapid on- duced colitis, such inhibition of NF-κB–dependent immune
rate with the slow off-rate from IL-2Rβ /γc dimers results in modulators limits tissue damage and disease severity.377 It will
high-affinity binding that is vital for responding to the very therefore be important to determine the full scope of immu-
low concentrations of IL-2 that are physiologically present in nologic actions that are mediated by Epo.
vivo. Moreover, as activated T cells express approximately 10
times as many low-affinity than high-affinity receptors, IL-
2Rα may serve as an efficient means of recruitment and con- CYTOKINE AND CYTOKINE RECEPTOR PLEIOTROPY
centration of IL-2 on the cell surface, allowing more efficient AND REDUNDANCY
formation of IL-2/IL-2Rβ /γc signaling complexes. It is well recognized that many cytokines exhibit the phenom-
As noted previously, IL-2Rα is not a type I cytokine re- ena of cytokine “pleiotropy” and “redundancy.”304 Cytokine
ceptor and in fact has homology to the recognition domain pleiotropy refers to the ability of a cytokine to exert many dif-
of complement factor B.371 Subsequently, the IL-15 receptor α ferent types of responses, often on different cell types, whereas
chain was shown to also have a similar structure to IL-2Rα,155 cytokine redundancy refers to the fact that many different
with both IL-2Rα and IL-15Rα having what are called “sushi” cytokines can induce similar actions. One set of cytokines
domains, which contribute to ligand binding. The fact that that exhibit cytokine pleiotropy is the γc family of cytokines.
IL-2 and IL-15 both have related α chains is consistent with For example, IL-2 can induce T-cell growth, augment B-cell
the close relationship between IL-2 and IL-15, and that the Ig synthesis, increase the cytolytic activity of lymphokine-
receptors for both IL-2 and IL-15 contain both IL-2Rβ and γc. activated killer and NK cells, and play an essential role in me-
As IL-2Rα cannot transduce a signal by itself, the detection diating activation-induced cell death; IL-4 can induce B-cell
of IL-2Rα on the cell surface does not necessarily reflect IL-2 growth and Ig class switch; and IL-7 not only plays a major
responsiveness. Because IL-2Rα was discovered before IL-2Rβ role in thymocyte development but also can stimulate ma-
and γc, many papers in the literature have evaluated IL-2 recep- ture T cells, and at least in the mouse, can act as a pre–B-cell
tor expression based on IL-2Rα expression alone, and this may growth factor. The gp130 family cytokines also exhibit broad
not reflect IL-2 responsiveness. IL-2Rα is also expressed on a actions. For example, IL-6 exerts effects ranging from that of

a comitogen for thymocyte activation to that of a mediator of IL-13Rα2, GM-CSF, type I and type II IFNs, IL-22, and
the acute-phase response in liver. Regarding cytokine redun- TNF.381–383 As is clear from this list of cytokines, soluble re-
dancy, it has already been highlighted that IL-2, IL-4, IL-7, ceptors are not restricted to receptors for type I cytokines,
and IL-15 receptors contain γc can act as T-cell growth factors, and in the case of IL-2, the principal soluble receptor protein
and that IL-3 has actions that overlap with IL-5 and GM-CSF. is IL-2Rα , which is not a type I cytokine receptor. Soluble
The recognition that cytokines not only have overlapping receptors can be created by alternative splicing that trun-
actions but also share receptor components, led to the con- cates the protein N-terminal to the transmembrane domain,
cepts of “cytokine receptor pleiotropy” and “cytokine recep- resulting in a secreted protein rather than a membrane-an-
tor redundancy.”304 The first of these terms can be defined by chored membrane in the case of IL-4Rα , IL-5Rα , IL-6Rα ,
the ability of a single cytokine receptor subunit to function IL-7Rα , IFNAR-2, and GM-CSFRα. Alternatively, they can
in more than one receptor. Thus, examples include the shar- be created by proteolytic cleavage of the membrane recep-
ing of γc, βc, gp130, IL-4Rα , LIFRβ, and OSMRβ, as summa- tor as is found for the receptors for IL-2Rα and TNFRI and
rized previously, as well as the sharing of IL-2Rβ by IL-2 and TNFRII382 (Table 25.9). Although it is conceivable that a dis-
IL-15 receptors, the sharing of IL-4Rα and IL-13Rα in type tinct gene might encode the soluble forms of a receptor, no
II IL-4 receptors and IL-13 receptors, and the sharing of IL- examples have been reported. In the cases where proteolytic
7Rα by the IL-7 and TSLP receptors. Another way of viewing cleavage occurs, the identity of the proteases has not been
receptor pleiotropy is that certain receptor chains are useful identified. The major questions related to these soluble re-
“modules” that function in more than one context. ceptors are as follows: 1) Do they have physiologic or patho-
The final term, cytokine receptor subunit redundancy, is physiologic functions? 2) How do their affinities compare
the one with fewest examples. There is one well-documented to the corresponding cell surface receptor? 3) Do they have
example in mice but not in humans. IL-3 signals through IL- diagnostic, prognostic, and therapeutic applications?
3Rα plus either βc or an alternative unique IL-3Rβ that shares Unfortunately, there is little information available on the
91% amino acid identity with βc and appears to be a com- in vivo role of soluble receptors. In general, when analyzed
pletely functionally redundant protein for IL-3 signaling, but using in vitro studies, soluble receptors can compete with
IL-3Rβ cannot substitute for βc in the context of IL-5 or GM- their corresponding cell surface receptors, thereby serving
CSF signaling.239 Other potential examples exist. For example, negative regulatory roles. For example, this appears to be the
in type I and type II IL-4 receptors, IL-4Rα coordinates with case for the IL-22 soluble receptor. However, soluble IL-6Rα
either γc or IL-13Rα1, respectively, and there are two types exerts an agonistic role because, as summarized previously,
of OSM receptors, both of which contain gp130, but one of IL-6 signaling occurs equally well via gp130 when the soluble
which contains a specific OSM receptor while the other con- rather than transmembrane form of IL-6Rα interacts with
tains LIFRβ. What remains unknown, however, is whether IL-6. Nevertheless, a mutated form of IL-6Rα that cannot
the signals mediated by these different types of receptors are interact with gp130 but still binds IL-6 can effectively in-
truly identical so that there is redundancy, or whether there hibit the actions of IL-6.384 In the case of IL-2Rα , there is no
are distinctive features to the signals that IL-4 and OSM reported physiologic function for soluble IL-2Rα , as the af-
induce via the different receptors. Indeed, mice lacking IL- finity of the released receptor is, as expected, similar to that
13Rα1 have exacerbated Th2 responses and is essential for of the low-affinity receptor (Kd ≈ 10−8 M), making it unlikely
allergen-induced airway hyperreactivity, suggesting that type to effectively compete with the high-affinity cell surface re-
2 IL-4 receptor signaling (either via IL-4 or IL-13) is not iden- ceptor (Kd ≈ 10−11 M). However, this and other soluble recep-
tical to that mediated by the type 1 IL-4 receptor (via IL-4).378 tors could serve as cytokine carrier proteins and potentially
In addition to these examples related to type I cytokines, could increase stability of a cytokine by protecting it from
the IL-10 subfamily of type II cytokines is interesting in that proteolysis.382 This theoretically could be similar to the role
IL-10Rβ (IL-10R1), IL-20Rα (IL-10R1), IL-20Rβ (IL-20R2), of IL-2/anti–IL-2 conjugates in exhibiting higher activity.385
and IL-22Rα (IL-22R) are each shared receptor components, Moreover, there are potential diagnostic and prognostic uses
collectively affecting signaling in response to IL-10, IL-19, for measuring the level of shed receptors (Table 25.10).
IL-20, IL-22, and IL-24.304,379 Specifically, IL-10 signals via a
receptor containing IL-10R1 and IL-10R2, IL-19 signals via a
receptor containing IL-20R1 and IL-20R2, IL-20 signals via
receptors containing IL-20R2 and either IL-20R1 or IL-22R, TABLE 25.9 Soluble Cytokine Receptors
IL-22 signals via receptors containing IL-22R and IL-10R2,
and IL-24 signals through receptors containing IL-20R2 Generated by Alternative Generated by Proteolytic
Splicing Cleavage of Mature Receptor
plus either IL-20R1 or IL-22R. Two other IL-10 family mem-
bers, IL-28 and IL-29, both signal via IL-28R + IL-10R2.380 sIL-4Rα sIL-2Rα
Interestingly, IL-10 is believed to be a dimer; IL-26 may also sIL-5Rα sTNFR
be a dimer, but the other family members are monomers.379 sIL-6Rα
sIL-7Rα
sGM-CSFRα
SOLUBLE RECEPTORS sIFNAR-2
Soluble forms of many cytokine receptors have been iden- GM-CSF, granulocyte macrophage-colony stimulating factor; IL, interleukin; TNFR,
tified, including those for IL-1, IL-2, IL-4, IL-5, IL-6, IL-7, tumor necrosis factor receptor.

Soluble Interleukin-2 Receptors in TABLE 25.11 Type II Cytokines

TABLE 25.10
Human Disease
Chromosomal
Malignancies Location (h/m)
Hematologic
Type I IFNs
Adult T-cell leukemia
IFNα Many genes 9p22/4
Hairy cell leukemia
IFNβ Single gene 9p21/4
Acute lymphocytic leukemia
IFNω Single gene 9p21/4
Chronic lymphocytic leukemia (B cell)
IFNτ Many genes
Acute myelogenous leukemia
Chronic myelogenous leukemia (especially in blast crisis) Type II IFN
Malignant lymphomas IFNγ Single gene 12q14/10
Hodgkin disease
Non-Hodgkin lymphomas IL-10 family cytokines
Nonhematologic IL-10 Single gene 1q31–32/1
Adenocarcinoma of lung, breast, pancreas IL-19 Single gene 1q32/1
Small cell bronchogenic carcinoma IL-20 Single gene 1q32/1
Ovarian, cervical, and endometrial cancers IL-22 Single gene 12q14–15/10
Nasopharyngeal carcinoma IL-24 Single gene 1q32/1
Melanoma IL-26 Single gene 12q14–15/
IL-28 Single gene 19q13/7
Infections IL-29 Single gene 19q13
Human immunodeficiency virus
Tuberculosis IFN, interferon; IL, interleukin.
Rubeola
Infectious mononucleosis
merly designated as an IFNα. There are multiple (at least 12)
Other diseases IFNαs in mice and humans. In contrast, the other type I IFNs,
End-stage renal disease IFNβ and IFNω, are each encoded by single human and mouse
Rheumatoid arthritis genes near the IFNα cluster, and in addition, there are mul-
Systemic lupus erythematosus
tiple pseudogenes most closely related to IFNα and IFNω. The
Scleroderma
type I IFNs are clustered on human chromosome 9 and mouse
Sarcoidosis
chromosome 4. Type II interferon is IFNγ,314 which is encoded
After transplantation by a single gene on human chromosome 12 and mouse chro-
mosome 10. IFNα /β is produced after viral infection in many
After interleukin-2 administration cell types, although plasmacytoid DCs appear to be the largest
In adults, the mean sIL-2Rα levels are 280 ± 161 units/mL producers of IFNα /β. These type I IFNs inhibit viral replica-
(levels tend to be higher in pediatric populations). The
situations where the levels exceed 5000 units/mL are adult
tion and induce the apoptosis of virally infected cells. In addi-
T-cell leukemia, hairy cell leukemia, chronic myelogenous tion, IFNα /β mediate the activation of macrophages and NK
leukemia, and after interleukin-2 administration. The situ- cells, and they additionally affect the proliferation and survival
ations where levels are between 1000 and 5000 units/mL of CD8 + T cells.
include acute myelogenous leukemia, chronic lymphocytic The grouping of the IFNαs and IFNβ together as type I
leukemia, non-Hodgkin lymphomas, acquired immunodefi- IFNs is logical not only because of the similar amino acid se-
ciency syndrome associated with Kaposi sarcoma, tuber- quences and structures of these IFNs but also based on the fact
culosis, rubeola, and end-stage renal disease. Data are
from Kurman et al.381 and Fernandex-Botran et al.382 that they share the same receptor and induce essentially the
same signals.397,398 Although DNA array analysis does show
some differences in the genes induced by IFNα and IFNβ,
the basis for these differences are unclear.399 These signals
include not only antiproliferative and antiviral activities but
INTERFERONS (TYPE II CYTOKINES) AND THEIR also the ability to stimulate cytolytic activity in lymphoyctes,
RECEPTORS NK cells, and macrophages. In contrast, IFNγ has a distinct
IFNs represent an evolutionarily conserved family (Table 25.11) receptor. Type I and type II IFN receptors share a sufficient
of cytokines that are related to the IL-10 family of type II cyto- degree of similarity to each other so as to form a family.400
kines. IFNs were discovered in 1957 on the basis of their antivi- The structure of the IFNγ receptor401 is shown in Figure 25.2.
ral activity and were the first cytokines that were discovered.386 IFN receptors are referred to as type II cytokine receptors,
IFNs are known as either type I or type II IFNs,387–396 where and there are substantial differences between these receptors
type I interferons include IFNα (originally known as leuko- and the type I cytokine receptors.8 Because both type I and
cyte IFN) and IFNβ (originally known as fibroblast IFN), and type II IFNs bind to type II cytokine receptors, IFNs are oc-
IFNε, IFNκ, IFNω, IFNδ, and IFNτ. IFNδ and IFNτ are absent casionally referred to as type II cytokines but more gener-
in humans. IFNω is closely related to the IFNαs and was for- ally are referred to as IFNs. Based on the similarity of the

IL-10 receptor to the IFNγ receptors,8 IL-10 was designated cells as well as Th17 cells. IFNγ is produced by a range of
as a type II cytokine, and indeed when its x-ray crystal was cells, including Treg cells and in this context appears to re-
determined, IL-10 was found to be topologically related to sult in apoptosis of naive cells and Th2 effector T cells. It is
IFNγ.16 As noted previously, more recently a series of IL-10– produced by NK cells and can mediate NK-dependent lysis
related cytokines has been identified, including IL-19, IL-20, of tumor cells, augment IgG2a isotype switching of B cells,
IL-22, IL-24, IL-26, IL-28A, IL-28B, and IL-29, and these are and activate DCs. Moreover, IFNγ can either contribute to
also designated as type II cytokines, with IL-28A, IL28B, and disease protection or can be protective, depending on the
IL-29 also known as IFNλs. Among type II cytokines, IFNγ context.413 Finally, IFNγ- or anti-IFNγ –based therapy are
has helices similar to those of the type I short-chain helical being tried in a range of conditions.314
cytokines, but its short helices that occupy the AB and CD IFNγ is encoded by four exons on chromosome 12. IFNγ
loops positions exhibit the long-chain cytokine-like AB over forms a functional homodimer with an apparent molecu-
CD topology. IL-10 and IFNαβ have long-chain structures.8 lar weight of 34 kDa, whereas little of the monomeric form
Thus, the theme of short-chain and long-chain type I cyto- can be detected and it is not biologically active. Each IFNγ
kines also extends to the IFNs and the IL-10 family of type monomer has six α helices, four of which resemble the short-
II cytokines. chain helical cytokines, and there is no β sheet structure.
Type I IFNs signal through a receptor known as the type The subunits interact in an antiparallel fashion. In con-
I IFN receptor.388,392,394,396,397 The receptor consists of at least trast to the ability of many different cells to produce IFNα ,
two chains.402–405 In contrast to the α and β chain nomen- IFNγ is more restricted, with it being produced primarily
clature typical for type I cytokine receptors, IFN receptor by NK cells, CD8 + T cells, and the Th1 subclass of CD4 +
chains are officially denoted as IFNAR-1 (previously also de- T cells.388 Many signals, including antigen stimulation, IL-
noted IFN-αR1, IFNAR1, and IFN-Rα) and IFNAR-2 (pre- 12, and IL-18, can induce the production of IFNγ. IFNγ ex-
viously also known as IFNα / β receptor, IFNαR2, IFNAR2, erts its effects through specific receptors that are expressed
and IFNRβ).406 IFNAR-2 has both short and long forms as on all types of cells except erythrocytes. Interestingly,
well as a soluble form.407 The long form has a much larger even platelets express IFNγ receptors, raising the possibil-
cytoplasmic domain and serves a more important role in ity that they can serve a function in transporting IFNγ in
signal transduction. Whereas IFNAR-1 cannot bind IFNα , the circulation.388 The functional human receptor consists
IFNAR-2 binds with low affinity, and the combination of of two chains414 : IFNGR-1, formerly also denoted IFNγR1
both chains results in high-affinity binding407 and function. or IFNγRα406,415 a 90 kDa protein whose gene is located on
As detailed in the following, IFNAR-1 binds the Janus fam- human chromosome 6q16 to 6q22 and mouse chromosome
ily tyrosine kinase TYK2, whereas IFNAR-2 binds JAK1. In 10,412 and IFNGR-2, also denoted as IFNγRβ,416,417 located on
addition to these cellular receptors, it is interesting that vac- human chromosome 21q22.1 and mouse chromosome 16.412
cinia virus and other orthopoxviruses encode a soluble form IFNGR-1 is required for ligand binding, whereas IFNGR-2
of type I IFN receptor that is related to the IL-1 receptors plays a role in signaling. JAK1 associates with the Leu-Pro-
and that is capable of binding IFNα , IFNβ, and IFNω.408,409 Lys-Ser sequence in the membrane proximal region of the
This form of IFN receptor is therefore not a member of the cytoplasmic domain of IFNGR-1,418 whereas JAK2 binds to
type II cytokine family but instead is an Ig superfamily pro- IFNGR-2.419 The fact that IFNγ is a homodimer explains
tein. IFNα s are typically species specific, although IFNα8 how its binding induces the homodimerization of IFNGR-1,
is a human type I IFN that can bind to the mouse receptor. which then allows the recruitment of IFNGR-2. Thus, the
IFNAR-1 confers species specificity of binding. As noted pre- functional IFNγ receptor is believed to contain two mole-
viously, type I IFNs are induced by viral infection. Infection cules each of IFNGR-1 and IFNGR-2. Normal IFNγ produc-
with viruses results in the generation of nonself RNAs in the tion is dependent on IL-12, and defective IFNγ signaling is
cell, which are recognized by retinoic acid inducible gene- associated with failure to appropriately clear mycobacterial
I (RIG-I)–like receptors, which include RIG-I, MDA5, and and other infections (discussed subsequently).
LGP2, each of which contains a helicase domain. Retinoic Both type I and type II IFNs are important for cancer im-
acid inducible gene-I–like receptors are essential for the pro- munoediting, a process in which neoplastic growth is sup-
duction of tpe I interferons as well as IL-6, and cells lacking pressed by the immune system and tumor immunogenicity
both RIG-I and MDA5 are completely defective in the pro- is shaped. IFNγ can suppress tumor development, but type I
duction of type I IFNs.410 IFNs are required to initiate the antitumor response, at least
IFNγ was first recognized in 1965 as immune IFN, named in part via actions on DCs.420
as IFNγ in 1980, cloned in 1982, and then confirmed to be
the same as primary macrophage activating factor.388,397,411–413
In addition to its antiviral activity, IFNγ also has a range of Interleukin-10, a Type II Cytokine, and the
actions on proinflammatory and anti-inflammatory host Related Cytokines Interleukin-19, Interleukin-20,
responses.393 For example, IFNγ can prime macrophages to Interleukin-22, Interleukin-24, Interleukin-26,
manifest antimicrobial and antitumor effects. Moreover, Interleukin-28A, Interleukin-28B, and Interleukin-29
following IL-12–mediated differentiation of Th1 cells, IFNγ IL-10 is a type II cytokine that originally was identified as
is a major secreted product. This production is critical for cytokine synthesis inhibitory factor,380,421,422 produced by
host response to a range of infectious pathogens, including, Th2 cells. However, it then was observed to be produced
for example, T. gondii. IFNγ inhibits the generation of Th2 by Treg cells as well, and it is now clear that IL-10 is much

more broadly expressed, including by Th1, Th17 cells, B is believed to have a role in psoriasis, and in psoriatic skin,
cells, macrophages, NKT, NK, Treg, monocytes, and my- expression of IL-20 is primarily detected in DCs.379,438 IL-20
eloid DCs but not plasmacytoid DCs, keratinocytes, as well also appears to contribute to rheumatoid arthritis, athero-
as mast cells and eosinophils.423,424 IL-10 production by Treg sclerosis, and stroke; recently, IL-20 has been implicated
cells helps to explain their suppressive effects, and IL-10 in bone loss, with elevated IL-20 in patients with osteope-
production by Treg cells is required for their suppression of nia and osteoporosis in in ovariectomized mice. Moreover,
Th17-mediated inflammation.425 The IL-10 gene contains IL-20R1–deficient mice have increased bone density as did
five exons and is located on chromosome 1 in both mice and treatment with anti–IL-20.439
humans.422 IL-10 has an open reading frame of 178 amino IL-22 was found as an IL-9–induced cytokine that can in
acids, including the signal peptide, and the mature protein turn induce acute-phase reactant production by hepatocytes
is 18 kDa. Human IL-10 receptor maps to 11q23.3. IL-10 can and accordingly that it might be involved in inflammatory
inhibit the production of a number of cytokines, including responses. It was first observed to be produced by activated
IL-2, IL-3, IFNγ, GM-CSF, and TNF-α. Production of IL-10 Th1 cells but was subsequently noted to be more broadly pro-
is substantially dependent on the transcription factor E4BP4 duced by CD4 + and CD8 + T cells, NK cells, lymphoid tissue-
(also called NFIL-3).426 IL-10 production is induced by inducer cells and mucosal RORγ t+NKp46 + cells in humans
pathogen-derived products, with TLR2 signaling being crit- and mice,440–442 and to be produced by Th17 cells. IL-22 is
ical for its induction by antigen-presenting cells and macro- often coexpressed with IL-10, and STAT3, the Notch pathway,
phages, although TLR-independent induction of IL-10 has and the aryl hydrocarbon receptor promoter production of
also been observed.423 Interestingly, IL-21 has been shown to these cytokines; in contrast, TGF-β, ICOS, and IL-27 aug-
be a potent inducer of IL-10 and to mediate IL-27 and part of ment IL-10 production yet inhibit IL-22.424 IL-22 mediates
the IL-6–mediated induction of IL-10 as well.189,190 IL-10 in- cross talk between leukocytes and tissue epithelia, with tar-
hibits monocyte/macrophage/DC-dependent T-cell prolif- get cells primarily in the digestive and respiratory systems,
eration, in part, by markedly decreasing synthesis of a vari- but it also can act on keratinocytes, can promote innate im-
ety of cytokines, expression of costimulatory molecules, and munity,443 and can cooperate with IL-17A to effect epithelial
chemokines, to name just some of its inhibitory effects.380 and healing responses. In particular, IL-22 can contribute to
In addition to these indirect effects on T cells, IL-10 can epithelial resistance to injury after microbial infection of the
exert direct stimulatory effects on thymocytes and T cells in gut or lungs, and thereby may play a role in host defense and
vitro and promote the development of B1 cells and activity tissue repair.442 Interestingly, the transcription factor c-Maf
of NK cells. IL-10–deficient mice develop a form of inflam- mediates suppression of IL-22 production by TGF-β.444
matory bowel disease that is similar to Crohn disease.422,427 IL-24 was originally discovered as melanoma differentia-
Interestingly, the BCRF1 protein that is encoded by Epstein- tion-associated antigen 7. IL-24 is induced in peripheral blood
Barr virus is very similar to IL-10 and shares many of its bio- mononuclear cells by mitogen stimulation and is produced
logic properties as a macrophage “deactivating” factor and not only by Th2 cells but also by myeloid cells. Overexpression
as a costimulator of proliferation of B cells.428 The Epstein- of IL-24 can induce apoptosis in a wide range of tumor cells,
Barr virus IL-10 homologue is a selective agonist, although including, for example, melanoma and malignant gliomas,
its binding to the IL-10 receptor is somewhat impaired.428 and infection with adenoviral-drive melanoma differentia-
IL-10 is a major inhibitor of Th1 functions.422 Although it tion-associated antigen 7 can sensitize tumors to ionizing ra-
was originally suggested that IL-10 might also favor Th2 de- diation.380 Like IL-19, IL-20, and IL-22, IL-24 is expressed in
velopment, IL-4 is the major mediator of Th2 cell develop- psoriatic but not normal skin and like these other cytokines,
ment, and IL-10 instead plays a major role in limiting and functional receptors are highly expressed on keratinocytes.
terminating inflammatory responses.422 IL-10, IL-19, IL-20, IL-22, and IL-24 are all increased in sy-
The IL-10 receptor is most closely related to IFN recep- novial fluid cells in rheumatoid arthritis and thus are poten-
tors, making it a type II cytokine receptor429,430 and cor- tial mediators of this autoimmune disease. IL-20 and IL-22 in
responding to the close structural relationship of IL-10 to particular may be associated with lupus.379
IFNγ. The receptor for IL-10 consists of an IL-10R1 chain IL-26 was originally discovered as AK155. Relatively little
and IL-10R2422 ; this latter protein is the same as CRF2-4, is known regarding the function of IL-26.
which was first identified as an “orphan” IFN receptor fam- IL-28A and IL-28B (IFNλ1 and IFNλ2) and IL-29
ily member that is located on chromosome 21 within 35 kb (IFNλ3) are also called type III IFNs and are induced in
of IFNGR-2.431,432 The major STAT protein used by IL-10 is plasmacytoid DCs by viral infection and exhibit antiviral
STAT3.423 activity, analogous to type I IFNs; nevertheless, they bind to
A series of IL-10–related cytokines have been been iden- a receptor consisting of IL-10R2 and IL-28R445 and differ in
tified, including IL-19, IL-20, IL-22, IL-24, IL-26, IL-28A, terms of potency and kinetics from IFNα .379 IL-28A, IL-28B,
IL-28B, and IL-29.379,424,433–437 IL-19 was discovered as a gene and IL-29 induce relatively weak antiviral responses, and
that was induced in LPS-stimulated monocytes. IL-19 has the IL-28R chain is relatively restricted in its expression, as
been suggested to potentially be involved in the pathogen- compared to the IFNα receptor. IL-28A appears to enhance
esis of chronic inflammatory diseases, such as psoriasis. antiviral activity induced by TLR3 and TLR9 and to be im-
Like IL-19, IL-20 is produced by myeloid cells; it was also portant for host defense to influenza A.379 Interestingly, the
found to be produced by epidermal cells, with overexpres- genes encoding IL-10, IL-19, IL-20, and IL-24 colocalize at
sion resulting in aberrant epidermal differentiation. IL-20 human chromosome 1q32, whereas those encoding IL-22

and IL-26 are at 12q14 to 12q15 near the IFNG gene, and Zap70 and related proteins, PI 3-kinase, IRS-1 and IRS-
IL-28 and IL-29 are at 19q13.380,446 Thus, these type II cyto- 2, and phosphatases. Each of these pathways will be dis-
kines can be subdivided into three different groups based at cussed in turn. The JAK-STAT pathways are important
least in part on chromosome localization. for both type I and IFNs/type II cytokines, while the IRF
Overall, IL-10 family cytokines are diverse and can be proteins play central functions more so for the IFNs than
subdivided into the IL-19, IL-20, IL-22, IL-24, and IL-26 subtype I cytokines.
family that acts on tissue epithelial cells and in the case of
IL-20 in wound healing, whereas IL-28A, IL-28A, and IL-29
exhibit antiviral activity and are more IFN-like in their OVERVIEW OF JAKS AND STATS
biology.379 The JAK-STAT pathway30,355,397,447–449 is particularly exciting
in that it serves as a rapid mechanism by which signals can
be transduced from the membrane to the nucleus. JAK ki-
SPECIES SPECIFICITY OF CYTOKINES nases are known as Janus family tyrosine kinases, and the
There are no general rules for the species specificity of acronym “STAT” denotes signal transducer and activator of
human and mouse cytokines, and how the cytokines and transcription. STAT proteins are substrates for JAK kinases.
their receptor chains have coevolved. To provide illustra- A tremendous amount of information is now available on
tive examples of each situation, human IL-2 can stimulate JAKs and STATs that demonstrate their importance related
both human and mouse cells, whereas mouse IL-2 exhibits to development, differentiation, proliferation, cellular trans-
little action on human cells.59 Conversely, human IL-12 does formation, and tumorigenesis.
not work on mouse cells, whereas mouse IL-12 is biologi-
cally active on both mouse and human cells.311 This selective
property of IL-12 is dependent on the species origin of p35. JAKs
Finally, IL-4 exhibits rather strict specificity so that human The JAK kinases are 116 to 140 kDa, and comprise ap-
and mouse IL-4 only induce responses on human and mouse proximately 1150 amino acids.355,397 The seven regions of
cells, respectively.92,93 As noted previously, IFNα s are gener- conserved sequences in JAK kinases, denoted JH1 to JH7,
ally specifies specific, although IFNα8 is not. In addition to are depicted in Figure 25.6. One of the hallmark features of
these examples, varying degrees of relative species specificity these kinases is that in addition to the presence of a catalytic
have been demonstrated, depending on the cytokine. Thus, tyrosine kinase domain (JH1), there is also a pseudokinase
a cytokine may or may not be restricted in its specificity to region (JH2). The name Janus kinase reflects the two faces
its own species, and if it does occur, cross-species activity of the mythological Roman god, with one face representing
may have attenuated potency. the true kinase and the other the pseudokinase. Although
the JH nomenclature was used historically, it has obvious
limitations in that except for the JH1 catalytic and JH2
SIGNALING THROUGH INTERFERON AND CYTOKINE pseudokinase domains, and it remains unclear whether or
RECEPTORS not the other JH regions correspond to discrete domains.
Our understanding of signaling through IFN and cyto- Modeling has indicated five important regions, including an
kine receptors has tremendously increased in the past few N-terminal domain, a FERM (4.1, ezrin, radixin, moesin)
years. Multiple signaling pathways/molecules have been domain, and SH2, kinase-like (which spans two of the JH
observed for various cytokines. Collectively, these in- domains), and kinase domains.450
clude the JAK-STAT pathway, IRF family proteins, Ras/ There are four mammalian JAK kinases, JAK1,451 JAK2,452
mitogen-activated protein (MAP) kinase pathway, Src and JAK3,453,454 and TYK2,455 each of which was identified as part
FIG. 25.6. Schematic of JAK Kinases. Shown are the locations of the seven JH domains. JH1 is the
catalytic domain. JH2 is the pseudokinase domain, the presence of which prompted the naming of
this family as Janus family tyrosine kinases. As noted in the text, the JH nomenclature has limitations
and in fact masks the presence of an SH2 domain that spans parts of the JH4 and JH3 domains. Also
shown is the conserved tyrosine (Y1007 in JAK2) whose phosphorylation is required for maximal
catalytic activity.

reasonable to assume that at least two JAK molecules (either

TABLE 25.12 Features of Janus Family Kinases two molecules of one JAK or one molecule each of two dif-
Chromosomal
ferent JAKs) will be activated, as one JAK will be associated
Inducible versus Location with each receptor chain. In accord with their ubiquitous
Kinase Constitutive Size (H/M) expression, JAK1, JAK2, and TYK2 are activated by a variety
of different sets of cytokines (see Tables 25.12 and 25.13).
JAK1 Constitutive 135 kDa 1p31.3/4
For example, JAK1 is activated not only by type I and II IFNs
JAK2 Constitutive 130 kDa 9p23–24/19
but also by the γc family of cytokines (eg, IL-2, IL-4, IL-7,
JAK3 Inducible 116 kDa 19p13/8
IL-9, IL-15, and IL-21), whereas JAK2 is activated not only
TYK2 Constitutive 140 kDa 19p13.2/9
by IFNγ but also by growth hormone, Epo, prolactin, and
the hematopoietic cytokines, IL-3, IL-5, and GM-CSF.355,447
TYK2 is somewhat more restricted in that it is activated by
of a study intended to identify new kinases.355 At least one IFNα /β and IL-12/IL-23; the significance of its activation
JAK kinase is activated by every IFN and cytokine, and by gp130 family cytokines is less clear. Interestingly, JAK1,
some cytokines activate two or three JAK kinases.355,397,447,448 JAK2, and TYK2 are recruited by each of the cytokine re-
Table 25.12 lists a number of features of each JAK kinase, ceptors that share gp130 as a signal transducing molecule,
whereas Table 25.13 summarizes the JAK kinases that are raising the question as to whether these cytokines require all
activated by a variety of cytokines. three JAK kinases for optimal function or whether any one
Because JAK1, JAK2, and TYK2 are ubiquitously ex- or two is/are sufficient. At least for IL-6, JAK1 is vital,463,464
pressed, each cell type expresses either three or in some whereas the importance of JAK2 and TYK2 is less clear.
cases all four JAK kinases (eg, lymphoid cells that also JAK3 is different from the other JAK kinases in that it is
express JAK3). The JAK kinases that are activated by a much more inducible. Moreover, JAK3 is only activated by
specific cytokine are those that can bind to its receptor’s cytokines whose receptors contain γc.30,45 Interestingly, each
cytoplasmic domain. JAK kinases physically bind to the cytokine whose receptor contains γc activates not only JAK3
membrane proximal Box 1/Box 2 region of the cytoplasmic but also JAK1. The basis for the activation of both JAK1 and
domains,454–459 and the N-terminal region of the JAK is re- JAK3 by IL-2, IL-4, IL-7, IL-9, IL-15, and IL-21 is that JAK1
quired for this function.460–462 In cytokine receptors, the Box associates with each of the distinctive signaling chains (IL-
1 region is proline rich,26 suggesting that JAK kinases may 2Rβ, IL-4Rα , IL-7Rα , IL-9Rα , and IL-21R), whereas JAK3
have SH3-like domains in their N-terminal regions to medi- associates with γc.30,45 Although this could be a coincidental
ate these interactions. As each cytokine or IFN receptor is result of coevolution of these cytokine systems, these obser-
a homodimer, heterodimer, or higher order oligomer, it is vations raise the possibility that JAK1 is the JAK kinase that
most efficiently functionally cooperates with JAK3.
Given the wide range of cytokines that activate any par-
ticular JAK kinase and that in some cases multiple cytokines
TABLE 25.13 Cytokines and the JAKs They Activate can activate the same set of JAKs, it is clear that the JAK
Cytokine JAK Kinase(s) Activated kinases by themselves do not determine signaling speci-
ficity. For example, IL-2, IL-4, IL-7, IL-9, IL-15, and IL-21
IFNα/β JAK1, TYK2 all activate JAK1 and JAK3 but induce a range of actions.
IFNγ JAK1, JAK2 Moreover, JAK2 is the only JAK kinase that is activated by
Growth hormone JAK2 growth hormone and erythropoietin, cytokines with differ-
Prolactin JAK2 ent target cells and biologic functions. Interestingly, JAK2 is
Erythopoietin JAK2 the JAK that interacts with all cytokine receptors that form
Thrombopoietin JAK2 homodimers. Given that homodimeric receptors are likely
IL-10 JAK1, TYK2 the oldest cytokine receptors in evolution, this suggests that
IL-12, IL-23 JAK2, TYK2
JAK2 might be the “primordial” JAK kinase from which
G-CSF JAK1, JAK2
others evolved.
γc family
IL-2, IL-4,a IL-7, IL-9, IL-15, IL-21 JAK1, JAK3
βc family Importance of JAK Kinases in Signaling
IL-3, IL-5, GM-CSF JAK2, ?JAK1 In addition to the activation of JAK kinases by multi-
gp130 family ples cytokines and IFNs, a variety of other data indicate
IL-6, IL-11, CNTF, LIF, OSM, CT-1 JAK1, JAK2, TYK2 their importance for signaling. One of the vital series of
TSLP JAK1, JAK2 experiments that led to the establishment of the critical
CNTF, ciliary neurotrophic factor; CT-1, cardiotrophin-1; G-CSF, granulocyte-colony role of JAK kinases in IFN signaling involved a group of
stimulating factor; GM-CSF, granulocyte macrophage-colony stimulating factor; mutant cell lines that were defective for IFN signaling, but
IFN, interferon; IL, interleukin; LIF, leukemia inhibitory factor; OSM, oncostatin M;
TSLP, thymic stromal lymphopoietin. wherein signaling could be rescued by genetic comple-
a
Note that IL-4 activates JAK1 and JAK3 when it acts through the type I IL-4 receptor mentation.397 Defective signaling in response to IFNα and
(IL-4Rα + γc, found for example on T cells). However, JAK3 is not activated when
IL-4 signals through the type II IL-4 receptors (IL-4Rα + IL-13Rα1, a form of recep-
IFNβ was found in a mutant fibroblast cell line (U1 cells)
tor that is expressed on a number of non-T cells, including fibroblasts). lacking TYK2; defective signaling in response to IFNα ,

IFNβ, and IFNγ was found in a mutant cell line lacking could result in clinical disease, ranging from moderate-to-
JAK1 (U4 cells), and defective IFNγ signaling was found severe immunodeficiency.
in cells lacking JAK2.397 A variety of other data have indi- Analogous to the similarity of X-linked SCID and JAK3-
cated the importance of JAK kinases in cytokine signaling deficient SCID, mice deficient in either γc or JAK3 also have
pathways. First, dominant negative JAK2 inhibits signal- indistinguishable phenotypes.211,212,470–472 These in vivo data
ing by erythropoietin and growth hormone,465,466 while a underscore the vital role of JAK3 in mediating γc-dependent
dominant negative JAK3 inhibits signaling in response to functions and suggested that JAK3 is essential for most,
IL-2467; as noted previoulsy, JAK1 is vital for IL-6 signaling. if not all, γc functions. Some in vitro data indicate that γc
Second, humans 468,469 and mice470–472 deficient in JAK3 ex- may do more than to recruit JAK3,467,483 and interestingly,
hibit developmental and signaling defects. Third, humans γc can interact with calpain,484 but the recruitment of JAK3
with an activating mutation in JAK2 develop polycythemia is clearly essential, and the defects in T-cell and NK-cell
vera,473 and a JAK2 inhibitor inhibited the growth of acute development associated with γc or JAK3 deficiency are in-
lymphoblastic leukemia cells in vitro.474 Fourth, a patient distinguishable. Because JAK3 deficiency is not clinically
with mutation in TYK2 developed a form of immunodefi- or phenotypically more severe than γc deficiency in humans
ciency.475 Fifth, in Drosophila, the hopscotch gene encodes and mice, it seems likely that γc is the major, if not only, pro-
a JAK kinase, wherein loss-of-function alleles result in tein with which JAK3 associates. Moreover, as had been pre-
lethality and decreased proliferation, whereas a gain-of- dicted based on the association of γc with JAK3,41 JAK3 in-
function allele, hopscotch Tumorous-lethal results in melanotic hibitors are immunosuppressive, and one JAK inhibitor has
tumors and hypertrophy of the hematopoietic organs.476,477 now been approved by the FDA.485 Although JAK3 was sug-
Sixth, in zebrafish, JAK1 is vital for normal cell migra- gested to associate with CD40,486 a functional role for JAK3
tion and anterior specification.478 Seventh, as discussed in in CD40 signaling has been questioned.487 The phenotypes
the following, JAK kinases are constitutively activated in of mice lacking each of the four JAK kinases are summa-
many cell lines infected with a number of viruses, includ- rized in Table 25.14.
ing HTLV-I, v-Abl, spleen focus-forming virus, and with
v-Src.479–482 These data together underscore the vital roles JAK2 Mutations and Translocations
of JAKs in cytokine signaling. Depending on the function The JAK2V617F mutations has now been identified in most
of the particular cytokine (eg, development, differentia- of patients with polycythemia vera,473,488 some of whom have
tion, or proliferation), the particular JAK kinases poten- homozygous and some have heterozygous mutations. The
tially may be involved in a variety of processes, and when homozygous patients had duplications of the mutated allele.
dysregulated, in at least certain settings, they appear to This mutation results in a constitutively activated kinase
contribute to cellular transformation. and presumably confers factor-independent growth, some-
thing that has been demonstrated in transfected cell lines.
JAK3 Mutations Result in an Autosomal Recessive Moreover, patients with polycythemia exhibit erythropoi-
Form of Severe Combined Immunodeficiency that is etin-independent BFU-E. Furthermore, the JAK2 locus is
Indistinguishable from X-Linked Severe Combined involved in a chromosomal translocation to create the TEL-
Immunodeficiency JAK2 protein that is causally related to a human leukemia,489
A very large number of different mutations in γc have been further underscoring the relationship of JAK2 to the growth
observed in X-linked SCID. As might be expected, in cases of hematopoietic cells.
where it has been examined, amino acid substitutions in
the extracellular domain result in defective cytokine bind- TYK2 Mutations and Human Immunodeficiency
ing. In contrast, mutations or truncations in the γc cyto- A mutation in TYK2 has now been identified in a patient
plasmic domain result in defective signaling. Analysis of with primary immunodeficiency.475 The patient had clinical
a family in which a number of males exhibit a moderate defects more severe than anticipated from TYK2-deficient
form of X-linked combined immunodeficiency revealed mice, with increased susceptibility to viral infections and
that this disease also resulted from a mutation in γc41 (Leu atypical mycobacterial infections and defective signal-
271 → Gln) that resulted in a decrease, but not total loss ing in response to IL-23, IL-6, and IL-10. Additionally, the
of JAK3 association, in contrast to the loss of JAK3 inter- patient had hyper-IgE production, which perhaps corre-
action seen with mutations in the γc cytoplasmic domain sponds to enhanced Th2 cell–mediated allergic inflamma-
that cause X-linked SCID. Thus, the severity of the immu- tion in TYK2− / − mice.490 Because this was a single patient,
nodeficiency inversely correlated with the degree of JAK3 it remains to be determined if other individuals with TYK2
activation.30,45 Moreover, it was predicted that JAK3 was deficiency will have this full range of clinical problems.
required for T-cell and NK-cell development, and that mu-
tations in JAK3 might result in a clinical phenotype indis-
tinguishable from that in X-linked SCID.41 Indeed, this is ACTIVATION OF JAKS AND
the case in humans.468,469 As one would hypothesize, many THE JAK-STAT PARADIGM
mutations in JAK3 have been identified.355 Presumably, any The paradigm of JAK-STAT activation is shown in
mutation that interferes with its ability to interact with γc, Figure 25.7, which also shows activation of other pathways,
with its catalytic activity, or with recruitment of substrates such as the Ras-MAP kinase and PI 3-kinase/AKT/p70 S6

TABLE 25.14 Phenotypes of Mice Deficient in Type I and Type II Cytokines, Their Receptors, JAKs and STATs
Type I Cytokines and Their Receptors
gc family
IL-2205,692,693 Normal thymic and peripheral T-cell development. Decreased polyclonal T-cell responses in vitro, but
more normal in vivo responses to pathogenic challenges. Autoimmunity with marked changes in
levels of serum immunoglobulin isotypes. Ulcerative colitis–like inflammatory bowel disease.
IL-4208,209 Defective Th2 cytokine responses and class switch; defective IgG1 and IgE production.
IL-2/IL-4210 Some features of both IL-2 and IL-4 knockout mice. No gross abnormalities of T-cell development.
IL-7104 Greatly diminished thymic and peripheral T-cell development and B lymphopoiesis, resulting in pro-
found lymphopenia.
IL-9133,135 No T-cell defect. Defect in pulmonary goblet cell hyperplasia and mastocytosis following challenge
with Schistoma mansoni eggs. No defect in eosinophilia or granuloma formation.
IL-15151 Defective NK-cell development. Defect in CD8 memory T-cell homeostasis.
IL-2Rα217 Normal initial lymphoid development, but massive enlargement of peripheral lymphoid organs, poly-
clonal T- and B-cell expansions, and activated T cells, with impaired activation-induced cell death.
Autoimmunity with increasing age, including hemolytic anemia and inflammatory bowel disease.
IL-2Rβ218 Severe autoimmunity including autoimmune hemolytic anemia. Death within approximately 3 months.
Deregulated T-cell activation. Dysregulated B-cell differentiation and altered Ig profile.
γc211,212 Greatly diminished thymic development but double negative, double positives, and single positives all
represented. Age-dependent accumulation of peripheral CD4+ T cells with an activated memory
phenotype. Greatly diminished numbers of conventional B cells, although B1 cells are present. No
NK cells or γδ cells. Absent gut-associated lymphoid tissue, including Peyer patches.
IL-4Rα694 Like IL-4−/− mice, they have defective Th2 cytokine responses and class switch; defective IgG1 and
IgE production. In addition, they cannot expel Nippostrongylus brasiliensis, presumably because of
defective IL-13 signaling.
IL-7Rα103 Greatly diminished thymic and peripheral T-cell development and B lymphopoiesis, resulting in pro-
found lymphopenia.
IL-15Rα150 Defective NK cell development. Defect in CD8 memory T-cell homeostasis.
IL-13366,695 Defective Th2 cell development and the ability to expel helminths. Impaired diarrhea to OVA.
IL-4/IL-13 DKO695 Produce almost no IgE, highly resistant to OVA-induced diarrhea.
IL-13Rα1378,695 Exacerbated Th2 responses, with diminished mortality after infection with S. mansoni and greater suscepti-
bility to N. brasiliensis. However, partially impaired allergic diarrhea to OVA.
IL-21R182,184 Decreased IgG1, elevated IgE.
IL-21R/IL-4 DKO182 Panhypogammaglobulinemia (mimicks the B-cell phenotype in humans with XSCID).
bc family
IL-5227 Decreased basal level of eosinophils and defective induction of eosinophils following infectious
challenge. Developmental defect in CD5+ B1 cells. Normal antibody and cytotoxic T-cell responses.
GM-CSF696,697 Normal basal hematopoiesis. Unexpected abnormalities of the lung; abnormal pulmonary homeostasis.
IL-3Rβ698 No defects (due to redundant function of βc).
βc243,698 Defective responses to IL-5 and GM-CSF but normal responses to IL-3 (due to redundant function
of IL-3Rβ). Diminished eosinophils: both basal levels and in responses to infectious challenge.
Unexpected abnormalities of the lung characterized by pulmonary proteinosis and reduced phago-
cytosis by alveolar macrophages. In other words, the defects are a combination of those found in
the IL-5– and GM-CSF–deficient mice.
IL-3Rβ + βc DKO698 Same phenotype as βc-deficient mice, except that they cannot respond to IL-3.
gp130 family
IL-6699 Impaired acute-phase responses following infection or tissue damage. Decreased numbers of
hematopoietic progenitor cells.
gp130700,701 Embryonic lethal. Extreme hypoplastic development of the myocardium; although the ventricular wall
was very thin, trabeculation within the ventricle chamber was normal. Hematologic abnormalities
characterized by greatly reduced CFU-S and somewhat reduced CFU-Gm and BFU-E. Markedly
diminished size of thymus and numbers of thymocytes. Reduced primordial germ cells in embryonic
gonads. Diminished size of placenta.
LIF702–704 Decreased hematopoietic progenitor cells. Normal sympathetic neurons but deficient neurotransmitter
switch in vitro. Defective blastocyst implantation.
LIFRβ705,706 Postnatal lethality. Normal hematopoietic and germ cell compartments but multiple neurologic,
skeletal, placental, and metabolic defects. The greater severity than found in LIF-deficient mice re-
flects that LIFRβ is shared by several cytokines, including CNTF, LIF, OSM, and CT-1.
CNTF707 Progressive atrophy and loss of motor neurons.

TABLE 25.14 Phenotypes of Mice Deficient in Type I and Type II Cytokines, Their Receptors, JAKs and STATs (Cont.)
CNTFRα708 Severe motor neuron deficient resulting in perinatal mortality. The more severe phenotype than in
CNTF-deficient mice was unexpected and suggests another cytokine may utilize CNTFRα.
IL-11Rα709 Blastocysts can implant but decidualization cannot occur, associated with failure of pregnancy. Fetal
lethal phenotype.
IL-12 p40710 Impaired but not completely lacking in their ability to produce IFNγ and to mount a Th1 response in
vivo. Elevated secretion of IL-4; normal production of IL-2 and IL-10. Substantially decreased CTL
responses. Resistant to infection with intracellular pathogens.
IL-12Rβ1711 Defective IL-12 signaling. IL-2 but not IL-12 could augment NK activity. Defective IFNγ production in
response to ConA or anti-CD3. Severe defect in Th1 differentiation.
IL-12 p35175,712 Selective loss of IL-12 function without loss of IL-23. Less resistant to infection than p40 and IL-12Rβ1 KO.
IL-12Rβ2713 Selective loss of IL-12 function without loss of IL-23. Less resistant to infection than p40 and IL-12Rβ1 KO.
IL-23 p19175,714 Complete resistance to EAE, defective DTH response.
EBI3715 Increased pathologic features of colitis and shorter survival than in p28 KO mice, indicating that IL-35
rather IL-27 is protective.
IL-27 p28715 Decreased pathologic features of colitis and longer survival than in EBI3 KO mice, indicating that
IL-35 rather than IL-27 is protective.
Epo, Tpo, G-CSF, and M-CSF
Epo716 Embryonic lethal. Complete block of fetal liver erythropoiesis, resulting in severe anemia, yet normal
development of BFU-E and CFU-E progenitor cells.
EpoR717 Same as Epo-deficient mice.
TpoR718 Decreased megakaryocytes and platelets, but other hematopoietic cells are present in normal numbers.
G-CSF719 Neutropenia and impaired neutrophil mobility. Diminished granulocytes and macrophage precursors.
M-CSF720,721 Osteopetrosis, absence of teeth. Females are infertile, suggesting an unexpected role for M-CSF.
M-CSF + GM-CSF721 A combination of defects of both M-CSF and GM-CSF, with osteopetrosis and alveolar proteinosis. Early
death of pneumonia.
Type II cytokines and their receptors
IFNAR1722–724 Normal development. Defective immune defense against most viral infections tested, including lymphocytic
choriomeningitis virus, Semliki Forest virus, Theiler virus, vesicular stomatitis virus. Normal resistance to
Listeria monocytogenes, Leishmania major, Mycobacterium bovis, and Mycobacterium avium.
IFNγ725–727 Normal lymphoid development. Impaired resistance to Listeria monocytogenes, Leishmania major, M.
bovis, and M. avium. Can mount curative responses to a number of viruses. CD4+ effector cells de-
fault to the Th2 pathway after infection with Leishmania. Succumb to infection with Toxoplasmosis
gondii.
IFNGR1723,724,728 Normal lymphoid development. Impaired resistance to Listeria monocytogenes, Leishmania major,
M. bovis, and M. avium Can mount curative responses to a number of viruses.
IL-10422,427 Normal lymphocyte development and antibody responses. Chronic enterocolitis, anemia, and growth
retardation. Augmented inflammatory responses.
IL-19729 Increased susceptibility to dextran sulfate sodium–induced colitis, increased accumulation of
macrophages, and production of proinflammatory cytokines and IFNγ.
IL-20R139 Increased bone density with defective osteoclast differentiation.
IL-20R2730 Anti-TCR stimulation results in elevated IL-2 and IFNγ production, with increased antigen-specific CD4
and CD8 cells that produce IFNγ, but diminished IL-10.
IL-22731 Increased susceptibility to infection with Citrobacter rodentium, with augmented bacterial burden, in-
testinal epithelial damage, and mortality.
IL-28A732 Normal clearance of a range of viruses but diminished antiviral activity in response to TLR3 and TLR9
agonists. Expression of IL-28α on nonhematopoietic cells appears to be the most critical.
STATs and JAKs
STAT1557,558 Defective responses to both type I and type II IFNs. Defective response to certain viruses and bacterial
antigens.
STAT2559 Defective signaling in response to type I IFNs; interestingly, the defect in not as severe in STAT2-
deficient macrophages as it is in STAT2-deficient fibroblasts.
STAT3560–564 Embryonic lethal. Embryos implant but cannot grow. The fact that this phenotype is even more severe than
that seen with gp130 suggests a role for STAT3 via a gp130-independent cytokine. By Cre-lox methodol-
ogy, STAT3 was also selectively targeted in T cells, which exhibit defective IL-2–induced proliferation that
correlates with a defect in IL-2–induced IL-2Rα expression; as expected, T cells also exhibit defective
signaling in response to IL-6. STAT3-deficient neutrophils and macrophages show defective IL-10 signal-
ing. STAT3 is also essential for normal involution of the mammary epithelium, for wound healing, and for
normal hair cycle processes. STAT3-deficient DCs exhibit defective Flt3L-dependent expansion of DCs.
(continued)

TABLE 25.14 Phenotypes of Mice Deficient in Type I and Type II Cytokines, Their Receptors, JAKs and STATs (Cont.)
STAT4566,567 Defective Th1 development. Essentially, the same phenotype as in IL-12–deficient mice, including de-
fective IL-12–mediated boosting of NK cell cytolytic activity.
STAT5A511,571 Defective lobuloalveolar development in the mammary gland, a syndrome resulting from defective pro-
lactin signaling. Defective IL-2–induced IL-2Rα expression and associated defects in IL-2–induced
T-cell proliferation. Defective superantigen-induced expansion of Vβ8 T cells. Defective antigen-
induced recruitment of eosinophils into the lung as well as antigen-induced IgG1 production.
STAT5B512,572 Defective growth analogous to Laron dwarfism, a disease of defective growth hormone signaling.
Defective IL-2–induced IL-2Rα expression. More severe defects in IL-2–induced T-cell proliferation
and NK-cell proliferation. Defective antigen-induced recruitment of eosinophils into the lung as well
as antigen-induced IgG1 production. Defective NK cytolytic activity.
Partial STAT5A/ Defective signaling in response to prolactin and growth hormone. Absent NK-cell development. Major
STAT5B513,514 defect in T-cell proliferation and TCR signaling. Anemia.
Complete STAT5A/ Fetal lethality; absent T-cell development.
STAT5B573
STAT6568–570 Defective Th2 development. Essentially, the same phenotype as IL-4–deficient mice. Defective B-cell
proliferation.
JAK1463 Perinatal lethality, with defective signaling by gp130-dependent cytokines (IL-6, IL-11, CNTF, OSM, LIF,
CT-1, and
NNT-1/BSF-3). Defective signaling by γc-dependent cytokines (IL-2, IL-4, IL-7, IL-9, IL-15, IL-21; of
these, IL-2 and IL-7 were formally evaluated).
JAK2733,734 Fetal lethality with profound anemia due to defective signaling in response to Epo.
JAK3470–472 Very similar and possibly identical to γc-deficient mice. Defective signaling in response to γc-dependent
cytokines (IL-2, IL-4, IL-7, IL-9, IL-15, IL-21). Greatly diminished T cells in thymus and spleen, but
then age-dependent peripheral expansion of CD4+ T cells. Unlike humans with JAK3 mutations,
Jak3-deficient mice have greatly diminished B-cell numbers as well.
TYK2735–737 Mice lacking TYK2 exhibit diminished signaling in response to IFNα/β and IL-12, but it is not abrogated.
Primarily STAT3 activation is diminished, even though STAT1/STAT2 and STAT4, respectively, are
the STAT proteins that are primarily activated by IFNα/β and IL-12. Additionally, there are dimin-
ished responses to IFNγ and IL-18.
BFU, burst forming unit; BSF-3, B cell-stimulating factor-3; CD, cluster of differentiation; CFU, colony forming unit; CNTF, ciliary neurotrophic factor; CT-1, cardiotrophin-1; CTL, cytotoxic T
lymphocyte; DC, dendritic cell; DKO, double knockout; DTH, delayed type hypersensitivity; EAE, experimental autoimmune encephalomyelitis; Epo, erythropoietin; G-CSF, granulocyte-
colony stimulating factor; GM-CSF, granulocyte macrophage-colony stimulating factor; IFN, interferon; Ig, immunoglobulin; IL, interleukin; KO, knockout; LIF, leukemia inhibitory factor;
M-CSF, macrophage-colony stimulating factor; NK, natural killer; NNT-1, novel neurotrophin-1; OSM, oncostatin M; STAT, signal transducer and activator of transcription; TCR, T-cell
receptor; Th, T helper; Tpo, thrombopoietin; XSCID, X-linked severe combined immunodeficiency.
kinase pathways. Following IFN or cytokine engagement, The previous discussion assumes that transphosphoryla-
dimerization or higher order oligomerization of receptor tion of JAK kinases is a mechanism for the amplification of
complexes is induced. This in turn allows the juxtaposi- catalytic activity. Unless other kinases are involved, how-
tioning of JAK kinases, facilitating transphosphorylation ever, implicit to this idea is that the JAK kinases themselves
and activation. In receptors with only two chains, the direct must exhibit some basal activity that is amplified to a higher
transphosphorylation of one JAK by the other seems likely to level by autophosphorylation or transphosphorylation. It is
occur. In more complex receptors, such as the IFNγ system, reasonable to speculate that both models may be operative,
because the receptor is a heterotetramer with two IFNGR-1 depending on the system. Consistent with JAK kinases being
chains that each bind JAK1 and two IFNGR-2 chains that activated by phosphorylation, mutagenesis of a critical tyro-
each bind JAK2, it is not clear if JAK1 and JAK2 transacti- sine in TYK2492 or JAK2493 (eg, tyrosine 1007 in the case of
vate each other or if one of the JAK kinases plays a dominant JAK2) in the activation loop of the kinase domain inhibits
role. Indeed, one study suggests that JAK2 may phosphory- activity. It is also conceivable that phosphorylation of other
late both itself and JAK1, thereby increasing the catalytic tyrosines on the JAKs may create appropriate motifs for the
activities of both kinases.491 JAK1 in turn then phosphory- recruitment of additional signaling molecules.
lates IFNGR-1, allowing the recruitment of STAT1. In this The function of the pseudokinase domain remains un-
model, it is additionally suggested that JAK2 phosphorylates clear. No other metazoan protein tyrosine kinases contain
STAT1.491 Interestingly, a kinase-dead mutant of JAK1 was such a domain. The JH2 lacks the third glycine in the criti-
able to mediate the induction of certain IFNγ-induced cal Gly-X-Gly-X-X-Gly motif, is missing an aspartic acid that
genes, indicating a potential “structural” role for JAK1, but serves as the proton acceptor that is typically conserved in
catalytically active JAK1 was essential for the establishment the catalytic loop of both tyrosine and serine kinases, and
of the antiviral state,491 emphasizing the essential role of is missing the conserved phenylalanine in the Asp-Phe-Glu
both JAK1 and JAK2 for normal IFNγ function. motif that binds adenosine triphosphate355 together, pre-

FIG. 25.7. Schematic of Cytokine Signaling Showing Multiple Signaling Pathways Activated by Interleukin
(IL)-2. Shown is the association of JAK1 and JAK3 with different chains of the receptor. Activation of JAK
kinases results in tyrosine phosphorylation of IL-2Rβ. This allows the docking of STAT5 proteins via their SH2
domain. The STATs themselves are tyrosine phosphorylated, dimerize, and translocate to the nucleus where
they modulate expression of target genes. The schematic also indicates that another phosphotyrosine medi-
ates recruitment of SHC, which then can couple to the Ras/Raf/MEK/mitogen-activated protein kinase pathway.
Also shown is the important phosphatidyl inositol 3-kinase pathway. These and other pathways are activated
by many type I cytokines.
sumably explaining the lack of catalytic function of the JH2 sponse to cytokines that are mitogenic growth factors (eg,
pseudokinase domain.452 Despite the lack of catalytic activ- IL-2, IL-3, etc.), and in the antiviral response (IFNs). Studies
ity, there are data in support of vital functions for this rein zebrafish revealed not only a role for JAK1 in early verte-
gion. While the kinase domain alone can act as an active ki- brate development but also revealed that during early devel-
nase, it is interesting that a mutation in the JAK kinase JH2 opment JAK1 kinase was exclusively of maternal origin.478
domain can hyperactivate the Drosophila (hopTum-l /DSTAT) These developmental roles for JAK1 in zebrafish are consis-
JAK-STAT pathway, and that the corresponding Glu695 to tent with the importance of a JAK kinase in early Drosophila
Lys mutation in mouse JAK2 also resulted in increased au- development as well.496
tophosphorylation of JAK2 and phosphorylation of STAT5
in transfected cells.494 Moreover, the JH2 domain may play
an important role in mediating the interaction of JAKs with SIGNAL TRANSDUCER AND ACTIVATOR OF
STAT proteins.495 TRANSCRIPTION PROTEINS ARE SUBSTRATES
Given that JAK kinases are associated with a wide range FOR JAKS THAT AT LEAST IN PART HELP
of IFN and cytokine receptors, it was predicted that JAKs DETERMINE SPECIFICITY
would mediate signals involved in multiple processes. These Given that there are only four JAK kinases but scores of
include important roles in development (as demonstrated by cytokines, it is clear that JAKs by themselves cannot fully
the lack of T-cell and NK-cell development associated with determine specificity. Indeed, different cytokines with dif-
JAK3 deficiency, a defect at least partially due to defective ferent actions activate the same JAKs. One level of specificity
signaling in response to IL-7 and IL-15), signaling in re- comes from the same JAKs having different substrates, de-

pending on the receptor. The best characterized substrates than STAT5A and STAT5B within a single species. The same
for JAKs are the STAT proteins,355,447,448 and these appear to is true for mouse and human STAT5B, suggesting that there
provide some, but not all of the specificity, particularly given has been evolutionary pressure to maintain the difference
that there are only seven STAT proteins. Among the mutant between STAT5A and STAT5B and that these two proteins
cell lines with defects in IFN signaling, in addition to the might have certain distinctive functions. In this regard,
ones with defects in JAKs noted previously, others were STAT5A and STAT5B knockout mice exhibit both similarities
defective in STAT proteins, providing perhaps the earliest as well as some differences in their phenotypes511–514 ; however,
data proving a vital role for STAT proteins in signaling in given different levels of each STAT5 protein in certain tissues,
response to IFNs. it remains to be determined if the different phenotypes result
STAT proteins are classically considered to be latent tran- from different intrinsic actions of STAT5A and STAT5B ver-
scription factors that initially exist in the cytosol but then sus differences in the total level of STAT5.
must be activated by phosphorylation, with subsequent trans- As active STAT proteins exist as dimers, the ability of
location to the nucleus. STATs were first discovered as factors at least some STATs to form heterodimers (eg, STAT1 with
that bound to the promoters of interferon-inducible genes. STAT2 or STAT3448) increases the number of different com-
The seven mammalian STAT proteins are STAT1,497 STAT2,498 plexes that can form. In addition, further complexity can be
STAT3,499,500 STAT4,501,502 STAT5A,503–507 STAT5B,505,507,508 and generated by the ability of at least some of the STATs to exist
STAT6.509 Table 25.15 summarizes the cytokines that activate in alternatively spliced forms,508,515 some of which are inactive.
each of the STATs. Additionally, some STATs (eg, STAT3) are A schematic of STAT proteins is shown in Figure 25.8. The
known to have more than one isoform with distinct func- STATs can be divided into two basic groups: those that are lon-
tions.510 Although the STATs conserve a reasonable level of ger (STAT2 and STAT6, approximately 850 amino acids) and
homology, STAT5A and STAT5B are unusually closely re- those that are shorter (STAT1, STAT3, STAT4, STAT5A, and
lated, being 91% identical at the amino acid level.505–508 It is STAT5B, between 750 and 800 amino acids). The STAT genes
interesting that mouse and human STAT5A are more related cluster in three locations. Mouse STAT2 and STAT6 are both
located on chromosome 10; STAT1 and STAT4 are located on
chromosome 1; and STAT3, STAT5A, and STAT5B are located
on chromosome 11.516 Correspondingly, human STAT5A and
Cytokines and the Signal STAT5B are closely positioned on chromosome 17q.508
TABLE 25.15 Transducer and Activator of The classic model is that in order for STATs to be “acti-
Transcriptions They Activate vated” and to be able to function as transcriptional activa-
tors, a number of cellular events must occur. They first bind
Cytokine STATs Activated to phosphorylated tyrosines on cytokine receptors, are tyro-
IFNα/β STAT1, STAT2, STAT4 sine phosphorylated, dissociate, dimerize, translocate from
IFNγ STAT1 the cytosol to the nucleus, bind to target DNA sequences,
Growth hormone STAT5A, STAT5B and activate gene expression. A number of conserved struc-
Prolactin STAT5A, STAT5B tural features common to all STATs help to explain these
Erythopoietin STAT5A, STAT5B functions. These include an SH2 domain, a conserved ty-
Thrombopoietin STAT5A, STAT5B rosine residue, a DNA-binding domain, and a C-terminal
IL-10 STAT3 transactivation domain, and an N-terminal STAT tetramer-
IL-12 STAT4, STAT3 ization region (Fig. 25.8). In addition, there is now evidence
G-CSF STAT3 for the biologic function of nuclear STATs that are not phos-
phorylated, as discussed subsequently.
gc family
IL-2, IL-7, IL-9, IL-15, IL-21 STAT5A, STAT5B, STAT3,
STAT1 Docking of Signal Transducer and Activator of
IL-4 STAT6, STAT5A, STAT5B Transcription on Receptors or Other Molecules, their
IL-13 STAT6 Tyrosine Phosphorylation, and Dimerization
TSLP STAT5A, STAT5B
Each STAT protein has an SH2 domain that plays two impor-
bc family tant roles: 1) for receptor docking, as for example, has been
IL-3, IL-5, GM-CSF STAT5A, STAT5B shown for STAT1 docking on IFNGR-1,517 STAT2 docking on
IFNAR-1,518 STAT3 docking on gp130,519 STAT5A and STAT5B
gp130 family docking on IL-2Rβ and IL-7Rα,356,520 and STAT6 docking on
IL-6, IL-11, CNTF, LIF, OSM, CT-1 STAT3 IL-4Rα521; and 2) for STAT dimerization that is mediated by
Leptin STAT3 the SH2 of one STAT interacting with the conserved phos-
CNTF, ciliary neurotrophic factor; CT-1, cardiotrophin-1; G-CSF, granulocyte-colony phorylated tyrosine of another STAT protein. In the case of
stimulating factor; GM-CSF, granulocyte macrophage-colony stimulating factor; the IFNα receptor, no STAT1 docking site on IFNAR-1 or
IFN, interferon; IL, interleukin; LIF, leukemia inhibitory factor; STAT, signal trans-
ducer and activator of transcription; TSLP, thymic stromal lymphopoietin. IFNAR-2 has been identified, and it is believed that STAT1 may
Note that for IL-2, IL-7, IL-9, and IL-15, STAT5A and STAT5B appear to be the major interact with STAT2 after STAT2 is itself tyrosine phosphory-
STATs that are activated, although STAT3 in particular and STAT1 to a lesser
degree can also be activated. For IL-21, STAT3 is the dominant STAT protein acti-
lated.355 It is also possible that STATs can dock on JAKs, given
vated, followed by STAT1 and STAT5A/STAT5B.44,585 the ability to directly coprecipitate JAKs and STATs.480,495 After

FIG. 25.8. Architecture of a Typical Signal Transducer and Activator of Transcription (STAT)
Protein. Shown are the locations of the following important regions: 1) The N-terminal region
has been shown to mediate the interaction of STAT dimers bound to adjacent gamma acti-
vated sites (known to be important for STAT1 but presumably true for all STATs), 2) the DNA–
binding domain, 3) the SH2 domain that mediates STAT docking on receptors and STAT
homodimerization/heterodimerization following tyrosine phosphorylation, and 4) the location
of the conserved tyrosine whose phosphorylation allows the SH2-mediated dimerization.
At least in STAT1 and STAT3, Serine 727, which is C-terminal to the conserved tyrosine, is
an important site for phosphorylation. In the case of STAT1, p48 interacts downstream of
the STAT dimerization domain. CBP/p300 interacts with two sites at both the N-terminus
and C-terminus. Although it has been suggested that the region between the DNA–binding
domains and the SH2 domain is an SH3 domain, this remains unproven, and no interactions
with proline-rich regions have been reported; as a result, the labeling of this region as an
SH3 domain has been omitted. Note that this structure is typical of that for STAT1, STAT3,
STAT4, STAT5A, and STAT5B. The main features are conserved in STAT2 and STAT6, but
these are approximately 50 to 100 amino acids longer.
docking has occurred, a conserved tyrosine (tyrosine 701 in The mechanism for nuclear translocation was originally
STAT1, tyrosine 694 in STAT5A, etc.) can be phosphorylated. mysterious given the absence of an obvious nuclear lo-
This phosphorylation is required for SH2 domain–mediated calization signal.524 However, it was shown that tyrosine-
STAT dimerization, and the phosphorylation likely occurs phosphorylated STAT1 dimers can directly interact with
while the STAT is docked on the receptor in physical proxim- importin-α5, allowing internalization. This suggested that
ity to receptor-associated JAK kinases. Following STAT phos- there indeed was a nuclear localization signal that is normally
phorylation, the STAT protein dissociates from the receptor, masked, and mutation of Leu407 does not interfere with ty-
and its dimerization with itself or another STAT is presumably rosine phosphorylation, dimerization, or DNA binding but
then favored over its reassociation with the cytokine receptor prevents nuclear localization.524 Following its dephosphory-
chain given that STAT dimerization involves two reciprocal lation, nuclear STAT1 is exported to the cytosol by a process
phosphotyrosine-SH2 interactions, whereas docking on a re- that is dependent on the chromosome region maintenance 1
ceptor involves only one interaction. Thus, efficient activation (CRM1) export reporter.525 Thus, both import and export of
of STATs requires the presence in STATs of a conserved SH2 STAT1 appear to be regulated processes.526
domain and a critical tyrosine. Whereas most STAT dimers directly bind to DNA, in the
Whereas some receptor proteins such as IFNGR-1517 case of IFN-α / β, STAT1-STAT2 heterodimers are formed,
and IL-7Rα356 have a single STAT docking site (for STAT1 and these bind DNA in conjunction with a 48 kDa DNA-
and STAT5, respectively), a number of receptor molecules, binding protein known as IRF9; the STAT1-STAT2-IRF9
including IL-2Rβ,356,520 IL-4Rα ,521 gp130,519 Epo recep- complex is known as IFN-stimulated gene factor 3 (ISGF3).448
tor (EpoR),522 and IL-10R1,523 have more than one docking In the case of other STAT dimers, accessory proteins are not
site for their respective STATs. The presence of more than required for DNA binding. The motif recognized by ISGF3
one site not only provides functional redundancy but also complexes is AGTTTNCNTTTCC (known as an ISRE, for
potentially could allow the simultaneous activation of two IFN-stimulated response element), whereas the other STAT
STAT molecules, providing a high local concentration of complexes bind more semipalindromic TTCNmGAA motifs
phosphorylated STATs to facilitate their dimerization. that are generally denoted as gamma-activated site (GAS)
motifs for IFNγ-activated sequences, reflecting their origi-
nal discovery in the context of IFNγ.355,397,447
Signal Transducer and Activator of Transcription A series of chimeric STAT proteins were used to delin-
Nuclear Translocation and Deoxyribonucleic eate a DNA-binding domain of approximately 180 amino
Acid Binding acids, with two conserved subdomains. Although multiple
Following dimerization, the STATs translocate into the nu- STATs bind to the same motifs, their relative efficiencies
cleus where they can bind to DNA as transcription factors. vary, indicating fine specificities conferred by the different

tion. Previously, tyrosine phosphorylation was primarily

associated with membrane proximal events, but the tyrosine
phosphorylation of STATs is associated with a membrane
proximal event, as the STATs dock on cell surface recep-
tors. This phosphorylation then allows the rapid dimeriza-
tion that facilitates nuclear localization and DNA binding.
STATs can interact directly with JAK kinases (first shown
for STAT5 and JAK3),480 suggesting that STATs may at times
dock on JAKs rather than on receptors.
Nonphosphorylated Signal Transducer and Activator

of Transcription Proteins Can Be Nuclear
In addition to the classical tyrosine phosphorylation–medi-
ated dimerization and nuclear translocation, it is now clear
that STAT proteins can exist in the nucleus even without
being tyrosine phosphorylated and in that context can mod-
ulate gene expression.536–538 Moreover, nonphosphorylaed
STATs (eg, STAT4) can form N-domain–mediated dimers.539
Optimal Binding Sites for Signal Transducer and

Activator of Transcription Proteins
The optimal binding motifs for several STAT proteins
have been determined. For STAT1, STAT3, and STAT4,
a TTCCSGGAA motif was defined,448,530 while STAT5A
and STAT5B optimally bind a TTCYNRGAA motif529 and
STAT6 binds a TTCNTNGGAA motif (where Y is C or T and
R is G or A).521,529 Whereas dimeric STAT protein binding
strongly preferred canonical motifs, the range of sequences
FIG. 25.9. Three-Dimensional Structure of a Signal Transducer and recognized by STAT5 tetramers is broader, often occurring
Activator of Transcription 1 Dimer Bound to Deoxyribonucleic Acid. to two imperfect motifs or with one canonical motif with an
Reproduced from Chen et al.527 and Becker et al.,528 with permission of associated “TTC” or “GAA“ half GAS motif. The optimal
Dr. Kuryian and Cell Press.
inter-GAS motif spacing is 11–12 bp, with relative second-
ary peaks of 6–7 bp and 16 bp as well.529a The presence of
suboptimal GAS motifs spaced at appropriate distances to
DNA-binding domains. For example, whereas STAT1 ho- allow tetrameric binding may allow greater specificity via
modimers favor a TTCN3GAA motif, STAT6 prefers a cooperative binding of STAT oligomers. A subset of genes
TTCN4GAA motif.521 These differences partially explain bind STAT5 tetramers and STAT5 tetramers are critical for
why different STATs modulate the expression of nonidenti- cytokine responses and normal immune function.
cal sets of target genes. The structures of STAT1 and STAT3β
bound to DNA527,528 (Fig. 25.9) almost resembles that of a ver-
tebral column, wherein the DNA represents the spinal cord. Transcriptional Activation by Signal Transducer and
The N-terminal and coiled-coil domains are spatially the Activator of Transcription
furthest from the DNA, whereas the DNA-binding domain, In addition to tyrosine phosphorylation, some STATs can be
linker, and SH2 domain surround the DNA, with the stabil- phosphorylated on serine, and STAT1 and STAT3 phosphory-
ity apparently being provided by the SH2-phosphotyrosine lation at Serine 727 in the C-terminal transactivation domain
interaction between the STAT monomers and each STAT is required for full activity.540–542 The Serine 727 region resem-
monomer–DNA interaction with the DNA, presumably via a bles a MAP kinase recognition site, and one study indicated
“half GAS site,”529 as discussed under STAT tetramerization. that MAP kinase activity is required for IFNα /β-induced
N-terminal regions mediate cooperative DNA binding of gene expression.543 Serine 727 mediates the interaction of
STAT proteins when multiple STAT-binding sites occur in STAT1 with MCM5, a member of the minichromosome
close proximity,530,531 for example, in the IFNγ gene.530 In the maintenance family of proteins; this interaction presumably
IL-2Rα gene, IL-2 response elements have been described in is important for maximal transcriptional activation.544 In
both the 5′ regulatory region and the first intron, each of contrast, STAT2 is not serine phosphorylated.448 STAT5A and
which has tandem GAS motifs that functionally cooperate STAT5B are serine phosphorylated,545 and this phosphoryla-
to mediate IL-2–induced IL-2Rα transcription.532–535 tion may be important in cellular transformation.546,547
STAT proteins were the first transcription factors In addition to this regulated modification of the STAT
that were identified as targets for tyrosine phosphoryla- proteins, STATs can also interact with other factors. For

example, as noted previously, STAT1-STAT2 heterodimers similar to that of IL-12–deficient mice (ie, defective Th1 de-
bind DNA only after interacting with IRF9 to form ISGF3.448 velopment), consistent with the observation that STAT4 is
STAT1 interacts with and synergizes with Sp1 for transcrip- activated by IL-12.566,567 Analogously, STAT6-deficient mice
tional activation in the ICAM-1 gene548 and with TRADD to exhibit a phenotype similar to that of IL-4–deficient mice (ie,
influence IFNγ signaling.549 An alternatively spliced shorter defective Th2 development),568–570 in keeping with the obser-
form of STAT3, denoted STAT3β, associates with c-Jun to vation that STAT6 is only activated by IL-4 and the closely
enhance transcriptional activity,550 and full-length STAT3 related cytokine, IL-13. Interestingly, mice lacking STAT5A
can also associate with c-Jun.551 Moreover, certain STATs exhibit a defect in prolactin-mediated effects, including de-
can interact with the potent transcriptional coactivators fective lobuloalveloar proliferation,571 whereas mice lack-
CBP/p300.552–554 In the case of STAT1, this is mediated by ing STAT5B have defective growth similar to that found in
interactions involving both the N- and C-terminal regions Laron dwarfism.572 Thus, although STAT5A and STAT5B
of STAT1 and the CREB and E1A binding regions of CBP, appear to always be coordinately induced, each of these
respectively.553 STAT5A has been shown to associate with the STATs is important in vivo. In addition to these defects, both
glucocorticoid receptor.555 Additionally, the IL-2–response STAT5A- and STAT5B-deficient mice have defects in T-cell
element in the IL-2Rα gene requires not only STAT5 bind- development and signaling.511,512 STAT5A-deficient mice
ing but also the binding of Elf-1, an Ets family transcrip- have diminished numbers of splenocytes and exhibit a de-
tion factor, to a nearby site.532 Thus, active STAT complexes fect in IL-2–induced IL-2Rα expression.511 STAT5B-deficient
appear to involve the coordination of STAT proteins with mice have similar defects but also have diminished thymo-
other factors. The corepressor silencing mediator for reti- cytes.512 Most dramatically, these mice have a major defect
noic acid receptor and thyroid hormone receptor was identi- in the proliferation of freshly isolated splenocytes512 and de-
fied as a potential STAT5-binding partner that binds to both fective NK-cell development.512 Mice lacking both STAT5A
STAT5A and STAT5B and potentially plays a negative regu- and STAT5B in lymphoid cells exhibit a dramatic defect in
latory role.556 T-cell development,573 and even hypomorphic expression of
both STAT5A and STAT5B is sufficient to result in a lack of
NK-cell development.513 Presumably, the lack of T-cell devel-
Specificity of Signal Transducer and opment relates to defective IL-7 signaling, whereas the lack
Activator of Transcription Proteins of NK-cell development relates to defective IL-15– dependent
Analogous to the JAKs, the same STATs are activated by STAT5 activation. STAT5A/STAT5B double knockout mice
multiple cytokines. The phenotypes of mice lacking expres- exhibit lethality that is associated with severe anemia that
sion of each of the seven STAT proteins are known, and these develops in these mice.574
knockout models have helped to clarify which cytokines
critically depend on a given STAT protein. STAT1 knockout
mice exhibit defects that are very selective for the actions of Signal Transducer and Activator of Transcription
type I and type II IFNs,557,558 suggesting that STAT1 is only Proteins are Evolutionarily Old
vital for the actions of IFNs, even though a variety of other Just as Drosophila has a JAK kinase, there is a Drosophila
cytokines have been reported to activate STAT1. Although STAT, denoted as either DSTAT or STAT92E.575,576 The exis-
it is possible that STAT1 plays an important but redundant tence of JAK kinases and STAT proteins in lower organisms
role for at least some of these other cytokines, the pheno- suggest that the system is evolutionarily old. A STAT has
type of STAT1-deficient mice indicates a need for caution been identified in Dictyostelium that recognizes the sequence
in the interpretation of in vitro experiments that use high TTGA,577 has highest sequence similarity to STAT5B, and
concentrations of cytokines or cell lines. STAT2-deficient can bind mammalian IFN-stimulated response elements.577
mice exhibit defects consistent with selective inactivation of Interestingly, Saccharomyces cerevisiae do not appear to have
IFNα / β signaling.559 STAT3-deficient mice exhibit fetal le- STATs as no SH2 domains have been identified in the entire
thality; the embryos implant, but they exhibit defective de- S. cerevisiae genome.
velopment and growth.560 Deletion of STAT3 within specific
lineages has revealed that mice lacking STAT3 in T cells561
have normal lymphoid development but exhibit a defect in What Types of Functions Are Mediated by Signal
IL-2–induced IL-2Rα expression, somewhat analogous to Transducer and Activator of Transcriptions?
what is seen in STAT5A- and STAT5B-deficient mice (see First and foremost, STAT proteins can translocate to the nu-
the following discussion). They also exhibit defective Th17 cleus and bind to regulatory regions of target genes and influ-
differentiation due to defective signaling by IL-6 and IL- ence transcription. Although STATs are generally activators
21.169 Neutrophils and macrophages lacking STAT3 exhibit of transcription, they can repress. For example, STAT1 can
defective signaling to IL-10, and it is known that STAT3 is function as a transcriptional repressor of the c-myc gene.578
important for the normal involution of the mammary epi- STAT proteins were discovered based on the study of IFN-
thelium, for wound healing, and for normal hair cycle.562,563 inducible genes as part of studies intended to understand
DCs lacking STAT3 exhibit a defect in Flt3L-dependent dif- the cellular differentiation events that lead to development
ferentiation,564 and antigen-presenting cells that are deficient of the antiviral state. In addition to roles in differentiation,
in STAT3 exhibit disruption of priming of antigen-specific STAT proteins can contribute, directly or indirectly, to sur-
CD4 + T cells.565 STAT4-deficient mice exhibit a phenotype vival and/or mitogenic/proliferative responses that typify he-

matopoietic and immunologic cytokines, such as IL-3, IL-5, OTHER LATENT TRANSCRIPTION FACTORS AS
GM-CSF, IL-2, and IL-4.355 First, a number of in vitro systems EXAMPLES OF CYTOPLASMIC TO NUCLEAR
have demonstrated that viruses or viral oncogenes are associ- SIGNALING (NF-kB, NFAT, AND SMADS)
ated with activated JAK-STAT pathways, suggesting a role for
An exciting feature of STAT proteins is that they exist in an
STATs in cellular transformation. Indeed, STAT3 and STAT5
inactive form in the cytosol and then are rapidly translocated
have been implicated as oncogenes, with compelling data for
to the nucleus. The rapid activation within minutes of sig-
transformation capability of STAT3 in cell lines, persistently
nals from cell membrane to nuclear DNA binding makes the
tyrosine-phosphorylated STATs in several types of leukemia
STAT acronym apt, analogous to the urgency associated with
and lymphoma, and the development of T-lymphoblastic
“STAT” emergency physician orders in clinical medicine.
lymphomas in STAT5 transgenic mice.579 The role of STAT3
The rapid activation of STAT proteins is somewhat analo-
in the tumor microenvironment makes it an attractive target
gous to several other transcription factors.587 NF-κ B also
for cancer immunotherapy.580 Second, there is diminished
undergoes rapid nuclear translocation but by a completely
proliferation in a number of the STAT knockout mice that
different mechanism from STATs. In contrast to STAT pro-
have been analyzed. For example, STAT4-deficient cells ex-
teins where the tyrosine phosphorylation of the STATs is an
hibit diminished cellular proliferation to IL-12566,567; STAT6-
initiator of nuclear translocation, for NF-κ B, it is the serine
deficient cells exhibit diminished proliferation to IL-4568,569 ;
phosphorylation and/or uquitination of Iκ B that results in
and STAT5-deficient mice have diminished T-cell prolifera-
its dissociation and/or destruction, allowing the release and
tion to IL-2.511,512 Some of these effects are indirect, based
translocation of NF-κ B. There is the classical NF-κ B path-
on modulation of receptor expression. For example, STAT6-
way used by many cytokines, which involves the IKKα /
deficient mice exhibit decreased IL-4Rα expression. Similarly,
IKKβ /IKKγ-dependent phosphorylation of Iκ B, resulting
the absence of STAT5 results in decreased IL-2–induced IL-
in the activation of NFκ B1 (p50)/Rel-A (p65) heterodimers
2Rα expression as well as decreased proliferation.511,512 Thus,
as well as the alternative pathway in which a homodimer of
in at least some cases, decreased proliferation results at least
IKKα mediates the activation of NFκ B2 (p52)/Rel-B het-
in part from decreased expression of receptor components.
erodimers.588–594 A third example of cytosolic to nuclear
STAT5A and STAT5B regulate BCL-XL induction, indicating
translocation occurs with nuclear factor of activated T-cell
their ability to affect cell survival.574 Finally, STAT1 has been
(NFAT) family proteins,595–597 which are vital for regulating
linked to cell growth arrest and induction of the cdk inhibi-
transcription of a number of cytokines, including for ex-
tor p21WAF1/CIP1,581 and activation of STAT1 occurs in thana-
ample IL-2, IL-4, and GM-CSF. NFAT is translocated to the
tophoric dysplasia type II dwarfism as the result of a mutant
nucleus where it associates with activator protein 1 (AP-1)
fibroblast growth factor receptor.582 In this chondrodysplasia,
family proteins to form a functional complex. It is the activa-
the mutant fibroblast growth factor receptor induces nuclear
tion of calcineurin and dephosphorylation of NFAT that al-
translocation of STAT1, expression of p21WAF1/CIP1, and growth
lows its nuclear translocation. A fourth example of cytosolic
arrest, suggesting a possible relationship to the disease. Thus,
to nuclear translocation occurs with the SMAD proteins that
different STATs may potentially mediate either growth ex-
mediate TGF-β signaling, which contributes to Th17 differ-
pansion or growth arrest. Moreover, STATs may potentially
entiation (discussed subsequently). For SMADs, the phos-
play other types of roles, as well. For example, STAT3 has
phorylation is on serine and the kinase is intrinsic to the
been reported to serve as an adapter to couple PI 3-kinase to
receptor, but like STATs, activation of these latent transcrip-
the IFNAR-1 component of type I IFN receptors.583
tion factors is rapid and initiated by the binding of a growth
A more complete understanding of the actions of the dif-
factor. Thus, several mechanisms, each involving phosphory-
ferent STAT proteins may be facilitated with a compilation
lation or dephosphorylation, have evolved to allow cytoplas-
of the genes that are regulated by each STAT. Progress in
mic to nuclear translocation of latent transcription factors.
this area has been made for a number of cytokines by mi-
croarray or RNA-Seq analysis (eg, for the IFNs399). It is also
important to recognize that not all cytokine signals are de-
OTHER SUBSTRATES FOR JAKS
pendent on STATs, as illustrated, for example, by STAT1-
independent IFN signals,584 STAT5-independent IL-2 sig- As JAKs are potent cytosolic tyrosine kinases, it is evident
nals,520 and STAT1/STAT3-independent IL-21 signals.585 that the JAKs may do more than just phosphorylate tyro-
sine residues on receptors where STAT proteins dock as well
as phosphorylating the STATs. In vitro data indicate that
Do Other Proteins Bind to GAS Motifs? JAK kinases also can phosphorylate receptor tyrosines other
At least one non-STAT protein can bind to GAS motifs. The than the docking sites for STATs. For example, in the case of
BCL6 gene is often mutated or has undergone translocations IL-2Rβ, JAK1 can phosphorylate not only tyrosines 392 and
in diffuse large cell B-cell lymphomas. Interestingly, BCL6 510, which are STAT docking sites, but also tyrosine 338,
binds to GAS motifs capable of binding STAT6 and specifi- which is a docking site for SHC,520 all of which are required
cally can inhibit IL-4 action.586 Mice lacking BCL6 expres- for maximal proliferation. JAK kinases are known to auto-
sion exhibit defective germinal center formation, suggesting phosphorylate themselves or transphosphorylate other JAKs.
that formation of germinal centers may be at least partially Other molecules are potentially phosphorylated by JAK ki-
dependent on BCL6 regulated (presumably negative) con- nases, including the STAM adapter molecule598 and the p85
trol of certain STAT-responsive genes.586 subunit of PI 3-kinase.599

Interferon Regulatory Factor Family Proteins ber of phosphotyrosine docking sites, particularly for the
IRFs form a family of nine proteins that are regulated by type p85 subunit of PI 3-kinase, and they presumably serve to
I IFNs.395,600,601 They each have a well-conserved N-terminal recruit important accessory molecules. Interestingly, both
DNA-binding domain that forms a helix-loop-helix struc- insulin and IL-4 could induce tyrosine phosphorylation
ture as well as a C-terminal interaction domain. IRFs are of an IRS-1–like molecule in hematopoietic cells. The 32D
critical for a number of actions, including type I IFN–de- myeloid progenitor cells lack IRS-1 and could only signal in
pendent gene transcription.395 Select IRFs play critical roles response to insulin or IL-4 when they were transfected with
in TLR-mediated IFN induction. IRFs can have negative as IRS-1.620 Both the insulin receptor and IL-4Rα proteins con-
well as positive effects. For example, IFNγ-mediated repres- tain NPXY sequences that are important for IRS-1 or IRS-2
sion of IL-4 is mediated by IRF protein(s).602 binding; in IL-4Rα , this is contained within a sequence de-
noted as the I4R motif.621
Other cytokines have subsequently been shown to ac-
OTHER SIGNALING MOLECULES IMPORTANT tivate IRS-1 and/or IRS-2. For example, growth hormone
FOR CYTOKINES can induce phosphorylation of IRS-1622 and IRS-2623 ; IFNγ
Other Tyrosine Kinases besides JAKS and LIF induce phosphorylation of IRS-2623 ; and the γc-de-
In addition to their activation of JAK kinases, a number of pendent cytokines IL-2, IL-7, and IL-15 can induce tyrosine
cytokines can activate Src family kinases. For example, IL-2 phosphorylation of IRS-1 and IRS-2 in T cells.624 The signifi-
can activate p56lck603,604 in T cells and p59fyn and p53/p56lyn cance of these findings remain unclear, as 32D cells reconsti-
in B-cell lines.605,606 The activation of some of these kinases tuted with a complete IL-2 receptor can proliferative vigor-
has been reported to be mediated by associating with the “A” ously in response to IL-2,608 whereas as noted previously, in
region of IL-2Rβ. Another tyrosine kinase, Syk, has been re- these same cells, IL-4 responsiveness requires coexpression
ported to associate with the S region of IL-2Rβ.607 However, of both IL-4Rα and IRS-1, indicative of differing roles for
the significance of these interactions is less clear than that for IRS proteins for different cytokines.
JAK kinases. First, cells lacking Lck can vigorously prolifer-
ate in response to IL-2.608,609 Second, when the A region is Phosphatidylinositol 3-Kinase
deleted, proliferation still occurs, albeit at a lower level than
PI 3-kinase is a lipid kinase that consists of an 85 kDa regu-
seen with wild-type IL-2Rβ.610 However, Y338, which is re-
latory subunit and a 110 kDa catalytic subunit.625 PI 3-kinase
quired for the recruitment of SHC, is in the A region and is
phosphorylation and activation can be induced by a number
required for normal proliferation.520 Thus, it is possible that
of cytokines,626–629 and the use of inhibitors, such a wort-
the decrease in proliferation associated with deletion of the
mannin or LY294002, have demonstrated its importance in
A region relates more to the loss of Y338 than it does to the
signaling for at least certain cytokines. IRS-1 has multiple
loss of association of Lck. Moreover, in Il2rb − / − mice recon-
docking sites for PI 3-kinase (YXXM motifs) and thus for
stituted with an IL-2Rβ A-region mutant, proliferation is in-
some cytokines, such as IL-4, the association of IRS-1 might
creased rather than decreased.611 Additional investigation is
be the mechanism by which PI 3-kinase can be recruited.
required to clarify the role of activation of Src family kinases
by IL-2 and other cytokines as well as the significance of the
Syk interaction with IL-2Rβ. As Syk and JAK1 both associ- The Ras/Mitogen-Activated Protein Kinase Pathway
ate with the S region of IL-2Rβ, mutations that delete the S Another major signaling pathway for a number of cytokines
region simultaneous prevent both associations, making it is the Ras/MAP kinase pathway.630 This pathway presum-
hard to determine the specific role of Syk. Syk-deficient mice ably is used by cytokines whose receptors recruit the SHC
exhibit normal IL-2 proliferation,612 further suggesting that adaptor molecule, which in turn mediates the recruitment of
Syk may not play an important role in IL-2–induced prolif- Grb2 and Sos, eventually leading to the activation of Ras. In
eration. The G-CSF receptor also can form a complex with turn, Ras couples to the MAP kinase pathway through a well-
Lyn and Syk,613 but again Syk-deficient mice do not exhibit defined signaling cascade. Certain cytokines, such as IL-2
a defect in G-CSF signaling.612 βc has also been reported to and IL-3, appear to use this pathway, whereas others, such
interact with Src family kinases614 ; gp130 has been reported as IL-4, do not, indicating that this pathway is differentially
to associate with a number of other kinases, including Btk, important depending on the cytokine that is being used.
Tec, and Fes615–617; and IL-4Rα has been shown to interact
with Fes.618 However, relatively little is known about the sig-
nificance of tyrosine kinases besides JAKs in cytokine signal- DOWNMODULATION OF CYTOKINE SIGNALS
ing. Additionally, PI 3-kinase, p38, ERK, and JNK can each Much of the previous discussion has centered on the mech-
become activated, leading to activation of many downstream anisms by which cytokines induce signals. However, the
transcription factors, including NF-κ B, AP1, PU.1, IRF1, mechanisms by which cytokine signals can be terminated
IRF4, and IRF8, to mediate a program of gene expression. are also extremely important. There are multiple levels at
which negative regulation can occur. These include 1) reg-
IRS Proteins ulating a balance between the production (transcriptional
IRS-1 was discovered as a tyrosine phosphorylated substrate and translational control) of the cytokine, its receptor, and/
of the insulin receptor.619 IRS proteins have a large num- or downstream signaling molecules and the degradation

of these same molecules; and 2) regulation of the activa- ing. PTP1B has been reported to dephosphorylate JAK2 and
tion state of the receptor and downstream signaling mol- TYK2, and TCPTP has been reported to dephosphorylate
ecules. For example, these can be mediated by transcrip- JAK1 and JAK3.449 In the cytosol, Shp-2 and PTP1B have also
tional repressors or molecules that either inhibit cytokine been reported to dephosphorylate cytosolic STAT5 and nu-
signaling (eg, SOCS proteins) or reverse activated states (eg, clear STAT1 and STAT3, whereas TCPTP has been reported
phosphatases). to dephosphorylate STAT1 and STAT3 in both cytosol and
Transcriptional control of cytokine production is a widely nucleus, but additional work is needed to determine whether
used mechanism. Many T-cell–derived cytokines such as these events are occurring physiologically in vivo.449
IL-2, IL-3, and IL-4 are only produced by activated T cells, In addition to dephosphorylation, another mode of nega-
and their production is lost with the loss of activation. IL-15 tive regulation is by degradation. In addition to the degra-
provides an example where translation of the protein is care- dation of receptor molecules, STAT1 itself is a target of the
fully regulated, in part by the existence of multiple upstream ubiquitin-proteasome pathway.644 Finally, it is possible that
antithymocyte globulins.147 Most cytokine receptor chains regulation also can occur at the level of alternative splicing.
are constitutively expressed, but some like IL-21R and IL-2Rβ In this regard, alternatively spliced versions of some of the
are regulated in part by signals that act through the TCR. STATs448,508,515 have been reported.
The most regulated receptor chain may be IL-2Rα , whose
expression is absent on resting lymphocytes but strongly in-
duced following stimulation with antigens, mitogens, and THE CIS/SOCS/JAB/SSI FAMILY OF INHIBITORY
certain cytokines, but the transcriptional/translational con- ADAPTER PROTEINS
trol of most cytokine receptors is poorly studied. In 1995, the protoype molecule for an interesting class
Because phosphorylation events are vital for the creation of proteins was discovered. The prototype molecule was
of phosphotyrosine docking sites, dephosphorylation is an named CIS, for cytokine inducible, SH2-containing pro-
obvious mechanism of control. Indeed, two tyrosine phos- tein, and it was shown to negatively regulate the actions of
phatases, Shp-1 (formerly also known as SHP, HCP, SH- a set of cytokines.645,646 CIS is rapidly induced by a variety
PTP1, and PTP1C) and Shp-2 (formerly also known as Syp of cytokines, including IL-2, IL-3, GM-CSF, and Epo, to
and PTP1D) have been shown to play roles related to cy- physically associate with both the βc and Epo receptors.645,646
tokine signaling.631,632 Shp-1 mutations cause the motheaten Subsequently, a related protein, variably denoted SOCS-1,
(me) and viable motheaten (me v) phenotypes in mice.633,634 JAB (JAK-binding protein), and SSI-1 (STAT-induced STAT
The viable motheaten mouse had a less severe phenotype inhibitor-1) was identified that could negatively regulate the
that is associated with increased numbers of erythroid pro- activity of other cytokines, including IL-6647–649 and IL-2,
genitor cells and hyperresponsiveness to Epo,635 suggesting among others. Interestingly, this protein could associate
that Shp-1 might normally diminish responsiveness to Epo. with JAK family kinases, whereas this function has not been
Indeed, it was demonstrated that Shp-1 binds directly to reported for CIS. A total of eight CIS/SOCS/JAB/SSI fam-
the EpoR when Y429 is phosphorylated.636 This tyrosine is ily members have been identified that collectively regulate
located in a “negative” regulatory region of the EpoR, and signals in response to multiple cytokines.650–654 These pro-
when mutated, Epo-responsive cells can grow in lower con- teins are now generally known as SOCS proteins and share
centrations of Epo. Following Shp-1 binding to Y429, de- a central SH2 domain and a region known as a C-terminal
phosphorylation and inactivation of JAK2 is facilitated.636 region known as a SOCS box. Additional SOCS box–con-
Thus, the negative regulation of Epo signaling appears to taining proteins lack SH2 domains and include proteins
be at the level of a receptor-dependent inactivation of a JAK known as ASBs (ankyrin repeat-containing proteins with a
kinase. Shp-1 has also been shown to interact with βc and SOCS box), SSBs (SPRY domain-containing proteins with a
to mediate diminished IL-3–induced signaling637 and to be SOCS box), and WSBs (WD40 repeat–containing proteins
able to associate with both TYK2638 and JAK2.639 with a SOCS box). SOCS proteins tend to be expressed at
Shp-2 has generally been considered primarily an very low levels in nonactivated cells and to be induced by
“activating” phosphatase; it is therefore interesting that it cytokines and pathogens. Following their induction, they
can also interact with JAK1, JAK2, and TYK2.640 In addi- negatively influence cytokine signaling, serving as negative
tion to the presumed dephosphorylation of JAK kinases by feedback regulators. Their actions can extend beyond type I
phosphatases, STAT proteins also appear to be regulated at cytokines, for example, SOCS-1 can negatively regulate LPS
the level of tyrosine dephosphorylation,641 and interestingly, responses.655
it has been shown that dephosphorylation of phosphotyro- Knockout mice for a number of the SOCS proteins have
sine on STAT1 dimers requires extensive spatial reorienta- been prepared, alone and in combination. CIS knockout
tion of the two monomers within the dimer, something that mice are relatively normal, but CIS transgenic mice ex-
is facilitated by the N-terminal domain that is involved in hibit low body weight, failure of lactation, and diminished
tetramerization.642 Finally, another type of phosphatase, numbers of NK and NKT cells as well as altered TCR-
namely the lipid phosphatases, known as SHIP and SHIP2, mediated responses, presumably related to its ability to in-
can act as negative regulators of cytokine signals.643 hibit STAT5-dependent responses including those related
In addition to Shp-1 and Shp-2, CD45, PTP1B, and T-cell to growth hormone, prolactin, IL-15, and IL-2, as well as
PTP (TCPTP) have been reported to regulate JAK kinases. others. SOCS-1 knockout exhibit multiorgan inflamma-
CD45 appears to play roles related to Epo and IFN signal- tion and neonatal lethality as the result of augmented re-

sponsiveness to IFNγ, and this lethality can be prevented if STAT5A or STAT5B alone might not cause a phenotype due
the mice are crossed to IFNγ-deficient mice. Interestingly, to potential redundancy, mice lacking STAT5A or STAT5B
SOCS-1− / − mice also indicate an essential role for SOCS-1 exhibit defects in T- and NK-cell numbers as well as T-cell
in thymocyte differentiation, and there are diminished proliferation and NK cytolytic activity,511,512 and mice lack-
numbers of maturing B cells. Deletion of SOCS-1 within ing both STAT5A and STAT5B have an even more profound
thymocytes/T cells/NKT cells results in multiple lymphoid defect, with absent T-cell573 and NK-cell development.513
abnormalites with increased CD8 + T cells due to increased Moreover, STATB mutations have been found and cause
sensitivity to γc-dependent cytokines, including IL-7.656 growth hormone insensitivity, IGF-1 deficiency, and im-
SOCS-2 knockout mice exhibit gigantism, ostensible due to mune dysfunction.573a,573b
dysregulated growth hormone signaling. SOCS-2 knockout It can be predicted that human immunodeficiencies
T cells exhibit enhanced Th2 differentiation, and knockout might also result from mutations in some of the cytokines
mice exhibit elevated Th2 responses after helminth infec- whose receptors contain γc or from mutations in other com-
tion.657 SOCS-3 knockout mice exhibit embryonic lethality ponents of the receptors for these cytokines. Indeed, as noted
due to placental insufficiency and dysregulated responses to previously, patients with IL-2 deficiency exhibit a SCID-like
LIF and IL-6 as well as hematopoietic defects. Conditional syndrome, due to inadequate function of their T cells, and
SOCS-3 knockout mice have revealed a physiologic role recently, an unusual immunodeficiency has been found to
for SOCS-3 in regulating G-CSF signaling in myeloid cells result from a mutation in IL-2Rα .369 One patient with de-
and for “emergency” granulopoiesis658 as well as for IL-6/ fective IL-2Rβ expression also had an immunodeficiency
gp130 signaling.659,660 SOCS-6 knockout have mild growth syndrome characterized by autoimmunity,668 somewhat
retardation.449,652,661 Additional SOCS knockout mice and analogous to IL-2Rβ –deficient mice. Given that mutations
various combinations continue to be generated, adding in IL-7Rα in humans cause T–B + NK+ SCID,206 mutations
additional information to this important area of negative in IL-7 might be predicted to cause a similar syndrome. The
regulation of STAT-dependent signaling. one major difference might be that IL-7–deficient humans
might not be capable of receiving a successful bone marrow
transplant if stromal IL-7 is required for the graft. Such pa-
PROTEIN INHIBITORS OF ACTIVATED tients have not yet been identified. Given the defective NK-
SIGNAL TRANSDUCER AND ACTIVATOR OF cell development and CD8 + memory T-cell development in
TRANSCRIPTION PROTEINS IL-15– and IL-15Rα –deficient mice, it is likely that these
Protein inhibitors of activated STAT (PIAS) proteins are types of defects would also occur in humans lacking either
other negative regulators of cytokine actions,449,662–666 with of these proteins. However, again, such patients have not
PIAS-1 being an inhibitor of STAT1-binding activity and yet been identified. Although IL-9 transgenic mice develop
PIAS-3 being a similar inhibitor of STAT3-binding activ- lymphomas,669 IL-9–deficient mice exhibit defects related to
ity. These PIAS proteins block the DNA-binding activity of mast cells and mucous production rather than lymphoid de-
STAT dimers. Two other members, PIASx (which has both α fects. Thus, defects related to the IL-9 system seem unlikely
and β splice variants) and PIASy, have also been described, as causes of SCID. At present, defective expression of IL-2,
with PIASx inhibiting STAT4 and PIASY inhibiting STAT1. IL-2Rα , IL-2Rβ, IL-7Rα , γc, and JAK3 are the only cytokine-
They appear to act at least in part as transcriptional core- related mutations that have been found to cause SCID, with
pressors by recruiting other proteins such as histone deacet- TYK2 being implicated in another form of immunodefi-
ylase. In addition, PIASx and PIASy have added actions not ciency. More time will be required to determine whether
restricted to the context of STAT inhibition. For example, mutations in other cytokines, cytokine receptors, JAKs, or
PIASy is a nuclear matrix-associated SUMO E3 ligase (a STATs can also cause SCID.
ubiquitin-related protein) that can repress the activitity of
the Wnt-responsive transcription factor, LEF1.665 PIASy co-
expression results in the covalent modification of LEF1 by Defects in the Ability to Clear Mycobacterial
SUMO. Thus, as a class, PIAS proteins have more than one Infections and Chronic Mucocutaneous Candidiasis
type of action.449,667 A number of immunodeficiencies have been characterized
where affected individuals cannot properly clear mycobacte-
rial infections. These have also turned out to be diseases of
DISEASES OF CYTOKINE RECEPTORS
defective cytokine signaling. Mutations have been found in
AND RELATED MOLECULES the components of either IL-12 itself or in the IFN or IL-12/
A Range of Cytokine-Related Causes of Severe IL-23 receptors, with mutations having been found in either
Combined Immunodeficiency the gene encoding the p40 subunit of IL-12 (which as noted
As detailed previously, mutations in the γc cause X-linked previously is also a component of IL-23),670 in IL-12Rβ1 (a
SCID, and mutations in JAK3 cause a similar T–B + NK– component of both the IL-12 and IL-23 receptors),671 or in
SCID. Given that γc-dependent cytokines activate primarily either the IFNGR-1 or IFNGR-2 components of IFNγ recep-
STAT5A and STAT5B as signaling molecules downstream of tors.672,673 The critical role of IL-12 for Th1 cell–mediated
the JAKs, it remains an open question as to whether mu- differentiation and production of IFNγ provides the expla-
tations in these STAT proteins also cause human disease. nation for finding similar clinical syndromes in humans
Although one could hypothesize that mutations in either lacking the p40, IL-12Rβ1, IFNGR1, or IFNGR-2. Moreover,

one patient with a mutation in the STAT1 gene was also iden- major example is Castleman disease, which is associated
tified with a similar clinical syndrome,674 indicating that as with overproduction of IL-6 by lymph node cells, leading
anticipated, STAT1 is a critical mediator of IFNγ signaling. to the successful treatment of this disease with IL-6 recep-
Interestingly, this patient had a mutation on only one STAT1 tor blockade.257,689 Anti–IL-6R–based therapy also has utility
allele, but the mutation was a dominant negative mutations for rheumatoid arthritis and possibly for Crohn disase.257,690
that selectively inhibits the formation of STAT1 dimers Blocking the IL-2 receptor has been used in the treatment
(hence abrogating IFNγ signaling) but yet had at most only of patients with adult T-cell leukemia and other neoplasias,
a modest effect on the ability to form ISGF3, hence leaving and the use of humanized and conjugated antibodies have
signaling in response to IFNα / β relatively intact. Chronic produced responses in a number of individuals. Humanized
mucocutaneous candidiasis can be caused by autosomal anti-Tac monoclonal antibody, marketed under the name
dominant IL-17F or autosomal recessive IL-17RA deficiency daclizumab, has shown very strong efficacy in the treatment
or by gain-of-function heterozygous STAT1 mutant alleles of allograft rejection as well as in T-cell–mediated autoim-
that impair IL-17 immunity.675 mune disorders, including multiple sclerosis, uveitis, and
Somatic mutations that activate STAT3 occur in benign tropical spastic paraparesis. Conjugated antibodies to both
liver adenomas, suggesting a role for IL-6–STAT3 pathway IL-2Rα and IL-2Rβ are also being tested for use in a variety
in human hepatocellular tumorigenesis.676 of malignant disorders, wherein the malignant cells express
these proteins. Thus, there is the possibility of therapy, ei-
Mutations in the WSX-1/TCCR Type I Receptor ther based on blocking the cytokine or based on eliminating
the responding cells.691
Interestingly, there is a type I cytokine receptor denoted as
TCCR or WSX-1 that is related to IL-12β2 and its mutations
results in defective Th1-related responses and diminished CONCLUSION
IFNγ production, resulting in susceptible to Leishmania Type I cytokines and IFNs are involved in the regulation of
major and Listeria monocytogenes.254,677,678 TCCR/WSX-1 is an enormous number of immunologic and nonimmuno-
an essential component of the receptor for IL-27300 and also logic processes. There has been a progressive transition from
has been found to be required for resistance to Trypanosoma viewing these as discrete molecules with special actions to
cruzi.679,680 sets of molecules that can be grouped according to shared
receptor components and common signaling pathways.
Other Diseases Associated with Cytokine Receptors Signaling is one area where our understanding has greatly ex-
A number of other diseases have been reported that re- panded; the pathways that are activated are similar for many
lated to cytokine receptors. First, mutations in the growth cytokines, even when the biologic functions they induce are
hormone receptor have been found in a form of dwarfism dramatically different. Although some of the differences can
(Laron dwarfism) 681 in which target cells cannot respond be explained by “compartmentalization” according to which
to growth hormone. Interestingly, some aspects of STAT5B cells produce the cytokine and which cells express receptors
deficiency are related to this syndrome. Second, a single that allow them to respond to the cytokine, a tremendous
patient with a form of congenital neutropenia (Kostmann amount still needs to be learned about how distinctive sig-
syndrome) has been found to have a mutation in one of his nals are triggered as well as more regarding the sets of genes
G-CSF receptor alleles.682 Third, a kindred of patients with that are induced by each cytokine. These will provide vital
familial erythrocytosis has truncation in the erythropoietin information important to the quest to completely under-
receptor, resulting in hypersensitivity to Epo.683 Fourth, an stand the mechanisms by which type I cytokines and IFNs
altered virally encoded form (v-mpl) of the thrombopoietin can effect their actions. At the same time, the generation of
receptor (c-mpl) was originally identified as the oncogene knockout mice for most cytokines and their receptors, as
of the myeloproliferative leukemia virus.684 Finally, gain- well as many signaling molecules, has provided in vivo clues
of-function mutations in IL7R685 and genomic aberrations as to vital functions served by these cytokines. Caution is
of TSLPR (also known as CRLF2) 686–688 have been found in clearly needed, however, in generalizing from these fi ndings
childhood acute lymphoblastic leukemias, sometimes asso- to human biology, given some apparently major differences
ciated with mutations in JAK2 or JAK1 or TLSPR itself. in roles served, such as the essential role played by IL-7 in
both humans and mice for T-cell development, whereas IL-7
is also essential for B-cell development in mice but not in
Modulation of Cytokines and the Clinic humans. The identification of so many humans disorders
Certain diseases are associated with increased levels of cy- associated with cytokines and cytokine receptors has tre-
tokines or other situations where treatment with an an- mendously helped to teach us more about normal human
ticytokine receptor antibody is a rationale therapy. One biology as well.

CHAPTER
26 The Interleukin-1 Family
Charles A. Dinarello • Mihai G. Netea
INTRODUCTION with a signal peptide (Fig. 26.1), although an intracellular

More than any other cytokine family, the interleukin (IL)-1 form also exists.1 IL-1Ra is produced in health and is found
family of ligands and receptors is primarily associated with circulating in mice and humans where the antagonist serves
acute and chronic inflammation. Also, more than any other as a brake on inflammation driven by endogenous IL-1α
cytokine family, the IL-1 family plays a fundamental role or IL-1β. IL-1Ra binds to the IL-1RI and blocks the recep-
in the nonspecific innate response to infection that facili- tor from binding to either IL-1α or IL-1β (see Fig. 26.1C).
tates specific immunologic responses such as antibodies and Mice as well as humans born with a deficiency in functional
cytotoxic T-lymphocytes. This nonspecific response to infec- IL-1Ra exhibit increased systemic and local inflammation;
tion is now termed the “innate immune response.” The cy- in humans, a deficiency in IL-1Ra is lethal. The IL-36 re-
tosolic segment of each member of the IL-1 receptor family ceptor antagonist (IL-36Ra), another member of the IL-1
contains the toll-IL-1-receptor (TIR) domain. This domain family, inhibits the activity of endogenous IL-36α , β, and γ.
is also present in each toll-like receptor (TLR), receptors that Although IL-36Ra is not readily secreted, individuals with a
respond to microbial products, viruses, and nucleic acids. mutation in IL-36Ra develop a severe form of psoriasis. One
TIR is the functional domain for both the TLR and IL-1 re- may conclude that most members of the IL-1 family primar-
ceptor families, as mutations in this domain result in loss of ily promote inflammation and enhance specific acquired
response to IL-1 and TLR agonists. The biologic properties immune responses. But there are also members that provide
of both IL-1 family ligands and TLR agonists characteristi- a brake on inflammation. The primary characteristics of the
cally are proinflammatory and act as adjuvants for specific each member of the IL-1 family are depicted in Table 26.1.
immune responses to antigen. Thus, the IL-1 family of li-
gands and receptors is fundamental to innate immunity. INTERLEUKIN-1 FAMILY AND INNATE RESPONSES
Of the 11 members of the IL-1 family, IL-1β has emerged as Independent of the type of organism or its products, the
a therapeutic target for an expanding number of systemic innate response is one of inflammation in which the host
and local inflammatory conditions termed “autoinflamma- musters its defenses to increase the production and infi l-
tory” diseases. These diseases are distinct from autoimmune tration of phagocytic cells to the area of the invading mi-
diseases and include rare hereditary conditions. But autoin- crobe in an attempt limit infection and kill-off the invader.
flammatory diseases are also common diseases such as heart Systemically, the liver increases the synthesis of acute phase
failure, gouty arthritis, and type 2 diabetes. For these, neu- proteins, include antiproteases. Even in humans, in most
tralization of IL-1β results in a rapid and sustained reduc- cases this process protects the subject without the use of an-
tion in disease severity. Another member of the IL-1 family, tibiotics. For example, a break in the skin allows bacteria
IL-1α , is also a mediator of inflammation but is classified as to gain access to the dermis and subsequent inflammation
an “alarmin” because the cytokine is present in most cells provides activation of complement, the release of preformed
and readily released upon cell death. Whereas treatment for cytokines from keratinocytes, an increase in vascular wall
autoimmune diseases often includes immunosuppressive adhesions molecules, and the extravasation of neutrophils.
drugs, neutralization of IL-1β or blocking the IL-1 receptor This response has functioned to battle against invaders for
is mostly anti-inflammatory. millions of years and can be traced back to fruit flies.
With one exception, all members of the IL-1 family are The skin, lung, and intestinal tract each provide a first
initially translated as precursors lacking a signal peptide line of defense against microbial invasion and the lining
for secretion via the Golgi. The precursors are found in the cells, whether keratinocytes of the skin, the alveolar epithe-
cytosol and exit the cell following death by necrosis, not lial cells of the pulmonary tree, or the epithelial cells of the
apoptosis. For example, once released, IL-1α , IL-33, and entire gastrointestinal tract, each contain preformed IL-1α ,
IL-36 can be processed extracellularly by neutrophil prote- IL-18, and IL-33 as well as the members of the IL-36 subfam-
ases into active cytokines. Although IL-1β is primarily pro- ily. Because these members of the IL-1 family are each pre-
cessed intracellularly by the cysteine protease caspase-1, the formed in these cells, their release is a consequence of injury
IL-1β precursor can also be cleaved extracellularly into an and is immediate. Therefore, they are termed “alarmins” as
active cytokine by similar serine proteases of neutrophils. they alert the host to initiate the response. There are other
The one member of the IL-1 family that is readily secreted “alarmins” from the lining cells that participate in defense,
is the IL-1 receptor antagonist (IL-1Ra). IL-1Ra is translated for example, defensins, which are directly antimicrobial.
639

AXD IL-1β 269

IL-1
AXD IL-1α 271
subfamily
AXD IL-33 270
IL-1 Receptor antagonist
AXD IL-18 193

IL-18
subfamily
AXD IL-37 192 FIG. 26.1. Organization of the Interleukin (IL)-1 Family into
Three Subfamilies. The number of amino acids of the full-
length of each member is shown at the C-terminal end.
AXD IL-36Ra 155 The consensus sequence (A-X-D) is common to all IL-1
family members and serves to locate the N-terminus nine
IL-36 amino acids forward from this site, as shown by the dark
AXD IL-36α, β, γ 158
subfamily vertical bar. The N-terminus results in propieces of various
lengths. The IL-1Ra has a bonafide signal peptide and is
AXD IL-38 155 shown by comparison.
Each of the constitutively present IL-1 family members in IL-1 receptor family and TLR family signaling; a mutation
lining cells is present as a precursor. In the case of IL-1α , the in the TIR domain severely impairs responses to IL-1 fam-
precursor is fully active; in the case of the other members, ily ligands as well to a large number of microbial products.7
the precursors are weakly active at first but are converted to The TIR domain binds MyD88 (see Fig. 26.2A and E), itself
more active cytokines upon the infi ltration of neutrophils a TIR domain–containing protein, through TIR/TIR interac-
and processing by extracellular neutrophil proteases. In the tions triggering a cascade of kinases that propagate the IL-1
end, the infection is contained, the invading microorganism signal and result in transcription of a large number of genes,
is eliminated, and skin begins its process of repair. the majority of which code for other cytokines, chemokines,
Following the cloning of the mouse IL-1 receptor,2 the and a host of inflammatory mediators. Of these is IL-1 and
cytosolic domain of the IL-1 receptor was found to be ho- other members of the IL-1 family such as IL-36 and IL-18.
mologous to toll of the fruit fly.3 Moreover, at the same time, The “innate immune response” regulates to the “acquired
the TIR domain for IL-1 signaling (Fig. 26.2A) was shown by immune response.” The late Charles Janeway proposed that
Heguy to be required for IL-1 signaling.4 Toll had been ini- the innate response assists the host in mounting an acquired
tially studied since its discovery in 1985 because of its cen- immune response. This relationship between a nonspecific
tral role in establishing dorsal ventral polarity in Drosophila. cytokine providing help for a specific response to a micro-
Only since 1996 was toll linked to survival in fruit flies in- bial antigen is simply the adjuvant property of some cyto-
fected with fungi.5 However, it had already been reported, kines. The adjuvant property of some cytokines functions
back in 1988, that a member of the IL-1/TLR family, human by upregulating lymphocyte growth factors such as IL-2,
IL-1β, protected mice from lethal Pseudomonas infection.6 IL-4, and IL-6, or lymphocyte receptors resulting in expan-
As noted previously, the TIR domain is essential for both sion of lymphocyte clones, which will either rid the host of
the invading microorganism with neutralizing antibodies or
in generation of cytotoxic T cells to eliminate viral infec-
tions. In 1979, purified human IL-1β, a nonspecific macro-
TABLE 26.1 Interleukin-1 Family Members phage product, was shown to augment the T-cell response to
Family Name Name Property
specific antigen.8 It was nearly 20 years later that TLR were
identified as inducing IL-1β from monocytes.
IL-1F1 IL-1α Agonist
IL-1F2 IL-1β Agonist
IL-1F3 IL-1Ra Receptor antagonist ORGANIZATION OF THE INTERLEUKIN-1 FAMILY OF
IL-1F4 IL-18 Agonist LIGANDS AND THE CONSENSUS SEQUENCE
IL-1F5 IL-36Ra Receptor antagonist As depicted in Figure 26.1, the IL-1 family can be divided into
IL-1F6 IL-36α Agonist subfamilies according to the length of the precursor and the
IL-1F7 IL-37 Anti-inflammatory
length of the propiece for each precursor. The IL-1 subfamily
IL-1F8 IL-36β Agonist
is comprised of IL-1α , IL-1β, and IL-33. This subfamily has
IL-1F9 IL-36γ Agonist
the longest proteins with the longest propieces. In the case of
IL-1F10 IL-38 Receptor antagonist
IL-1β, the propiece is cleaved intracellularly by caspase-1 and
IL-1F11 IL-33 Agonist
then the mature cytokine is secreted. In the case of IL-1α ,
IL, interleukin. cleavage appears to be by the membrane protease calpain, but

CHAPTER 26 THE INTERLEUKIN-1 FAMILY | 641
A B C D
IL-1α/β IL-1α/β IL-1Ra IL-33
IL-1RAcPb
IL-1RAcP
IL-1RAcP
IL-1RAcP
IL-1RI
IL-1RI
IL-1RI
ST2
TIR TIR TIR TIRb TIR TIR TIR TIR
MyD88 MyD88
no signal
IRAKs IKKβ signal IRAKs IKKβ
inhibition
proinflammatory proinflammatory
signal signal
FIG. 26.2. Interleukin (IL)-1 Subfamily. A: IL-1α or IL-1β binds to the IL-1RI and recruits the coreceptor IL-1RAcP. The
heterodimeric IL-1 receptor complex results in a close approximation of the toll-IL-1-receptor (TIR) domains on each
receptor chain (arrows), resulting in the binding of intracellular MyD88 to the complex followed by phosphorylation of
MyD88. Subsequent phosphorylations of IRAKs and IKKβ, increased NF-κB, and IL-1R AcP-1 translocation to the nucleus
take place followed by expression of proinflammatory genes. B: In the central nervous system, IL-1α or IL-1β bind IL-1RI
recruiting IL-1RAcP but can also recruit the coreceptor IL-1RAcPb. IL-1RAcPb contains an altered TIR domain, which re-
sults in a reduced signal. C: IL-1Ra binds to IL-1RI. There is no recruitment of the co-receptor IL-1RAcP, no approximation
of the TIR domains, and no signal. D: IL-33 binds to its specific receptor, ST2, and recruits the coreceptor IL-1RAcP, the
TIR domains approximate, and signal transduction is initiated resulting in the induction of proinflammatory gene profile.
extracellular neutrophil proteases can also cleave the IL-1α example, with the tenth amino acid before A-X-D consen-
precursor. Extracellular neutrophil proteases account for the sus site as the N-terminus, the specific activity of the IL-36
cleavage of the propiece of IL-33. The exception in the IL-1 subfamily (IL-36α , IL-36β, IL-36γ and the IL-36Ra) is low.
subfamily is IL-1Ra, which contains a signal peptide. However, with the ninth amino acid as the N-terminus, there
The IL-18 subfamily is comprised of IL-18 and IL-37. was a marked increased in the activity.9 In the case of IL-1β,
By comparison, this subfamily has a smaller propiece. IL-18 the ninth amino acid before the A-X-D site coincides exactly
requires the cleavage of its propiece by caspase-1 in order to with the N-terminal alanine generated by the caspase-1 site.
be active. IL-37 is part of the IL-18 subfamily because the
cytokine binds to the IL-18Rα chain. It is unclear how the
propiece of IL-37 is removed. The IL-36 subfamily com- THE INFLUENCE OF INTERLEUKIN-1 FAMILY ON
prised of IL-36α , β, and γ, as well as IL-36Ra. In addition, TH17 RESPONSES
IL-38 likely belongs to this family due to its binding to the The IL-1 family plays a significant role in interferon (IFN)γ
IL-36R. The IL-36 subfamily has the shortest propiece. production, which is essential for the defense against intra-
A consensus sequence in all members of the IL-1 family cellular pathogens. On the other hand, Th2 cells are charac-
is A-X-D, where A is an aliphatic amino acid such as isoleu- terized by the production of IL-4 and are important in the
cine, methionine, or leucine; X is any amino acid; and D is host defense against parasitic infections. For more than one
aspartic acid. The aspartic acid of the consensus sequence decade, the dichotomy between Th1 and Th2 has been the
is not the aspartic acid of the caspase-1 cleavage recogni- focus of studies on differentiation of cluster of differentia-
tion site. The A-X-D motif is conserved in the IL-1 family tion (CD)4 + T-lymphocytes. More recently, Th17 helper cells
where it plays a role in three-dimensional structure of the have been described and are characterized by their produc-
active cytokine. The actual N-terminus is often located nine tion of IL-17. IL-17 plays a major role in neutrophil recruit-
amino acids before the A-X-D site. By eliminating the amino ment and host defense against extracellular bacteria and
acids before the N-terminus, the first beta-sheet structure fungi. Th17 cells produce a distinct cytokine profi le, namely
common to all members of the IL-1 family can form. For IL-17A, IL-17F, IL-21, and IL-22. The cytokines produced

by Th17 cells, in addition to activating neutrophils, are also INTERLEUKIN-1a

crucial for nonimmune cells, for example, induction of de- From an evolutionary viewpoint, IL-1α is the oldest member
fensins by IL-22 in epithelial cells and keratinocytes, which of the IL-1 family and its primary amino acid sequence is
are part of mucosal and skin defenses. It has become appar- closely related to that of the fibroblast growth factor family.
ent that Th17 responses are associated with chronic inflam- Like fibroblast growth factor, IL-1α does not have signal pep-
mation and autoimmune diseases such as multiple sclerosis, tide, binds to nuclear deoxyribonucleic acid (DNA), exits the
type 1 diabetes, Crohn disease, and psoriasis. Furthermore, cell upon death, and binds to its receptor as an unprocessed
Th17 responses are fundamental for host defense against precursor. As shown in Figure 26.2A, IL-1α binds to the IL-
many microorganisms, although they also contribute to the 1RI and recruits the IL-1R accessory protein (IL-1RIAcP)
inflammation during infection. to form a heterodimeric complex, which signals to induce
Whereas IL-4 and IL-12 were the first cytokines described inflammation. In health, primary cells contain constitu-
as influencing Th-cell differentiation, cytokines of the IL-1 tive levels of the IL-1α precursor but not IL-1β.18 The IL-1α
family also influence cytokine differentiation. IL-18 was precursor is present in keratinocytes, thymic epithelium,
initially described as IFNγ-inducing factor due to its strong hepatocytes, endothelial cells, the epithelial cells of mucus
stimulatory effect on Th1/IFNγ responses.10 It is now known membranes, including the entire gastrointestinal tract, and
that IL-18 is, in fact, a crucial cytokine directing the devel- fibroblasts, regardless of their location. The propiece of IL-
opment of Th1 cells, and one role of IL-12 is the induction 1α precursor can be cleaved extracellularly by neutrophil
of the expression of IL-18 receptors. In contrast, binding of proteases, a step that increases its biologic activity. However,
IL-33, another member of the IL-1 family, to its ST2 recep- IL-1α can also be active as a membrane-associated cytokine.
tor plays a role in inducing Th2 responses,11 and it thus ap- Most cell lines, including tumor cell lines, contain constitu-
peared as if distinct members of the IL-1 family of cytokines tive levels of IL-1α .19–21 Using an epithelial cell line, what was
directed Th1 versus Th2 differentiation. Considering these considered to be intrinsic IFNγ activities depended largely on
effects of IL-18 and IL-33, it came as no surprise that IL-1, constitutively expressed IL-1α . IFNγ activities were inhib-
the most well-known member of the family, participates in ited by antibodies to IL-1α , but not to IL-1β.20 The concept
the function of Th cells. that IL-1α acts as an autocrine growth factor assumes that
Known for over 30 years that IL-1 enhances T-cell activa- the intracellular IL-1α precursor regulates normal cellular
tion and recognition of antigen, one of the early names of IL-1 differentiation, particularly in epithelial and ectodermal
was lymphocyte activation factor. The specificity of this re- cells. In support of the concept, an antisense oligonucleotide
sponse was, however, not known. Although initially only IL- to IL-1α reduces senescence in endothelial cells.22 In fibro-
23, IL-6, IL-21, and transforming growth factor-β were sug- blasts, the constitutive IL-1α precursor binds to HAX-1, a
gested to play a role in the development of Th17 responses in nonreceptor substrate for tyrosine kinases in hematopoietic
mice, there is no dearth of data that a more complex picture cells. In fibroblasts, the IL-1α HAX-1 complex translocates
exists. Thus, IL-1β, IL-6, and transforming growth factor-β to the nucleus.23 Although the concept is that IL-1α acts as
have been reported to induce the development of Th17 cells, an autocrine growth factor in fibroblasts or endothelial cells
whereas IL-23 has been reported to be important for the in vitro, the data should be interpreted carefully as mice de-
maintenance of Th17 cells. The combination of IL-23 and IL- ficient in IL-1α show no demonstrable defects in growth and
1β induce the development of human Th17 cells expressing development, including skin, fur, epithelium, and gastro-
IL-17A, IL-17F, IL-22, IL-26, the chemokine CCL20, and tran- intestinal function.24 However, mice deficient in IL-1α still
scription factor RORγ t.12 Interestingly, these cells also released retain the N-terminal propiece, which functions as nuclear
IFNγ, displaying a phenotype common to both Th17 and Th1 factor.21 In fact, in another study, the N-terminal propiece
cells.12 The strong capacity of IL-1 to induce Th17 differentia- of IL-1α was shown to bind HAX-1.25
tion has been also linked to its well-known capacity to induce Is there is a role for intracellular precursor IL-1α in nor-
the release of prostaglandins, as reviewed in Dinarello.13 PGE2 mal cell function? The IL-1α precursor is present in cells that
induced by COX-2 is a stimulator of Th17 induction, and in- also contain large amounts of the intracellular form of the
hibitors of cyclooxygenase decrease IL-17 production.14 On the IL-1Ra (icIL-1Ra), as reviewed in Arend.1 This form of the IL-
other hand, engagement of the aryl hydrocarbon receptor, a 1Ra also binds to the IL-1 receptor and prevents signal trans-
pathway demonstrated to be crucial for the generation of Th17 duction. In fact, icIL-1Ra is thought to compete with the in-
cells, has been shown to strongly induce IL-1β.15 In addition to tracellular pool of precursor IL-1α for nuclear binding sites.
inducing IL-17 production from the Th17 subset of lympho-
cytes, IL-1β is required for the production of IL-17 by natural
killer (NK) T cells16 and of IL-22 from NK cells.17 Membrane-Associated Interleukin-1a
Thus, cytokines of the IL-1 family have an important role Precursor IL-1α can be found on the surface of several cells,
in the differentiation of the Th subsets, with IL-1β strongly particularly on monocytes and B-lymphocytes, where it is
inducing Th17 responses, IL-18 being crucial for the genera- referred to as membrane IL-1α .26 Membrane IL-1α is bio-
tion of Th1 cells, and IL-33 being important in Th2 responses. logically active27; its biologic activities are neutralized by an-
Interestingly, reciprocal regulation has been demonstrated be- tibodies to IL-1α but to IL-1β. Endothelial cells undergoing
tween the various Th subsets, with cytokines released by Th2 stress-induced apoptosis release membrane apoptotic body-
cells inhibiting Th1 responses, whereas IFNγ release from Th1 like particles containing nuclear fragments and histones as
cells impairing both Th2 and Th17 responses. well as full-length IL-1α precursor and the processed ma-

ture form.28 When injected into mice, apoptotic body-like Studies in Interleukin-1a–Deficient Mice
particles containing the IL-1α precursor induce neutro- Mice deficient in IL-1α are born healthy and develop nor-
philic infi ltration that was prevented by neutralization of mally. In some models of local inflammatory responses,
IL-1α but not IL-1β.28 wild-type and IL-1α –deficient mice develop fever and acute-
phase proteins, whereas IL-1Β –deficient mice do not.24 In
Processing and Secretion of Interleukin-1a addition, although the inflammation-associated induction
of glucocorticoids was suppressed in IL-1β –deficient mice,
Although the IL-1α precursor is biologically active, the pro- this suppression was not observed in IL-1α –deficient mice.
cessed form is more active. Furthermore, the binding of IL- However, expression of IL-1β messenger ribonucleic acid
1α to the IL-1RI has been modeled using recombinant IL-1α (mRNA) in the brain decreased 1.5-fold in IL-1α –deficient
with an N-terminus at 113. The processing of the IL-1α pre- mice, whereas expression of IL-1α mRNA decreased more
cursor is accomplished by calpain II, a membrane-associated, than 30-fold in IL-1β –deficient mice. These data suggest
calcium-dependent cysteine protease.29 In macrophages that IL-1β exerts greater control over production of IL-1α
treated with hydroquinone, calpain II levels fall and are as- than does IL-1α over the production of IL-1β. In caspase-1–
sociated with inhibition of IL-1α precursor processing.29 Not deficient mice, IL-1α production is also reduced,43 further
surprisingly, calcium influx induced IL-1α secretion of the suggesting that production of IL-1α is under the control of
processed form.30 The secretion of IL-1α requires the pres- IL-1β. It is important that caspase-1–deficient mice are also
ence of IL-1β, as IL-1β –deficient mice do not secrete IL-1α.31 deficient in caspase-11.44
IL-1α binding to IL-1β has been reported in which IL-1β acts In mice fed a high-fat diet, serum amyloid A protein, a
as a chaperone for the secretion mechanism via caspase-1.31 marker of inflammation in atherogenesis, was markedly lower
In another study, IL-1β was shown to bind to, and enhance in IL-1α –deficient mice compared to wild-type or IL-1β –
the activity of, high-mobility group protein B1 (HMGB1).32 It deficient mice.45 IL-1α –deficient mice had significantly higher
is thus possible that both IL-1α exits the cell bound to IL-1β levels of non–high-density lipoprotein cholesterol. The ben-
and HMGB1. eficial effect of IL-1α deficiency was due to hematopoietic cells
transferred from the bone marrow of IL-1α –deficient mice,
Biologic Functions of Constitutive Interleukin-1a: resulting in a reduction in aortic lesion size twice that ob-
Interleukin-1a and Sterile Inflammation served in mice transplanted with IL-1β –deficient bone mar-
row cells. Therefore, IL-1α appears to play a greater role in the
Large numbers of reports use bacterial and fungal products pathogenesis of lipid-mediated atherogenesis than IL-1β, and
to induce cytokines as models of inflammatory disease; this may be due to an effect of membrane IL-1α.
however, most inflammatory diseases are sterile. For exam-
ple, the inflammation associated with atherosclerosis, myo-
cardial infarction, stroke, cancer, and renal and liver failure
is sterile. The hypoxic insult that takes place in ischemia
INTERLEUKIN-1b
results in local necrosis and release of cellular contents, in- Interleukin-1b: The Master Cytokine in the
cluding nucleic acids. Members of the IL-1 family contribute Interleukin-1 Family
to sterile inflammation, and IL-1α plays a significant role in More than any other member of the IL-1 family, IL-1β has
this regard. Upon cell death by necrosis, the IL-1α precursor been the focus of most studies. IL-1β is a highly inflam-
is released33,34 and binds to the IL-1 receptor on nearby tissue matory cytokine, particularly in humans, as reviewed in
macrophages and epithelial cells, triggering a response.35,36 Dinarello.46 As shown in Figure 26.2A, IL-1β and IL-1α bind
For example, infi ltration of neutrophils occurs first and to the same IL-1RI and trigger a proinflammatory signal. The
followed by influx of monocytes.35 Extracts of tumor cells interest in IL-1β is also due, in part, to it being a secreted cy-
induce neutrophilic inflammation, which does not occur in tokine from macrophages and to the importance of the mac-
mice deficient in IL-1RI and is prevented by neutralization rophage in antigen presentation before the era of dendritic
of IL-1α , not neutralization of IL-1β.37 Sterile inflammation cells. The inactive IL-1β precursor is converted into an active
is independent on TLR2 and TLR4.37 cytokine by the intracellular cysteine protease caspase-1. In
Thus, IL-1α, either the unprocessed precursor or the cal- particular, persons with activating mutations in one of the
pain cleavage form, is classified as an “alarmin” because the key genes that control the activation of caspase-1 can develop
cytokine is preformed and triggers an inflammatory response life-threatening systemic inflammation, which is reversed by
rapidly. Endothelial cells subjected to nutritional stress re- either blocking the IL-1 receptor or through the use of a neu-
lease inflammatory apoptotic bodies, which contain both tralizing antibody to IL-1β. Other chronic inflammatory dis-
the precursor and processed forms of IL-1α.28 Inflammatory eases are mediated by IL-1β, as neutralizing antibodies have
apoptotic bodies induce chemokine and neutrophilic infil- been used to treat a broad spectrum of diseases.
tration into the peritoneal cavity, both of which are IL-1α The IL-1β –mediated illnesses fall into the category of
dependent.28 Platelets also contain IL-1α as well as IL-1β.38 “autoinflammatory” diseases, which are to be distinguished
Platelet-derived IL-1 induces chemokines such as IL-8 from from the classic “autoimmune” diseases. Although inflam-
endothelial cells39 and monocyte chemotactic protein (MCP- mation is common to both autoinflammatory and autoim-
1) from monocytes.40 Platelet-derived IL-1α is important in mune diseases, in the case of IL-1–mediated disease, there is
brain injury in stroke models41 and in atherosclerosis.42 no evidence for role of adaptive immunity in its induction.

Interleukin-1b is an Inducible Cytokine (NLRP3). NLRP3 is also termed cryopyrin as the gene was
Unlike IL-1α , the IL-1β precursor is not present in health. initially discovered in patients with “familial cold autoin-
Also unlike IL-1α , IL-1β is primarily a product of monocytes, flammatory syndrome,” a genetic disease characterized by
macrophages, and dendritic cells, as well as B-lymphocytes constitutional symptoms, fevers, and elevated acute-phase
and NK cells. In health, circulating human blood monocytes proteins following exposure to cold.61
or bone marrow cells do not constitutively express mRNA As monocytes exit the bone marrow, they circulate in the
for IL-1β. Endothelial cells, skin keratinocytes, fibroblasts, bloodstream for approximately 3 days. In the absence of dis-
and epithelial cells contain constitutive IL-1α and consti- ease, it is likely that these cells do not enter tissues but are
tutive IL-33 as precursors as well as mRNA, but these cells destroyed in the spleen or undergo apoptosis. There is no
do not express IL-1β mRNA even upon stimulation with dearth of reports that circulating human blood monocytes
TLR ligands. Melanoma cells do express IL-1β as a precur- release processed IL-1β upon stimulation starting 4 hours
sor, and the more aggressive and metastatic the melanoma, after stimulation with TLR agonists and continue to release
the greater the likelihood of active caspase-1 and IL-1β se- the cytokine during the following 20 to 40 hours. Following
cretion.47 In the bone marrow neutrophil precursors, IL-1β lipopolysaccharide (LPS), IL-1β mRNA levels rise rapidly
gene expression is inducible but mature neutrophils in the within 15 minutes but begin to decline after 4 hours due
circulation no longer produce IL-1β. Neutrophil IL-1β plays to the short half-life of their mRNA or the action of micro
a pathologic role in the severe inflammation of mice with a RNA. In contrast, using IL-1 itself as a stimulant, IL-1β
mutant form of the phosphatase SHP1.48 Several malignant mRNA levels are sustained for over 24 hours.52 Raising intra-
tumors do express IL-1β as part of their neoplastic nature, cellular cAMP levels with histamine enhances IL-1–induced
particularly acute myelogenous leukemia, melanoma, mul- IL-1 gene expression and protein synthesis. Monocytes of
tiple myeloma, and juvenile myelogenous leukemia, each of patients with autoinflammatory diseases such as cryopyrin-
which exhibit constitutive expression of IL-1β. Unlike most associated periodic syndrome (CAPS) and hyper IgΔ syn-
cytokine promoters, IL-1β regulatory regions are distrib- drome (HIDS) release IL-1β even without TLR stimulation
uted over several thousand base pairs upstream from the during a 24-hour incubation.62,63
transcriptional start site. In addition to a cAMP response When obtained from the venous blood of healthy sub-
element, there are NF-κ B–like and activating protein-1 jects, human blood monocytes contain active caspase-1.
sites. IL-1β gene regulation has been reviewed in detail.49 Active caspase-1, as determined by its cleavage into the
Although steady-state mRNA levels for IL-1β may be pres- active dimer, is present even in the absence of stimula-
ent, there is distinct dissociation between transcription and tion.64 Active caspase-1 present in freshly obtained mono-
translation of the IL-1β precursor. Non-TLR ligands such cytes is nevertheless dependent on the presence of the
as the complement component C5a, hypoxia, adherence to key components of the inflammasome, namely ASC and
surfaces, or clotting of blood induce the synthesis of large NLRP3.64 However, during subsequent incubation, extra-
amounts of IL-1β mRNA in monocytic cells without sig- cellular levels of adenosine triphosphate (ATP) increase in
nificant translation into the IL-1β protein. In these cells, the the supernatant as IL-1β also increases and inhibition of
IL-1β mRNA assembles into large polyribosomes, but there ATP by oxidized ATP reduces the secretion of IL-1β.64 The
is no significant elongation of the peptide.50 This failure to inhibition of IL-1β secretion by oxidized ATP is consistent
complete the translation into IL-1β protein may be due to with the role of the P2X7 receptor, which binds ATP and
the instability element present in the coding region. This in- opens the potassium channel for release of intracellular
stability region is also found in IL-18 and IL-37, and appears potassium. The presence of active caspase-1 in circulat-
to limit the mRNA of these cytokines.51 However, comple- ing blood monocytes suggests that the rate limiting step
tion of translation of the mRNA into the respective cyto- in the processing and release of IL-1β is at the level of gene
kines can be accomplished by adding low concentrations of expression.
TLR ligands or IL-1 itself to the “primed” monocytes.52 However, upon differentiation of the same blood mono-
cytes into macrophages in vitro, TLR-induced IL-1β release
requires activation of caspase-1 by exogenous ATP.64 The
Processing and Secretion of Interleukin-1b assembly of the inflammasome components with inactive
via the Caspase-1 pro–caspase-1 takes place following a fall in intracellular
Nearly all microbial products induce IL-1β via TLR acti- potassium triggered by ATP binding to the P2X7 receptor.
vation; in addition, IL-1 (either IL-1α or IL-1β) induces ATP activation of the P2X7 receptor opens the potas-
itself both in vivo and in monocytes in vitro.53 Other stud- sium channel, and simultaneously, as potassium levels fall,
ies supporting this concept of IL-1–induced IL-1 have been caspase-1 is activated by the inflammasome.65–69 Without
reported.54–57 Regardless of the stimulus, processing and exogenous ATP, there is little or no processing of the IL-1β
secretion of IL-1β requires conversion of procaspase-1 to ac- precursor in differentiated monocyte-derived macrophages.
tive caspase-1, although in some studies processing of the Alveolar macrophages obtained from the lungs of healthy
IL-1β precursor is caspase-1 independent.58 The activation human also do not release IL-1β with LPS stimulation un-
to active caspase-1 is dependent on a complex of intracel- less exogenous ATP is added.64 In addition to ATP activation
lular proteins termed the “inflammasome” by the late Juerg of P2X7, activation of IL-1β processing can also take place
Tschopp.59,60 The critical component of the inflammasome with a cathelicidin-derived peptide termed LL37, which is
is NACHT, LRR, and PYD domains containing protein 3 released from neutrophils.69

The cleavage of the IL-1β precursor by active caspase-1 receptor polymorphism is associated with increased mor-
can take place in the specialized secretory lysosomes or in tality in patients undergoing allogenic stem cell transplan-
the cytoplasm. However, more than one pathway seems tation.79 Bacteremia was documented in 68% of patients
available for processed IL-1β to exit the cell. These include by with this polymorphism compared to 18% in wild-type
exocytosis of the secretory lysosomes,65,66 shedding of plasma control patients.79
membrane microvesicles, and direct release via transporters In mice deficient in P2X7 receptors, inflammation,
or multivesicular bodies containing exosomes.70 In general, pain, and IL-1β –mediated IL-6 production are markedly
the release of processed IL-1β takes place before there is sig- reduced.80 In addition to a fall in intracellular potassium,
nificant release of lactate dehydrogenase,71 although in vitro ATP triggers formation of peroxynitrite, which is required
cell death eventually takes place. Pyroptosis is a caspase-1– for caspase-1 activation because peroxynitrite scavengers
dependent form of cell death and is induced by certain bac- prevent IL-1β secretion.81 Pannexin-1, a mammalian pro-
teria using Ipaf, a member of the Nod-like receptor (NLR) tein that functions as a hemichannel for the uptake of dyes,
family of intracellular receptors.72 An increase in intracellu- is required for caspase-1 processing and release of IL-1β via
lar calcium is also required for the mature IL-1β to exit the the P2X7 receptor.82 Pannexin-1 can also function for LPS-
cell and is phospholipase C dependent.66 induced IL-1β synthesis in the absence of TLR4.83 P2X7
receptor activity is also regulated by “regeneration and tol-
erance factor.”84
Gain of Function Mutation in Cryopyrin
Diseases associated with single amino acid activating mu-
tations in cryopyrin are termed CAPS. In monocytes from Non–Caspase-1 Processing of Interleukin-1b
patients with CAPS, activation of caspase-1 occurs without Non–caspase-1 mechanisms also exist to generate active
a requirement for a rapid fall in the level of intracellular forms of IL-1β. For example, sterile inflammation induces
potassium.57 Therefore, mutated cryopyrin allows for the fever, elevated IL-6, and increased production of hepatic
assembly of the complex of interacting proteins in the pres- acute–phase proteins. These responses are absent in mice de-
ence of normal intracellular levels of potassium. Although ficient in IL-1β but present in mice deficient in caspase-1.85,86
often studied using LPS-induced synthesis of the IL-1β pre- Sterile inflammation is often associated with neutrophilic
cursor,73 it is unlikely that LPS plays a role in autoinflam- infi ltration and neutrophils produce IL-1β. Because neutro-
matory diseases. On the other hand, spontaneous secretion phils are short-lived cells dying within hours upon emigra-
of IL-1β from monocytes of patients is due to endogenous tion, release of the IL-1β precursor from intracellular stores
IL-1β stimulation. In patients with CAPS, there is a de- is not unexpected. Processing of the IL-1β precursor extra-
crease in steady state levels of pro–caspase-1 mRNA with cellularly into an active cytokine has been reported for the
IL-1Ra treatment,54 suggesting that IL-1β stimulates its own common neutrophil protease, proteinase-3.86,87 Proteinase-3
production and processing. Thus, in any disease process also contributes to the processing of IL-18.88 Other proteases
that includes an increase in the steady state levels of pro– such as elastase, matrix metalloprotease 9, and granzyme A
caspase-1 mRNA, components of the inflammasome or the process the IL-1β precursor extracellularly. In addition, a
IL-1β precursor explain the “autoinflammatory” nature of mast cell chymase generates active IL-1β.
the disease. Type 2 diabetes appears to be an example of an Mice with a targeted IKKβ deletion in myeloid cells are
autoinflammatory disease where glucose induces IL-1β pro- more susceptible to LPS-induced shock than control mice,55
duction from the insulin-producing beta cell and IL-1β in- and markedly elevated levels of IL-1β are found in the cir-
duces the beta cell to produce its own IL-1β.74 culation associated with a prominent neutrophilia.55 The el-
evated levels of IL-1β are lethal as blockade with IL-1Ra pro-
tects these mice from death. The source of the IL-1β in these
Polymorphisms in P2X7 and the Activation mice is the neutrophil. When incubated with proteinase-3,
of the Inflammasome cleavage of the IL-1β precursor is observed yielding mo-
Patients with classic autoinflammatory diseases such as lecular weights of 25,000 and 15,000 Daltons.55 Because the
Familial Mediterranean Fever (FMF) or CAPS have nearly cleavage of the IL-1β precursor by proteinase-3, elastase, and
identical clinical parameters, secrete more IL-1β, and re- cathepsin G are within three amino acids of the caspase-1
spond dramatically to IL-1 receptor blockade yet have no cleavage site, the products of the non–caspase-1 cleavage are
mutation in NALP3. It is therefore possible that mutations biologically active.86,87 Therefore, in inflammatory condi-
in P2X7 itself or regulation of the other genes control- tions such as urate crystal arthritis, which is characterized by
ling potassium channels75 may account for dysfunctional a prominent neutrophilic infi ltration, proteinase-3 cleavage
secretion of IL-1β. For example, monocytes from patients of extracellular IL-1β precursor likely takes place.89 Mice de-
with rheumatoid arthritis are more sensitive to release of ficient in caspase-1 are not protected against urate-induced
IL-1β following ATP activation of the P2X7 receptor com- inflammation. Although IL-1Ra is effective in treating gout,
pared to monocytes from healthy controls.76 However, IL-1Ra would be equally effective in any disease with extra-
monocytes from subjects with a P2X7 Glu496Ala loss-of- cellular processing of the precursor.90–92 The importance of
function polymorphism secrete significantly less IL-1β.77 extracellular processing of the IL-1β precursor by serine pro-
Monocytes from subjects homozygous for this polymor- teases may explain, in part, the anti-inflammatory proper-
phism also released significantly less IL-18.78 Another P2X7 ties of alpha-1-antitrypsin.93

Reactive Oxygen Species and Interleukin-1b findings may reflect the fact that IL-1β is secreted into the
Processing microenvironment resulting in the emigration of monocytes
Is there a role for reactive oxygen species (ROS) in the ac- and neutrophils, whereas IL-1α remaining cell-associated is
tivation of the IL-1β inflammasome? It was reported that less likely to affect the microenvironment.
uric acid crystals added to human monocytes result in the
generation of ROS, which bind to and activate NLRP3 with Interleukin-1a and Autophagy
subsequent secretion of IL-1β.94 However, mice deficient in Autophagy is an ancient process of recycling cellular com-
ROS production exhibit a proinflammatory phenotype.95 ponents, such as cytosolic organelles and protein aggregates,
Humans with chronic granulomatous disease (CGD) due through degradation mediated by lysosomes. Autophagy is
to mutations in p47-phox cannot generate ROS and are activated in conditions of cell stress, hypoxia, starvation,
severely affected by inflammatory granuloma. Uric acid or growth factor deprivation; it promotes cell survival by
crystal activation of primary monocytes from persons with generating free metabolites and energy through degrada-
CGD produced fourfold higher levels of IL-1β compared to tion of the endogenous cellular components.104 However, in
monocytes from unaffected persons.96 In contrast to previ- addition to its role in the pathophysiology of cancer, neuro-
ous studies,94 the small molecule ROS inhibitor diphenyl- degenerative diseases, or aging, autophagy is also a modula-
eneiodonium, which reduces the production of IL-1β, does tor of inflammation.105 A role for autophagy in production
so due to inhibition of IL-1β gene expression rather than of proinflammatory cytokines, particularly of IL-1β, has
decreased caspase-1 activation.96 Another study identified emerged with deletion of ATG16-L1. For example, macro-
phagocyte oxidase–defective monocytes from CGD patients phages from ATG16L1-deficient mice produce higher levels
as a source of elevated IL-1β.97 These findings support the of IL-1β and IL-18 after stimulation with TLR4 ligands.106
concept that ROS likely dampens inflammasome activation The data suggest that higher activation of caspase-1 in the
and may explain the presence of an inflammatory pheno- ATG16L1-deficient mice accounts for the higher produc-
type characterized by granulomas and inflammatory bowel tion level.106 This observation was related to the specific
disease occurring in patients with CGD. In fact, patients degradation of the IL-1β precursor in autophagosomes in
with CGD-related inflammatory bowel disease improve mouse macrophages.107 Additional studies in the ATG16L1-
upon IL-1 receptor blocking therapy.98 deficient mice point toward a regulatory effect of autophagy
on caspase-1 activation through modulation of the NLRP3
inflammasome.94,108,109
Effects in Mice Deficient in Interleukin-1b This role of autophagy in the secretion of IL-1β was also
After 10 years of continuous breeding, mice deficient in observed in human primary monocytes, in which specific
IL-1β exhibit no spontaneous disease. However, upon chal- inhibition of autophagy leads to increased production of
lenge, IL-1β –deficient mice exhibit specific differences from IL-1β.110 However, in the same cells, tumor necrosis factor
their wild-type controls. The most dramatic is the response (TNF) α production was decreased by autophagy inhibition.
to local inflammation induced by a subcutaneous injection These data suggest divergent effects of autophagy on the
of an irritant. Within the first 24 hours, IL-1β –deficient mice production of these two important proinflammatory cyto-
do not manifest an acute-phase response, do not develop an- kines. In mice, the increase in IL-1β production is ascribed
orexia, have no circulating IL-6, and have no fever.85,99 These to increased activation of the inflammasome, but in human
findings are consistent with those reported in the same model cells, it is IL-1β mRNA transcription that is elevated when
using anti–IL-1R type I antibodies in wild-type mice.85,99 autophagy was inhibited, whereas no effects were observed
IL-1β –deficient mice also have reduced inflammation due on caspase-1 activation.106,107,110 Despite these differences be-
to zymosan-induced peritonitis.85,100 In contrast, IL-1β – tween mouse and human cells, the inhibition of autophagy
deficient mice have elevated febrile responses to LPS, IL-1β, increases the production of IL-1β but not TNFα .
or IL-1α compared to wild-type mice.101 Nevertheless, IL- The modulation of inflammation by autophagy in hu-
1β –deficient mice injected with LPS have little or no expres- mans has been studied in Crohn disease. Genome-wide
sion of leptin mRNA or protein.102 association studies in large cohorts of patients with Crohn
Mice deficient in IL-1β were compared to mice deficient disease have revealed that genetic variants in two autophagy
in IL-1α after exposure to chemical carcinogens.103 In IL- genes, ATG16L1 and IRGM, result in increased suscepti-
1β –deficient mice, tumors developed slower or did not de- bility to the disease. A nonsynonymous polymorphism in
velop in some mice. A deficiency in IL-1α , on the other hand, ATG16L1 on chromosome 2q37.1 and two polymorphisms
did not impair tumor development compared to wild-type in IRGM on chromosome 5q33.1 were significantly associ-
mice injected with the same carcinogen. In IL-1Ra–deficient ated with Crohn disease risk.111,112 Another study revealed a
mice, tumor development was the most rapid. A leukocyte significant association of Crohn disease susceptibility with
infi ltrate was found at the site of carcinogen injection. The an intronic polymorphism in the autophagy gene ULK1.113
neutrophilic infi ltrate was almost absent in IL-1β –deficient Moreover, autophagy defects have been reported in indi-
mice, whereas in IL-1Ra–deficient mice, a dense neutrophilic viduals bearing nucleotide oligomerization domain (NOD)2
infi ltrate was observed. In wild-type mice, the leukocytic in- mutations and are consistent with the concept that impaired
fi ltrate was sparse and the infiltrate that was observed in IL- bacterial clearance and increased bacterial persistence are
1α –deficient mice was similar to that of control mice. These part of the pathogenesis of Crohn disease.114

The mechanism through which polymorphisms in au- of estrogens may be related to the large number of studies
tophagy genes influence susceptibility to Crohn disease ap- on the effect of estrogens to regulate IL-1 and inflammation.
pears to involve IL-1β production. The ATG16L1 300Ala risk Similar to most members of the IL-1 receptor family,
allele was associated with elevated production of IL-1β and ST2 is comprised of three extracellular immunoglobulin
IL-6; however, this finding was only observed in cells stimu- domains and an intracellular TIR domain. Although the
lated with the NOD2 ligand muramyl dipeptide. In contrast, name ST2 is still used, the correct term is the IL-33 receptor
the expected levels of IL-1β and IL-6 were produced upon α chain (IL-33Rα). As shown in Figure 26.2D, the IL-33Rα
stimulation with TLR2 and TLR4 ligands.115 The increased chain similar to the IL-1R1 in that it is the ligand binding
production of IL-1β was associated with an increase in the chain for IL-33 but requires the IL-1RAcP to signal.121,122
steady state levels of IL-1β mRNA rather than increased Before the discovery of IL-33, several studies suggested
activation of the inflammasome.115 Studying the same that the putative ligand (IL-33) for the ST2 orphan receptor
polymorphism (ATG16L1 Thr300Ala) in human dendritic was playing a role in allergic type diseases. It became clear
cells, Cooney et al. reported defective NOD2-induced, but that activation of ST2 was uniquely driving Th2 responses.
not TLR-induced, autophagy and antigen presentation.116 Structurally, IL-33 is closer to IL-18 than IL-1β. Biologically,
Furthermore, effects of this polymorphism on antibacte- IL-33 is closest to IL-1α , as the precursors for IL-1α and
rial autophagy in epithelial cells have been observed.117 The IL-33 are constitutively present in all endothelial cells.
specific effect of the ATG16L1 polymorphism on the NOD2 As discussed in the following, like IL-1α , IL-33 functions as
pathway, and not on TLR-induced stimulation, is likely a DNA-binding molecule. The dominant property of IL-33
related to the fact NOD2 and ATG16L1 form a protein com- is the induction of IL-4, IL-5, and IL-13, as well as other
plex that is essential for NOD2-induced autophagosome for- properties anticipated for a Th2 type cytokine. Diseases
mation.118 Because the ATG16L1 Thr300Ala polymorphism thought to be due to increased immunoglobulin production
affects protein stability,119 defective induction of autophagy may also be related to IL-33. IL-33 induces the production of
and therefore enhanced IL-1β mRNA transcription upon IL-6, IL-1β, and PGE2 from mast cells.
triggering of NOD2 may be due to the presence of defective
complex.
Interleukin-33 and Th2 Responses
The properties of recombinant IL-33 recapitulate much of
INTERLEUKIN-33
the existing data that ST2 promotes Th2-type responses.
Interleukin-33 as a Member of the Interleukin-1 For example, before its discovery, a role for IL-33 in the Th2
Subfamily response was observed using soluble extracellular forms of
Formerly termed IL-1F11, IL-33 belongs to the IL-1 subfam- ST2.11 However, IL-33 has properties that go beyond its role
ily and has been studied for its role in the Th2 paradigm in the Th2 paradigm because, similar to IL-1α , IL-1β and
of immune responses. IL-1β is also linked to the Th2 re- IL-36, IL-33 forms a heterodimeric complex with IL-1RAcP
sponse. The existence of IL-33 was predicted in 1994 fol- for signal transduction.121,122 Although the IL-1RAcP is ex-
lowing the discovery of a novel member of the IL-1 receptor pressed on most nucleated cells, ST2 is somewhat restricted
family termed ST2.120 ST2 is the ligand binding chain for to low expression on most cells with the notable exception
IL-33 (Table 26.2) and is structurally similar to the ligand of mast cells.
binding chain of IL-1α and IL-1β. In addition, the corecep- There are several mechanisms by which IL-33 favors
tor for IL-33 is the IL-1RAcP, which is also the coreceptor the Th2 response. Similar to IL-1β, IL-33 induces IL-6, an
for IL-1α and IL-1β. It was not until 2005 that IL-33 was re- adjuvant for antibody production. IL-33 induction of IL-6
ported as the ligand for ST2.11 ST2 is regulated by the estro- is prevented by a blocking antibody to IL-1RAcP.122 IL-33
gen inducible transcription factor Fos,120 and this property initiates signal transduction via activation of NF-κ B, which
TABLE 26.2 Interleukin-1 Receptor Family

Name Designation Ligands Coreceptor
IL-1RI IL-1R1 IL-1α, IL-1β, IL-1Ra IL-1RAcP (IL-1R3)
IL-1RII IL-1R2 IL-1β, IL-1β precursor IL-1RAcP (IL-1R3)
IL-1RAcP IL-1R3 IL-1α, IL-1β, IL-33, IL-36 Not applicable
ST2/IL-33Rα IL-1 R4 IL-33 IL-1RAcP (IL-1R3)
IL-18Rα IL-1R5 IL-18, IL-37 IL-18Rβ (IL-1R7)
IL-1Rrp-2 IL-1R6 IL-36α, β, γ IL-1RAcP (IL-1R3)
IL-18Rβ IL-1R7 IL-18 Not applicable
TIGIRR-2/IL-1RAPL IL-1R8 Unknown Unknown
TIGIRR-1 IL-1R9 Unknown Unknown
SIGIRR TIR8 Unknown Unknown
IL, interleukin; SIGIRR, single immunoglobulin IL-1–related receptor; TIGIRR, three immunoglobulin IL-1–related receptor.

is typical of IL-1α , IL-1β and IL-18,11 but other studies have T cells and blood monocytes. One of the most studied prop-
shown that antibody cross-linking of ST2 does not result in erties of IL-33 is the induction of IL-5 and IL-13 and their
activation of NF-κ B but rather activating protein-1. IL-33 respective roles in lung inflammation such as allergic type
treatment also increased serum immunoglobulin A and im- asthma. For example, instillation of IL-33 into the airways
munoglobulin E, an expected response for a switch from triggers an immediate allergic response in the lung of naïve
Th1 to Th2. mice and worsens the response in mice sensitized to anti-
gen peripherally but challenged by exposure of antigen in
the lung.127
Processing of the Interleukin-33 Precursor Mice deficient in ST2 do not develop a Th2 response to
Initially, IL-33 was considered closely related to IL-1β and Schistosoma egg antigen. Indeed, several studies have fo-
IL-18 because the IL-33 precursor contains a caspase-1 site, cused on the role of IL-33 in the pathogenesis of helminth
which upon activation would cleave the IL-33 precursor worm infections. The Th2 response by the host contrib-
and release the active cytokine,11 similar to that for IL-1β utes to the elimination of these worm infestations, which
and IL-18. Indeed, the first recombinant forms of IL-33 are worldwide and afflict hundreds of millions. The role of
were produced with an N-terminus at the caspase-1 site.11 IL-33 in the induction of IL-4, IL-5, and IL-13 is of para-
Although recombinant IL-33 was active, the concentrations mount importance in terms of pulmonary and intestinal
required for activity were considerably higher than those of complications that reduce lifespan. Using mice deficient
other members of the IL-1 family. Indeed, subsequent stud- in IL-33, a crucial role was demonstrated in mice to rid an
ies revealed that caspase-1 actually results in loss of IL-33 infection with Strongyloides venezuelensis.128 The infection
activity and that the full-length IL-33 precursor binds to induces a unique class of cells called natural helper cells or
ST2 and is active,123 similar to the ability of the IL-1α pre- nuocytes, which upon activation by IL-33 produce IL-5 and
cursor to bind to IL-1RI. In addition, it was reported that IL-13, resulting in eosinophilic infi ltration into the lungs. In
the caspase-1 cleavage site at 178 is similar to the consensus this model, pulmonary inflammation causes damage via eo-
sequence for caspase-3 and that intracellular IL-33 precur- sinophilic infi ltration, which is IL-33 and IL-5 dependent.128
sor is a substrate for caspase-3.123 Mice injected with human IL-33 exhibit impressive
Using immobilized IL-33 precursor, neutrophil protein- pathologic changes in the arterial walls, lungs, and intes-
ase 3 (PR3) was isolated from human urinary proteins.124 tinal tissues.11 Of particular relevance to the concept that
Neutrophil PR3 is known to process the IL-1β precursor IL-33 drives a Th2 response, eosinophilic infi ltration was
into an active cytokine.86 PR3 converted human and mouse a prominent finding in the lung and in allergic rhinitis as
precursor IL-33 proteins to biologic active forms; however, well as allergic conjunctivitis.129 These initial observations
increasing the incubation time of PR3 abrogated IL-33 ac- have been confirmed by other reports.130 Although the inter-
tivities.124 Using the consensus amino acid sequence sites pretation of in vivo effects following the administration of
for PR3, six human and mouse recombinant IL-33 proteins an exogenous cytokine should be conservative, the findings
were produced and assessed for biologic activities; varying are clearly consistent with IL-33 being a proinflammatory
levels of activity were reported.124 Another study also dem- ligand of the IL-1 receptor family. Even before the ability
onstrated cleavage of the IL-33 precursor by neutrophil pro- to test IL-33–mediated activation, others had reported that
teases such as PR3, neutrophil elastase, and cathepsin G,125 neutralization of the putative ST2 ligand using soluble ST2
resulting in the generation of IL-33 with different N-termini markedly reduced joint inflammation, synovial hyperpla-
and varying levels of activity. These studies support the con- sia, and joint erosion when given in the therapeutic phase of
cept that extracellular IL-33 is released as a precursor, is rap- collagen-induced arthritis in mice.126
idly processed by neutrophil enzymes, and generates active
forms with varying levels of activity. The implications for
generation of active IL-33 by neutrophil enzymes for Th2 Interleukin-33 as an Anti-inflammatory Cytokine
polarization remain unclear. It may be more relevant to Members of the IL-1 family of ligands bind to their spe-
study the effect of proteases from eosinophils in the process- cific cell surface receptors and recruit an accessory chain.
ing of the IL-33 precursor. Nevertheless, the IL-33 precur- The IL-1RIAcP is used by IL-1α and IL-1β but also by
sor binds to ST2 and recruits the accessory chain for signal IL-36 and IL-33. The accessory chain for IL-18 is related to
transduction, but compared to IL-33 generated by neutro- the IL-1RIAcP but is encoded by a distinct gene. We now
phil proteases, the activity of IL-33 precursor is weak.124,125 recognize that other members of the IL-1 receptor family
There was no dearth of studies on ST2 tissue–specific lo- will bind more than one cytokine. The best example is IL-
calization, regulation of its expression, effects in transgenic 1α and IL-1β. Both bind with similar affinities to IL-1RI,
mice overexpressing ST2, as well as deletion, neutralization, but the three-dimensional structure of IL-1α and IL-1β
and antibody cross-linking of ST2. Elevated levels of the sol- are hardy identical.131 The IL-1β precursor binds to IL-1RII
uble form of ST2 were present in the circulation of patients as well as a processed form with the first 112 amino acids
with a various inflammatory diseases and that exogenous cleaved from the precursor. IL-37 binds to the IL-18 receptor
administration of pharmacologic doses of soluble ST2 neu- alpha chain,132 and both IL-36 and IL-38 bind to the IL-36
tralized endogenous levels of the then putative ligand IL-33 receptor.133
and reduced inflammation.126 IL-33 activates Th2 lympho- IL-33 forms a complex with ST2 IL-1RIAcP but also with
cytes, mast cells, basophils, and eosinophils, as well as NK single immunoglobulin IL-1–related receptor (SIGIRR).134

This complex plays a role in the Th2 response by reducing cloning of “IFNγ-inducing factor” in 1995,144 the name was
IL-33 signaling,134 and consistent with these observations, changed to IL-18. Surprisingly, the new cytokine was related
Th2 responses are increased in mice deficient in SIGIRR. to IL-1 and particularly to IL-1β. Similar to IL-1β, IL-18
Furthermore, there is high expression of SIGIRR in Th2 lacks a signal peptide, is first synthesized as an inactive pre-
polarized cells and in models of Th2 antigen sensitization; cursor, and remains as an intracellular cytokine. The ter-
SIGIRR-deficient mice exhibit a greater Th2 response.134 tiary structure of the mature form of IL-18 closely resembles
The complex with SIGIRR and IL-33 may explain the anti- that of IL-1β,144 although the IL-18 precursor is closely re-
inflammatory properties of IL-33. ST2 can sequester TLR lated to the IL-37 precursor. Since 1995, many studies have
adaptor molecules such as MyD88 and Mal.135 used neutralization of endogenous IL-18 or IL-18–deficient
In mice deficient in ST2, there is myocardial hypertro- mice to demonstrate the role for this cytokine in promoting
phy, ventricle dilation, and fibrosis upon pressure overload, inflammation and immune responses.145 However, the biol-
suggesting that IL-33 plays a protective role in the heart.136 ogy of IL-18 is hardly the recapitulation of IL-1β. There are
Furthermore, elevated levels of the extracellular domain of several unique and specific differences between IL-18 and
ST2 predict outcomes in patients with systolic heart failure IL-1β. For example, in healthy human subjects and also in
or following a myocardial infarction.136 In a model of cardio- healthy mice, gene expression for IL-1β in blood mononu-
myocyte hypertrophy induced by chronic administration of clear cells and hematopoietic cells is absent, and there is no
phenylephrine, administration of recombinant IL-33 inhib- evidence that the IL-1β precursor is constitutively present in
ited the phosphorylation of Iκ B and reduced the hypertro- epithelial cells.146 In contrast, in the same blood cells, large
phy and fibrosis.136 One of the more challenging aspects of amounts of the IL-18 precursor are present. Peritoneal mac-
the properties of IL-33 to act as a Th2 cytokine is its role rophages and mouse spleen contain the IL-18 precursor in
as an antagonist in the ApoE-deficient mouse model of ath- the absence of disease.146 The IL-18 precursor is also present
erosclerosis. In this model, arterial wall plaques of mice on in keratinocytes and nearly all epithelial cells. In this regard,
a high-fat diet contain IL-33 and ST2. In mice treated with IL-18 is similar to IL-1α and IL-33.
IL-33, the atherosclerotic plaques were markedly reduced.137
In mice treated with soluble ST2 to neutralize IL-33, the dis- Processing of the Interleukin-18 Precursor
ease worsened.137 The IL-18 precursor has a molecular weight of 24,000 and is
processed by caspase 1 cleavage into a mature molecule of
18,000. Compared to wild-type mice, following an injection
Interleukin-33 as a Transcription Factor of endotoxin into caspase-1–deficient mice, circulating IFNγ
Similar to IL-1α , there is another side to IL-33. Although is absent. IL-12–induced IFNγ is also absent in caspase-1–
IL-33 binds to its specific surface receptor, IL-33 is identical deficient mice.147 Importantly, any phenotypic characteristic
to a nuclear factor dominantly expressed in high endothelial of capsase-1–deficient mice must be studied as to whether
venules.138 This nuclear factor is termed NF-HEV. In addi- the deficiency is due to reduced IL-1β or IL-18 activity. For
tion to endothelial cells, constitutive nuclear localization of example, the caspase-1–deficient mouse is resistant to coli-
IL-33 has been reported in several cell types such as type tis,148 but the IL-1β –deficient mouse is susceptible in the
II lung epithelial cells,128 epithelial cells,139 and pancreatic same disease. Because neutralizing antibodies to IL-18 are
stellate cells.140 In fact, IL-33 binding to DNA and acting protective in the colitis model, caspase-1 deficiency appears
as a nuclear factor is similar to IL-1α binding to chroma- to prevent processing of IL-18.148,149 On the other hand, there
tin and functioning as a nuclear factor.21,33,141 A short IL-33 are examples where caspase-1 processing of IL-18 is not re-
peptide similar to a sequence in Kaposi sarcoma virus binds quired. For example, Fas ligand stimulation results in release
chromatin.142 The full-length IL-33 precursor, but not ma- of biologically active IL-18 in caspase-1–deficient murine
ture IL-33, binds to the N-terminal Rel homology domain macrophages.150 Similar to IL-1β processing, proteinase-3
of NF-κ B p65.143 In cells overexpressing the IL-33 precur- appears to activate processing to mature IL-18.88
sor, there was a reduction in IL-1β –induced TNFα .143 These Similar to IL-1α and IL-33, the IL-18 precursor is consti-
data are consistent with other data that IL-33 possesses anti- tutively expressed in endothelial cells, keratinocytes, and in-
inflammatory properties (see previous discussion) and the testinal epithelial cells throughout the gastrointestinal tract.
mechanism for this property of IL-33 appears to be nuclear Macrophages and dendritic cells are the primary sources for
sequestration similar to that of IL-1α .33 the release of active IL-18, whereas the inactive precursor
remains in the intracellular compartment of mesenchymal
cells. Also, similar to IL-1α and IL-33, the IL-18 precursor is
INTERLEUKIN-18 AND INTERLEUKIN-37 SUBFAMILY
released from dying cells and processed extracellularly, most
Interleukin-18 likely by neutrophil proteases such as proteinase-3.
Background The IL-1 family consensus sequence (A-X-D) in IL-18 is
IL-18 was first described in 1989 as “IFNγ-inducing factor” I-N-D at amino acid 50, but the N-terminus generated by
isolated in the serum of mice following an injection of endo- caspase-1 is 14 amino acids before the consensus sequence
toxin. The mice had been pretreated with Proprionibacterium rather than 9 amino acids (see Fig. 26.1). Although the IL-18
acnes, which stimulates the reticuloendothelial system, par- as well as the IL-1β precursor can be cleaved extracellularly
ticularly the Kupffer cells of the liver. Many investigators by proteianse-3, there is an additional enzyme that cleaves
concluded that the serum factor was IL-12. With molecular the IL-18 precursor into an active cytokine. Merprin β is a

member of the merpin family of zinc metalloproteinases, that for IL-1.154 There are differences between IL-1 and IL-18
which are expressed on the membrane surface of epithe- signaling that remains unexplained. With few exceptions,
lial cells of the intestine and kidney but also on myeloid IL-1α or IL-1β are active on cells in the low nanogram/mL
cells.151,152 Merprin α will cut the IL-1β precursor into an ac- range and often in the picogram/mL range. In contrast, IL-18
tive cytokine, and inhibitors of merprin α markedly reduce activation of cells expressing the two IL-18 receptor chains
serum levels of IL-1β in mice subjected to cecal ligation and requires 10 to 20 ng/mL and sometimes higher levels.155,156
puncture.151 Merprin β will cleave the IL-18 precursor and Although nearly all cells express the IL-1RI, not all cells
generate an active cytokine on cells bearing the IL-18R.153 In a express IL-1RAcP. Similarly, most cells express the IL-18Rα
model of dextran sulfate sodium (DSS) colitis, serum IL-18 is but not all cell express the IL-18Rβ. IL-18Rβ is expressed on
elevated on days 3 and 5, during which time the mice develop T cells and dendritic cells but not commonly expressed in
inflammatory colitis. However, in merprin β –deficient mice, mesenchymal cells. The best example is the A549 cell. This
there is a statistically significant reduction in serum IL-18.153 cell line, derived from a lung carcinoma epithelial cell, does
The role of IL-18 and caspase-1 in DSS colitis is discussed in not express IL-18Rβ,157 and there is no signal unless IL-12 is
the following. added to induce IL-18Rβ.10 In the absence of IL-18Rβ, IL-18
binds to IL-18Rα without a proinflammatory signal. In
Signal Transduction by Interleukin-18 A549 cells transfected with IL-18Rβ, IL-18 induces IL-8 and
As shown in Figure 26.3A, IL-18 forms a signaling complex a large number of genes. One of these genes is the former IL-
by binding to the IL-18 alpha chain (IL-18Rα), which is the 2-induced gene termed NK4,158 now termed IL-32.157 IL-32
ligand binding chain for mature IL-18; however, this binding is not a member of the IL-1 family but plays an important
is of low affinity. In cells that express the coreceptor, termed role in the regulation of cytokines such as IL-1β and TNFα .
IL-18 receptor beta chain (IL-18Rβ), a high-affinity complex
is formed, which then signals. The complex of IL-18 with Interleukin-18 as an Immunoregulatory Cytokine
the IL-18Rα and IL-18Rβ chains is similar to that formed Together with IL-12, IL-18 participates in the Th1 para-
by other members of the IL-1 family with the coreceptor, digm. This property of IL-18 is due to its ability to induce
the IL-1R accessory chain IL-1RAcP. Following the forma- IFNγ either with IL-12 or IL-15. Without IL-12 or IL-15,
tion of the heterodimer, the TIR domains approximate, IL-18 does not induce IFNγ. IL-12 or IL-15 increases the
and it appears that the cascade of sequential recruitment of IL-18Rβ, which is essential for IL-18 signal transduction.
MyD88, the four IRAKs, and TRAF-6 followed by the deg- Without IL-12 or IL-15, IL-18 plays a role in Th2 diseases.159
radation of Iκ B and release of NFκ B are nearly identical as The importance of IL-18 as an immunoregulatory cytokine
IL-18BP
IL-18BP
IL-18
IL-37
B D
A IL-18 C IL-37 E IL-37
FIG. 26.3. Interleukin (IL)-18 Subfamily.

IL-18Rα
IL-18Rα
IL-18Rα
IL-18BP
IL-18Rβ
SIGIRR A: IL-18 binds to the IL-18Rα chain and

recruits the coreceptor IL-18Rβ. The
signaling cascade of the IL-18 receptor
complex is nearly the same as that of IL-
1α and IL-1β, resulting in the expression
of proinflammatory genes. B: The natu-
ral occurring IL-18BP binds IL-18, thus
neutralizing the activity of the cytokine.
C: IL-37 also binds to the IL-18Rα but with
an affinity lower than that of IL-18 bind-
ing to the same receptor. Furthermore,
the binding of IL-37 to IL-18Rα does
TIR TIR TIR TIR not recruit the coreceptor IL-18Rβ and
therefore there is no proinflammatory
signal. The anti-inflammatory properties
MyD88 of IL-37 require single immunoglobulin
signal IL-1–related receptor, which may act as
IRAKs IKKβ
inhibition a “decoy” for MyD88. D: IL-18BP also
binds to IL-37, thus preventing binding
of IL-37 to the IL-18Rα. E: IL-37 binds to
IL-18BP forming a complex, which then
proinflammatory binds to the IL-18Rα, enhancing the
signal anti-inflammatory property of IL-18BP.

is derived from its prominent biologic property of inducing endothelial cells independently of IFNγ. Vascular call adhe-
IFNγ from NK cells. Macrophage-colony stimulating factor sion molecule-1 plays a major role in multiple sclerosis, other
induces human blood monocytes to develop into a subset of autoimmune diseases, as well as in the metastatic process.170
macrophages; these cells express a membrane-bound form
of IL-18.160 Membrane IL-18 is expressed in 30% to 40% Role of Interleukin-18 in Models of
of macrophage-colony stimulating factor–primed macro- Inflammatory Bowel Disease
phages. In contrast, monocytes, dendritic cells, and mono- Inflammatory bowel disease such as Crohn disease is a com-
cytes differentiated into M1 macrophages did not express plex autoimmune disease. Treatment is initially based on
membrane IL-18. Although the expression of membrane immunosuppressive drugs. Not surprisingly, anticytokines
IL-18 is caspase-1 dependent,160 LPS treatment was neces- such as neutralizing monoclonal antibodies to TNFα171 or to
sary for the release of membrane IL-18.160 A major immu- IL-12/23 provide effective treatment for many patients.172,173
noregulating role for IL-18 is on the NK cell. Upon shedding The reduction of IFNγ in Crohn disease is linked to the
of membrane IL-18 into a soluble form, NK cells expressed clinical response to these agents.173 IL-18 is found in affected
CCR7 and produced high levels of IFNγ. As expected, IFNγ intestinal lesions from patients with Crohn disease as a ma-
production was prevented by neutralization of IL-18. This ture protein, but the IL-18 precursor form is present in unin-
mechanism may account for the role of IL-18 as major IFNγ- volved intestinal tissues.174 This observation was confirmed
inducing factor from NK cells and the role of NK cells in the in a similar assessment of mucosal biopsies from patients
pathogenesis of autoimmune diseases. with Crohn disease.175 Antisense RNA to IL-18 decreased
The induction of IFNγ by IL-18 has been studied with IFNγ production in lamina propria mononuclear cells.175
coinducer IL-12. For example, mice injected with the combi- A commonly used mouse model for colitis is DSS, which
nation of IL-18 plus IL-12 develop high levels of IFNγ and die is added to the drinking water and which damages the intes-
with hypoglycemia, intestinal inflammation, and inanition.161 tinal wall. Thus in DSS-induced colitis, the epithelial barrier
In leptin-deficient mice, IL-18 plus IL-12 induce acute pancre- defenses against luminal bacterial products are breeched. In
atitis.162 Several human autoimmune diseases are associated this model, reducing IL-18 with a neutralizing antibody is
with elevated production of IFNγ and IL-18. Diseases such as protective and linked to a reduction in IFNγ.149 Blocking
systemic lupus erythematosus, rheumatoid arthritis, type 1 IL-18 with the IL-18 binding protein (IL-18BP) also reduces
diabetes, Crohn disease, psoriasis, and graft-versus-host dis- colitis induced by antigen sensitization.176 Because genera-
ease are thought to be mediated, in part, by IL-18. tion of active IL-18 requires caspase-1, studies have also been
performed in mice deficient in caspase-1 and subjected to
Proinflammatory Properties of Interleukin-18 DSS colitis. Nevertheless, despite many studies, the role of
IL-18 exhibits characteristics of other proinflammatory cy- caspase-1 in DSS colitis remains unclear. The first study
tokines, such as increases in cell adhesion molecules, nitric showed that mice deficient in caspase-1 were protected.148,177
oxide synthesis, and chemokine production. Blocking IL-18 In addition, treatment of mice with a specific caspase-1
activity reduces metastasis in a mouse model of melanoma; inhibitor was also effective in protecting against the coli-
this is due to a reduction in IL-18–induced expression of vas- tis.178–180 In both studies, the effect of caspase-1 deficiency
cular call adhesion molecule-1.163 A unique property of IL-18 was linked to reduced IL-18 activity, whereas reducing IL-1
is the induction of Fas ligand (FasL), which may account for activity with the IL-1Ra was ineffective.148 In support of the
the hepatic damage that takes place in macrophage activa- role of IL-18 in DSS colitis, inhibition of endogenous mer-
tion syndrome.150,164 The induction of fever, a well-studied prin β to reduce the generation of active IL-18 was protective
property of IL-1α and IL-1β as well as TNFα and IL-6, is in DSS colitis.153
not a property of IL-18. Injection of IL-18 into mice, rab- However, a conundrum has developed whether cas-
bits, or humans does not produce fever.165,166 Unlike IL-1 and pase-1 deficiency is protective or detrimental in DSS coli-
TNFα , IL-18 does not induce cyclooxygenase-2 and hence tis. DSS colitis is not the optimal model for Crohn disease,
there is no production of prostaglandin E2.156,167 IL-18 has as the model is one of rapid loss of the protective barrier
been administered to humans for the treatment of cancer of the intestinal epithelium exposing the lamina propria
in order to increase the activity and expansion of cytotoxic mononuclear cells to a large amount and variety of bacte-
T cells. Not unexpectedly and similar to several cytokines, rial products. Using the same DSS model, mice deficient in
the therapeutic focus on IL-18 has shifted from its use as an the adapter protein inflammasome component ASC experi-
immune stimulant to inhibition of its activity.145,168 enced increased disease, morbidity, and precancerous lesions
Because IL-18 can increase IFNγ production, blocking compared to wild-type mice exposed to DSS.181 Similarly,
IL-18 activity in autoimmune diseases is an attractive thera- mice deficient in caspase-1 died rapidly from DSS com-
peutic target as anti–IL-12/23 reduces the severity of Crohn pared to wild-type mice,182 whereas mice deficient in cas-
disease as well as psoriasis. As discussed subsequently, there pase-12, in which caspase-1 is enhanced, were protected.182
appears to be a role for blocking IL-18 in Crohn disease. Administration of exogenous IL-18 restored mucosal heal-
However, there are several activities of IL-18 that are inde- ing in caspase-1–deficient mice.182 Also, mice deficient in
pendent of IFNγ. For example, IL-18 inhibits proteoglycan NLRP3 were more susceptible to either DSS or 2,4,6-trini-
synthesis in chondrocytes,169 and proteoglycan synthesis trobenzene sulfonic acid (TNBS)-induced colitis and exhib-
is essential for maintaining healthy cartilage. IL-18 also ited decreased IL-1β as well as decreased beta-defensins.183
increases vascular call adhesion molecule-1 expression in Macrophages from NLRP3-deficient mice failed to respond

to muramyl dipeptide.183 Mice deficient in NLRP6 are also unique mechanism is responsible for the higher food intake
more vulnerable to DSS,37,184 and the susceptibility appears in the IL-18–deficient animals appears to be due to a cen-
to be due to lack of sufficient IL-18. tral nervous system loss of appetite control. IL-18– deficient
How to reconcile these data in mouse models of colitis mice eat throughout the day, whereas wild-type mice eat
was addressed by Siegmund185? It is likely that IL-18 being once, nocturnally.
constitutive in the intestinal epithelium has a protective
role in that the cytokine contributes to maintaining the Interleukin-18 as a Protected Cytokine
intestinal barrier. With loss of the barrier, the microbial As stated previously, mice deficient in caspase-1 experience
products stimulate macrophages in the lamina propria and increased disease severity when subjected to DSS colitis and
caspase-1–dependent processing of IL-18 results in inflam- that administration of exogenous IL-18 restored mucosal
mation. In this model, inhibition of IL-18 production in healing in these mice.182 In addition, mice deficient in NLRP3
caspase-1–deficient mice or treatment of wild-type mice were more susceptible to DSS colitis, which is thought to be
with anti–IL-18 antibodies or caspase-1 inhibitors is protec- due to decreased IL-18.183 Mice deficient in NLRP6 are also
tive. Worsening of disease in mice deficient in caspase-1 or more vulnerable to DSS,37,184 and the susceptibility appears
NLRP3 or NLRP6 may lower the levels of active endogenous to be due to lack of sufficient IL-18. Thus, there are a growing
IL-18 needed to protect the epithelial barrier. Similarly, ac- number of studies that support a protective role for IL-18.
tive endogenous IL-1β may be needed to protect to maintain The fact that mice deficient in IL-18 develop a metabolic
the epithelial barrier by inducing growth factors. syndrome–like phenotype is consistent with a role for IL-18
Although it remains unclear why caspase-1 deficiency in homeostasis. A study in age-related macular degenera-
worsens DSS colitis, in humans with Crohn disease, natali- tion is also consistent with a protective role for IL-18. In that
zumab, the antibody that blocks the very late antigen-4, is study, drusen, which is mixture of complement-derived and
highly effective in treating the disease. Very late antigen-4 is apolipoproteins and lipids, was shown to activate NLRP3
the α4 subunit of the β -1 integrin. Anti–very late antigen-4 and induce the production of mature IL-1β and IL-18.188 In a
binds to the surface of macrophages and other myeloid cells, mouse model of “wet” age-related macular degeneration, the
and prevents the binding of these cells to the very late anti- disease was worse in mice deficient in NLRP3 but not in IL-
gen-4 receptor on endothelial cells known as vascular cell 1RI–deficient mice.188 Therefore, IL-18 rather than IL-1α or
adhesion molecule-1. Thus, the antibody disables the func- IL-1β were protective; upon administration of IL-IL-18, the
tion of vascular cell adhesion molecule-1 and prevents the disease severity improved. Taken together, there is a case for
passage of macrophages and other myeloid cells into tissues IL-18 being a protective rather than inflammatory cytokine.
such as the intestine in Crohn disease and the brain in mul-
tiple sclerosis. Because IL-18 induces vascular cell adhesion Interleukin-18 Binding Protein
molecule-1, blocking IL-18 would also reduce the passage of The discovery of the IL-18BP took place during the search for
cells through the endothelium into to intestine. the soluble receptors for IL-18.189 IL-18BP is a constitutively
secreted protein with an exceptionally high affinity for IL-18
Interleukin-18, Hyperphagia, and the (400 pM) (see Fig. 26.3B). Present in the serum of healthy hu-
Metabolic Syndrome mans at a 20-fold molar excess compared to IL-18,190 IL-18BP
Whereas there is no constitutive gene expression for IL-1β may contribute to a default mechanism by which a Th1 re-
in freshly obtained human peripheral blood mononuclear sponse to foreign organisms is blunted in order to reduce
cells (PBMC), the same cells express constitutive mRNA triggering an autoimmune responses to a routine infection.
for IL-18.146 In western blot analysis from the same cells, Although IL-18BP is readily secreted, it falls into the func-
the IL-18 precursor was present but not the IL-1β precursor. tional category of being a shed soluble receptor. As shown in
Similar observations were also made in mice.146 These fi nd- Figure 26.3B, IL-18BP contains only one immunoglobulin G
ings suggest that IL-18 may act as regulator of homeostasis. domain, whereas the type II IL-1 receptor contains three do-
Starting at age 16 weeks of age, IL-18– deficient mice start mains. In this regard, the single immunoglobulin G domain
to overeat, become obese, and exhibit lipid abnormalities; of IL-18BP is similar to SIGIRR, which also has a single im-
there is increased atherosclerosis, insulin resistance, and munoglobulin G domain and also functions as a decoy recep-
diabetes mellitus reminiscent of the metabolic syndrome.186 tor. The salient property of IL-18BP in immune responses is in
IL-18Rα –deficient mice also develop a similar phenotype. downregulating Th1 responses by binding to IL-18 and thus
The higher body weight is attributed to enhanced food in- reducing the induction of IFNγ.159 Because IL-18 also affects
take, in which the IL-18–deficient mice begin to diverge Th2 responses, IL-18BP also has properties controlling a Th2
from wild-type animals at a relatively early age, and to reach cytokine response.159 IL-18BP has a classic signal peptide and
values 30% to 40% higher than that of wild-type mice. therefore is readily secreted. Serum levels in healthy subjects
Others have observed similar fi ndings.187 A striking fi nding are in the range of 2,000 to 3,000 pg/mL compared to the levels
was an increase of more than 100% in the percent of adi- of IL-18 in the same sera of 80 to 120 pg/mL/mL. Moreover,
pose tissue in the IL-18–deficient animals that was accom- IL-18BP binds IL-18 with an affinity of 3 to 5 nM,189 an affin-
panied by fat deposition in the arterial walls. The insulin ity significantly higher than that of IL-18Rα. Because a single
resistance in these mice is corrected by exogenous recom- IL-18BP molecule binds a single IL-18 molecule, one can calcu-
binant IL-18. Mice deficient in IL-18 respond normally to a late bound versus free IL-18 in a mixture of both molecules.190
challenge with exogenous leptin, suggesting that expression If one examines immunologically mediated diseases where
of the leptin receptor is unaffected. The unexpected and IFNγ plays a pathologic role such as Wegener granulomatosis

and systemic lupus erythematosus, one must consider the not been documented with any certainty. It is likely, how-
level of free IL-18 compared to IL-18 bound to IL-18BP. In ever, that similar to IL-1α and IL-33, with loss of membrane
fact, in these diseases both IL-18BP and IL-18 are high,191,192 integrity upon cell death, the IL-37 precursor exits from
but the level of IL-18BP is not sufficiently high enough to the cell. The recombinant form of the IL-37 precursor sup-
neutralize IL-18 and therefore, the level of free IL-18 is higher presses LPS-induced IL-1β, IL-6, and TNFα . However, this
than in healthy subjects. In macrophage activation syndrome is observed primarily in macrophages that have been differ-
where IFNγ plays a pathologic role, both IL-18BP and IL-18 entiated into the M1 phenotype by 5 days in the presence of
are also high but the clinical and hematologic abnormalities granulocyte macrophage-colony stimulating factor. There
correlate with elevated free IL-18.164 are two consensus sequences (A-X-D) in N-terminal domain
A unique property of IL-18BP is that the molecule also of IL-37: IHD and LED. A recombinant form of IL-37 with
binds IL-37193 and in doing so, enhances the ability of IL- an N-terminus nine amino acids from the IHD site is active
18BP to inhibit the induction of IFNγ by IL-18. IL-37 binds in suppressing LPS-induced TNFα and IL-6. Whether this
to the IL-18Rα with a very low affinity but in mice express- short form of recombinant IL-37 exists in nature is unclear.
ing human IL-37, a profound anti-inflammatory effect is
observed,194 particularly of LPS-induced cytokines and Interleukin-37 Reduces Interleukin-1b and
dendritic cell maturation.194 Human IL-37–expressing mice Lipopolysaccharide-Induced Inflammation in Vivo
are also resistant to colitis.195 Thus, the anti-inflammatory A mouse homologue for human IL-37 has not been identi-
property of IL-37 can be affected by the concentration of IL- fied. Therefore, to define the in vivo functional role of IL-37,
18BP. As the concentration of IL-18BP increases and binds a strain of transgenic mice was generated.194 The full-length
IL-37, there is the possibility that IL-37 becomes less avail- IL-37 complementary DNA was inserted into a vector using
able as an anti-inflammatory cytokine. Indeed, this has been the standard cytomegalovirus (CMV) promoter for consti-
observed in mice injected with IL-18BP. At low dosing of tutive expression of the transgene in all cells. Both heterozy-
IL-18BP, there is reduced inflammation in a model of rheu- gous and homozygous IL-37 transgenic mice (IL-37 tg) mice
matoid arthritis but as the doing of IL-18BP increases, the breed normally and exhibit no obvious phenotype. Despite
anti-inflammatory properties of IL-18BP are lost.196 the presence of the CMV promoter, the IL-37 transcript is
IL-18BP is highly regulated at the level of gene expression not constitutively expressed in the tissues of the IL-37 trans-
and unexpectedly, IFNγ increases gene expression and syngenic mice. The failure to express IL-37 is likely due to a
thesis of IL-18BP.197,198 Therefore, IFNγ driving an increase in functional instability sequence found in IL-37, which limits
the natural and potent inhibitor of IL-18 falls into the category the half-life of IL-37 mRNA.51 Nevertheless, upon stimula-
of a negative feedback loop. The concept is supported by clini- tion with LPS or IL-1β, levels of IL-37 increase after 4 to 24
cal data showing that patients being treated with IFNα for hours. Once the transcript is present, the IL-37 precursor
hepatitis have elevated levels of IL-18BP.199,200 IL-27, like IFNγ, can be found in peripheral blood cells taken from the trans-
functions as both a pro- as well as an anti-inflammatory cyto- genic mice.203
kine and both may accomplish their roles as anti-inflamma- IL-37 transgenic mice are protected against LPS challenge
tory cytokines at the level of increased production of IL-18BP. compared to similarly challenged wild-type mice. IL-37
In the skin, IL-27 also acts through a negative feedback loop transgenic mice exhibit significantly less hypothermia, aci-
for inflammation. IL-27 is acting, as is IFNγ, by induction of dosis, hyperkalemia, hepatitis, and dehydration.194 In addi-
IL-18BP gene expression and synthesis.201 tion, circulating cytokines are significantly reduced as well
as cytokines induced in whole blood cultures and in lung
Viral Interleukin-18 Binding Protein and spleen cell homogenates. In addition to LPS-induced
Natural neutralization of human IL-18 by IL-18BP takes cytokines, whole blood cultures from IL-37 transgenic mice
place during a common viral infection. In Molluscum produce significantly less IL-6 and TNFα when stimulated
contagiosum infection, characterized by raised but bland by IL-1β or the combination of IL-12 plus IL-18. The anti-
eruptions, there are large numbers of viral particles in the inflammatory activity of IL-37 was not limited to a reduc-
epithelial cells of the skin, but histologically there are few tion of the cytokines and chemokines of innate immunity.
inflammatory or immunologically active cells in or near Dendritic cells isolated from the spleen of IL-37 transgenic
the lesions. Clearly, the virus fails to elicit an inflammatory mice upon LPS stimulation revealed a marked reduction
or immunologic response. Amino acid similarity exists be- (75% and 89%) in expression of CD86 and MHC II, respec-
tween human IL-18BP and a gene found in various members tively.194 The total numbers of dendritic cells, macrophages,
of the poxviruses; the greatest degree of homology is found NK cells, and CD4 + T cells were similar in all strains and
to be expressed by Molluscum contagiosum gene.202 The abil- experimental conditions.
ity of viral IL-18BP to reduce the activity of mammalian
IL-18 likely explains the lack of inflammatory and immune A Role for Interleukin-37 During Experimental Colitis
cells in the infected tissues and provides further evidence for IL-37 transgenic mice have been subjected to DSS-induced
the natural ability of IL-18BP to interfere with IL-18 activity. colitis. Despite the presence of a CMV promoter to drive
expression of IL-37, mRNA transcripts were not present in
colons in the resting state.195 Expression was observed only
Interleukin-37 upon disruption of the epithelial barrier, with a six- to sev-
IL-37 was formerly termed IL-1F7. IL-37 lacks a signal pep- enfold increase on days 3 and 5 after continuous exposure
tide and has a caspase-1 site, but the secretion of IL-37 has to DSS. During the development of colitis, clinical disease

scores were reduced by 50% and histologic indices of colitis IL-37 does not act as a classical receptor antagonist for IL-18
were one-third less in IL-37 transgenic mice compared with in that the ability of recombinant IL-18 to induce IFNγ is
wild-type counterparts. Reduced inflammation was associ- not inhibited by high concentrations of IL-37. However, in
ated with decreased leukocyte recruitment into the colonic the presence of low concentrations of IL-18BP, recombinant
lamina propria. In addition, release of IL-1β and TNFα from IL-37 modestly reduces IL-18–induced IFNγ.193 The concept
ex vivo colonic explant tissue was decreased 5- and 13-fold, that IL-37 binds to the IL-18Rα and reduces cytokine pro-
respectively, compared with wild-type mice, whereas IL-10 duction is supported, in part, with the finding embryonic
was increased 6-fold. However, IL-10 was not required fibroblasts from mice deficient in IL-18Rα produce 10-fold
for the anti-inflammatory effects of IL-37 because IL-10 more IL-6 in response to IL1β than do wild-type embryonic
receptor antibody blockade did not reverse IL-37–mediated fibroblasts.208 In addition, silencing of IL-18Rα in primary
protection. Mechanistically, IL-37 originating from hemato- human blood monocytes results in a fourfold increase in
poietic cells was sufficient to exert anti-inflammatory effects the secretion of LPS-induced IL-1β, IL-6, IFNγ, and CD40
because wild-type mice reconstituted with bone marrow ligand.208 Thus, the seemingly paradoxical hyperresponsive
from IL-37 transgenic mice were protected from colitis. state in cells deficient in IL-18Rα supports the concept that
IL-18Rα participates in both pro- and anti-inflammatory
A Nuclear Role for Interleukin-37 responses and that the endogenous ligand IL-37 engages the
In stable transfectants of human IL-37 in RAW macro- IL-18Rα to deliver an inhibitory signal.
phages stimulated with LPS, levels of TNFα , IL-1α , IL-6,
as well as the chemokine MIP-2 were substantially reduced Role of Single Immunoglobulin Interleukin-1–Related
(72% to 98%) compared with LPS-stimulated cells trans- Receptor in the Anti-inflammatory Property of
fected with the empty plasmid.204 Similar to IL-1α and Interleukin-37
IL-33, IL-37 translocates to the nucleus following stimula- The mechanism by which an IL-1β or an LPS signal is sup-
tion.204 In mouse RAW macrophages stably expressing IL-37, pressed by IL-37 requires an understanding of SIGIRR
the mature carboxyl-terminal was detected in the nucleus. (Fig. 26.4G). The IL-1RAcP serves as the coreceptor for
Furthermore, a specific caspase-1 inhibitor markedly re- IL-1α , IL-1β, IL-36α , IL-36β, IL-36γ, and IL-33, each a proin-
duced nuclear entry of IL-37.204 The data demonstrate that flammatory cytokine. However, in the IL-1 family of recep-
IL-37 translocates to the nucleus after caspase-1 processing tors, three coreceptors contain unusually long intracellular
and may act as a transcriptional modulator reducing the domains. These are SIGIRR (Fig. 26.4G), a variant of the IL-
production of LPS-stimulated proinflammatory cytokines, 1RAcP termed IL-1RAcPb (Fig. 26.4F) and receptors termed
consistent with IL-37 being an anti-inflammatory member “three immunoglobulin IL-1–related receptors” (TIGIRRs).
of the IL-1 family. As shown in Figure 26.4H, there are two TIGIRRs: TIGIRR-1
IL-37 was identified in a proteomics-based search for pro- and TIGIRR-2. IL-1RacPb is expressed only in the brain,
teins that interacted with Smad3.205 To test for a functional and TIGIRR has limited expression. However, SIGIRR is
interaction of Smad3 with IL-37, IL-37 was transfected into expressed in most cells. The TIR domain of the three core-
A549 cells. IL-37 colocalized with phospho-Smad3 and ceptors is also different from that of other members of the
was found in perinuclear and cytosolic regions, and a IL- IL-1 coreceptor family in that the TIR domain contains an
37-Smad3 complex was observed.194 A specific inhibitor of amino acid sequence different from that of wild-type TIR.209
Smad3 reversed the inhibition of IL-6 and IL-1β expression As shown in Figure 26.2B, IL-1RAcPb forms the expected
in RAW cells stably transfected with IL-37. In stable human complex with IL-1 and IL-1R1 but does not recruit MyD88
macrophage lines expressing IL-37, depletion of Smad3 by or phosphorylate IRAK4.209 Therefore, most IL-1 signaling is
lentiviral introduction of short hairpin RNA that inhib- arrested. But as some genes are increased in response to the
its Smad3 expression prevented the ability of IL-37 to re- formation of the IL-1RI/IL-1RAcPb complex, partial IL-1
duce IL-1β - or LPS-induced production of IL-8, IL-6, and signaling must take place. Nevertheless, IL-1RAcPb func-
TNF. These in vitro findings were confi rmed in vivo. IL-37 tions as an inhibitory receptor chain but only in the brain.
transgenic mice were pretreated intranasally with a Smad3 Mice deficient in IL-1RAcPb exhibit a normal inflammatory
specific small interfering RNA and then challenged with response in the periphery but greater neurodegeneration in
intranasal LPS. The reduction of lung cytokines in IL-37 the brain. As such, IL-1RAcPb could play a role in chronic
transgenic mice was reversed in transgenic mice with a lung inflammatory responses in the brain by “buffering” IL-1–
knockdown of Smad3.194 mediated neurodegeneration.
Like IL-1RAcPb, SIGIRR contains the same amino acids
Role of Interleukin-18Ra for Interleukin-37 differences in its TIR domain, termed TIRb. Compared to
From the first reports on IL-37, it was observed that the re- wild-type TIR, TIRb likely has reduced binding of MyD88.209
combinant forms bound to the IL-18Rα .132,206 This is shown In addition to an altered TIR domain, SIGIRR has a car-
in Figure 26.3C and E. The binding of IL-37 to IL-18Rα has boxyl extension of 140 amino acids. Carboxyl extensions
also been observed in cells from IL-37 transgenic mice using are also present in IL-1RAcPb as well as the two TIGIRRs.
immunofluoresence, immunoprecipitation, and fluores- TIGIRR-2, which is associated with an X-linked cognitive
cence resonance energy transfer analysis.207 IL-37 specifi- deficiency, is apparently independent of IL-1 function. Little
cally binds to the third domain of the IL-18Rα .193 Despite is known whether these C-terminal segments contribute to
these studies showing binding of IL-37 to the IL-18Rα chain, the inhibitory properties of these receptors. Nevertheless,

A B IL-1β C
IL-1RAcP
sIL-1RII
sIL-1RII
sIL-1RI
IL-1Ra
IL-1β
IL-1RII IL-1β TIGIRR-1/2

IL-1β
IL-1RAcPb
IL-1RAcP
D E F G SIGIRR H
no signal no signal
TIR TIRb TIRb TIRb

reduced
signal
FIG. 26.4. Anti-Inflammatory Receptors of the Interleukin (IL)-1 Receptor Family. A: Soluble (extracellular) IL-1RII
(sIL-1RII) binds and neutralizes either the precursor or mature form of IL-1β. B: A complex of IL-1β, sIL-1RII, and sIL-
1RAcP neutralizes IL-1β activity. C: Soluble (extracellular) IL-1RI (sIL-1RI) binds IL-1Ra and prevents IL-1Ra from its func-
tion as a receptor antagonist. Similarly, IL-1α can bind to the sIL-1RI (not shown). D: IL-1RII expressed on the membrane
binds mature IL-1β acting as a decoy receptor, preventing IL-1β from binding to its signaling receptor. E: A complex of
membrane IL-1RII, IL-1β, and membrane IL-1RAcP also prevents IL-1β from signaling. F: The IL-1RAcPb is expressed in
brain tissue (see Fig. 26.2B). G: Single immunoglobulin IL-1–related receptor is expressed in myeloid cells and epithelial
cells and function in damped inflammation due to the TIRb domain. H: Three immunoglobulin IL-1–related receptor is
also an anti-inflammatory member of the IL-1 due to the toll-IL-1-receptor b domain.
it seems likely that the alternative sequence in the TIRb signal of SIGIRR is established in several mouse models of
domain of SIGIRR may act as a partial decoy for MyD88. inflammation in which SIGIRR-deficient mice exhibit more
MyD88 is phosphorylated upon TLR4 as well as IL-1β and inflammation compared to wild-type control mice.210 In
IL-18 signaling, and results in downstream phosphorylation differentiated human blood M1 macrophages, recombinant
of IRAK-4. In cells expressing SIGIRR and activated by IL-37 IL-37 suppresses LPS-induced TNFα and IL-6 production
binding to the IL-18Rα , the signal from either IL-1 or LPS 50% to 70%. A source of IL-18BP in this culture may be fetal
initiates phosphorylation of MyD88. However, the decoy ef- calf serum. During the differentiation of macrophages into
fect by the mutated TIRb of SIGIRR reduces the degree of the M1 subset by granulocyte macrophage-colony stimulat-
phosphorylation of MyD88 and thus the phosphorylation ing factor, it is possible that SIGIRR expression increases
of IRAK-4. The reduction, however, is partial. Indeed, the whereas the level of IL-18Rβ decreases (see Fig. 26.3A). In
suppression by IL-37 added to blood macrophages is in the the absence of IL-18Rβ, a proinflammatory complex is not
range of 20% to 50% and unlike the total loss of activation formed with IL-18Rα and thus IL-37 binding to IL-18Rα
by a deficiency in MyD88. may recruit or activate SIGIRR. Thus, expression of SIGIRR
Upon binding to the IL-18Rα , the IL-37 precursor may ac- and the absence of IL-18Rβ would best explain the inhibi-
tivate SIGIRR and provide a negative signal (see Fig. 26.3C). tory properties of recombinant IL-37 reducing the response
An inhibitory signal from IL-37 is enhanced by a low con- to LPS induction of IL-6 and TNFα.
centration of IL-18BP.193 As shown in Figure 26.3D, IL-18BP
binds IL-37193 and likely presents the complex of the cyto-
kine with the binding protein to the IL-18Rα (see Fig. 26.3E). INTERLEUKIN-36 SUBFAMILY
Because A549 cells express SIGIRR, it is likely that the in- The IL-1 family members IL-1F5, IL-1F6, IL-1F8, and IL-1F9
hibitory signal from IL-37 activates SIGIRR or alternatively are now termed IL-36Ra, IL-36α , IL-36β, and IL-36γ, respec-
IL-37 recruits SIGIRR as the accessory chain. The inhibitory tively.211 As shown in Figure 26.5A, each member of the IL-36

A IL-36 α /β/γ B IL-36Ra C IL-38
IL-1RAcP
IL-1RAcP
IL-1RAcP
IL-36R
IL-36R
IL-36R
TIR TIR TIR TIR TIR TIR
MyD88
no signal no signal
IRAKs IKKβ
FIG. 26.5. Interleukin (IL)-36 Subfamily. A: IL-36α, IL-36β, or IL-36γ bind to the IL-36R and recruit the
coreceptor IL-1RAcP. The heterodimeric IL-36 receptor complex results in a close approximation of the
toll-IL-1-receptor (TIR) domains on each receptor chain (arrows), resulting in the binding of intracellular
MyD88 to the complex with phosphorylation of MyD88. Subsequent phosphorylations of IRAKs and IKKβ,
increased NF-κB, and activating protein-1 translocation to the nucleus take place, followed by expression
of proinflammatory genes. B: IL-36Ra binds to IL-36R. There is no recruitment of the coreceptor IL-1RAcP,
no approximation of the TIR domains, and no signal. C: IL-38 binds to IL-36R but there is no signal and acts
similar to IL-36Ra.
subfamily binds to the IL-1Rpr2, now termed IL-36R.212 the various N-termini of the IL-36 subfamily; nevertheless,
Because IL-38 also binds to the IL-36R (Fig. 26.5C),133 IL-38 each member has a unique N-terminus with a different
is included in the IL-36R subfamily. The IL-36 subfamily levels of biologic activity.9 What determines the N-terminus
is closely related to the IL-1 subfamily because, similar to with optimal biologic activity in the IL-36 subfamily? Each
the IL-1α and IL-1β and IL-33, the IL-36R forms a signaling member of the IL-36 subfamily contains the IL-1 fam-
complex with the IL-1RAcP (Fig. 26.5A).122,212 Thus of the 11 ily consensus sequence of A-X-D. The aspartic acid is not
members of the entire IL-1 family, 6 members use IL-1RAcP the recognition amino acid for caspase-1 or caspase-3, but
as the coreceptor for signal transduction. rather participates in the stabilization of the fi rst beta sheet
of the three-dimensional structure that characterizes the
entire IL-1 family.
Interleukin-36 The “A” of the consensus sequence is for any aliphatic amino
The IL-36R is the ligand binding chain and therefore is acid, for example, leucine or isoleucine. Nine amino acids be-
comparable to the ligand binding IL-1R1 and the IL-18Rα . fore the “A” of the consensus sequence is the N-terminal site,
However, two members of the IL-36 subfamily bind to the which results in the cytokine with the greatest activity.9 For
IL-36R but do not signal, the IL-36 receptor antagonist example, the biologic activity of IL-36γ increases by a factor
(IL-36Ra) and IL-38 (see Fig. 26.5B and C). As such, these of 1,000 when the N-terminus is at the site nine amino acids
function as receptor antagonists.9,133 An unusual property of before the consensus sequence and in the case of the IL-36β,
IL-38 is that low concentrations (1 to 10 ng/mL) are able to there is a 10,000-fold increase.9 The nine amino acid forward
reduce the activity of endogenous IL-36,133 whereas in the site from the consensus sequence is not only the N-terminus
case of IL-1Ra, higher concentrations are required to prevent for the agonist members of the IL-36 family but also the IL-36
the activating of endogenous IL-1α or IL-1β. receptor antagonist (IL-36Ra),9 which increases from a low
None of the members of the IL-36 subfamily have a sig- level of blocking of IL-36 family ligands to a high degree of
nal peptide, indicating the generation of the N-terminus blockade. It is unclear what specific protease cleaves at this
and secretion via the Golgi. In addition, each member of site as the amino acid is different for each member of the
the IL-36 subfamily has an unusually short propiece com- IL-36 subfamily. Moreover, the site for the N-terminus of the
pared to those of the IL-1α , IL-1β and IL-33 (see Fig. 26.1). IL-36Ra (valine) is but one amino acid from the N-terminal
Like IL-1β and IL-18, there is no true caspase-1 cleavage precursor methionine and yet IL-36Ra with an N-terminus at
site for generation of an N-terminus in the IL-36 subfam- the valine site is 10,000-fold more potent that the IL-36 precur-
ily. It remains unknown which specific proteases generate sor. It is also an unusual situation that proteases that usually

are inflammatory in processing members of the IL-1 family in into the category of being an alarmin in the skin, particu-
that case of IL-36Ra generate an anti-inflammatory molecule. larly due to infection.221 There is also a role for IL-36 in the
production of IL-17: studies suggest that each of the IL-36
ligands is expressed in the skin and dependent on IL-22.222
Interleukin-36a, b, and g, Proinflammatory Members of Furthermore, the expression of IL-36 ligands in the psoriatic
the Interleukin-36 Subfamily skin correlated with IL-17.222 Similar to other models in the
IL-36 α was highly expressed in the murine model of glo- IL-1 family, auto- and coinduction accounts for a role in a
merulonephritis213 where the cytokine was localized to the pathologic process.
kidney epithelium and also in CD3 T cells surrounding the Overexpression of IL-36 in mice results in inflamma-
tubules. IL-36 α is also found in the kidneys of the MRL/ tory skin lesions that resemble psoriasis in humans, as re-
lupus, nephritic syndrome, and streptozotocin-induced viewed in Towne and Sims.223 Likewise, mice deficient in
diabetic models.213 IL-36γ increases IL-8, CXCL3, and the endogenous IL-36Ra results in a severe lesion similar to
Th17 chemokine CCL20 in human lung fibroblasts, 214 and that of humans with pustular psoriasis. The role of IL-36
thus may account for acute neutrophilic lung infl amma- in pustular psoriasis may include IL-1α , as humans with
tion. In addition to CD4 + T cells, human articular chon- pustular psoriasis responds to an antibody that neutralizes
drocytes and synovial fibroblasts express the IL-36R.215 In IL-1α . Both IL-36 and IL-1α are found in the keratinocytes
chondrocytes, there is also constitutive gene expression of in healthy skin.
IL-36β. Following stimulation with IL-1β or TNFα , levels
of the IL-36β precursor rise intracellularly but the cyto-
kine is not secreted. Although IL-36β levels were detected A Role for Interleukin-36R in Metabolic Regulation
in the joint fluids of patients with rheumatoid arthritis as Obesity is characterized by chronic low-grade inflamma-
well as in serum samples, there was no correlation with tion originating from expanding adipose tissue. Human
disease severity.215 It is likely that IL-36 ligands are func- adipogenic tissue levels of IL-36 α is primarily present in
tional only when released from dying cells and can be adipose tissue resident macrophages and induced by in-
processed extracellularly by enzymes present in infl am- flammation; however, IL-36β is absent.224 IL-36 α , but
matory conditions such as the joints of patients with rheu- not IL-36β, reduces adipocyte differentiation, as shown
matoid arthritis. It unclear to what extent IL-36β plays a by a significant decrease in PPARγ gene expression. Both
role in joint disease, although constitutive expression in IL-36 α and IL-36β induce inflammatory gene expression
primary chondrocytes may indicate a role for the cytokine in mature adipocytes.224 Therefore, IL-36 α and IL-36β are
in osteoarthritis.215 present in adipose tissue and are involved in the regula-
High levels of this cytokine are found in mouse embry- tion of adipose tissue gene expression. Importantly, IL-36 α
onic tissues rich in epithelial cells.216 In humans, IL-36 α is inhibits PPARγ expression, which may lead to reduced adi-
observed in keratinocytes, not fibroblasts, and is thought pocyte differentiation suggesting metabolic effects of this
to contribute to the infl ammation of psoriasis. Upon cytokine.
forming the heterodimer with IL-36R and IL-1RAcP, IL- Although IL-36Ra is known to occupy the IL-36R and
36 α activates NF-κ B similar to that of IL-1β. 212 In addi- act as a receptor antagonist, earlier studies revealed that
tion to NF-κ B activation, IL-36 α also activates MAPK, IL-36Ra inhibited the induction of IL-1β by LPS. The role
JNK, and ERK1/2. 212 In the mouse, bone marrow–derived of IL-36Ra was also examined in the brain. IL-36Ra in-
dendritic cells and CD4 + T cells express IL-36 receptors jected into the rat brain induced IL-4 and also in glial
in health. In a comparison with IL-1 as a stimulant, the cells in vitro. 225 Moreover, the reduction in LPS-induced
three IL-36 ligands are more active in inducing IL-1β, IL-1β was not observed in cells deficient in IL-4 and also
IL-6, IL-12, TNFα , and IL-23. 217 In addition, IL-36 li- not observed in cell deficient in SIGIRR. 225 However,
gands induced the production of IFNγ, IL-4, and IL-17 these unique properties of IL-36Ra were not observed in
from CD4 + T cells. Not unexpectedly, cytokines induced peripheral monocytes or dendritic cells but only in the
by IL-36 lands were prevented by 100- to 1,000-fold excess brain.
IL-36Ra. 217
Role of Interleukin-36 in Human Disease
Interleukin-36 in Psoriasis The importance of any cytokine in human biology can be
Several studies implicate IL-36 ligands in the pathogenesis found in mutations that result in a profound clinical pic-
of psoriasis.218–220 IL-36γ is highly expressed in keratino- ture. In the case IL-1, a mutation in the naturally IL-1Ra
cytes from healthy human skin and increases upon stimu- (see Fig. 26.5B) results in severe systemic inflammation
lation with TLR polyI:C.221 Furthermore, polyI:C induced with erosive bone lesions, sterile meningitis, and death;
caspase-3, which resulted in cell death and the release of the syndrome is called deficiency of IL-1Ra.226,227 In case
IL-36γ. Unexpectedly, stimulation of IL-36γ gene expression of the IL-36 family, persons with a mutation in the natu-
was dependent on caspase-1.221 The caspase-1 dependency rally occurring IL-36Ra suffer with a severe form of pus-
may be due to IL-18 as this cytokine is constitutively pres- tular psoriasis.228,229 These human studies are consistent
ent in keratinocytes as is IL-1α. With the release of IL-36γ with the data from transgenic mice overexpressing IL-36 α
by polyI:C and the subsequent death of the cell, IL-36γ falls in the skin and with the ability of IL-36Ra to suppress the

severity of the inflammation.218 In mice overexpressing tokines IL-22 and IL-17. The dose-response suppression of
IL-36a in the skin, crossing the mice to generate a strain of IL-38 as well as that of IL-36Ra of Candida-induced IL-22
mice heterozygous for natural IL-36Ra knockout results in and IL-17 was not that of the classic IL-1 receptor antago-
worsening of the skin lesions.218 nist, because low concentrations were optimal for inhibit-
ing IL-22 production.133 These data provide evidence that
IL-38 binds to the IL-36R, as does IL-36Ra, and that IL-38
Interleukin-38 and IL-36Ra have similar biologic effects on immune cells
IL-38 is the name for the IL-1 family member 10. During by engaging the IL-36 receptor.
the nomenclature revision of the IL-1 family in 2010,211 the
term IL-38 was assigned to IL-1F10 without any known bio-
logic function. Since then, IL-38 has been shown to bind to CONCLUSION
the IL-36 receptor (formerly IL-1Rrp2) (see Fig. 26.5C).133 In conclusion, IL-1 family of cytokines is one of the most
In order to fi nd the receptor for IL-38, each member of the important families of proinflammatory mediators, that
IL-1 receptor family was immobilized and recombinant have profound local and systemic effects. The importance of
IL-38 precursor containing 152 amino acids was added and these molecules has been demonstrated in homeostasis, as
binding assessed using an antibody to the ligand. IL-38 well as in autoinflammatory diseases, and treatment strate-
bound only to the IL-36 receptor, as did IL-36Ra.133 To as- gies based on modulation of IL-1 cytokines have begun to be
sess the biologic function of IL-38, heat-killed Candida albi- employed successfully. Furthermore, important and novel
cans was used to stimulate memory T-lymphocyte cytokine biologic roles are currently being described for the members
production in freshly obtained human peripheral blood of IL-1 family of cytokines, and it is to be expected that this
mononuclear cells from healthy subjects. The addition of will represent one of the most dynamic areas of research in
recombinant IL-38 inhibited the production of T-cell cy- immunology in the coming years.

Tumor Necrosis Factor–Related

CHAPTER
27 Cytokines in Immunity
Carl F. Ware
INTRODUCTION TUMOR NECROSIS FACTOR LIGAND FAMILY

Lymphocytes are highly mobile, transiting the vascular and The TNF-related cytokines are type II transmembrane pro-
lymphatic circulatory systems with limited stopovers in or- teins (intracellular N-terminus) with a short cytoplasmic
ganized lymphoid tissues (lymph nodes, spleen, and Peyer’s tail (15 to 25 residues in length) and a larger extracellular
patches) in which they commune to initiate immune re- region (∼150 amino acids) containing the signature TNF
sponses. Cytokines, both secreted and membrane-anchored homology domain where the receptor binding sites are lo-
proteins, serve as the communication media of the immune cated (Fig. 27.1). The TNF homology domain assembles into
system. Cytokines in the tumor necrosis factor (TNF) super- trimers, the functional unit of the ligand. Atomic analysis
family (TNFSF) help orchestrate the development, homeo- of several members of the family1–5 revealed the ligands have
stasis, and effector actions of cells in the immune system. a highly conserved tertiary structure folding into a β sheet
The diversification of the TNFSF is evidenced by its roles in sandwich, yet amino acid sequence conservation is limited to
regulating cells of neuronal, skeletal, and ectodermal origin. < 35% among the family members. The conserved residues
This chapter’s focus is on members in the TNFSF that are defining this superfamily are primarily located within the
primarily involved in regulating immunity. internal β strands that form the molecular scaffold, which
The ligands belonging to the TNFSF initiate intercel- promote assembly into trimers. The residues in the loops
lular communication by binding specific receptors on the between the external β -strands are variable and in specific
surface of the receiving cell. The TNF-related ligands are loops make contact with the receptor. Although most of the
membrane-bound and require cell-cell contact to initiate TNF ligands self-assemble into homotrimers, heterotrimers
signaling, although some ligands may be secreted in soluble can also form between LTα and LTβ6 and APRIL and BAFF.7
form influencing cellular responses at locations distant from The stoichiometry of the LT heterotrimer is 1:2 (LTαβ2),
the source of production. The TNF receptors (TNFRs) form which imparts its distinct receptor specificity from the LTα
a corresponding superfamily of cognate membrane proteins homotrimer. Interestingly, complement component C1q
that initiate intracellular signaling pathways that influence and several related proteins are structurally related to the
cellular growth, differentiation, and survival. Each ligand- TNF family, containing a TNF homology domain, a rather
receptor pair forms a “system” with over 40 distinct ligand- surprising finding given the apparent functional divergence
receptor systems. Many of the ligands and receptors engage between the complement and TNF systems.8 Alternate splic-
more than one cognate, thereby forming “circuits” of signaling can also generate distinct ligands, such as the splice form
ing systems, which often function together as coordinated or joining TWEAK and APRIL (TWE-PRIL),9 the alternate
integrated networks with other cytokines to regulate a specific ligands for ectodysplasin receptor,10 and a cytosolic form of
cellular process. The signals delivered to the cytosol activate LIGHT.11
transcription factors such as NF-κB and AP1, which in turn The genetic organization of TNF ligands is highly con-
initiate the expression of hundreds of genes that alter cellular served and typically encoded in three or four exons, with
differentiation. Additionally, some TNFRs activate cell death the fourth exon encoding most of the extracellular TNF
pathways, both apoptotic and necrotic, terminating cellular homology domain. The genes encoding TNF, LTβ, and
life. Genetic mutations have revealed the importance of indi- LTα reside adjacent to each other in a compact loci in the
vidual TNFSF members in human immune responses and de- class III region of the major histocompatibility complex
velopment. Moreover, the beneficial effect of TNF antagonists (MHC) (in humans at chromosome 6p21) sandwiched
in patients with certain autoimmune diseases brings the need between the antigenic-peptide presenting MHC proteins
to understand the complexities of the TNFSF into sharp focus. encoded by MHC class I and II genes.6,12 Three other ge-
The nomenclature of the TNFSF is a bit of a morass for netic clusters of TNF-related cytokines are found within
students new to the field, although introduction of a nu- the corresponding MHC paralogous regions, located on
merical system standardizing gene names (www.genenames. chromosomes 1, 9, and 19. These genetic clusters share a
org/) helps in accessing genomic databases. However, the remarkably conserved gene structure and transcriptional
use of more common, often colorful, acronyms continues. orientation, and a similar function linked to regulating
The name, rank, and genomic identification of each ligand cellular immune responses.11 The evolutionary pressures
(Table 27.1) and receptor (Table 27.2) is tabulated along with retaining this genetic configuration of the MHC in gen-
additional pertinent information for human and mouse genes. eral is encompassed in the paralogy theory of genome
659

Tumor Necrosis Factor Superfamily Tumor Necrosis Factor Receptor

TABLE 27.1 TABLE 27.2
Chromosomal Locations Superfamily
Chromosomal Location Chromosomal Location
Gene Name/ Human Mouse Ligand Gene Name/ Human Mouse Gene
Alias Symbol Aliases Symbol
TNF 6p21.3 ch17 (19.06 cm) TNFSF1A TNFR-1, p55-60 12p13.2 ch6 (60.55 cM) TNFRSF1A
LTα 6p21.3 ch17 (19.06 cm) TNFSF1B TNFR2, p75-80 1p36.3- ch4 (75.5 cM) TNFRSF1B
LTβ 6p21.3 ch17 (19.06 cm) TNFSF3 36.2
OX40-L 1q25 ch1 (84.90 cM) TNFSF4 LTβR 12p13 ch6 (60.4 cM) LTβR
CD40-L, CD154 Xq26 chX (18.0 cM) TNFSF5 OX40 1p36 ch4 (79.4 cM) TNFRSF4
Fas-L 1q23 ch1 (85.0 cM) TNFSF6 CD40 20q12- ch2 (97.0 cM) CD40
CD27-L, CD70 19p13 ch17 (20.0 cM) TNFSF7 q13.2
CD30-L, CD153 9q33 ch4 (32.20 cM) TNFSF8 FAS, CD95 10q24.1 ch19 (23.0 cM) TNFRSF6
4-1BB-L 19p13 ch17 (20.0 cM) TNFSF9 DcR3 20q13 unknown TNFRSF6B
TRAIL 3q26 ch3 TNFSF10 CD27 12p13 ch6 (60.35) TNFRSF7
RANK-L, 13q14 ch14 (45.0 cM) TNFSF11 CD30 1p36 ch4 (75.5 cM) TNFRSF8
TRANCE 4-1BB 1p36 ch4 (75.5 cM) TNFRSF9
TWEAK 17p13 ch11 TNFSF12 TRAILR-1, DR4 8p21 Unknown TNRSF10A
APRIL/TALL2 17P13.1 ch13 TNFSF13 TRAIL-R2, DR5 8p22-p21 ch14 (D1) TNFRSF10B
BAFF, BLYS, 13q32-q34 ch8 (3 cM) TNFSF13B TRAILR3, DcR1 8p22-p21 ch7 (69.6 cM) TNFRSF10C
TALL1 TRAILR4, DcR2 8p21 ch7 (69.6 cM) TNFRSF10D
LIGHT 19p13.3 ch17 (D-E1) TNFSF14 RANK, 18q22.1 ch1 TNFRSF11A
TL1A 9q33 ch4 (31.80 cM) TNFSF15 TRANCE-R
G1TRL, A1TRL 1q23 Unknown TNFSF18 OPG, TR1 8q24 ch15 TNFRSF11B
EDA1 Xq12-q13.1 chX (37.0 cM) EDA1 FN14 16p13.3 ch17 TNFRSF12A
EDA2 Xq12-q13.1 cX (37.0 cM) EDA2 TRAMP, DR3, 1p36.3 ch4 (E1) TNFRSF25
LARD
CD, cluster of differentiation; LT, lymphotoxin; TNF, tumor necrosis factor; TNFSF, TACI 17p11.2 ch11 TNFRSF13B
tumor necrosis factor superfamily; TRAIL, TMF-related apoptosis inducing ligand.
From Ware CF. The TNF superfamily-2008. Cytokine Growth Factor Rev. 2008;19:183–186. BAFFR 22q13.1- ch15 TNFRSF13C
q13.31
HVEM, HveA, 1p36.3- ch4 TNFRSF14
ATAR p36.3
P75NTR, NGFR 17q12-q22 ch11 (55.6 cM) TNFRSF16
evolution.13 Conservation of gene structure of the TNF BCMA 16p13.1 ch16 (B3) TNFRSF17
ligands outside these MHC paralogs is limited. An evolu- AITR, GITR 1p36.3 ch4 (E) TNFRSF18
tionarily conserved pathway in Drosophila melanogaster, RELT 11q13.2 unknown TNFRSF19L
Eiger-Wengen system, is structurally and functionally TROY, TAJ 13q12.11- ch14 TNFRSF19
related to the TNFSF-related signaling pathway, although q12.3
EDAR 2q11-q13 ch10 EDAR1
this system is predominantly expressed in the nervous
EDA2R Xq11.1 ch X EDA2R
system in invertebrates.14–16
DR6 6P12.2-21.1 ch17 TNFRSF21
IGFLR1, 19q13.12 ch7 IGFLR1
TUMOR NECROSIS FACTOR RECEPTOR FAMILY Tmem149
Members of the TNF receptor superfamily (TNFRSF) in- CD, cluster of differentiation; HVEM, herpesvirus entry mediator; LT, lymphotoxin; OPG,
clude proteins of vertebrate and viral origin. Most of the osteoprotegerin; TNFR, tumor necrosis factor receptor; TNFRSF, tumor necrosis
factor receptor superfamily; TRAIL, TMF-related apoptosis inducing ligand.
signaling receptors in the TNFRSF are type I transmem- From Ware CF. The TNF superfamily-2008. Cytokine Growth Factor Rev. 2008;19:183–186.
brane glycoproteins (N-terminus exterior to the cell).
However, several TNFRSF members lack a membrane-
anchor domain, are proteolytically cleaved from the sur-
face, or are anchored via glycolipid linkage (eg, TNF-related The cysteine-rich domain (CRD) in the extracellular, li-
apoptosis inducing ligand [TRAIL]R3). These soluble recep- gand-binding region defines membership in the TNFRSF (see
tors, termed “decoy receptors,” retain their ligand-binding Fig. 27.1). Each CRD typically contains six cysteine residues
properties and compete with cellular receptors for the spe- forming three disulfide bonds. The CRD is pseudorepeated
cific ligands, thus earning the title of decoy. The structural in different members ranging from one to six. Based upon
motifs in the cytoplasmic domains further categorize the the crystal structures solved for several TNFRSF members,
TNFR into two groups based on their signaling properties: the CRD confers an elongated shape and sidedness to the ect-
those contain a death domain (DD) and others that engage odomain.17,18 The crystal structure of the complex between
TNFR-associated factors (TRAFs). TNFR1 and one of its ligands, the LTα homotrimer, revealed

CHAPTER 27 TUMOR NECROSIS FACTOR–RELATED CYTOKINES IN IMMUNITY | 661
FIG. 27.1. Structure of Tumor Necrosis Factor (TNF) and

TNF Receptors (TNFRs). Top: TNF. The β-sandwich
of the TNF monomer (1A8M.pdf) (shown as a ribbon
diagram) contains two stacked β-pleated sheets each
formed by five antiparallel β strands (wide ribbons given
letter designations, A–G) that fold into a Greek key or
“jelly-roll” topology.219 The inner β-sheet (strands A, A′,
H, C, and F) is involved in contacts between adjacent
subunits, which promotes assembly into a trimer. The
trimer is formed such that one edge of each subunit
is packed against the inner sheet of its neighbor. The
outer β sheet (strands B, B′, D, E, and G) is surface
exposed. The trimeric structure is characteristic of all
TNF-related ligands. The type II configuration of TNF
(N-terminus inside the cell) anchors TNF to the mem-
brane. The TNF trimer is ∼60 Å in height with a rela-
tively flat base residing close to the membrane, resem-
bling a bell-shape (shown as surface representation
with different shades of gray used for each subunit of the
trimer). The surface exposed loops between A-A′ and
D-E strands are involved in receptor binding. TNF is
cleaved by TNFα-converting enzyme (TACE), a member
of the ADAM family of metalloproteinases (ADAM17)
involved in processing of many cell surface proteins.
TACE is a type 1 transmembrane protein that cleaves
TNF between residues Val77 and Ala76, when all three
sites in the trimer are cleaved TNF is released from the
membrane. Bottom: TNFR and ligand complex. The
ectodomain of TNFR1 forms an elongated molecule
with CRD1 proximal to the N-terminus (ribbon diagram).
The face of TNFR1 on the left engages LTα. In the li-
gand-receptor complex, the elongated receptor (dark)
lies along the cleft formed between adjacent subunits.
Shown (space-filling) is a single TNFR1 in complex
with two subunits of LTα (lighter shades); the receptor
N-terminus points upward, closest to the base of the
ligand (transorientation). In the exploded view, TNFR1
is removed from in front of the ligand revealing the con-
tact residues in the ligand, which are primarily located
in the D–E and the A-A″ β-strands (dark shade). TNFR1
is rotated 180 degrees, exposing the contact residues
in the receptor (light shade). Structures from 1TRN.pdf17
as visualized with MacPyMOL (www.pymol.org).
that residues in CRD2 and 3 of TNFR1 contact the ligand. with the receptor, with the receptor binding site formed as
Variation in binding interactions has been identified; for ex- a composite of adjoining subunits in the trimeric ligand17
ample, the receptors for BAFF have one functional CRD.2,19 (see Fig. 27.1).
The trimeric architecture of the TNF ligands, containing
three equivalent receptor-binding sites, provides the basis
RECEPTOR-LIGAND COMPLEX for initiating signaling through aggregation or clustering of
The binding specificity of the various members of the TNF receptors. This concept is supported by the finding that re-
ligand and receptor superfamilies show monotypic interac- ceptor-specific bivalent antibodies can act as agonists mim-
tion, yet several members interact with multiple partners20 icking the signaling activity of the natural ligand.21 Indeed,
(Figs. 27.2 and 27.3). The binding interactions between TNF- antibodies or peptide mimetic to TNFR can function as
related ligands and TNFR are typically high affinity, with antagonists, blocking the ability of the natural ligand to
equilibrium binding constants measured in the high pM to bind the receptor while simultaneously activating the recep-
low nM range. The membrane position of the ligands fur- tor as an agonist.22 Some ligands such as FasL, TRAIL, and
ther enhances the binding interaction with their receptors. BAFF form higher ordered oligomers of the basic trimer.
In the membrane anchored position, the ligands and recep- These higher ordered oligomers promote supraaggregation
tors must be in trans to form a complex. In the LTα-TNFR1 of receptors, enhancing or sustaining signaling pathways
complex, the surface loops between A-A″ and D-E β strands in the receptor-bearing cell.23,24 Overexpression of TNFR
contain many of the amino acid residues that make contact in cells can also lead to ligand-independent signaling, a

FIG. 27.2. The Tumor Necrosis Factor (TNF) Superfamily–Major Histocompatibility Complex (MHC) Paralogs. Members of the TNF ligand
superfamily (above) and their corresponding receptors (below) are identified by connecting arrows. The ligands are grouped according to
their chromosomal locations in the MHC paralogous regions. The number of cysteine-rich domains are depicted for each TNF receptor,
and TNF receptors containing a death domain are identified by cylinder in the cytoplasmic tail. (Modified from Ware336).
feature bestowed in part by the propensity of the cytosolic LIGHT and LTα, it also engages two members of immuno-
tails to self-associate.25 In the physiologic setting, the expres- globulin (Ig) superfamily, B- and T-lymphocyte attenuator
sion level and compartmentalization of these receptors are (BTLA)31 and cluster of differentiation (CD)160.32 BTLA bind-
tightly controlled. A subregion in the first CRD of TNFR1 ing to HVEM occurs in CRD1, on the opposite face of where
known as the preligand assembly domain may restrict the LIGHT/LTα bind in CRD2 and 3. The BTLA binding site in
orientation of the unligated receptor to prevent spontane- CRD1 of HVEM is a region also targeted by herpesviruses.33–35
ous activation.26 It is not known if this mechanism applies to Recent evidence indicates the neurotrophin and chondrocyte
all TNFRSF. Interestingly, some patients with periodic fever growth factor–like protein progranulin interacts with TNFR1
syndrome have mutant TNFR1 that form abnormal disul- and TNFR2 and competes with TNF for binding.36
fide-linked oligomers that are retained intracellullarly and An insulin growth factor–like protein was recently
provoke misfolded protein response.27 identified as a ligand for a TNFR-like type 1 transmem-
brane protein (insulin growth factor–like R, formerly
TMEM149).37 Insulin growth factor–like R has conserved
Alternate Ligands positioning of cysteines delineating a CRD1, but atypical
There is significant divergence in the ligands and the mech- CRD2/3. Insulin growth factor–like messenger ribonucleic
anisms of ligand binding by the TNFR family. A major acid (mRNA) is expressed in psoriatic skin lesions and the
branch point is exemplified by the ligands for p75 neuro- receptor is detected in T cells suggesting a possible role in
trophin receptor (nerve growth factor and the other neuro- skin inflammation.
trophins), which are structurally unrelated to TNF ligand Death receptor-6 (DR6) engages the growth factor–like
family. Molecular contacts between NGF and p75NTR domain in β-amyloid precursor protein38,39 and thus may
occur through two spatially separated binding regions function as a negative regulator more like p75 neurotrophin
located at the first and second CRD and the junction be- receptor.40 Emerging evidence suggests DR6 has a role in neu-
tween the CRD3 and CRD4.28 The p75NTR functions in roimmune function. In a mouse experimental autoimmune
complex with two other proteins, Nogo66 and LINGO, to encephalomyelitis (EAE) model, DR6-deficiency was shown
engage myelin-associated inhibitory factors. Taj/TROY can to protect against central nervous system demyelination
supplant p75NTR in this complex.29 Like p75 NTR, DR6, and leukocyte infi ltration, but also enhanced overall CD4 +
ILGF1R, and TROY/Taj do not bind any of the known TNF T-cell proliferation and TH2 differentiation, underscoring a
ligands but do engage other ligands. The pathways activated role for DR6 in mediating TH1-specific immune responses
by p75NTR and TROY/Taj systems show both positive and in EAE progression.41 Taj/TROY and RELT also do not bind
inhibitory regulation of axonal regeneration.29,30 any of the known TNFSF ligands; however, recent evidence
The herpesvirus entry mediator (HVEM; TNFRSF14) indicates that Taj/TROY functions more like p75NTR, ca-
provides an example of a TNFR system that binds alternate pable of binding myelin inhibitory factors.29,30 A role for Taj/
ligands. Although HVEM engages two TNF-related ligands, TROY and RELT in immune function is presently unclear.

FIG. 27.3. The Tumor Necrosis Factor (TNF)

Superfamily—Part 2. Members of the TNF
ligand superfamily (above ) and their cor-
responding receptors (below ) are identi-
fied by connecting arrows. The ligands are
grouped according to their chromosomal
locations. The number of cysteine-rich do-
mains are depicted for each TNF receptor,
and TNF receptors containing a death do-
main are identified by cylinder in the cyto-
plasmic tail. (Modified from Ware336).
The engagement of ligands distinct from TNFSF members SIGNALING PATHWAYS AND
implicates a higher level of integration with other signaling CELLULAR RESPONSES
pathways.
Regulation
The cellular response activated by a TNF-related cytokine
Viral Orthologs depends on several factors including the temporal patterns
TNFR-like proteins are found in the genomes of several viral of expression of the ligands and receptors on the interact-
pathogens representing captured cellular genes that have ing cells, and the cellular context (the state of differentia-
evolved as part of that pathogen’s immune evasion strategy tion of the responding cell). Regulation is achieved at the
(Table 27.3). Poxviruses were the first pathogens identified level of transcriptional and translational controls, and by
harboring a version of a cellular TNFR.43 Poxvirus TNFR modulating the availability of the ligand or receptors (Fig.
displays significant sequence homology to TNFR2 and binds 27.4). For some ligands, transcriptional activation is a
TNF and LTα. The rabbit poxvirus protein T2 is secreted by critical feature controlling the duration of mRNA expres-
virally infected cells and contributes to the virulence of in- sion. Some ligands exhibit inducible, transient expression
fection.44,45 Smallpox virus, the former scourge of mankind, of mRNA following signals from antigen receptors or in-
also harbored viral versions of TNFR2.46 nate receptor recognition systems. The half-life of mRNA
TABLE 27.3 Viral Orthologs and Modulators of the Tumor Necrosis Factor Superfamily
Virus Name/ORF Orthologa Mechanism
Poxvirus
Myxoma T2 Soluble TNFR2 TNF and LTα decoy
Variola crm B (G2R) Soluble TNFR2 TNF and LTα decoy
Cowpox vCD30 Soluble CD30 CD30 ligand inhibitor
Herpesvirus
HSV1&2 Glycoprotein D BTLA Entry; HVEM blockade
HCMV UL144 HVEM BTLA activation
EBV LMP1 CD40 intracellular TRAF activation
γHV vFLIP FLIP Caspase 8 blockade
Adenovirus E3-10.4, 14.5, 6.7 ? Fas and TRAILR downmodulation
Retrovirus
EIAV Envelope gp90 ? HVEM entry factor
FIV Envelope gp95 ? Ox40 entry factor
ASLV Envelope gp85 ? TRAILR entry factor
Rabies virus Envelope RVG ? NTRp75 entry factor
ASLV, avian sarcosis and leukosis virus; BTLA, B- and T-lymphocyte attenuator; CD, cluster of differentiation; EBV, Epstein-Barr virus; HCMV, human cytomegalovirus; HSV, herpes
simplex virus; γHV, equine gamma herpesvirus; EIAV, equine infectious anemia virus; FIV, feline immunodeficiency virus; TNFR, tumor necrosis factor receptor; TRAF, TNFR-
associated factor; TRAIL, TMF-related apoptosis inducing ligand; ?, no homology recognized.
a
Relationship determined by sequence or structural homology to the indicated TNF superfamily member.

FIG. 27.4. Regulation of Tumor Necrosis Factor (TNF) Bioavailability. The expression
of TNF is regulated at the transcriptional and translational levels, and bioavailability
by altering its physical location and cellular receptors. TNF transcription is regu-
lated by the action of multiple transcription factors including nuclear factors of ac-
tivated T cells (NFAT), activated protein-1 (AP1), and NF-κB. NFAT is a predominant
acting transcription factor regulating TNF transcription in T- and B-lymphocytes,
and NF-κB and AP1 are important in myeloid lineage cells following activation via
innate pattern recognition receptors, such as the toll-like receptors. TNF messen-
ger ribonucleic acid (mRNA) stability is controlled by an adenylate-uridylate–rich
element (ARE) in the 3′ untranslated region present in many transiently expressed
inflammatory genes.337 Stability of TNF mRNA is decreased by the action of tristetra-
prolin, and T-cell intracellular antigen silences translation of TNF mRNA through the
ARE.338 TACE proteolytically controls TNF at the membrane, generating the soluble
form of TNF. TACE also cleaves TNFR1 and 2,339 downregulating cell surface recep-
tors and releasing soluble receptors that retain TNF binding activity. Soluble TNFR
can stabilize the TNF trimer at sub saturating concentrations, and at higher, saturat-
ing concentrations act as decoys competing for TNF binding to cellular receptors.340
for TNF is short, controlled by an adenylate-uridylate–rich including transforming growth factor α , L-selectin, and
element in the 3′ untranslated region.47 Deletion of the TNFR1 and 2. Production of soluble TNFR1 and 2 may be
adenylate-uridylate–rich element in TNF mRNA in mice important in regulating TNF bioavailability (see Fig. 27.4).
leads to a profound inflammatory disease.47,48 TNF mRNA Most of the TNF-related ligands are expressed by DCs,
is inducible in macrophages by multiple pathways, particu- activated lymphocytes, and myeloid cells, particularly mac-
larly innate activation pathways such as toll-like receptor rophages, but can also be produced by nonlymphoid cells.
signaling, whereas other ligands like 41BBL and OX40L are TNF is an example of a ligand expressed by many cell types
constitutively expressed in differentiated antigen-presenting depending on the stimulus. Expression of TNFR is wide-
dendritic cells (DCs). T- and B-lymphocytes require activa- spread. TNFR1 is expressed on most cells, while TNFR2 is
tion prior to expression of TNF. In these cells, signals via the limited to cells of hematopoietic origin and is expressed fol-
antigen receptor utilizing nuclear factor of activated T cells lowing activation of B or T cells. Macrophages are a primary
(NFAT) transcription factors are needed for the induction source of TNF in response to toll-like receptor signaling,
of TNF transcription. The inducible or constitutive patterns and T cells produce TNF when activated by antigenic stim-
of expression are also observed with some receptors as well. uli. Fibroblasts also produce TNF in response to virus infec-
Posttranslational regulation of signaling is achieved by tion, and nonlymphoid tumor cells may ectopically express
proteolytic cleavage of the ligand or receptor from the cell’s TNF. FasL is another example of a ligand with a varied cel-
surface, which places the protein into the soluble phase, lular expression pattern. FasL is expressed by effector T cells
where its half-life may be dramatically shortened. TNF and and natural killer (NK) cells as a component of their cell
Fas ligand, for example, are shed from the surface by mem- lytic activity, yet FasL mRNA is also detected in reproduc-
brane proteases. ADAM17 (also known as TNFα-converting tive organs and epithelium of the eye, which may use this
enzyme [TACE]), the enzyme that processes TNF into a solu- TNFSF member to kill organ infi ltrating inflammatory cells
ble form, is involved in cleaving multiple cell surface proteins as a mechanism to dampen inflammation.

Signal Transduction Pathways viruses have evolved many different strategies to alter pro-
TNF receptors initiate signaling pathways that alter gene apoptotic pathways (eg, viral orthologs of XIAP and FLIP
expression patterns, changing the differentiation status of prevent premature death of the cell) (see Fig. 27.6).
a cell, as well as apoptotic pathways that terminate cellular
life. The propagation of signals from receptor to enzymati- Cell Survival Signaling
cally active proteins is mediated by two distinct types of Activation of transcription factors by TNFR utilizes the
signaling motifs in the cytosolic domain of the TNFR: DDs TRAF adaptors. There are seven members of the TRAF fam-
and TRAF-binding motifs. TNFR that contain a DD include ily (given numerical designations) that play key roles in reg-
Fas, TNFR1, DR3, and TRAILR1 and 2. Other TNFRs have ulating TNFR signaling and activation of host defenses. Each
TRAF recruitment motifs. Three basic schemes are used by TRAF appears to play different roles in modulating signal-
TNF receptors to connect to enzymatically-driven signaling. Each TRAF protein contains a TRAF homology domain
ing pathways (Fig. 27.5A). Adaptor molecules are required that binds TNFR and a RING and/or zinc fi nger domain
to establish the signaling connections between the receptor characteristic of E3 ubiquitin ligases.52–54 Ubiquitination
and signaling enzymes. The DD connects TNFR to cyto- plays an essential role in regulating signal transduction by
solic proteins containing a Death Effector Domain (DED), TNFR55,56 and in the pathogen recognition receptors as part
which in turn link to caspases (cysteine based, aspartic acid of innate host defenses.57
specific proteinases). Alternately, the TRAF proteins con- Clustering of the TNFR promotes the recruitment of
nect to the cytosolic domain of the TNFR, altering ubiq- TRAFs into a complex with the receptor through two dif-
uitin-dependent pathways that regulate key serine kinases ferent mechanisms. TRAF2, 3, and 5 bind directly to the re-
that activate NFκ B and AP1 transcription factors. The ceptor’s cytosolic domain recognizing a consensus PXQS/T
third scheme involves an indirect link between the death motif,58 whereas TNFR1 uses TRADD to couple to TRAF2.
domain and TRAFs via the adaptor TNFR-associated death TRAF3 and TRAF6 are preassociated with different pro-
domain (TRADD). teins, including protein kinases involved in multiple sig-
The apoptotic and NF-κ B pathways activated by the naling pathways including the pathogen recognition and
TNFR family help regulate cellular homeostasis by con- interferon (IFN) pathways.53,59 TRAF adaptors link TNFR
trolling cell death and survival. In the immune system, directly to protein kinase cascades, which in turn lead to ac-
apoptosis is essential for homeostasis and for eliminating tivation of transcription factors including NF-κ B and AP1.
antigen-bearing cells from the host. Many TNFR can in- The NF-κ B family of transcription factors control ex-
duce activation of survival or death pathways depending in pression of genes critical for cell survival, inflammatory, and
part on the differentiated state of that cell. In all nucleated immune responses. In mammalian cells, the NF-κ B fam-
cells, apoptosis is the default pathway; that is to say, all of the ily consists of five members: RelA, RelB, c-Rel, p50/NFκ B
constituents of the pathway are expressed and ready to be B1, and p52/NFκ B B2 (proteolytic processing of p105 and
activated (Fig. 27.6), whereas cellular survival requires tran- p100 yields the active forms p50 and p52, respectively).60,61
scriptional control of genes that encode regulatory proteins Activation of NF-κ B releases inhibitors that restrict nuclear
that suppress the progression of the apoptotic pathway. translocation.62 Homo- and heterodimers of NF-κ B fam-
ily members are held inactive in the cytosol by inhibitors
Apoptosis of κ B (Iκ B), such as Iκ Bα , that mask nuclear localization
Ligation of TNFR promotes assemble of the death-induc- motifs on the NF-κ B dimers. A complex consisting of the
ing signaling complex (DISC) that promotes dimeriza- kinase catalytic subunits IKKα and IKKβ and the regula-
tion of procaspase 8 forming an active enzyme complex.49 tory/scaffold subunit IKKγ form the IKK complex that me-
Activated caspase 8 acts directly to cleave procaspase 3 and diates phosphorylation and ubiquitination of Iκ B, leading to
7, which are known as the executioner caspases as they di- its degradation and release of the active transcription factor.
rectly cleave critical cellular substrates leading to apoptotic Iκ B is the common target of a variety of signals that control
death. Caspase 8 also cleaves BID, a crucial connector to the the activation of the RelA NF-κ B transcription factor, which
mitochondria-associated death mechanism, which greatly in turn regulates expression of many proinflammatory genes
amplifies the apoptotic process.50 A cell must be capable of within a signal responsive cell.63
actively transcribing and translating genes to resist apopto- A distinct pathway regulates the activation of RelB NF-
sis signaling. A variety of genes can inhibit apoptosis and cell κ B through the NF-κ B–inducing kinase (NIK) and IKKα
death pathways. For example, the cellular inhibitor of apop- kinases. In unstimulated cells, NIK levels are maintained at
tosis (XIAP) is a direct caspase inhibitor that is regulated at vanishing low levels by ubiquitination and proteosome deg-
the level of gene expression by transcription factor NF-κ B. radation through an E3 ligase complex comprised of TRAF3,
Another regulator, FLIP (FLICE inhibitory protein), is also TRAF2, and CIAP1/2.52 Receptor ligation releases the active
an NF-κ B–regulated gene that contains a DED and a pseu- form of NIK by competitively displacing TRAF3, preventing
docaspase domain that attenuates the apoptotic pathway by further ubiquitination and allowing NIK to accumulate.64
blocking conversion of procaspase 8 to the active form.51 NIK phosphorylates IKKα , which induces the proteosome
Many viral pathogens parasitize the transcriptional capa- dependent processing of p100 to p52, degrading the inhibi-
bilities of the cell and prevent the cell from making new sur- tory domain of p100 and allowing the RelB/p52 complex to
vival proteins, allowing apoptosis to proceed. As expected, activate gene transcription.

TNFR1 LT␤R
IKK␣␤␥ NIK
complex TRAF
IKK␣ IKK␣
p100
RelB
I␬B␣
p50 p65
p100
SLC
ELC
MIP-1␤ SDF-1␣
MIP-2 ICAM
p65 p100 RelB
p65 VCAM-1 p50
BAFF p52
p50
B
FIG. 27.5. Signaling Pathways and NF-jB. A: Adaptors link tumor necrosis factor receptors (TNFRs) to proteinases and kinases (upper
panel). Three basic schemes link activated TNFRs to signaling pathways: the death inducing signaling complex formed with Fas is
initiated by ligand clustering of Fas, promoting death domain (DD) interactions, which recruits Fas-associated death domain (FADD)
(heterotypic interaction). The death effector domain of FADD links to the death effector domain of procaspase 8. The proximity of
multiple procaspase 8 domains forms an active enzyme that can process other caspase 8 molecules. By contrast, TNFRs bind TNF-
associated factor (TRAF) adaptors via short peptide motifs that release TRAF from associated kinases, such as NF-κB–inducing kinase
(NIK). The third scheme is a combination of DD and TRAF recruitment. TNFR1 uses the DD protein TRADD to recruit RIP and TRAF2,
promoting the activation of NF-κB. There are currently seven TRAF members with distinct interaction patterns with the TNFR family.
TRAF proteins may function as regulators of key kinases. TRAF2, 3, and 6 function as modulators of several different kinases involved
in toll-like receptor signaling, induction of type 1 interferon responses, and signaling by some TNFR family members. TRAF proteins
contain an N-terminal RING finger motif, a coiled coil domain (isoleucine zipper), and the receptor association domain (TRAF) domain.
The TRAF are trimers formed through their TRAF and coiled domains. The TRAF domain can bind to several different TNFR through a
relatively short proline-anchored sequence that is responsible for binding directly to the mushroom head of various TRAF molecules.341
The zinc RING of TRAF6 functions together with Ubc13 and Uev1A as ubiquitin E3 ligase targeting proteins for proteosome degradation.
TRAF2, 3 and cIAP form the ubiquitin E3 ligase that targets NIK. Modified with permission from Ware CF. Tumor necrosis factors. In:
Bertino JR, ed. Encyclopedia of Cancer. San Diego, CA: Academic Press, Inc.: 2002;475–489. B: NF-κB activation. TNFR1 and LTβR induce
distinct forms of the NF-κB family of transcription factors. TNFR1 signaling is a potent activator of RelA/p50 but does not activate the
RelB pathway, whereas lymphotoxin βR can activate both. The RelA and RelB forms of the κB family control transcription of distinct
sets of genes; however, the two pathways are interrelated through the control of p100 expression by the RelA/p50 complex. Modified
from Dejardin et al.66

As examples, TNFR1 and LTβR activate separate yet

related pathways that lead to distinct forms of NF-κ B, the
RelA/p50, and RelB/p52 complexes.65 Each form of NF-κ B
activates a large number of genes with distinct roles in phys-
iology (see Fig. 27.5B). TNFR1 signaling rapidly mobilizes
(within minutes) the RelA/p50 complex, which controls
expression of many proinflammatory and survival genes.
By contrast, the processing of p100 and accumulation of
nuclear RelB/p52 takes several hours after the initial stimu-
lus. RelB-dependent genes are often involved in lymphoid
tissue organogenesis and homeostasis. NIK is also required
for NF-κ B RelB and RelA activation by CD27, CD40, and
BAFF-R, but not by TNFRI, which is restricted to activating
RelA/p50 complex.66,67 Components of the NF-κ B pathway,
including TRAF3, cIAP, NIK, and A20, are frequently mu-
tated in cancer, leading to constitutive expression of survival
factors such as the BCL2 family.68
THE IMMEDIATE LYMPHOTOXIN AND

A
TUMOR NECROSIS FACTOR FAMILY
TNF (formerly TNFα or TNFSF2) and lymphotoxin (LT) α
(formerly TNFβ or TNFSF1) were originally pursued and
characterized as inducers of tumor cell death, holding prom-
ise as antitumor therapeutics. However, the potent inflam-
matory action of TNF, particularly in the cardiovasculature,
was quickly realized by the response of cancer patients in-
jected with recombinant protein. We now recognize that
TNF and LTα are two components of an interconnected
network of “signaling circuits” that include LT-β, LIGHT
(TNFSF14), and their specific receptors and regulatory pro-
teins (Fig. 27.7). Each individual pathway has unique and
cooperative signaling activities with other members of this
immediate family. The immunologic processes controlled
by this cytokine network are extensive, ranging from the de-
velopment and homeostasis of lymphoid organs to the mo-
bilization of innate defense systems to cosignaling activity
B promoting adaptive immune responses.69–71
TNF, LTα , LTβ, and LIGHT define the immediate group
FIG. 27.6. Apoptosis and Survival Signaling. A: Ligation of tumor of TNF-related ligands that bind four cognate cell surface
necrosis factor (TNF) receptor 1, Fas, or TNF-related apoptosis in- receptors with distinct but shared specificities. TNF and
ducing ligand 1 and 2 activates apoptosis pathways. The nascent
death-inducing signaling complex can directly cleave the execu- LTα both bind two distinct receptors, TNFR1 (p55-60,
tioner caspases 3, 6, and 7. Caspase 8 also cleaves BID to tBID, TNFR1A) and TNFR2 (p75-80, TNFR1B). The heteromeric
activating the mitochondrial regulated apoptotic pathway, dramati- LTαβ2 complex binds the LTβR, which also binds LIGHT
cally accelerating cellular death. A number of viral proteins inter- (TNFSF14). LIGHT also engages the HVEM (TNFRSF14),
fere with apoptosis induced through TNF pathway including human which acts as a ligand for BTLA and CD160,31 an Ig super-
papilloma virus, hepatitis C virus, herpesvirus vFLIP, and virus or- family member. Two distinct human herpesviruses, herpes
thologs of BCL2 and vMIA act on the mitochrondria. Adapted with
simplex virus and cytomegalovirus, target the HVEM-BTLA
permission.111 B: The components of the apoptotic pathway are pre-
formed in the cytosol enabling the cell to respond rapidly to death pathway using different mechanisms35,72 (see Fig. 27.7).
signaling. In contrast, the key regulators of the apoptotic pathway, TNF mediates a diverse range of cellular and physiologic
including Bcl2 and inhibitor of apoptosis, require new gene expres- responses linked to acute and chronic inflammatory pro-
sion controlled by NF-κB. An important NF-κB–induced gene is cesses. The diversity of responses is due in part to the broad
cellular FLIP, which contains a pseudocaspase domain. The cFLIP expression of TNFR1 and the release of TNF as a soluble me-
protein interferes with caspase 8 activation, blocking the prodeath diator where it can act in a systemic fashion. TNF production
pathway. If the cell is transcriptionally inactive or unable to acti-
vate NF-κB, which often occurs in a pathogen-infected cell, then
is triggered by many of the innate recognition systems, such
the proapoptotic pathway dominates the signals emanating from the as the toll-like receptors and by T and B cells. In chronic in-
TNF receptor 1 signaling complex leading to apoptosis and curtail- flammation, TNF production can lead to tissue damage and
ing parasitism of the cell. organ failure. Genes expressed in response to TNF signaling

FIG. 27.7. The Immediate Tumor Necrosis Factor (TNF) Lymphotoxin (LT) Network. The car-
toon depicts the signaling network formed between TNF, LTα, LTβ, and LIGHT, and their
receptors. Each arrow indicates a ligand-receptor interaction. The network is defined by
the extensive cross-utilization of ligand and receptors. Herpesvirus entry mediator (HVEM)
forms a switch between positive cosignaling through LIGHT-HVEM interaction and inhibitory
signaling through B- and T-lymphocyte attenuator (BTLA). LIGHT bound to HVEM activates
TNF receptor–associated factor–dependent activation of NF-κB, whereas HVEM-BTLA acts
through an immunoreceptor tyrosine-based inhibitory motif of BTLA to recruit the phospha-
tase SHP2, attenuating kinases activated by T-cell receptor signaling. The herpes simplex
viron envelope protein gD attaches to HVEM acting as an entry step for infection. UL144
gene of human cytomegalovirus binds BTLA, but not LIGHT, selectively mimicking the inhibi-
tory pathway of HVEM-BTLA. Fas Ligand and TL1A are included in this network through their
interaction with decoy receptor-3.
coordinate the physiologic responses during inflammation Lymphotoxinà-LymphotoxinaR, A Mammalian Organ

(Table 27.4). The responses will reflect the characteristic of Development Pathway
the inflamed organ, the local or systemic source of TNF, and Gene-deficient mice sharing a common phenotype of no
the duration of TNF signaling. Acute inflammatory responses lymph nodes revealed the framework of a signaling pathway
involve rapid changes in hemodynamics (plasma leakage and involved in mammalian organ development. LTα-, LTβ-, or
edema) and leukocyte adherence, extravasation, and organ LTβR-deficient mice fail to develop secondary lymphoid tis-
infiltration induced by TNF. TNF production during chronic sues.76,77 Several other knockout mice, including the tran-
inflammation may contribute to systemic metabolic derange- scriptional regulators Ikaros, ID2, and RORγ t, also lack lymph
ments and wasting (cachexia) or loss of organ structure and nodes,78–82 as do mice deficient in components of the NF-κB
function (bone erosion in joints of patients with rheumatoid
arthritis [RA]). In some models, TNF can promote an inflam-
matory environment that promotes tumor formation.73–75
Elucidation of the functions associated with this TNF/LT Physiologic Correlates of Tumor
signaling network has been aided by studies with genetically TABLE 27.4 Necrosis Factor–Mediated
modified mice engineered with null or transgene expression Gene Induction
of the cytokine or receptor (Table 27.5). Deficiency in TNF or
TNFR1, but not TNFR2, have similar phenotypes with altera- Induced Gene Response
tions in host defense to intracellular bacterial pathogens, like iNOS Vasodilation, edema
Listeria monocytogenes and Mycobacterium tuberculosis, but VCAM-1 Leukocyte margination and
surprisingly modest susceptibility to some viral pathogens. extravasation
These results demonstrated a role for TNF in acute-phase re- IL-8 Leukocyte chemotaxis
sponse of the host defense system. In contrast, LTα-deficient MHC-1 Antigen presentation
mice showed a failure in formation of peripheral lymphoid Caspase 8 activation Apoptosis
organs, a phenotype not observed in mice deficient in either LPL Cachexia
TNFR or TNF, which implicated the LTαβ complex signaling
Caspase 8, cysteine-dependent aspartic acid specific proteinase-8; IL-8, interleukin-8 (CXC
through the LTβR as a key developmental pathway for lym- chemokine); iNOS, inducible nitric oxide synthetase; LPL, lipoprotein lipase; MHC-1,
phoid organogenesis. major histocompatibility complex-1; VCAM-1, vascular cell adhesion molecule-1.

TABLE 27.5 Phenotypes in Mice Deficient in Lymphotoxin and Tumor Necrosis Factor Immediate Family
Phenotypes
Gene Deletion LNa PPb Architecturec NKd NKTe DCf
LTα − − Disrupted Impaired Impaired CD8-DC
LTβ − − Disrupted Impaired Impaired CD8-DC
LIGHT + + + + + +
LTβ-Bg + + Disrupted nr nr nr
LTβ-Tg + + + nr nr nr
TNFh + + + + + Maturation
LTβR − − Disrupted − − CD8-DC
TNFR1 + − Disrupted MZ + + Maturation
TNFR2 + + + + + +
−, absent; +, normal; CD, cluster of differentiation; DC, dendritic cell; LN, lymph node; LT, lymphotoxin; MZ, marginal zone; NK, natural killer; NKT, natural killer T; nr, not reported;
PP, Peyer’s patches; TNF, tumor necrosis factor; TNFR, tumor necrosis factor receptor.
a
LTβ−/− mice have ∼75% of mesenteric LN; LIGHT/LTβ−/− mice have fewer mesenteric nodes than LTβ−/− mice.76,342,343
b
See Rutschmann et al.344 and Neumann et al.345
c
Architecture of the splenic white pulp includes T- and B-zone segregation, MZ, germinal center and follicular DC network.
d
NK-cell deficiency includes reduced cell numbers and enhanced tumor susceptibility.346
e
NKT cells Vα14 subset.347
f
CD8− DC subsets in spleen are diminished from failure to proliferate.95
g
LTβ conditionally deleted in B cells or T cells.348 LTβ-B cells showed partial disruption in architecture; normal for LTβ-T, but combined knockout in both B and T cells was
worse than LTβ-B.
h
Normal architecture observed in TNF point mutant344; abnormal architecture in TNFR1−/− mice.345
Modified with permission from Ware.70
activation pathway, including TRAF6, NIK, IKKα, and Rel B, and CCL21 act through the chemokine receptor CCR7 ex-
and their gene targets. Target genes including CXCR5, the repressed on T cells, and CXCL13 binds CXCR5 on B cells to
ceptor for CXCL13, show defective lymphoid organ structure, promote localization in the follicles. An LT-chemokine cir-
as do CCR7−/− mice (receptor for CCL19/CCL21). The devel- cuit is formed by migration of B cells to LTβR+ stromal cells
opmental program of secondary lymphoid organ formation expressing CXCL13, which in turn induces expression of
initiates at 9 days post coitus progressing in organized fash- LTαβ on B cells.89–91 Circulating B cells lack surface LTαβ,
ion from dorsal to lateral movement, with intestinal Peyer’s but expression is regained upon reentry into the CXCL13
patches forming during the first postnatal week.83 The defect rich microenvironment.91 The formation of the splenic mi-
is irreversible in that transferring LTαβ-sufficient bone mar- croarchitecture depends on B-cell expression of LTαβ, which
row into an adult LT-mutant mouse failed to induce lymph induces differentiation of specialized stromal cells (eg, secre-
node formation. Accumulating studies indicate that the for- tion CCL20 and CXCL13 chemokines) in the spleen during
mation of lymphoid organs involves two distinct cell types, postnatal maturation. Remodeling of the microarchitecture
an embryonic LTβR-expressing mesenchymal stromal cell of the secondary lymphoid organs occurs during immune re-
that responds to LTαβ expressed in a cell of hematopoieti- sponses, which requires both TNF and LT pathways signaling
cally derived lineage, termed the lymphoid tissue inducer cell. on fibroblastic reticular cells.92 A viral pathogen, cytomega-
The lymphoid tissue inducer cell develops separately from lovirus, can induce specific changes in the splenic microenvi-
lymphocytes and myeloid cells in a pathway dependent on ronment through modulation of CCL21 expression.93
ID2 and RORγ t and different cytokines including interleukin The LTβR and TNFR pathways facilitate lymphocyte
(IL)7, RANK ligand (TRANCE, TNFSF11), and TNF. These entry into lymphoid tissues in part by modulating expres-
cytokines induce surface LTαβ in lymphoid tissue inducer sion of adhesion molecules, such as peripheral node and
cells that differentially engenders formation of lymph nodes mucosal addressins on high endothelial venules.94 LTβR sig-
and Peyer’s patches.84 Cells of the lymphoid tissue inducer lin- naling is necessary to maintain networks of follicular DCs
eage are maintained in the adult tissues at low levels (0.5%), involved in capturing antigen and immune complexes that
presumably aiding in the homeostasis of lymphoid organs.85,86 aid in activating B cells. These cellular interactions are fur-
The microarchitecture of the white pulp in the spleen87 ther enhanced by LTβR signals that provide growth signals
is disrupted in LT- and TNF-deficient mice.88 Multiple ab- for some conventional myeloid DC (CD8− subsets) within
normal features of the architecture are observed in LT- and the lymphoid organ, whereas TNF plays a role in differentia-
TNF-deficient mice, including missing macrophages in the tion of DC progenitors in bone marrow.95
marginal sinus and the loss of positional segregation of T T and B cells can communicate with stromal and myeloid
and B cells into discrete zones. The segregation of T and B cells via the LTαβ -LTβR pathway and thus modify their im-
cells into discrete compartments depends on expression of mediate microenvironment during the course of an immune
the tissue organizing chemokines CCL19 and CCL21, which response. Nonlymphoid tissues suffering from chronic
attract T cells, and CXCL13, which attracts B cells. CCL19 inflammation associated with autoimmune disease, graft

rejection, or microbial infection often contain organized ac- is rapidly lethal owing to profound changes in blood circula-
cumulations of lymphocytes reminiscent of secondary lym- tion. However, mice survive lipopolysaccharide in the genetic
phoid organs, called tertiary lymphoid organs. Activated T absence of TNFR1 or if treated with an TNF-neutralizing
and B cells provide the source of LTαβ that helps drive the antibody, indicating that host-derived TNF mediates patho-
process of forming these structures, but as antigen is cleared, genesis.107–109 On the other hand, TNFR1 is essential for re-
immune responsiveness subsides and these structures resistance to infection with a live organism, such as Listeria
solve. A gradation of features may be found in the tertiary monocytogenes, through multiple processes including en-
lymphoid organs including presence of DCs, expression of hancement of phagocytosis and bacteriocidal destruction
chemokines, high endothelial venules, segregated regions by macrophages. By contrast, T-cell immunity is not overtly
of T and B cells, and germinal centers, but these structures impaired in TNFR1−/− mice. Humans treated with TNF
typically lack the permanence of a lymph node. TNF also inhibitors show some increase in susceptibility to selective
contributes to formation of granuloma that assists in wall- pathogens, particularly Mycobacterium tuberculosis, rein-
ing off bacteria.96 Thus, the LTαβ and TNF pathways opera- forcing the role of TNF as a critical host defense system.110
tive in embryonic life also play critical roles in the adult in In contrast, LT-deficient mice showed significant vari-
the formation of tertiary lymphoid structures. ability in susceptibility to individual pathogens (Table 27.6).
Increased susceptibility resulted from either developmen-
Influence of the Lymphotoxinà Pathway on tally controlled aspects of lymphoid organ structure (eg,
lymphocytic choriomeningitis virus, Leishmania) or a re-
Lymphocyte Development
quirement for LTαβ-LTβR pathway as an effector system in
Mounting evidence indicates the LTαβ -LTβR pathway con- innate and adaptive immune systems (eg, murine cytomeg-
tributes to the ontogeny of unconventional T cells, including alovirus). Viewed from an evolutionary perspective, this
γδ T cells and invariant NK T cells, whereas conventional variation in the requirement for LT signaling may reflect
T cell subsets are normal in mice deficient in the TNF and specific contributions from the pathogen used to evade the
LTβR pathways. The LTβR pathway seems to operate at dis- broader TNF- and LT-dependent pathways.35,111
tinct levels during thymic development.97 Double positive
thymocytes regulate the differentiation of early thymocyte
progenitors and γδ T cells via the LTβR pathway,98 yet the TUMOR NECROSIS FACTOR SUPERFAMILY AND
LTβR is not expressed in thymocytes, suggesting an indirect T-CELL COSIGNALING
mechanism. In this regard, LTβR signaling is required for the Antigen recognition together with cooperating signaling
proper formation and function of the thymic stroma, which “cosignaling” systems determine the quality of the adap-
influences T-cell development.99 The thymic medulla ap- tive immune responses. Lymphocyte responses to antigen
pears to control the export of invariant NK T cells from the are dynamic processes that start with the activation of naïve
thymus.100,101 In addition, LTβR signaling in thymic stroma cells and transition through effector and memory phases.
affects central tolerance to peripherally restricted antigens, Cosignaling systems assist these phases by promoting more
which may be either dependent or independent upon the au- efficient engagement of antigen-binding TCR molecules
toimmune regulator Aire.102 Thymic differentiation depends to enhance initial cell activation and cell division (clonal
on the LTβR pathway to mediate the cellular communica- expansion), augment cell survival (clonal contraction or
tion between lymphoid and stromal compartments. memory cell differentiation), or induce effector functions
such as cytokine secretion or killer function. Negative sig-
Autoimmunity nals (inhibitory cosignaling) may also be delivered to T cells
depending upon the particular system, which may prevent
Dysregulated expression of several members of the TNFSF initial cellular activation or eliminate excess activated cells
leads to autoimmune-like diseases in humans and animal to dampen inflammation. Cosignaling can be quantitative,
models. For example, enforced expression of TNF or LIGHT, modifying thresholds of common signaling intermediates,
which overrides the normal transient expression, causes se- or qualitative, involving signals distinct from other cosig-
vere autoimmune and inflammatory processes in mice.103–105 naling systems or the TCR. Cosignaling receptors and li-
LTα or LTαβ transgenic expression in the pancreas leads to gands can be up- or downregulated at the transcriptional
insulitis and formation of tertiary lymphoid structures.106 and protein levels depending on the stage of the T-cell re-
These types of results have implicated members of the TNF sponse and the inflammatory milieu. In the absence of co-
superfamily as immune regulators and support the notion signaling, T cells may become unresponsive (anergic) or die.
that LTαβ and LIGHT pathways contribute to inflammation The TNFRSR is one of two major families of cosignaling
and tissue destructive processes. regulators that modulate T cells. The other cosignaling sys-
tems belong to the Ig superfamily, such as CD28,112 cytotoxic
Infectious Diseases T-lymphocyte (CTL)A-4,113 ICOS,114 PD1,115 and BTLA.116,117
TNF is a major inflammatory cytokine required for the acute- TNFRSF members involved in T-cell cosignaling include
phase response to bacterial infection. For instance, lipopoly- OX40, 41BB, DR3, CD27, CD30, and HVEM,118–122 whereas
saccharide in gram-negative bacteria is a potent inducer of CD40 and BAFFR are more involved in cosignaling in B lym-
TNF secretion through the TLR4 innate recognition system. phocytes.123,124 However, considerable crossover of activities
In mice, lipopolysaccharide induces a shock syndrome that in both lymphocyte populations can be demonstrated. Death

TABLE 27.6 Lymphotoxins in Host Defense: Mouse Models

Pathogena Mouse Modelb Susceptibility Mechanism Reference
Herpesvirus
349
MHV68 LTα−/− Minimal nd
350
HSV-1 LTα−/− Increased Decreased effector CD8+ T cells
351–353
MCMV LTα−/−;B-LTβ Increased IFN response; adaptive immunity lost
354
MCMV LTβR−Fc Tg Increased Poor innate defenses
354–356
LCMV LTβ−/−; LTα−/− Increased Defective architecture
357
LCMV LTβR-Fc Decreased Decreased CD8+/IFNγ
358
Theiler virus LTα−/− Increased Defective architecture
LTβR−Fc
359
Influenza LTα−/− Minimal nd

(Medicalstudyzone - Com) Paul's Fundamental Immunology PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(Medicalstudyzone - Com) Paul's Fundamental Immunology PDF

Uploaded by

Copyright:

Available Formats

Www.Medicalstudyzone.

Paul_FM_final.indd i 9/17/12 3:25 PM

This PDF was created and uploaded by

Furthermore You can also request a specific

Paul_FM_final.indd ii 9/17/12 3:25 PM

Paul_FM_final.indd iii 9/17/12 3:25 PM

Acquisitions Editor: Julie Goolsby

© 2013 by LIPPINCOTT WILLIAMS & WILKINS, a WOLTERS KLUWER business

Two Commerce Square

Library of Congress Cataloging-in-Publication Data

Fundamental immunology / editor, William E. Paul. — 7th ed.

Paul_FM_final.indd iv 9/27/12 12:50 AM

Paul_FM_final.indd v 9/17/12 3:25 PM

Paul_FM_final.indd vi 9/17/12 3:25 PM

Eitan M. Akirav, PhD Jean-Laurent Casanova, MD, PhD

Ira Berkower, MD, PhD Sangeeta S. Chavan, PhD

Paul_FM_final.indd vii 9/17/12 3:25 PM

Ken T. Coppieters Martin F. Flajnik, PhD

Paul_FM_final.indd viii 9/27/12 12:51 AM

Marc K. Jenkins, PhD Judy Lieberman, MD, PhD

Paul_FM_final.indd ix 9/17/12 3:25 PM

Philip M. Murphy, MD Luke A.J. O’Neill, PhD

Moon H. Nahm, MD John J. O’Shea, MD

Luigi D. Notarangelo, MD Jeffrey V. Ravetch, MD, PhD

Paul_FM_final.indd x 9/17/12 3:25 PM

Jamie Rossjohn, PhD Alan Sher, PhD

Ellen V. Rothenberg, PhD Ethan M. Shevach, MD

David L. Sacks, PhD Megan Sykes, MD

Paul_FM_final.indd xi 9/17/12 3:25 PM

David Wald, MD, PhD Kathryn J. Wood, DPhil

Marsha Wills-Karp, PhD

Paul_FM_final.indd xii 9/17/12 3:25 PM

Paul_FM_final.indd xiii 9/17/12 3:25 PM

Paul_FM_final.indd xiv 9/17/12 3:25 PM

Paul_FM_final.indd xv 9/17/12 3:25 PM

Paul_FM_final.indd xvi 9/17/12 3:25 PM

Contributing Authors vii Section IV. T-Lymphocytes

Section I. Introduction to Immunology

2 History of Immunology . . . . . . . . . . . . . . . . . . . . . . .22 13 T-Lymphocyte Developmental Biology . . . . . . . .325

3 Lymphoid Tissues and Organs . . . . . . . . . . . . . . . .47

Section III. Immunoglobulins and 16 Dendritic Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . .381

6 Immunoglobulins: Molecular Genetics . . . . . . .150 18 CD1d–Restricted Natural Killer T Cells . . . . . . .432

7 Antigen-Antibody Interactions and 19 Macrophages and Phagocytosis . . . . . . . . . . . . .448

9 B-Lymphocyte Receptors, Signaling 21 The Major Histocompatibility

10 B-Lymphocyte Responses . . . . . . . . . . . . . . . . . . .261 22 Antigen Processing and Presentation . . . . . . . .524

Paul_FM_final.indd xvii 9/17/12 3:25 PM

Section VI. Induction, Regulation, 36 Complement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .863

Response 37 Cell-Mediated Cytotoxicity . . . . . . . . . . . . . . . . . .891

Section VII. Immunity to Infectious Agents

25 Type I Cytokines and Interferons, and 39 Immunity to Viruses . . . . . . . . . . . . . . . . . . . . . . . .937

33 Regulatory/Suppressor T Cells . . . . . . . . . . . . . . .795 46 Transplantation Immunology . . . . . . . . . . . . . . .1154

34 The Mucosal Immune System . . . . . . . . . . . . . . .833 47 Cancer Immunology . . . . . . . . . . . . . . . . . . . . . . .1200

35 Neurophysiologic Reflex 48 Inborn Errors of Immunity . . . . . . . . . . . . . . . . . .1235

Paul_FM_final.indd xviii 9/17/12 3:25 PM

Nothing is as powerful as an idea whose time has come.

Everything should be made as simple as possible, but not simpler.

In the fields of observation, chance favors the prepared mind.

In all things of nature there is something of the marvelous.

Paul_FM_final.indd xix 9/17/12 3:25 PM

Paul_FM_final.indd xx 9/17/12 3:25 PM

Paul_CH01_final.indd 1 9/17/12 5:16 AM

Immune Responses Against Self-Antigens can Result

Paul_CH01_final.indd 2 9/17/12 5:16 AM