Food

Microbiological Analysis : New Technologies IFT Basic Symposium

title:
Series ; 12
author: Tortorello, Mary Lou
publisher: CRC Press
isbn10 | asin: 0824700872
print isbn13: 9780824700874
ebook isbn13: 9780585350264
language: English
subject Food--Microbiology, Food--Analysis, Microorganisms--Identification.
publication date: 1997
lcc: QR115.F638 1997eb
ddc: 664/.001/579
subject: Food--Microbiology, Food--Analysis, Microorganisms--Identification.
 Food Microbiological Analysis

 Basic Symposium Series
Edited by
INSTITUTE OF FOOD TECHNOLOGISTS
221 N. LaSalle St.
Chicago, Illinois

1. Foodborne Microorganisms and Their Toxins: Developing Methodology, edited by Merle D. Pierson and Norman J.
Stern

2. Water Activity: Theory and Applications to Food, edited by Louis B. Rockland and Larry Beuchat

3. Nutrient Interactions, edited by C. E. Bodwell and John W. Erdman

4. Food Toxicology: A Perspective on the Relative Risks, edited by Steven L. Taylor and Richard A. Scanlan
5. Biotechnology and Food Process Engineering, edited by Henry G. Schwartzberg and M. A. Rao

6. Sensory Science Theory and Applications in Foods, edited by Harry T. Lawless and Barbara P. Klein

7. Physical Chemistry of Foods, edited by Henry G. Schwartzberg and Richard W. Hartel

8. Flavor Measurement, edited by Chi-Tang Ho and Charles H. Manley

9. Protein Functionality in Food Systems, edited by Navam S. Hettiarachchy and Gregory R. Ziegler

10. Bioseparation Processes in Foods, edited by Rakesh K. Singh and Syed S. H. Rizvi

11. Food Lipids and Health, edited by Richard E. McDonald and David B. Min

12. Food Microbiological Analysis: New Technologies, edited by Mary Lou Tortorello and Steven M. Gendel

 Page i

Food Microbiological Analysis

New Technologies

edited by
Mary Lou Tortorello
Steven M. Gendel

National Center for Food Safety and Technology

Food and Drug Administration
Summit-Argo, Illinois

 Page ii

Library of Congress Cataloging-in-Publication Data

Food microbiological analysis: new technologies/edited by Mary Lou

Tortorello, Steven M. Gendel.
p. cm. (IFT basic symposium series: 12)
Includes index.
ISBN 0-8247-0087-2 (hardcover: alk. paper)
1. FoodMicrobiology. 2. FoodAnalysis. 3. Microorganisms
Identification. I. Tortorello, Mary Lou. II. Gendel,
Steven M. III. Series.
QR115.F638 1997
664'.001 '579dc21 97-10654
CIP

The publisher offers discounts on this book when ordered in bulk quantities. For more information, write to Special
Sales/Professional Marketing at the address below.

This book is printed on acid-free paper.

Copyright © 1997 by MARCEL DEKKER, INC. All Rights Reserved.

Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic or
mechanical, including photocopying, microfilming, and recording, or by any information storage and retrieval system,
without permission in writing from the publisher.

MARCEL DEKKER, INC.

270 Madison Avenue, New York, New York 10016
http://www.dekker.com

Current printing (last digit):

10 9 8 7 6 5 4 3 2 1

PRINTED IN THE UNITED STATES OF AMERICA

 Page iii

PREFACE
The Institute of Food Technologists (IFT) sponsors a Basic Symposium in conjunction with its annual meeting for the
purpose of providing an overview of current topics of importance to food scientists. Each year the proceedings of the
symposium are published; this book is the result of the 1996 Basic Symposium.
Microbiology was last the topic of an IFT Basic Symposium in 1985, and our aim in 1996 was to provide an update of
some of the issues and technologies that have emerged in microbiological analysis since then, with special emphasis on
food safety applications. Symposium registrants heard presentations from leading experts on various subjects of
microbiological analysis. These subjects constitute the individual chapters of this book, authored by the symposium
presenters. The chapters are grouped into three parts: "Detection and Identification Systems", "Genetic Techniques of
Analysis", and "Methods for Assessment of Microbial Growth and Viability''.
The first part deals with general systems of microbiological analysis, beginning with an overview of recent
developments provided by Daniel Y. C.

 Page iv

Fung. Through his many research publications and reviews, and his legendary Rapid Methods Workshop held each
year at Kansas State Univesity, Dan has preached the gospel of rapid methods to more microbiologists than perhaps
anyone else. Although the reader unfortunately misses the sparkle of his live presentation, Dan's chapter provides a
thorough summary of the area, with a few predictions of future developments added to light the imagination.

Rapid detection of microorganisms that may be present in food at very low levels (e.g., pathogens) depends, to a large
extent, on the preparation of the food prior to analysis. Analytical methods that can boast of theoretical detection limits
at the single-cell level often fall short in practical application because of the interfering substances present in foods.
More research is needed on techniques for separating microorganisms from the food matrix and for concentrating them
for rapid detection. This generally neglected but crucial issue was the topic of Tony Sharpe's presentation at the
symposium. The thoughtful comments and insights in his chapter could have been written only by one with a great deal
of experience in food microbiology methods development. Dr. Sharpe is truly an expert and outstanding contributor to
the field.

The heavy focus on optical methods in Part I of the book is due to their potential as extremely rapid, direct systems of
detection. "Have You Hugged Your Microscope Today?" was the headline of the IFT Meeting Newsletter account of the
presentation by this co-editor. Her chapter in this book emphasizes the advantages of using modern innovations in
conjunction with the oldest microbiological toolthe microscopeto achieve not only specific, rapid detection but also
quantitation of microorganisms in foods. The flow cytometer is an instrument, borrowed primarily from immunological
research applications, that is frequently cited as an emerging technology in food microbiology. Rich Raybourne's
chapter includes a helpful discussion of the basics of flow cytometry and its use for the detection and, through the
instrument's cell sorting capacity, isolation of microbial cells in food. Continuing the subject of optical detection, Nile
Hartman provides details of his biosensor for detection and quantitation of microbial cells in food. Most efforts in
biosensor research are geared toward analysis of soluble substances (for example, microbial toxins), but Nile's work is
distinguished for development of the technology for analysis of whole microbial cells in foods. His optical biosensor is
currently undergoing field trials in food production facilities.
Part I also includes discussions of the well-established workhorses of the microbiological lab, those systems of
identification that enjoy a high level of commercial popularity because of their ease of use, high-volume through-

 Page v

put, and reliability. The chapter on immunodiagnostics, by Barb Robison, is an excellent overview of commercially
available "kits" based on the specific recognition afforded by antibody-antigen reactions. Some of these kits were just
in development in 1985, the last time the IFT Basic Symposium addressed food microbiology, when ELISA technology
was only beginning to make its impression on the food microbiological world. Now we depend on them heavily as
rapid screening assays and as confirmatory techniques of identification. It is difficult to imagine a more comprehensive
review of currently available automated systems of identification than the carefully written chapter by Charlie Stager.
Marketed primarily to the clinical laboratory, these automated systems are making inroads into the food microbiology
arena. The chapter will serve as a valuable reference for anyone contemplating an investment in such a system.

There has been a great deal of activity in the development of nucleic acid-based methods useful for detection and
identification of foodborne pathogens. Some of the technologies that are well established today, such as assays based
on DNA probe hybridization, were relatively new marvels in 1985. Some technologies, now in an explosive stage of
growth in applications (e.g., PCR, genetic fingerprinting), were just being invented then. Part II presents chapters
concerned with various genetic techniques of analysis.

The extensive chapter on gene probes for detection of foodborne pathogens, by Karen Jinneman and Walt Hill,
discusses the basics of nucleic acid hybridization, from probe design to assay format, followed by an all-inclusive
review of gene probe applications. The current literature on detection of classical and emerging foodborne pathogens or
their products is thoroughly summarized, and the chapter includes a section on the expanding literature on foodborne
viruses. A similarly exhaustive review of the literature on PCR and other nucleic acid amplification techniques is
provided in the chapter by Mike Johnson and colleagues. There is increasing interest in the use of ribosomal RNA as a
target for detection because of its natural abundance in the cell and because of its ability to probe at various levels of
specificities. The chapter by Mark Mozola provides an elegant explanation of the theory of ribosomal RNA probe
design as well as applications for detection of the various foodborne pathogens. Together, these three chapters will
provide the reader with a comprehensive reference on the current state of gene detection methods in food safety
microbiology.
Bioinformatics is perhaps the most rapidly growing field in the life sciences today. As more and more genetic sequence
information is generated, there will be increased need for its assimilation into searchable and readily accessible
electronic databases. Steve Gendel summarizes in his chapter the various publicly accessible sequence databases and
their special functions.

 Page vi

With plenty of illustrations and information on how to log on to these databases, his chapter will serve as a valuable
guide for both the Internet novice and the experienced user.

From time to time it is reassuring to have at one's fingertips an instructional review of a particular topic of interest. Tim
Barrett's chapter is just such a primer of molecular fingerprinting techniques. The various methods currently in use are
diligently described, with a balanced perspective of their advantages and drawbacks, and the application of
fingerprinting technology in epidemiological investigations of foodborne disease outbreaks is discussed.

The microbial ecology of foods is a relatively unexplored frontier in food safety microbiology, and its investigation
will be made possible only by making available reliable methods for assessing microbial growth, viability, and
survival. Part III collects the symposium presentations related to these methods.
Part III begins with a discussion of the technology of bioluminescence and its use in food microbiological analysis.
Featured in the 1985 IFF Basic Symposium, bioluminescence is a technology with a long history of potential. Only
recently have there been a number of commercial ventures that exploit the phenomenon, and the acceptance of these
portable devices is increasing, particularly as effective tools in food sanitation and HACCP programs. The chapter by
Gordon Stewart and colleagues takes a different tack, however, with its emphasis on the use of bioluminescence as a
tool for studying microbial injury and survival at the molecular level.

Microorganisms are faced with a variety of stresses and inhibitory agents, particularly in a food production
environment, that may weaken but not necessarily kill them. As the food industry increasingly moves toward minimal
processing, we need to devote more effort to studying stressed or injured cells and consider their implications in food
safety microbiology. A preeminent contributor to the field of microbial ecology, Rita R. Colwell, has written
extensively on the viable but nonculturable state of microbial cells. Rita's chapter reminds us of the shaky foundation
of our belief in the standard plate count and reviews the methods used for assessing nonculturable populations of cells.

The measurement of microbial growth and activity by impedance has been practical for many years, and it was a
mature technology when it was included in the 1985 Symposium agenda. The commercial success of impedance-based
instruments has facilitated the push in development toward greater automation. Pat Rule revisits the basics of
impedance microbiology and provides updated examples that show the versatility of the technology.

Biofilm formation can be said to be the natural and preferred form of

 Page vii

growth for microorganisms in their habitats, and food production facilities provide microorganisms with a variety of
niches for biofilm establishment. There has been a recent surge in interest in experimental methods for studying
biofilms, and Edmund Zottola's chapter reviews these techniques. His beautiful photomicrographs are a stellar addition
to the book.

Microbiological food safety has always been a concern for the food industry, but never before has there been such
public awareness of the issue. There will be continued interest in development of improved assays for microbial
analysis in foods, and continued competition in the marketplace for these assays. An expanding number of
commercially available, affordable test kits based on molecular technologies will make it feasible to bring rapid,
specific microbiological analysis to virtually every food microbiology laboratory.

My co-editor, Steve Gendel, and I wish to thank the IFT staff who have been helpful not only in administration of the
1996 Basic Symposium but also in the task of publishing the proceedings. Topping the list are Anna May Schenck and
Patricia Schenck, who provided incomparable editing and indexing skills. We also thank Dean Duxbury and the
members of the IFT Basic Symposium Committee, who guided the development of this Symposium topic. Rich
McDonald, our FDA colleague and IFT Committee liaison, provided support and encouragement and was a calming
influence on more than one occasion.

MARY LOU TORTORELLO

 Page ix

CONTRIBUTORS
Timothy G. Aldsworth Department of Applied Biochemistry and Food Science, The University of Nottingham,
Leicestershire, England
Timothy J. Barrett Foodborne and Diarrheal Diseases Branch, Division of Bacterial and Mycotic Diseases, National
Center for Infectious Diseases, Centers for Disease Control and Prevention, Atlanta, Georgia

Rita R. Colwell University of Maryland Biotechnology Institute, College Park, Maryland

Christine E. R. Dodd Department of Applied Biochemistry and Food Science, The University of Nottingham,
Leicestershire, England

Daniel Y. C. Fung Department of Animal Sciences and Industry, Kansas State University, Manhattan, Kansas

Steven M. Gendel National Center for Food Safety and Technology, Food and Drug Administration, Summit-Argo,
Illinois

 Page x

Paula T. Gibson Department of Applied Biochemistry and Food Science, The University of Nottingham,
Leicestershire, England

Nile F. Hartman Optoelectronics and Chemical Sciences Division, Georgia Tech Research Institute, Atlanta, Georgia

Walter E. Hill Seafood Products Research Center, Seattle District Office, Office of Regulatory Affairs, U.S. Food and
Drug Administration, Bothell, Washington

Karen C. Jinneman Seafood Products Research Center, Seattle District Office, Office of Regulatory Affairs, U.S. Food
and Drug Administration, Bothell, Washington
Michael G. Johnson Department of Food Science, University of Arkansas, Fayetteville, Arkansas

Mark A. Mozola GENE-TRAK Systems, Hopkinton, Massachusetts

Richard B. Raybourne Food and Drug Administration, Laurel, Maryland

Barbara J. Robison Organon Teknika Corporation, Durham, North Carolina

P. Rule bioMérieux Vitek, Inc., Hazelwood, Missouri

Rachel L. Sharman Department of Applied Biochemistry and Food Science, The University of Nottingham,
Leicestershire, England

A. N. Sharpe Bureau of Microbial Hazards, Food Directorate, Health Protection Branch, Health Canada, Ottawa,
Ontario, Canada

Charles E. Stager Department of Pathology, Ben Taub General Hospital, Baylor College of Medicine, Houston, Texas

Gordon S. A. B. Stewart Department of Applied Biochemistry and Food Science, The University of Nottingham,
Leicestershire, England
Mary Lou Tortorello National Center for Food Safety and Technology, Food and Drug Administration, Summit-Argo,
Illinois

Rong-Fu Wang Division of Microbiology, National Center for Toxicological Research, Food and Drug Administration,
Jefferson, Arkansas

Debra K. Winters Department of Food Science, University of Arkansas, Fayetteville, Arkansas

Edmund A. Zottola Department of Food Science and Nutrition, University of Minnesota, St. Paul, Minnesota

 Page xi

CONTENTS

Preface iii

Contributors ix
Part I: Detection and Identification Systems

1. Overview of Rapid Methods of Microbiological Analysis 1

Daniel Y. C. Fung
2. Separation and Concentration of Pathogens from Foods 27
A. N. Sharpe
3. A New Look at an Old Technique: Application of Microscopy to Food 45
Microbiological Analysis
Mary Lou Tortorello

 Page xii

4. Flow Cytometry in Food Microbiology: Detection of Escherichia coli O157:H7 57

Richard B. Raybourne
5. Optical Biosensors for Microbiological Analysis 69
Nile F. Hartman
6. Immunodiagnostics in the Detection of Foodborne Pathogens 77
Barbara J. Robison
7. Automated Microbial Identification Systems 91
Charles E. Stager
Part II: Genetic Techniques of Analysis
8. Applications of Gene Probes for the Detection of Foodborne Pathogens 115
Karen C. Jinneman and Walter E. Hill
9. PCR and Nucleic Acid Amplification Methods 183
Michael G. Johnson, Rong-Fu Wang, and Debra K. Winters
10. Detection of Microorganisms in Foods Using DNA Probes Targeted to Ribosomal 207
RNA Sequences
Mark A. Mozola
11. Gene Sequence Databases and Food Microbiology 229
Steven M. Gendel
12. Molecular Fingerprinting of Foodborne Pathogenic Bacteria: An Introduction to 249
Methods, Uses, and Problems
Timothy J. Barrett
Part III: Methods for Assessment of Microbial Growth and Viability

13. Bioluminescence: lux as an Enabling Tool for the Microbiological Analysis of 265
Food
Gordon S. A. B. Stewart, Timothy G. Aldsworth, Rachel L. Sharman, Paula T. Gibson,
and Christine E. R. Dodd
14. Detection of Viable but Nonculturable and Stressed Microbial Cells 289
Rita R. Colwell

 Page xiii

15. Measurement of Microbial Activity by Impedance 305

P. Rule
16. Special Techniques for Studying Microbial Biofilms in Food Systems 315
Edmund A. Zottola
Index 347

 Page 1

1
Overview of Rapid Methods of Microbiological Analysis
Daniel Y. C. Fung
Kansas State University
Manhattan, Kansas

Introduction

Rapid methods and automation in microbiology are dynamic fields of study that address the utilization of
microbiological, chemical, biochemical, biophysical, immunological, and serological methods for the study of
improving isolation, early detection, characterization, and enumeration of microorganisms and their products in
clinical, food, industrial, and environmental samples. In the past 10 years, food microbiologists have started to adapt
rapid and automated methods in their laboratories (Fig. 1.1). Conventional methods of detection, enumeration
identification, and characterization of microbes are described in reference books such as Compendium of Methods for
the Microbiological Examination of Foods (Vanderzant and Splittstoesser, 1992), Official Methods of Analysis of the
AOAC (AOAC, 1990),

 Page 2

Fig. 1.1
Development of interest in rapid methods.
(From Fung, 1995.)

Bacteriological Analytical Manual (FDA, 1996), Standard Methods for the Examination of Dairy Products (APHA,
1992), and Modern Food Microbiology (Jay, 1996).

Important publications on the subject of rapid methods for medical specimens, water, food, industrial, and
environmental samples include a series of papers by Fung and colleagues (Fung, 1991, 1992, 1994, 1995; Fung et al.,
1989), and books such as Mechanizing Microbiology (Sharpe and Clark, 1978), Foodborne Microorganisms and Their
Toxins: Developing Methodology (Pierson and Stern, 1986), Rapid Methods in Food Microbiology (Adams and Hope,
1989), Instrumental Methods for Quality Assurance in Foods (Fung and Matthews, 1991), and Rapid Analysis
Techniques in Food Microbiology (Patel, 1994). Hartman et al. (1992) had an excellent chapter on rapid methods and
automation in the Compendium (Vanderzant and Splittstoesser, 1992). Swaminathan and Feng (1994) also provided
updated materials in rapid methods.

In a recent article in Food Technology, Fung (1995) described updated information concerning rapid detection of
foodborne organisms (Table 1.1). The historical progression of developments in rapid methods is listed in Table 1.2
(1992). Key issues and concerns in rapid methods are listed in Table 1.3 (1992).

The purpose of this chapter is to review the basic principles and practical applications of a variety of instruments and
procedures directly and indirectly related to improved methods for microbiology in quality assurance and research in
food microbiology.

 Page 3

TABLE 1.1 Needed Information Concerning Foodborne Pathogens

Morphology under magnification and on agar plates
Gram reaction and special staining properties
Biochemical activity profile and special enzyme systems
Pigment production, bioluminescence, chemiluminescence, and fluorescent compound
production
Nutritional and growth factor requirements
Temperature and pH requirements and tolerance
Fermentation products, metabolites, and toxin production
Antibiotic sensitivity pattern (antibiogram)
Gas requirements and tolerance
Genetic profile: DNA/RNA sequences and fingerprinting
Pathogenicity to animals and humans
Serology and phage typing
Cell wall, cell membrane, and cellular components
Growth rate constant and generation time
Ecological niche and survival ability
Motility and spore formation
Extracellular and intracellular products
Response to electromagnetic fields, light, sound, and radiation
Resistance to organic dyes and special compounds
Impedance, conductance, and capacitance characteristics
Others
Source: Fung, 1995.

Improvements in Sampling and Sample Preparation

One of the most useful instruments developed for sample preparation is the Stomacher (Tekmar, Cincinnati, OH). This
instrument is designed to massage food samples in a sterile bag. The food sample is first placed in the sterile disposable
plastic bag to which appropriate sterile diluents are added. The bag with the food is placed in the open chamber. After
the chamber is closed, the bag is massaged by two paddles for a suitable time period, usually 15 minutes. No contact
occurs between the instrument and the sample. During massaging, microorganisms are dislodged into the diluent for
further microbiological manipulation. Massaged slurries are then used for microbiological analysis. Sharpe and Jackson
(1975) and Emswiler et al. (1977) have shown that satisfactory results can be obtained by this method

 Page 4

TABLE 1.2 Historical Progression of Developments in Rapid Methods

Diagnostic test for liquid, semisolid, and solid media
Large tubes: One reaction per tube
Large tubes: Multiple reactions per tube
Small tubes: One reaction per tube
Small tubes: Multiple reaction per tube
Wells in tray with different configurations:
One type of reaction for many organisms per tray
Many types of reactions for one type of organism per tray
Many types of reactions for a few organisms per tray
Diagnostic kits
Inoculations into diagnostic tests
One inoculation per tube
Multiple inoculations (manually or by instruments)
Liquid in a tray
Solid in agar
Agar in multiple compartments
Liquid dispensing to multiple wells
Automated instruments for monitoring
Cell mass
Cell components
Cell metabolites
Cell activities
Serological and immunological techniques
Immunoblotting
Electrophoresis
RIA and ELISA
Magnetic immunocapture
Genetic techniques
DNA probes, RNA probes
Polymerase chain reaction and variations (e.g., Taqman)
Restriction fragment length polymorphism; ribotyping
Source: Fung, 1992

 Page 5

TABLE 1.3 Key Issues and Concerns

What is ''rapid"?
A. Speed of reaction: instantaneous, seconds, minutes, or hours versus
hours, days, weeks
B. Numbers of samples per oepration: 10, 50, 96, or more versus one at a
time
What is "automation"?
A. Manual: by humans, one or more sample at a time
B. Semi-automation: by humans, with the aid of some form of
instrumentation
C. Complete automation: by robots
Concepts involving the living cell
A. Living cell versus dead cell
B. Growing cell versus nongrowing cell
C. Meaning of viable cell count
D. Correlation between total count and other parameters
E. Amplification of cells
F. Concentration of cells
G. Signal versus background
H. Sensitivity versus detection time
Source: Fung, 1992.

compared with the conventional blending of foods. A similar system named Masticator Homogenizer is marketed by
IUL Instruments (Erglanger, KY). Recently Sharpe (personal communication, 1996) described a new instrument called
the Pulsifier, which can dislodge bacteria from food by pulsification in a bag.

One instrument used for sample preparation is the Gravimetric Diluter (Spiral Biotech, Bethesda, MD). One of the
routine procedures in food microbiological work is to aseptically measure a sample of food (e.g., 5 g of meat) and then
aseptically add an exact amount of sterile diluent (e.g., 45 mL) to make a desired dilution (1:10). With the Gravimetric
Diluter, the analyst needs only to aseptically place an amount of food (e.g., 5.3 g), into a Stomacher bag, set a desired
dilution (1:10), and set the instrument to deliver the appropriate amount of sterile diluent (e.g., 47.7 mL). Thus, the
dilution operation can be automatically done. The dilution factor can be chosen by the analyst to satisfy the need (1:10,
1:50, 1:100, etc.) simply by programming the instrument. Manninen and Fung (1992) evaluated the Gravimetric
Diluter and found that, depending on the volume tested, the

 Page 6

accuracy of delivery for most samples was in the range of 90100%. A new version of this instrument called Diluflo has
been in use satisfactorily in the author's laboratory since 1992. Recently Pbi of Italy marketed a similar instrument
named Dilumacher for automatic microbiological dilution.

Alternative Methods for Viable Cell Count Procedure

The conventional viable cell count or standard plate count method is time-consuming in terms of both operation and
collection of data. Several methods have been explored to improve the efficiency of operation of the viable cell count
procedure.

The spiral plating method is an automated system for obtaining viable cell counts (Spiral Biotech, Bethesda, MD). The
instrument can spread a liquid sample on the surface of agar contained in a Petri dish in a spiral shape (the Archimedes
spiral) with a concentration gradient starting from the center and decreasing as the spiral progresses outward on the
rotating plate. The volume of liquid deposited at any segment of the agar plate is known. After the liquid containing
microorganisms is spread, the agar plate is incubated overnight at an appropriate temperature for the colonies to
develop. The colonies appearing along the spiral pathway can be counted either manually or electronically. The spiral
system has been used in the United States for a variety of foods with satisfactory results (Schalkowsky, 1986).

New versions of the spiral plater recently introduced include the "Autoplater" (Spiral Biotech, Bethesda, MD) and the
Whitley Automatic Spiral Plater (Bioscience International, Inc., Rockville, MD). With these instruments an analyst
need only to present the liquid sample, which the instrument completely and automatically processes, then resterilizing
the unit for the next sample. A newer version of this system has been introduced by IUL in which the stylus is
automatically sliced after each use, thus by-passing the sterilization of the stylus by disinfectants used by the original
spiral platers.

One of the problems of the spiral system is clogging of the dispensing stylus by food particles. Konuma and Kurata
(1982) modified a stomacher bag by placing a filter in the bag so that the homogenized liquid poured from the bag will
be free of particles. Since then commercial companies have manufactured stomacher bags with filters in place. After
filtration the liquid presented to the spiral system will not clog the stylus.

There are many instruments now available for rapid enumeration of bacterial colonies in agar plate. Spiral Biotech
(Bethesda, MD) markets a

 Page 7

laser system called CASBA II, which has been in use in the author's laboratory for many years with satisfactory results.
Recently the same company introduced CASBA III, which includes a high-resolution scanner with a high-performance
personal computer and WindowsTM-based colony image analysis software. Protos colony counter (Bioscience
International, Inc., Rockville, MD) is another automatic colony counter, which can count 600 plates per hour at 0.5
seconds per plate. Countermat (IUL Instruments, Erlanger, KY) is another automatic colony counter, which has a
sample changer attached to the counting system to automatically place growth plates one after another under the
counter for colony enumeration. Many other companies also market automatic plate counters.

Another alternative method for viable cell count is the ISOGRID system (QA Laboratories Ltd., San Diego, CA). This
system consists of a square filter with hydrophobic grids printed on the filter to form 1600 squares for each filter. Food
samples are weighed, blended, and enzyme treated before passage through the membrane filter containing the
hydrophobic grids by vacuum. The filter is then placed on agar containing a suitable nutrient for growth of the bacteria,
yeasts, or molds. The hydrophobic grids prevent colonies from growing further than the square grids; thus all colonies
have a square shape. This facilitates counting of the colonies both manually and electronically. This method has been
successfully used to make viable cell counts for a variety of foods, including milk, meat, black pepper, flour, peanut
butter, mushrooms, rice, fish, shrimp, and oyster (Entis, 1983, 1984; Entis et al., 1982; Sharpe and Peterkin, 1988).
Other applications of the Isogrid system include determination of total coliforms, fecal coliforms, E. coli, Salmonella
spp., etc. (Sharpe, 1991).

Rehydratable nutrients are imbedded into a series of films in the Petrifilm System (3M Co., St. Paul, MN). The top
layer of the protective cover is lifted and 1 mL of liquid sample is introduced to the center of the unit, and the cover is
then replaced. A plastic template is placed on the cover to make a round mold. The rehydrated medium with nutrient
will support the growth of microorganisms after suitable incubation time and temperature. The colonies can be counted
directly in the unit. The unit is about the size and thickness of a plastic credit card, thus providing great savings of
space in storage and incubation. Petrifilm units have been developed for total bacterial count, coliform count, fecal
coliform count, yeast and mold counts, and hemorrhagic E. coli O157:H7 kit. Petrifilm has been successfully used to
count organisms from milk and meat (Fung et al., 1987; Ginn et al., 1986; Smith et al., 1986).
Redigel (RCR Scientific, Inc., Goshen, IN) is another convenient viable cell count system. This system consists of
sterile nutrients with a pectin gel in

 Page 8

a tube. The tube is ready to be used any time and no heat is needed to "melt" agar. A 1-mL food sample is first pipetted
into the tube. After mixing, the entire content is poured into a special Petri dish previously coated with calcium. When
liquid comes in contact with the calcium, a Ca-pectate gel is formed and the complex swells to resemble conventional
agar. After an appropriate incubation time and temperature, the colonies can be counted exactly like conventional
standard plate count method. Besides total count, Redigel also has systems for coliform, fecal coliform, yeast and
molds, lactic acid bacteria, Staphylococcus aureus, and Salmonella. Fung and Chain (1991) compared 17 different
foods (pasteurized milk, raw milk, cheddar cheese, chocolate chips, rice, wheat germ, corn meal, whole wheat, flour,
peanuts, pecans, ground beef, chicken, ground black pepper, thyme, broccoli, mushrooms, and turkey pot pie20
samples each) and obtained a correlation of 0.964.

The four aforementioned methods have potential as an alternative to the conventional agar pour plate method. Chain
and Fung (1991) made a comprehensive analysis of all four methods against the conventional method on seven
different foods (skinless chicken breast, fresh ground beef, fresh ground pork, packaged whole shelled pecans, raw
milk, thyme, and whole wheat flour20 samples each) and showed that the new systems and the conventional method
were highly comparable and exhibited a high degree of accuracy and agreement (r = 0.95+). It should be noted that
these newer methods require training and experience before satisfactory results can be consistently obtained. They are
good methods if operated carefully.
In the Direct Epifluorescent Filter Technique (DEFT) method, the liquefied sample is first passed through a filter that
retains the microorganisms, the filter is then stained with acridine orange, and the slide is observed under ultraviolet
microscopy. "Live" cells usually stain orange-red, orange-yellow, or orange-brown, whereas "dead" cells fluoresce
green. The slides can be read by the eye or by a semi-automated counting system marketed by Bio-Foss. A "viable cell
count" can be made in less than an hour. With the use of an image analyzer, an operator can count 50 DEFT slides per
hour (Pettipher, 1986). This method has been used satisfactorily for counting viable cells in milk and other food
samples such as fish (Pettipher, 1989). Recently the Nordic countries are using this method for quality assurance of
ground beef. Tortorello and Gendel (1993) further developed this procedure by using fluorescent antibodies in
conjunction with DEFT to enumerate E. coli O157:H7 in milk and juice.
These alternative methods for viable cell counts of microorganisms in foods provide an intermediate step toward
replacing the cumbersome conventional method. These methods are now accepted for a variety of food products, and
their use will continue to be adapted in years to come.

 Page 9

Instruments for Estimation of Microbial Populations and Biomass

Many methods have been developed in recent years to estimate the total number of microorganisms by parameters
other than the viable colony count, as described in the previous section. In order for a new method to be acceptable, it
should have some direct correlation with the total viable cell count. Thus, standard curves correlating parameters such
as adenosine triphosphate (ATP) level, detection time of electrical impedance or conductance, generation of heat,
radioactive CO2, etc., with viable cell counts of the same sample series must be made. In general, the larger the number
of viable cells in the sample, the shorter the detection time of these systems. A scattergram is then plotted and used for
further comparison of unknown samples. The assumption is that as the number of microorganisms increase in the
sample, these physical, biophysical, and biochemical events will also increase accordingly.
Theoretically, these methods can detect as little as one viable cell in the sample if the incubation period is long enough
(days or weeks). On the practical side, the limit is usually 103104 cells/mL When a sample has 105106 organisms/mL
detection can be achieved in about 46 hr.

All living things utilize ATP. In the presence of a firefly enzyme system (luciferase and luciterin system), oxygen, and
magnesium ions, ATP will facilitate the reaction to generate light. The amount of light generated by this reaction is
proportional to the amount of ATP in the sample; thus, the light units can be used to estimate the biomass of cells in a
sample. The light emitted by this process can be monitored by a variety of fluorimeters. These procedures can be
automated for handling of large numbers of samples. Some of the instruments can detect as little as 102103 fg. The
amount of ATP in one colony-forming unit has been reported as 0.47 fg with a range of 0.221.03 fg.

Using this principle, many researchers have tested the efficacy of using ATP to estimate microbial cells in foods and
beverages. Littel et al. (1986) indicated that the ATP procedure was able to predict bacterial levels within 0.5 log10, of
the actual count for beef and chicken samples. Minimum sensitivity is 5 × 104 colony-forming units/g of meat sample.
Ward et al. (1986) also found positive correlation between the ATP method and the conventional method in evaluating
fish samples.

Lumac (Landgraaf, the Netherlands) markets several models of ATP instruments and provides customers with test kits
with all necessary reagents, such as a fruit juice kit, hygiene monitoring kit, etc. The reagents are injected into the
instrument automatically, and readout is reported as relative light units (RLUs). By knowing the number of
microorganisms

 Page 10

responsible for generating known RLUs, one can estimate the number of microorganisms in the food sample. In some
food systems, such as wine, the occurrence of any living matter is undesirable; thus, monitoring of ATP can be a useful
tool for quality assurance in the winery. Recently, much interest has been expressed in using ATP estimation not for
total viable numbers but as a sanitation check by companies such as Lumac, BioTrace (Plainsboro, NJ), Lightning
(IDEXX, Westbrook, ME), Hy-Lite (Glengarry, Biotech, Cornwall, Canada), Charm 4000 Luminometer (Charm
Sciences, Malden, MA), and others. BioProbe is a system that can directly monitor the ATP level in flat surfaces
without using a swab for sampling.

As microorganisms grow and metabolize nutrients, large molecules change to smaller molecules in a liquid system and
cause a change in electrical conductivity and resistance in the liquid as well as at the interface of electrodes. By
measuring the changes in electrical impedance, capacitance, and conductance, the number of microorganisms in the
liquid can be estimated, because the larger the number of microorganisms in the fluid, the faster the change in these
parameters, which can be measured by sensitive instruments. A detailed analysis on the subject of impedance,
capacitance, and conductance in relation to food microbiology has been made by Eden and Eden (1984).

The Bactometer (bioMerieux Vitek, Inc., Hazelwood, MO) is an instrument designed to measure impedance changes in
foods. Samples are placed in the wells of a 16-well module. After the module is completely or partially filled, it is
plugged into the incubator unit to start the monitoring sequence. At first, there is a stabilization period for the
instrument to adjust to the module, then a base line is established. As the microorganisms metabolize the substrates and
reach a critical number (1056 cells/mL), change in impedance increases sharply, and the monitor screen shows a slope
similar to the log phase of a growth curve. The point at which the change in impedance begins is the "detection time,"
and this is measured in hours from the start of the experiment. The detection time is inversely proportional to the
number of microorganisms in the sample. By knowing the number of microorganisms per milliliter in a series of liquid
samples and the detection time of each sample, one can establish a standard curve. From the curve one can decide the
cutoff points to monitor certain specifications of the food products. For example, if one finds that meat with 106
organisms/g when applied to the instrument will result in a detection time of 5 hr, then one can use 5 hr as a cutoff
point for an indicator that meat samples have fewer or more than 106 organisms/g. A similar meat sample with a
detection time of 3 hr would indicate that the meat has more than 106 organisms/g. Conversely, a similar meat with a
detection time of 7 hr would indicate that the meat has less than 106 organisms/g.

 Page 11

Impedance methods have been used to estimate bacteria in milk, dairy products, meats, and other foods (Bishop and
White, 1985; Eden and Eden, 1984; Waes and Bossuyt, 1984; Zindulis, 1984). Of particular interest is the application
of this method for determining the shelf life potential of pasteurized whole milk by Bishop et al. (1984).

The Malthus system (Crawley, UK) works by measuring the conductance of the fluid as the organisms grow in the
system. It also generates a conductance curve similar to the impedance curve of the Bactometer, and it uses detection
time for monitoring the density of microorganisms in the food. The major difference between the two systems, besides
the scientific principle (impedance versus conductance), is the incubation units. In the Bactometer system, the size of
the well (about 2 mL capacity) in the 16-well module is fixed. No modification is possible because the module is
designed to fit into the incubator chamber. The unit (with two separate chambers) can be operated at two different
temperatures at the same time. The temperature is controlled by heating air. The Malthus system, however, allows
analysts to choose three sizes ranging from 2- to 100-mL samples, depending on the sample involved. The temperature
unit is controlled by heating water. Another important difference is that the modules of the Bactometer are disposable,
whereas the jars, tubes, and electrodes of the Malthus system are autoclavable and reusable. In terms of performance,
the systems are equivalent in sensitivity and detection time.
The Malthus system has been used for microbial monitoring of brewing liquids (Evans, 1988), milk (Visser and
deGroote, 1984), and hygiene monitoring (McMurdo and Whyward, 1984). Gibson and colleagues (1987, 1980;
Ogden, 1986) have done a considerable amount of work using the Malthus system to study seafood microbiology.

Besides estimating viable cells in foods, both the Bactometer and the Malthus systems can detect specific organisms by
the use of selective and differential liquid media. New developments of these two systems are constantly being made.
For example, Malthus has developed a tube system to detect CO2 production by yeast using indirect conductance
measurements. They also introduced disposable units in the system. An automatic instrument for measuring direct and
indirect impedance has been developed in Europe and named RABIT (Rapid Automated Bacterial Impedance
Technique).
An instrument called the "Omnispec bioactivity monitor system" (Wescor, Inc., Logan, UT) is a tri-stimulus reflectance
colorimeter that monitors dye pigmentation changes mediated by microbial activity. Dyes can be used that produce
color changes as a result of pH changes, changes in the redox potential of the medium, or the presence of compounds
with free amino groups. Samples are placed in microtiter wells or other types of

 Page 12

containers and are scanned by an automated light source with computer interface during the growth stages (024 hr).
The change of color or hue (a*, b*, L) can be monitored similar to impedance curve and conductance curve. Manninen
and Fung (1992b) evaluated this system in a study of pure cultures of Listeria monocytogenes and food samples and
found high correlation coefficients (r) of 0.900.99 for pure bacterial cultures and 0.82 for minced beef between the
colony counts predicted by the colorimetric technique and the results of the traditional plate count method. They also
showed that detection times for bacterial cultures such as Enterobacter aerogenes, E. coli, Hafnia alvei, and several
strains of L. monocytogenes were substantially (224 hr) shorter using the instrument than using the traditional method
and concluded that the colorimetric detection technique employed by the "Omnispec" system simplifies the analyses,
saves labor and materials, and provides a high sampling capacity. Tuitemwong (1993) completed an extensive study
using Omnispec 4000 to monitor growth responses of food pathogens in the presence or absence of membrane-bound
enzymes. This instrument is highly efficient in large-scale studies of microbial interaction with different compounds in
liquid and food.

The catalase test is another rapid method for estimation of microbial populations in certain foods. Catalase is a very
reactive enzyme. Microorganisms can be categorized as either catalase positive and catalase negative. Both groups are
important in food microbiology; however, under certain food-storage conditions, a certain group predominates. Most
perishable foods (commercial as well as domestic) are cold stored under aerobic conditions. The organisms causing
spoilage of these foods are psychrotrophs. The predominant psychrotrophic bacteria are Pseudomonas spp., which are
strongly catalase positive. Other important psychrotrophs such as Micrococcus, Staphylococcus, and a variety of
enterics are also catalase positive. Thus, one can make use of the presence of catalase to estimate the bacterial
population. Catalase activity can be detected by a simple capillary tube method (Fung, 1985).

Another way to use the catalase test is as an index of the cleanliness of meat-processing areas. Preliminary data from
the author's laboratory indicate that a simple swab-catalase test can determine the degree of cleanliness after clean-up
of a meat-processing plant. A moist swab was applied to a 2 x 2 in. area and then the swab was placed in 2 mL of 3%
H2O2 in a test tube. The amount of bubbles generated in the tube can be scaled as 0, 1+, 2+, 3+, 4+, and 5+ depending
on the activity of sample. A bacterial count of the immediate adjacent area was also taken. The results showed that
prior to slaughter, the room had very low catalase activity, which indicated that bacteria as well as blood or other
particles that may carry catalase activity were present in low number. During the slaughter operation, the catalase

 Page 13

activity was high because of meat particles, blood, and bacteria in the environment. After proper clean-up, the catalase
activity in the area fell to the original level. Bacterial counts did not directly correlate with the catalase activity. This is
to be expected because the swab picks up material other than bacteria that can generate bubbles in the presence of
H2O2. Further work is in process to ascertain the value of this simple test for environmental sanitation monitoring.

Another exciting development in the catalase test is to use it to monitor end-point heating temperature of food. Ang et
al. (1994) recently showed that heating poultry to 71°C (a legal requirement for these products) will destroy both
bacterial and animal catalase. The test is 99% accurate and is simple and inexpensive to perform.

Miniaturized Microbiological Techniques

Identification of microorganisms is an important part of quality assurance and control programs in the food industry.
The author has developed many miniaturized methods to reduce the volume of reagents and media (from 510 mL to
about 0.2 mL) for microbiological testing in microtiter plates. The basic components of the miniaturized system are the
microtiter plates for test cultures, a multiple inoculation device, and containers to house solid media (large Petri dishes)
and liquid media (another series of microtiter plates). The procedure involves placing liquid cultures (pure cultures) to
be studied into sterile wells of a microtiter plate to form a master plate. Each microtiter plate can hold up to 96 different
cultures, 48 duplicate cultures, or various combinations as desired. The cultures are then transferred by a sterile
multipoint inoculator (96 needles protruding from a template) to solid or liquid media. Sterilization of the inoculator is
by alcohol flaming. Each transfer represents 96 separate inoculations in the conventional method. After incubation at
an appropriate temperature, the growth of cultures on solid media or liquid media can be observed and recorded, and
the data can be analyzed. These miniaturized procedures save a considerable amount of time in operation, effort in
manipulation, materials, labor, and space. These methods are ideal for studying large numbers of isolates or for
research involving challenging large numbers of microbes against a host of test compounds.

The miniaturized methods have been used to study large numbers of isolates from foods (Fung and Hartman, 1975; Lee
et al., 1982, 1985) and to develop bacteriological media and procedures (Chein and Fung, 1991). Miniaturized methods
for studying food yeast were also developed in the author's laboratory (Fung and Liang, 1989; Lin and Fung, 1985,
1987).

 Page 14

Currently, these miniaturized methods are being used to study food mycology (Hart and Fung, 1990; Hart et al., 1991).
Many useful microbiological media were discovered through this line of research. For example, an aniline blue
Candida albicans medium was developed and marketed by Difco under the name of Candida Isolation Agar. The
sensitivity and specificity were 98.0 and 99.5%, respectively, with a predictive value of 99.1% (Goldschmidt et al.,
1991).

On the commercial side, many diagnostic kits to identify microorganisms have been developed and marketed since the
1970s. Currently, API, Enterotube, R/B, Minitek, MicroID, and IDS are available. Most of these systems were first
developed for the identification of enterics (Escherichia coli, Salmonella, Shigella, Proteus, Enterobacter spp., etc.).
Later many of the companies expanded the capacity to identify nonfermentor, anaerobes, gram-positive organisms, and
even yeasts and molds. Most of the early comparative analyses centered around evaluation of these kits for clinical
specimens. Cox et al. (1977) and Fung and Cox (1981) studied these systems from the standpoint of food microbiology
and concluded that these systems generally provide 9095% accuracy when compared with conventional methods.
Comparative analysis of diagnostic kits and selection criteria for miniaturized systems were made by Fung et al. (1984)
and Cox et al. (1984). They concluded that these miniaturized systems are accurate, efficient, labor saving, space
saving, and cheaper than the conventional procedure. Their usefulness in clinical and food microbiological laboratories
will continue to be important.

Novel Techniques

Many sophisticated instruments have been developed to identify isolates from clinical specimens such as Sensititre
(Radiometer Amer, Westlake, OH) and Biolog (Hayward, CA). One of the most automated systems for the
identification of isolates (clinical and foods) is the Vitek system. This system depends on the growth of target
organisms in specially designed media housed in tiny chambers in a plastic "card." The card is then inserted into the
incubation chamber. The instrument periodically scans the wells of the cards and sends information to the computer,
which then matches the data base and identifies the unknown cultures in the cards. The system is entirely automated
and computerized and provides hard copies for recordkeeping. Most evaluations of the usefulness of the Vitek system
had previously used clinical specimens. Bailey et al. (1985) were successful in utilizing the Vitek system to identify
Enterobacteriaceae from foods. The system is capable of identifying enterics, yeast, Bacillus, selected gram-positive
pathogens, and other organisms.

 Page 15

Concepts and applications of miniaturized kits, immunoassays, and DNA hybridization for recognition and
identification of foodborne bacteria are reviewed by Cox et al. (1987). The DNA probe (Genetrak, Hopkinton, MA) is a
sensitive method to detect pathogens such as Campylobacter, Salmonella, Listeria, and Escherichia coli. At first, the
system utilized radioactive compounds for assay. The second generation of probes uses enzymatic reactions to detect
the presence of pathogens. Another major change in this area is the development of probes to detect target RNA. In a
cell there is only one copy of DNA, however, there may be 1,00010,000 copies of ribosomal RNA. Thus, the new
generation of probes is designed to probe target RNA. After enrichment of cells (either in liquid or colonies on solid
agar), DNA or RNA of target cells can be extracted and released into the liquid and then detected by the appropriate
DNA and/or RNA probes. These methods are currently being automated. Currently, kits are available for Salmonella,
Listeria, Campylobacter, and Yersinia, among others. As the need arises, more organisms will be added to the list.

Polymerase chain reaction (PCR) systems are the latest development in DNA amplification technology and have
recently gained much attention in food microbiology. Originally the procedures were highly complicated and a very
clean environment was needed to perform the test. Recently, much research has been directed at simplifying the
procedure for laboratory analysts. DuPont recently commercialized a system called BAXTM for PCR. According to
Scott Fritschel of QualiconTM LLC (a subsidiary unit of DuPont):
The BAXTM screening system is designed to work from overnight enrichment broths. In the case of the BAXTM system for Salmonella,
the food is enriched overnight in any standard non-selective broth (lactose broth, buffered peptone water, etc.). A 1:10 dilution into BHI is
performed followed by a 3 hour grow-back incubation. An aliquot is taken into a tube and treated with lytic enzymes and heat. An aliquot
of the cell lysate is used to rehydrate the PCR reagents housed in a tube and a control tablet and the sample is placed into the thermal
cycler along with the corresponding control tube. Thermal cycling, UV visualization of the gel and photography to document the results
are then performed to detect PCR products.

These types of developments will help to transfer PCR technology to food laboratories in the near future. Some
obstacles to PCR technology include the occurrence of inhibitors in the food samples and the fact that DNA from dead
cells can also be amplified, thus giving a false-positive result for a food that may be safe to eat. For pure cultures this
technology is very useful.

 Page 16

Dilution of samples (1:10) will reduce the inhibitors and enrichment of target cultures will ensure amplification of live
cells even in the presence of other organisms.

Another important development, also by DuPont, is the RiboPrinterTM Characterization System. Again according to
Scott Fritschel,
The RiboPrinterTM is designed to accept isolated colonies of bacteria as the sample. A sample is prepared as follows:

(1) A colony of bacteria is picked from an agar plate using a sterile plastic stick (provided in the sample kit).

(2) Cells from the stick are suspended in a buffer solution by mechanical agitation.

(3) An aliquot of the cell suspension is loaded into the sample carrier to be placed into the instrument. That's it! Each sample carrier has
space for eight individual colony picks.

In both products, we (QualiconTM) have invested a lot of research in allowing the user to work from samples that microbiologists are
comfortable handling. We do not want to require that the user be familiar with the techniques of molecular biology in order to benefit
from the power that these techniques can bring to food microbiology.

Modification of the basic PCR procedure includes reverse transcriptase PCR, nested/multiplex PCR, Randomly
Amplified Polymorphic DNA (RAPD), fluorescent probes in PCR, etc. When these methods will find their way to the
food microbiology laboratory is not certain.
The enzyme-linked immunosorbent assay (ELISA) method commercialized by Organon Teknika (Durham, NC)
utilizes two monoclonal antibodies specific for Salmonella detection. In a comparative study involving 1289 samples,
Eckner et al. (1987) found that there was no significant difference between the conventional method and the ELISA
method for food samples except cake mix and raw shrimp. Another ELISA system, the Tecra system (International
BioProducts, Redmond, WA), was developed in Australia and uses polyclonal antibodies to detect Salmonella. These
methods have also been used to detect Listeria and E. coli. Many companies are providing a host of monoclonal and
polyclonal antibodies for a variety of diagnostic tests, some including food pathogens (Fung et al., 1988).
In the VIDASTM system, all intermediate steps are automated. The VIDAS (Vitek ImmunoDiagnostic Assay System) is
a multiparametric immunoanalysis system. The system utilizes the enzyme-linked fluorescent immunoassay (ELFA)
method. According to the manufacturer, ''The end result of the test protocol is a fluorescent product and the VIDAS
reader utilizes a

 Page 17

special optical scanner that measures the degree of fluorescence. From the moment the solid phase receptacles and the
reagent strips are placed in the instrument, the VIDAS is fully automated." This is a revolutionary development
because one of the drawbacks of the ELISA test is the many steps necessary for adding reagents and washing test
samples. Many automated ELISA instruments are on the market now, including TECRA OPUS (International
BioProducts, Redmond, MD), Bio-tek Instruments (Highland Park, VT), and Automated EIA Processor (BioControl,
Bothell, WA).

In a related development, several self-contained small units (REVEAL, VIP, etc.) have been marketed recently. After
enrichment (with or without Oxyrase-type stimulation), an analyst only needs to apply a small aliquot (boiled or
unboiled) to the kit. Reaction occurs in a few minutes. These kits have been used to rapidly screen E. coli O157:H7 and
Salmonella in ground beef.

The UNIQUE system for Salmonella is another way of using immunocapture technology. In this system a dip stick
with antibody against Salmonella is applied to the preenriched liquid. The antibody captures the Salmonella if present.
This charged dip stick is then placed in a fresh enrichment broth and the cells are allowed to multiply for a few hours.
After the second enrichment step, the dip stick with a much larger population of Salmonella attached to it will be
subject to further ELISA procedures. The entire test is housed in a convenient plastic self-contained unit. This type of
method is very useful for a small laboratory with low-volume testing of pathogens.

Immunomagnetic capture methods have attracted much attention lately. VICAM and DYNAL developed magnetic
beads coated with a variety of antibodies to capture target cells or cellular components in foods. After the magnetic
beads have had a chance to interact with potential target cells in a tumbling apparatus for an hour, they are physically
separated from the food or liquid by a powerful magnet applied to the side of the test tube. Further microbiological
procedures such as direct plating or ELISA tests can be performed on these charged beads. Theoretically, one can
detect one living cell in a sample by this method without the enrichment step of the food sample. This is especially
important for foods with very low numbers of target pathogen such as Listeria or Salmonella.
Motility enrichment is a very useful concept in the rapid isolation and identification of food pathogens. Fung and Kraft
(1970) described a motility flask system for the rapid detection and isolation of Salmonella spp. from mixed cultures
and poultry products. The system involves a flask with a side arm that contains several agar layers. Lactose broth is
placed in the flask, and then a sample, (with or without salmonellae) is inoculated into the lactose broth. When
salmonellae are present, they will swim through the first level of agar containing selenite cysteine and sodium lauryl
sulfate, which inhibits other organisms but allows salmonellae to pass through.

 Page 18

Once salmonellae pass the first layer, they can grow and metabolize compounds in the second and third layers. By
looking at the color changes in the second and third agar layers, one can determine if salmonellae were in the sample
put into the lactose broth. Further serological tests can then be performed to confirm the presence of Salmonella.

In more recent years, a commercial system called the Salmonella 1-2 test (BioControl, Bothell, WA) was developed
that utilizes motility as a form of selection. The food sample is first preenriched for 24 hr in lactose broth, and then 0.1
ml is inoculated into one of the chambers in an L-shaped system. The chamber contains selective enrichment liquid
medium. There is also a small hole connecting the liquid chamber with the rest of the system, which contains a soft
agar through which salmonellae can migrate. An opening on the top of the second chamber allows the analyst to
deposit a drop of polyvalent anti-H antibody. If the sample contains salmonellae from the lower side of the L unit, they
will migrate through the hole and up the agar column. Simultaneously, the antibody against flagella of salmonellae will
diffuse downward through the agar. When the antibody meets the salmonellae, a visible "immunoband" will form. The
presence of an immunoband in this system is a positive test for Salmonella spp. The system is easy to use and has
gained popularity because of its simplicity.

Oxyrase® (Mansfield, OH), a membrane fraction of E. coli, was found in the author's laboratory to stimulate the
growth of a large number of important facultative anaerobic food pathogens. In the presence of a hydrogen donor such
as lactate, Oxyrase® can convert O2, to H2O, thereby reducing the oxygen tension of the medium and creating
anaerobic conditions that favor the growth of facultative anaerobic organisms. In a medium containing 0.1 units/mL of
the enzyme, the growth of Listeria monocytogenes, E. coli O157:H7, Salmonella typhimurium, Streptococcus faecalis,
and Proteus vulgaris were greatly enhanced; colony counts were greater by 12 log units, depending on the initial count
and the strain studied, after incubation in tile presence of the enzyme for 58 hr at 3542°C compared with control
without Oxyrase.®

By combining the Oxyrase® enzyme and a unique U-shaped tube, Yu and Fung (1991a,b, 1992) developed an effective
method to detect Listeria monocytogenes and other Listeria spp. from laboratory cultures and meat systems.
Niroomand and Fung (1994) studied the effects of Oxyrase® in stimulation of growth of Campylobacter from foods.
Tuitemwong et al. (1994) also found that these membrane fragments and those obtained from Acetobacter and
Gluconobacter can stimulate growth of starter cultures in food fermentation.

There are many other systems that involve modern biochemistry, chemistry, and immunology. For example, one can
use protein profiles for micro-

 Page 19

bial "fingerprinting" (AMBIS system, San Diego, CA) or cell composition as a way to identify bacterial cultures
(Hewlett-Packard, Palo Alto, CA).

Conclusions
This chapter describes a variety of methods that are designed to improve current methods, explore new ideas, and
develop new concepts and technologies for the improvement of applied microbiology. Although many of these
methods were first developed for clinical microbiology, they are being increasingly used for food microbiology. After
being involved in the field of rapid methods and automation in microbiology for 25 years, this author summarizes his
ideas concerning the attributes for an ideal automated microbiology assay system in Table 1.4. A new publication,
Journal of Rapid Methods and Automation, was started in 1992 by the author to encourage the rapid dissemination of
information concerning current developments in
TABLE 1.4 Attributes of an Ideal Automated Microbiology Assay System

1. Accuracy for the intended purposes

Sensitivityminimal detectable limits
Specficity of test system
Versatilitypotential applications
Comparison to referenced methods
2. Speed in productivity: in obtaining results
in number of samples processed per run; per day
3. Cost: initial, per test, reagents, others
4. Acceptability:
by scientific community
by regulatory agencies
5. Simplicity of operation:
sample operation
operation of test equipment
computer versatility
6. Training: on site; how long; quality of training personnel
7. Reagents: reagent preparation/stability/availability/consistency
8. Company reputation
9. Technical service: speed and availability; cost and scope of technical
background
10. Utility and space requirement
Source: Fung, 1992.

 Page 20

rapid methods and automation. This field will certainly grow, and many food microbiologists will find these new
methods very useful in their routine work in the immediate future. Many methods described here are already being
used by applied microbiologists nationally and internationally.

Acknowledgment
Contribution No. 96-552-B, Agricultural Experiment Station, Kansas State University, Manhattan, Ks. 66506. This
material is based on work supported by the Cooperative State Research Service, U.S. Department of Agriculture, under
agreement No. 8890341874511.

References

Adams, M.R. and Hope, C.F.A. 1989. Rapid Methods in Food Microbiology. Elsevier, Amsterdam.

Ang, C.Y.W., Liu, F., Townsend, W.E., and Fung, D.Y.C. 1994. Sensitive catalase test for end-point temperature of
heated chicken meat. J. Food Sci. 59: 494.

AOAC. 1990. Official Methods of Analysis of the AOAC (Association of Official Analytical Chemists). Vol. I and II.
Association of Official Analytical Chemists, Arlington, VA.

APHA. 1992. Standard Methods for the Examination of Dairy Products. American Public Health Association,
Washington, D.C.

Bailey, J.S., Cox, N.A., Thomson, J.E., and Fung, D.Y.C. 1985. Identification of Enterobacteriaceae in foods with the
Automicrobic System. J. Food Protect. 48: 147.

Bishop, J.R. and White, C.H. 1985. Estimation of potential shelf-life to cottage cheese utilizing bacterial numbers and
metabolites. J. Food Prot. 48: 663.

Bishop, J.R., White, C.R., and Firstenberg-Eden, R. 1984. A rapid impedimetric method for determining the potential
shelf-life of pasteurized whole milk. J. Food Prot. 47: 471.
Chain, V.S. and Fung, D.Y.C. 1991. Comparison of Redigel, Petrifilm, Spiral Plate System, ISOGRID and standard
plate count for the aerobic count on selected foods. J. Food Prot. 54: 208.

Chein, S.P. and Fung, D.Y.C. 1991. Acriflavin violet red bile agar for the isolation and enumeration of Klebsiella
pneumoniae. Food Microbiol. 7: 73.

 Page 21

Cox, N.A., McHan, F., and Fung, D.Y.C. 1977. Commercially available minikits for the identification of
Enterobacteriaceae: A review. J. Food Prot. 40: 866.

Cox, N.A., Fung, D.Y.C., Goldschmidt, M.S., Bailey, J.S., and Thomson, J.E. 1984. Selecting a miniaturized system
for identification of Enterobacteriaceae. J. Food Prot. 47: 74.

Cox, N.A., Fung, D.Y.C., Bailey, J.S., Hartman, P. A., and Vasavada, P.C. 1987. Miniaturized kits, immunoassays and
DNA hybridization for recognition and identification of foodborne bacteria. Dairy Food Sanit. 7: 628.

Eckner, K.F., Flowers, R.S., Robinson, B.J., Mattingly, J.A., Gabis, D.A., and Silliker, J.A. 1987. Comparison of
Salmonella Bio-EnzaBead immunoassay method and conventional culture procedure for detection of Salmonella in
foods. J. Food Prot. 50: 379.

Eden, R. and Eden, G. 1984. Impedance Microbiology. Research Studies Press, Letchworth, UK.

Emswiler, B.S., Pierson, C.J., and Kotula, A.W. 1977. Stomaching vs. blending. Food Technol. 31(10): 40.

Entis, P. 1983. Enumeration of coliforms in nonfat dry milk and canned custard by hydrophobic grid membrane filter
method: Collaborative study. J. Assoc. Off. Anal. Chem. 66: 897.

Entis, P. 1984. Enumeration of total coliforms and Escherichia coli in foods by hydrophobic grid membrane filters
collaborative study. J. Assoc. Off. Anal. Chem. 67: 812.

Entis, P., Brodsky, M.H., and Sharpe, A.N. 1982. Effect prefiltration and enzyme treatment on membrane filtration of
foods. J. Food Prot. 45: 8.

Evans, H.A.V. 1985. A note on the use of conductivity in brewery microbiology control. Food Microbiol. 2: 19.

FDA. 1996. Bacteriological Analytical Manual, 6th ed. Association of Official Analytical Chemists. Food and Drug
Administration, Arlington, Va.

Fung, D.Y.C. 1985. Procedures and methods for one-day analysis of microbial loads in foods. In: Rapid Methods and
Automation in Microbiology and Immunology, K.-O. Habermehl (Ed.), p. 656664. Springer-Verlag, Berlin.

Fung, D.Y.C. 1991. Rapid methods and automation for food microbiology. In: Instrumental Methods for Quality
Assurance in Foods. D.Y.C. Fung and R. F. Matthews (Ed.). Marcel Dekker, Inc., New York.

Fung, D.Y.C. 1992. Historical development of rapid methods and automation in microbiology. J. Rapid Methods
Autom. Microbiol. 1(1): 1.

Fung, D.Y.C. 1994. Rapid methods and automation in food microbiology: A review. Food Rev. Int. 10(3): 357375.

 Page 22

Fung, D.Y.C. 1995. What's needed in rapid detection of food borne pathogens. Food Technol. 44(6): 6467.

Fung, D.Y.C. and Chain, V.S. 1991. Comparative analysis of Redigel and aerobic plate count methods for viable cell
count of selected foods. Food Microbiol. 8: 299.

Fung, D.Y.C. and Cox, N.A. 1981. Rapid identification systems in the food industry: Present and future. J. Food Prot.
44: 877.

Fung, D.Y.C. and Hartman, P.A. 1975. Miniaturized microbiology techniques for rapid characterization of bacteria. In:
New Approaches to the Identification of Microorganisms, C.G. Heden and T. Illeni (Ed.), p. 347370. Wiley, New York.
Fung, D.Y.C. and Kraft, A.A. 1970. A rapid and simple method for the detection and isolation of Salmonella from
mixed cultures and poultry products. Poultry Sci. 49: 46.

Fung, D.Y.C. and Liang, C. 1989. A new fluorescent agar for the isolation of Candida albicans. Bulletin d'information
des Laboraores des Service Veterinaries (France). No. 29/30, p. 12.

Fung, D.Y.C. and Matthews, R.F. (Ed.). 1991. Instrumental Methods for Quality Assurance in Foods. Marcel Dekkel,
Inc., New York.

Fung, D.Y.C., Goldschmidt, M.C., and Cox, N.A. 1984. Evaluation of bacterial diagnostic kits and systems at an
instructional workshop. J. Food Prot. 47: 68.

Fung, D.Y.C., Hart, R.A., and Chain, V. 1987. Rapid methods and automated procedures for microbiological evaluation
of seafood. In: Seafood Quality Determination, D.E. Kramer and J. Liston (Ed.), p. 247253. Elsevier, Amsterdam.

Fung, D.Y.C., Bennett, R., and Lehleitner, G.C. 1988. Rapid diagnosis in bacteriology: Contribution of polyclonal and
monoclonal antibodies. In: Biologie Prospective, le Colloque de Pont-e-Mousson, M.M. Gaeteau, J. Henry, and G.
Siest (Ed.), p. 2126. John Libbey, London.

Fung, D.Y.C., Cox, N.A., Goldschmidt, M.C., and Bailey, J.S. 1989. Rapid methods and automation: A survey of
professional microbiologists. J. Food Prot. 52: 65.

Gibson, D.M. and Hobbs, G. 1987. Some recent developments in microbiological methods in seafood quality. In:
Seafood Quality Determination, D.F. Kramer and J. Listen (Ed.), p. 283298. Elsevier, Amsterdam.
Gibson, D.M. and Ogden, I.D. 1980. Assessing bacterial quality of fish by conductance measurement. J. Appl.
Bacteriol. 49: 12.

Ginn, R.E., Packard, V.S., and Fox, T.L. 1986. Enumeration of total bacteria

 Page 23

and coliforms in milk by dry rehydratable film method: Collaborative study. J. Assoc. Off. Anal. Chem. 69: 527.

Goldschmidt, M.C., Fung, D.Y.C., Grant, R., White, J., and Brown, T. 1991. New aniline blue dye medium for rapid
identification and isolation of Candida albicans. J. Clin. Microbiol. 29(6): 10981099.

Hart, R.A. and Fung, D.Y.C. 1990. Evaluation of dye media selective for Aspergillus and/or Penicillium. In: Proc. Ann.
Mtg. Am. Soc. Microbiol. Anaheim, CA, May 1317, p. 279.

Hart, R.A., Mo, O., Borius, F., and Fung, D.Y.C. 1991. Comparative analysis of Trypan blue agar and Congo red agar
for the enumeration of yeast and mold using HGMF system. J. Food Safety 11: 227.
Hartman, P.A., Curiale, M.S., Sharpe, A.N., Swaminathan, B., Eden, R., Cox, N.A., Goldschmidt, M.C., and Fung,
D.Y.C. 1992. Rapid methods and automation. In: Compendium of Methods for the Examinations of Foods, (C.
Vanderzant and D. Splittstoesser (Ed.). American Public Health Association, Washington, D.C.

Jay, J.M. 1996. Modern Food Microbiology, 5th ed. Chapman and Hall, New York.

Konuma, H.A. and Kurata, H. 1982. Improved Stomacher 400 bag applicable to the spiral plate system for counting
bacteria. Appl. Environ. Microbiol. 44: 765.

Lee, C.Y., Fung, D.Y.C., and Kastner, C.L. 1982. Computer-assisted identification of bacteria on hot-boned and
conventionally processed beef. J. Food Sci. 47: 363.

Lee, C.Y., Fung, D.Y.C., and Kastner, C.L. 1985. Computer-assisted identification of microflora on hot-boned and
conventionally processed beef: Effect of moderate and slow initial chilling rate. J. Food Sci. 50: 553.

Lin, C.C.S. and Fung, D.Y.C. 1985. Effect of dyes on growth of food yeast. J. Food Sci. 47: 770.

Lin, C.C.S. and Fung, D.Y.C. 1987. Critical review of conventional and rapid methods for yeast identification. CRC
Crit. Rev. Microbiol. 14(4): 273.

Littel, K.J., Pikelis, S., and Spurgash, A. 1986. Bioluminescent ATP assay for rapid estimation of microbial numbers in
fresh meat. J. Food Prot. 49: 18.

Manninen, M.T. and Fung, D.Y.C. 1992a. Use of the Gravimeter diluter in microbiological work. J. Food Prot. 55: 59.
Manninen, M.T. and Fung, D.Y.C. 1992b. Estimation of microbial numbers from pure bacterial cultures and from
minced beef samples by reflectance colorimetry with Omnispec 4000. J. Rapid Methods Autom. Microbiol. 1(1): 41.

 Page 24

McMurdo, I.H. and Whyward, S. 1984. Suitability of rapid microbiological methods for the hygiene management of
spray drier plant. J. Soc. Dairy Technol. 34(1): 4.

Niroomand, F. and Fung, D.Y.C. 1994. Effect of oxygen reducing membrane fragments on growth of Campylobacter.
J. Rapid Methods Autom. Microbiol. 2: 247.

Ogden, I.D. 1986. Use of conductance methods to predict bacterial counts in fish. J. Appl. Bacteriol. 61: 36.

Patel, P.D. 1994. Rapid Analysis Techniques in Food Microbiology. Chapman and Hall, New York.
Pettipher, G.L. 1986. Review: The direct epifluorescent filter technique. J. Food Technol. 21: 535.

Pettipher, G.L. 1989. The direct epifluorescent filter technique. In: Rapid Microbiological Methods, M.R. Adams and
C.F.A. Hope (Ed.). Elsevier, New York.

Pierson, M.D. and Stern, N.J. 1986. Foodborne Microorganisms and Their Toxins: Developing Methodology. Marcel
Dekker, New York.

Schalkowsky, S. 1986. Plating systems. In: Foodborne Microorganisms and Their Toxins: Developing Methodology,
M.D. Pierson and N.J. Stern (Ed.), p. 1626. Marcel Dekker, New York.

Sharpe, A.W. 1991. Rapid methods: Consideration to adoption by regulatory agencies. Proceedings 105 AOAC Annual
International Meetings, Aug. 1215, 1991, Phoenix, AZ.

Sharpe, A.W. and Clark. D.S. (Ed.). 1978. Mechanizing Microbiology. Thomas, Springfield, IL.

Sharpe, A.W. and Jackson, A.K. 1975. Automation requirements in microbiological quality control of foods. In:
Automation in Microbiology and Immunology, C.G. Heden and T. Illeni (Ed.), p. 117124. Wiley, New York.

Sharpe, A.W. and Peterkin, P.I. 1988. Membrane Filter Food Microbiology. Research Studies Press, Letchworth, UK.

Smith, L.B., Fox, T.L., and Busta, F.F. 1986. Comparison of a dry medium culture plate (Petrifilm SM plates) method
to the aerobic plate count method for enumeration of mesophilic aerobic colony-forming units in fresh ground beef. J.
Food Prot. 48: 1044.
Swaminathan, B. and P. Feng. 1994. Rapid detection of food-borne pathogenic bacteria. Rev. Microbiol. 48: 401.

Tortorello, M. and Gendel, S.M. 1993. Fluorescent antibodies applied to direct epifluorescent filter techniques for
microscopic enumeration of Escherichia coli O157:H7 in milk and juice. J. Food Prot. 56: 672.

 Page 25

Tuitemwong, K. 1993. Characteristics of food grade membrane bound enzymes and applications in food microbiology
and food safety. Ph.D. Dissertation, Kansas State University Library, Manhattan, KS.

Tuitemwong, K., Fung, D.Y.C., and Tuitemwong, P. 1994. Acceleration of yoghurt fermentation by bacterial
membrane fraction biocatalysis. J. Rapid Methods Autom. Microbiol. 3: 127,

Vanderzant, C. and Splittstoesser, D. (Ed.) 1992. Compendium of Methods for the Examination of Foods. American
Public Health Association, Washington, D.C.

Visser, I.J.R. and deGroote, J. 1984. Prospects for the use of conductivity as an aid in the bacteriological monitoring of
pasteurized milk. Antonin van Leewenhoek 50: 202.

Waes, G.M. and Bossuyt, R.G. 1984. Impedance measurements to detect bacteriophage problems in cheddar cheese. J.
Food Prot. 47: 349.

Ward, D.R., LaRocco, K.A., and Hopson, D.J. 1986. Adenosine triphosphate bioluminescent assay to enumerate
bacterial numbers on fresh fish. J. Food Prot. 49: 647.

Yu, L.S.L. and Fung, D.Y.C. 1991a. Oxyrase® enzyme and motility enrichment Fung-Yu tube for rapid detection of
Listeria monocytogenes and Listeria spp. J. Food Safety 11: 149.
Yu, L.S.L. and Fung, D.Y.C. 1991b. Effect of Oxyrase® enzyme in Listeria monocytogenes and other facultative
anaerobes. J. Food Safety 11: 163.

Yu, L.S.L. and Fung, D.Y.C. 1992. Growth kinetics of Listeria in the presence of Oxyrase® enzyme in a broth model
system. J. Rapid Methods Autom. Microbiol. 1(1): 15.

Zindulis, J. 1984. A medium for the impedimetric detection of yeasts in foods. Food Microbiol. 1: 159.

 Page 27

2
Separation and Concentration of Pathogens from Foods
A. N. Sharpe
Health Protection Branch, Health Canada
Ottawa, Ontario, Canada

Introduction

The ability to detect (presence or absence) bacterial pathogens at the level of one viable cell in 25 g of food within 30
min would permit, for example, rejection of shipments at a loading bay, diversion of materials to no, mild, or complete
processing, even monitoring of Hazard Analysis Critical Control Points (HACCP). But current identification kits and
systems are mainly immunological, DNA hybridization, or conductance-based, needing levels of at least 105
organisms/mL for reliable detection. To reach 105/mL from 1 cell/25 g means a concentration gain of about 107. This is
22 doublingsor the familiar overnight enrichment. DNA amplification methods cannot be applied directly to 25-g food
samples. But assuming we could satisfactorily extract the target microbe into a sample aliquot suitable for the
Polymerase Chain Reaction and amplify it without noise, a 5-min PCR cycle would take 2 hr to achieve a
concentration gain of, say 106, plus the time needed to detect the amplified product. So PCR-type reactions are
potential candidates for rapid tests. Immunomagnetic separation methods, combined with PCR,

 Page 28

seem to be exploding in popularity at the moment, though I am not convinced of their ultimate potential and ultimately
I believe physical or chemical ways of separating pathogens from foods will provide the necessary concentration gains
for detection faster than DNA amplifications.

The Case for Separation and Concentration

Current detection methods require many microbial cells or DNA/RNA templates because they use reagent volumes of
10100 mL or more. It is the need to amplify microbial signals to detectable levels in these volumes that wastes time.
But the detectability of microbes depends not on their number, but rather on their concentration, which in turn affects
the signal-to-noise ratio. An alternative approach uses physical, chemical, or immunological methods to improve the
microbial signal-to-noise ratio by separating them from the food components and reducing the volume of the
suspension in which they exist.

Separating and concentrating a target species from food would enable us to employ many rapid identification
procedures that currently are impractical owing to poor signal-to-noise ratios, for example, by microscopes or flow
cytometers. The concentration gains needed to improve microbial signal-to-noise ratios for direct detection are
enormousin the order of 107 or greater. Even the ''DEFT-type" enzyme/surfactant/membrane filtration
treatmentcurrently one of the best separation methods and quite successful for spoilage organismsneeds improving by
several orders of magnitude before it can be applied to direct pathogen detection at the limiting regulatory level. In
various parts of the world (Europe in particular), workers are examining the use of lectins, immunomagnetic particles,
standing ultrasonic waves, dielectrophoresis, and other techniques to separate pathogens.

General Approaches to Removal, Separation and Detection

Among the approaches to direct detection of pathogens in foods are:

1. Extraction of whole target cells from the food, and identification after a concentration step(s), utilizing the cell's
phenotypy such as serological or enzymological properties.

2. Chemical extraction (most probably of DNA or RNA) to provide a relatively consistent base for analysis, regardless
of the food/species starting point.

 Page 29

3. Direct detection of pathogens in the food, by analyses which confer on the target cells some property (light emission,
radioactivity, etc.) that is detectable against the background of the food.

For the forseeable future, it will be necessary first to remove microbes from a food into liquid suspension, and then
concentrate them. Reliable detection of pathogens then depends on removing them quantitatively from the food into a
primary suspension, then concentrating them quantitatively from this primary suspension into the detection device.

There is an urgent need for new methods to generate primary suspensions. To suspend a high proportion of
microorganisms from foods by present technology (rinses, stomachers, blenders, etc.), one must produce large volumes
of suspension (100250 mL). Yet most powerful separation techniques (immunomagnetic beads, flow cytometers, etc.)
cannot handle more than a few milliliters of suspension, and the cost of using expensive microbial-capture reagents in
large volumes is prohibitive. An important problem to solve or to avoid in separation (and also the most difficult first
step) is how to reduce a primary suspension to a volume of a few milliliters or less. This is extremely difficult because
the primary suspension is likely to have a very variable composition and because its volume prevents the use of
expensive materials. Once the suspension is reduced to a few milliliters, one can consider more "sophisticated"
techniques to further concentrate microorganisms.

How Efficiently Must Pathogens Be Extracted from Foods?

This important factor determines the credibility of all methods not depending on microbial multiplication. Much
depends on whether the goal is pathogen detection (presence/absence) or pathogen enumeration. For years, simple
detection was acceptable, but now we want to determine or monitor critical control points, and eventually regulatory
approval of HACCP-based food operations could be based on pathogen enumeration.

Any rapid pathogen detection method we develop will be measured against the performance of traditional detection
methods based on microbial multiplication, for which it is assumed that single viable cells can be detected, given
sufficient time and suitable preenrichment and selective enrichment conditions. In the traditional microbiological
approach, even if organisms exist in the food in a tightly bound or protected state, they will, under the conditions of the
incubation, escape into the menstruum. The test of a rapid method is not if there are >100 CFU/g in the analytical
sample, but when there is only a single viable cell; if it does not find that

 Page 30

single cell, it will prove to be inferior to the traditional method. Not to be regarded as inferior to the traditional
microbiology, either a "rapid" method must be capable of collecting and detecting every cell of the target organism or
the probability of analytical samples containing this limiting concentration of cells must be low.

For example, if the target organism occurs in the lot mainly as sporadic colonies of a few million cells, then samples
will tend to be either "clean" or heavily contaminated. Only a very small proportion of samples will contain target cells
at the limit of detectability of the rapid method, and it will not matter if the rapid method only detects samples of >100
organisms/g. However, if the target pathogen occurs uniformly through a lot at a level of 1 cell/g, then a method based
on separation might perform very poorly. Much will depend on the distribution of target cells in a typical lot.

Separation Efficiency and Speed

Currently, efficient primary detachment methods yield relatively large suspension volumes. Without a "capture means"
(filter, column, beads, etc.) of surface area and volume commensurate with the volume being processed, one has a low
probability of making contact with the target pathogen ("hits"). If the method cannot reach all of the target cells
quickly, it is condemned to being slow. How can we quickly capture target cells from large volumes at affordable
costs? My theory (Sharpe, 1991) is that separation/concentration processes will be faster if they are broken down into
several smaller steps, each of lesser efficiency, as follows.

Considering any separation process (centrifugation, membrane filtration, foam flotation, electrophoresis, ion-exchange,
affinity chromatography, etc.) without regard to the detection method (microscopy, flow-cytometry, luminescence,
etc.), one can calculate how long it takes to extract target cells in one stage or by two or more less ambitious stages.

One can consider a single-stage concentration process as the operation of passing a capture element (in reality, the final
volume) through a sample (which is N times greater in volume) until it has gone through the whole sample and
captured the target microorganisms from it. Assuming it takes t sec for the capture element to pass through its own
volume in the sample (specific sweep time) and that this remains constant during the capture pass, the time T required
for the capture element to reach the end of its pass is:

Consider now concentrating the sample in two stages, each of which is less efficient (i.e., yields a smaller increase in
analyte concentration) than a

 Page 31

single-stage process, i.e., instead of one we use n capture elements for the first stage, and after the sample has been
concentrated into these, we concentrate these n elements into one as in a single-stage process. One can show that the
total time required to separate microbes from the sample is:

T being a minimum when n = N1/2, i.e., when the overall concentration factor is achieved in two stages of
approximately equal effectiveness. Continuing, one can show that for a three-stage process the total separation time is
minimum when n = N1/3 and that p multiplexed processes execute in a minimum time when n = N1/p.
One can put this into perspective by imagining an extreme case of concentrating pathogens from 1000 mL of sample
suspension into a volume of 1.10-9 mL (one high-power field) for microscopy, ignoring practicalities such as the need
to prepare the product of one stage for introduction to the next (e.g., eluting from a column). In a typical example,
membrane filtration, where a 1-mL sample (1 × 1 × 1 cm) is reduced to 1.10-4 mL when it is captured as a layer
approximately 1 mm thick in 10 sec, the value of t = 1/1000 sec. Other processes (e.g., stirring with antibody-coated
beads) have different specific sweep times, but we can ignore this in illuminating the relative efficiency of multiplexed
procedures. By using t = 0.001 sec. for all steps in concentrating a 1000-mL sample into a 1.10-9-mL microscope field,
we find that for single-, two-, and three-stage processes, total concentration times are 1.109 sec (32 years), 2.103 sec (33
min), and 30 sec, respectively.

Even allowing for different specific sweep times for different processes and the practicalities of handling suspensions,
the effect of multiplexing separation steps shown by this simple treatment is so dramatic that one should keep it in
mind. That is, one should not expect too much of any one stage or the overall process slows down.

Primary Microbial Removal Methods

Because they cause less tissue disruption than bladed blenders while removing microbial cells, instruments of the
Stomacher type are useful in separating and concentrating microbes from foods. But since any rapid method will be
compared with tests in which presumably single viable cells are detectable, one should be concerned as to whether
stomaching removes the majority of target organisms from the food.
The original description and evaluation of the Stomacher (Sharpe and

 Page 32

Jackson, 1972) used peptone diluent, and later Sharpe and Harshman (1976) recommended addition of 1% Tween 80
surfactant to improve microbial yields from foods of moderate to high fat concentrations. The Stomacher has been
extensively compared against other methods, with 8 studies in which it yielded about the same count as a bladed
blender, 6 in which there was no difference, and about 20 in which it was slightly lower. Until recently, bladed blenders
were assumed to yield the closest representation of the actual count of separable microbes, but the study of Lillard
(1988) suggests that even blending does not give a representative count. We can infer that rapid tests for pathogens
based on primary removal techniques that are only as good as blending might not yield the number of positives as
traditional techniques based on preenrichment.

There are much data on enzyme/surfactant treatments of suspensions to improve their membrane filterability (e.g.,
Sharpe et al., 1979; Peterkin and Sharpe, 1980; Peterkin et al., 1982; Cousins et al., 1979; Entis et al., 1982; Pettipher
and Rodrigues, 1982; Rodrigues and Kroll, 1985; Pettipher, 1989). What effect digestion would have on CFU/g if
applied to food samples before stomaching is unknown. Are there reagents or treatments that would (by releasing
firmly attached microbes) yield a count much more representative of the total? Obvious problems include how to apply
potentially expensive reagents to foods with minimum waste. One solution for intact raw materials such as skin, meat
cuts, raw vegetables, etc., would be to mix the enzyme into a viscous solution or "sauce"; many samples could simply
be coated by being dipped into the sauce and left for the required period before stomaching. However, data obtained for
the Rotorinser (see below) suggest that enzyme pretreatment of carcass surfaces did not greatly increase the yield of
total aerobic bacteria.

Swabs, Stomachings, Blendings, and a Mass Action Effect

While not wishing to suggest that bacteria "communicate" with one another while they are being removed from foods,
one has to wonder whether some sort of mass action effect determines detachment of attached cells into their
environment. Data by Price (1938), Pohle and Stuart (1940), Cade (1952), Ulrich (1961), Leistner and Szentkuti
(1970), Sheena and Stiles (1983), Lillard (1988), etc., and my own (unpublished) observations indicate that on rinsing,
stomaching, or blending, bacteria reach a limiting concentration in suspension, which decreases only slowly on
repeating the procedure scores of times. Adding up all of the counts, one achieves a value

 Page 33

much higher than is suggested by a single blending. Thus, novel techniques are needed to pull, say, 90% of the
organisms off a carcass.

Poor Performance by Swabs

Swabs are nondestructive and yield minimal debris in the removed microbes, but efficiencies of microbe removal are
very far below the level of efficiency we need in a sampling technique. No swab techniques, except perhaps for the
highly standardized "Danish swab method" (Olgaard, 1977) or the "total object swab" (Mossel and Buchli, 1964), yield
even reproducible fractions of the surface count provided by more rigorous methods. In the most detailed and
controlled study on swabs (Ingram and Roberts, 1976) a "wet and dry'' swab method for carcasses and carcass meats
gave total counts from the two swabs ranging from 124% (average 10%) for fresh beef carcass, 2752% (average 37%)
for fresh mutton, 1367% (average 44%) for fresh pork, and 2589% (average 39%) for chilled pork belly, compared
with counts from the excised, blended surface. Bearing in mind the mass action concept, low and variable counts from
swabs are explainable by our inability to treat surfaces with sufficient diluent, or even to distribute and dilute the
organisms that have been removed throughout the volume of the diluent on the swab.

The Rotorinser uses this starting concept and provides representative carcass samples by bringing a relatively large
diluent volume, plus efficient distribution of the removed microbes in it, to improve a "swab." It holds 10 mL or more
of diluent in an open-cell urethane foam cylinder, which scrubs the test surface, and compressions and rotations of the
cylinder against it pump liquid repeatedly through the foam to bring microbes from the test surface into equilibrium (!)
with the diluent. Because the foam cylinder retains diluent when the process is stopped, the Rotorinser can be used on
surfaces at varying angles. In trials on pork skin and beef carcass surfaces, the Rotorinser actually proved more
efficient by removing more microbes than was excision followed by stomaching (Sharpe et al., 1996). It will be
available commercially in the near future.

Spray Methods

Sprays also are relatively nondestructive and yield low debris levels. Clark (1963) developed a spray gun which uses
air at a pressure of 35207 kPa (530 psi) to spray diluent over the surface of samples, with washings being collected as
they ran off. Bacterial yields as high as or higher than a blender

 Page 34

were reported. Other workers claimed various successes with similar techniques (Baumgart and Kussman, 1975; Hess
and Lott, 1970; Leistner et al., 1977; Reuter et al., 1979; Thran, 1979).

Ultrasound
There are few studies (Puleo et al., 1967a,b; Pike et al., 1972; Masurovsky and Jordan, 1960; Huhtanen, 1968) on
insonation for removal of microbes, with only one being very relevant. Sharpe and Kilsby (1970) compared the
efficiencies of an ultrasonic tank, vortex stirring and conventional blending, for removal of bacteria from a variety of
foods. Insonation gave good recovery for peas and beans, was less effective for intact meat pieces that could clump
together to hide surfaces from the energy source, and worse still with comminuted meat. However, it did yield bacterial
suspensions with a very low debris content. The dispersive (even chemical) effects of insonation occur only at intensity
levels high enough to cause cavitation, and cavitation quickly is lethal to bacteria. The sonication conditions must be a
compromise between effectiveness and lethality.

Electrical Detachment

Tanikawa et al. (1966, 1967) claimed to be able to depurate clams and other seafood by electrophoresis of bacteria
away from the clams, patenting the process, but there seems to have been little mention of the technique since. For low-
level contamination complete elimination of E. coli from clam or oysters was reported, while for higher levels (105/g)
9899.9% removal was seen. It is not clear from the paper whether the observed count reduction was entirely due to
microbial electrophoresis. One possibility might be the use of antibodies or other molecules with specific affinities,
which carry strongly electropositively or electronegatively charged groups (e.g., polyamine or polysulfonate). Most
biological structures are electronegative; by modifying the surface charge of microbes, one might induce them either to
migrate in the opposite direction of most food debris or in the same direction but fastereither way effecting some
separation.

Sticky Tapes
Any technique removing microbes into a form in which they are more accessible to separation is interesting. Adhesive
tapes have often been used for removing microbial cells (usually with some of the surface), for direct microscopy, or
for plating and incubating in the manner of an agar contact

 Page 35

technique. Imam and Gould (1990) describe the adhesion of amylolytic bacteria to a starchbased film, with the
intriguing observation that killed or inactivated cells did not adhere, suggesting that adhesion required cell viability.
The system was tested with an Arthrobacter spp., but perhaps it would be possible to tailor "adhesive substrates" for
specific microbes.

Gas Bubbles
Household hintstype books often suggest soaking stubborn deposits in Coca-Cola, and there are cleaning fluids on the
market for which the advertising suggests the same effect. The disruptive power of expanding gas bubble nuclei to
dislodge materials from surfaces should not be overlooked. One can observe the effect of the generation of gas bubbles
by, for example, soaking a reaction flask that has an obstinate tar-cake in (NH4)2CO3 solution for a time, then treating it
with dilute acid; some of the effect may be due to the alkali, but it is difficult not to feel that the sudden generation of
gas within the deposit caused a disruption. Presumably, the closer organisms are to the site of nucleation of a bubble,
the greater the force they would experience. Catalase-positive organisms might be a superb subject for a quick study;
they would be right at the spotcan they be lifted off with dilute H2O2?

Separation of Cells Once Suspended

After extraction of microbes from the food into suspension, the next step is to bring them into a more handleable state.
No doubt everyone keeps a close eye on the immunomagnetic bead literature, for which publications directly related to
food microbiology have multiplied rapidly (e.g., Johne and Jarp, 1988; Lund et al., 1988; Skjerve et al., 1990; Cudjoe
et al., 1991, Luk and Lindberg, 1991; Skjerve and Olsvik, 1991; Tomoyasu, 1992; Chapman and Siddons, 1996; etc.),
with kits available for enumerating L. mono-cytogenes and enterohemorrhagic E. coli and other pathogens. Apart from
price the main problems with immunomagnetic beads are the small volume that can be sampled because of the limited
range of a magnetic field, and a tendency for less-than-quantitative attachment even with great excesses of bead to
target cell.

Membrane Filtration
This is, of course, a prime separation method, capable of improving direct microscopic methods by several orders of
magnitude, for example for the

 Page 36

Direct Epifluorescent Filter Technique (Cousins et al., 1979; Pettipher and Rodrigues, 1982; Pettipher, 1989; Rodrigues
and Kroll, 1985; and many more). The DEFT has been improved even further (Tortorello and Gendel, 1993; Tortorello
and Stewart, 1994) by direct detection of E. coli O157 in juice and meat at levels down to 10 cells/g. Filtration of foods
can be a problem and much work has been done on this (e.g., by the DEFT workers, Sharpe et al., 1979; Peterkin and
Sharpe, 1980; Peterkin et al., 1982; Entis et al., 1982). Membrane filtration of whole chicken rinses after 30 min of
trypsin digestion has been reported (Yamaguchi et al., 1992), although counts of Salmonella were low compared with
an MPN procedure. The filterability of chicken rinses is very variable and enzyme digestion somewhat tedious and
hard on Salmonella cells. My own lab has been studying improvement of whole carcass rinses by other means, with
some success (unpublished).

Centrifugation
Despite being a tedious, messy step, centrifugation has a role in pathogen separation. Stannard and Wood (1983)
reported successful use of centrifugation at 2000 × g for 10 sec as the first step in separating meat particles from
bacteria, prior to estimating biomass by ATP measurement. Virtually all meat solids sedimented during this time,
whereas the number of bacteria did not decrease. Basel et al. (1983) described a density gradient method using a
colloidal silica gradient material. After 15 min the clear supernatant from food suspensions had the same count as the
Uncentrifuged suspension. Linhardt (1987) used the automated density gradient centrifugation apparatus of the
Bactoscan instrument to concentrate food-related microorganisms from liquid samples, with recovery rates of 78134%
(cf plate count), but did not describe actual food applications. Sharp et al. (1991) found recoveries of Lactobacillus
plantarum of 109% versus 9% for stomaching, by applying differential centrifugation to artificially inoculated grass
silage. Sharma et al. (1993) reported a sedimentation field-flow fractionation applied to separating pure bacterial
cultures and estimating their biomass. Cells were injected into an open, unpacked channel, first sedimented by a low
(510 RCF) centrifugal field, then fractionated by the parabolic fluid-velocity field as diluent passes through the
chamber. This suggests that if bacteria from a food sample are truly planktonic, there is the possibility of separating
them from food debris.

Our problem is that foods producing fluffy, fibrous debris may trap bacteria as they settle. In our experiments,
supernatants of suspensions of Citrobacter freundii, Bacillus subtilis, or Staphylococcus aureus showed little or no

 Page 37

reduction in count at RCFs of 32 × g, recoveries better than 80% for minimum centrifugation times (i.e., time to
required speed + coasting time) at 130 × g, but erratic recoveries after 35 min at that speed. In further experiments
where stomached, naturally contaminated food suspensions were centrifuged at 32 or 130 × g for the minimum
possible times and 290 × g for 2.5 min, considerable losses of count occurred. The losses were greater for yeast and
fungal cultures. This would seem to set limits on the centrifuge speeds that can be used without the assistance of
density gradients.

Ion-Exchange and Other Columns

It is difficult to generate enthusiasm about ion-exchange columns/beads or about lectin-based columns/beads. Some
success was reported using lectins in separating bacteria from food suspensions (Patchett et al., 1991; Payne et al.,
1992). One might be rather more optimistic about more specific affinity substrates such as those coated with antibodies
or other specific attractants (e.g., the polymyxin capture method for Salmonella antigen) (Blais and Yamazaki, 1991).

Macrodialysis Experiments

We did many experiments sealing food samples (110 g) in small pouches of a polyester cloth of supposed 1 mm mesh
size (Cadisch and Sons, Finchley, London, UK) and stomaching these on the assumption that only 1-mm particles, free
cells, or cells attached to nothing larger than a 1-mm debris particle escape through the pouch walls. The process is not
filtration, since no net flow of liquid occurs, but liquid and bacteria oscillate into and out of the pouch, a process rather
like dialysishence the name. Since the liquid volume outside the pouch is probably 100 times the volume inside it, at
equilibrium most of the bacteria capable of passing through the pouch wall are found outside the pouch. The technique
could have had many uses in separation, but it turned out that the mesh had much larger openings (about 10 mm) than
claimed by the supplier. The technique is still capable of producing relatively clean suspensions prior to further
separation steps.

Standing Wave Ultrasound

Several groups have described the use of standing ultrasound waves to move suspended particles around (e.g., Coakley
et al., 1989; Grundy et al., 1989; Whitworth et al., 1991). Particles concentrate at nodes, which can be moved by
varying the frequency. The idea of detaching cells from the test surface

 Page 38

and then moving them off to a place of collection by the same ultrasonic forces is almost too tempting. Particle size,
concentration, energy levels, and other factors affecting efficiency were described by Miles et al. (1995). I have yet to
see a practical application of the technique.

Dielectrophoresis
The polarization and subsequent uneven forces acting on slightly conducting particles suspended in liquid in a
nonuniform electric field (e.g., between plate and pin electrodes) cause them to migrate to or from the pin electrode.
This occurs regardless of whether the applied field is d.c. or a.c. If the electrical conductivity of the particle is greater
than that of the medium, the particle moves towards the pin electrode (positive dielectrophoresis). If its conductivity is
less than that of the surrounding medium, the particle moves away from the pin electrode (negative dielectrophoresis).
Voltage gradients of hundreds to thousands of volts/cm may be needed to levitate or move cells, and to prevent
electrolysis effects high-frequency a.c. fields (100 Hz to 100 MHz) may be used. Electrode assemblies are often not
much larger than the cells they are used with and may be fabricated on silicon semiconductor chips. With complex
electrodes (e.g., arrays of four) applied fields can also rotate cells. The interaction of particles modifies applied fields,
allowing a degree of analysis of the situation (Huang et al., 1992; Fuhr et al., 1994), for example, the death or
transformation of mammalian cells (Pethig, 1991). The UK MAFF dielectrophoresis research group has just filed
patent applications, so perhaps there is considerable potential in this technique.

Biphasic Partitioning

Bacteria, viruses, and other bodies partition themselves between the solid phases of aqueous biphasic systems (e.g., of
polysaccharide and gelatin mixtures), permitting some degree of separation (Magnusson and Stendahl, 1985;
Mattiason, 1983; Betts, 1993). Salmonella and E. coli have been separated as have rough/smooth mutants of
Salmonella typhimurium (Stendahl et al., 1973a,b), but I have not seen practical applications to foods.

Foam Flotation
Many ores, particularly sulfide ores, are concentrated from the worthless gangue in which they occur by foam flotation.
Ores are finely crushed and dispersed in aqueous solutions of surfactants, which are chosen to adsorb

 Page 39

preferentially to the ore. When the slurry is bubbled with air, coated ore particles stick to the bubbles and are carried to
the surface. A closely allied effect is the concentration of biological molecules at the air-water interfaces of aerosols
and foams. There is considerable literature on the concentration of biological molecules in sea surf, splashes from
raindrops, sewage sprays, etc., in which concentration gains up to 2000-fold are claimed (e.g., Woodcock, 1955; Smith
and Hugh-Jones, 1969; Blanchard and Syzdek, 1970; Baylor et al., 1977). The phenomenon has actually been used to
separate microbes from water (Hejkal et al., 1980) and microstructures from the Dutch Elm fungus organism
Ceratocystis ulmi, which could not be separated by other means (Richards and Takai, 1973). Purification of
wastewaters by foaming with quaternary ammonium compunds is now common practice. In my own lab, simple
hydrophobic particles such 3-mm polystyrene beads were completely and instantly separated by foaming. Bacteria,
being more hydrophilic, are more problematic, but we have obtained concentration factors (foam to bulk) of viable E.
coli of up to 168 times and sufficient experience to suggest pursuing the approach further. Unfortunately, the best
reagents appear to be the quaternaries used for water purification, their disadvantage being that they are also lethal to
food pathogensfor water purification this is an advantage.

References

Basel, R.M., Richter, E.R., and Banwart, G.J. 1983. Monitoring microbial numbers in food by density centrifugation.
Appl. Environ. Microbiol. 45: 11561159.

Baumgart, J. 1980. Möglichkeiten der Schnellbestimmung von Mikro-organismen. Fleischwirtschaft 7: 13191323.

Baumgart, J. 1977. Zur mikrobiologischen Stabilität von Feinkostsalaten. Alimenta 16: 197200.

Baumgart, J. 1973. Der Stomacher ein neues Zerkleinerungsgerät zur Herstellung von Lebensmittelsuspensionen für
die Keimzahlbestimmung. Fleischwirtschaft 53: 16001601.

Baumgart, J. and Kussmann, H. 1975. Eine Sprühmethode zur Ermittlung des Oberflächenkeimgehaltes bei
Schlachttieren. Fleischwirtschaft 55: 113114.
Baylor, E.R., Baylor, M.B., Blanchard, D.C., Syzdek, L.D., and Appel, C. 1977. Virus transfer from surf to wind.
Science 198: 575580.

Betts, R.P. 1993. The separation and rapid detection of microorganisms. In: Rapid Methods and Automation in
Microbiology and Automation, R.C. Spencer, E.P. Wright, and S.W.B. Newsom (Ed.), p. 107119. Athenaeum

 Page 40

Press, UK. (Proceedings of the Seventh International Congress on Rapid Methods and Automation in Microbiology
and Automation, London, UK, Sept. 1215, 1993.)

Blais, B.W. and Yamazaki, 1991. Application of polymyxin-coated polyester cloth to the semiquantitation of
Salmonella in processed foods. Int. J. Food Microbiol. 14: 4350.

Blanchard, D.C. and Syzdek, L. 1970. Mechanism for the water-to-air transfer and concentration of bacteria. Science
170: 626628.

Cade, A.R. 1952. An in-vivo method for determining the degerming efficiency of soaps containing hexachlorophene.
American Society for Testing and Materials, Tech. Publication No. 115, p. 3339.

Chapman, P.A. and Siddons, 1996. Evaluation of a commercial enzyme immunoassay (EHEC-TEK) for detecting
Escherichia coli O157 in beef and beef products. Food Microbiol. 13: 175182.

Clark, D.S. 1963. Method of estimating the bacterial population on surfaces. Can. J. Microbiol. 11: 407413.

Coakley, W.T., Bardsley, D.W., and Grundy, M.A. 1989. Cell manipulation in ultrasonic standing wave fields. J. Chem.
Tech. Biotechnol. 44: 4362.

Cousins, C.M., Pettipher, G.L., McKinnon, C.H., and Mansell, R. 1979. A rapid method for counting bacteria in raw
milk. Dairy Ind. Int. 44: 2739.

Cudjoe, K.S., Thorsen, L.I., Sorensen, T., Reseland, J., Olsvik, O., and Granum, P.E. 1991. Detection of Clostridium
perfringens type A enterotoxin in faecal and food samples using immunomagnetic separation (IMS)-ELISA. Int. J.
Food Microbiol. 12: 313322.

Entis, P., Brodsky, M.H., and Sharpe, A.N. 1982. Effect of pre-filtration and enzyme treatment on membrane filtration
of foods. J. Food Prot. 45: 811.

Fuhr, G., Glasser, H., Muler, T., and Schnelle, T. 1994. Cell manipulation and cultivation under a.c. electric field
influence in highly conductive culture media. Biochim. Biophys. Acta 1201: 353360.

Grundy, M.A., Coakley, W.T., and Clarke, D.J. 1989. Rapid detection of hepatitis B virus using a haemagglutination
assay in an ultrasonic standing wave field. J. Clin. Lab. Immunol. 30: 9396.
Hejkal, T.W., LaRock, P.A., and Winchester, J.W. 1980. Water-to-air fractionation of bacteria. Appl. Environ Microbiol
39: 335338.

Hess, E. and Lott, G. 1970. Kontamination des Fleisches während und nach der Schlachtung. Fleischwirtschaft 50:
4750.

Huang, Y., Holzel, R., Pethig, R., and Wang, X.B. 1992. Differences in the AC electrodynamics of viable and non-
viable yeast cells determined through combined dielectrophoresis and electrorotation studies. Phys. Med. Biol. 37:
14991517.

 Page 41

Hutanen, C.N. 1968. Effect of low-frequency ultrasound and elevated temperatures on isolation of bacteria from raw
milk. Appl. Microbiol. 16: 470475.

Imam, S.H. and Gould, J.M. 1990. Adhesion of an amylolytic Arthrobacter sp. to starch-containing plastic films. Appl.
Environ. Microbiol. 56: 872876.

Ingram, M. and Roberts, T.A. 1976. The microbiology of the red meat carcass and the slaughterhouse. Roy. Soc.
Health J. 96: 270276.

Johne, B. and Jarp, J. 1988. A rapid assay for protein-A in Staphylococcus aureus strains, using immunomagnetic
monosized polymer particles. Acta Pathol. Scand. 96: 17.

Leistner, L. and Szentkuti, L. 1970. Zwei Methoden zur bakteriologischen Untersuchung von Schlachtgeflügel.
Fleischwirtschaft 50: 81.

Leistner, L., Bem, Z., and Dresel, J. 1977. Schnellmethoden zur Keim-zahlbestimmung bei Fleisch. Mitteilungsblatt
der Bundesanstalt für Fleischforschung, Kulmbach, No. 56(1): 6

Lillard, H. S. 1988. Comparison of sampling methods and implications for bacterial decontamination of poultry
carcasses by rinsing. J. Food Prot. 51: 405408.
Linhardt, C.H. 1987. Density gradient centrifugation as a means to concentrate food-related microorganisms from
liquid samples. J. Appl. Bacteriol. 63(6): XXIII (conference paper).

Luk, J.M.C. and Lindberg, A.A. 1991. Rapid and sensitige detection of Salmonella (O:6,7) by immunomagnetic
monoclonal antibody-based assays. J. Immunol. Methods 137: 18.

Lund, A., Hellemann, A.L., and Vartdal, F. 1988. Rapid isolation of K88+ Escherichia coli by using immunomagnetic
particles. J. Clin. Microbiol. 26: 25722575.

Magnusson, K.-E. and Stendahl, O. 1985. Partitioning of bacteria, virus, and phage. In: Partitioning in Aqueous Two-
Phase Systems, H. Walter, D.E. Brooks, and D. Fisher (Ed.), p. 415452. Academic Press, New York.

Masurovsky, E.B. and Jordan, W.K. 1960. Studies on the removal of Staphylococrus aureus from milk-contact surfaces
by ultrasonic cleaning methods. J. Dairy Sci. 11: 15451559.
Mattiasaon, B. 1983. Applications of aqueous two-phase systems in biotechnology. Trends Biotechnol. 1: 1620.

Miles, C.A., Morley, M.J., Hudson, W.R., and Mackey, B.M. 1995. Principles of separating microorganisms from
suspensions using ultrasound. J. Appl. Bacteriol. 78: 4754.

Mossel, D.A.A. and Buchli, K. 1964. The Total Object Swab ("TOS") technique. Lab. Pract. 13: 11841187.

 Page 42

Olgaard, K. 1977. Determination of relative bacterial levels on carcasses and meatsa new quick method. J. Appl. Bact.
42: 321329.

Patchett, R.A., Kelly, A.F., and Kroll, R.G. 1991. The adsorption of bacteria to immobilized lectins. J. Appl. Bacteriol.
71: 277284.

Payne, M.J., Campbell, S., Patchett, R.A., and Kroll, R.G. 1992. The use of immobilized lectins in the separation of
Staphylococcus aureus, Escherichia coli, Listeria and Salmonella spp. from pure cultures and foods. J. Appl. Bacteriol.
73: 4152.
Payne, M.J. and Kroll, R.G. 1991. Methods for the separation and concentration of bacteria from foods. Trends Food
Sci. Technol. (Dec.): 315319.

Peterkin, P.I. and Sharpe, A.N. 1980. Membrane filtration of dairy products for microbiological analysis. Appl.
Environ. Microbiol. 39: 11381143.
Peterkin, P.I., Sharpe, A.N., and Warburton, D.W. 1982. Inexpensive treatment of frozen dairy products for membrane
filtration. Appl. Environ. Microbiol. 43: 486487.

Pethig, R. 1991. Dielectric-based biosensors. Biochem. Soc. Trans. 19: 2125.

Pettipher, G.L. 1989. The Direct Epifluorescence Filter Technique. In: Rapid Methods in Food Microbiology, M.R.
Adams and C.F.A. Hope (Ed.), p. 1956. Elsevier, New York.

Pettipher, G.L. and Rodrigues, U.M. 1982. Rapid enumeration of microorganisms in foods by the Direct Epifluorescent
Filter Technique. Appl. Environ. Microbiol. 44: 809813.

Pike, E.B., Carrington, E.G., and Ashburner, P.A. 1972. An evaluation of procedures for enumerating bacteria in
activated sludge. J. Appl. Bact. 35: 309321.

Polhe, W.D. and Stuart, L.S. 1940. The germicidal action of cleaning agentsa study of a modification of Price's
procedure. J. Int. Dis. 67: 275281.

Price, P.B. 1938. The bacteriology of normal skin; a new quantitative test applied to a study of the bacterial flora and
the disinfectant action of mechanical cleaning. J. Inf. Dis. 63: 301318.
Puleo, J.R., Favero, M.S., and Petersen. 1967a. Use of ultrasonic energy in assessing microbial contamination on
surfaces. Appl. Microbiol. 15: 13451351.

Puleo, J.R., Petersen, M.S., and Tritz, G.J. 1967b. Fesibility of using ultrasonics for removing viable microorganisms
from surfaces. Contam. Control 6: 5867.

Reuter, G., Sasse, D., and Sibomana, G. 1979. Entwicklung und Prüfung eines Abspulgerätes zur Erfassung des
Oberflächenkeimgehaltes an Schlachttierkörpern. Arch. Lebensmitt. 30: 126129.

 Page 43

Richards, W.C. and Takai, S. 1973. Novel technique for isolating microstructures present in shake cultures of the
fungus Ceratocystis ulmi. Appl. Microbiol. 26: 443444.

Rodrigues, U.M. and Kroll, R.G. 1985. Increased selectivity, sensitivity and rapidity in the Direct Epifluorescent Filter
Technique (DEFT). J. Appl. Bacteriol. 59: 493499.

Sharma, R.V., Edwards, R.T., and Beckett, R. 1993. Physical characterization and quantification of bacteria by
sedimentation field-flow fractionation. Appl. Environ. Microbiol. 59: 18641875.

Sharp, R., Gilbert, H.J., and O'Donnell, A.G. 1991. An improved method for the extraction and recovery of
microorganisms from grass silages. Lett. Appl. Microbiol. 13: 227229.

Sharpe, A.N. 1991. On multiplexed separations in quick detection of micro-organisms in foods. Food Microbiol. 8:
167170.

Sharpe, A.N. and Harshman, G.C. 1976. Recovery of Clostridium perfringens, Staphylococcus aureus and molds by
the Stomacher: Effect of fat content, surfactant concentration and blending time. Can. Inst. Food Sci. Technol. J. 9:
3034.

Sharpe, A.N. and Jackson, A.K. 1972. Stomaching: A new concept in bacteriological sample preparation. Appl.
Microbiol. 24: 175178.

Sharpe, A.N. and Kilsby, D.C. 1970. Ultrasound and vortex stirring as bacteriological sampling methods for foods. J.
Appl. Bacteriol. 33: 351357.

Sharpe, A.N. and Peterkin. 1988. Membrane Filter Food Microbiology. Innovation in Microbiology Research Studies
Series, Research Studies Press, Letchworth, UK.

Sharpe, A.N., Bin Kingombe, C.I., Watney, P., Parrington, L.J., and Dudas, I. 1996. Efficient non-destructive sampler
for carcasses and other surfaces. J. Food Prot. 59: 18.

Sharpe, A.N., Peterkin, P.I., and Dudas, I. 1979. Membrane filtration of food suspensions. Appl. Environ. Microbiol.
37: 2135.

Sheena, A.Z. and Stiles, M.E. 1983. Comparison of barrier creams and germicides for hand hygiene. J. Food Prot. 46:
943946.
Skjerve, E. and Olsvik, O. 1991. Immunomagnetic separation of Salmonella from foods. Int. J. Food Microbiol. 14:
1118.

Skjerve, E., Rorvik, L.M., and Olsvik, O. 1990. Detection of Listeria monocytogenes in foods by immunomagnetic
separation. Appl. Environ. Microbiol. 56: 34783481.

Smith, L.P. and Hugh-Jones, M.E. 1969. The weather factor in foot and mouth disease epidemics. Nature 223: 712715.

Stannard, C.J. and Wood, J.M. 1983. The rapid estimation of microbial

 Page 44

contamination of raw meat by measurement of adenosine triphosphate (ATP). J. Appl. Bacteriol. 55: 429438.

Stendahl, O., Magnusson, K-E., Cunningham, C., and Edebo, R. 1973a. Characterisation of mutants of Salmonella
typhimurium by counter current distribution in aqueous two-polymer phase systems. Infect. Immun. 7: 573577.

Stendahl, O., Tagesson, C., and Edebo, C. 1973b. Partition of Salmonella typhimurium in a two-polymer aqueous
phase system in relation to phagocystosis. Infect. Immun. 8: 3641.

Tanikawa, E., Motohiro, T., and Akiba, M. 1967. Removal of bacteria from food raw materials by electrophoresis. II.
Applied study with baby clams and oysters. Food Technol. 21 (3A): 121A125A.
Tanikawa, E., Motohiro, T., and Akiba, M. 1966. Removal of bacteria from food raw materials by electrophoresis. I.
Factors affecting the electrophoretic mobility of certain species of bacteria. J. Food Sci. 31: 595604.

Thran, V. 1979. Mikrobiologische Untersuchung von Oberflächenein Probennahmegerät. Fleischwirtschaft 59: 950953.
Tomoyasu, T. 1992. Development of the immunomagnetic enrickment method selective for Vibrio parahaemolyticus
serotype K and its application to food poisoning study. Appl. Environ. Microbiol. 58: 26792682.
Tortorello, M.L. and Gendel, S.M. 1993. Fluorescent antibodies applied to Direct Epifluorescent Filter Technique for
microscopic enumeration of Escherichia coli O157:H7 in milk and juice. J. Food Prot. 56: 672677.

Tortorello, M.L. and Stewart, D.S. 1994. Antibody-direct epifluorescent filter technique for rapid direct enumeration of
Escherichia coli O157:H7 in beef. Appl. Environ. Microbiol. 60: 35533559.

Ulrich, J.A. 1961. Technics of skin sampling for microbial contaminants. Health Lab. Sci. 1: 133136.

Whitworth, G., Grundy, M.A., and Coakley, W.T. 1991. Transport and harvesting of suspended particles using
modulated ultrasound. Ultrasonics 29: 439444.

Woodcock, A.H. 1955. Bursting bubbles and air pollution. Sewage In. Wastes 27: 11891192.

Yamaguchi, W., Kopek, J.M., and Waldrup, A.L. 1994. Comparison of a membrane filtration procedure and a MPN
procedure for enumerating Salmonella in whole carcass rinse samples. J. Rapid Meth. Autom. Microbiol. 2: 287297.

 Page 45

3
A New Look at an Old Technique: Application of Microscopy to Food Microbiological Analysis
Mary Lou Tortorello
National Center for Food Safety and Technology
Food and Drug Administration
Summit-Argo, Illinois

Introduction

The microscope is the most widely recognized definitive tool of microbiology and a symbol of scientific investigation
in general. The discovery of the microbial world by Anton van Leeuwenhoek in 1674 was made by using a simple
hand-held microscope. Indeed, microorganisms are defined by their inability to be seen without a microscope's power.
Yet few practicing microbiologists routinely use the microscope, and most food microbiological analysis is performed
without the aid of this fundamental instrument. The reasons for its underutilization are clear. Traditionally, microscopy
has been regarded as an analytical method having poor sensitivity, poor specificity, an inability to discriminate between
live and dead cells, and involving too much tedium and technical skill. Furthermore, early traumatic experiences with
poor-quality microscopes in high school biology classes may also contribute to a general distaste for the microscope as
an analytical tool. Notwithstanding the importance of better educational experiences in acceptance of the

 Page 46

instrument, a variety of modern innovations may be coupled with the microscope to minimize its shortcomings and to
provide greater application as an extremely rapid method in food safety microbiology.

General Considerations for Rapid Microbiological Analysis of Foods

The microbiological analysis of foods is laborious and time-consuming because of the procedures necessary to
overcome two problems: (1) the indigenous microbial population usually vastly outnumbers the pathogenic species,
which makes it difficult to find and identify the pathogens, and (2) food is a complex matrix, which interferes with the
performance of instruments that might otherwise provide rapid analytical capability.

Most standard methods of microbiological analysis employ measures to circumvent both problems. Selective
enrichment allows the small number of pathogenic cells present in the food to increase in number while keeping in
check the growth of other microbial species. The enrichment culture is then plated onto a differential medium to obtain
isolated colonies, and in doing so, the microbial cells are effectively purified from the food matrix. The isolated
colonies that exhibit morphologies typical of the pathogen are then cultured and subjected to various biochemical tests
for proof of species identification. The entire scheme is costly and time-consuming, primarily because of the several
steps of microbial growth. It is obvious, then, that a rapid microbiological method should aim to decrease or eliminate
these growth steps, while providing the sensitivity and specificity necessary to detect and discriminate low numbers of
cells in a complex food matrix. Three objectives must be accomplished in developing a rapid microbiological assay:
separation of microbial cells from the food, discrimination of target microorganisms from the background population,
and rapid detection of the target microorganisms.

Microscopy: The Oldest Rapid Method

Microscopy is a very rapid method of detecting the presence of microbial cells. It is difficult to imagine a more rapid
assay than simply ''looking" to see if microbes are present. Microscopy is a "direct" method of detection and, therefore,
allows examination of microbial populations in situ in food systems (Stringer et al., 1995). With the development of
scanning confocal laser microscopy, in situ analysis of complex, three-dimensional microenvironments, such as food
system biofilms, is possible (Lawrence et al., 1991).

 Page 47

Compared with standard cultural methods, however, microscopy has certain disadvantages that must be overcome.

Poor Sensitivity

Sensitivity may be defined as the lowest level of microbial cells detectable by a method and is often referred to as the
"detection limit." A method is said to have high sensitivity if it can detect a very low number of cells. Typically the
sensitivity of a method is reported as the lowest number of cells detectable per unit weight or volume of food.
Conventional culture methods are extremely sensitive and typically have the ability to detect at the single cell level,
because of enrichment procedures that allow a single cell to undergo cell division and increase in number to detectable
levels. Microscopy is often regarded as a rather insensitive detection method. For example, the Breed Count, a direct
microscopic method for counting total bacteria in milk, generally provides a detection limit no better than
approximately 106 cells per mL (Cousins et al., 1979). For pathogen detection, this level of sensitivity clearly won't
work.
Sensitivity may be improved, however, by concentrating the cells prior to detection. The direct epifluorescent filter
technique (DEFT), for example, uses membrane filtration of the food to concentrate and collect microbial cells on the
filter's surface (Pettipher, 1986). A 1000-fold increase in sensitivity over other direct microscopic methods may be
achieved by using fluorescent dyes to stain the cells and epifluorescence illumination for their visualization (Pettipher
et al., 1980). Filter concentration of the sample provides an additional benefit, which affects sensitivity; i.e., many
substances present in the food and irrelevant to the microbiological assay are eliminated by their passage through the
filter. These substances often interfere with the outcome of analytical techniques, such as the enzyme-linked
immunosorbent assay (ELISA) and the polymerase chain reaction (PCR), and are an important reason for the
requirement for enrichment (and sometimes isolation) steps prior to the use of these assays. Filtration, then, allows a
degree of concentration and clean-up in a single, simple step.
Poor Specificity

The ability to differentiate particular types of microbial cells from all others is the quality known as specificity. In
microscopy, specificity is determined by the types of chemical staining reagents used for visualization of the cell. The
DEFT uses acridine orange, a stain that binds to nucleic acids (Hobbie et al., 1977; Kepner and Pratt, 1994) and,
therefore, provides an increased level of differentiation over that provided by general stains, which, for example,

 Page 48

may not distinguish between cells and noncellular matter. Extremely specific differentiation, however, may be achieved
through the use of molecular probes, such as antibodies and oligonucleotides, which may be designed to distinguish
microbial cells at the level of genus, species, or strain. The antibody-direct epifluorescent filter technique (Ab-DEFT),
for example, uses fluorochrome-labeled antibody directed against the cell wall antigen O157, and provides a very rapid
and specific detection method for the foodborne pathogen Escherichia coli O157:H7 (Tortorello and Stewart, 1994).

Antibodies and oligonucleotides have advantages and disadvantages in their use as molecular probes. Antibodies are
protein molecules which bind to specific cellular structures or products, generally called antigens. Antigens tend to be
the proteins, polysaccharides, and their derivatives, which make up the architecture of the cell. The identification of an
antigen that is unique to a particular microbial group (genus, species, strain, etc.) is the first step in the design of an
antibody-based molecular probe. Antigens are often present in the cell in amounts sufficient for detection by
microscopy: however, they may not always be present in the cell. Expression of the antigen by the cell may be
governed, for example, by environmental conditions. In fact, antibody probes are used to study regulation of expression
and to track antigen expression microscopically (Harry et al., 1995). Antigen-antibody binding does not involve
reactions that destroy metabolic processes, and therefore cell viability determinations may be made in conjunction with
group-specific identification.

Oligonucleotides are short nucleotide polymers, which bind to specific segments of nucleic acids by complementary
base pairing. For use as molecular probes, the oligonucleotides are usually approximately 20 nucleotide bases in
length. As with antigens, identification of a sequence of DNA or RNA that distinguishes a specific microbial group is
required for oligo-nucleotide probe design. Unlike antigens, however, the genetic sequences are usually present at low
levels in the cell, perhaps only a single copy, and it is impossible to detect a single molecule by light microscopy.
Amplification of the gene or of the signal provided by the probe is a strategy that may be incorporated to increase
detectability (Hodson et al., 1995; Matsuhisa et al., 1994). There are, however, relatively abundant amounts of RNA
present in the ribosomes of the cell, and group-specific detection of microbial cells based on recognition of unique
ribosomal RNA regions is possible using complementary oligonucleotide probes without any amplification steps
(DeLong et al., 1989). Oligonucleotide binding usually involves reactions that result in destruction of the cell's
viability; therefore viability determinations are more limited, but not impossible, when oligonucleotides are used as
probes.

 Page 49

Inability to Determine Cell Viability

One of the most common misconceptions about the microscope is the perceived inability to distinguish between viable
and nonviable cells, but several methods exist to do so. Viability determination by microscopy may be done through
redox dyes, such as tetrazolium compounds, which accept electrons from respiring cells and form colored precipitates,
which accumulate in the cell (Bovill et al., 1994). The ability of viable cells possessing intact membranes to exclude
certain dyes is the basis of another group of viability stains (Haugland, 1992). Cells that are capable of undergoing cell
division increase in length in the presence of certain cell division inhibitors but do not complete the formation of new
cell walls; thus elongated cells are indicative of viable cells (Frank et al., 1992). Dyes that are substrates for certain
cellular enzymes may be used to indicate enzyme activity in viable cells. Incubation of viable cells in the presence of
nutrients gives rise to the formation of microcolonies, which may be detected microscopically (Rodrigues and Kroll,
1990).

Labor and Technical Difficulty

The development of computer-controlled image analysis systems for microscopes diminishes the argument that
microscopy is a tedious chore. A CCD (charge-coupled device) camera captures the microscopic image and sends a
digitized version to a computer. Image analysis software allows automatic counting or sizing of digitized images of
cells (Sieracki et al., 1985; Viles and Sieracki, 1992; Wirtanen and Matilla-Sandholm, 1993). Capture of images in
color or through the use of appropriate filters provides the ability to distinguish between cells that may be differentially
stained, e.g., viable versus nonviable cells or pathogenic versus nonpathogenic cells. The technology also exists for
computer-controlled scanning of slides by automatic movements of the microscope stage (X-Y axis) and focusing knob
(Z axis). All of the steps of the DEFT have been completely automated in a commercial instrument, from sample
application and membrane filtration, through staining, microscope scanning, and image analysis-based enumeration of
cells (Pettipher et al., 1992).

Analysis of Beef for E. Coli O157:H7 by Microscopy

E. coli O157:H7 has caused foodborne illnesses that have been traced primarily to the consumption of undercooked
ground beef, and its detection is an important food safety concern. The following steps summarize the stan-

 Page 50

dard method for detection of the microorganism in beef, as recommended by the U.S. Department of Agriculture
(USDA) (Okrend and Rose, 1989): 24-hr enrichment of the beef in modified EC broth supplemented with novobiocin;
plating of the enrichment culture onto MacConkey sorbitol-BCIG agar for colony isolation; transfer of suspect colonies
to phenol red sorbitol agar and eosin methylene blue agar; biochemical and serological testing of the suspect colonies.
The standard method requires 48 hr for presumptive results and a minimum of 4 days for confirmation of
presumptively positive cultures. A number of rapid methods based on ELISA technology, which provide presumptive
results in approximately 24 hr, have been commercialized (Mermelstein, 1996). These "screening" tests are confirmed
in the traditional manner (4 days minimum) if presumptively positive results are obtained.

The Ab-DEFT is an example of how microscopy can be utilized, despite the traditional misconceptions, as a rapid
screening test for detection of food pathogens. The specific detection of E. coli O157:H7 in beef may be achieved in
less than 1 hr by using fluorescein-labeled polyclonal antibody with specificity for the O157 antigen. Beef is
homogenized in buffer in a Stomacher blender, followed by treatment with trypsin and Triton X-100 for 10 min at
50°C to facilitate membrane filtration. The homogenate is passed through a 5-mm pore nylon prefilter, followed by
filtration through a 0.4-mm polycarbonate filter. The polycarbonate filter is stained with the fluorescent antibody for
10 min, and the filter is examined using epifluorescence microscopy. It is possible to achieve a detection limit of
approximately 16 cells/g of beef by using this procedure (Tortorello and Stewart, 1994). Coupling of the Ab-DEFT
with a 10-hr enrichment of the beef in nonselective medium increases sensitivity to levels comparable to the standard
USDA method (1 cell/25 g of beef), but increases the total time to obtain presumptive results to approximately 11 hr
(Restaino et al., 1996). Immunomagnetic separation (IMS) may be used to increase the efficiency of isolating the
pathogen from the enrichment culture and, by subsequent plating of the immunomagnetic beads onto MacConkey
sorbitol-BCIG Agar, may be used to obtain confirmation of presumptively positive screening results in approximately
24 hr (Restaino et al., 1996). The Ab-DEFT-IMS screening and confirmation protocols, as recommended for use in a
beef-processing facility, are illustrated in Fig. 3.1.

Microscopy as a Quantitative Technique

Perhaps the most unique and powerful aspect of microscopy is its ability to provide not only rapid detection, but also
rapid quantitation of microbial

 Page 51

Fig. 3.1
Application of the antibody-direct epifluorescent filter technique
(Ab-DEFT) and immunomagnetic separation (IMS) for 11-hr screening
and 35-hr confirmation of E. coli O157:H7 in beef. mBPW, Modified
buffered peptone water; MSA-BCIG, MacConkey-sorbitol agar
supplemented with 5-bromo-4-chloro-3-indolyl-b-D-glucuronic acid.

populations in foods. When used with membrane filtration to concentrate the cells and with identifiers such as
molecular probes, microscopy allows the unparalleled ability to enumerate specific microorganisms very rapidly. In
addition, viability determinations may be performed simultaneously to obtain an estimate of the number of viable cells
of a specific population (Pyle et al., 1995).

Figure 3.2 illustrates how microscopy may be used in a quantitative sense, i.e., for monitoring cell counts of E. coli
O157:H7 in beef in a long-term survival study. During a 15-day period of storage in beef at 4°C, total cell counts were
obtained microscopically by DEFT (acridine orange stain) and E. coli O157:H7 counts were obtained by the Ab-DEFT
(fluorescent antibody stain). For comparison, plating on nonselective and selective media was used to estimate total
and specific E. coli O157:H7 viable plate counts, respectively. During the first 2 days of storage, total microbial counts
in the beef increased approximately 10-fold, and then stabilized for the duration of the experiment. This trend was
shown by both the total microscopic counts obtained by acridine orange stain and total plate count. However, the initial
decrease in survival of E. coli O157:H7 indicated by the specific plate count through day 2 was not mirrored by the
Ab-DEFT count in the

 Page 52

Fig. 3.2
Microbiological analysis of ground beef
inoculated with E. coli O157:H7 and stored
at 4°C for 15 days. Open circles, Total cell
count by microscopy, acridine orange stain;
open triangles, total viable plate count on
nonselective medium; closed circles, E. coli
O157:H7 cell count by microscopy, fluorescent
antibody stain; closed triangles, E. coli O157:H7
viable plate count on selective medium.

same period. Rather, the Ab-DEFT count was relatively unchanged over this 2-day period. After their initial decline,
the specific plate count data appeared to stabilize, but the Ab-DEFT counts appeared to slowly decline until they
converged with the specific plate count at about day 7. After this point and through day 15, the counting results
obtained by the two methods were virtually the same.

The following equation may be used to calculate microbial cell concentration from the number of cells counted in a
microscope field of view:

The Filter Microscope Factor (FMF) is a function of the area of the membrane filter and the area of the microscopic
field of view:

Microscope field areas vary for each microscope objective lens. Furthermore, every lens system is unique, and
microscope field areas must always be

 Page 53

determined individually by using a stage micrometer. For all microscope systems, the higher the magnification of the
objective lens, the smaller the microscope field area.

Typical microscope field areas (p × radius2) for several lenses are indicated below.

Objective lens Area of microscope field (mm2)

40× 0.294

63× 0.124
100× 0.052

The higher the magnification of the objective lens, the larger the filter microscope factor. Corresponding filter
microscope factors, calculated using the area of a 25-mm membrane filter, are indicated below:

Objective lens Filter microscope factor

40× 1670

63× 3959
100× 9440

Quantitation of a microbial cell population by microscopy is illustrated in the following example. Filtration of a 10-mL
sample volume and use of the 100x lens results in an average count of 20 cells per microscope field. With appropriate
substitutions in the equation presented above, the cell concentration is calculated as:

The amount of sample analyzed ultimately influences the sensitivity of the method: the larger the sample size observed,
the greater the chances of detecting lower concentrations of cells. Three factors govern the amount of sample analyzed:
volume of sample filtered, magnification, and number of microscope fields observed.

It is easy to see from the equation how the sample volume influences sensitivity. A sample with a cell concentration an
order of magnitude less than that of the above example, i.e., 1.9 × 103 cells/mL, could be analyzed under the same
conditions (100× lens, filter microscope factor 9440) and with similar results (average count of 20 cells per field) if a
100-mL sample volume were filtered instead of 10 mL. It theoretically would take a 1000-mL volume to obtain similar
results for a sample having 1.9 × 102 cells/mL Filtration of such a large volume is difficult, even with the best of
solubilization conditions. Therefore, other manipulations of the system may be per-

 Page 54

formed to increase the amount of sample analyzed and achieve lower detection limits.

The second factor governing the amount of sample analyzed is the magnification used. It is important to determine the
lowest magnification usable for detecting a particular microbial species in any system. The size of the microscope field
area is directly proportional to the amount of sample analyzed; i.e., the larger the field of view (lower the
magnification), the greater the sample size observed. In the example shown in the equation, if the 63× lens (filter
microscope factor 3959) were used instead of the 100×, and if the average count of 20 cells per field and 10-ml sample
remained the same, the concentration of cells detected would decrease to 7.9 × 103 cells/ mL. If the 40× lens (filter
microscope factor 1670) were used instead of the 100×, the level of detection theoretically would decrease to 3.3 × 103
cells/mL.

The third factor influencing sample size, and therefore sensitivity, is the number of microscope fields observed.
Certainly, the greater the number of microscope fields observed, the larger the sample size analyzed, and therefore the
greater the chances of detecting low numbers of cells. To continue from the previous example, by using the 40× lens
and a 10-mL volume of a sample containing 3.3 × 103 cells/mL, a sufficient number of fields was counted to yield an
average count of 20 cells per field. A cell concentration lower by an order of magnitude, i.e. 3.3 × 102 cells/mL, could
be detected by scanning a larger number of fields to result in an average count of 2 cells per field. If a sufficient
number of fields were counted to yield an average count of 0.2 cells per field, an additional log-fold increase in
sensitivity (33 cells/ mL) theoretically could be achieved.

The strategy of manipulating magnification, number of fields counted, and sample volume was utilized in developing
the Ab-DEFT for detection of E. coli O157:H7 and in demonstrating the feasibility of attaining a sensitivity of
approximately 16 cells/g of beef (Tortorello and Stewart, 1994). Because this level of sensitivity is not sufficient for
analysis of this pathogen, however, the Ab-DEFT may be modified to include a period of enrichment (Restaino et al.,
1996). The modified procedure may be improved, perhaps to allow even shorter enrichment times, by adopting a
similar strategy of manipulation of the above parameters.
For most microbial groups, specific quantitation is often accomplished by a particular adaptation of the most probable
number (MPN) technique. Although the MPN is a time-consuming, cumbersome, and rather imprecise method for
enumeration of specific microbial populations, it is, like conventional techniques for detection, extremely sensitive, due
to its reliance on enrichment culture. Nevertheless, it is ripe for replacement by a more efficient method. The need to
develop better quantitative tests for certain microorganisms, e.g., Listeria monocytogenes, is recognized by many in the

 Page 55

field of food safety microbiology. Quantitative microscopy may provide a rapid, specific alternative to the MPN
technique.

References
Bovill, R.A., Shallcross, J.A., and Mackey, B.M. 1994. Comparison of the fluorescent redox dye 5-cyano-2,3-
ditolyltetrazolium chloride with p-iodonitrotetrazolium violet to detect metabolic activity in heat-stressed Listeria
monocytogenes cells. J. Appl. Bacteriol. 77: 353358.

Cousins, C.M., Pettipher, G.L., McKinnon, C.H., and Mansell, R. 1979. A rapid method for counting bacteria in raw
milk. Dairy Indust. Int. 44: 2739.

DeLong, E.F., Wickham, G.S., and Pace, N.R. 1989. Phylogenetic stains: Ribosomal RNA-based probes for the
identification of single cells. Science 243: 13601363.

Frank, J.F., Gassem, M.A., and Gillett, R.A.N. 1992. A direct viable count method suitable for use with Listeria
monocytogenes. J. Food Prot. 55: 697700.

Harry, E.J., Pogliano, R, and Losick, R. 1995. Use of immunofluorescence to visualize cell-specific gene expression
during sporulation in Bacillus subtilis. J. Bacteriol. 177: 33863393.

Haugland, R.P. 1992. Handbook of Fluorescent Probes and Research Chemicals, 5th ed. Molecular Probes, Inc.,
Eugene, OR.
Hobbie, J.E., Daley, R.J., and Jasper, S. 1977. Use of nucleopore filters for counting bacteria by fluorescence
microscopy. Appl. Environ. Microbiol. 33: 12251228.

Hodson, R.E., Dustman, W.A., Garg, R.P., and Moran, M.A. 1995. In situ PCR for visualization of microscale
distribution of specific genes and gene products in prokaryotic communities. Appl. Environ. Microbiol. 61: 40744082.

Kepner, Jr., R.L. and Pratt, J.R. 1994. Use of fluorochromes for direct enumeration of total bacteria in environmental
samples: past and present. Microbiol. Rev. 58: 603615.

Lawrence, J.R., Korber, D.R., Hoyle, B.D., Costerton, J.W., and Caldwell, D.E. 1991. Optical sectioning of microbial
biofilms. J. Bacteriol. 173: 65586567.

Matsuhisa, A., Saito, Y., Ueyama, H., Aikawa Y., and Ohono, T. 1994. Detection of staphylococci in mouse phagocytic
cells by in situ hybridization using biotinylated DNA probes. Biotechnic Histochem. 69: 3137.

Mermelstein, N.H 1996. Testing for E. coli O157:H7 in meat. Food Technol. 50: 36.

 Page 56

Okrend, A.J.G. and Rose, B.E. 1989. Isolation and identification of Escherichia coli O157:H7 from meat. Revision 3,
Laboratory Communication No. 38. FSIS, Microbiology Division, U.S. Department of Agriculture, Washington, DC.

Pettipher, G.L. 1986. Review: the direct epifluorescent filter technique. J. Food Technol. 21: 535546.

Pettipher, G.L., Mansell, R., McKinnon, C.H., and Cousins, C.M. 1980. Rapid membrane filtration-epifluorescent
microscopy technique for direct enumeration of bacteria in raw milk. Appl. Environ. Microbiol. 39: 423429.

Pettipher, G.L., Watts, Y.B., Langford, S.A., and Kroll, R.G. 1992. Preliminary evaluation of COBRA, an automated
DEFT instrument, for the rapid enumeration of microorganisms in cultures, raw milk, meat and fish. Lett. Appl.
Microbiol. 14: 206209.

Pyle, B.H., Broadaway, S.C., and McFeters, G.A. 1995. A rapid, direct method for enumerating respiring
enterohemorrhagic Escherichia coli O157:H7 in water. Appl. Environ. Microbiol. 61: 26142619.

Restaino, L., Castillo, H.J., Stewart, D., and Tortorello, M.L. 1996. Antibody-direct epifluorescent filter technique and
immunomagnetic separation for 10-h screening and 24-h confirmation of Escherichia coli O157:H7 in beef. J. Food
Prot. 59: 10721075.

Rodrigues, U.M. and Kroll, R.G. 1990. Rapid detection of salmonellas in raw meats using a fluorescent antibody-
microcolony technique. J. Appl. Bacteriol. 68: 213223.

Sieracki, M.E., Johnson, P.W., and Sieburth, J.M. 1985. Detection, enumeration and sizing of planktonic bacteria by
image-analyzed epifluorescence microscopy. Appl. Environ. Microbiol. 49: 799810.

Stringer, S.C., Dodd, C.E.R., Morgan, M.R.A., and Waites, W.M. 1995. Detection of microorganisms in situ in solid
foods. Trends Food Sci. Technol. 6: 370374.

Tortorello, M.L. and Stewart, D.S. 1994. Antibody-direct epifluorescent filter technique for rapid, direct enumeration
of Escherichia coli O157:H7 in beef. Appl. Environ. Microbiol. 60: 35533559.

Viles, C.L. and Sieracki, M.E. 1992. Measurement of marine picoplankton cell size by using a cooled, charge-coupled
device camera with image-analyzed fluorescence microscopy. Appl. Environ. Microbiol. 58: 584592.

Wirtanen, G. and Mattila-Sandholm, T. 1993. Epifluorescence image analysis and cultivation of foodborne biofilm
bacteria grown on stainless steel surfaces. J. Food Prot. 56: 678683.

 Page 57

4
Flow Cytometry in Food Microbiology: Detection of Escherichia coli O157:H7
Richard B. Raybourne
Food and Drug Administration
Laurel, Maryland

Introduction

Since its development in the 1970s as a commercially available technology for cell analysis, flow cytometry (FC) has
held significant potential as a tool for microbiologists. In the intervening 20 years, however, analysis of bacteria has
lagged far behind the advances made in the analysis of eukaryotic cells by immunologists and cell biologists: of over
23,000 Medline references to FC, only 571 also referenced bacteria. There are two major reasons for this discrepancy.
The optical configurations of most commercially available flow cytometers are optimized for analysis of intact cells in
the size range of eukaryotic cells, especially lymphocytes, rather than for analysis of small cells such as bacteria.
Second, there is a lack of well-characterized, specific molecular probes for bacterial analysis; the almost universal use
of FC in immunology is driven by the availability of fluorochrome-labeled probes for specific molecules. Our
laboratory became involved in analysis of bacteria by FC as a result of developing methods to measure uptake and
survival of facultative intracellular bacteria by phagocytic cells (Raybourne and Bun-

 Page 58

ning, 1994). As part of these studies, it was necessary to analyze fluorochrome-labeled bacteria by FC. This led to
firsthand experience with the above problems and an interest in investigating the practicality of FC as a method for
detection and isolation of pathogenic bacteria in an applied setting such as food microbiology.

Because of limited exposure among food microbiologists, it is useful to briefly describe the principles involved in FC.
The operation of a typical commercially available flow cytometer depends upon the movement of particles (cells,
beads, nuclei) in an aqueous stream which is stabilized by being drawn into the center of a surrounding stream in a
laminar flow allowing the particles to move in single file in a fixed position. The particle-containing stream intersects a
focused, high-intensity, stable light beam, usually produced by a laser. The most widely used laser is an argon laser
emitting a 488 nm beam. As the particles pass through the beam, two light scatter signals are emitted: forward light
scatter (FLS), at a low angle in the direction of the beam, and side light scatter (SS), at about 90 degrees to the beam.
FLS is generally proportional to the size of the cell, while SS increases with granularity. In addition, light from
fluorochromes such as those attached to antibodies or bound to DNA will also be emitted if the beam is of the
appropriate excitation wavelength. These light signals are collected by lenses and focused onto photomultiplier tubes
(PMTs) or photodiodes that convert photons to electrical pulses. In order to separate and independently analyze the
signals from side-scattered laser light and multiple fluorochromes that may be produced by the same particle, an optical
path is configured using appropriate dichroic and wavelength-pass filters to direct the desired wavelength to each PMT.
The signals from PMTs are then amplified using logarithmic and linear amplifiers. Linear amplification is suited to
pulses of similar size, while log amplification provides greater sensitivity for weak signals or when signals are of
widely divergent size. In order to distinguish noise and debris from the particles (cells) of interest, a noise
discrimination setting is established. This setting is established for one parameter being measured; if the signal from a
cell is above the discriminator setting, the instrument is triggered to collect all other signals derived from that cell. If it
falls below the discriminator setting, all signals from the cell are discarded. Typically FLS signals are used to trigger
the instrument, since all cells will produce this signal. The signals are then converted from analog or pulse form to
digital channel numbers, based on pulse height. The number of pulses or counts falling into each channel is used to plot
a histogram for the cells being analyzed. Because multiple parameter data are collected and stored for each cell
(listmode data), simultaneous measurements from different parameters can be used to produce two-parameter
histograms. Populations of cells defined on the basis of one- or two-

 Page 59

parameter histograms can be analyzed for a third measured parameter (gating). Listmode data files stored on disk can
be replayed with different gating and different combinations of parameters, making it possible to reanalyze the sample
multiple times. Flow cytometers with cell-sorting capability can be used to physically isolate cells identified on the
basis of any combination of parameters. This is accomplished by breaking the sample stream into electrically charged
droplets, each containing a single cell. The droplets are then deflected by charged plates into an appropriate vessel for
further analysis.

Analysis of E. coli O157:H7 and Non-O157:H7 by FC

Detection by Light Scatter and Fluorescence

We used a conventionally configured Epics Elite (Coulter Corp., Hialeah FL,) flow cytometer with sorting capability to
determine the feasibility of detecting E. coli O157:H7 in mixtures with E. coli strain HB101 in phosphate-buffered
saline (PBS). This bacterial strain was chosen because of the availability of a fluorescent conjugated antibody, specific
for O157:H7, which had been previously used in a direct antibody-based epifluorescence microscopy technique (Ab-
DEFT) for detection of O157:H7 in beef (Tortorello and Stewart, 1994). A series of mixtures of O157:H7
(streptomycin-resistant strain DHS-1) and HB101 were diluted to a concentration of approximately 106 cells/mL in
Dulbecco's PBS and the fluorescein isothiocyanate (FITC)-conjugated rabbit anti-O157:H7 antibody added at a final
dilution of 1:2000. After 30 min, the samples were analyzed by FC. Bacterial cells were found to be at or near the
limits of the FLS detection capability of the instrument. It was necessary to use log amplified FLS signals to achieve
resolution of the bacteria above the noise discrimination level for FLS triggering. Gating on the bacterial population
defined by the FLS-SS histogram demonstrated excellent detection of the bacteria labeled with the FITC-conjugated
anti-O157:H7 antibody. An operator-created region gives the number of cells in the positive population and the
percentage it represents of the total bacterial cell number collected during the time of the sample run.

Enumeration of Fluorescent E. coli O157:H7 by FC

Subsequent experiments were designed to determine the minimum frequency at which O157 could be distinguished
from the HB101 background in mixtures and to determine the degree of correlation between Ab-DEFT and FC in
determining the number of O157:H7 present. For most FC

 Page 60

applications, only the identification and relative proportion of a given population of cells are needed. Therefore, the
flow cytometer used in these experiments does not deliver a precise volume of sample so that the absolute number of
cells/unit volume in any population cannot be directly determined. In order to derive the actual number of bacteria/ml
from the flow cytometer, 10-mM fluorescent beads were added to the bacterial samples. These beads were easily
distinguished from bacteria by fluorescence intensity and by light scatter, allowing us to establish the number of beads
and bacteria detected by the cytometer during a given sample run. The number of beads/mL was then determined by
hemacytometer counts under a fluorescent microscope. Using the number of fluorescent O157:H7 detected during the
sample run, the number/mL of O157:H7 was calculated as a function of the ratio of beads detected during the run
versus the actual number of beads/mL determined microscopically. The number detected was derived from the region
containing the fluorescent bacteria in a sample with a high frequency of O157:H7. The region was kept constant
throughout the experiment. The number of events in this region in samples with no O157:H7 was used to establish the
background of the system. All counts were based on acquisition of at least 20,000 FLS events. The elapsed time
required for this was approximately 12 min for each sample. In two experiments, good correlation was found between
the Ab-DEFT and FC O157:H7 counts and the viable plate count. The Ab-DEFT and FC absolute numbers showed
relatively small deviation from the calculated CFU of O157:H7. The number of O157:H7 detected by FC was slightly
higher than by Ab-DEFT, and both fluorescence methods tended to produce counts that were somewhat higher than the
number of CFU added (Table 4.1). This may be due to the presence of nonviable O157:H7 in the inoculum or the
inherent imprecision of the methods. In samples with no O157:H7 added, Ab-DEFT background counts were
essentially zero, while FC detected fluorescence events in the region set for FITC-positive O157:H7 (Table 4.1).
TABLE 4.1 Inherent Background and Lower Limit of Detection of Fluorescent E. coli O157:H7 by FC and Ab-DEFT
CFU/mL added Ab-DEFT counts/mL FC counts/mL
58,000 170,000 240,000
5,800 110,000 68,000
0 33 33,000

 Page 61

The source of this background is possibly particulate debris that falls in the same FLS range as bacteria and
nonspecifically binds the antibody. In the Ab-DEFT procedure, these would be morphologically distinguished from
bacteria and ignored. Use of an additional parameter such as a DNA-binding dye to distinguish bacteria might alleviate
this problem. The lower limit of detection by FC appears to be on the order of 5 × 104/mL O157:H7 in a background of
5 × 107/mL HB101, or about 0.1% FITC-positive cells. This level of sensitivity is similar to that reported by
McClelland and Pinder (1994a,b) with Salmonella using a custom-designed flow cytometer. Those studies also
involved addition of the fluorescent DNA-binding dye ethidium bromide to the bacterial samples. As a result, all
bacteria exhibit red fluorescence from ethidium bromide, while those labeled with the FITC-antisalmonella antibody
are defined in a two-parameter histogram as double positive. One advantage to this approach is that all bacteria are
positive for one fluorescence parameter, which provides an additional option, other than light scatter, for gating on the
bacteria. In other FC studies, different bacterial strains were found to be distinguishable based on their staining with
another DNA-binding dye, propidium iodide (Van Dilla et al., 1983; Donnelly and Baigent, 1986). One disadvantage of
DNA-binding dyes is toxicity. Because one of our goals was to sort and culture bacteria from the antibody-stained
mixtures, we did not use this method.

Detection of O157:H7 in Beef Enrichment Cultures

Mixtures of two pure bacterial strains provide the optimal scenario for detection by FC. The bacterial population
isolated from a food matrix such as ground beef is much more complex and provides a more realistic assessment of the
FC technique for food microbiology. Direct detection in freshly isolated samples would be impractical given that the
lower limits of FC detection in mixtures reported by us and others (McClelland and Pinder, 1994b) are likely to greatly
exceed the infective dose of many pathogens. Enrichment cultures could be used to bring low numbers of O157:H7 up
to the level detectable by FC. This approach has been used to detect Salmonella in milk and cheese (McClelland and
Pinder, 1994b) and L. monocytogenes in milk (Donnelly and Baigent, 1986). In order to control precisely the level of
O157:H7 present, known numbers of O157:H7 CFU (DHS-1) were added to enrichment cultures, which were prepared
from O157:H7-negative ground beef according to the USDA-recommended procedure. The mixtures were then
processed for the Ab-DEFT procedure (Tortorello and Stewart, 1994). Prior to the final filtration step, an aliquot was
taken for FC analysis as described previously. The FITC-stained O157:H7 were well separated from the negative
population. However, a less bright population was present in

 Page 62

Fig. 4.1
Fluorescence histogram of a mixture of nonselective beef enrichment culture and
E. coli O157:H7 stained with FITC-labeled anti-O157:H7 antibody. Region A
represents unstained bacteria and region E bacteria stained with the antibody.

the beef samples that was not seen in the pure culture mixtures (Fig. 4.1). This population was easily distinguished
from the brighter staining O157:H7, and was present in all samples regardless of the number of O157:H7 added. This
population may be a result of either weak cross-reaction of the antibody or nonspecific binding due to the processing
steps associated with the beef extracts such as trypsin and Triton × 100 treatment. There was good agreement in the
number of O157:H7/mL between Ab-DEFT, FC, and plate counts (Fig. 4.2). The higher fluorescence background seen
in FC produced roughly equivalent fluorescence counts in the mixture with no O157:H7 and in the mixture with the
fewest O157:H7 making the lower limit of detection by FC about 104 O157:H7/mL.

Sorting Viable O157:H7 from Beef Enrichments

Another potentially useful aspect of FC technology is the ability to sort populations of interest. From the food
microbiology perspective, this is valuable because it provides the opportunity to not only identify the presence of the
organism of interest, but to isolate and culture it for subsequent analysis. FITC-positive bacteria of the streptomycin-
resistant DHS-1 strain

 Page 63

Fig. 4.2
Comparison of the number of E. coli O157:H7 detected by FC
and Ab-DEFT in a series of beef-enrichment mixtures (B-H) with
a decreasing proportion of O157:H7. Plate represents the viable
plate count of the added O157:H7 for each mixture. Results are for
two experiments.

were sorted from the brightest LGFL population (Fig. 4.1) of three beef-enrichment mixtures with differing numbers of
O157:H7. Either 100 or 500 bacteria were sorted into PBS in four replicate wells of a 96-well plate. The sorted bacteria
were plated on agar medium with or without streptomycin. The number of colonies with and without streptomycin was
equivalent, indicating that the bacteria sorted were virtually pure strep-resistant O157:H7. There was an approximate
fivefold difference in colony number between 100 and 500 sorted cells. Only about one half the number of colonies as
compared to the number of sorted bacteria were present. Assuming that the sorting was 95100% numerically accurate,
the viability of O157:H7 in these mixtures was about 50%. These results were similar at all frequencies of O157,
indicating that the purity of the sort was not affected by decreasing O157:H7 to the 2% level.

 Page 64

Analysis of Bacteria by FC with Enhanced Light Scatter Resolution

Other studies using FC in microbial analysis (McClelland and Pinder, 1994a,b) have frequently depended on the use of
a custom-designed flow cytometer with special optical or electronic properties designed for microbial detection, which
are not commercially available. Because development of such a ''home-built" instrument is beyond the capabilities of
most microbiology laboratories not devoted to this purpose, we emphasized the use of off-the-shelf technology in the
preceding experiments. Recently a flow cytometer with light scatter sensitivity and resolution suitable for analysis of
small particles down to 0.2 µM (Steen, 1990) was commercially marketed. We were able to obtain, through the
courtesy of the manufacturer (BioRad Laboratories, Hercules, CA), an early version of this instrument (BioRad Bryte
HS) for use in our later experiments. The primary factors accounting for the increased FLS and SS are the use of
microscope optics and dark field illumination, which reduces background and increases sensitivity, and the use of a
PMT for FLS detection (Steen, 1990). Another advantage of this instrument is that the sample is delivered into the
sheath using a precision-controlled microsyringe. This allows the exact sample volume run to be known so that counts/
µL can be directly determined without the use of calibration beads.
Mixtures of O157:H7 and HB101 were stained with the FITC anti-O157:H7 antibody and analyzed on this instrument.
The FLS histogram resolved two populations of bacteria. The FLS-LGFL histogram demonstrated that the O157:H7
was the higher FLS population. Analysis of beef-enrichment-O157:H7 mixtures showed a similar improved resolution
of O157:H7 by FLS. Counts of O157:H7 in beef enrichment with this flow cytometer were equivalent to those
achieved with Ab-DEFT and the standard flow cytometer (Table 4.2). At the lowest concentrations of O157:H7 and in
the uninnoculated beef enrichment, higher background was associated with both FC instruments than with Ab-DEFT.
One solution to this problem may lie in the use of additional parameters to distinguish bacteria from debris.

Multiparameter Analysis

In addition to rapid real-time analysis of large numbers of cells, another powerful aspect of FC technology is that
multiparameter data for each cell are collected and stored in the computer or on disk, allowing for the possibility to
gate on one or more parameters for the purpose of increasing

 Page 65

TABLE 4.2 Comparison of Three Fluorescence-Based

Methods of Detecting E. coli O157:H7 in Beef-
Enrichment Culture (Counts/mL × 106)
CFU/mL × 106 Ab-DEFT FC-STDa FC-HRb
37 30 63 34
7.4 4.6 9 4.1
3.7 2.9 3.6 2.5
1.8 0.94 2.1 1.7
aAnalysis by flow cytometer with standard light scatter
resolution.
bAnalysis by flow cytometer with enhanced light scatter
resolution and metered sample volume.

the specificity of the analysis. Our experiments with the high-FLS-resolution flow cytometer provide a simple example.
Knowing that O157:H7 has a distinct FLS profile allows reanalysis of the listmode data, specifically gating on the
appropriate region of the FLS histogram. Using this multiparameter approach has the effect of increasing the frequency
of O157:H7 in the gated population and enhancing sensitivity (Fig. 4.3). Light scatter measurements have been used to
distinguish bacterial species in pure culture (Steen, 1990), however, the practicality of this approach for other
organisms in complex mixtures has not yet been determined. It is likely the FLS profile could be used in combination
with other parameters that can be used to differentiate bacterial genera or species. In addition to antibodies such as we
have used, DNA-binding dyes that bind preferentially to A-T or C-G have been used to distinguish bacterial species
(Van Dilla et al., 1983). The fluorescent in situ hybridization technique (FISH) utilizing probes for 16S rRNA has been
adapted for FC (Amann et al., 1990) and offers the potential advantage of species-specific identification. Another
parameter that can be determined by FC is viability. A series of mixtures of freshly isolated and killed E. coli were
stained with a combination of dyes that cause viable cells to emit green and nonviable cells to emit red (Bac Light
live/dead kit, Molecular Probes, Eugene, Ore.). The LRFL-LGFL histogram showed two distinct populations, which
correlated with the mixture proportions of live and dead bacteria (Fig. 4.4). This technique has been used for antibiotic
susceptibility testing and may have applications in food microbiology.

Page 66
 Fig. 4.3
Frequency of FITC-labeled O157:H7 (boxed population
in lower panels) in beef-enrichment mixture based on
gating off FLS properties (boxed region in upper panels).
Gating is for the entire FLS population (A) or the region
enriched for O157:H7 (B).

Summary

FC as a tool for food microbiology offers the advantage of rapid and objective sample analysis. Analysis of 2050 × 104
individual bacteria can be accomplished in 12 min. The ability to collect in real time and store multiparameter data for
subsequent reanalysis and gating and the ability to sort bacteria of interest from mixed populations are unique features
of this technology. FC enumeration of E. coli O157:H7 in nonselective beef enrich-

 Page 67

Fig. 4.4
Two-parameter (LGFL x LRFL) FC analysis of
live and ethanol-killed E. coli (A) 8:2 live:dead;
(B) 2:8 live:dead. Dead bacteria stained red
(y axis); live bacteria stained green (x axis).

ment cultures was equivalent to Ab-DEFT, especially at higher concentrations of O157:H7. The major drawbacks are
the background fluorescence counts seen with FC but not with Ab-DEFT, which limit sensitivity, and a shortage of
specific antibody probes for important bacterial pathogens. Other areas in food microbiology where FC may be
particularly useful are in the analysis and sorting of subpopulations of bacteria from complex mixtures and in the
development of automated systems for quality control.

References

Amann, R.I., Binder, B.J., Olson, R.J., Chisholm, S.W., Devereux, R., and Stahl, D.A. 1990. Combination of 16S
rRNA-targeted oligonucleotide probes with flow cytometry for analyzing mixed microbial populations. Appl. Environ.
Microbiol. 56: 19191925.

Donnelly, C.W. and Baigent, G.J. 1986. Method for flow cytometric detection of Listeria monocytogenes in milk.
Appl. Environ. Microbiol. 52: 689695.

McClelland, R.G. and Pinder, A.C. 1994a. Detection of Salmonella typhimurium in dairy products with flow cytometry
and monoclonal antibodies. Appl. Environ. Microbiol. 60: 42554262.

McClelland, R.G. and Pinder, A.C. 1994b. Detection of low levels of specific Salmonella species by fluorescent
antibodies and flow cytometry. J. Appl. Bacteriol. 77: 440447.

 Page 68

Raybourne, R.B. and Bunning, V.K. 1994. Bacterium-host cell interation at the cellular level: fluorescent labeling of
bacteria and analysis of short-term bacterium-phagocyte interaction by flow cytometry. Infect. Immun. 62: 665672.

Steen, H.B. 1990. Light scattering measurement in an arc lamp-based flow cytometer. Cytometry. 11: 223230.

Tortorello, M.L. and Stewart, D.S. 1994. Antibody-direct epifluorescent filter technique for rapid, direct enumeration
of Escherichia coli O157:H7 in beef. Appl. Environ. Microbiol. 60: 35533559.

Van Dilla, M.A., Langlois, R.G., Pinkel, D., Yajko, D., and Hadley, W.K. 1983. Bacterial charaterization by flow
cytometry. Science 220: 620622.

 Page 69

5
Optical Biosensors for Microbiological Analysis
Nile F. Hartman
Georgia Tech Research Institute
Atlanta, Georgia
With recent advancements in the field of optics, new and innovative sensing technologies have emerged. In some
instances, these technologies are proving to be effective as rapid response biosensing tools and, in practice, may prove
to be valuable for food safety and inspection. As biosensors, the optical technologies are capable of detecting biological
contaminants including foodborne pathogens at concentration levels below those achievable with current ELISA
methods. Equally important, sample preparation and chemistry may be dramatically simplified. In this paper, emerging
optical biosensor technologies are reviewed and their potential applicability evaluated.

Definition
By definition, a biosensor consists of a transducer capable of detecting a biological reaction. Of the emerging optical
techniques, the so-called evanescent field sensors have proven to be effective transducers. As biosensors,

 Page 70

Fig. 5.1
Evanescent field interaction at a dielectric interface.

they typically rely on immunoassay reactions occurring at a solid support surface. In general, the sensing mechanism
depends on (1) perturbing the properties of a guided optical wave, (2) altering the performance of an optical device, or
(3) inducing fluorescence at the interface of an optical waveguide.

The basis for evanescent field sensing is illustrated in Fig. 5.1. An optical ray reflected from a surface has associated
with it an electric field extending through the dielectric boundary or interface into the adjacent medium. The strength of
the field decays exponentially with distance from the surface, and its penetration depth is dependent on the optical
properties of the propagation media as well as the surrounding media. Typically the field depth is less than one
wavelength of the light wave. Most importantly, changes occurring at or on the surface of the interface, such as the
selective binding of a protein to covalently bound antibodies, can perturb the properties of an optical wave through
interaction with the evanescent field. Alternatively, the evanescent field can be used to induce fluorescence in labeled
molecules bound to a waveguide surface. An important distinction between many of the emerging optical techniques
and other biosensing methods is their ability to directly detect binding of an antigen or an antibody to a solid support
surface. As a result, no labeling or tagging is required for detection.

Description of Optical Technologies

The emerging optical technologies are based on optical waveguides including both planar and cylindrical fiber formats
and resonant thin film structures. For many applications, the planar waveguide configurations are advantageous as they
provide direct access to a waveguide surface and greater flexibility in choice of waveguide material systems. These are
critical to optimizing overall device sensitivity. The evanescent wave biosensing approaches include the following
device technologies:

 Page 71

Grating-based sensors
Resonant mirror sensors
Surface-enhanced plasmon resonance
fiber optic fluorescence systems
integrated optic interferometers

Of these techniques, only the surface-enhanced plasmon resonance (SPR) and the fiber optic fluorescence sensors have
been commercialized to any degree (e.g., BIAcoreTM, Pharmacia Biosensor AB, Piscataway, N.J., and Analyte 2000
Immunoassay System, Research International, Inc., Woodinville, Wash.). The grating-based devices, the resonant
mirror system, and SPR are all similar in operation and physical structure. Functionally, they behave as nonintegrating
type sensors. In contrast, the IO interferometer and the fiber optic fluorescence systems function as integrating devices,
that is, signal is directly proportional to optical path length or surface area. As a result, the most promising of these
techniques based on demonstrated and projected sensitivity are the fiber optic fluorescence and the integrated optic
interferometric sensors. The following discussion focuses on these two methods, although the grating-based device, the
resonant mirror system, and the SPR are briefly described.
Grating-Based Sensors

The grating-based sensor (ASI AG, 1993), illustrated in Fig. 5.2, consists of a periodic grating structure formed on an
optically flat surface. Because of

Fig. 5.2
Grating-based biosensor configuration.
(From Artificial Sensing Instruments, 1993.)

 Page 72

the periodic nature of the grating, incident monochromatic light is scattered (diffracted) at a well-defined angle. That
angle is determined by the wavelength of the incident light, the grating period, and the refractive index of the
surrounding media at the device surface. By attaching an antibody to the surface in the grating region, the effective
refractive index at the grating surface changes with the binding of an antigen to the antibody. The index change
manifests itself as a slight change in the angular position of the diffracted signal. In actual practice this device is
relatively insensitive when compared to the detection levels realistically required for the detection of foodborne
pathogens.

Surface Plasmon Resonance

SPR sensors are also relatively simple devices, similar to grating sensors. In operation, an optical wave incident on a
dielectric/thin metal film interface excites a plasmon wave in the thin metal film. The excitation process only occurs at
a very specific angle for a given optical wavelength. At the excitation condition, the optical energy is strongly absorbed
and a strong dip in reflected signal intensity is observed. The resonance condition is extremely sensitive to any changes
on the surface of the metal film. In the case of antigen binding to an antibody covalently attached to the surface, the
change in effective refractive index is sufficient to detune the device from resonance. In practice, the detuning causes
the resonance condition to be shifted to a slightly different angular position. As noted previously, the SPR technique
has been commercialized by Pharmacia Biosensor AB (Piscataway, NJ.). Sensitivities of a few hundred ng/ml have
been realized with these instruments for typical proteins (Fontana et al., 1990). Again, these levels are not sufficient for
a practical, rapid-response biosensor to be used for food inspection.

Resonant Mirror Sensors

The resonant mirror device is also somewhat similar to the SPR and grating-based sensors in both configuration and
operation (Hartman et al.). Its physical configuration is depicted in Figure 5.3. In operation, total internal reflection
occurs at the boundary between a dielectric material and a thin resonant layer. With a change in effective refractive
index in the sensing layer, light undergoing total internal reflection within the resonant layer undergoes a phase shift.
Using two polarized light beams (polarized perpendicular and parallel to the plane of incidence), the relative phase
shift due to an index change is easily determined. Protein molecules have been detected at less than 0.1 nM levels. In
practice, sensitivities at fM levels or lower will be required for food safety and inspection.

 Page 73

Fig. 5.3
Physical configuration of a resonant mirror sensor.

Integrated Optic Interferometer

The integrated optic (IO) waveguide interferometer detects refractive index changes occurring on or at the surface of an
optical waveguide. The combination of the inherent stability provided by the integrated optic platform, the unique
sensitivity of the interferometer, and the reduction of thermomechanical noise allows index changes of 10-6 or less to be
detected. The IO interferometer is also an integrating device as the detected signal strength is directly proportional to
the length of the sensing layer. This is an important distinction between this technology and the prior sensing methods.
Because of these features, the IO interferometer detection capability is in the fM or sub-ng/mL range.

The basic IO interferometric sensor structure is illustrated in Figure 5.4. It consists of a supporting substrate, a central
waveguide film, and a covering superstrate film designed to be interactive with the species to be sensed. The optical
source is typically a solid-state diode laser with its output beam coupled into the waveguide by means of a surface
grating coupler. The grating coupler provides a practical and easy-to-use method of coupling light into a waveguide.

In a typical sensor configuration, the waveguide is a thin layer (<200 nm) of slightly higher refractive index material
deposited on the surface of an optically transparent substrate. For biosensing, the chemically selective layer may be a
covalently attached antibody that binds with a specific antigen. The selective binding alters the effective refractive
index of the waveguide and the phase or velocity of the guided wave. The resulting phase change is easily

 Page 74

Fig. 5.4
Interferometric sensor package cross-sectional view.

detected by integrating the sensing guided wave with an unperturbed reference guided wave.

At the current level of development the IO sensor has proven to be capable of detecting Salmonella at concentrations of
less than 104 microbes/mL. Work to date has used commercial polyclonal and monoclonal antibodies attached to the
surface of the IO biosensor by means of silane-coupling chemistry. This method of attachment effectively isolates the
antibody from the polar surface of the waveguide and, thus, minimizes conformational changes and/or denaturing of
the protein.
For testing purposes, standard samples of whole Salmonella cells were prepared and characterized by flow cytometry.
A typical sensor response to the whole cell samples is shown in Figure 5.5. The observed signal phase shift was quite
significant, approximately 0.5p radians after 1 min, with an equilibrium value of greater than 4.5p radians after several
minutes. Note

Fig. 5.5
Interferometer response to whole Salmonella cells.

 Page 75

that the measurements were made in a ''broth" containing other proteins and fats and no problems were observed with
nonspecific binding or surface contamination. Because the minimum detectable phase shift (due to noise) is
approximately 0.02p radians, the observed results imply that, at least for this system, the IO biosensor should be
capable of detecting as few as 104 cells/mL, which is an order of magnitude enhancement of sensitivity relative to
ELISA techniques. With improved surface activity and better sample-delivery approaches, the projected detection level
for Salmonella is expected to be in the range of 1001000 cells/mL and require only tens of minutes.

Although the IO interferometer provides direct detection of antigen molecules binding to the waveguide surface,
sandwich assays may be performed as well. The sandwich assay offers enhanced signal levels and reduced false-
positives and serotyping capabilities. Experimental measurements made by coming back with an antibody after
selective binding of an antigen to a functionalized waveguide surface has resulted in an additional signal gain of
approximately one order of magnitude.

Fiber Optic Fluorescence Probes

The fiber optic fluorescence bioassay methods have received considerable attention in the past years due to military
interest as candidate systems for detection of biological warfare agents. This technology has received considerable
attention at the Naval Research Laboratory, where a tapered fiber system has been developed that provides increased
signal collection efficiency (Golden et al., 1992). The fiber probe is functionalized with an appropriate antibody or
receptor molecule. Immunoassays may then be performed using competitive or sandwich assays. This technology is
being commercialized in the form of a four-probe sensor approximately the size of a shoebox. The instrument utilizes a
diode laser as the excitation source and a long-wave pass filter to block the excitation signal and transmit the collected
fluorescence signal from the labeled and captured antigen. To further enhance the signal-to-noise ratio, phase-locked
detection is utilized. This technology has proven to be quite effective and has been used to detect common pathogens
and toxins. Microbial concentrations in the range of a few hundred cells/mL have been detected, as have toxins
(botulinum and aflatoxin) at sub-ng/mL concentrations (Anderson, 1996).

Summary

Optically based biosensor technologies utilizing immunoassay chemistries are emerging from the laboratories. These
technologies are potentially capable of providing rapid-response systems suitable for detection and

 Page 76

identification of common foodborne pathogens. They offer sensitivity, direct detection capabilities, rapid response, and
multianalyte tracking capabilities. In particular, the fiber optic fluorescence approach and the integrated optic
interferometer are capable of detection sensitivities that exceed currently available ELISA test kits. The fiber optic
technique is much further along in development, whereas the integrated optic system is a relatively new technology in
the early development stage. Both systems are capable of detection sensitivities in the 100 cell/ml range. The integrated
optic system is a direct detection scheme, although sandwich assays may implemented for serotyping and signal
amplification. It also requires no special sample preparation as the presence of other proteins and fatty materials show
no interference. At the present time, neither of the systems is likely to detect at the 1 microbe/25 g level proposed as a
quality standard.

References
Anderson, G. 1996. Fiber Optic Biosensors, The 5th Biennial DOD Photonics Conference, March 2628, McLean, VA.

ASI AG, Zurich, Switzerland. 1993. Artificial sensing instruments. Biosensors Bioelectronics 8(7/8).

Fontana, et al. 1990. Appl. Optics 29(31).

Golden, et al. 1992. Fluorometer and tapered fiber optic probes for sensing in the evanescent wave. Optic. Eng. 31 (7):
14581462.
Hartman, et al. Rapid response biosensor for detection and identification of common foodborne pathogens. SPIE Proc.
2345: 128.

 Page 77

6
Immunodiagnostics in the Detection of Foodborne Pathogens
Barbara J. Robison
Organon Teknika Corporation
Durham, North Carolina

Introduction

Today more than ever before, the public is concerned about the safety of the food that it consumes. Food producers and
processors want to provide safe food, but they also want to minimize production costs by shipping their products more
quickly. Prior to the 1980s, foods were tested for the presence of pathogens primarily by conventional culture
procedures, which are labor intensive and time consuming. There was obviously a clear need for rapid, accurate
microbiological methods. Advances in biotechnology have made it possible to utilize the power of immunology to
detect foodborne pathogens (Feng, 1996).
To understand the principles of immunological methods, one must first be familiar with the antigen-antibody
relationship. An antigen can be defined as a substance that is capable of provoking the production of antibodies.
Antigens are typically high molecular weight proteins, carbohy-

 Page 78

drates, or polypeptides that are recognized as foreign by the host. An antigen molecule may consist of a number of
chemically unique components termed antigenic determinants. The antibody molecules produced as a result of
exposure to the antigenic determinants are a group of serum proteins, or immunoglobulins, made by B lymphocytes. Of
the classes of immunoglobulins produced, IgG is by far the most common. This molecule resembles the letter Y in its
shape, with the antigen-binding sites present at the ends of the arms of the Y. These binding sites account for the unique
specificity of the antibody. This recognition of an antigen by an antibody is the principle that makes immunological
methods such powerful diagnostic tools. To detect foodborne pathogens, antibodies are commonly produced that detect
flagellar components, cell wall antigens, or polysaccharides. These antigens are generally heat stable, and the samples
can be boiled without affecting the antigen. Such preparations are no longer a biohazard and can be safely handled in
the laboratory.

The first use of immunological methods for diagnostic purposes occurred in the early 1900s, when researchers
discovered that the serum and urine of patients with typhoid fever contained a soluble substance that would precipitate
when mixed with rabbit anti-Salmonella antiserum (Marcon, 1995). In fact, most of the early methodology utilizing
antigen-antibody reactions was available in the clinical laboratory long before such methods came into use by food
microbiologists. As a result of the growing interest in detecting infectious agents more rapidly, the technology for
producing large quantities of antibodies with high purity, avidity (ability of an antibody to bind antigen), and specificity
was developed. In particular, the development of techniques to produce monoclonal antibodies has contributed to the
improvement of the specificity of enzyme immunoassays. Since the antibodies used in a diagnostic test are the key to
its success, improvements in antibody production and purification have resulted in an explosion of diagnostic kits,
particularly in the food industry. For example, during the years 19391985, only three methods had been validated by
the AOAC that utilized commercial kits; from 19861993, that number jumped to 30 (Andrews, 1996).

There are a number of formats available today that employ antibodies to specifically detect certain foodborne
pathogens. The detection systems discussed in more detail here include the following: (1) enzyme-linked
immunosorbent assays, both colorimetric and fluorogenic; (2) particle agglutination tests, including latex agglutination
and single-use devices; (3) precipitation tests, including immunoimmobilization; (4) direct immunostaining formats;
(5) antibody-coated paramagnetic beads; and (6) antibody-based biosensors.

 Page 79

ELISA Methods

The immunological method that enjoys the most widespread use is the enzyme-linked immunosorbent assay (ELISA).
The ELISA technique was first described in 1971 by two independent groups, Engvall and Perlmann in Sweden and
van Weemen and Schuurs in Holland. The principles on which it is based are essentially very similar to the fluorescent
antibody technique, i.e., an antigen is specifically detected due to the discriminatory power of an antibody, and a
"label" attached to the antibody allows for the reaction to be visualized. Enzyme labels are particularly useful because
they are not hazardous, they have long shelf lives, and they have the capacity of amplification. Depending on the
substrate used, enzyme assays can be either colorimetric or fluorogenic. As a general rule, horseradish peroxidase is the
most sensitive enzyme for colorimetric detection, while b-galactosidase is most sensitive for fluorometric assays
(Kurstak, 1986). ELISA assays are classified as either competitive or noncompetitive, depending on whether the
technique involves a reaction step in which unlabeled antigen and enzyme-labeled antigen compete for a limited
number of antibody-binding sites or whether the antigen to be detected is allowed to react alone with an excess of
immune reactant. Most ELISA procedures used for microbial antigen detection require physical separation of the free,
labeled antibody from the antibody-antigen complexes. For this reason, the assays are referred to as heterogeneous. The
technique most commonly used to detect bacterial antigens in foods is a version of the noncompetitive ELISA called
the antigen sandwich (Fig. 6.1). In this configuration, relevant antibody is adsorbed onto the surface of the wells of a
microtiter plate. The sample is then added to the well and the microtiter plate is incubated; if antigen is present, it will
bind to the antibody on the well. Following a wash step to remove unbound antigen, an enzyme-labeled antibody
(conjugate) is added, which also binds to the antigen. After a wash step to remove unbound conjugate, the substrate is
added. In the presence of the enzyme, the substrate is converted to a colored end product. After a specified period of
time, the reaction is stopped by adding a stop solution. The color intensity is measured on a spectrophotometer at a
specified wavelength, and a numerical result is obtained. Alternatively, the color reaction can be determined visually
rather than by an instrument; however, visual reading introduces an element of subjectivity into the determination
(Robison, 1995). Examples of ELISA tests utilizing this format are Salmonella-TekTM, Listeria-TekTM, and EHEC-
TekTM from Organon Teknika (Fig. 6.2); TECRA® Salmonella, Listeria, Bacillus cereus enterotoxin, and
staphylococcal enterotoxin from Biotech Australia; RIDASCREENTM ELISA for staphylococcal enterotoxins from

 Page 80

Fig. 6.1
Diagram of the antigen sandwich format of an ELISA assay.
(Reprinted by permission of Organon Teknika Corp.)

Fig. 6.2
ELISA kits for the detection of foodborne pathogens. Test kits
shown are for the detection of Salmonella, Listeria, and E. coli O157:H7.
(Reprinted by permission of Organon Teknika Corp.)

 Page 81

R-Biopharm GmbH; Premier EHEC Enzyme Immunoassay from Meridian Diagnostics; and AssuranceTM Salmonella,
Listeria, and E. coli O157:H7 EIA from BioControl Systems.

One of the advantages of ELISA over other rapid detection methods is the fact that it can be automated. The VIDAS®
and the mini VIDAS® (bioMerieux Vitek, Inc.) are examples of automated systems (Fig. 6.3). Although the principles
of the assays run on the VIDAS are the same as described above, there are some notable differences. These instruments
perform all enzyme-linked fluorescent immunoassay (ELFA) and measure a fluorogenic end product rather than a
colorimetric one. The substrate used in these assays is 4-methylumbelliferyl phosphate, which is converted to the
fluorogenic compound 4-methylumbelliferone in the presence of the enzyme alkaline phosphatase. The reagent system
consists of a predispensed disposable reagent strip and a solid phase receptable (SPR) rather than a microtiter plate for
immobilization of the target antigen. The reagent strip contains all of the individual reagents needed for the
immunoassay. The SPR is a pipette-tip-like disposable device, which is coated on its inside surface with antibody used
to capture the target antigen. The SPR also serves as a pipettor for accurate sampling and transfer of reagents during the
assay. The VIDAS instrument consists of an assay processor module,

Fig. 6.3
The VIDAS® automated immunoassay system. This instrument
can process up to 30 immunoassays and can be used to detect
Salmonella, Listeria, Staphylococcal enterotoxins, E. coli O157,
and Campylobacter.
(Reprinted by permission of bioMerieux Vitek, Inc.)

 Page 82

which executes all of the procedures necessary to perform heterogeneous immunoassays. The mini VIDAS is a smaller
version of the VIDAS and can run 12 tests at once compared to 30 with the VIDAS. Assays currently available on
these systems include Salmonella, Listeria monocytogenes, E. coli O157, staphylococcal enterotoxin, and
Campylobacter (bioMerieux Vitek, Inc.).

Particle Agglutination Assays

The most straightforward assay format utilizing antibodies is latex agglutination. This assay makes use of the fact that
latex (polystyrene) beads bind antibody molecules to their surface. Each latex bead is typically 1 mm in diameter and
can carry thousands of antibody molecules. When thoroughly mixed, the antibody-coated latex particles form a
uniform, milky suspension. Presence of the specific antigen, however, results in binding of the antigen to the antibody
and also cross-linking of antigens and antibody. This secondary cross-linking results in the formation of a lattice
structure that precipitates out of solution, forming a visible agglutination reaction (Fig. 6.4) (Marcon, 1995). Although
they are very simple to use, the sensitivity of these assays is usually at least 10 times less than that of ELISA assays. A
number of manufacturers provide these assays for detection of such pathogens as Salmonella, Listeria, Staphylococcus
aureus, and E. coli O157.

Fig. 6.4
Latex agglutination assay for detection of E. coli O157. The latex
in well 1 has agglutinated, indicating a positive reaction, while
there is no agglutination in well 2, indicating a negative response.
(Reprinted by permission of Unipath Limited.)

 Page 83

Latex particles carrying antibodies can also be found in single-use, or lateral flow, devices. These devices utilize a
format termed immunochromatography. This format does not require agglutination of the antigen/antibody, but rather
the latex acts as a carrier for the antigen/antibody complex. One such commercially available test kit is the
ClearviewTM Listeria Device from Oxoid. This device contains specific monoclonal antibodies to a flagellar antigen
found in Listeria species. Heated sample extract is added onto a pad in the sample window, which also contains blue
latex labeled with anti-Listeria antibodies. The extract rehydrates the complex, and the specific antigen reacts, if
present, with the antibody. The complex moves through the pad by capillary action to a test strip containing an
immobilized line of antibody midway along the Result Window. A further reaction between antigen/latex complex and
the fixed antibody results in a blue line in the Result Window (Fig. 6.5). If no Listeria antigen is present, the Result
Window will remain clear. The device also provides an integral

Fig. 6.5
A single-use device utilizing immunochromato-
graphy. The device on the left shows a line in
the Result Window indicating a presumptive
positive reaction for Listeria, while the device
on the right shows no line and is negative.
(Reprinted by permission of Unipath Limited.)

 Page 84

control. The appearance of a blue line in the Control Window indicates that the test has been conducted properly
(Oxoid, 1994). Antibodies can also be bound to other particles such as gold sol and used in this type of format. Tests
for E. coli O157, Listeria, and Salmonella in this format are also available from BioControl, Neogen, and LUMAC.

Precipitation Tests
In precipitation reactions, the antigens and antibodies are not bound to a solid phase or carrier. Rather, soluble antigens
react with antibodies to form aggregates. These reactions normally occur only when the ratio of antibody to antigen is
optimal. Immunodiffusion tests are precipitation reactions carried out in an agar medium. The BioControl Salmonella
1-2 Test® is a specialized immunoimmobilization assay that uses the principles of immunodiffusion. This test is
conducted in a small disposable plastic unit with two compartments: one for selective enrichment and one for the
immunoimmobilization process to take place. There is a separator between the two chambers, which is removed prior
to inoculation. Prior to inoculation, anti-Salmonella antibody is added to the motility chamber. After preenrich-

Fig. 6.6
A device utilizing immunoimmobilization. The device on
the left does not contain Salmonella, and therefore there
is no interaction between the antigen and antibody. The
device on the right is positive; flagella of Salmonella
are immobilized when they come into contact with the
anti-Salmonella antibody, forming an "ImmunoBand."

 Page 85

ment, a sample is inoculated into the chamber containing the selective enrichment medium. If salmonellae are present
in the test sample, the motile organisms present in the selective media will migrate through the motility chamber, and
eventually encounter the antibodies diffusing through the gel. The antibodies bind to the flagella of the salmonellae and
immobilize the cells into a radial pattern that reflects the antibody diffusion path (BioControl Systems, 1989). This
''immunoband" is presumptive for the presence of Salmonella (Fig. 6.6).

Direct Immunostaining

Immunoblotting is defined as the transfer of antigenic material from a surface to a nitrocellulose membrane. Once the
antigens are present on the membrane, antibodies specific for the antigen can be added and an ELISA assay can be
performed on the nitrocellulose membrane. The only difference from the ELISAs described earlier is that the substrate
must not be water soluble.

An example of a rapid method that utilizes this type of format is the PetrifilmTM Test Kit-HEC for detection of E. coli
O157. The procedure involves inoculation of Petrifilm E. coli Count plates with the test sample (3M HealthCare,
1994). Bacteria present, including any E. coli O157:H7, form colonies on the Petrifilm plate. Colony antigens are then
transferred from the Petrifilm plate to a reactive disc. If E. coli O157 antigens are present, they will be transferred to
the disc and will capture the enzyme-labeled antibody, which is subsequently added. The antigen-antibody complexes
are visually detected when the bound enzyme converts an identifying substrate to a permanent black spot on the disc.
Each spot indicates an O157:H7 presumptive-positive colony, which can be confirmed by returning to the Petrifilm
plate and culturing the suspect colonies onto MSA-BCIG agar plates, followed by biochemical and serological
identification (Fig. 6.7).

Antibody-Coated Paramagnetic Beads

Antibodies can also be coated onto the surface of superparamagnetc polystyrene beads. The advantage of using these
magnetic beads is that specific bacteria bound to the antibodies on the beads can be isolated from heterogeneous
bacterial soups by the simple application of a magnetic field (Fig. 6.8). This immunoseparation technique produces a
significant degree of selective enrichment when applied as part of an isolation process and allows even more rapid
detection of the target organism. Once the organisms have been physically removed from the competing microflora,
they can be di-

 Page 86

Fig. 6.7
A 3MTM PetrifilmTM E. coli Count plate on the left showing positive
E. coli colonies and its corresponding reactive disk.
(Reprinted by permission of 3M Health Care.)

Fig. 6.8
Salmonella cells on the surface of paramagnetic beads.
(Reprinted by permission of Dynal, Inc.)

 Page 87

rectly plated on selective agars or further enriched and detected by ELISA or genetic probes (Dynal, 1991). Dynal®
provides antibody-coated beads (Dynabeads®) for the detection of Salmonella and E. coli O157. When the beads are
used in combination with the Salmonella-Tek and EHEC-Tek ELISA assays, Salmonella and E. coli O157:H7 can be
detected in 24 hr, a significant improvement over the time needed for the traditional culture method. Another
manufacturer, VICAM, provides kits that utilize antibody-coated magnetic beads to detect Listeria.

Biosensors

One of the newest developments in immunology-based detection systems is the BIAcore system from Pharmacia.
BIAcore is a microchip-based system for analyzing the formation of macromolecular complexes, for example, the
interaction between an antibody and an antigen (Malmqvist, 1993). The BIAcore system contains a sensor microchip,
an LED light source emitting polarized light, an automated fluid-handling system, and a diode-array position-sensitive
detector. The sensor chip consists of a glass support, an overlaid gold film, and a carboxymethylated dextran matrix to
which biomolecules (antigens or antibodies) can be coupled. The detection principle of this system relies on surface
plasmon resonance (SPR), an optical technique that measures changes in the refractive index at the sensor chip surface
(Fig. 6.9). To study interactions between two molecules, one of the molecules is covalently immobilized on the sensor
chip and the second molecule is injected over the chip surface. Interactions between the two molecules result in mass
changes at the chip surface, which translate to refractive index changes and changes in the SPR angle. The changes in
SPR angles are monitored as resonance units (RU) (Raghavan and Bjorkman, 1995). Although the sensitivity of this
system is similar to that of enzyme immunoassays, the main advantages of this technology are that it measures the
molecular interactions as they occur, no labeling of antigen or antibody is required, and the biosensor is reusable. Like
ELISA, a number of configurations are possible with this technology, including a direct assay (antibody bound to the
sensor to capture antigen), a sandwich assay as described for ELISA, and an inhibition assay (antibody bound to the
sensor, which competes with antibody added to the sample).

Two groups are currently investigating the use of BIAcore for detecting foodborne pathogens. The USDA-ARS has
looked at detection of cultures of E. coli O157:H7 (Fratamico, 1995), while researchers at Leatherhead Food Research
Association in England have successfully demonstrated that the system can detect Salmonella species. All work to date
has been with pure

 Page 88

Fig. 6.9
Surface plasmon resonance, an optical technique that measures
changes in the refractive index at the sensor chip surface.
(Reprinted by permission of Pharmacia Biosensor AB.)

cultures of Salmonella, and BIA detected the serotypes tested at 104 cfu/mL, with no cross-reactivity with 10 other
bacterial species (Haines and Patel, 1995).

Conclusions
The power of immunology has been increasingly utilized to provide us with methods that can rapidly and accurately
detect a number of foodborne pathogens. The wide range of formats makes it possible for almost any laboratory, large
or small, to test their food products and get them to the marketplace more quickly. Certainly if the future is as fruitful
as the past 15 years have been, the food microbiologist of the twenty-first century will have a very interesting job
indeed.

Acknowledgments

I thank the following individuals for providing product materials which were extremely useful in the preparation of this
manuscript: Becky Brown,

 Page 89

bioMerieux Vitek, Inc.; Mike Robinson, Pharmacia Biosensor AB, DeAnn Benesh, 3M Health Care; Heidi Israel,
Dynal; and Keith May, Unipath. In addition, I thank R. Durham and J. DiGuiseppi for their careful review of this
manuscript.

References
Andrews, W.H. 1996. Evolution of methods for the detection of Salmonella in foods. JAOAC Int. 79(1): 412.

BioControl Systems product literature. 1989.

bioMerieux Vitek, Inc. VIDAS and mini VIDAS product literature.

Dynal A.S. 1991. Dynabeads anti-Salmonella package insert. 1991.

Feng, P. 1996. Emergence of rapid methods for identifying microbial pathogens in foods. JAOAC Int. 79(3): 809812.

Haines, J. and Patel, P.D. 1995. Detection of food borne pathogens using BIA. BIA J. 2(2): 31.

Kurstak, E. 1986. Enzyme Immunodiagnostics. Academic Press, Inc., Orlando, FL.

Malmqvist, M. 1993. Biospecific interaction analysis using biosensor technology. Nature 361: 186187.

Marcon, M.J. 1995. A. Direct microbial antigen detection. In Textbook of Diagnostic Microbiology. C.R. Mahon and
G. Manuselis, Jr. (Ed), p. 119140. W.B. Saunders Co., Philadelphia, PA.

Oxoid Listeria Rapid Test package insert. 1994.

Raghavan, M. and Bjorkman, P.J. 1995. BIAcore: A microchip-based system for analyzing the formation of
macromolecular complexes. Structure 3: 331333.

Robison, B.J. 1995. Use of commercially available ELISA kits for detection of foodborne pathogens. In: Methods in
Molecular Biology, Vol. 46: Diagnostic Bacteriology Protocols, J. Howard and D.M. Whitcombe (Ed.), p. 123132.
Humana Press, Inc., Totowa, NJ.

3M Health Care, PetrifilmTM test kit package insert. 1994.

 Page 91

7
Automated Microbial Identification Systems
Charles E. Stager
Ben Taub General Hospital and
Baylor College of Medicine
Houston, Texas

Introduction

Prior to the 1970s, microorganisms were primarily identified with conventional biochemical tests and often required
several days to weeks for identification. In the early 1970s, API (bioMerieux Vitek) introduced lyophilized substrates
in microcupules that allowed many clinical isolates to be identified in 2448 hr. Subsequends, with use of a heavy
inoculum, preformed enzymes could be detected, allowing some bacteria to be identified in 4 hr. In the latter half of the
1970s, automated identification systems were introduced. Our discussion will be limited to identification systems that
have automated result entry, have a data base for the identification of a large variety of different microorganisms and
are available in the United States.
Table 7.1 compares the features of the automated identification systems to be discussed. With the exception of the
Alamar system, descriptions of the components, principle of operation, kinds of microorganisms identi-

 Page 92

TABLE 7.1 Comparison of Features of Automated Identification Systems

Values for automated identification system
Sensititre-
Feature Viteka autoSCAN.4b WalkAwaya Sensititreb ARISa Alamarb Biologb MIDIa
Capacity 32, 60, or 120 Unlimited 40 or 96 Unlimited 192 Unlimited Unlimited 60
IDc (no. of substrates tested)d
GN 105 (29) 115 (27) 115 (27e, 34f) 140 (32) 140 (32) 32 (24) 567 (95) ~300 (NA)g
GP 47 (29) 48 (27) 48 (27e, 34f) No No No 255 (95) ~150 (NA)
ANA No No No No No No No ~600 (NA)
FAS No 20 (18) 20 (18) No No No 49 (95) ~40 (NA)
ENV 58 (29) No No No No No ~400 (95) ~200 (NA)
Yeasts 36 (26) 42 (27) 42 (27) No No No 267 (95) 194 (NA)
MYCO No No No No No No No 28 (NA)
Inoculation Automated Manual Manual Automated Automated Automated Manual Automated
Incubation
GN 418 hr 1642hr 2 or 1642 hr 5 or 18 hr 5 or 18 hr 1820 hr 4 or 24 hr 30 min
GP 415 hr 1642 hr 2 or 1642 hr 4 or 24 hr 30 min
ANA 30 min
FAS 4 hr 4 hr 4 or 24 hr 30 min
ENV 624 hr 4 or 24 hr 30 min
Yeast 24 or 48 hr 4 hr 4 hr 24, 48, or 72 hr 30 min
MYCO 30 min
Manual reagent addition No Yes No Yes Yes Yes No No

(table continued on next page)

 Page 93

(table continued from previous page)

TABLE 7.1 Comparison of Features of Automated Identification Systems
Values for automated identification system
Sensititre-
Feature Viteka autoSCAN.4b WalkAwaya Sensititreb ARISa Alamarb Biologb MIDIa
Additional tests required before incubation Yes Yes Yes No No Yes No No
Storage temp. 4°C RTe,h, 4°Ci RTe, 4°Ci RT RT RT 4°C RT
Other features
Susceptibility testing Yes Yes Yes Yes Yes Yes No No
Urine screen/ identification Yes No No No No No No No
DMS Yes Yes Yes Yes Yes Yes Yes Yes
Computer interface Yes Yes Yes Yes Yes Yes Yes Yes
List price $52,000j $49,900 $85,000m $15,800 $37,500 $30,000 $30,756 $52,000
$80,000k $125,000n
$97,000l
aOn-line incubation.
bOff-line incubation.
cID, number of groups, genera, or species identified.
dGN, aerobic gram-negative bacilli; GP, aerobic gram-positive bacteria; ANA, anaerobic bateria; FAS, fastidious bacteria; ENV, environmentally, isolated bacteria; MYCO,
mycobacteria.
eConventional identification panel.
fFluorogenic identification panel.
gNA, not applicable.
hRT, room temperature.
iAll rapid identification panels.
jList price of Vitek system with 32-card capacity.
kList price of Vitek system with 60-card capacity.
lList price for Vitek system with 120-card capacity.
mList price for WalkAway-40 with 40-panel capacity.
nList price for WalkAway-96 with 96-panel capacity.

 Page 94

fied, and inoculum preparation for the automated identification systems discussed here are found in a review by Stager
and Davis (1992) and will not be repeated. The Alamar system will be described below.

Stager and Davis (1992) reviewed the automated systems available from the late 1970s to the spring of 1992. This
discussion includes the studies published on automated systems since the spring of 1992.

Vitek

Visser et al. (1992) compared the Vitek Gram Negative Identification Card (GNI; bioMerieux Vitek) for identification
of 207 members of the Enterobacteriaceae and 41 nonfermentative gram-negative bacilli that had been maintained
frozen at -70°C. The Enterotube II system (Hoffmann-LaRoche, Switzerland) and Oxi-ferm tube (Hoffmann-LaRoche)
and/ or API 2ONE systems (bioMerieux Vitek) were used for reference identification. If identification to species level
was discrepant, standard methods at the Dutch National Institute of Public Health and Environmental Hygiene were
used. The Vitek correctly identified 96% of the Enterobacteriaceae and 95% of the nonfermentative gram-negative
bacilli. Misidentification with a high probability occurred with one Citrobacter freundii strain identified as
Enterobacter cloacae. Three Salmonella strains were reported as unidentified. One Pseudomonas aeruginosa was
misidentified, and one Xanthomonas maltophilia was correctly identified but with low probability. Vitek produced final
identification results at 4 hr for less than half of the strains and had identified about two-thirds of the strains within 6
hr.

O'Hara et al. (1993) compared the Vitek GNI versus conventional biochemicals for identification of biochemically
typical and atypical members of the family Enterobacteriaceae (N = 229) and common non-glucose-fermenting gram-
negative bacilli (N = 23). All strains were included in the Vitek data base. At the end of the initial incubation, 88.6% of
the Enterobacteriaceae and 100% of the non-glucose fermenters were correct to the genus and species level. After
additional biochemical tests, as required by Vitek's protocol, 92.1% of the Enterobacteriaceae were correct to genus
and species level. Two atypical Salmonella enteritidis strains were misidentified as an Escherichia coli and a C.
freundii. A Shigella flexneri was misidentified as an E. coli, and a Shigella sonnei was unidentified. One Salmonella
arizonae, Salmonella cholerae-suis, and atypical S. enteritidis were each unidentified.

Simoons-Smit and Maclaren (1994) tested 297 clinical isolates of Enterobacteriaceae and 62 nontermentative Gram-
negative bacilli using the Vitek GNI and conventional biochemical tests. Additional biochemical tests were not used
with the Vitek GNI. Vitek identified 95% of the Enterobacteriaceae and 82.5% of the nonfermentative gram-negative
bacilli. Most of the strains

 Page 95

not identified were with Acinetobacter spp., Pseudomonas spp., or X. maltophilia (15 strains), which were recorded as
unidentified or nonviable. Only seven strains were misidentified.

Robinson et al. (1995) reported on the accuracy of the Vitek GNI for the identification of gram-negative bacilli,
including 381 members of the family Enterobacteriaceae and 131 nonenteric bacilli. The Vitek identified 92.4 and
96.1% of the Enterobacteriaceae following 10 and 18 hr of incubation, respectively, without supplemental tests.
Similarly, the Vitek identified 84.0 and 93.9% of the nonenteric bacilli following 10 and 18 hr of incubation. With
supplemental testing, Vitek identified 97.9% of the Enterobacteriaceae and 97.7% of the nonenteric bacilli.
Supplemental tests were required for 1% of the isolates. Errors in identification were randomly distributed.

Rhoads et al. (1995) compared the Vitek GNI with the MicroScan WalkAway system. Discordant results were
arbitrated with the API 20E identification system. For urinary isolates, 493 samples were tested. Sixteen genera and
species were tested, with E. coli, Klebsiella pneumoniae, and Proteus mirabilis accounting for 63.1, 16, and 6.9% of
the isolates tested, respectively. One isolate was misidentified at the genus level, and one isolate was not identified by
the Vitek system. From body sites other than urine, there were 331 isolates tested, representing 21 genera and species.
Acinetobacter baumanii, E. cloacae, E. coli, K. pneumoniae, and P. aeruginosa accounted for 21.8, 9.7, 19.9, 16, and
11.8% of the isolates tested, respectively. The Vitek system misidentified nine isolates at the genus level and one isolate
at the species level and did not identify eight isolates. Vitek correctly identified 98.4% of the Enterobacteriaceae and
94.4% of the nonenteric bacilli.

Kiska et al. (1996) compared the Vitek GNI with conventional biochemicals for 150 nonfermenters including 58
isolates of Burkholderia cepacia recovered from respiratory secretions from cystic fibrosis patients. Sixty percent of
the isolates were correctly identified by the Vitek system. Only 50% of the B. cepacia strains were correctly identified.
Seven B. cepacia isolates were incorrectly identified as Burkholderia pickettii, while 10 isolates were not identified.
The addition of lysine decarboxylase, ONPG, and cetrimide growth tests improved the accuracy of the Vitek system to
71% for B. cepacia. The Vitek system correctly identified only 13 of 24 Alcaligenes xylosoxidans strains. Negative OF
glucose and xylose test results were primarily the cause of the misidentifications.
Vitek has a new version of the GNI card called the GNI+. The GNI+ can identify glucose-fermenting organisms in 28
hr and glucose-nonfermenting organisms in 412 hr. Twenty new species have been added to the data base. Moss et al.
(1996) evaluated the GNI+ with 3,179 strains of organisms from 167 species. The strains were obtained from clinical
(48.8%), stock (43.9%), industrial (4.5%), ATCC (1.9%), veterinary (0.6%), and miscellaneous (0.3%) sources. The
reference system for identification was not mentioned.

 Page 96

TABLE 7.2 Performance of Automated Identification Systems for Identification of Members of the Family Enterobacteriaceae
% of identifications that were:
System No. of organisms tested Correct Inconclusive Not made Incorrect Ref.
Vitek GNI 207 94a 2b 1.5 2.5 Visser, 1992
Vitek GNI 229 92.1c 4.8 3.1 O'Hara, 1993
Vitek GNI 297 95d 2.7 0.3 2.0 Simoons-Smit, 1994
Vitek GNI 381 97.9e 1.6 0.5 Robinson, 1995
Vitek GNI 700 98.4f 0.6 1.0 Rhoads, 1995
Sensititre AP80 879 92.5g 0.3h 0.5 6.7 Staneck, 1993
Sensititre AP80-VET 108 91.7i Patten, 1995
WalkAway-96 R-GNB 207 93f 5j 2 Visser, 1992
WalkAway-96 R-GNB 374 90.1d Kelly, 1992
WalkAway-96 R-GNB 364 95.6k 4.4 York, 1992
WalkAway-96 R-GNB 467 91.9c 8.1 O'Hara, 1992
WalkAway-96 R-GNB 229 96.5c 3.5 O'Hara, 1993
WalkAway-96 GNB 700 96.9f 0.7 2.4 Rhoads, 1995
Biolog 511 77l 12 11 Holmes, 1994

(table continued on next page)

 Page 97

(table continued from previous page)

aPercent correct identification with first-choice likelihood ³ 90%.
bPercent correct identification with low probability (< 90%).
cPercent correct identification with required supplemental tests.
dIdentification likelihood not stated. Supplemental tests not performed.
ePercent correct identification with first-choice likelihood ³ 90%. Supplemental tests performed as required
fPercent correct identification with first-choice likelihood ³ 85%. Supplemental tests not performed.
gListed as ''acceptable identification" or better. Supplemental tests performed as required.
hListed as "good likelihood but low selectivity."
iIdentification likelihood not stated. Supplemental tests performed as required.
jPercent correct identification with low probability (< 85%). Supplemental tests not performed.
kPercent correct identification with first-choice likelihood ³ 85%. Supplemental tests performed as required
lPercent correct identification at 24 hr with a similarity index ³ 0.500. Supplemental tests not performed.

 Page 98

TABLE 7.3 Performance of Automated Identification Systems for Identification of Strains of Gram-negative Organisms that were not Members of the Family Enterobacteriaceae
% of identifications that were:
System No. of organisms tested Correct Inconclusive Not made Incorrect Ref.
Vitek GNI 41 95a 2b 3 Visser, 1992
Vitek GNI 23 100c O'Hara, 1993
Vitek GNI 81 82.5d 17.3 1.2 Simoons-Smit, 1994
Vitek GNI 131 97.7e 2.3 Robinson, 1995
Vitek GNI 124 94.4f 3.2 2.4 Rhoads, 1995
Vitek GNI 141g 60h 22i 15 20 Kiska, 1996
Sensititre AP80 144 84.7j 1.4k 0.7 13.2 Staneck, 1993
Sensititre AP80-VET 162 72.8l Patten, 1995
WalkAway-96 R-GNB 41 90f 7m 3 Visser, 1992
WalkAway-96 R-GNB 121 93.4d Kelly, 1992
WalkAway-96 R-GNB 36 100n York, 1992
WalkAway-96 R-GNB 23 100c O'Hara, 1993
WalkAway-96 GNB 124 92.8f 4.0 3.2 Rhoads, 1995
Biolog 278 56.5o 28.0 15.5 Holmes, 1994

(table continued on next page)

 Page 99

(table continued from previous page)

aPercent correct identification with first-choice likelihood ³ 90%.
bPercent correct identification with low probability (< 90%).
cPercent correct identification with required supplemental tests.
dIdentification likelihood not stated. Supplemental tests not performed.
ePercent correct identification with first-choice likelihood ³ 90%. Supplemental tests performed as required.
fPercent correct identification with first-choice likelihood ³ 85%. Supplemental tests not performed.
gFifiy-eight isolates were B. cepacia.
hPercent correct identification with first-choice likelihood ³ 90%. Supplemental tests not performed.
iLow level of discrimination; the correct identification was listed among two or more choices.
jListed as "acceptable identification" or better. Supplemental tests performed as required.
kListed as "good likelihood but low selectivity."
lIdentification likelihood not stated. Supplemental tests performed as required.
mPercent correct identification with low probability (<85%). Supplemental tests not performed.
nPercent correct identification with first-choice likelihood ³ 85%. Supplemental tests performed as required.
oPercent correct identification at 24 hr with a similarity index ³ 0.500. Supplemental tests not performed.

 Page 100

GNI+ correlations were 96.4% correct, 2.6% incorrect, and 0.9% unidentified. The GNI+ identified 40.1% of the
organisms in 3 hr versus 19.4% for the GNI.

Refsahl and Andersen (1992) evaluated the Vitek Gram Positive Identification Card (GPI) for identification of 131
coagulase-negative staphylococci. The staphylococci were isolated from blood cultures (N = 68), vascular catheters (N
= 6), osteomyelitis foci (N = 13), postoperative and other wounds (N = 15), and urine specimens (N = 29). The isolates
consisted of Staphylococcus epidermidis (N = 96), Staphylococcus hominis (N = 9), Staphylococcus saprophyticus (N
= 8), Staphylococcus haemolyticus (N = 7), Staphylococcus warneri (N = 6), Staphylococcus simulans (N = 3),
Staphylococcus capitis (N = 1), and Staphylococcus auricularis (N = 1). The Vitek system correctly identified 96.9%
of the S. epidermidis and 94.7% of all isolates tested to the species level.

Bannerman et al. (1993) compared the Vitek GPI with conventional biochemicals with 500 sequential clinical isolates
of coagulase-negative Staphylococcus. The isolates included S. epidermidis (N = 322), S. haemolyticus (N = 68), S.
hominis (N = 37), S. warneri (N = 20), Staphylococcus lugdunensis (N = 14), S. simulans (N = 12), S. capitis subsp.
ureolyticus (N = 10), S. capitis subsp. capitis (N = 8), S. saprophyticus (N = 5), Staphylococcus cohnii subsp.
urealyticum (N = 2), S. cohnii subsp. cohnii (N = 1), and S. auricularis (N = 1). The species identified with the greatest
accuracy were S. epidermidis (92%), S. haemolyticus (95%), S. capitis subsp. capitis (88%), and S. saprophyticus
(100%). S. hominis (63%) was identified with the least accuracy. The overall agreement for those species included the
Vitek data base was 89%. For all strains tested, 77.2% were correctly identified to species and an additional 8.8% were
correctly identified but with low probability.

Perl et al. (1994) evaluated the Vitek GPI, testing 94 blood cultures isolates of S. epidermidis and 183 other coagulase-
negative "non-epidermidis" species. All organisms were identified with conventional biochemicals. The Vitek correctly
identified 64% of the S. epidermidis. Nineteen of 34 S. epidermidis not correctly identified were listed as unidentified.
The Vitek correctly identified 68% of the remaining species, but only S. saprophyticus were identified with a
probability of >90%. The Vitek correctly identified 66.8% of all isolates tested.
Buschelman et al. (1993) evaluated the Vitek GPI with 208 enterococcal nosocomial isolates. Seventy-five percent
were Enterococcus faecalis, and 20% were Enterococcus faecium. A modified conventional test method served for the
reference identification. Of those species of enterococci in the Vitek data base, 196 of 199 (98.5%) were correctly
identified. Twelve organisms (5.8% overall) were misidentified. A Vitek veterinarian software program correctly
identified three of four Enterococcus casseliflavus and two strains

 Page 101

TABLE 7.4 Performance of Automated Identification Systems for Identification of Staphylococcus Species
No. of organisms tested % of identifications that were:
System Correct Inconclusive Not made Incorrect Ref.
Vitek GPI 131 94.7a 0.9 4.6 Refsahl, 1992
Vitek GPI 500 77.2 8.8b 5.0 9.0 Bannerman, 1993
Vitek GPI 277 66.8a 14.8 18.4 Perl, 1994
WalkAway-96 GPB 233 53.6c 24.9d 6.4 15.0 Grant, 1994
WalkAway-96 R-GPB 233 50.6c 24.9d 5.2 19.3 Grant, 1994
aIdentification likelihood not stated.
bAn acceptable likelihood for identification but biochemical pattern closely resembles two species.
cCorrect species was first option and identification probability was ³85%.
dCorrect species was one of the choices, and none of the species listed achieved an identification probability of ³85%.

 Page 102

each of Enterococcus hirae and Enterococcus gallinarum. Four strains (1.9%) were still incorrectly identified.

Sader et al. (1995) compared the Vitek GPI with conventional biochemicals for the identification of 369 nosocomial
enterococcal strains derived from 97 medical centers. Only seven isolates (three E. faecalis, one E. faecium, two
Enterococcus raffinosus, and one Enterococcus solitarius) were misidentified. The latter two species were not in the
Vitek data base. The Vitek system correctly identified 97.6% of the isolates.

De la Higuera et al. (1995) tested 160 strains of Streptococcus mutans obtained from clinical oral isolates with the
Vitek GPI. The strains had been identified with conventional biochemicals. Vitek correctly identified 116 of the 160
(72.5%) strains. Of the remaining 44, 45.5% were identified as Streptococcus bovis, 9.1% as Streptococcus
intermedius, 9.1% as Streptococcus anginosus. 4.5% as Streptococcus sanguis I, 9.1% as Streptococcus sanguis II,
13.6% as Streptococcus uberis, and 9.1% as Streptococcus constellatus. The greatest discrepancy was found in the
inulin fermentation test, which by Vitek was positive in only 32 of 116 (27.6%) strains.
Wan-qing et al. (1992) evaluated the Vitek Yeast Biochemical Card (YBC) with 50 fresh clinical isolates and 77
stocked clinical isolates. A conventional method served as a reference identification. The Vitek YBC correctly
identified 96.1% of the isolates. One hundred twenty-three (96.9%) were identi-
TABLE 7.5 Performance of Automated Identification Systems for Identification of Yeasts and Yeastlike Organisms
System No. of organisms tested % Correctly identified Ref.
Vitek YBC 127 96.1a Wan-qing, 1992
Vitek YBC 222 76.1b Dooley, 1994
Vitek YBC 409c 89.7b Fenn, 1994
Vitek YBC 150 95d Riddle, 1994
WalkAway-96 YIP 150 82d Riddle, 1995
aIdentification likelihood not stated; requirements for microscopic morphological examinations or supplemental physiological
tests not stated.
bPercent correct identification with required microscopic morphological examinations or supplemental physiological tests.
cGerm tube-negative yeasts were evaluated.
dPercent correct identification with first-choice likelihood ³ 85%. Microscopic morphological examinations routinely performed
and supplemental physiological tests performed when required.

 Page 103

fied in 24 hr and four (3.1%) in 48 hr by the Vitek system. One isolate each of Candida albicans, Candida
parapsilosis, Candida stellatoidea, and Geotrichum were misidentified, and one isolate of Rhodotorula rubra was
unidentified.

Dooley et al. (1994) evaluated the Vitek YBC for identification of 222 germ tube-negative yeasts isolated from clinical
specimens. The results were compared with the API 20C (bioMerieux Vitek) and morphology, with additional
biochemical reactions performed as required. One hundred and fifteen of 123 (93%) commonly isolated yeasts were
correctly identified, with only C. albicans, Candida tropicalis, and Cryptococcus neoformans mis- or unidentified more
than once. However, only 54 of 99 (55%) less commonly isolated yeasts included in the YBC. data base were correctly
identified. The YBC did not identify 10 of 24 (42%) Candida krusei, 4 of 5 (80%) Candida lambica, 7 of 8 (88%)
Trichosporon beigelii, or 10 of 12 (83%) Cryptococcus isolates (non-C. neoformans species). Forty-two of 53 (79%)
identification failures gave no identification by the end of 48 hr; the remaining identification failures gave definite but
incorrect identifications. The Vitek correctly identified 76.1% of all isolates tested.

Fenn et al. (1994) evaluated the Vitek YBC using 409 germ tube-negative yeasts that represented nine genera and 21
species. The API 20C and morphology agars were the reference standards. The Vitek YBC identified 367 (89.7%)
correctly, with 77.4% being identified in 24 hr. The Vitek YBC misidentified 30 isolates, while 14 isolates yielded no
identification. Morphology agars were required for identification with 50 isolates when the Vitek YBC was used. Vitek
had some problems in the identification of C. tropicalis, C. krusei, Trichosporon spp., and some Cryptococcus spp.

Riddle et al. (1994) reported on the accuracy of the Vitek YBC using 150 clinical yeast isolates [30 C. albicans, 67
Candida spp. (not albicans), 26 Torulopsis glabrata, 13 C. neoformans, four Saccharomyces cerevisiae, six T. beigelii,
three Rhodotorula spp., and one Geotrichum spp.]. The API 20C and morphology agar served as the reference
identification. On initial testing 128 (85%) isolates were correctly identified by Vitek. After repeat testing 142 (95%)
were correctly identified. Yeasts most commonly misidentified by Vitek were C. tropicalis and T. glabrata.

Sensititre

Staneck et al. (1993) reported on the performance of the Sensititre AP80 panel (AccuMed International, Inc., Westlake,
OH) using 879 members of the family Enterobacteriaceae and 144 nonenteric organisms. The isolates consisted of 20
genera and 40 species. The API 20E or Rapid NFF (bio-Merieux Vitek) served as the reference identification.
Conventional bio-

 Page 104

chemical tests were used to arbitrate isolates showing discrepancies. Among the members of the family
Enterobacteriaceae, 93% of the results were final at 5 hr, and among the nonenteric bacilli, 69% of the results were
final at 5 hr. The organisms that most frequently required 18 hr of incubation were Acinetobacter spp., S. sonnei, and X.
maltophilia. Agreements with reference results for members of the family Enterobacteriaceae were 95.3 and 92.5% at
the genus and species level, respectively, and for the nonenteric organisms, the agreements with reference results were
95.1 and 84.7%, respectively. Among Enterobacteriaceae misidentified at the genus level, C. freundii, E. cloacae, and
E. coli made up over half of the isolates, perhaps because they were among the most frequently tested isolates. Genus-
level errors occurred for six nonenteric organisms. A total of 62 supplemental tests were required for 38 of all isolates
(12 pigmentation, 15 motility, 10 oxidase, and 25 indole tests).

Sensititre has recently obtained FDA approval for an identification panel for gram-negative bacilli (AP80-VET) and
gram-positive bacteria (AP90-VET) of veterinary importance.

Patten et al. (1995) evaluated the AP80-VET for identification of 270 clinical veterinary isolates, representing 23
genera and 40 species. Conventional biochemical methods served as a reference identification. Of the
Enterobacteriaceae tested (N = 108), Sensititre correctly identified to species 69.4% at 5 hr and 91.7% at 18 hr. Of the
remaining isolates tested (N = 162), Sensititre correctly identified to species 47.5% at 5 hr and 72.8% at 18 hr.

Alamar

Alamar (AccuMed International, Inc., Westlake, OH) is a newly introduced system and will be described here in detail.
Alamar has a microtiter panel with 168 wells. Susceptibility panels are available for both gram-negative and gram-
positive bacteria. Results are interpreted at 1820 hr. Alamar also has a gram-negative bacilli identification panel with
24 substrates. The substrates include 15 conventional substrates, 8 inhibitory compounds, and a positive control. There
must be manual entry of the oxidase, Gram stain, and catalase reaction into the computer. The Alamar identification
panel data base includes information for the identification of 33 members of the family Enterobacteriaceae and 11
members of other gram-negative bacilli organisms. This identification panel can also identify six members of gram-
positive cocci but does not have FDA approval. The inoculum for the identification panel is prepared from selective or
nonselective agar media. Several colonies are suspended in 6 mL of saline and adjusted to the

 Page 105

equivalent of a 0.5 McFarland standard. Twenty-five microliters is inoculated into 25 mL of Alamar Broth in the
automated panel inoculator (PIPETar). One hundred microliters is automatically dispensed into each microwell of the
identification panel. Two drops of mineral oil are added to the arginine, lysine, and ornithine wells, the panel is
covered, and then incubated 1820 hr at 35°C in a non-CO2 incubator. The substrates are incorporated into cellulose
disks, which also contain an oxidation-reduction indicator that changes color in response to chemical reduction of
growth medium resulting from bacterial growth. Reduction related to growth causes the indicator to change from
oxidized (blue, nonfluorescent) to reduced (red, fluorescent). The panels can be read visually. The reduced indicator
also fluoresces and can be detected in the READar fluorometer instrument. The READar scan-head contains eight
light-emitting diode-photodetector pairs (one pair for each row) to illuminate the solution and to measure the resulting
fluorescence. It requires approximately 20 sec to read a panel. Fluorescence intensity data is transmitted to a 486-33
MHz computer. An algorithm is used to generate an identification, which is printed. A data management center is also
available.

Killian et al. (1996) evaluated the Alamar identification panel with 373 members of the family Enterobacteriaceae and
175 gram-negative bacilli that were non-Enterobacteriaceae. The reference system for identification was not
mentioned. Alamar correctly identified 96.5% of the Enterobacteriaceae to the species level and 86.9% of the non-
Enterobacteriaceae to the species level. To determine the percent agreement between readings obtained by the
automated reader and visually, 10 organisms were tested for each substrate. The agreement between visual and
instrument readings was 99.0%.

Walkaway

Visser et al. (1992) reported on the accuracy of the WalkAway-96 Rapid Gram Negative Bacilli panel (R-GNB) using
207 members of the Enterobacteriaceae and 41 nonfermentative gram-negative bacilli that had been maintained frozen
at -70°C. The Enterotube II system and Oxi-ferm tube and/or API 2ONE systems were used for reference
identification. If identification to species level was discrepant, standard methods at the Dutch National Institute of
Public Health and Environmental Hygiene were used. The WalkAway correctly identified 93% of the
Enterobactcriaceae and 90% of the nonfermentative gram-negative bacilli in 2 hr. Misidentification with a high
probability occurred with one C. freundii identified as E. cloacae. A Shigella dysenteriae was identified as a
Salmonella paratyphi A, a C. freundii as E. cloacae,

 Page 106

TABLE 7.6 Studies Comparing Automated Identification Systems

System
(% identification accuracy)
No. of organisms tested Vitek WalkAway-96 p-value Ref.
207a GNI (94) R-GNB (93) > 0.50 Visser, 1992
41b GNI (95) R-GNB (90) > 0.30 Visser, 1992
229a GNI (92.1) R-GNB (96.5) < 0.05 O'Hara, 1993
23b GNI (100) R-GNB (100) NDc O'Hara, 1993
700a GNI (98.4) R-GNB (96.9) > 0.05 Rhoads, 1995
124b GNI (94.4) R-GNB (92.8) > 0.50 Rhoads, 1995
150 YBC (95) YIP (82) < 0.001 Riddle, 1994
aMembers of the family Enterobacteriaceae.
bGram-negative organisms that were not members of the Enterobacteriaceae.
cNot done.

and a Serratia liquefaciens as Serratia marcescens. WalkAway misidentified an Acinetobacter calcoacetius as X.

maltophilia.

Kelly and Leicester (1992) tested the WalkAway-96 R-GNB using 374 members of the Enterobacteriaceae and 121
nonfermentative gram-negative bacilli. Cathra Repliscan (Cathra, St. Paul, MN) and conventional biochemical methods
served as the reference identification. WalkAway correctly identified 90.1% of the Enterobacteriaceae and 93.4% of the
nonfermentative gram-negative bacilli. The WalkAway identified C. freundii, Mor-ganella spp., Proteus vulgaris,
Shigella spp., and Vibrio parahaemolyticus with less than 80% accuracy.

York et al. (1992) evaluated the WalkAway-96 R-GNB for 364 members of the Enterobacteriaceae and 36 Aeromonas
isolates, by using the API 20E and/or tube biochemical tests as the reference identification system. The WalkAway
correctly identified 95.6% of the Enterobacteriaceae and 100% of the Aeromonas isolates in 2 hr. Of the 16
misidentifications, S. liquefaciens accounted for 6. Further tests to complete the identification was required for 63
isolates (17%). Approximately half (N = 32) of the additional tests were required to separate Citrobacter diversus from
Citrobacter amalonaticus.
O'Hara and Miller (1992) evaluated the WalkAway-96 R-GNB for identification of 467 biochemically typical and
atypical stock cultures of Enterobacteriaceae. The organisms had been identified by standard conventional methods. At
2 hr, 353 of 467 (75.6%) strains were correctly identified to

 Page 107

genus and species. After performing recommended biochemical tests, another 76 (16.3%) strains were correctly
identified to genus and species. Twenty-two isolates (4.7%) were correct at the genus level but incorrect at the species
level, and 16 isolates (3.4%) were misidentified. Of these 16 isolates, 9 were incorrect at 2 hr and 7 were incorrect after
additional biochemical testing.

O'Hara et al. (1993) compared the WalkAway-96 R-GNB versus conventional biochemicals for identification of
biochemically typical and atypical members of the family Enterobacteriaceae (N = 229) and common nonglucose-
fermenting gram-negative bacilli (N = 23). All strains were included in the WalkAway data base. At the end of the
initial incubation, 82.1% of the Enterobacteriaceae and 100% of the non-glucose fermenters were correct to the genus
and species level. Alter additional biochemical tests, as required by WalkAway's protocol, 96.5% of the
Enterobacteriaceae were correct to the genus and species level. One atypical S. enteritidis strain was misidentified as an
Enterobacter gergoviae.

McGregor et al. (1995) evaluated the WalkAway-96 R-GNB with 64 Enterobacteriaceae, 17 pseudomonads, 8
Vibrionaceae and 11 miscellaneous gram-negative bacilli. They also tested the Rapid Gram Positive Bacteria panel (R-
GPB) with 51 streptococci, 28 staphylococci, 15 enterococci, and 6 miscellaneous gram-positive organisms.
Conventional methods served for the reference identification. The R-GNB correctly identified 81% to species and 93%
to genus; 6% were misidentified and 1% were not identified. Nine isolates required a total of 12 additional tests to
complete organism identification. The R-GPB correctly identified 90% to species; 7% were in probable agreement, 2%
were misidentified, and 1% were not identified. The two organisms that were misidentiffed were a Streptococcus
pneumoniae (S. sanguis) and a S. sanguis (S. bovis). A Streptococcus milleri was not identified by WalkAway, although
the organism was included in the data base. Two additional tests were required to complete identifications: a serotype
for Streptococcus pyogenes and a coagulase test.

Rhoads et al. (1995) compared the WalkAway-96 GNB and Vitek GNI. Discordant results were arbitrated with the API
20E. For urinary isolates, 493 samples were tested. Sixteen genera and species were tested, with E. coli, K.
pneumoniae, and P. mirabilis accounting for 63.1, 16, and 6.9% of the isolates tested, respectively. The WalkAway
misidentified six isolates at the genus level and one at the species level. Seven isolates were identified as ''slow
growers" after 18 hr of incubation and required 42 hr of incubation. These isolates were correctly identified as A.
baumanii (N = 6) and A. xylasoxidans (N = 1). The WalkAway gave no identification for four isolates. From body sites
other than urine, there were 331 isolates tested, representing 21 genera and species. A. baumanii, E. cloacae, E. coli, K.
pneumoniae, and P. aeruginosa

 Page 108

accounted for 21.8, 9.7, 19.9, 16, and 11.8% of the isolates, respectively. Nine isolates were misidentified at the genus
level and five at the species level. Six isolates were not identified. Eighty-two isolates (24.8%) required greater than 18
hr of incubation. Sixty-one (74.4%) were A. baumanii, nine were P. aeruginosa, four were X. maltophilia, and three
were Alcaligenes spp. Four isolates were not identified after additional incubation. The WalkAway correctly identified
96.9% of the Enterobacteriaceae and 92.8% of the nonenteric bacilli.

Grant et al. (1994) compared the WalkAway-96 GPB and R-GPB with 233 strains of coagulase-negative staphylococci,
representing 14 species. A modified conventional method served as the reference. The GPB correctly identified 53.6%
of the isolates, 24.9% were inconclusive, 6.4% were not identified, and 15% were incorrectly identified. Of the 35
incorrect identifications, S. hominis accounted for 14 (40%) and S. warneri 7 (20%). Twelve of the S. hominis strains
were incorrectly identified as S. warneri and five of the S. warneri were misidentified as S. hominis. The R-GPB
correctly identified 50.6% of the isolates, 24.9% were inconclusive, 5.2% were not identified, and 19.3% were
incorrectly identified. Sixteen (35.5%) of the 45 incorrect identifications involved S. hominis, eight (17.8%) involved
S. warneri, and five (14.3%) involved S. cohnii. Eleven of the S. hominis were misidentified as S. warneri.

Visser et al. (1992) evaluated the WalkAway-96 R-GPB using 31 Staphylococcus aureus and 12 E. faecalis. Reference
identification used conventional biochemical methods. One hundred percent of the isolates were identified to the
species level in 2 hr.

Riddle et al. (1994) evaluated the WalkAway-96 Yeast Identification Panel (YIP) using 150 clinical yeast isolates [30
C. albicans, 67 Candida spp. (not albicans), 26 T. glabrata, 13 C. neoformans, four S. cerevisiae, six T. beigelii, three
Rhodotorula spp., and one Geotrichum spp.] The API 20C and morphology agar served as the reference identification.
On initial testing, 101 (67%) isolates were correctly identified by WalkAway. After repeat testing, 123 (82%) were
correctly identified. The yeasts most commonly misidentified by WalkAway were C. tropicalis, T. glabrata, and C.
parapsilosis.

Biolog

Holmes et al. (1994) evaluated the Biolog system (Biolog, Inc., Hayward, CA) with 789 strains commonly encountered
in the clinical laboratory. The taxa comprised 35 Enterobacteriaceae (511 strains), 15 nonfermenters (214 strains), and
five oxidase-positive fermenters (64 strains). The strains had

 Page 109

been previously identified with conventional tests. The Biolog panels were read at 4 and 24 hr with the MicroPlate
reader and at 24 hr visually. Plates read by the automated reader at 24 hr gave the following performances:
Enterobacteriaceae, 77% correct, 11% incorrect, and 12% not identified; oxidase-positive fermenters, 89% correct, 7%
incorrect, and 4% not identified; biochemically active nonfermenters, 79% correct, 8% incorrect, and 13% not
identified; and biochemically inactive nonfermenters, 22% correct, 39% incorrect, and 39% not identified. The results
were poor at 4 hr with the automated reader and visual results at 24 hr were somewhat better than with the automated
reader at 24 hr.

Miller et al. (1993) evaluated the Biolog system for identification of members of the family Micrococcaceae. They
tested 113 isolates, which included 33 type strains of coagulase-positive and coagulase-negative staphylococci, 5
strains of Micrococcus, and one strain of Stomatococcus mucilaginosus. All strains analyzed were in the Biolog data
base. All isolates had been previously identified by the CDC reference method. The strains were blind coded and tested
at each of two laboratories with readings obtained at 24 hr on the automatic plate reader. Overall correct identification
to the species level upon initial testing were 47.7 and 59.3%, respectively, at the two laboratories. After repeat testing
of isolates that were reported as correct at the genus level only, incorrect, or unidentified, the overall accuracies
increased to 69.0 and 74.3%, respectively, at the two laboratories. Error rates were 7.1 and 9.7%, respectively, at the
two sites. After repeat testing, there was no significant difference between the two laboratories (78 of 113 vs. 84 of
113; p > 0.05).

There have been a large number of publications on the Biolog identification system in the last several years. The
studies have involved environmental, industrial, food, and basic research applications.

MIDI

Kellogg et al. (1996) evaluated the MIDI Microbial Identification System (MIDI; Microbial ID, Inc., Newark, Del.)
with 363 isolates of non-Enterobacteriaceae gram-negative bacilli. Conventional tests were used for the reference
identification. MIDI, which chromatographically analyzes cellular fatty acids, identified 328 (90.4%), of which 327
(99.7%) and 253 (77.1%) were correctly named to the genus and species level, respectively. Nine isolates (one
Acinetobacter junii, two Acinetobacter radioresistens, one Capnocytophaga sputigena, one EO-2, and four
Roseomonas spp.) were not in the MIDI data base. Of these, six (67%) were correctly not identified and three

 Page 110

(all Acinetobacter spp.) were correct only to the genus level. Ten isolates of Pseudomonas putida were misidentified as
Pseudomonas fluorescens.

Tunér et al. (1992) compared the MIDI to conventional biochemical tests for the identification of 46 clinical isolates
and 6 ATCC strains of fusobacteria. MIDI correctly identified 89% of the isolates. Two of four clinical Fusbacterium
necrophorum strains were identified as Fusobacterium nucleatum. Two strains of F. nucleatum were identified as
Fusobacterium perfoetens.

Allen et al. (1995) evaluated the MIDI system for identification of 216 strains representing 18 species of Clostridium.
Traditional biochemical methods were used for reference identification. MIDI correctly identified 86% of the strains to
the species level without supplemental tests. An additional 6% of the strains were correct to the species level when a
few supplemental tests were performed. Only 3% of strains were incorrectly identified to the genus level.
As with Biolog, a number of additional publications with MIDI have demonstrated a range of applications for this
system.

Studies Comparing Automated Identification Systems

Since the spring of 1992, four studies have compared the identification accuracy of two automated identification
systems when testing the same isolates. The identification accuracy for each of the automated systems, the percentile
(p-value) of the chi-square distribution as determined by the chi-square test for each study, and the cited reference are
shown in Table 7.6. The details of how each automated system performed versus the reference system are found earlier
in this chapter where that particular automated system is reviewed.

Conclusion

It is difficult to conclude from the literature which automated identification system performs best for a particular
application. Among the many reasons for this is that the collection of microorganisms evaluated is never the same in
the various studies. Also, the manufacturer will often change the data base, alter thresholds on the reader, or modify,
add, or replace some of the substrates by the time a study is reported. Certainly, one should consider most seriously
studies that have evaluated or compared the latest software, data base, biochemical configuration, or other performance
characteristics of a particular system.

 Page 111

Acknowledgements

I would like to thank Melinda Sanchez for typing the manuscript.

References
Allen, S.D., Siders, J.A., Riddell, M.J., Fill, J.A., and Wegener, W.S. 1995. Cellular fatty acid analysis in the
differentiation of Clostridium species in the clinical microbiology laboratory. Clin. Infect. Dis. 20 (Suppl 2): S198201.

Bannerman, T.L., Kleeman, K.T., and Kloos, W.E. 1993. Evaluation of the Vitek Systems Gram-Positive Identification
Card for species identification of coagulase-negative staphylococci. J. Clin. Microbiol. 31 (5): 13221325.

Buschelman, B.J., Bale, M.J., and Jones, R.N. 1993. Species identification and determination of high-level
aminoglycoside resistance among enterococci. Diagn. Microbiol. Infect. Dis. 16: 119122.
De la Higuera, A., Liébana, J., Gutierrez, J., García-Mendoza, A., and Castillo, A. 1995. Evaluation of the
Automicrobic System for the identification of' Streptococcus mutans. Eur. J. Clin. Microbiol. Infect. Dis. 14:
11021105.

Dooley, D.P., Beckius, M.L., and Jeffery, B.S. 1994. Misidentification of clinical yeast isolates by using the updated
Vitek Yeast Biochemical Card. J. Clin. Microbiol. 32(12): 28892892.

Fenn, J.P., Segal, H., Barland, B., Denton, D., Whisenant, J., Chun, H., Christofferson, K., Hamilton, L., and Carroll,
K. 1994. Comparison of updated Vitek Yeast Biochemical Card and API 20C yeast identification systems. J. Clin.
Microbiol. 32(5): 11841187.

Grant, C.E., Sewell, D.L., Pfaller, M., Bumgardner, R.V., and Williams, J.A. 1994. Evaluation of two commercial
systems for identification of coagulase-negative staphylococci to species level. Diagn. Microbiol. Infect. Dis. 18: 15.

Holmes, B., Costas, M., Ganner, M., On, S.L.W., and Stevens, M. 1994. Evaluation of Biolog system for identification
of some gram-negative bacteria of clinical importance. J. Clin. Microbiol. 32(8): 19701975.
Kellogg, J.A., Bankert, D.A., Brennerman, T.M., Grove, M.A., Wetzel, S.L., and Young, K.S. 1996. Identification of
clinical isolates of non-Enterobacteriaceae gram-negative rods by computer-assisted gas-liquid chromatography. J.
Clin. Microbiol. 34(4): 10031006.

 Page 112

Kelly, M.T. and Leicester, C. 1992. Evaluation of the Autoscan Walkaway System for rapid identification and
susceptibility testing of gram-negative bacilli. J. Clin. Microbiol. 30(6): 15681571.

Kiska, D.L., Kerr, A., Jones, M.C., Caracciolo, J.A., Eskridge, B., Jordan, M., Miller, S., Hughes, D., King, N., and
Gilligan, P.H. 1996. Accuracy of four commercial systems for identification of Burkholderia cepacia and other gram-
negative nonfermenting bacilli recovered from patients with cystic fibrosis. J. Clin. Microbiol. 34(4): 886891.

McGregor, A., Schio, F., Beaton, S., Boulton, V., Perman, M., and Gilbert, G. 1995. The MicroScan WalkAway
diagnostic microbiology systeman evaluation. Pathology 27: 172176.
Miller, J.M., Biddle, J.W., Quenzer, V.K., and McLaughlin, J.C. 1993. Evaluation of Biolog for identification of
members of the family Micrococcaceae. J. Clin. Microbiol. 31(12): 31703173.

Moss, N.S., Wilder, D., Combs, D., Monroe, D., Mayer, J., and Morris, R. Evaluation of the Vitek GNI+ card, abstr. C-
389, p. 70. Abstr. 96th Annu. Meet. Am. Soc. Microbiol. 1996. American Society for Microbiology, Washington, D.C.
O'Hara, C.M. and Miller, J.M. 1992. Evaluation of the autoSCAN-W/A System for rapid (2-hour) identification of
members of the family Enterobacteriaceae. J. Clin. Microbiol. 30(6): 15411543.

O'Hara, C.M., Tenover, F.C., and Miller, J.M. 1993. Parallel comparison of accuracy of API 20E, Vitek GNI,
MicroScan Walk/Away Rapid ID, and Becton Dickinson Cobas Micro ID-E/NF for identification of members of the
family Enterobacteriaceae and common Gram-negative, non-glucose-fermenting bacilli. J. Clin. Microbiol. 31 (12):
31653169.

Patten, V.H., Shin, S.J., Cole, J., Watson, C.W., and Fales, W.H. 1995. Evaluation of a commercial automated system
and software for the identification of veterinary bacterial isolates. J. Vet. Diagn. Invest. 7(4): 506508.

Perl, T.M., Rhomberg, P.R., Bale, M.J., Fuchs, P.C., Jones, R.N., Koontz, F.P., and Pfaller, M.A. 1994. Comparison of
identification systems for Staphylococcus epidermidis and other coagulase-negative Staphylococcus species. Diagn.
Microbiol. Infect. Dis. 18: 151155.

Refsahl, K. and Andersen, B.M. 1992. Clinically significant coagulase-negative staphylococci: Identification and
resistance patterns. J. Hosp. Infect. 22: 1931.

Rhoads, S., Marinelli, L., Imperatrice, A., and Nachamkin, I. 1995. Comparison of MicroScan WalkAway System and
Vitek System for identification of gram-negative bacteria. J. Clin. Microbiol. 33(11): 30443046.

Riddle, D.L., Giger, O., Miller, L., Hall, G.S., and Woods, G.L. 1994. Clinical

 Page 113

comparison of the Baxter MicroScan Yeast Identification PanelTM and the Vitek Yeast Biochemical CardTM. Clin.
Microbiol. Infect. Dis. 101(4): 438442.

Robinson, A., McCarter, Y.S., and Tetreault, J. 1995. Comparison of Crystal Enteric/Nonfermenter System, API 20E
System, and Vitek Automicrobic System for identification of gram-negative bacilli. J. Clin. Microbiol. 33(2): 364370.

Sader, H.S., Biedenbach, D., and Jones, R.N. 1995. Evaluation of Vitek and API 20S for species identification of
enterococci. Diagn. Microbiol. Infect. Dis. 22: 315319.

Simoons-Smit, A.M. and Maclaren, D.M. 1994. Comparison of Vitek and Cobas Micro systems with a semiautomated
conventional microsystem for identification and susceptibility testing of gram negative bacilli. J. Clin. Pathol. 47:
7175.

Stager, C.E. and Davis, J.R. 1992. Automated systems for identification of microorganisms. Clin. Microbiol. Rev. 5(3):
302327.
Staneck, J.L., Wechbach, L.S., Tilton, R.C., Zabransky, R.J., Bayola-Mueller, L., O'Hara, C.M., and Miller, J.M. 1993.
Collaborative evaluation of the Radiometer Sensititre AP80 for identification of gram-negative bacilli. J. Clin.
Microbiol. 31 (5): 11791184.

Tunér, K., Baron, E.J., Summanen, P., and Finegold, S.M. 1992. Cellular fatty acids in Fusobacterium species as a tool
for identification. J. Clin. Microbiol. 30(12): 32253229.
Visser, M.R., Bogaards, L., Rozenberg-Arska, M., and Verhoef, J. 1992. Comparison of the autoSCAN-W/A and Vitek
Automicrobic Systems for identification and susceptibility testing of bacteria. Eur. J. Clin. Microbiol. Infect. Dis.
11(11): 979984.

Wan-qing, L., Zhi-qiang, Z., and Hu, Z. 1992. Research on rapid identification of automicrobiology system for
pathogenic yeast. Chin. Med. J. 105(4): 319321.

York, M.K., Brooks, G.F., and Fiss, E.H. 1992. Evaluation of the autoSCAN-W/A rapid system for identification and
susceptibility testing of gram-negative fermentative bacilli. J. Clin. Microbiol. 30(11): 29032910.

 Page 115

8
Applications of Gene Probes for the Detection of Foodborne Pathogens
Karen C. Jinneman and Walter E. Hill
Seafood Products Research Center
Seattle District Office, Office of Regulatory Affairs
U.S. Food and Drug Administration
Bothell, Washington

Introduction

Gene probes in a variety of hybridization formats have been used for the detection, identification, and characterization
of specific DNA segments. The versatility of gene probe analyses is demonstrated by the many clinical and
environmental applications that have been developed. The continuing development of gene probes and hybridization
applications for food analyses has the potential to improve the rate of identification and characterization of foodborne
pathogenic microorganisms. This is especially important because the Centers for Disease Control and Prevention report
that the etiological agent was confirmed in only 38% of the recognized outbreaks of
Reference to any commercial materials, equipment or processes are for illustrative purposes only and does not in any way constitute
approval, endorsement, or recommendation by the authors or the U.S. Food and Drug Administration.

 Page 116

foodborne disease in the United States (Bean et al., 1990), illustrating the challenge of detecting disease-causing
microbes in foods.

The application of gene probes to the analysis of foods has been hampered by a number of difficulties. Assay
sensitivity is dependent on sample preparation and is affected by low levels of target; it is critical to make the target
nucleic acids available and free of inhibiting substances. In addition, laboratory safety issues associated with the use of
radiolabeled probes limited the widespread implementation of gene probe hybridization assays. Current protocols to
address these issues such as the development of non-isotopically labeled probes followed by the production of
commercially available diagnostic kits for food analyses are making the use of gene probes more common. Many gene
probes are used to detect specific foodborne pathogens or virulence genes in cells from enrichment broths or colonies
on an agar plate. However, additional applications such as incorporation into polymerase chain reaction (PCR) analyses
to confirm the identity of an amplified product and improve PCR test sensitivity or for the characterization of specific
strains in probe-based subtyping methods such as ribotyping are also available.

The primary principle upon which gene probe tests are based is that the complementary sequences of single-stranded
nucleic acid molecules can reform the double helix structure under appropriate conditions. The target of gene probe
hybridization assays is a particular region of the DNA or RNA of an unknown organism, which is tested for the
presence of a specific genetic sequence usually associated with a defined genetic function. The gene probe is an
appropriately labeled nucleotide sequence, which can range in size from a few bases (1520) to several thousand. The
labels can vary widely and may be a reporter molecule, which can be directly detected (such as a radioactive moiety),
or a modifying group (such as biotin), which can be indirectly detected following the hybridization. The formation of a
double-stranded structure between the labeled nucleic acid probe and a complementary target sequence is an indication
that the sample contains an organism harboring that particular genetic element. A variety of hybridization formats and
conditions to effect accuracy (stringency) can be applied to optimize the specificity and sensitivity of the target-probe
hybridization, the removal of unhybridized probe molecules, and the detection of hybridized target-probe complexes.
The sensitivity is usually in the range of 104106 copies of the target gene.
The remaining portion of this chapter will discuss gene probe analysis mechanics and hybridization kinetics and how
these factors play a role in probe test development including target selection and type of probe used; a variety of
hybridization formats are described. The types of labels used on gene probes and various detection methods will be
discussed. Most of this

 Page 117

cluster is dedicated to a survey of gene probe methods along with some technological information needed to
understand these methods and specific applications for a number of foodborne pathogenic bacteria and viruses.

DNA Hybridization Principles

Gene probes are used for the detection of particular genes by the mechanism of DNA hybridization. The gene probes
are labeled so that they can be detected in order to determine whether they have formed DNA hybrids with the DNA of
the sample.

Mechanism of DNA Hybridization

DNA molecules usually exist as a double-stranded helix held together by hydrogen bonds between the nucleotides
adenine and thymine and/or guanine and cytosine. The strands can be separated (denatured) by increasing the pH or the
temperature. If individual strands contain complementary nucleotide sequences, they can form hydrogen-bonded pairs
and reform the double helix when the pH or the temperature is lowered. When the strands are from the same source,
this process is called reanealling (Hall and Spiegelman, 1961; Britten and Kohne, 1968). If the strands reforming the
double helix originate from different sources (such as different strains or even different genera), the resulting molecules
are called hybrids and the process is termed hybridization. When a gene probe (a labeled oligonucleotide or DNA
fragment whose function is usually known) is found stably associated with the target DNA, it is concluded that the
DNA in the sample possesses the same information encoded in the probe.

Hybridization Kinetics

The speed at which DNA hybridization reactions proceed depends on a large number of factors such as the size and
concentration of the gene probe and its target, the salt concentration, and the temperature. If the probe is in excess over
the target sequences, doubled-stranded probes exhibit similar hybridization and reanealling rates within the first 4
hours; at longer times a significant amount of probe reanealling occurs so that these molecules do not interact with the
target. However, when synthetic oligonucleotide probes are used, all the probe molecules are identical and, therefore,
not complementary; no probe strand reanealling occurs.
Meinkoth and Wahl (1984) estimated the time needed for half of the probe molecules to hybridize with an immobilized
target as t1/2 = ln2/kC, where k is a rate constant and C is the probe concentration. The rate

 Page 118

constant is a function of probe length, probe genetic complexity, temperature, ionic strength, solution viscosity, and pH.
For Na+ concentrations in the range 0.41.0 M and a hybridization temperature of 25°C below the Tm (see below),
Wetmur and Davidson (1968) determined the rate constant to be 3.5 × 105. Therefore, t1/2 for a 20-base pair probe at a
concentration of 4 ng/mL (about 6 × 10-10 molar) is roughly 4 hours. By comparison, the t1/2 for a 500-base pair probe at
the same molar concentration is about 21 hours (Keller and Manak, 1989).

Hybridization Reaction Factors

The accuracy of a hybridization reaction is hard to define, so several different terms need to be discussed.
Stringency

The factors that affect the stability of the double-stranded DNA hybrids determine the degree of stringency of the
hybridization reaction. The nucleotide sequences of two complementary strands do not have to form hydrogen-bonded
pairs between every A, T, G, or C with 100% accuracy. However, if the hybridization conditions are such that all pairs
must be correctly matched for the hybrid molecule to be stable in the double-stranded form (as might be the case with
oligonucleotide probes), the reaction is said to have high stringency. Such conditions might be appropriate if a DNA
hybridization assay requires that only exact sequence matches should be detected, as a false-positive result can be
caused by a closely related sequence. When the temperature is too high, even perfectly matched hybrids will have
difficulty remaining in the double-stranded state and false-negative reactions may be observed. When it is important to
detect a family of related sequences (such as detecting an entire genus or when performing a Southern blot as described
below), lower stringency conditions might be chosen.
Generally, stringency can be increased by increasing the temperature at which the hybridization is conducted or by
decreasing the salt concentration. Likewise, stringency can be decreased by lowering the temperature or increasing the
salt concentration.

Normally, a suitable hybridization temperature is determined empirically, but there are a number of algorithms for
estimating this temperature. The rate of hybridization is related to the temperature, but increasing the temperature too
much in an attempt to speed up the reaction will lead to a decrease in hybridization efficiency because even perfectly
matched hybrids will be unstable. Hybridization usually proceeds well at 25°C below the Tm (the ''melting"
temperature) for long gene probes. The Tm is the temperature where half of the hydrogen bonds have been broken so
that half of the molecules are single stranded; the Tm also depends in part on the fraction of

 Page 119

G-C base pairs, as these pairs are held together by three hydrogen bonds, while A-T pairs are held together by only
two. For a long DNA molecule (i.e., greater than about 500 base pairs) containing about 50% G-C base pairs, the Tm is
about 95°C. If the hybridization is to proceed for many hours at this temperature, the DNA can break down. By using
formamide (which helps increase stringency), equivalent stringencies can be obtained at much lower temperatures. For
example, in 0.75 M [Na+] and 50% formamide the molecule described above has a Tm of about 65°C so that the
hybridization can be carried out at about 40°C (Tm -25°C). For short DNA probes such as synthetic oligonucleotides,
other empirical equations have been proposed (Rychlik and Rhoads, 1989) and hybridizations are often conducted at
5°C, below the calculated Tm.

Specificity

The specificity of a gene probe depends in part on the uniqueness of the sequence of nucleotides that comprise the
target and the conditions under which the hybridization reaction is performed. Conserved regions of genes encoding
ribosomal RNA are widely distributed in nature and, therefore, would not make good targets where a high degree of
specificity is required. In other words, specificity refers to the ability of a gene probe system to exclude organisms not
among those to be detected. A test with high specificity would be expected to yield a small fraction of false-positive
results.

Sensitivity

In reference to gene probes, "sensitivity" can have two quite different meanings. The first is related to specificity, in
that it refers to the fraction of organisms that one expects to detect that are positive in the assay. Gene probe tests with
high sensitivity in this sense will be able to detect a high percentage of the target organisms; in other words, very few
false negatives will occur.

The other definition of sensitivity refers to the number of organisms (or copies of target DNA) needed to give a
positive result. A number of factors affect sensitivity including the amount of signal produced by a particular label. The
ability of the detection system to register a positive result, how many cells are assayed, the assay format, and the
number of target copies per cell will be discussed in following sections. Generally, test sensitivity is in the range of
104106 target copies.

Probe Test Development

During the development of gene probe tests, many decisions must be made. These include the target gene sequences to
be sought in the assay, the type of gene probe, the format, the label, and the detection system. The develop-

 Page 120

ment of the gene probe itself is often the most time-consuming, labor intensive, and costly part of the process.

Target Selection

The major criterion driving the selection of a suitable genetic target is the particular information for which the probe-
based test is being designed to provide. For example, if a particular taxonomic group of organisms is to be detected,
then it would be reasonable to select as a target those genes conserved throughout the group but that vary enough to be
distinguished from other groups that are evolutionarily closely related. The conserved regions of ribosomal RNA genes
have been used to this advantage (DeLong et al., 1989). The use of ribosomal RNAs themselves as targets has an
additional advantage because there are usually between 1,000 and 10,000 copies of these molecules per cell. This
feature can greatly increase test sensitivity in the sense of being able to detect fewer cells. Genes sufficiently unique to
a particular group of organisms such as coliform bacteria have also been targeted (Bej et al., 1990, 1991). Alternatively,
if one wishes to detect and identify pathogenic strains of a particular species, then it is important to select as targets
those genes that are major contributors to the strains ability to cause disease (Moseley et al., 1982; Yamamoto et al.,
1990).
It is important to stress that although gene probe tests can reveal the presence of specific nucleotide sequences, they
cannot fully demonstrate the ability of the organism to function as a pathogen. For example, the target gene may
contain a small mutation which renders the virulence determinant nonfunctional, or the cell may have a defect in the
expression of the gene product; in both cases such strains may not be able to cause disease. Until more confidence in
gene probe tests is gained by experience, the results of these tests are often confirmed by a method based on a different
scientific principle such as a cell or whole animal bioassay or perhaps by an immunological test.
Probe Types

"Natural" Probes

Often the exact nucleotide sequence of a target is not known so a synthetic oligonucleotide probe must be ruled out.
Generally, the target sequence will have been cloned into a suitable plasmid vector for propagation and further study. It
would be convenient if the entire plasmid, including the target insert, could be used as a probe. However, unwanted
cross-reactions due to hybridization between the vector sequences and the DNA of the organism in which the target
gene resides may yield false-positive results or at least generate an unacceptable level of background signal. Therefore,
the cloned target insert usually must be isolated from the

 Page 121

vector, often by restriction endonuclease cleavage and separation of the products by gel electrophoresis. The sought-
after fragment must then be recovered and purified from the gel itself (Hengen, 1994; Krowczynska et al., 1995). The
recent use of capillary electrophoresis might make this task simpler as fractions containing separated fragments can be
collected from the column (Karger et al., 1995). The methods used to label these cloned probes can be quite different
from those preferred in the labeling of synthetic oligonucleotides (see below).

Synthetic Oligonucleotides

The advent of DNA-sequencing methods (Sanger et al., 1992) coupled with the development of automated laboratory
techniques for the cyclic solid-state synthesis of polynucleotides (Anderson et al., 1995) has ushered in the era of
"synthetic" gene probes. This technology has brought the cost of custom synthetic oligonucleotides to a point where
these sequence-specific heteropolymers can be considered as just another laboratory reagent.
A large number of nucleotide sequences are now available in internationally accessible databases such as GenBank®.
Release 94, April 1996, contains 685,693 entries encompassing a total of 499,127,741 nucleotides, and the amount of
data is currently increasing at a rate of about 100% per year; 27,358 of the entries (52,716,800 nucleotides) are from
bacterial sequences, and 30,232 (32,687,772 nucleotides) are from viruses. It is possible to select and conduct the
testing of probe sequences "in cybero" by using a suite of specially designed software programs (Devereux et al.,
1984). Thus, common sequences can be easily avoided, and while this approach cannot guarantee that a particular
sequence is unique, it provides support. There are a number of programs (e.g., Rychlik and Rhoads, 1989) for
estimating optimum hybridization temperatures for oligonucleotides, but trial and error appears to yield satisfactory
results.
Gene Probe Labels

During the early use of gene probe tests, radioactive isotopes were usually incorporated as the reporter molecule. The
most common radioactive isotope employed for use in gene probe applications for food analyses is 32P. High specific
activity and the emission of high-energy ß-particles, which result in short autoradiographic exposure times, are the
primary advantages of 32P-labeled probes. The primary disadvantages of this isotope are the relatively short 14.3-day
half-life and the handling and disposal of radioactive material. Until recently it has been difficult for nonradiolabeled
probes to achieve the sensitivity available with isotopically labeled probes.

The appearance of gene probe hybridization assays that utilize non-radiolabeled probes has increased in recent years.
Factors that encouraged

 Page 122

the development of non-isotopic-labeled probes include (1) the elimination of licensing and waste-handling issues
associated with radiolabeled probes, (2) the long-term (12 years) stability of nonradioactively labeled probes, and (3)
the ability to label large (&x109g) amounts of DNA at once allowing for multiple tests to be conducted from the same
lot of labeled probe.

The main disadvantage of nonradiolabeled probes can be reduced sensitivity of the probe, which has been primarily
attributed to interference from endogenous substances such as the food itself. This has presented significant challenges
to the application of nonradiolabeled probes to food analyses. Foods are complex matrices, and foodborne pathogens
are often present in low numbers: this makes concentrating, extracting, and making the target nucleic acids available
for gene probe hybridizations difficult. Thus the choice of probe target, hybridization format, detection method, probe
label, and reporter molecule detection system all play critical and interdependent roles in the optimization and
application of gene probe-based hybridization analyses.

A variety of gene probe systems are available that use either direct or indirect detection methods. In the direct systems
gene probe labels are bifunctional conjugates, which are prepared by covalently linking the oligo- or polynucleotide
sequence directly with the signal-generating detector molecule. The advantage is that only a single reaction is required
for detection. However, the reporter molecule must be able to retain its activity during the hybridization procedure.
Indirect systems link a modifier group to the oligo- or polynucleotide probe sequence. A detection conjugate that
recognizes the probe modifier group is covalently linked to a signal-generating detector molecule and is used to detect
the hybridized target-probe complex.

The most common nonradioactive nucleic acid probe-labeling systems applied for the detection of foodborne
pathogens involve either biotin (Chollet and Kawashima, 1985; Saez-Llorens et al., 1989; Emond et al., 1993; Tsen et
al., 1991, Luk et al., 1994) or digoxigenin (Kessler, 1991; Schultz et al., 1994; Jackson, 1991; Kim et al., 1991a,b;
Jaton et al., 1992; Herman and DeRidder, 1992; Luk et al., 1994). Oligonucleotide probes have also been labeled with
horseradish peroxidase (Datta et al., 1993; Herman and DeRidder, 1992; Blais and Phillippe, 1993) or alkaline
phosphatase (Jablonski et al., 1986; Panda et al., 1990; Oberhelman et al., 1993, Wright et al., 1993) for use in
hybridization assays for the detection of foodborne pathogens.

The gene probe label and particular labeling procedure used depend on a number of variables. The length of the probe
to be labeled, whether it is a fairly long (1001000 bp) cloned fragment or a short (<50 bp) oligonucleotide probe, will
in part determine how the label molecules are incorporated

 Page 123

and the number that can be added to each probe molecule. The labeling procedure also depends on the nucleic acid
(DNA or RNA) being labeled. The hybridization format (e.g., Southern blot, dot or slot blot, colony lift analysis, liquid
format, or target capture) also affects the choice of the optimum gene probe label for the application. Optimization of
labeling procedures is important to generate probes with maximum signal intensity and reduced background.
Commercial kits and reagents provide the analyst with multiple enzymatic or chemical labeling options for the
incorporation of many different molecules for labeling gene probes. The scope of this section encompasses gene probe
labels and labeling methods most often used for food analyses. For a broader discussion of all gene probe labels refer
to Symons (1989), Keller and Manak (1989), or Kricka (1992).

Gene Probe-Labeling Techniques

Random Primed DNA Labeling: Polynucleotide probes 10010,000 bp in length may be labeled by this procedure. The
template DNA is purified, linearized, and denatured. Short (68 bp) random nucleotide segments, representing all
possible combinations of sequences, are annealed to the template DNA to produce short regions of double-stranded
DNA followed by single-stranded stretches. The Klenow fragment of DNA polymerase synthesizes a complementary
strand incorporating the available deoxyribonucleoside triphosphates, some of which are labeled with a modifying
group or reporter molecule (Feinberg and Vogelstein, 1983). Probes of indeterminate length can be generated; the size
depends on where the DNA synthesis is initiated along the length of the DNA template. After the reaction is
terminated, the labeled probe is precipitated from the reaction for use in hybridization procedures. This labeling
technique produces a probe with multiple reporter molecules and high specific activity. It is appropriate for use in many
different hybridization formats.

Nick Translation: Another enzymatic incorporation procedure for labeling polynucleotide probes of several hundred
base pairs is nick translation (Rigby et al., 1977). The double-stranded template DNA is first nicked with DNase. E.
coli DNA polymerase I is used to incorporate deoxyribonucleotides, which include labeled DNA precursors, to the 3'-
hydroxyl terminus created by the nick in the double-stranded DNA. The 5' to 3' exonucleolytic activity of the DNA
Polymerase I simultaneously removes nucleotides from the 5' edge of the nick, thus providing an important "path-
clearing" function for the continuing addition of nucleotides to the 3' terminus. It is important to use probe-labeling
conditions that favor the incorporation of the labeled dNTP over available unlabeled dNTPs to maximize the specific
activity of the probe (Symons, 1989).

 Page 124

3' End Oligonucleotide Labeling: A single dideoxynucleotide with modifying group or reporter molecule can be
coupled to the 3' end of each oligonucleotide by terminal transferase (Tu and Cohen, 1980). The single
dideoxynucleotide at the 3' end prevents additional extension of the probe.

3' End Tailing: Terminal transferase is used to add a tail of between 10 and 100 bases to the 3' end of a single-stranded
oligonucleotide when a mixture of labeled and unlabeled nucleotides is used in the labeling reaction (Bollum, 1974;
Roychoudhury et al., 1979).

5' End Labeling: Oligonucleotides can be labeled at the 5' end leaving the 3' end free to act as primer in amplification
reactions such as PCR. This method requires a free -NH2 group at the 5' end of the oligonucleotide, which can be
prepared by the appropriate programming of DNA synthesizers (Agrawal et al., 1986).
In Vitro Transcription of RNA: A RNA polymerase generates an RNA transcript and incorporates labeled nucleotides
into the probe (Melton et al., 1984; Morris et al., 1986; Kreig and Melton, 1987).

Labeling cDNA Probes: Heat-denatured RNA is used as a template; enzyme avian myoblastosis virus (AMV) reverse
transcriptase is used to synthesize a cDNA probe (Taylor et al., 1976).

Labeling DNA by PCR: Large quantities of labeled probe can be rapidly produced by one of two strategies. During a
standard PCR reaction a labeled oligonucleotide primer is incorporated into the amplification product, which is then
used as a gene probe (Kricka, 1992), or appropriately labeled dNTPs are incorporated into the amplified product, which
can be denatured and used as a probe in hybridization protocols (Lo et al., 1988; Seibl et al., 1990). DNA probes can be
prepared from an RNA template using PCR that has been preceded by reverse transcription.

Photoactivation: Long-wave UV illumination is a nonenzymatic method to covalently link a modifying group onto a
DNA or RNA probe at a rate of one modifying group for every 200400 bases. The most common photo-labeling
methods use an aryl azide compound coupled to digoxigenin (Muhlegger et al., 1990) or biotin (Forster et al., 1985).
Other applications use compounds such as psoralen-biotin, which intercalate into nucleic acids; when photoactivated
they are covalently linked with flanking bases (Sheldon et al., 1985).

Other Nonenzymatic Labeling: Both DNA and RNA probes in the single-stranded form can be labeled
nonenzymatically with active enzymes such as horseradish peroxidase (HRP) and alkaline phosphatase (AP). AP can
be covalently cross-linked to oligodeoxynucleotides via a linker arm with a terminal reactive primary amine (Jablonski
et al., 1986). Also, DNA can be chemically cross-linked with HRP by the addition of glutaraldehyde (Renz and Kurz,
1984).

 Page 125

Detection of Gene Probes

As with every step of a gene probe hybridization analysis, the detection part of the procedure depends on several
variables. The hybridization format used, the degree of sensitivity required for the application, and the type of label
incorporated into the probe all play a role in the type of detection method used. With respect to formats, when the target
is fixed, the labeled but unbound probe molecules must be washed away. When both target and probe are in solution,
more innovative approaches must be found. Direct and indirect detection method approaches are the two basic
detection strategies.

Direct detection protocols can be used when the probe is labeled with the reporter molecule itself. Radioactive probes
are the primary example of this type of detection strategy. The radioactive isotope emits particles, which can be
detected primarily by autoradiography (x-ray film) and more uncommonly by scintillation counting or Geiger-Müeller
tube counting. Either colorimetric or chemiluminescent substrates can be acted upon directly by nonradioactive probes
labeled with an enzyme such as HRP or AP. The Accuprobe kit (Ninet et al., 1992) hybridization format involves the
selective degradation of unhybridized probe; therefore, the hybridized probe target complexes can be directly detected.
The hybridization protection assay uses an acridinium ester covalently linked to the probe. After hybridization, the
solution is made alkaline and the labels on unhybridized probe are degraded, while those protected by the double helix
of the probe-target complex are protected (Ninet et al., 1992). Another unique hybridization format (Bassler et al.,
1995) uses PCR to effectively degrade an internally hybridized probe, thereby removing the quenching effect of one
molecule and allowing for the direct detection of a second reporter molecule.

Indirect detection methods are more commonly used in gene probe hybridizations using nonradiolabeled probes. The
two most common non-isotopic modifying groups incorporated into gene probes are digoxigenin and biotin; both are
used with indirect detection protocols. Streptavidin complexed to an active enzyme is used to specifically detect biotin-
labeled probes, and an enzyme-antibody complex specific for digoxigenin is used to detect probes labeled with that
moeity. Enzyme-antibody complexes specific to duplex nucleic acid hybrids have also been used in some hybridization
formats (Blais and Phillippe, 1993; Blais, 1994). HRP or AP are commonly used for detection. Another hybridization
format uses two sequential hybridizations (Wilson et al., 1990; Rose et al., 1991; Curiale et al., 1990; D'Aoûst et al.,
1995; King et al., 1989). The first hybridization is usually a "capture" step; the probe is either labeled with biotin or a
poly dT tail. Stringent hybridization conditions can be used to achieve maximum specificity without interfering with
the activity of the detection enzyme. Following the first hybridization the hybridized target is often captured with an

 Page 126

appropriately coated solid support. The captured target is then transferred to a second hybridization reaction where a
second probe labeled with the reporter molecule is used. This has been nicknamed a "dipstick" format.

Both colorimetric and chemiluminescent substrates are available to detect enzymes such as HRP and AP. Colorimetric
substrates often consist of 5-bromo-4-chloro-3-indoyl phosphate (BCIP)/nitroblue tetrazolium (NBT) solutions, where
results are developed in place. Chemiluminescent dioxetene substrates are cleaved by AP to give a localized light
emission detected by exposure of x-ray film. Although chemiluminescent detection involves an additional step, the
sensitivity is improved and multiple exposures can be made. In a study comparing the quantitation of PCR-produced
DNA, electrochemiluminescence was judged to be 10100 times more sensitive than fluorescence detection by using
ethidium bromide staining (Yu et al., 1995).

Formats for Gene Probe Use

Most gene probe hybridization formats fall into two groups. In the first group, one of the reactants is immobilized and
the other is free in solutionthe liquid-solid format. In some methods, the target is immobilized (such as in DNA colony
hybridization) and in others the probe might be fixed to the solid support such as a "dipstick," microtiter dish well, or
membrane. These simplify the removal of unbound probe by facilitating the physical removal and/or washing steps. In
the other class, both the gene probe and the target DNA are free in solution. This is often referred to as a liquid-liquid
format or a homogeneous system. A disadvantage of this system is the difficulty of removing labeled but unbound
probe molecules, but several innovative techniques are dealing with this problem (see below).

Liquid-Solid Format
DNA annealing experiments were first conducted in solution, but soon thereafter one of the DNAs was affixed to a
solid matrix for nucleotide similarity studies (Brenner et al., 1969). For screening bacterial cultures with a gene probe,
dot blots were often used. Extracted (but not necessarily highly purified) DNA is spotted onto and then fixed to a
membrane. A technique was developed for assaying colonies directly (Grünstein and Hogness, 1975), which consisted
of lysing the cells in colonies in situ and fixing them to a membrane or paper (Maas, 1983) matrix. Thus the
hybridization was carried out with the cells in the original array.

Recently, formats that use a DNA probe affixed to a dipstick have been developed for commercial applications. Some
of the systems can be rather complex involving multiple probes in a "sandwich" arrangement reminiscent of antigen-
antibody sandwich formats. In one such configuration, two probes are used. One (the "capture" probe) is affixed to a
dipstick, which is

 Page 127

then immersed in the test sample. If complementary target DNA is present, it will hybridize to the capture probe and be
removed from the solution when the solid support is withdrawn. In a variation on this theme, two probes
complementary to nonoverlapping regions on the target are used. The probes can become linked into the same complex
only when the target sequence is present and both probes can bind to it. A region of one of the probes (the capture
probe) is designed to be complementary to a sequence on a dipstick; therefore, the whole complex can be "captured"
and removed from the solution. The other probe is the reporter probe complexed with a molecule that can be easily
detected such as an enzyme. Such systems can be constructed to yield very low background signals.

Southern blots (Southern, 1975) also fall into this category of gene probe formats. DNA is digested with a restriction
endonuclease, the fragments are separated by gel electrophoresis, and the resulting fragment pattern is transferred to a
solid matrix where they are hybridized to a gene probe. Ribotyping is a type of Southern blot; separated restriction
fragments are probed, usually with E. coli ribosomal RNA, to reveal the distribution of ribosomal RNA genes (Grimont
and Grimont, 1986). Both of these methods are normally applied for subtyping species and for generating
epidemiological data.

Liquid-Liquid Format

When both the target and the probe are in solution, a kinetic advantage is gained (Wetmur and Davidson, 1968;
Meinkoth and Wahl, 1984), but until recently this format suffered from the technical problem of removing labeled but
unhybridized probe molecules. Several systems have been developed to overcome this difficulty. One method uses two
fluorescently labeled probes such that the emission from the first probe excites the emission of the second, and it is this
latter wavelength that is monitored. The probes are labeled such that, when complexed with the target, a label on the 5'
end of one will be in close proximity to the label on the 3' end of the other. Only when the probes are in this
configuration will any appreciable emission at the second wavelength occur so that unbound probes contribute only a
negligible amount of fluorescence (Cardullo et al., 1988; Heller and Morrison, 1985). It appears that this system would
allow continuous monitoring of hybridization reactions.
In the "hybridization protection assay," the probe is labeled so that the labeling moiety will be on the interior of the
double helix when the probe molecule hybridizes with a target molecule (Ou et al., 1990). Prior to detection, the
hybridization solution is treated to destroy the fluorescence of unbound probe labels, but those labels associated with
hybridized probe molecules will be protected.

 Page 128

Gene Probe Applications

In this section, we describe gene probe-based tests for most of the important foodborne pathogenic bacteria and
viruses. However, some of the organisms discussed here are not considered foodborne pathogens in the traditional
sense. They are usually found in the environment and, therefore, may contaminate food. Not all the organisms listed
here have been detected in foods by the use of gene probes, but probes have been developed to detect them in patient
samples or in the environment. It is hoped that these successes will serve as the motivation for applying these tools for
the testing of foods. One further caveat: The amount of discussion here is not proportional to the relative public health
significance of these microorganisms as foodborne pathogens but more or less is a function of the amount of published
information.
In several instances, PCR tests have been developed in conjunction with probe-based tests. In fact, many PCR primers
serve as perfectly good oligonucleotide probes, and probes increase the sensitivity of PCR-based tests by detecting
levels of amplified product not observable by gel electrophoresis. However, PCR-based methods are beyond the scope
of this chapter and will be discussed only tangentially.

Probe-based methods, especially in the colony hybridization format, are well-suited for the enumeration of bacteria in
foods. These quantitative results are often much more reliable than quantitative methods based on gene amplification
assays. In addition, colony hybridization methods do not raise questions about the presence of DNA in dead cells, or
viable but noncultural cells in a sample might be a concern with PCR-based and other amplification procedures.
Nonisotopic labels are proving to be dependable and now are as sensitive as radioisotopic labels. Some gene probe kits
are commercially available for the detection of a few bacterial species in some foods.

Bacteria

Aeromonas species

Although usually not considered to be foodborne pathogens, members of the genus Aeromonas may be found in both
fresh and brackish waters and, therefore, could enter foods (Wadström and Ljungh, 1991; Kirov, 1993). These
organisms can cause both intestinal and extra-intestinal infections. The taxonomy of this group is not well delineated,
although characterization of the nucleotide sequence for the region encoding 23S rRNA and the intergenic spacer
region of the 16S-23S rRNA genes of A. hydrophila are helping to clarify taxonomic relationships within the genus
(Martinetti-Lucchini and Altwegg, 1991; East and Collins, 1993). A

 Page 129

universal probe to the 16S and 23S rRNA sequences produced ribotyping patterns that were not useful in elucidating
the taxonomy of Aeromonas sp. but could be applied in epidemiological studies of A. hydrophila (Carey et al., 1994).

A number of extracellular substances such as hemolysins and cytotoxins are elaborated by aeromonads and appear to
be important pathogenic factors (Stelma et al., 1986). Also, the nucleotide sequence of the gene encoding a cytolytic
enterotoxin shares some similarity to an enterotoxin gene of C. perfringens and the listeriolysin gene of L.
monocytogenes (Chopra et al., 1993). This nucleotide sequence data allows the construction of a virulence factor gene
probe(s) for use in screening food samples.

Aerolysin is produced by some strains of A. hydrophila associated with human disease; however, the presence of an
aerolysin gene may not always correlate with strain virulence (Chopra et al., 1991a). There is considerable nucleotide
sequence divergence observed in the regions upstream from the structural genes for the aerolysin and a cytolytic
enterotoxin, which suggests that these structurally similar toxins may differ in the regulation of their expression
(Chopra et al., 1993). A DNA probe to an extracellular hemolysin demonstrated that the particular hemolysin gene was
present in 43 of 62 (70 percent) hemolysin-producing strains of A. hydrophila, however, all 43 hemolysin-producing
strains of A. salmonicida examined reacted with this probe; three negative strains did not react but three nonhemolytic
strains of this species did (Hirono and Aoki, 1991). Once the variation in the nucleotide sequences of these virulence
genes is better understood, useful gene probes for pathogenic strains will be developed for their detection and
enumeration.

Bacillus cereus

Capable of producing emetic or diarrheal syndromes, Bacillus cereus elaborates an emetic toxin or an enterotoxin
(Beecher and Macmillan, 1991; Drobniewski, 1993; Reed, 1994b). Vehicles implicated in outbreaks include
contaminated meat products (Luby et al., 1993; Sharma et al., 1994), poultry (Slaten et al., 1992), dairy products
(Granum et al., 1993; Sutherland, 1993; Becker et al., 1994), and rice-based dishes (Centers for Disease Control and
Prevention, 1994). Nucleotide sequencing of the 23S rRNA genes of B. anthracis and B. cereus revealed only two
differences (Ash and Collins, 1992). The intergenic spacer region of the 16S and 23S ribosomal RNA genes of B.
thuringiensis and B. cereus was compared; it was concluded that the difference was not sufficient to construct a
species-specific oligonucleotide probe (Bourque et al., 1995). Another presumptive virulence determinant, cereolysin
AB (a cytolytic protein responsible for hemolysis), has been targeted with oligonucleotide probes (Schraft and Griffith,
1995).

 Page 130

Even though there are no reports as yet of gene probe test procedures being applied to the testing of foods implicated in
foodborne disease outbreaks, the methodology is being developed and the tools are becoming available.

Campylobacter spp

Among the genus Campylobacter, the species C. jejuni and C. coli appear to be the most important agents of acute
diarrheal disease in humans and animals (Karmali et al., 1983). These two species are among the most frequently
isolated ones that cause diarrheal disease (Tauxe, 1991; Taylor, 1991), and are also found in foods (Butzler and
Oosterom, 1991). These campylobacters have been found to contaminate milk, poultry (Shane, 1992), pork, untreated
water (Park et al., 1991), and shellfish (Abeyta et al., 1993). To enrich for these species, microaerophilic conditions are
required; they may exist in a viable but nonculturable form (McKay, 1992). The enrichment procedure can be lengthy,
and subsequent growth and biochemical characterization are needed for identification.
An AP-labeled oligonucleotide probe, commercially available and specific for thermotolerant campylobacters, was
tested. This probe system detected all 71 C. jejuni and C. coli strains examined (Freier et al., 1989; Cudjoe et al.,
1991). Additional testing of this probe system shows that it is specific for some species of Campylobacter, but a loss of
sensitivity may be encountered when stool samples are tested (Olive et al., 1990; Tenover et al., 1990; Thorne et al.,
1990). Another commercial nonradioactive DNA probe was used to test enrichment broths inoculated with
contaminated chicken carcasses and feces (Stern and Mozola, 1992). To yield positive results a level of approximately
106 colony-forming units (CFU)/mL was needed in the enrichment. A biotinylated commercial C. jejuni probe was
used to characterize strains by RFLP analysis (Geilhausen et al., 1990). This technique appeared more sensitive than
serotyping and might have epidemiological applications.
The gene encoding the major flagellar protein, flagellin, in C. jejuni has been a popular target. Two copies of the
structural genes (flaA and flaB) are found as tandem repeats and are 95% similar to each other (Nuijten et al., 1990).
The flaA gene has been used as a Campylobacter-specific target both for probe and PCR tests. Oyofo et al. (1992) used
a nonradioactively labeled probe targeted to an internal region amplified by PCR to develop a test that was specific for
C. jejuni and C. coli. This probe could detect as little as 6 fg of purified C. coli DNA, equivalent to about four cells.
However, in stools seeded with C. coli, the limit of detection was between 30 and 60 cells in the PCR part of the assay.
Western blots indicated that if isolates from a single strain of C. jejuni expressed either of two flagellar antigens (59 or
62 kDa),

 Page 131

they still reacted with a flagellin-specific gene probe (King and Clayton, 1991).

Several probes targeting the genes encoding proteins in outer membranes have been developed. A nonradioactive probe
to the gene for a 46 kDa outer membrane protein could detect 105 C. jejuni cells on filters, but a fragment from C. coli
was 100-fold less sensitive (Taylor and Hiratsuka, 1990). The C. jejuni probe exhibited 93% sensitivity against culture-
positive stool samples but yielded 15% false-positive results on culture-negative samples. A gene probe to an 18 kDa
outer membrane protein (Omp 18) from C. jejuni showed that similar genes were present in all species of the
thermotolerant campylobacters. It was also determined that the nucleotide sequence of the Omp 18 gene is highly
conserved in strains of C. jejuni isolated from humans, dogs, cats, calves, and chickens but is different in other
Campylobacter species (Burnens et al., 1995). Antisera against a 24 kDa cloned protein reacted with C. jejuni and not
C. coli. The corresponding gene, mapA, was found by a probe assay to be present in 120 clinical isolates of C. jejuni
but not in 126 other campylobacters, which included 34 C. coli strains (Stucki et al., 1995). The recA gene of
Helicobacter pylori is 75% similar to that from C. jejuni. This gene appears to be a contributing factor in the survival
of H. pylori in a low-pH environment (Thompson and Blaser, 1995). The recA gene in Campylobacter may be a
virulence factor as well. Probes not targeted to a particular gene have also been developed. Somewhat surprisingly, a
biotin-labeled probe composed of C. jejuni genomic DNA reacted only slightly with C. coli DNA and was sensitive
enough to detect 102 CFU/g in chicken fecal material (Chuma et al., 1993).
PCR primers to repetitive extragenic palindromes (REP) or enterobacterial repetitive intergenic consensus (ERIC)
sequences have been developed (Giesendorf et al., 1993). It was noted that isolate-specific fragments were generated
and that these could be used as species-specific probes. This was confirmed by hybridization to Southern blots with C.
jejuni, C. coli, and C. lari DNA.

Clostridium botulinum
Bcause neurotoxins elaborated by C. botulinum in low acid-containing foods (Hutchinson, 1992) can cause severe and
sometimes fatal paralytic disease, the rapid detection and identification of these toxins in foods is critical for an
appropriate public health response to minimize the spread of illness during a foodborne outbreak. Bioassays may be too
slow in confirming the presence of the toxin(s) to adequately control the outbreak.

While most enteric foodborne pathogens cause disease as the result of an active infection, the effects of C. botulinum-
caused illness (with the excep-

 Page 132

tion of infant botulism) are due to an intoxicationthe ingestion of preformed toxin. Therefore, viable cells or spores of
C. botulinum might not be present in a food suspected of being contaminated with the etiological agent of an outbreak.
As PCR can be used to amplify the DNA of viable but nonculturable or even nonviable microbes, this method can be
employed to demonstrate if a food may have been contaminated with C. botulinum harboring a particular toxin gene. In
this case gene probe methods may not be as useful as techniques based upon gene amplification.

Some methods using both gene probes and PCR have been reported. For example, by targeting the 16S rRNA gene
sequences, a PCR-based method also employing a capture probe immobilized to a microtiter plate to detect clostridia
was developed with a claimed sensitivity of about five cells or spores (Galindo et al., 1993). The previously published
nucleotide sequence of C. botulinum type A neurotoxin (Binz et al., 1990; Thompson et al., 1990) was used by Ferreira
et al. (1992, 1994) for selecting primers to amplify a 1.3 kilobase pair (kbp) region. Commercially canned products
were inoculated with 100 viable spores per gram. After incubation samples were centrifuged, pellets were washed in
buffer and then they were heated at 99°C for 10 minutes to release DNA for PCR amplification. PCR products were
detected by dot blots by using a digoxigenin-labeled internal gene probe. Most food samples gave positive results after
one day of incubation, while others required more than 15 days. Some spores may not be detected by PCR because the
DNA cannot be efficiently released by the lysing procedure used to examine vegetative cells. The variable results
obtained with some food products could reflect differences in the ability of C. botulinum to sporulate in them, so care
must be taken when confronted by negative results. A degenerate primer pair was used to amplify DNA from C.
botulinum toxin types A, B, E, F, and G (Fach et al., 1995). The PCR products were used as gene probes for the
identification of each toxin type. Assay sensitivity was between 1 and 10 cells per reaction.

Clostridium perfringens

C. perfringens is the causative agent of gastroenteritis (Reed, 1994a; Borriello, 1995) and can be associated with
undercooked meat products (Regan et al., 1995). The sequence of an enterotoxin gene (cpe) from C. perfringens,
believed to be responsible for food poisoning in humans, has been determined; synthetic probes were developed and
tested (Van Damme-Jongsten et al., 1990). More recently, it has been determined that this gene may be either
chromosomally located or borne on plasmids (Cornillot et al., 1995). A nonisotopic enterotoxin probe was labeled with
digoxigenin during PCR amplification and used in a colony hybridization assay to detect and enumerate C. perfringens
in raw ground

 Page 133

beef seeded with known numbers of cells (Baez and Juneja, 1995). Sensitivity was below 10 CFU/g.

A digoxigenin-labeled probe for the alphatoxin gene (plc) was generated by using PCR (Schlapp et al., 1995). All 296
C. perfringens strains tested were positive, but none of 11 other clostridial species reacted. Much more work has been
done with PCR-based methods (Hill and Jinneman, 1996).

Escherichia coli

Many E. coli are nonpathogenic; however, it has been discovered that some E. coli can cause disease by several
mechanisms. The four main classes of enterovirulent E. coli associated with foodborne illnesses are enterotoxigenic E.
coli (ETEC), enteropathogenic E. coli (EPEC), enteroinvasive E. coli (EIEC), and enterohemorrhagic E. coli (EHEC).
Doyle and Padhye (1989) and Neill et al. (1994) provide a more complete review of these foodborne pathogens. Each
of these classes of enterovirulent E. coli is able to cause illness by a variety of virulence mechanisms. Many of the
genetic elements identified as being responsible for these various virulence mechanisms have been used as targets in
the development of molecular detection and identification methods including gene probe hybridization. These methods
have also been continually refined to improve their sensitivity and specificity and to apply them directly to analyses of
foods. The direct application of gene probe tests to food samples may enable a reduction in the number of selective
enrichment steps and thus provide a more rapid analysis. Less reliance on selective enrichments may also limit the loss
of plasmids encoding virulence-determining genes (Hill and Carlisle, 1981). Tests designed to detect either a single
class or several classes of enterovirulent E. coli using gene probe hybridization tests have been developed.

EPEC: The pathogenicity of EPEC is attributed to genes present on a 5570 MDa EPEC adherence factor plasmid and
the chromosomally located eae gene cluster, which is necessary for the attachment and effacing mechanisms steps in
causing disease (Donnenberg et al., 1993). DNA hybridization procedures have been developed to test for the presence
of a conserved section of the EAF plasmid (Natero et al., 1985) and for the eae gene cluster (Jerse et al., 1990a,b,
1991).

EIEC: These enterovirulent E. coli cause an invasive infection resulting in a dysenteric form of diarrheal disease in
humans. The invasive ability is attributed to several genes for outer membrane polypeptides encoded on a large 120140
MDa plasmid (Doyle and Padhye, 1989). Both the disease and this virulence plasmid are very similar to the disease
caused by and the virulence plasmid found in Shigella. Many gene probe methods have been developed for the
simultaneous detection of EIEC and Shigella (Boileau et

 Page 134

al., 1984; Small and Falkow, 1986; Taylor et al., 1988; Panda et al., 1990; Venkatesan et al., 1988, 1989; Oberhelman
et al., 1993).

ETEC: Two classes of enterotoxins are produced by enterovirulent E. coli from plasmid-encoded genes. The heat-labile
enterotoxin (LT) genes (elt) include both human (LTH or LT1) and piglet (LTP of LT2) versions and produce an
enterotoxin that is immunologically related to cholera toxin. The other class of enterotoxins are heat stable (ST), have a
low molecular weight, and are nonantigenic. As with the LT-encoding genes, several forms of the genes (est) for the ST
enterotoxins exist; the STaI and STaII forms are usually of human or animal origin, while the STb form is commonly
from a porcine source. The existence of so many different toxin gene types has complicated the development of gene
probe assays for the detection of these enterotoxin genes.

The detection of a single gene or combinations of genes using a colony blot hybridization format has been incorporated
into food analyses (Hill, 1981; Hill et al., 1986; Schultz et al., 1994; Saez-Llorens et al., 1989). The simultaneous
detection of a variety of individual LT and ST genes or alleles was accomplished with the use of a combination of
oligonucleotide probes that respond similarly under the same hybridization conditions (Hill et al., 1986, Schultz et al.,
1994). Approximately equal sensitivity for colony blots from stool samples spiked with 106108 CFU/g for both 32P or
digoxigenin-labeled oligonucleotide probes was reported when the colony lift filters were treated with proteinase K
prior to hybridization with the digoxigenin labeled probe (Schulz et al., 1994).

The generation of a fused gene probe for detecting ETEC harboring the genes for either LT or ST or both is an
alternative to the probe mixture approach (Saez-Llorens et al., 1989). Unique gene sequences from the estA and estB
genes (for ST) were cloned into a plasmid. A single hybrid RNA probe was generated by using RNA polymerase to
transcribe the linearized plasmid and to incorporate radioactive or biotinylated ribonucleotides. The greater stability of
RNA:DNA duplexes than DNA:DNA duplexes means that the hybridization stringency conditions can be increased
with an RNA probe. However, a disadvantage of RNA probes is the ubiquitous presence of RNAses. An SDS wash of
the nitrocellulose membrane before hybridization improved the specificity of the biotin-labeled probe.
EHEC: This class of enterohemorrhagic E. coli can cause hemorrhagic colitis and may lead to the serious illnesses of
hemolytic uremia syndrome (HUS) or thrombotic thrombocytopenic purpura (TTP). These strains of E. coli are
classified by their ability to elaborate one or more types of a cytotoxin belonging to the Shiga toxin (Stx) family. These
toxins are also referred to as Shiga-like toxins (SLT) or verocytotoxins (VT). Calderwood et al. (1996) have proposed
the new nomenclature of Stx to reflect the shared properties

 Page 135

and close relationship of this toxin to the Shiga toxin family. Shiga toxins 1 and 2 (Stx1 and Stx2) are the two major E.
coli Shiga toxin types. The Stx1 gene shares 99% sequence similarity and the toxin is immunologically
indistinguishable from Shiga toxin. Shiga toxin 2 (Stx2) is immunologically distinct from Stx and Stx1, and their genes
share only 58% sequence similarity. A variant of the Stx2 gene (Stx2v) has also been reported, primarily in E. coli
strains associated with edema disease in pigs (Donnenburg et al., 1993). A number of gene probe methods have focused
on the detection of these related genes (Samadpour et al., 1990, 1994; Cordovez et al., 1992; T'oth et al., 1994; Thomas
et al., 1991, 1993; Bettleheim et al., 1990).

Most illnesses caused by EHEC infection involve little or no fever, suggesting that this organism is able to colonize the
intestine and then elaborate toxins. The genetic elements attributed to the adherence ability of EHEC have been
identified on a 60 MDa plasmid. Most members of the O157:H7 serogroup have a conserved 3.4 kbp segment of this
60 MDa plasmid (Levine et al., 1987). An EHEC invasin encoded by the eae gene may also contribute to the virulence
of this class of E. coli (Louie et al., 1993). The EHEC O157:H7 eaeA gene has been cloned and sequenced (Beebakhee
et al., 1992; Yu and Kaper, 1992; Louie et al., 1994). Sequence similarity between the EHEC and EPEC eae genes is
greatest in the first 2.2 kbp (97%) but less similar (59%) in the 800 bp of the 3' end. Unique EHEC strain sequences for
hylA and hylC genes have also been identified (Schmidt et al., 1993, 1994, 1995). A number of genetic probe methods
for the detection, identification, and/or characterization of EHEC have been developed that target these genetic
elements (Willshaw et al., 1994; Schmidt et al., 1994; Beutin et al., 1995).

Poly- and oligonucleotide probes labeled with radioactive or nonradioactive reporter molecules can detect the EHEC
stx1, stx2, EHEC eaeA, uidA, and/or EHEC plasmid genes (Samadpour et al., 1990, 1994; Cordovez et al., 1992; T'oth et
al., 1994; Wilson et al., 1992; Thomas et al., 1991, 1993; Willshaw et al., 1993; Scotland et al., 1993; Bettelheim et al.,
1990; Beutin et al., 1995; Gicquelois et al., 1990; Willshaw et al., 1994; Feng, 1993). Oyster and raw hamburger
samples were directly analyzed with radiolabeled stx1 and stx2 gene probes with a minimum detection level of 1.3
CFU/g of E. coli O157:H7 (Samadpour et al., 1994). Both colony lift and dot blot hybridization formats were
compared by testing spiked food samples enriched in mTSB. Although the dot blot analysis was faster because colony
growth on an agar plates was not required, it did not allow recovery of the isolate from the food sample. EHEC stx
genes were detected by gene probe analysis in 17% of the food samples, including raw meat, poultry, and seafood
samples (Samadpour et al., 1994; Willshaw et al., 1993).
The EHEC serogroup O157:H7 is the one most commonly associated

 Page 136

with illness in humans, therefore many of the reported EHEC analyses are specific for this serogroup. Most O157:H7
analyses screen for distinct phenotypic characteristics such as the inability to utilize sorbitol and the lack of a
functional ß-glucuronidase. Recent reports have described some ß-glucuronidase positive strains of this serotype
(Gunzer et al., 1992; Schmidt et al., 1993; Karch et al., 1993; Hayes et al., 1995). A specific base substitution at
position 92 in the uidA (ß-glucuronidase) gene has been identified as specific for E. coli O157:H7 serotype (Feng,
1993). This base substitution does not cause the lack of ß-glucuronidase activity, as it has been found in both ß-
glucuronidase negative and positive strains (Hayes et al., 1995). Although this genetic distinction is not in a virulence
gene, it does provide a target for the development of genetically based tests specific for E. coli O157:H7. An
oligonucleotide probe hybridization assay has been developed that will detect and identify E. coli O157:H7 based on
this single base substitution (Feng, 1993). In addition, a 2.0 kb Sma I fragment E. coli O157 plasmid gene probe was
radiolabeled and reported to be specific for serogroup O157 EHEC when stringent hybridization conditions were used
(Huck et al., 1995).

Gene probes have also been used to improve the sensitivity and specificity of PCR analyses by detecting PCR-
amplified products (Lang et al., 1994; Begum et al., 1993; Brian et al., 1992). An unique gene probe application used
PCR to incorporate digoxigenin-11-dUTP into an amplified product from the stx gene. The digoxigenin amplified
product was detected by hybridization with specific internal gene sequences immobilized on nitro-cellulose membranes
(Jackson, 1991).

Gene probe hybridization has also been used to provide the specificity needed to distinguish between PCR amplified
products from the stx1 and stx2 genes. A PCR analysis used degenerate primers from the conserved regions of these
genes. The amplified product was hybridized to internal oligonucleotide probes that shared only 30% sequence
homology, and, therefore, it was possible to distinguish between the stx genes (Karch and Meyer, 1989).

Gene probes have also been used to more fully characterize bacterial strains and provide epidemiological information
in foodborne outbreaks. A series of probes to a number of genetic elements can be used to establish a genotype for each
strain tested that is based on the overall presence or absence of each genetic element. In the case of EHEC strains, stx
genotypes have been established using this approach (Paros et al., 1993; Osterhoff et al., 1989; Samadpour, 1995).

Probe-based restriction fragment length polymorphism subtyping techniques have also been developed. The general
principle is to identify a subset of restriction fragments, which can vary in number and size between

 Page 137

different strains. Total genomic DNA is digested with a restriction enzyme; the fragments are separated by gel
electrophoresis and transferred to a membrane for hybridization with a gene probe. The most common probes used for
this purpose are the rRNA genes, and this procedure is termed ribotyping (Grimont and Grimont, 1986). The rRNA
genes are selected because they contain conserved sequences for which a probe can be prepared and used with many
bacterial species. In addition, there are generally between 7 and 20 copies of the rRNA operon per bacterial genome,
which provides multiple targets for the gene probe, thereby increasing the number of fragments composing a
characteristic pattern for a given bacterial isolate. For a given bacterial genus the choice of restriction enzyme(s) is
determined empirically based on the generation of a range of ribotype patterns that permit adequate discrimination
between individual strains.

Many epidemiological studies of EHEC have focused on the single E. coli O157:H7 serogroup. Because members of
this serogroup are very closely related, it is difficult to develop subtyping methods capable of distinguishing strains.
Ribotyping has not been able to provide a high level of discrimination for strains of this serogroup (Martin et al.,
1996). Other genes that have been selected from probe-based subtyping methods include the l phage (Paros et al.,
1993; Samadpour et al., 1993; Grimm et al., 1995), insertion sequences, specifically IS3 and IS30 (Whittam, 1995),
and stx1 and stx2 (Samadpour, 1995; Grimm et al., 1995).

Listeria

Listeria monocytogenes is the only species of the six comprising the genus Listeria to be considered an important
foodborne pathogen for humans. The presence of this species in food and its significance as a foodborne pathogen has
prompted the development and application of gene probe methods for food analyses. A number of these techniques
have been reported for the detection and identification of the genus Listeria and more specifically for L.
monocytogenes.
Microwaving enhanced the lysis of the gram-positive Listeria cells prior to hybridization procedures (Datta et al.,
1987). A 32P-labeled, 500 bp fragment probe based on a putative-hemolysin gene was used effectively in colony
hybridization to identify and enumerate L. monocytogenes in naturally contaminated dairy products with levels, which
ranged from 6 × 103 to 3.6 × 106 CFU/g (Datta et al., 1988a). Several oligonucleotide probes were generated from this
500 bp region, one of which was used to detect L. monocytogenes in artificially contaminated raw milk and soft
cheeses (Datta et al., 1988b). Nonradioactive gene probes have also been developed for this target region. The full-
length 500 bp fragment was digoxigenin labeled by random priming and used in a colony-lift hybridization application
(Kim et al., 1991a). Cross-reactivity with one L. seeligeri isolate was eliminated using

 Page 138

stringent hybridization conditions. Sequences specific for L. monocytogenes were identified within this region and
developed into oligonucleotide probes labeled with digoxigenin by terminal deoxynucleotidyl transferase. Cross-
reactions with L. seeligeri were not observed with these oligonucleotide sequences under stringent hybridization
conditions (Kim et al., 1991b). The 500 bp fragment used by Datta et al. (1988a) is part of the sequence that encodes a
major secreted polypeptide of approximately 60 kDa designated the msp gene (Flamm et al., 1989). The msp gene [also
called iap for invasion-associated protein (Kuhn and Goebel, 1989)] is conserved among L. monocytogenes strains, and
stringent gene probe hybridization assays can be specific for this species. A unique internal region of the iap gene was
used as a probe specific for L. monocytogenes (Kohler et al., 1990).

The listeriolysin O gene is also a target for gene probes specific for L. monocytogenes (Mengaud et al., 1988;
Liemeister-Watcher and Chakraborty, 1989; Datta et al., 1990). Again, highly stringent hybridization conditions are
required to make these gene probes specific for L. monocytogenes. An oligonucleotide probe identified by Datta et al.
(1990) was used in a colony blot hybridization test to detect L. monocytogenes in ground beef samples (Mohamood et
al., 1992). The ability of the plating media to suppress background growth and the stringency of hybridization
conditions affected probe efficiency. These probes were also labeled with HRP (Datta et al., 1993); the colony blot
hybridization format provides enough target copies to produce a signal, in spite of the lower sensitivity of the
nonradioactive probes compared to radioactive probes. The most suitable matrix for hybridization assays that use
nonradioactive probes are charged nylon membranes (Datta et al., 1993).

Oligonucleotide probes for listeriolysin O and the iap genes were labeled either directly by coupling to HRP or with
digoxigenin incorporated into the probe. These probes were compared for detecting L. monocytogenes in soft cheese in
colony-lift hybridization formats (Herman and DeRidder, 1992). Background signals were produced by both
nonradiolabeled probes, but the removal of excess cellular debris from colony blots by rubbing with a soft paper
soaked in 5X SSC solution reduced the background signal with the digoxigenin system.

Commercially prepared kits make gene probe technology more readily available. Listeria genus-specific regions of the
16S rRNA gene are the target for the Gene-Trak product. Because the target is present in multiple copies per cell, gene
probe assay sensitivity is improved. When first available with a 32P-labeled synthetic probe (Klinger and Johnson,
1988), the Gene-Trak Isotopic Assay was compared to the FDA cultural method and found generally equivalent when
testing spiked food samples but less sensitive when applied to naturally contaminated seafood samples (McKee et al.,
1989). A

 Page 139

colorimetric version of the Gene-Trak gene probe hybridization is now available (King et al., 1989). When the Gene-
Trak system was compared to the International Dairy Federation cultural method with 250 naturally contaminated
cheese and environmental samples, a 5.9% false-negative rate was reported (Url et al., 1993). The possibility that the
samples did not reach the minimum detection limit of 106 cells/mL during the enrichment step was cited as a possible
explanation.

The AccuProbe L. monocytogenes culture identification test (Gen-Probe Inc., San Diego, CA) is a second
commercially available system based on a liquid hybridization format. A single-stranded, acridinium ester-labeled
DNA probe complementary to the rRNA of the target organism is used. A stable DNA:RNA hybrid is formed between
rRNA released from the cells and the DNA probe. Unhybridized probes are selectively inactivated by differential
hydrolysis with an alkaline H2O2 solution. the DNA:RNA hybrids are detected by quantitation of the light emitted by
the acridinium ester label (Ninet et al., 1992). A minimum of 105 CFU from a colony resuspended in physiological
saline could be detected. When samples from three nonselective enrichment broths and the FDA selective Listeria
enrichment broth were tested, the minimum detection level was 106 CFU per assay. However, the selective enrichment
broths UVM and PALCAMY interfered with the AccuProbe hybridization kit sensitivity (Ninet et al., 1992). The
AccuProbe L. monocytogenes product performed well for the specific identification of L. monocytogenes compared to a
slot blot hybridization analysis using the probe developed by Datta et al. (1988a) and biochemical characterization tests
(Johnson and Lattuada, 1993).

Another hybridization format used monoclonal antibodies to hetero-duplex nucleic acids for the capture of hybrid
molecules between Listeria rRNA and a biotinylated rRNA probe. The probe target was a unique Listeria species-
specific spacer sequence between the 16S and 23S rRNA genes. A minimum detection level of 5 × 102 CFU/mL in
artificially contaminated meat homogenates was reported (Emond et al., 1993).

Gene probes have also been used to improve the specificity and sensitivity of PCR analyses by detecting and
confirming the amplified product. Both nested PCR and digoxigenin-labeled internal probes were able to detect PCR-
amplified product from £ 10 cells in sterile cerebral spinal fluid (CSF) (Jaton et al., 1992). A 32P-labeled internal probe
to detect amplified product was also reported (Bessesen et al., 1990). A third variation was an unlabeled RNA probe
complementary to the amplified hylA fragment. A polyclonal anti-RNA-DNA IgG antibody-peroxidase conjugate was
used for hybridization detection and was reported to increase sensitivity when compared to conventional agarose gel
electrophoresis (Blais and Phillippe, 1993). This approach was further refined by using one PCR primer spiked onto
the T7

 Page 140

RNA polymerase. Thus, the amplified product can produce an RNA transcript, which is captured by a complementary
DNA probe immobilized in microtiter wells. An immunoenzymatic assay with anti-RNA-DNA antibodies detects as
little as 0.15 ng of amplified product representing a 10-fold improvement over conventional gel electrophoresis (Blais,
1994).

An unique alternate approach uses an internal probe included in the PCR amplification. This probe is labeled with both
a fluorescent reporter dye [6-carboxyfluorescein (FAM), tetrachloro-6-carboxyfluorescein or hexachloro-6-
carboxyfluorscein] and fluorescent quencher dye [6-carboxy-tetramethylrhodamine (TAMARA)], which are coupled to
individual nucleotides. The juxtaposition of these two labels permits the TAMARA to quench the fluorescence of the
FAM dye. During PCR amplification, the internal probe anneals to the template during each cycle. The Taq polymerase
5'-to-3' nuclease activity digests the internal probe during the extension step of each cycle. This increases the
fluorescence of the reporter FAM dye by removing the quenching dye from its nearby location. The amount of reporter
molecule signal is quantitative for the amount of the PCR product produced (Bassler et al., 1995).

Gene probes have been particularly useful for subtyping L. monocytogenes by ribotyping. A variety of probe labels
have been used for this application including 32P (Swaminathan et al., 1988), biotin (McLauchlin et al., 1988), and
digoxigenin (Graves et al., 1991; Graves et al., 1994). For ribotyping, the restriction enzyme EcoR1 provided the most
discrimination among Listeria isolates. The RiboprintTM procedure, which uses gene probe technology, has been
automated and is available commercially (Weidmann et al., 1996; Bruce et al., 1995; Hubner et al., 1995).

Salmonella

The development of DNA hybridization assays have focused on the simultaneous detection of all Salmonella spp. in
foods because most Salmonella spp. are associated with illness in humans. One of the first DNA-DNA hybridization
applications for detection of Salmonella spp. in foods was reported by Fitts et al. (1983). This assay used a 32P-labeled
probe complementary to chromosomally located Salmonella-specific sequences and a dot blot hybridization format
from preenrichment cultures. A commercial kit (GeneTrak) was developed using this probe. The test compared
favorably with standard culture methods on more than 1600 samples representing 23 food types (Flowers et al., 1987).
This kit was also used to detect the DNA of Salmonella spp. extracted directly from filter-concentrated water samples
(Knight et al., 1990).

Several improvements were incorporated into later versions of the Gene-Trak kit. The probe target sequence was
changed to Salmonella rRNA genes, which are present in multiple copies per cell. A colorimetric detection

 Page 141

system with equivalent sensitivity was substituted for the isotopic system because of the more abundant target
sequence. The Gene-Trak colorimetric assay format incorporated a ''capture" probe for the rRNA target sequences with
a poly dA tail. The hybridized target DNA complex was captured from the liquid environment with a poly dT coated
dipstick. The captured target was then transferred to a new tube for detection. Finally, a "reporter" probe labeled with
fluorescein isothiocyanate (FITC) covalently coupled to the oligonucleotide is hybridized with the target and detected
with use of anti-FITC horseradish peroxidase conjugate and added substrate-chromagen. This version of the Gene-Trak
kit was compared with conventional cultural and other rapid method tests based on EIA techniques for the detection of
Salmonella from foods (Wilson et al., 1990; Rose et al., 1991; Curiale et al., 1990; D'Aoûst et al., 1995). D'Aoûst et al.
(1995) explored the use of alternative cultural conditions to improve the Gene-Trak method performance and perhaps
reduce the preenrichment and selective enrichment steps required.

Other probes and a number of strategies have been identified to improve gene probe analyses for the detection of
Salmonella spp. (Scholl et al., 1990; Tsen et al., 1991; Olsen et al., 1991; Pumfrey and Nelson et al., 1991). Wicking
was used to effectively separate DNA from substances that compromise DNA retention and hybridization reactivity
(Scholl et al., 1990). Cells are rapidly lysed and DNA denatured using an alkali-surfactant reagent. The released DNA
migrates onto Hybriwix filters via vertical capillary action. For colony blot analyses the transfer of small colonies
allowed more complete hydrolysis by proteinase K and reduced background in a biotin labeled DNA probe
hybridization procedure (Tsen et al., 1991). A minimum detection level of 1.6 Salmonella CFU/g food was reported for
both dot blots from preenrichments and selective enrichments using colony blots.
Genes associated with the virulence of Salmonella spp. have also been used as targets for gene probe hybridization
method development. Probes have been used to detect the invA-D genes needed for invasion of intestinal epithelial
cells (Galan and Curtiss, 1991; Rahn et al., 1992). PCR was used to label part of the invA gene with digoxigenin for
use as a probe which was specific for Salmonella in a nitrocellulose filter colony hybridization assay (Bulte and Jakob,
1995). A comparison of gene probes to the Salmonella virulence genes invA, pagC, and spvC reported that pagC was
detected the most often, 99% of 103 Salmonella isolates tested (Nolan et al., 1995). An 18 bp synthetic oligonucleotide
that incorporated a 1 bp difference in the Salmonella plasmid virulence gene (spvA) produced a gene probe specific for
S. enteritidis (Hanes et al., 1995).
As with other bacterial genera, different detection gene probe protocols have been employed for the detection of PCR-
amplified products from

 Page 142

Salmonella. These ranged from using digoxigenin-labeled probes to detect internal sequences of the amplified products
(Fluit et al., 1993) to the "capture" of amplified sequences with oligonucleotides covalently bound to microtiter wells
followed by hybridization and detection with an internal AP-labeled probe and chromagen substrate (Cano et al.,
1993). Enzyme-based chemiluminescent and colorimetric assays were compared for the detection of PCR-amplified
Salmonella spp. gene fragments generated using biotin-labeled primers (Soumet et al., 1995). Internal oligonucleotides
immobilized on microtiter wells captured the biotinylated, amplified products, which were detected by using an
enzyme-streptavidin conjugate. Chromogenic substrates for peroxidase and AP for colorimetric tests and Lumi-Phos
530 chemiluminescent substrates were compared. The chemiluminescent method was the most sensitive, allowing for
detection of 1.8 fmol of target DNA per microtiter plate well while minimum detection levels of 810 fmol for AP and
34 fmol using peroxidase were reported for the colorimetric methods. Gel agarose analysis had a detection limit of
approximately 8.8 fmol amplified product. Luk (1994) described a PCR method that used biotinylated primers and
digoxigenin dUTP incorporated in the PCR reaction. Biotin-labeled amplification product is captured by streptavidin-
coated microtiter wells, and the "captured" amplification product was detected by an antidigoxigenin-AP conjugate.

An oligonucleotide ligation assay (OLA) described by Stone et al. (1994) for the detection of PCR-amplified products
also makes use of gene probes. Two adjacent oligonucleotide probes are hybridized to the denatured PCR amplification
product. One probe was biotinylated on the 5' end and referred to as the "capture probe"; the "reporter probe" was
labeled with digoxigenin on its 3' end. Ligation between the two probes was accomplished only if there was complete
complementarity. This technique improves the sensitivity to allow detection of single base pair differences.
Streptavidin-coated microtiter wells are used to capture the ligated probe complex, which is detected with a
chromagen-substrate complex.

Salmonella subtyping is another area in which gene probes have been applied. Ribotyping has been used to subtype
Salmonella typhi strains (Altwegg et al., 1989; Nastasi et al., 1991). Stanley et al. (1995) reported that probe-based
subtyping with an IS200 probe was more discriminatory than ribotyping and plasmid profiling for S. panama strains.

Shigella
The genus Shigella, the causative agent of shigellosis, consists of four species: S. dysenteriae, S. flexneri, S. boydii, and
S. sonnei. As with other foodborne pathogens discussed here, many of the genetic techniques used in Shigella analyses
target both plasmid and chromosomally located,

 Page 143

virulence-associated genes (Maurelli and Lampel, 1994). All Shigella spp. are very closely related to E. coli, with
greater than 80% DNA sequence similarity. Shigella and enterovirulent E. coli both possess similar plasmids required
for virulence. This genetic relatedness allows the development of probe methods capable of detecting both Shigella and
some EIEC.

The earliest reported DNA hybridization methods used 32P-labeled probes targeting genetic sequences complementary
to the virulence plasmid for the detection of Shigella spp. and EIEC (Boileau et al., 1984; Small and Falkow, 1986).
Wood et al. (1986) compared these two probes with the Serèny invasiveness test for the identification of 42 Shigella
and 29 E. coli isolates. Three Serèny test-negative isolates hybridized with both probes, suggesting that additional
factors may be required for virulence. The Boileau et al. (1984) probe was applied to the direct detection of Shigella
and EIEC in stool blots; it was reported that the specificity of the probe for Shigella and EIEC remained high, but the
sensitivity was poor compared to cultural results. This finding was attributed in part to inhibition of Shigella and EIEC
by other fecal bacteria (Taylor et al., 1988). Panda et al. (1990) compared an AP-conjugated oligonucleotide probe
selected from the 2.5 kb probe identified by Small and Falkow (1986) to the Serèny test by using a dot blot format for
the detection of both Shigella and EIEC. The probe hybridized with all 52 Serèny test-positive isolates and 4 of 21
Serèny test-negative isolates. Invasion plasmid antigen (ipa) genes have also been used as a target for probe
development to specifically detect Shigella and EIEC (Venkatesan et al., 1988, 1989; Oberhelman et al., 1993). The
ipaH gene is a multiple-copy element encoding a 60 kDa antigen and is located both chromosomally and on the
invasion plasmid. Due to the dual location of this gene it was selected as a probe target and has been labeled both with
32P (Venkatesan et al., 1989) and suberyl-AP (Oberhelman et al., 1993).

Staphylococcus aureus

Like C. botulinum, disease caused by Staphylococcus aureus, results from ingestion of preformed toxins (an
intoxication) rather than by infection. Enterotoxins produced in the food are later ingested; live organisms at the time of
ingestion are not required. S. aureus, a relatively common cause of food poisoning, can produce several toxins. The
most important agents of foodborne illness carried by this species are the family of heat-stable staphylococcal
enterotoxins types A through E. The consumption of the preformed toxin can cause vomiting and diarrhea. The time of
onset is rapid, a matter of a few hours, when compared with other enterotoxigenic organisms such as E. coli and V.
cholerae, which cause illness only after infection, colonization, and subsequent production of toxin within the lumen of
the host's intestine. Therefore, the identification of toxin type for

 Page 144

epidemiological purposes is important. It is also very useful to determine if nonviable cells are present in a food
suspected of being the vehicle of a foodborne disease outbreak.

There has been considerably more progress reported about the design of PCR-based methods for the detection of
enterotoxigenic S. aureus than has been published on gene probes (Hill and Jinneman, 1996). Nevertheless, gene
probes specific for the toxin types have been developed (Neill et al., 1990; Jaulhac et al., 1992). A S. aureus-specific
gene probe based on a unique region of the 16S rRNA gene sequence was reported (Bentley et al., 1993).

More recently, Wilson et al. (1994) developed a digoxigenin-labeled probe based on total genomic DNA. The probe
was specific for S. aureus in meat and milk even though other staphylococcal species were present. It is also important
to keep in mind that PCR primers, or the regions amplified by PCR, can usually be enlisted as serviceable gene probes
for specific targets. Usually gene probes are not sufficiently sensitive to detect the relatively low levels of
contaminating bacterial cells directly in foods. A growth period is often required in conjunction with the colony
hybridization format. PCR, on the other hand, can detect DNA of dead cells; this may provide important
epidemiological information, which would not be available if analysts had to rely upon traditional growth-related
microbiological methods.

Vibrio spp

The diarrheal disease cholera is endemic in some parts of Asia. However, the disease occurred only sporadically in the
Western Hemisphere until a recent outbreak in Latin America (Tauxe and Blake, 1992). It now appears to have a
significant foodborne aspect (Popovic et al., 1993). The large number of persons affected worldwide continues to make
Vibrio cholerae O1 and other pathogenic members of this genus subjects of intense investigation.

The most important pathogens within this genus for humans are V. cholerae, V. parahaemolyticus, and V. vulnificus.
These usually occur naturally in the marine environment. Therefore, typical indicators of fecal pollution do not reliably
correlate with the presence or levels of these species. Rapid screening methods are needed for detecting pathogenic
strains and for monitoring seafoods and seawater, notably those within shellfish-growing areas, and especially in
warmer waters.

A considerable amount of nucleotide sequence information for virulence factors of the important species pathogenic to
humans has accumulated. Molecular genetic subtyping techniques such as ribotyping have been used to create profiles
for epidemiological purposes (Wachsmuth et al., 1994). The regions of the 16S rRNA gene that exhibit sequence
variation

 Page 145

have been used as target sites for both species-specific and genus-specific gene probes and PCR primers (Dorsch et al.,
1992). Coelho et al. (1994) constructed specific synthetic oligonucleotides for the detection of Vibrio.

The classical cholera enterotoxin appears to be the major virulence factor of enterotoxigenic V. cholerae O1. The genes
that encode this toxin (ctxA and ctxB) have been the target of many gene-probe and PCR-based tests, although these
genes show sequence homology with the heat-labile enterotoxin of E. coli and a gene in Salmonella (Moseley and
Falkow, 1980; Chopra et al., 1991b). Wright et al. (1992) developed an AP-labeled oligonucleotide probe to a 23-base
region of the ctxA gene, which did not hybridize with the DNA of heat-labile enterotoxin-producing strains of E. coli. A
3-hour hybridization test with an AP-labeled, 30-base oligonucleotide probe to ctx was also reported by Yoh et al.
(1993). An enzyme-labeled ctx gene probe was a rapid means of detecting V. cholerae in clinical stool samples (Miyagi
et al., 1994). Said et al. (1994) collected evidence that gene probe-negative strains of V. cholerae non-O1 produce a
toxin that reacts with antiserum to cholera toxin but is not identical to it.

Hoge et al. (1990) and Takeda et al. (1991) developed oligonucleotide probes specific for a heat-stable enterotoxin
produced by some strains of V. cholerae non-O1 (NAG-ST) and used it to demonstrate that some V. cholerae O1 strains
also contained the gene for this toxin. However, there may be additional virulence factors; a strain of non-O1 V.
cholerae serovar O5 isolated from a patient with diarrhea did not produce the classical cholera toxin nor the NAG-ST
(Bhattacharya et al., 1992). Strains previously found to be negative for the NAG-ST by the suckling mouse assay were
retested with a NAG-ST probe and 2.3% of the strains were positive (Pal et al., 1992). Concentrated cell culture
supernatants were then assayed and found to possess ST activity.

Investigations prompted by the epidemic in Latin America have resulted in a rapidly increasing body of molecular data
on the genetic epidemiology of V. cholerae. Many of these tests are based on PCR (Berche et al., 1994; Makino et al.,
1995; Usera et al., 1994; Morgan and. Jones, 1995) or sequencing (Olsvik et al., 1993; Lebens and Holmgren, 1994).
However, Wachsmuth et al. (1993) concluded, based on RFLP analysis of rRNA and ctxA as well as DNA sequences of
the ctxB gene and multiple locus enzyme electrophoresis, that the Latin American clone of V. cholerae is closely
related to the seventh pandemic strain.
V. vulnificus is often recovered from marine molluscan shellfish harvested from U.S. coastal waters, especially in the
south during the warmer months. It appears that most persons are not at risk. However, the mortality rate for
susceptible individuals approaches 50% (Tackett et al., 1984). An important step in the use of molecular genetic
methods for the detection of this

 Page 146

pathogen was the cloning of a major virulence factor, the cytotoxin-hemolysin gene (hylA) (Wright et al., 1985). The
hylA gene of V. vulnificus has been sequenced (Yamamoto et al., 1990) and used as a target for PCR-based
amplification to test artificially contaminated oysters (Hill et al., 1991).

A nonisotopic, V. vulnificus-specific probe targeted to the hylA gene could detect about 102 CFU/mL in seeded oysters
(Wright et al., 1992, 1993). A cytotoxin-hemolysin probe was also used by Miceli et al. (1993) to enumerate V.
vulnificus in oysters. Coupled with a direct plating medium (VVE), gene probe analysis could detect 10 CFU per 100 g
of sample. The effectiveness of the selective agar, colistin-polymixin B-cellobiose agar, for V. vulnificus isolation has
been evaluated by the use of gene probes (Oliver et al., 1992). The cytotoxin-hemolysin probe was also useful for the
screening of environmental isolates (Kaysner et al., 1987; Morris et al., 1987). A fluorescently labeled eubacterial
oligonucleotide probe targeted to 16S rRNA genes was used to detect and enumerate V. vulnificus in liquid or colony
hybridization formats (Heidleberg et al., 1993). No statistically significant differences between colony hybridization
and acridine orange stain counts were detected.

After a prolonged period in a nutrient-poor medium, several species of microorganisms enter a viable but not culturable
state (VBNC) (Roszak and Colwell, 1987). Both V. cholerae and V. vulnificus can enter the VBNC, state in the
environment (Colwell et al., 1985; Oliver et al., 1991). It is possible that some bacterial species may also exhibit this
physiological phenomenon in foods. Such a quiescent metabolic state may make it difficult to demonstrate their
presence in food by using traditional culture methods and antibody-based assays. However, these cells still contain
DNA and may therefore be detectable by PCR. For example, V. vulnificus can be difficult to isolate from the marine
environment during the cooler months when it may be in a VBNC; however, this state can be reversed by resuscitation.
Nevertheless, V. vulnificus cells treated in the laboratory to render them viable but unable to form colonies upon plating
were more difficult to detect by PCR than expected (Nilsson et al., 1991; Brauns et al., 1991). The explanation for this
observation is not known, but it is possible that the DNA may have accumulated single-stranded nicks rendering it a
poor template for PCR.

The density of the halophile V. parahaemolyticus in marine environments is related to water temperature. Levels within
oysters may be 100 times higher than in the surrounding water (DePaola et al., 1990). The majority of V.
parahaemolyticus strains recovered from the environment do not express hemolysin (thermostable direct hemolysin,
TDH), a virulence factor (Miyamoto et al., 1969; Shirai et al., 1990). The tdh nucleotide sequence was determined
(Nishibuchi and Kaper, 1985) and specific oligonucleotide

 Page 147

probes were developed, but other Vibrio species showed some sequence similarity (Nishibuchi et al., 1985). Tada et al.
(1992) coupled a PCR assay and an AP-labeled probe to develop an assay for the tdh gene, which could detect seven
copies of the gene. A second hemolysin, the thermostable related hemolysin (TRH) has been identified, the gene (trh)
cloned and sequenced, and subsequently a gene probe developed (Nishibuchi et al., 1989). The trh gene exhibits about
68% nucleotide sequence similarity with the tdh gene, and variants have been discovered (Kishishita et al., 1992).

Lee et al. (1992) used a probe for the tdh gene to develop a specific hybridization assay for V. parahaemolyticus.
Enzyme-labeled probes for the tdh and trh genes were developed although the tdh probe hybridized with some strains
of V. hollisae and V. mimicus (Yamamoto et al., 1992).

Yersinia enterocolitica and Yersinia pseudotuberculosis

There are several serotypes of Yersinia enterocolitica in nature, but only a few are important agents of disease in
humans (Kapperud, 1991). A series of genetically related plasmids of 6070 kbp occur in several species in the genus
and are essential for causing disease. However, this extrachromosomal DNA can be lost during growth (especially at
37°C) or during prolonged storage of cultures in the laboratory (Zink et al., 1980).

Certain serovars of Y. enterocolitica are more prevalent in Europe, while other groups are found more frequently in
North America. These groups can be distinguished by the size of the PCR product generated (Ibrahim et al., 1992a).
This ability to differentiate between strains of different origins is useful not only epidemiologically but clinically as
well because diseases caused by the two groups of strains have different sequellae. Portnoy et al. (1981) determined
that a family of plasmids plays an important role in the pathogenicity of Y. enterocolitica. Regions of the plasmids
involved in virulence were mapped by insertion analysis (Portnoy and Falkow, 1981). Pooled restriction enzyme DNA
fragments were used as a probe to detect and differentiate virulent and nonvirulent strains and then to enumerate the
latter in artificially contaminated foods (Hill et al., 1983). The efficiency of the pooled fragments probe was
determined for seven artificially contaminated foods (66100%) and appears to be influenced by background microflora
titers > 107 CFU/g (Jagow and Hill, 1986).

DNA probes were made from eight regions of the virulence plasmid and from the ail and inv regions of the
chromosome. The chromosomally derived probes reacted with all species of yersiniae, but a similar probe from Y.
pseudotuberculosis hybridized with Y. pseudotuberculosis and Y. pestis but not Y. enterocolitica (Robins-Browne et al.,
1989). A virulence plasmid-borne region involved in cell cytotoxicity and the Serèny reaction for cell invasion was
used to detect pathogenic Y. enterocolitica strains seeded into foods

 Page 148

(Miliotis et al., 1989). A number of chromosomal genes are required for virulence including inv and ail, which encode
a factor required for the invasion of eukaryotic cells (Miller et al., 1989); yst encodes a Y. enterocolitica heat-stable
enterotoxin (Pierson and Falkow, 1990), although there may be more that one variant of this latter gene (Robins-
Browne et al., 1993; Ibrahim et al., 1992b). Probes derived from the yst sequence were efficient in distinguishing
pathogenic from nonpathogenic strains (Delor et al., 1990).

Probes for the ail gene in Y. enterocolitica and the inv gene in Y. pseudotuberculosis were found to be species-specific
(Feng, 1992). Two methods for isolating Y. enterocolitica from meats were monitored by using a gene probe to the
plasmid-encoded yadA gene in a colony hybridization format (Nesbakken et al., 1991). Nonradioactive probes based on
the inv and ail genes were useful for identifying pathogenic strains of Y. enterocolitica in swab samples obtained from a
pig slaughterhouse (Goverde et al., 1993). Disease-causing strains were carried by pigs and detected by using both a
PCR test for the ail gene as well as digoxigenin-labeled gene probes for the ail, inv, and yst genes (Kwaga et al., 1991).
Ibrahim et al. (1992a) used PCR to generate a digoxigenin-labeled probe for yst to distinguish between pathogenic and
nonpathogenic strains. The probe proved effective when inoculated fecal samples were examined. A region of the
virulence gene, yopA, was used to detect virulent strains (Kapperud et al., 1990). From studies on plasmid-harboring
and plasmid-less strains, it was concluded that this gene resides on an extrachromosomal element. When testing
artificially contaminated foods, detection efficiency ranged from 32 to 82% even when the indigenous microforal level
was 3 × 104 CFU/g.

The appropriate choice of a particular target gene for a probe-based assay or PCR amplification test is not always clear.
Most disease-causing bacteria carry a number of genetic functions, which they require to remain pathogenic towards
humans. If plasmids are involved in carrying genetic information for pathogenic phenotypes, the additional
complication of assuring that the plasmid is not lost during any enrichment steps must be considered. Which genes are
the most appropriate as probe/PCR targets for detecting Yersiniae: those that are plasmid encoded or those that are on
the chromosome? This dilemma is nicely illustrated by examining the following papers. While not discussing the
detection of Yersinia in foods, Wren and Tabaqchali (1990) developed a primer set to the regulatory gene, virF, located
on the virulence plasmid and found in association with virulent strains Y. enterocolitica. Of 59 isolates tested, 51 (86%)
were positive for the virulence plasmids using PCR amplification. The eight negative isolates had apparently lost the
virulence plasmid and, therefore, the PCR amplification target. The loss of virulence plasmids upon laboratory culture
at 37°C

 Page 149

occurs in Yersinia (Zink et al., 1980) and in E. coli during enrichment from foods as well (Hill and Carlisle, 1981; Hill
et al., 1985). The presence of the virF gene was also shown to be correlated with the crystal violet binding assay
(Koeppel et al., 1993).

While Fenwick and Murray (1991) developed a primer set to a chromosomally encoded virulence factor in Yersinia
that would not be lost during culture in the laboratory, evidence suggested that both the plasmid and chromosomally
encoded virulence genes are required to produce disease in humans. The lack of either gene renders the cell avirulent,
and therefore, a test for either one can result in a false-positive result if applied separately. It is not possible to
determine solely from probe or PCR results if a cell contained a virulence-associated plasmid while in a food sample
and then lost it during subsequent selection and enrichment. Some strains might harbor a virulence plasmid and be
defective in one or more chromosomal virulence factors; therefore, it would appear nonpathogenic.

Because multiple probes or primer sets (a multiplex PCR) can be used simultaneously, it would be a reasonable
strategy to test for the presence of both plasmid-encoded and chromosomally encoded genes. With a PCR-based test,
the size(s) of the product(s) would indicate which genes are present; gene probes could be labeled with different
enzymes that use chromogenic substrates distinguishable by color. The presence of either gene alone might indicate
pathogenic potential and may signal some cause for concern, but only strains harboring both chromosomal and plasmid
borne genes are considered pathogenic.

Viruses

Approximately 5% of confirmed foodborne outbreaks in the United States during 19831987 are attributed to viruses
(Bean et al., 1990). However, the extent of virus-caused disease may be grossly underestimated (Archer and Kvenberg,
1985) because of the scarcity of rapid and sensitive techniques to detect foodborne viruses. To further complicate the
situation, the infectious doses of some viruses are very low (Ward et al., 1986). Several of the tests used to detect
viruses rely upon tissue culture cell lines; however, some foodborne viruses of considerable public health import, such
as hepatitis A virus, rotaviruses, and Norwalk agent, are difficult to grow in the laboratory. Newer genetically based
techniques for the detection and identification of foodborne viruses are becoming more important for determining the
presence and the titers of these viruses in foods.

Tests using gene probes and/or PCR are being developed. When gene probes are used to detect PCR-amplified DNA,
sensitivity can be less than 10 virus particles, but in many cases, sensitivity is limited by the procedure used

 Page 150

to prepare samples. Although a RNA-containing virus cannot be detected directly by PCR, its genome can be copied
into a strand of DNA (cDNA) by using a reverse transcriptase, which serves as a template strand, and the segment is
then amplified as in the PCR.

Norwalk Virus

In the United States this virus may be the leading cause of nonbacterial gastroenteritis in adults (Kaplan et al., 1982).
However, Norwalk virus is difficult to detect because it cannot be grown in cell culture. A reverse transcriptase PCR
(RT-PCR) has been used to detect Norwalk virus in stools (Jiang et al., 1992) and in shellfish (Zhou et al., 1991). RT-
PCR sensitivity was judged to be 100-fold higher than the use of gene probes alone, which can detect between 104 and
105 hepatitis A virus particles (Jiang et al., 1987). Atmar et al. (1996) were able to boost the sensitivity of an RT-PCR-
based test by confirming PCR product with hybridization to a virus specific probe. Nirdnoy et al. (1995) used a
rotavirus-specific probe and a monoclonal enzyme immunoassay to detect virus in stool.
Small round structured viruses (SRSV) that resemble Norwalk viruses were detected using RT-PCR and subsequent
detection of the PCR product with Southern blotting and hybridization with a probe targeted to an internal region of the
amplified segment (Lees et al., 1995). Shellfish associated with four outbreaks of human gastroenteritis were positive.
Some SRSVs are difficult to detect because of nucleotide sequence diversity within the group. Ando et al. (1995)
devised primer sets for Norwalk viruses and Snow Mountain agent and then used four sets of internal nonisotopically
labeled gene probes to classify the viruses.

Rotavirus
An AP-labeled oligonucleotide probe was used in a dot blot format to detect as little as 1 ng of rotavirus RNA
(Yamakawa et al., 1989). In this system, 1 ng was equivalent to about 5 × 107 copies. By using individual genes as
probes, Lin et al. (1987) were able to subgroup rotaviruses isolated from humans. Based on a radioactive riboprobe
used under stringent hybridization conditions and the separation of hybrids by gel electrophoresis, three groups of
human rotaviruses were defined (Nakagomi et al., 1989). Biotinylated RNA probes have also been used (Bellinzoni et
al., 1989). A commercially available nonradioactive oligonucleotide probe for rotavirus was compared with
immunologically based kits (Arens and Swierkosz, 1989). All had sensitivities and specificities above 90%. Similar
results were reported by Olive and Sethi (1989), Eiden et al. (1987), and Lin et al. (1985).

Rosen et al. (1992) produced serotype-specific cDNA molecules by using RT-PCR to generate gene probes to detect
rotaviruses of porcine origin. Gouvea et al. (1990) used RT-PCR targeted to the major outer capsid glycoprotein gene
(vp7) for the detection of rotavirus in fecal samples from humans with a sensitivity of about 2 ng (equivalent to 108
viral genomes).

 Page 151

However, purification of templates by binding of rotaviral RNA to a CF11 cellulose column in buffered ethanol
increased sensitivity more than 1000-fold presumably through the removal of inhibitors (Wilde et al., 1990). A PCR
typing method was developed in which each serotype resulted in a characteristic size fragment.

To increase the sensitivity of PCR-based assays for Norwalk virus detection, De Leon et al. (1992) used an RT-PCR-OP
(oligoprobe) method. After removing RT-PCR inhibitors in fecal specimens by Sepahdex G-200 gel chromatography,
two sets of primers were used to amplify regions of the genes for RNA polymerase and a putative immunogenic
protein. The amplification products were detected by hybridization with a digoxigenin-labeled oligonucleotide. A stool
sample containing about 106 virions/mL as determined by immunoelectron microscopy and a human infectivity titer of
about 104 infectious doses per milliliter could be diluted 104-fold and 10 mL still gave a positive amplification result.
Therefore, the sensitivity seems to be close to one infectious unit, which might represent 100 virions.

Hepatitis A Virus (HAV)

A number of seafoods (such as oysters) are consumed raw and can spread HAV to humans it grown in contaminated
waters (Desenclos et al., 1991). HAV is easily transmitted by the fecal-oral route, the period of communicability
precedes the onset of clinical symptoms, and it can survive up to several months in experimentally contaminated
environments. These factors make it difficult to control the spread of the disease.

There appears to be nucleotide sequence diversity among strains of HAV isolated from human and nonhuman (Lemon
et al., 1987) sources, which suggests the construction of universal gene probes to detect these viruses. Nevertheless, a
probe for a dot blot assay to monitor waterborne HAV has been developed (Metcalf and Jiang, 1988). A radioactive
cDNA probe was used to detect HAV in stool samples (Sjogren et al., 1987) and in estuarine samples (Jiang et al.,
1986). In the latter case, humic acids caused only slight decreases in sensitivity, and interference by organic
components could be lessened by proteinase K treatment and phenol extraction.

As most methods for obtaining viral nucleic acids from clinical samples are time-consuming, Jansen et al. (1990)
selectively captured virus particles in fecal samples by using anti-HAV monoclonal antibodies, released viral RNA by
heating, and amplified it by RT-PCR. Southern blot analysis of PCR products using gene probes yielded a detection
rate of 90% with a sensitivity of 330 virus particles. A comparison of solid-phase radioimmunoassay and cDNA-RNA
hybridization resulted in 50 and 65% positive samples, respectively. More recently Deng et al. (1994) used antigen
capture and RT-PCR to test seeded oyster samples. Using probes to detect the PCR products, a sensitivity of about four
virus particles was obtained.

 Page 152

Atmar et al. (1993) attempted to detect HAV in seeded oysters using a sample preparation method developed for
Norwalk virus; HAV could not be detected by RT-PCR when added to whole oysters but could be detected after being
spiked into extracts of oysters. It was determined that HAV was removed from the supernatant during the flocculation
step. This finding illustrates the difficulty in extrapolating methods from one microbe and sample type to others.

A single-stranded RNA probe specific for HAV had a sensitivity of 501000 infectious units and was used to detect the
virus in water samples during an outbreak (Shieh et al., 1991). LeGuyader et al. (1993), using riboprobes, found that
67% of shellfish examined contained HAV RNA. There was no correlation between viral contamination and fecal
coliform counts. Jehl-Pietri et al. (1993) compared immune electron microscopy (IEM), radioimmunoassay (RIA), and
hybridization with a digoxigenin-labeled cDNA probe for the detection of human HAV in raw and treated sewage. RIA
appeared the least sensitive, and no relationship was established between the presence of HAV and enterovirus
concentration.

Enteroviruses

These RNA-containing viruses are a diverse group of pathogens and common causes of childhood infections, the most
notable of which is polio. It is difficult to detect these viruses with gene probes as titers of virus are usually low in
clinical samples. Because the complete nucleotide sequences of several enteroviruses have been determined, RT-PCR
methods have been applied (Rotbart, 1990). A few highlights of this work will be mentioned.

A 154 bp conserved region the 5' end of the enteroviral genome was targeted. Gene probe assay sensitivity is in the
range of 106 molecules (about 104 particles capable of growth in tissue culture); more importantly, the time required to
identify these viruses was shortened from 48 days to less than 1 day. Using poliovirus seeded into oysters as a test
system, Atmar et al. (1993) found flocculation and polyethylene glycol precipitation left RT-PCR inhibitors in the
extracts, which could be removed by CTAB allowing detection of as few as 10 plaque-forming units. Similar results
were reported by Lees et al. (1994). By using positively charged nylon membranes to capture enteroviruses from water
samples, Gilgen et al. (1995) isolated RNA with oligonucleotides bound to magnetic beads. The detection limit was 20
TCID50, and 7 of 40 (17%) of water samples tested were positive.

Summary and Conclusions

Gene probe-based tests have been applied to detect and characterize most of the important pathogenic foodborne
bacteria and viruses. However, most

 Page 153

of the reported work is based on clinical applications and environmental testing. Nevertheless, the past decade has seen
a dramatic increase in the development of genetically based methods designed for the testing of foods for the presence
of disease-causing microorganisms.

Much knowledge regarding the mechanisms of pathogenicity, as well as genetic and nucleotide sequence information,
has been obtained from these microorganisms, which greatly facilitates the development of gene probe assays. Many
test formats have been developed. Among the most promising are those that couple DNA amplification with gene
probe detection of the amplified products. In many cases, sensitivity has reached the level of a single cell.

Detection sensitivity continues to improve as new nonradioactive labels are introduced; thus, concerns regarding the
use of radioisotopes have decreased. Still, a number of problems remain, the most challenging of which might be
sample preparation. Innovative techniques need to be developed to foster further improvement in rapid, convenient,
and sensitive assays for foodborne pathogens. If the increases in genetically based methods noted during this past
decade continue, these tests will play important roles in our understanding of foodborne disease and will speed
progress toward the improvement of food safety.

Acknowledgments

The authors thank Dr. Marleen Wekell, Director, Seafood Products Research Center, FDA Bothell, Washington, for
support during this endeavor. We appreciate the assistance provided by the interlibrary loan section of the Center for
Food Safety and Applied Nutrition Library, FDA, Washington D.C., and the Center for Academic Computing,
University of Washington, Seattle, Washington. Center secretary, Nancy Hill, provided some much needed assistance.

References

Abeyta, C., Jr., Deeter, F.G., Kaysner, C.A., Stott, R.F., and Wekell, M.M. 1993. Campylobacter jejuni in a Washington
State shellfish growing bed associated with illness. J. Food Prot. 56: 323325.

Agrawal, S., Christodoulou, C., and Gait, M.J. 1986. Efficient methods for attaching nonradioactive labels to the 5'
ends of synthetic oligodeoxy-ribonucleotides. Nucleic Acids Res. 14: 62276245.

Altwegg, M., Hickman-Brenner, F.W., and Farmer, III, J.J. 1989. Ribosomal

 Page 154

RNA gene restriction patterns provide increased sensitivity for typing Salmonella typhi strains. J. Infect. Dis. 160:
145149.

Anderson, N.G., Anderson, N.L., Taylor, J., and Goodman, J. 1995. Large-scale oligonucleotide synthesizers. I. Basic
principles and system design. Appl. Biochem. Biotechnol. 54: 1942.

Ando, T., Monroe, S., Gentsch, J.R., Jin, Q., Lewis, D.C., and Glass, R.I. 1995. Detection and differentiatioin of
antigenically distinct small round-structured viruses (Norwalk-like viruses) by reverse transcription-PCR and Southern
hybridization. J. Clin. Microbiol. 33: 6471.
Archer, D.L. and Kvenberg, J.E. 1985. Incidence and cost of foodborne diarrheal disease in the United States. J. Food
Prot. 48: 887894.

Arens, M. and Swierkosz, E.M. 1989. Detection of rotavirus by hybridization with a nonradioactive synthetic DNA
probe and comparison with commercial enzyme immunoassays and silver-stained polyacrylamide gels. J. Clin.
Microbiol. 27: 12771279.

Ash, C. and Collins, M.D. 1992. Comparative analysis of 23S ribosomal RNA gene sequences of Bacillus anthracis
and emetic Bacillus cereus determined by PCR-direct sequencing. FEMS Microbiol. Lett. 73: 7580.

Atmar, R.L., Metcalf, T.G., Neill, F.H., and Estes, M.K. 1993. Detection of enteric viruses in oysters by using the
polymerase chain reaction. Appl. Environ. Microbiol. 59: 631635.
Atmar, R.L., Neill, F.H., Woodley, C.M., Manger, R., Fout, G.S., Burkhardt, W., Leja, L., McGovern, E.R., Le
Guyader, R., Metcalf, T.G., and Estes, M.K. 1996. Collaborative evaluation of a method for the detection of Norwalk
virus in shellfish tissues by PCR. Appl. Environ. Microbiol. 62: 254258.

Baez, L.A. and Juneja, V.K. 1995. Nonradioactive colony hybridization assay for detection and enumeration of
enterotoxigenic Clostridium perfringens in raw beef. Appl. Environ. Microbiol. 61: 807810.

Bassler, H.A., Flood, S.J.A., Livak, K.J., Marmaro, J., Knorr, R., and Batt, C.A. 1995. Use of a fluorgenic probe in a
PCR-based assay for the detection of Listeria monocytogenes. Appl. Environ. Microbiol. 61: 37243728.

Bean, N.H., Griffin, P.M., Golding, J.S., and Ivey, C.B. 1990. Foodborne disease outbreaks, 5-year summary,
19831987. MMWR 39: 1557.

Becker, H., Schaller, G., von Wiese, W., and Terplan, G. 1994. Bacillus cereus in infant foods and dried milk products.
Int. J. Food Microbiol. 23: 115.
Beebakahee, G., Louie, M., DeAzavedo, J., and Brunton, J. 1992. Cloning and nucleotide sequence of the eae gene
homologue from enterohemorrhagic Escherichia coli serotype O157:H7. FEMS Microbiol. Lett. 70: 6368.

 Page 155

Beecher, D.J. and Macmillan, J.D. 1991. Characterization of the components of hemolysin BL from Bacillus cereus.
Infect. Immun. 59: 17781784.

Begum, D., Strockbine, N.A., Sowers, E.G., and Jackson, M.P. 1993. Evaluation of a technique for identification of
Shiga-like toxin-producing Escherichia coli by using polymerase chain reaction and digoxigenin-labeled probes. J.
Clin. Microbiol. 31: 31533156.

Bej, A.K., Steffan, R.J., DiCesare, J., Haff, L., and Atlas, R.M. 1990. Detection of coliform bacteria in water by
polymerase chain reaction and gene probes. Appl. Environ. Microbiol. 56: 307314.
Bej, A.K., McCarty, S.C., and Atlas, R.M. 1991. Detection of coliform bacteria and Escherichia coli by multiplex
polymerase chain reaction: Comparison with defined substrate and plating methods for water quality monitoring. Appl.
Environ. Microbiol. 57: 429432.

Bellinzoni, R., Xi, J.A., Tanka, T.N., Scodeller, E., and Estes, M.K. 1989. Rotavirus gene detection with biotinylated
single-stranded RNA probes. Mol. Cell. Probes 3: 233244.

Bentley, R.W., Harland, N.M., Leigh, J.A., and Collins, M.D. 1993. A Staphylococcus aureus-specific oligonucleotide
probe derived from 16S rRNA gene sequences. Lett. Appl. Microbiol. 16: 203206.

Berche, P., Poyart, C., Abachin, E., Lelievere, H., Vandepitte, J., Dodin, A., and Fournier, J.M. 1994. The novel
epidemic strain O139 is closely related to the pandemic strain O1 of Vibrio cholerae. J. Infect. Dis. 170: 701704.
Bessesen, M.T., Luo, Q., Rotbart, H.A., Blaser, M.J., and Ellison, R.T., III. 1990. Detection of Listeria monocytogenes
by using the polymerase chain reaction. Appl. Environ. Microbiol. 56: 29302932.

Bettelheim, K.A., Brown, J.E., Lolekha, S., and Echeverria, P. 1990. Serotypes of Escherichia coli that hybridized with
DNA probes for genes encoding Shiga-like toxin I, Shiga-like toxin II, and serogroup O157 enterhemorrhagic E. coli
fimbriae isolated from adults with diarrhea in Thailand. J. Clin. Microbiol. 28: 293295.

Beutin, L., Geier, D., Zimmermann, S., and Karch, H. 1995. Virulence markers of Shiga-like toxin-producing
Escherichia coli strains originating from healthy domestic animals of different species. J. Clin. Microbiol. 33: 631635.

Bhattachary, S.K., Bhattacharya, M.K., Ramamurth, T., Pal, A., Bag, P.K., Takeda, T., Shimada, T., Dutta, P., Debnath,
A., and Chakraborti, S. 1992. Acute secretory tavellers' diarrhea caused by Vibrio cholerae non-O1 which does not
produce cholera-like or heat-stable enterotoxins. J. Diarrhaoeal Dis. Res. 10: 161163.

 Page 156

Binz, T., Kurazono, H., Willie, M., Frevert, J., Wernars, K., and Hiemann, H. 1990. The complete sequence of
botulinum neurotoxin type A and comparison with other clostridial neurotoxins. J. Biol. Chem. 265: 91539158.

Blais, B.W. 1994. Transcriptional enhancement of the Listeria monocytogenes PCR and simple immunoenzymatic
assay of the product using anti-RNA:DNA antibodies. Appl. Environ. Microbiol. 60: 348352.

Blais, B.W. and Phillippe, L.M. 1993. A simple RNA probe system for analysis of Listeria monocytogenes polymerase
chain reaction products. Appl. Environ. Microbiol. 59: 27952800.

Boileau, C.R., D'Hauteville, H.M., and Sansonetti, P.J. 1984. DNA hybridization technique to detect Shigella species
and enteroinvasive Escherichia coli. Clin. Microbiol. 20: 959961.

Bollum, F.J. 1974. Terminal deoxynucleotidyl transferase. In: The Enzymes, P.D. Boyer, (Ed.), Vol. 10, Ch. 5.
Academic Press Inc., New York.

Borriello, S.P. 1995. Clostridial disease of the gut. Clin. Infect. Dis. 20: s242s250.

Bourque, S.N., Valero, J.R., Lavoie, M.C., and Levesque, R.C. 1995. Comparative analysis of the 16S to 23S
ribosomal intergenic spacer sequences of Bacillus thuringiensis strains and subspecies and of closely related species.
Appl. Environ. Microbiol. 61: 16231626.
Brauns, L.A., Hudson, M.C., and Oliver, J.D. 1991. Use of the polymerase chain reaction in detection of culturable and
nonculturable Vibrio vulnificus cells. Appl. Environ. Microbiol. 57: 26512655.

Brenner, D.J., Fanning, G.R., Rake, A.V., and Johnson, K.E. 1969. Batch procedure for thermal elution of DNA from
hydroxyapatite. Anal. Biochem. 28: 447459.

Brian, M.J., Frosolono, M., Murray, B.E., Miranda, A., Lopez, E.L., Gomez, H.F., and Cleary, T.G. 1992. Polymerase
chain reaction for diagnosis of enterohemorrhagic Escherichia coli infection and hemolytic-uremic syndrome. J. Clin.
Microbiol. 30: 18011806.

Britten, R.J. and Kohne, D.E. 1968. Repeated sequences in DNA. Science 161: 529540.

Bruce, J.L., Hubner, R.J., Cole, E.M., McDowell, C.I., and Webster, J.A. 1995. Sets of EcoRI fragments containing
ribosomal RNA sequences are conserved among different strains of Listeria monocytogenes. Proc. Natl. Acad. Sci.
USA 92: 55295533.
Bulte, M. and Jakob, P. 1995. The use of a PCR-generated invA probe for the detection of Salmonella spp. in
artificially and naturally contaminated foods. Int. J. Food Microbiol. 26: 335344.

 Page 157

Burnens, A., Stucki, U., Nicolet, J., and Frey, J. 1995. Identification and characterization of an immunogenic outer
membrane protein of Campylobacter jejuni. J. Clin. Microbiol. 33: 226232.

Butzler, J.P. and Oosterom, J. 1991. Campylobacter. Pathogenicity and significance in foods. Int. J. Food Microbiol.
12: 18.

Calderwood, S., Acheson, D.W.K., Keusch, G.T., Barrett, T.J., Griffin, P.M., Strockbine, N.A., Swaminathan, B.,
Kaper, J.B., Levine, M.M., Kaplan, B.S., Karch, H., O'Brien, A.D., Obrig, T.G., Takeda, Y., Tarr, P.I., and Wachsmuth,
I.K. 1996. Proposed new nomenclature for SLT (VT) family. ASM News 62: 118119.
Cano, R.J., Rasmussen, S.R., S'anchez-Fraga, G., and Palomares, J.C. 1993. Fluorescent detection polymerase chain
(FD-PCR) assay on microwell plates as a screening test for Salmonellas in foods. J. Appl. Bacteriol. 75: 247253.

Cardullo, R.A., Agrawal, S., Flores, C., Zamecnik, P.C., and Wolf, D.E. 1988. Detection of nucleic acid hybridization
by nonradioactive fluorescence resonance energy transfer. Proc. Natl. Acad. Sci. USA 85: 87908794.
Carey, P.E., Eley, A., and Wilcox, M.H. 1994. Assessment of a chemiluminescent universal probe for taxonomical and
epidemiological investigations of Aeromonas spp. isolates. J. Clin. Pathol. 47: 642646.

Centers for Disease Control and Prevention. 1994. Bacillus cereus food poisoning associated with fried rice at two
child day care centersVirginia, 1993. JAMA 27: 1.

Chollet, A. and Kawashima, E.H. 1985. Biotin-labeled synthetic oligodeoxy-ribonucleotides: Chemical synthesis and
uses as hybridization probes. Nucleic Acids Res. 13: 15291541.

Chopra, A.K., Houston, C.W., and Kurosky, A. 1991a. Genetic variation in related cytolytic toxins produced by
different species of Aeromonas. FEMS Microbiol. Lett. 78: 231238.

Chopra, A.K., Peterson, J.W., Houston, C.W., Pericas, R., and Prasad, R. 1991b. Enterotoxin-associated DNA sequence
homology between Salmonella species and Escherichia coli. FEMS Microbiol. Lett. 61: 133138.

Chopra, A.K., Houston, C.W., Peterson, J.W., and Jin, G.F. 1993. Cloning, expression, and sequence analysis of a
cytolytic enterotoxin gene from Aeromonas hydrophila. Can. J. Microbiol. 39: 513523.

Chuma, T., Yamada, T., Okamoto, K., Yugi, H., and Ohya, T. 1993. Application of a DNA-DNA hybridization method
for detection of Campylobacter jejuni in chicken feces. J. Vet. Med. Sci. 55: 10271029.
Coelho, A., Monen, H., Vicente, A.C., and Salles, C.A. 1994. An analysis of

 Page 158

the V1 and V2 regions of Vibrio cholerae and Vibrio mimicus 16S rRNA. Res. Microbiol. 145: 151156.

Colwell, R.R., Brayton, P.R., Grimes, D.J., Roszak, D.B., Huq, S.A., and Palmer, L.M. 1985. Viable but nonculturable
Vibrio cholerae and related pathogens in the environment: Implications for release of genetically engineered
microorganisms. Biotechnology 3: 817820.

Cordovez, A., Prado, V., Maggi, L., Cordero, J., Martinez, J., Misraji, A., Rios, R., Soza, G., Ojeda, A., and Levine,
M.M. 1992. Enterohemorrhagic Escherichia coli associated with hemolyticuremic syndrome in Chilean children. J.
Clin. Microbiol. 30: 21532157.
Cornillot, E., Saint-Joanis, B., Daube, G., Katayama, S., Granum, P.E., Canard, B., and Cole, S.T. 1995. The
enterotoxin gene (cpe) of Clostridium perfringens can be chromosomal or plasmid-borne. Mol. Microbiol. 15: 639647.

Cudjoe, K.S., Kapperud, G., and Olsvik, Ø. 1991. Identification of thermo-tolerant Campylobacter species from
poultry using an enzyme-labelled oligonucleotide DNA probe. Int. J. Food Microbiol. 12: 271274.
Curiale, M.S., McIver, D., Weathersby, S., and Planer, C. 1990. Detection of Salmonellae and other Enterobacteriaciae
by commercial deoxyribonucleic acid hybridization and enzyme immunoassay kits. J. Food. Prot. 53: 10371046.

D'Aoûst, J.Y., Sewell, A.M., Greco, P., Mazola, M.A., and Colvin, R.E. 1995. Performance assessment of the Gene-
Trak colorimetric probe assay for the detection of foodborne Salmonella spp. J. Food Prot. 58: 10691076.

Datta, A.R., Wentz, B.A., and Hill, W.E. 1987. Detection of hemolytic Listeria monocytogenes by using DNA colony
hybridization. Appl. Environ. Microbiol. 53: 22562259.

Datta, A.R., Wentz, B.A., and Hill, W.E. 1988a. Identification and enumeration of beta-hemolytic Listeria
monocytogenes in naturally contaminated dairy products. JAOAC 71: 673675.

Datta, A.R., Wentz, B.A., Shook, D., and Trucksess, M.W. 1988b. Synthetic oligoribonucleotide probes for detection
of Listeria monocytogenes. Appl. Environ. Microbiol., 54: 29332937.

Datta, A.R., Wentz, B.A., and Russell, J. 1990. Cloning of the listeriolysin O gene and development of specific gene
probes for Listeria monocytogenes. Appl. Environ. Microbiol. 56: 38743877.

Datta, A.R., Moore, M.A., Wentz, B.A., and Lane, J. 1993. Identification and enumeration of Listeria monocytogenes
by nonradioactive DNA probe colony hybridization. Appl. Environ. Microbiol. 59: 144149.
De Leon, R., Matsui, S.M., Baric, R.S., Herrmann, J.E., Blacklow, N.R.,

 Page 159

Greenberg, H.B., and Sobsey, M.D. 1992. Detection of Norwalk virus in stool specimens by reverse transcriptase-
polymerase chain reaction and nonradioactive oligoprobes. J. Clin. Microbiol. 30: 31513157.

DeLong, E.F., Wickham, G.S., and Pace, N.R. 1989. Phylogenetic strains: ribosomal RNA-based probes for the
identification of single cells. Science 243: 13601363.

Delor, I., Kaeckenbeeck, A., Wauters, G., and Cornelis, G.R. 1990. Nucleotide sequence of yst, the Yersinia
enterocolitica gene encoding the heat-stable enterotoxin, and prevalence of the gene among pathogenic and
nonpathogenic yersiniae. Infect. Immun. 58: 29832988.
Deng, M.Y., Day, S.P., and Cliver, D.O. 1994. Detection of hepatitis A virus in environmental samples by antigen-
capture PCR. Appl. Environ. Microbiol. 60: 19271933.

DePaola, A., Hopkins, L.H., Peeler, J.T., Wentz, B., and McPhearson, R.M. 1990. Incidence of Vibrio
parahaemolyticus in U.S. coastal waters and oysters. Appl. Environ. Microbiol. 56: 299302.
Desenclos, J.-C.A., Klontz, K.C., Wilder, M.H., Nainan, O.V., Margolis, H.S., and Gunn, R.A. 1991. A multistate
outbreak of hepatitis A caused by the consumption of raw oysters. Am. J. Public Health. 81: 12681272.

Devereux, J., Haeberli, P., and Smithies, O. 1984. A comprehensive set of sequence analysis programs for the VAX.
Nucleic Acids Res. 12: 387395.

Donnenberg, M.S., Tacket, C.O., James, S.P., Losonsky, G., Nataro, J.P., Wasserman, S.S., Kaper, J.B., and Levine,
M.M. 1993. Role of the eaeA gene in experimental enteropathogenic Escherichia coli infection. J. Clin. Invest. 92:
14181424.

Dorsch, M., Lane, D., and Stackebrandt, E. 1992. Towards a phylogeny of the genus Vibrio based on 16S rRNA
sequences. Int. J. Syst. Bacteriol. 42: 5863.

Doyle, M.P. and Padhye, V.V. 1989. Escherichia coli. In: Foodborne Bacterial Pathogens, M.P. Doyle (Ed.). Marcel
Decker, Inc., New York.

Drobniewski, F.A. 1993. Bacillus cereus and related species. Clin. Microbiol. Rev. 6: 324338.

East, A.K. and Collins, M.D. 1993. Molecular characterization of DNA encoding 23S rRNA and 16S-23S rRNA
intergenic spacer regions of Aeromonas hydrophila. FEMS Microbiol. Lett. 106: 129133.
Eiden, J., Sato, S., and Yoken, R. 1987. Specificity of dot hybridization assay in the presence of rRNA for detection of
rotaviruses in clinical specimens. J. Clin. Microbiol. 25: 18091811.

Emond, E., Fliss, I., and Pandian, S. 1993. A ribosomal DNA fragment of

 Page 160

Listeria monocytogenes and its use as a genus-specific probe in an aqueousphase hybridization assay. Appl.
Environ. Microbiol. 59: 26902697.

Fach, P., Gibert, M., Griffais, R., Guillou, J.P., and Popoff, M.R. 1995. PCR and gene probe identification of botulinum
neurotoxin A-, B-, E-, F-, and G-producing Clostridium spp. and evaluation in food samples. Appl. Environ.
Microbiol. 61: 389392.

Feinberg, A.P. and Vogelstein, B. 1983. A technique for radiolabeling DNA restriction endonuclease fragments to high
specific activity. Anal. Biochem. 132: 613.
Feng, P. 1992. Identification of invasive Yersinia species using oligonucleotide probes. Mol. Cell. Probes 6: 291297.

Feng, P. 1993. Identification of Escherichia coli serotype O157:H7 by DNA probe specific for an allele of uidA gene.
Mol. Cell. Probes 7: 151154.

Fenwick, S.G. and Murray, A. 1991. Detection of pathogenic Yersinia enterocolitica by polymerase chain reaction.
Lancet 337: 496497.

Ferreira, J.L., Hamdy, M.K., McCay, S.G., and Baumstark, B.R. 1992. An improved assay for identification of type A
Clostridium botulinum using the polymerase chain reaction. J. Rapid Methods Autom. Microbiol. 1: 2939.
Ferreira, J.L., Hamdy, M.K., McCay, S.G., Hemphill, M., Kirma, N., and Baumstark, B.R. 1994. Detection of
Clostridium botulinum type F using the polymerase chain reaction. Mol. Cell. Probes 8: 365373.

Fitts, R., Diamond, M., Hamilton, C., and Neri, M. 1983. DNA-DNA hybridization assay for detection of Salmonella
spp. in foods. Appl. Environ. Microbiol. 46: 11461151.

Flamm, R.K., Hinrichs, D.J., and Thomashow, M.F. 1989. Cloning of a gene encoding a major selected polypeptide of
Listeria monocytogenes and its potential use as a species-specific probe. Appl. Environ. Microbiol. 55: 22512256.

Flowers, R.S., Mazola, M., Curiale, M.S., Gabis, D.A., and Silliker, J.H. 1987. Comparative study of DNA
hybridization method and the conventional culture procedure for detection of Salmonella in foods. J. Food Sci. 52:
781785.

Fluit, A.C., Widjojoatmodjo, M.N., Box, A.T.A., Torensma, R., and Verhoef, J. 1993. Rapid detection of salmonellae
in poultry with the magnetic immuno-polymerase chain reaction assay. Appl. Environ. Microbiol. 59: 13421346.
Forster, A.C., McInnes, J.L., Skingle, D.C., and Symons, R.H. 1985. Non-radioactive hybridization probes prepared by
the chemical labeling of

 Page 161

DNA and RNA with a novel reagent, photobiotin. Nucleic Acids Res. 13: 745761.

Freier, S.M., Roszak, D.B., Risen, L.A., Marich, J.E., and Wong, D.F. 1989. A non-radioactive, highly sensitive
oligonucleotide probe to detect Campylobacter. Dupont Biotech Update 4: 1416.

Galan, J.E. and Curtiss III, R. 1991. Distribution of the invA, -B, -C and -D genes of Salmonella typhimurium among
other Salmonella serovars: invA mutants of Salmonella typhi are deficient for entry into mammalian cells. Infect.
Immun. 59: 29012908.

Galindo, I., Rangel-Aldao, R., and Ramirez, J.L. 1993. A combined polymerase chain reaction-colour development
hybridization assay in a microtitre format for the detection of Closoidium spp. Appl. Microbiol. Biotechnol. 39:
553557.

Geilhausen, B., Mauff, G., Vlaes, L., Goossens, H., and Butzler, J.P. 1990. Restriction fragment length polymorphism
for the identification of Campylobacter jejuni isolates. Int. J. Med. Microbiol. 274: 366371.
Gicquelois, K.G., Baldini, M.M., Martinez, J., Maggi, L., Martin, W.C., Prado, V., Kaper, J.B. and Levine, M.M. 1990.
Practical and economical method for using biotinylated DNA probes with bacterial colony blots to identify, diarrhea-
causing Escherichia coli. J. Clin. Microbiol. 28: 24852490.

Giesendorf, B.A.J., van Belkum, A., Koeken, A., Stegeman, H., Henkens, M.H.C., van der Plas, J., Goossens, H.,
Niesters, H.G.M., and Quint, A.G.V. 1993. Development of species-specific DNA probes for Campylobacter jejuni,
Campylobacter coli, and Campylobacter lari by polymerase chain reaction fingerprinting. J. Clin. Microbiol. 31:
15411546.

Gilgen, M., Wegmüller, B., Burkhalter, P., Buhler, H.P., M&umml;ller, U., Lüthy, J., and Candrian, U. 1995. Reverse
transcription PCR to detect enteroviruses in surface water. Appl. Environ. Microbiol. 61: 12261231.

Gouvea, V., Glass, R.I., Woods, P., Taniguchi, K., Clark, J.F., Forrester, B., and Fang, Z.-Y. 1990. Polymerase chain
reaction amplification and typing of rotavirus nucleic acid from stool specimens. J. Clin. Microbiol. 28: 276282.

Goverde, R.L.,Jansen, W.H., Brunings, H.A., Huis in 't Veld, J.H., and Mooi, F.R. 1993. Digoxigenin-labelled inv- and
ail-probes for the detection and identification of pathogenic Yersinia enterocolitica in clinical specimens and naturally
contaminated pig samples. J. Appl. Bacteriol. 74 301313.

Granum, P.E., Brynestad, S. and Kramer, J.M. 1993. Analysis of enterotoxin production by Bacillus cereus from dairy
products, food poisoning incidents and non-gastrointestinal infections. Int. J. Food Microbiol. 17: 269279.

 Page 162

Graves, L.M., Swaminathan, B., Reeves, M.W. and Wenger, J. 1991. Ribosomal DNA fingerprinting of Listeria
monocytogenes using a digoxigenin-labeled DNA probe. Eur. J. Epidemiol. 7: 7782.

Graves, L.M., Swaminathan, B., Reeves, M.W., Hunter, S.B., Weaver, R.E., Plikaytis, B.D., and Shuchat, A. 1994.
Comparison of ribotyping and multilocus enzyme electrophoresis for subtyping of Listeria monocytogenes isolates. J.
Clin. Microbiol. 32: 29362943.

Grimm, L.M., Goldoft, M., Kobayashi, J., Lewis, J.H., Alfi, D., Perdichizzi, A.M., Tarr, P.I., Ongerth, J.E., Moseley,
S.L., and Samadpour, M. 1995. Molecular epidemiology of a fast-food restaurant-associated outbreak of Escherichia
coli O157:H7 in Washington state. J. Clin. Microbiol. 33: 21552158.
Grimont, F. and Grimont, P.A.D. 1986. Ribosomal ribonucleic acid gene restriction patterns as potential taxonomic
tools. Ann. Inst. Pasteur Microbiol. 137B: 165175.

Grünstein, M. and Hogness, D.S. 1975. Colony hybridization: method for the isolation of cloned DNAs that contain a
specific gene. Proc. Natl. Acad. Sci. 72: 39613965.

Gunzer, F., Bohm, H., Russman, H., Aleksic, S., and Karch, H. 1992. Molecular detection of sorbitol-fermenting
Escherichia coli O157 in patients with hemolytic-uremic syndrome. J. Clin. Microbiol. 30: 18071810.

Hall, B.D. and Spiegelman, S. 1961. Sequence complementarity of T2-DNA and T2-specific RNA. Proc. Nat. Acad.
Sci. Wash. 47: 137146.
Hanes, D.E., Koch, W.H., Miliotis, M.D., and Lampel, K.A. 1995. DNA probe for detecting Salmonella enteritidis in
food. Mol. Cell. Probes 9: 918.

Hayes, P.S., Blom, K, Feng, P., Lewis, J., Strockbine, N.A., and Swaminathan, B. 1995. Isolation and characterization
of a b-D-glucoronidase-producing strain of Escherichia coli serotype O157:H7 in the United States. J. Clin. Microbiol.
33: 33473348.

Hengen, P.N. 1994. Recovering DNA from agarose gels. Trends Biochem. Sci. 19: 388389.

Heidelburg, J.F., O'Neill, K.R., Jacobs, D., and Colwell, R.R. 1993. Enumeration of Vibrio vulnificus on membrane
filters with a fluorescently labeled oligonucleotide probe specific for kingdom-level 16S rRNA sequences. Appl.
Environ. Microbiol. 59: 474476.

Heller, M.J. and Morrison, L.E. 1985. Chemiluminescent and fluorescent probes for DNA hybridization systems. In:
Rapid Detection and Identification of Infectious Agents, p. 245256. Academic Press, San Diego, Calif.
Herman, L. and De Ridder, H. 1992. DNA hybridization and amplification

 Page 163

techniques to accelerate the identification of Listeria monocytogenes. Milchwissenschaft 47: 354357.

Hill, W.E. 1981. DNA hybridization method for detecting enterotoxigenic Escherichia coli in human isolates and its
possible application to food samples. J. Food Safety 3: 233247.

Hill, W.E. and Carlisle, C.L. 1981. Loss of plasmids during enrichment for Escherichia coli. Appl. Environ. Microbiol.
41: 10461048.

Hill, W.E. and Jinneman, K.C. 1996. Detection of microorganisms in foods: Principles and application of genetic
techniques as these relate to the detection of food-associated microorganisms. In: The Microbiology of Food. B.M.
Lund, A.C. Baird-Parker, and G.W. Gould, (Ed.), Ch. 47. Chapman & Hall, London.

Hill, W.E., Payne, W.L., and Aulisio, C.C.G. 1983. Detection and enumeration of virulent Yersinia enterocolitica in
food by colony hybridization. Appl. Environ. Microbiol. 46: 636641.

Hill, W.E., Ferreira, J.L., Payne, W.L., and Jones, V.M. 1985. Probability of recovering pathogenic Escherichia coli
from foods. Appl. Environ. Microbiol. 49: 13741378.

Hill, W.E., Wentz, B.A., Payne, W.L., Jagow, J.J., and Zon, G. 1986. DNA colony hybridization method using synthetic
oligonucleotides to detect enterotoxigenic Escherichia coli: Collaborative study. JAOAC 69: 531536.
Hill, W.E., Keasler, S.P., Trucksess, M.W., Feng, P., Kaysner, C.A., and Lampel, K. 1991. Polymerase chain reaction
identification of Vibrio vulnificus in artificially contaminated oysters. Appl. Environ. Microbiol. 57: 707711.

Hirono, I. and Aoki, T. 1991. Nucleotide sequence and expression of an extracellular hemolysin gene of Aeromonas
hydrophila. Microb. Pathog. 11: 189197.

Hoge, C.W., Sethabutr, O., Bodhidatta, L., Echeverria, P., Robertson, D.C., and Morris, J.G., Jr. 1990. Use of a
synthetic oligonucleotide probe to detect strains of nonserovar O1 Vibrio cholerae carrying the gene for heat-stable
enterotoxin (NAG-ST). J. Clin. Microbiol. 28: 14731476.

Hubner, R.J., Cole, E.M., Bruce, J.L., McDowell, C.I., and Webster, J.I. 1995. Predicted types of Listeria
monocytogenes created by the positions of EcoRI cleavage sites relative to rRNA sequences. Proc. Natl. Acad. USA
92: 52345238.

Huck, L.G., Dorn, C.R., and Angrick, E.J. 1995. DNA probe for detection of serogroup O157 enterohemorrhagic
Escherichia coli. Int. J. Food Microbiol. 25: 277287.
Hutchinson, D.N. 1992. Foodborne botulism. Br. Med. J. 305: 264265.

 Page 164

Ibrahim, A., Liesack, W., Pike, S. and Stackebrandt, E. 1992a. The polymerase chain reaction: An epidemiological tool
to differentiate between two clusters of pathogenic Yersinia enterocolitica strains. FEMS Microbiol. Lett. 76: 6366.

Ibrahim, A., Liesack, W. and Stackebrandt, E. 1992b. Differentiation between pathogenic and non-pathogenic Yersinia
enterocolitica strains by colony hybridization with a PCR-mediated digoxgenin-dUTP-labeled probe. Mol. Cell. Probes
6: 163171.

Jablonski, E., Moomaw, E.W., Tullis, R.H., and Ruth, J.L. 1986. Preparation of oligonucleotide-alkaline phosphatase
conjugates and their use as hybridization probes. Nucleic Acids Res. 14: 61156128.
Jackson, M.P. 1991. Detection of Shiga toxin-producing Shigella dysenteriae type I and Escherichia coli by using
polymerase chain reaction with incorporation of digoxigenin-11-dUTP. J. Clin. Microbiol. 29: 19101914.

Jagow, J.A. and Hill, W.E. 1986. Enumeration by DNA colony hybridization of virulent Yersinia enterocolitica
colonies in artificially contaminated food. Appl. Environ. Microbiol. 51: 441443.
Jansen, R.W., Siegl, G., and Lemon, S.M. 1990. Molecular epidemiology of human hepatitis A virus defined by an
antigen-capture polymerase chain reaction method. Proc. Natl. Sci. USA 87: 28672871.

Jaton, K., Sahli, R., and Bille, J. 1992. Development of polymerase chain reaction assays for detection of Listeria
monocytogenes in clinical cerebrospinal fluid samples. J. Clin. Microbiol. 30: 19311936.

Jaulhac, B., Bes, M., Bornstein, N., Píemont, Y., Brun, Y., and Fleurette, J. 1992. Synthetic DNA probes for detection
of genes for enterotoxins A, B, C, D, E, and for TSST-1 in staphylococcal strains. J. Appl. Bacteriol. 72: 386392.

Jehl-Pietri, C., Hughes, B., Andre, M., Diez, J.M., and Bosch, A. 1993. Comparison of immunological and molecular
hybridization detection methods for the detection of hepatitis A virus in sewage. Lett. Appl. Microbiol. 17: 162166.

Jerse, A.E., Martin, W.C., Galen, J.E., and Kaper, J.B. 1990a. Oligonucleotide probe for detection of the
enteropathogenic Escherichia coli (EPEC) adherence factor of localized adherent EPEC. J. Clin. Microbiol. 28:
28422844.

Jerse, A.E., Yu, J., Tall, B.D., and Kaper, J.B. 1990b. A genetic locus of enteropathogenic Escherichia coli necessary
for the production of attaching and effacing lesions of tissue culture cells. Proc. Natl. Acad. Sci. 87: 78397843.

Jerse, A.E., Gicquelais, K.G., and Kaper, J.B. 1991. Plasmid and chromo-

 Page 165

somal elements involved in the pathogenisis of attaching and effacing Escherichia coli. Infect. Immun. 59:
38693875.

Jiang, X., Estes, M.K., Metcalf, T.G., and Melnick, J.L. 1986. Detection of hepatitis A virus in seeded estuarine
samples by hybridization with cDNA probes. Appl. Environ. Microbiol. 52: 711717.

Jiang, X., Estes, M.K., and Metcalf, T.G. 1987. Detection of hepatitis A virus by hybridization with single-stranded
RNA probes. Appl. Environ. Microbiol. 53: 24872495.

Jiang, X., Wang, J., Graham, D.Y., and Estes, M.K. 1992. Detection of Norwalk virus in stool by polymerase chain
reaction. J. Clin. Microbiol. 30: 25292534.

Johnson, J.L. and Lattuada, C.P. 1993. Comparison of nucleic acid hybridization assays and biochemical
characterization tests for the confirmation of Listeria monocytogenes. J. Food Prot. 56: 834840.

Kaplan, J.E., Feldman, R., Campbell, D.S., Lookabaugh, C., and Gary, G.W. 1982. The frequency of a Norwalk-like
pattern of illness in outbreaks of acute gasteroenteritis. Am. J. Public Health 72: 13291332.

Kapperud, G., Dommarsnes, K., Skurnik, M., and Hornes, E. 1990. A synthetic oligonucleotide probe and a cloned
polynucleotide probe based on the yopA gene for detection and enumeration of virulent Yersinia enterocolitica. Appl.
Environ. Microbiol. 56: 1723.

Kapperud, G. 1991. Yersinia enterocolitica in food hygiene. Int. J. Food Microbiol. 12: 5365.

Karch, H. and Meyer, T. 1989. Single primer pair for amplifying segments of distinct Shiga-like-toxin genes by
polymerase chain reaction. J. Clin. Microbiol. 27: 27512757.

Karch, H., Bohm, H., Schmidt, H., Gunzer, F., Aleksic, S., and Heesemann, J. 1993. Clonal structure and pathogenicity
of Shiga-like toxin-producing, sorbitol-fermenting Escherichia coli O157:H-.J. Clin. Microbiol. 31: 12001205.

Karger, B.L., Chu, Y.H., and Foret, F. 1995. Capillary electrophoresis of proteins and nucleic acids. Annu. Rev.
Biophys. Biomol. Struct. 24: 579610.

Karmali, M.A., Penner, J.L., Fleming, P.C., Williams, A., and Hennessy, J.N. 1983. The serotype and biotype
distribution of clinical isolates of Campylobacter jejuni and Campylobacter coli over a three year period. J. Infect. Dis.
147: 243246.
Kaysner, C.A., Abeyta, C., Jr., Wekell, M.M., DePaola, A., Jr., Stott, R.F., and Leitch, J. 1987. Virulent strains of
Vibrio vulnificus isolated from estuaries

 Page 166

of the United States West Coast. Appl. Environ. Microbiol. 53: 13491352.

Keller, G.H., and Manak, M.M. 1989. In: DNA Probes, p. 1314. Stockton Press, New York.

Kessler, C. 1991. The digoxigenin:antidigoxigenin (DIG) technologya survey on the concept and realization of a novel
bioanalytical indicator system. Mol. Cell. Probes 5: 161205.

Kim, C., Graves, L.M., Swaminathan, B., Mayer, L.W., and Weaver, R.E. 1991a. Evaluation of hybridization
characteristics of a cloned pRF106 probe for Listeria monocytogenes. Detection and development of a nonisotopic
colony hybridization assay. Appl. Environ. Microbiol. 57: 289294.
Kim, C., Swaminathan, B., Cassaday, P.K., Mayer, L.W., and Holloway, B.P. 1991b. Rapid confirmation of Listeria
monocytogenes isolated from foods by colony blot assay using a digoxigenin-labeled synthetic oligonucleotide probe.
Appl. Environ. Microbiol. 57: 16091614.

King, V. and Clayton, C.L. 1991. Genomic investigation of phenotypic variation in Campylobacter jejuni flagellin.
FEMS Microbiol. Lett. 68: 107111.

King, W., Raposa, S., Warshaw, J., Johnson, A., Halbert, D., and Klinger, J.D. 1989. A new colorimetric nucleic acid
hybridization assay for Listeria in foods. Int. J. Food Microbiol. 8: 225232.
Kirov, S.M. 1993. The public health significance of Aeromonas spp. in foods. Int. J. Food Microbiol. 20: 179198.

Kishishita, M., Matsuoka, N., Kumagai, K., Yamasaki, S., Takeda, Y., and Nishibuchi, M. 1992. Sequence variation in
the thermostable direct hemolysin-related hemolysin (trh) gene of Vibrio parahaemolyticus. Appl. Environ. Microbiol.
58: 24492457.

Klinger, J.D. and Johnson, A.R. 1988. A rapid nucleic acid hybridization assay for Listeria in foods. Food. Technol. 42:
6670.

Knight, I.T., Shults, S., Kasper, C.W. and Colwell, R.R. 1990. Direct detection of Salmonella spp. in estuaries by using
a DNA probe. Appl. Environ. Microbiol. 56: 10591066.

Koeppel, E., Meyer, R., Lüethy, J., and Candrian, U. 1993. Recognition of pathogenic Yersinia enterocolitica by crystal
violet binding and polymerase chain reaction. Lett. Appl. Microbiol. 17: 231234.
Kohler, S., Leimeister-Wachter, M., Chakraborty, T., Lottspeich, F., and Goebel, W. 1990. The gene coding for protein
p60 of Listeria monocytogenes and its use as a specific probe for Listeria monocytogenes. Infect. Immun. 58:
19431950.

Kreig, P.A. and Melton, D.A. 1987. In vitro RNA synthesis with SP6 RNA polymerase. Methods Enzymol. 155:
397415.

 Page 167

Kricka, L.J. (Ed.) 1992. Nonisotopic DNA Probe Techniques. Academic Press, Inc., San Diego, Calif.

Krowczynska, A.M., Donoghue, K., and Hughes, L. 1995. Recovery of DNA, RNA and protein from gels with
microconcentrators. Biotechniques 18: 698703.

Kuhn, M. and Goebel, W. 1989. Identification of an extracellular protein of Listeria monocytogenes possibly involved
in intracellular uptake by mammalian cells. Infect. Immun. 57: 5561.

Kwaga, J., Iversen, J.O., and Misra, V. 1992. Detection of pathogenic Yersinia enterocolitica by polymerase chain
reaction and digoxigenin-labeled polynucleotide probes. J. Clin. Microbiol. 30: 26682673.
Lang, A.L., Tsai, Y.L., Mayer, C.L., Patton, KC., and Palmer, C.J. 1994. Multiplex PCR for detection of the heat-labile
toxin gene and Shiga-like toxin I and II genes in Escherichia coli isolated from natural waters. Appl. Environ.
Microbiol. 60: 31453149.

Lebens, M. and Holmgren, J. 1994. Structure and arrangement of the cholera toxin genes in Vibrio cholerae O139.
FEMS Microbiol. Lett. 117: 197202.

Lee, C., Chen, L.H., Liu, M.L., and Su, Y.C. 1992. Use of an oligonucleotide probe to detect Vibrio parahaemolyticus
in artificially contaminated oysters. Appl. Environ. Microbiol. 58: 34193422.
Lees, D.N., Henshilwood, K., and Dor'e, W.J. 1994. Development of a method for detection of enteroviruses in
shellfish by PCR with poliovirus as a model. Appl. Environ. Microbiol. 60: 2993005.

Lees, D.N., Henshilwood, K., Green, J., Gallimore, C.I., and Brown, D.W. 1995. Detection of small round structured
viruses in shellfish by reverse transcription-PCR. Appl. Environ. Microbiol. 61: 44184424.

LeGuyader, F., Apaire-Marchais, V., Brillet, J., and Billaudel, S. 1993. Use of genomic probes to detect hepatitis A
virus and enterovirus RNAs in wild shellfish and relationship of viral contamination to bacterial contamination. Appl.
Environ. Microbiol. 59: 39633968.

Leimeister-Wachter, M. and Chakraborty, T. 1989. Gene probes for the detection of Listeria sp. Acta Microbiol. Hung.
36: 211213.

Lemon, S.M., Chao, S.F., Jansen, R.W., Binn, L.N., and LeDuc, J.W. 1987. Genomic heterogeneity among human and
nonhuman strains of hepatitis A virus. J. Virol. 651: 735742.
Levine, M.M., Xu, J.G., Kaper, J.B., Lior, H., Prado, V., Tall, B., Natrero, J., Karch, J., and Wachsmuth, K. 1987. A
DNA probe to identify enterohemorrhagic Escherichia coli of O157:H7 and other serotypes that cause

 Page 168

hemorrhagic colitis and hemolitc uremic syndrome. J. Infect. Dis. 156: 175182.

Lin, M., Imai, M. Bellamy, A.R., Ikegami, N., Furuichi, Y., Summers, D., Nuss, D.L., and Deibel, R. 1985. Diagnosis
of rotavirus infection with cloned cDNA copies of viral genome segments. J. Virol. 55: 509512.

Lin, M., Imai, M., Ikegami, N., Bellamy, R., Summers, D., Nuss, D.L., Deibel, R., and Furuichi, Y. 1987. cDNA probes
of individual genes of human rotavirus distinguish viral subgroups and serotypes. J. Virol. Methods 15: 285289.

Lo, Y.-M.D., Mehal, W.Z., and Fleming, K.A. 1988. Rapid production of vector-free biotinylated probes using the
polymerase chain reaction. Nucleic Acids Res. 16: 8719.
Louie, M., de Azavedo, J.C., Handelsman, M.Y., Clark, C.J., Ally, B., Dytoc, M., Sherman, P., and Brunton, J. 1993.
Expression and characterization of the eaeA gene product of Escherichia coli serotype O157:H7. Infect. Immun. 61:
40854092.

Louie, M., de-Azavedo, J., Clarcke, R., Borczyk, A., Lior, H., Richter, M., and Brunton, J. 1994. Sequence
heterogeneity of the eae gene and detection of verotoxin-producing Escherichia coli using serotype-specific primers.
Epidemiol. Infect. 112: 449461.

Luby, S., Jones, J., Dowda, H., Kramer, J., and Horan, J. 1993. A large outbreak of gastroenteritis caused by diarrheal
toxin-producing Bacillus cereus. J. Infect. Dis. 167: 14521455.

Luk, J.M.C. 1994. A PCR enzyme immuno-assay for detection of Salmonella typhi. BioTechniques 17: 10381042.

Maas, R. 1983. An improved colony hybridization method with significantly increased sensitivity for detection of
single genes. Plasmid 10: 296298.

Makino, S., Kurazono, T., Okuyama, Y., Shimada, T., Okada, Y., and Sasakawa, C. 1995. Diversity of DNA sequences
among Vibrio cholerae O139 Bengal detected by PCR-based DNA fingerprinting. FEMS Microbiol. Lett. 126: 4348.

Martin, I.E., Tyler, S.D., Tyler, K.D., Khakhria, R., and W.M. Johnson. 1996. Evaluation of ribotyping as
epidemiologic tool for typing Escherichia coli serogroup O157 isolates. J. Clin. Microbiol. 34: 720723.

Martinetti-Lucchini, G. and Altwegg, M. 1991. Use of ribosomal RNA gene restriction (rDNA) patterns as taxomonic
tools in the genus Aeromonas. Abstr. Ann. Mtg. Amer. Soc. Microbiol., p. 374.
Maurelli, A.T. and Lampel, K.A. 1994. Shigella. In: Foodborne Disease Handbook: Vol. 1., J.R. Groman and K.D.
Murrell (Ed.), p. 319343. Marcel Dekker, Inc., New York.

 Page 169

McKay, A.M. 1992. Viable but non-culturable forms of potentially pathogenic bacteria in water. Lett. Appl. Microbiol.
14: 129135.

McKee, C.R., Perez, J.C., Gipson, M.V., Noah, C.W., and Yetts, S.Y. 1989. Detecting Listera spp. in food samples
using the Gene-Trak isotopic Listeria assay DNA hybridization procedure and the FDA cultural procedure. FDA Lab.
Inform. Bull. 3348: 112.

McLauchlin, J., Saunders, N.A., Ridley, A.M., and Taylor, A.G. 1988. Listeriosis and food-borne transmission. Lancet
i: 177178.
Meinkoth, J. and Wahl, G. 1984. Hybridization of nucleic acids immobilized on solid supports. Anal. Biochem. 138:
267284.

Melton, D.A., Krieg, P.A., Rebagliati, M.R., Maniatis, T., Zinn, K., and Green, M.R. 1984. Efficient in vitro synthesis
of biologically active RNA and RNA hybridization probes from plasmids containing a bacteriophage SP6 promoter.
Nucleic Acids Res. 12: 70357056.

Mengaud, J., Vincente, M.F., Chenevert, J., Pereira, J.M., Geoffroy, C., Gicquel-Sanzey, B., Baquero, F., Perez-Diaz,
J.C., and Cossart, P. 1988. Expression in Escherichia coli and sequence analysis of the listeriolysin O determinant of
Listeria monocytogenes. Infect. Immun. 56: 766772.

Metcalf, T.G. and Jiang, X. 1988. Detection of hepatitis A virus in estuarine samples by gene probe assay. Microbiol.
Sci. 5: 296300.

Miceli, G.A., Watkins, W.D., and Rippey, S.R. 1993. Direct plating procedure for enumerating Vibrio vulnificus in
oysters (Crassostrea virginica). Appl. Environ. Microbiol. 59: 35193524.

Miliotis, M.D., Galen, J.E., Kaper, J.B., and Morris, J.G. Jr. 1989. Development and testing of a synthetic
oligonucleotide probe for the detection of pathogenic Yersinia strains. J. Clin. Microbiol. 27: 16671670.

Miller, V.L., Farmer, J.J., III, Hill, W.E., and Falkow, S. 1989. The ail locus is found uniquely in Yersinia
enterotocolitica serotypes commonly associated with disease. Infect. Immun. 57: 121131.

Miyagi, K., Matsumoto, Y., Hayashi, K., Yoh, M., Yamamoto, K., and Honda, T. 1994. Successful application of
enzyme-labeled oligonucleotide probe for rapid and accurate cholera diagnosis in a clinical laboratory. Microbiol.
Immunol. 38: 301304.

Miyamoto, Y., Kato, T., Obara, Y., Akigama, S., Takizawa, K., and Yamai, S. 1969. In vitro hemolytic characteristic of
Vibrio parahaemolyticus: Its close correlation with human pathogenicity. J. Bacteriol. 100: 11471149.
Mohamood, A., Datta, A.R., and Eribo, B.E. 1992. Application of synthetic Listeriolysin O gene probe to the
identification of veta-hemolytic Listeria monocytogenes in retail ground beef. J. Food. Prot. 55: 385388.

Morgan, S., and Jones, D.D. 1995. Detection of viable and nonviable Vibrio

 Page 170

cholerae in artificially contaminated shellfish using polymerase chain reaction. Abstr. Ann. Mtg. Am. Soc.
Microbiol., p. 391.

Morris, C.E., Klement, J.F., and McAllister, W.T. 1986. Cloning and expression of the bacteriophage T3 RNA
polymerase gene. Gene. 41: 193200.

Morris, J.G., Jr., Wright, A.C., Roberts, D.M., Wood, P.K., Simpson, L.M., and Oliver, J.D. 1987. Identification of
environmental Vibrio vulnificus isolates with a DNA probe for the cytotoxin-hemolysin gene. Appl. Environ.
Microbiol. 43: 193195.
Moseley, S.L. and Falkow, S. 1980. Nucleotide sequence homology between the heat-labile enterotoxin gene of
Escherichia coli and Vibrio cholerae deoxyribonucleic acid. J. Bacteriol. 144: 444446.

Moseley, S.L., Echeverria, P., Seriwatana, J., Tirapat, C., Chaicumpa, W., Sakulaipeara, T., and Falkow, S. 1982.
Identification of enterotoxigenic Escherichia coli by colony hybridization using three enterotoxin gene probes. J.
Infect. Dis. 145: 863867.

Muhlegger, K., Huber, E., von der Eltz, H., Ruger, R., and Kessler, C. 1990. Nonradioactive labeling and detection of
nucleic acids: IV. Synthesis and properties of the nucleotide compounds of the digoxigenin system and of
photodigoxigenin. Mol. Gen. Hoppe-Seyler 371: 939951.

Nakagomi, O., Nakagomi, T., Akatani, K., and Ikegami, N. 1989. Identification of rotavirus genogroups by RNA-RNA
hybridization. Mol. Cell. Probes 3: 251261.

Nastasi, A., Mammina, C., and Villafrate, M.R. 1991. rDNA fingerprinting as a tool in epidemiological analysis of
Salmonella typhi infections. Epidemiol. Infect. 107: 565576.

Natero, J.P., Baldini, M.M., Kaper, J.B., Black, R.E., Bravo, N., and Levine, M.M. 1985. Detection of an adherence
factor of enteropathogenic Escherichia coli with a DNA probe. J. Infect. Dis. 132: 560565.

Neill, R.J., Fanning, G.R., Delahoz, F., Wolff, R., and Gemski, P. 1990. Oligonucleotide probes for detection and
differentiation of Staphylococcus aureus strains containing genes for enterotoxins A, B, and C and toxic shock
syndrome toxin 1. J. Clin. Microbiol. 28: 15141518.

Neill, M.A., Tarr, P.I., Taylor, D.N., and Trofa, A.F. 1994. Escherichia coli. In: Foodborne Disease Handbook, Y.H.
Hui, J.R. Gorham, K.D. Murrell, and D.O. Cliver. (Ed.), Ch. 7. Marcel Dekker, Inc., New York.

Nesbakken, T., Kapperud, G., Dommarsnes, K., Skurnik, M., and Hornes, E. 1991. Comparative study of a DNA
hybridization method and two isolation procedures for detection of Yersinia enterocolitica, O:3 in naturally
contaminated pork products. Appl. Environ. Microbiol. 57: 389394.
Nilsson, L., Oliver, J.D., and Kjellberg, S. 1991 Resuscitation of Vibrio vul-

 Page 171

nificus from the viable but nonculturable state. J. Bacteriol. 173: 50545059.

Ninet, B., Bannerman, E. and Bille, J. 1992. Assessment of the Accuprobe Listeria monocytogenes culture
identification reagent kit for rapid colony confirmation and its application in various enrichment broths. Appl. Environ.
Microbiol. 58: 40554059.

Nirdnoy, W., Sethabutr, O., Sakulkaipeara, T., Harikul, S., Hoge, C.W., and Echeverria, P. 1995. Typing of human
group A rotavirus with alkaline phosphatase-labeled oligonucletoide probes or a monoclonal enzyme immunoassay in
unfrozen stools of children with diarrhea in Bangkok. J. Med. Virol. 45: 117120.
Nishibuchi, M. and Kaper, J.B. 1985. Nucleotide sequence of the thermostable direct hemolysin gene of Vibrio
parahaemolyticus. J. Bacteriol. 162: 558564.

Nishibuchi, M., Ishibashi, M., Takeda, Y. and Kaper, J.B. 1985. Detection of the thermostable direct hemolysin gene
and related DNA sequences in Vibrio parahaemolyticus and other vibrio species by the DNA colony hybridization test.
Infect. Immun. 49: 481486.

Nishibuchi, M., Hill, W.E., Zon, G., Payne, W.L., and Kaper, J.B. 1986. Synthetic oligodeoxyribonucleotide probes to
detect Kanagawa phenomenon-positive Vibrio parahaemolyticus. J. Clin. Microbiol. 23: 10911095.

Nishibuchi, M., Taniguchi, T., Misawa, Khaeomanee-lam V., Honda, T., and Miwatani, T. 1989. Cloning and
nucleotide sequence of the gene (trh) encoding the hemolysin related to the thermostable direct hemolysin of Vibrio
parahaemolyticus. Infect. Immun. 57: 26912697.

Nolan, L.K., Giddings, C.W., and Brown, J. 1995. The distribution of invA, pagC and spvC genes among Salmonella
isolates from animals. Vet. Res. Commun. 19: 167177.

Nuijten, P.J., van-Asten, F.J., Gaastra, W., and van der Zeijst, B.A. 1990. Structural and functional analysis of two
Campylobacter jejuni flagellin genes. J. Biol. Chem. 265: 1779817804.

Oberhelman, R.A., Kopecko, D.J., Vendatesan, M.M., Salazar-Lindo, E., Gotuzzo, E., Yi, A., Chea-Woo, E., Ruiz, R.,
Fernandez-Prada, C., Leon-Barua, R., and Sack, R.B. 1993. Evaluation of alkaline phosphatase-labelled ipaH probe for
diagnosis of Shigella infections. J. Clin. Microbiol. 31: 21012104.

Olive, D.M. and Sethi, S.K. 1989. Detection of human rotavirus by using an alkaline phosphatase-conjugated synthetic
DNA probe in comparison with enzyme-linked immunoassay and polyacrylamide gel analysis. J. Clin. Microbiol. 27:
5357.

 Page 172

Olive, D.M., Johny, M., and Sethi, S.K. 1990. Use of an alkaline phosphatse-labeled synthetic oligonucleotide probe
for detection of Campylobacter jejuni and Campylobacter coli. J. Clin. Microbiol. 28: 15651569.

Oliver, J.D., Nilsson, L., and Kjellberg, S. 1991 Formation of nonculturable Vibrio vulnificus cells and its relationship
to the starvation state. Appl. Environ. Microbiol. 57: 26402644.

Oliver, J.D., Guthrie, K., Preyer, J., Wright, A., Simpson, L.M., Siebeling, R., and Morris, J.G., Jr. 1992. Use of
colistin-polymixin B-cellobiose agar for isolation of Vibrio vulnificus from the environment. Appl. Environ. Microbiol.
58: 737739.
Olsen, J.E., Aabo, S., Nielsen, E.O., and Nielsen, B.B. 1991. Isolation of a Salmonella-specific DNA hybridization
probe. APMIS 99: 114120.

Olsvik, Ø., Wahlberg, J., Petterson, B., Uhlén-M., Popovic, T., Wachsmuth, I.K., and Fields, P.I. 1993. Use of
automated sequencing of polymerase chain reaction-generated amplicons to identify three types of choleratoxin
subunit B in Vibrio cholerae O1 strains. J. Clin. Microbiol. 31: 2225.

Osteroff, S.M., Kobayashi, J.M., and Lewis, J.H. 1989. Infections with Escherichia coli O157:H7 in Washington State.
JAMA 262: 355359.

Ou, C.Y., McDonough, S.H., Cabanas, D., Ryder, T.B., Harper, M., Moore, J., and Schochetman, G. 1990. Rapid and
quantitative detection of enzymatically amplified HIV-1 DNA using chemiluminescent oligonucleotide probes. AIDS
Res. Hum. Retroviruses 6: 13231329.

Oyofo, B.A., Thornton, S.A., Burr, D.H., Trust, T.J., Pavlovskis, O.R., and Guerry, P. 1992. Specific detection of
Campylobacter jejuni and Campylobacter coli by using polymerase chain reaction. J. Clin. Microbiol. 30: 26132619.

Pal, A., Ramamurthy, T., Bhadra, R.K., Takeda, T., Shimada, T., Takeda, Y., Nair, G.B., Pal, S.C., and Chakrabarti, S.
1992. Reassessment of the prevalence of heat-stable enterotoxin (NAG-ST) among environmental Vibrio cholerae non-
O1 strains isolated from Calcutta, India, by using a NAG-ST probe. Appl. Environ. Microbiol. 58: 24852489.

Panda, C.S., Riley, L.W., Kumari, S.N., Khanna, K.K., and Prakash, K. 1990. Comparison of alkaline phosphatase-
conjugated oligonucleotide DNA probe with the Serèny test for identification of Shigella strains. J. Clin. Microbiol. 28:
21222124.

Park, R.W.A., Griffiths, P.L., and Moreno, G.A. 1991. Sources and survival of Campylobacters: relevance to enteritis
and the food industry. J. Appl. Bacteriol. Symp. Suppl. 70: 97S106S.

Paros, M., Tarr, P.I., Kim, H., Besser, T.E., and Hancock, D.D. 1993. A comparison of human and bovine Escherichia
coli O157:H7 isolates by

 Page 173

toxin genotype, plasmid profile and bacteriophage restriction fragment length polymorphism profile. J. Infect. Dis.
168: 13001303.

Pierson, D.E. and Falkow, S. 1990. Nonpathogenic isolates of Yersinia enterocolitica do not contain functional inv-
homologous sequences. Infect. Immun. 58: 10591064.

Popovic, T., Olsvik, Ø., Blake, P., and Wachsmuth, K. 1993. Cholera in the Americas: Foodborne aspects. J. Food Prot.
56: 811821.

Portnoy, D.A. and Falkow, S. 1981. Virulence-associated plasmids from Yersinia enterocolitica and Yersinia pestis. J.
Bacteriol. 148: 877883.

Portnoy, D.A., Moseley, S.L., and Falkow, S. 1981. Characterization of plasmids and plasmid-associated determinants
of Yersinia enterocolitica pathogenesis. Infect. Immun. 31: 775782.

Pumfrey, L. and Nelson, C.E. 1991. Use of a most probable number method modified with a deoxyribonucleic probe to
monitor control by food preservatives of natural Salmonella contamination in animal meat meals. Poultry Sci. 70:
780784.

Rahn, K., DeGrandis, S.A., Clarke, R.C., McEwen, S.A., Gal'an, J.E., and Ginocchio, C., Curtiss, R. and Gyles, C.L.
1992. Amplification of an invA gene sequence of Salmonella typhimurium by polymerase chain reaction as a specific
method of detection of Salmonella. Mol. Cell. Probes. 6: 271279.
Reed, G.H. 1994a. Foodborne Illness: Clostridium perfringerns gastroenteritis. Dairy Food Environ. Sanit. 14: 1617.

Reed, G.H. 1994b. Foodborne illness: Bacillus cereus gastroenteritis. Dairy Food Environ. Sanit. 14: 87.

Regan, C.M., Syed, Q. and Tunstall, P.J. 1995. A hospital outbreak of Clostridium perfringens food
poisoningimplications for food hygiene review in hospitals. J. Hosp. Infect. 29: 6973.

Renz, M. and Kurz, C. 1984. A colorimetric method for DNA hybridization. Nucleic Acids Res. 12: 34353444.

Rigby, P.W.J., Dieckmann, M., Rhodes, C., and Berg, P. 1977. Labeling deoxyribonucleic acid to high specificity by
nick translation with DNA polymerase I.J. Mol. Biol. 113: 237257.
Robins-Browne, R., Miliotis, M.D., Cianciosi, S., Miller, V.L., Falkow, S., and Glenn-Morris, J.G. 1989. Evaluation of
DNA colony hybridization and other techniques for detection of virulence in Yersinia species. J. Clin. Microbiol. 27:
644650.

Robins-Browne, R.M., Takeda, T., Fasano, A., Bordun, A.M., Dohi, S., Kasuga, H., Fang, G., Prado, V., and Guerrant,
R.L. 1993. Assessment of

 Page 174

enterotoxin production by Yersinia enterocolitica and identification of a novel heat-stable enterotoxin projected by
a noninvasive Y. enterocolitica strain isolated from clinical material. Infect. Immun. 61: 764767.

Rose, B.E., Llabres, C.M., and Bennett, B. 1991. Evaluation of a colorimetric DNA hybridization test for detection of
Salmonellae in meat and poultry products. J. Food Prot. 54: 127130.

Rosen, B.I., Parwani, A.V., Gorziglia, M., Larralde, G., and Saif, L.J. 1992. Characterization of full-length and
polymerase chain reaction-derived partial-length Gottfried and OSU gene 4 probes for serotypic differentiation of
porcine rotaviruses. J. Clin. Microbiol. 30: 26442652.
Roszak, D.B. and Colwell, R.R. 1987. Survival strategies of bacteria in the natural environment. Microbial Rev. 51:
365379.

Rotbart, H.A. 1990. Enzymatic RNA amplification of the enteroviruses. J. Clin. Microbiol. 28: 438442.

Roychoudhury, R., Tu, C.-P.D., and Wu, R. 1979. Influence of nucleotide sequence adjacent to duplex DNA termini on
3-terminal labeling by terminal transferase. Nucleic Acids Res. 6: 13231333.

Rychlik, W. and Rhoads, R.E. 1989. A computer program for choosing optimal oligonucleotides for filter
hybridization, sequencing and in vitro amplification of DNA. Nuc. Acids Res. 17: 85438551.
Saez-Llorens, X., Guzman-Verduco, L.M., Shelton, S., Nelson, J.D., and Kupersztoch, Y.M. 1989. Simultaneous
detection of Escherichia coli heat-stable and heat-labile enterotoxin genes with a single RNA probe. J. Clin. Microbiol.
27: 16841688.

Said, B., Scotland, S.M., and Rowe, B. 1994. The use of gene probes, immunoassays and tissue culture for the
detection of toxin in Vibrio cholerae non-O1. J. Med. Microbiol. 40: 3136.

Samadpour, M., Liston, J., Ongerth, J.E., and Tarr, P.I. 1990. Evaluation of DNA probes for detection of Shiga-like-
toxin-producing Escherichia coli in food and calf fecal samples. Appl. Environ. Microbiol. 56: 12121215.

Samadour, M., Grimm, L.M., Desai, B., Alfi, D., Ongerth, J., and Tarr, P.I. 1993. Molecular epidemiology of
Escherichia coli O157:H7 strains by bacteriophage lambda restriction fragment length polymorphism analysis:
application to a multistate foodborne outbreak and a day-care center cluster. J. Clin. Microbiol. 31:31793183.

Samadpour, M., Ongerth, J.E., Liston, J., Tran, N., Nguyen, C., Whittam, T.S., Wilson, R.A., and Tarr, P.I. 1994.
Occurence of Shiga-like toxin-producing Escherichia coli in retail fresh seafood, beef, lamb, pork and poultry from
grocery stores in Seattle, Washington. Appl. Environ. Microbiol. 60: 10381040.

 Page 175

Samadpour, M. 1995. Molecular epidemiology of Escherichia coli O157:H7 by restriction fragment length
polymorphism using shiga-like toxin genes. J. Clin. Microbiol. 33: 21502154.

Sanger, F., Nicklen, S., and Coulson, A.R. 1992. DNA sequencing with chain-terminating inhibitors. 1977.
Biotechnology 24: 104108.

Schlapp, T., Blaha, I., Bauerfeind, R., Wieler, L.H., Schoepe, H., Weiss, R., and Baljer, G. 1995. Synthesis and
evaluation of a non-radioactive gene probe for the detection of C. perfringens alpha toxin. Mol. Cell. Probes 9: 101109.

Schmidt, H., Russmann, H., and Karch, H. 1993. Virulence determinants in nontoxinogenic Escherichia coli O157
strains that cause infantile diarrhea. Infect. Immun. 61: 48944898.

Schmidt, H., Karch, H., and Beutin, L. 1994. The large-sized plasmids of enterohemorrhagic Escherichia coli O157
strains encode hemolysins which are presumably members of the E. coli alpha-hemolysin family. FEMS Microbiol.
Lett. 117: 189196.
Schmidt, H., Beutin, L., and Karch, H. 1995. Molecular analysis of the plasmid-encoded hemolysin of Escherichia coli
O157:H7 strain EDL 933. Infect. and Immun. 63: 10551061.

Scholl, D.R., Kaufmann, C., Jollick, J.D., York, C.K., Goodrum, G.R., and Charache, P. 1990. Clinical application of
novel sample processing technology of the identification of Salmonellae by using DNA probes. J. Clin. Microbiol. 28:
237241.
Schraft, H. and Griffith, M.W. 1995. Specific oligonucleotide primers for detection of lecithinase-positive Bacillus spp.
by PCR. Appl. Environ. Microbiol. 61: 98102.

Schultz, C., Pool, G.J., van Ketel, R., de Wever, B., Speelman, P., and Dankert, J. 1994. Detection of enterotoxigenic
Escherichia coli in stool samples by using nonradioactively labeled oligonucleotide DNA probes and PCR. J. Clin.
Microbiol. 32:23932937.

Scotland, S.M., Willshaw, G.A., Smith, H.R., Said, B.A., Stokes, N., and Rowe, B. 1993. Virulence properties of
Escherichia coli strains belonging to seogroups O26, O55, O111 and O128 isolated in the United Kingdom in 1991
from patients with diarrhoea. Epidemiol. Infect. 111: 429438.

Seibl, R., Holtke, H.-J., Ruger, R., Meindl, A., Zachau, H.-G., RaBhofer, G., Roggendorf, M., Wolf, H., Arnold, N.,
Weinberg, J., and Kessler, C. 1990. Nonradioactive labeling and detection of nucleic acids: III. Application of
digoxigenin system. Mol. Gen. Hoppe-Seyler. 371: 939951.

Shane, S.M. 1992. The significance of Campylobacter jejuni infection in poultry: a review. Avian Pathol. 21: 189212.

 Page 176

Sharma, N.K., Gill, J.P., Joshi, D.V., and Kwatra, M.S. 1994. Bacteriological analysis of uncooked pork sausages in
relation to public health. Indian J. Pathol. Microbiol. 37: 191195.

Sheldon, E.L., Kellogg, D.E., Watson, R.E., Levinson, C.H., and Erlich, H.A. 1985. Process for labeling nucleic acids
using psoralen derivatives. Eur. Patent Appl. 0156287.

Shieh, Y.S., Baric, R.S., Sobsey, M.D., Ticehurst, J., Miele, T.A., DeLeon, R., and Walter, R. 1991. Detection of
hepatitis A virus and other enteroviruses in water by ssRNA probes. J. Virol. Methods 31: 199136.

Shirai, H., Ito, H., Hirayama, T., Nakamoto, Y., Nakabayashi, N., Kumagai, I., Takeda, Y., and Nishibuchi, M. 1990.
Molecular epidemiologic evidence for association of thermostable direct hemolysin (TDH) and TDH-related
hemolysin of Vibrio parahaemolyticus with gastroenteritis. Infect. Immun. 58: 35683573.

Sjogren, M.H., Tanno, H., Fay, O., Sileoni, S., Cohen, B.D., Burke, D.S., and Geighny, R.J. 1987. Hepatitis A virus in
stool during clinical relapse. Ann. Intern. Med. 106: 221226.
Slaten, D.D., Oropeza, R.I., and Werner, S.B. 1992. An outbreak of Bacillus cereus food poisoningare caterers
supervised sufficiently? Public Health Rep. 107: 477480.

Small, P.C. and Falkow, S. 1986. Development of a DNA probe for the virulence plasmid of Shigella spp. and
enteroinvasive Escherichia coli. In: Microbiology1986, L. Leive, P.F. Bonventre, S.A. Morello, S.D. Silver, and Wu,
W.C. (Ed.), p. 121124. American Society for Microbiology, Washington, D.C.
Soumet, C., Ermel, G., Bouth, P., Boscher, E., and Colen, P. 1995. Chemiluminescent and colorimetric enzymatic
assays for the detection of PCR amplified Salmonella spp. products in microplates. BioTechniques 19: 792796.

Southern, E.M. 1975. Detection of specific sequences among DNA fragments separated by gel electrophoresis. J. Mol.
Biol. 98: 503510.

Stanley, J., Baquar, N. and Burnens, A. 1995. Molecular subtyping scheme for Salmonella panama. J. Clin. Microbiol.
33: 12061211.

Stelma, G.N., Johnson, C.H. and Spaulding, P. 1986. Evidence for the direct involvement of beta-hemolysin in
Aeromonas hydrophila enteropathogenicity, Curr. Microbiol. 14: 7177.

Stern, N.J. and Mozola, M.A. 1992. Methods for selective enrichment of Campylobacter spp. from poultry for use in
conjuction with DNA hybridization. J. Food Prot. 55: 767770.
Stone, G.G., Oberst, R..D., Hays, M.P., McVey, S., and Chengappa, M.M.

 Page 177

1994. Detection of Salmonella serovars from clinical samples by enrichment broth cultivation-PCR procedure. J.
Clin. Microbiol. 32: 17421749.

Stucki, U., Frey, J., Nicolet, J., and Burnens, A.P. 1995. Identification of Campylobacter jejuni on the basis of a
species-specific gene that encodes a membrane protein. J. Clin. Microbiol. 33: 855859.

Sutherland, A.D. 1993. Toxin production by Bacillus cereus in dairy products. J. Dairy Res. 60: 569574.

Swaminathan, B., Graves, L.M., Carlone, G.M., and Plikaytis, B.D. 1988. Molecular epidemiology of Listeria
monocytogenes by riborosmal RNA typing. Abst. Ann. Meet. Am. Soc. Microbiol. D-61:81.
Symons, R.H. (Ed.). 1989. Nucleic Acid Probes. CRC Press Inc., Boca Raton, Fla.

T'oth, I., Karcagi, V., Nagy, B., Gad'o, I., and Milch, H. 1994. Examination of verocytotoxin producing capacity and
determination of the presence of Shiga-like toxin genes in human Escherichia coli isolates. Acta Microbiol. Immunol.
Hung. 41: 259264.

Tackett, C.O., Brenner, F., and Blake, P.A. 1984. Clinical features and an epidemiological study of Vibrio vulnificus
infections. J. Infect. Dis. 149: 558561.

Tada, J., Ohashi, T., Nishimura, N., Ozaki, H., Fukushima, S., Takano, J., Nishibuchi, M., and Takeda, Y. 1992. Non-
isotopic microtitre plate-based assay for detecting products of polymerase chain reaction amplification: application to
detection of the tdh gene of Vibrio parahaemolyticus. Mol. Cell. Probes 6: 489494.

Takeda, T., Peina, Y., Ogawa, A., Dohi, S., Abe, H., Mair, G.B., and Pal, S.C. 1991. Detection of heat-stable
enterotoxin in a cholera toxin gene-positive strain of Vibrio cholerae O1. FEMS Microbiol. Lett. 64: 2327.

Tauxe, R.V. 1991. Epidemiology of Campylobacter infections in the United States and other industrialized nations. In:
Campylobacter jejuni: Current Status and Future Trends, Nachamkin, I., Blaser, M.J. and Tompkins, L.S. (Ed.), p. 919.
American Society for Microbiology, Washington, D.C.

Tauxe, R..V. and Blake, P.A. 1992. Epidemic cholera in Latin America. JAMA 267: 13881390.

Taylor, J.M., Illmensee, R., and Summers, J. 1976. Efficient transcription of RNA into DNA by avian sarcoma virus
polymerase. Biochim. Biophys. Acta 442: 324.
Taylor, D.N., Echeverria, P., Sethabutr, O., Pitarangsi, C., Leksomboon, U., Blacklow, N.R., Rowe, B., Gross, R., and
Cross, J. 1988. Clinical and microbiologic features of Shigella and enteroinvasive Escherichia coli infections detected
by DNA hybridization. J. Clin. Microbiol. 26: 13621366.

 Page 178

Taylor, D.E. and Hiratsuka, K 1990. Use of non-radioactive DNA probes for detection of Campylobacter jejuni and
Campylobacter coli in stool specimens. 4: 261271.

Taylor, D.N. 1991. Campylobacter infections in developing countries. In: Campylobacter jejuni: Current Status and
Future Trends, I. Nachamkin, M.J. Blaser, and L.S. Tompkins (Ed.), p. 2030. American Society for Microbiology,
Washington, D.C.

Tenover, F.C., Carlson, L., Barbagallo, S., and Nachamkin, I. 1990. DNA probe culture confirmation assay for
identification of themophilic Campylobacter species. J. Clin. Microbiol. 28: 12841287.
Thomas, A., Smith, H.R., Willshaw, G.A., and Rowe, B. 1991. Non radioactively labelled polynucleotide and
oligonucleotide DNA probes, for selectively detecting Escherichia coli strains producing Vero cytotoxins VT1, VT2
and VT2 variant. Mol. Cell Probes 5: 129135.

Thomas, A., Smith, H.R., and Rowe, B. 1993. Use of digoxigenin-labelled oligonucleotide DNA probes for VT2 and
VT2 human variant genes to differentiate Vero cytotoxin-producing Escherichia coli strains of serogroup O157. J.
Clin. Microbiol. 31: 17001703.

Thompson, D.E., Brehm, J.K., Oultram, J.D., Swinfield, T.J., Shone, C.C., Atkinson, T., Melling, J., and Minton, N.P.
1990. The complete amino acid sequence of Clostridium botulinum type A neurotoxin, deduced by nucleotide
sequence analysis of the encoding gene. Eur. J. Biochem. 189: 7381.

Thompson, S.A. and Blaser, M.J. 1995. Isolation of the Helicobacterpylori recA gene and involvement of the recA
region in resistance to low pH. Infect. Immun. 63: 21852193.

Thorne, G.M., Macone, A., and Goldmann, D.A. 1990. Enzymatically labelled nucleic acid (NA) probe assays for
detection of Campylobacter spp. in human faecal specimens and in culture. Mol. Cell. Probes 4: 133142.

Tsen, H.Y., Wang, S.J., and Green, S.S. 1991. Salmonella detection in meat and fish by membrane hybridization with
chromogenic/phosphatase/biotin DNA probe. J. Food Sci. 56: 15191523.

Tu, C.-P.D. and Cohen, S. 1980. 3'-End labeling of DNA with [32P]cordycepin-5-triphosphate. Gene 10:177183.

Url, B., Heitzer, A., and Brandl, E. 1993. Determination of Listeria in dairy and environmental samples: Comparison
of a cultural method and a colorimetric nucleic acid hybridization assay. J. Food Prot. 56: 581584.

Usera, M.A., Echeita, A., Olsvik, Ø., Evins, G.M., Cameron, D.N., and Popovic, T. 1994. Molecular subtyping of
Vibrio cholerae O1 strains recently

 Page 179

isolated from patient, food and environmental samples in Spain. Eur. J. Clin. Microbiol. Inf. Dis. 13: 299303.

Van Damme-Jongsten, M., Rodhouse, J., Gilber, R.J., and Notermans, S. 1990. Synthetic DNA probes for detection of
enterotoxigenic Clostridium perfringens strains isolated from outbreaks of food poisoning. J. Clin. Microbiol. 28:
131133.

Venkatesan, M., Buysse, J.M., Vandendries, E., and Kopecko, D.J. 1988. Development and testing of invasion-
associated DNA probes for detection of Shigella spp. and enteroinvasive Escherichia coli. J. Clin. Microbiol. 26:
261266.
Venkatesan, M.M., Buysse, J.M., and Kopecko, D.J. 1989. Use of Shigella flexneri ipaC and ipaH gene sequences for
the general identification of Shigella spp. and enteroinvasive Escherichia coli. J. Clin. Microbiol. 27: 26872691.

Wachsmuth, I.K., Evins, G.M., Fields, P.I., Olsvick, Ø., Popovic, T., Bopp, C.A., Wells, J.G., Carrillo, C., and Blake,
P.A. 1993. The molecular epidemiology of cholera in Latin America. J. Infect. Dis. 167: 621626.
Wachsmuth, K., Olsvik, Ø., Evins, G.M., and Popovic, T. 1994. Molecular epidemiology of cholera. In: Vibrio
cholerae and Cholera: Molecular to Global Perspectives, I.K. Wachsmuth, P.A. Blake, and Ø. Olsvik (Ed.), p. 357370.
ASM Press, Washington, D.C.

Wadström, T. and Ljungh, Ä. 1991. Aeromonas and Pleisiomonas as food- and waterborne pathogens. Int. J. Food
Microbiol. 12: 303312.
Ward, R.L., Bernstein, D.J., Young, E.C., Sherwood, J.R., Knowlton, D.R., and Schiff, G.M., 1986. Human rotavirus
studies in volunteers: determination of infectious dose and serological response to infection. J. Infect. Dis. 154:
871880.

Weidmann, M., Bruce, J.L., Knorr, R., Bodis, M., Cole, E.M., McDowell, C.I., McDonough, P.L., and Batt, C.A. 1996.
Ribotype diversity of Listeria monocytogenes strains associated with outbreaks of Listeriosis in ruminants. J. Clin.
Microbiol. 34: 10861090.

Wetmur, J.G. and Davidson, N. 1968. Kinetics of renaturation of DNA. J. Mol. Biol. 31: 349370.

Whittam, T.S. 1995. Insertion sequence fingerprinting: a new subtyping system for Escherichia coli O157:H7 strains.
J. Food Prot. 57(Suppl): 64.

Wilde, J., Eiden, J., and Yoken, R. 1990. Removal of inhibitory substances from human fecal specimens for detection
of group A rotaviruses by reverse transcriptase and polymerase chain reaction. J. Clin. Microbiol. 28: 13001307.
Willshaw, G.A., Smith, H.R., Roberts, T., Thirlwell, J., Cheasty, T., and Rowe,

 Page 180

B. 1993. Examination of raw beef products for the presence of Vero cytotoxin producing Escherichia coli,
particularly those of serogroup O157. J. Appl. Bacteriol. 75: 420426.

Willshaw, G.A., Scotland, S.M., Smith, H.R., Cheasty, T., Thomas, A., and Rowe, B. 1994. Hybridization of strains of
Escherichia coli O157 with probes derived from the eaeA gene of enteropathogenic E. coli and the eaeA homolog from
a Vero cytotoxin-producing strain of E. coli O157. J. Clin. Microbiol. 32: 897902.

Wilson, S.G., Chan, S., Deroo, M., Vera-Garcia, M., Johnson, A., Lane, D., and Halbert, D.N. 1990. Development of a
colorimetric, second generation nucleic acid hybridization method for detection of Salmonella in foods and a
comparison with conventional culture procedure. J. Food Sci. 55: 13941398.
Wilson, J.B., McEwen, S.A., Clarcke, R.C., Leslie, K.E., Wilson, R.A., Waltner-Toews, D., and Gyles, C.L. 1992.
Distribution and characteristics of verocytotoxigenic Escherichia coli isolated from Ontario dairy cattle. Epidemiol.
Infect. 108: 423439.

Wilson, I.G., Gilmour, A., Cooper, J.E., Bjourson, A.J., and Harvey, J. 1994. A non-isotopic DNA hybridization assay
for the identification of Staphylococcus aureus isolated from foods. Int. J. Food Microbiol. 22: 4354.

Wood, P.K., Morris, J.G., Jr., Small, P.L.C., Sethabutr, O., Toledo, M.R.F., Trabulsi, L., and Kaper, J.B. 1986.
Comparison of DNA probes and the Sereny test for identification of invasive Shigella and Escherichia coli strains. J.
Clin. Microbiol. 24: 498500.

Wren, B.W. and Tabaqchali, S. 1990. Detection of pathogenic Yersinia enterocolitica by the polymerase chain reaction.
Lancet 336: 693.

Wright, A.C., Morris, J.G., Jr., Maneval, D.R., Jr., Richardson, K, and Kaper, J.B. 1985. Cloning of the cytotoxin-
hemolysin gene of Vibrio vulnificus. Infect. Immun. 50: 922927.

Wright, A.C., Guo, Y., Johnson, J.A., Nataro, J.P., and Morris, J.G., Jr. 1992. Development and testing of a
nonradioactive DNA oligonucleotide probe that is specific for Vibrio cholerae. J. Clin. Microbiol. 30: 23022306.

Wright, A.C., Miceli, G.A., Landry, W.L., Christy, J.B., Watkins, W.D., and Morris, J.G., Jr. 1993. Rapid identification
of Vibrio vulnificus on nonselective media with an alkaline phosphatase-labeled oligonucletotide probe. Appl. Environ.
Microbiol. 59: 541546.

Yamakawa, K., Oyamada, H., and Nakagomi, O. 1989. Identification of rotaviruses by dot-blot hybridization using an
alkaline phosphatase-conjugated synthetic oligonucleotide probe. Mol. Cell. Probes 3: 397401.

Yamamoto, K., Wright, A.C., Kaper, J.B., and Morris, J.G., Jr. 1990. The

 Page 181

cytolysin gene of Vibrio vulnificus: Sequence and relationship to the Vibrio cholerae El Tor hemolysin gene. Infect.
Immun. 58: 27062709.

Yamamoto, K., Honda, T., Miwatani, T., Tamatsukuri, S., and Shibata, S. 1992. Enzyme-labeled oligonucleotide probes
for detection of the genes for thermostable direct hemolysin (TDR) and TDH-related hemolysin (TRH) of Vibrio
parahaemolyticus. Can. J. Microbiol. 38: 410415.

Yoh, M., Miyagi, K., Matsumoto, Y., Hayashi, K., Takarada, Y., Yamamoto, K., and Hando, T. 1993. Development of
an enzyme-labeled oligonucleotide probe for the cholera toxin gene. J. Clin. Microbiol. 31: 13121314.
Yu, J. and Kaper, J.B. 1992. Cloning and characterization of the eae gene of enterohaemorrhagic Escherichia coli
O157:H7. Mol. Microbiol. 6: 411417.

Yu, H., Bruno, J.G., Cheng, T., Calomiris, J.J., Goode, M.T., and Gatto-Menking, D.L. 1995. A comparative study of
PCR product detection and quantitation by electrochemiluminescence and fluorescence. J. Bioluminesc.
Chemiluminesc. 10: 239245.

Zhou, Y.-J., Estes, M.K., Jiang, X., and Metcalf, T.G. 1991. Concentration and detection of hepatitis A virus and
rotavirus from shellfish by hybridization tests. Appl. Environ. Microbiol. 57: 29632968.

Zink, D.L., Feely, J.C., Wells, J.G., and Vanderzant, C. 1980. Plasmid-mediated tissue invasiveness in Yersinia
enterocolitica. Nature (London) 283: 224226.

 Page 183

9
PCR and Nucleic Acid Amplification Methods
Michael G. Johnson and Debra K. Winters
University of Arkansas
Fayetteville, Arkansas
Rong-Fu Wang
National Center for Toxicological Research
Food and Drug Administration
Jefferson, Arkansas

Introduction
Classic references on the history of the development of the polymerase chain reaction (PCR), for which K. B. Mullis
won the Nobel Prize, and on the enzymology of PCR are summarized by Atlas and Bej (1994). PCR may be
summarized as follows (Atlas and Bej, 1994):
The polymerase chain reaction (PCR) is an in vitro method for replication of a defined target DNA sequence so that its amount is
increased exponentially. Whereas previously only minute amounts of a specific gene could be obtained from a cell, now even a single
gene copy can be amplified to a million copies within a few hours by PCR.

PCR consists of repetitive cycles of DNA denaturation through melting at elevated temperature to convert double

 Page 184

stranded DNA to single stranded DNA, annealing of oligonucleotide primers to the target DNA, and extension of the DNA by nucleotide
addition from the primers by the action of DNA polymerase. . . . The target region for amplification is defined by unique oligonucleotide
primers that flank a DNA segment. The oligonucleotide primers are designed to hybridize to regions of DNA that flank a desired target
gene sequence, annealing to a complementary strand of the target sequence. The primers are then extended across the target sequence by
using a heat stable DNA polymerase (frequently Taq DNA polymerase, a thermostable and thermoactive enzyme from Thermus
aquaticus) in the presence of free deoxynucleoside triphosphates (dNTPs), resulting in a double replication of the starting target material.
By repeating the three stage process many times, a nearly exponential increase in the amount of target DNA results. . . . The product of
each PCR cycle is complementary to and capable of binding the primers, and so the amount of DNA is potentially doubled in each
successive cycle.

We will summarize selected papers about the use of PCR in the detection of foodborne organisms. Brief summaries
will be provided of selected research papers published from 1993 through MayJune 1996 about (1) major recent
reviews on PCR techniques and primers, (2) PCR for anaerobic bacteria, (3) PCR and RAPD assays for pathogenic
bacteria and viruses as natural contaminants in foods and water, (4) PCR for lactics and spoilage bacteria, (5) PCR for
yeasts and molds associated with foods, (6) effects of cellular stresses on PCR results, (7) practical problems and
solutions in PCR assays of foods, (8) multiple organism PCR assay strategies, (9) enhanced detection of PCR products,
(10) new developments in PCR, and (11) suggested future research.

Major Recent Reviews On PCR Techniques and Primers

Methods for setting up and optimizing the concentrations of Taq DNA polymerase enzyme, nucleotides, buffer, Ca and
Mg ions, and other factors are summarized in several good manuals (Atlas and Bej, 1994; Dieffenbach and Dveksler,
1995; Innis et al., 1990; 1995). Some theoretical papers on the sequences in 16S rRNA that are unique for each target
organism are discussed by Jensen et al. (1993), Gendel (1996), and for Campylobacter specifically by Van Camp et al.
(1993). A nice review of the various PCR assay formats used in food analyses was published by Wolcott (1991). The
uses of rRNA sequences to distinguish and separate strains within a species when

 Page 185

cleaved with EcoRI endonuclease were reported for Listeria monocytogenes by Bruce et al. (1995, 1996) and Hubner
et al. (1995).

Extensive reviews of PCR assays for foodborne pathogens were recently published by Hill (1996) and Olsen et al.
(1995). Primer sequences listed as specific for pathogens in PCR assays by Hill (1996) include Aeromonas hydrophila;
Brucella spp.; Campylobacter (C. jejuni and C. coli); Clostridium spp. (C. botulinum neurotoxins A and B); coliforms;
Escherichia coli and its toxins; Shigella spp. pathogenic factors; Helicobacter pylori; Listeria spp. (L. innocua, L.
monocytogenes cells and pathogenic factors); Salmonella spp.; Staphylococcus aureus toxic factors; Vibrio spp. (V.
cholerae, V. parahaemolyticus, V. vulnificus); Yersinia spp. (Y. enterocolitica and Y. pseudotuberculosis); viruses
(hepatitis A, enteroviruses, Norwalk virus, rotavirus); and parasites (Giardia spp., G. duodenalis). Olsen et al. (1995)
provides primer sets for additional organisms or additional primers for some of the same organisms listed by Hill as
follows: E. coli pathogenic types (enteropathogenic, enterotoxigenic, enterohemorrhagic/verocytotoxigenic and
enteroinvasive); Shigella flexneri and Shigella dysenteriae; other noncholerae Vibrio; other Campylobacter spp. (C.
hyointestinalis, C. lari, and C. upsaliensis); L. monocytogenes, additional pathogenic factors and cell surface proteins;
S. aureus, additional toxic or pathogenic factors; C. botulinum neurotoxins A, B, C, D, E, F, and G, and Clostridium
perfringens toxins and pathogenic factors. Gendel (1996) provides a 16S rRNA data base set of unique nucleic acid
base sequences, ''signature sequences," for possible use as PCR primers (not tested) for these additional organisms:
Bifidobacterium spp.; Brochothrix spp.; Carnobacterium spp.; other Clostridium spp. (C. bifermentans, C. difficile, and
C. sordelle); Lactobacillus spp., Leuconostoc spp.; Listeria ivanovii; and Vibrio anguillarum.

PCR for Anaerobic Bacteria

Very recently the major anaerobic bacteria in the human gut (feces) were assayed by a quantitative dilution PCR
method by Wang et al. (1996), who reported specific primers and positive PCR results for Bifidobacterium (B.
adolescentis and B. longum); Eubacterium (E. biforme and E. limosum); Fusobacterium prausnitzii; Lactobacillus
acidophilus; Bacteroides spp. (B. thetaiotaomicron, B. distasonis, and B. vulgatus), and Clostridium clostridiiforme.

PCR and RAPD for Pathogenic Bacteria and Viruses as Natural Contaminants in Foods and Water

Recent selected examples of PCR tests for pathogens as natural contaminantsrather than as added cellsin foods include
L. monocytogenes in

 Page 186

jelled pork tongue product of France (Jacquet et al., 1995), in silage (Wiedmann et al., 1996b), in raw milk (Harvey and
Gilmour, 1994), in shrimp-processing plants (Destro et al., 1996), in smoked salmon (Simon et al., 1996), in poultry
processing (Lawrence and Gilmour, 1994, 1995), and in food outlets (Niederhauser et al., 1994a); C. jejuni in milk
(Wegmuller et al., 1993), and in chicken washes (Winters and Slavik, 1995); C. botulinum neurotoxins in foods (Szabo
et al., 1993; Fach et al., 1995); B. cereus in milk (Schraft et al., 1995, 1996); Brucella spp. in milk (Rijpens et al.,
1996); E. coli O157:H7 in raw ground beef (Weagant et al., 1995) and enterohemorrhagic E. coli gene in raw ground
beef (Witham et al., 1996); Salmonella spp. in oysters (Bej et al., 1994) and on chicken skin (Mahon et al., 1994);
Shigella flexneri survival on prepared raw vegetables (Rafil et al., 1995); S. aureus genes for enterotoxins directly in
foods (Tsen et al., 1994, 1995); Vibrio cholerae 01, 0139, and non-01 strains in water (Rivera et al., 1995); V.
parahaemolyticus in shellfish (Lee et al., 1995); V. vulnificus in fish and oysters (Arias et al., 1995; Coleman et al.,
1996); hepatitis A in hardshell clams and shellfish (Goswani et al., 1993; Guyader et al., 1994); rotaviruses in shellfish
and swine/cattle (Guyader et al., 1994; Gouvea et al., 1994); enteroviruses (Atmar et al., 1993, 1995; Deng et al., 1994;
Gilgen et al., 1995; Jaykus et al., 1995); and a very recent collaborative study of PCR detection of Norwalk virus
(Atmar et al., 1996). A method to tell whether such viruses are still viable and infectious was addressed recently in a
coupled cell culturePCR assay (Reynolds et al., 1996).

Recent PCR assays for important parasites in water or feces, such as Cryptosporidium, the organism involved in
waterborne outbreaks in Milwaukee, were published by Johnson et al. (1995) and Leng et al. (1996) and on Leishmania
by Pogue et al. (1995).
Recent RAPD assay reports for distinguishing pathogens in model systems or tracing them in foods and food
processing plants include separation of L. innocua, a frequent co-contaminant, from L. monocytogenes (Czajka et al.,
1993); L. monocytogenes in shrimp, cheese, and poultry-processing plants or general samples (Farber and Addison,
1994; Destro et al., 1996; Wagner et al., 1996; Lawrence et al., 1993; Lawrence and Gilmour, 1995; Kerr et al., 1995);
combinations of L. monocytogenes, Y. enterocolitica, and verocytotoxic E. coli and S. enteritidis (Niederhauser et al.,
1994a); and S. aureus (Matthews and Oliver, 1994).
Other recent primer sets and PCR assays reported specific for pathogenic bacteria or viruses in foods or water
published from 1993 to 1996 include tests for: Arcobacter species (Harmon and Wesley, 1995); Pseudomonas
aeruginosa (Khan and Cerniglia, 1994); Bacillus cereus-specific lecithinase and emetic toxin assays (Schraft et al.,
1996, Schraft and Griffiths, 1995); mycoplasma in tissue culture (Dussurget and Rullard-Dussoix, 1994); L. ivanovii,
(Wang et al., 1993).

 Page 187

PCR for Lactic Acid and Spoilage Bacteria

Lactic acid bacteria and their metabolic products contribute not only flavor compounds but also bacteriocins, which can
help control growth of pathogens in foods. Some recent PCR assays for bacteriocin-producing potential of Lactococcus
spp. (Cocconelli et al., 1995; Klijn et al., 1995a) or of Lactobacillus spp. in meats (Rodriguez et al., 1995) as well as
for Lactobacillus spp. tolerant to 15% alcohol in sake fermentation (Nakagawa et al., 1994) have recently been
reported. To help judge the balance of fermentation cultures and their relative impacts on fermented dairy product
finished quality, a PCR assay was recently developed to differentially amplify the DNA from several species of
Lactobacillus: L. acidophilus, L. casei, L. delbrueckii, and L. helveticus. Further, a novel PCR assay strategy was
recently used to show the absence of enterotoxin production potential in two specially selected strains of Citrobacter
freundii and Klebsiella pneumoniae used to produce vitamin B12 in a tempeh fermentation (Keuth and Bisping, 1994).
PCR assays specific for genes coding for some enzymes important in the food industry are starting to be developed.
One example is the PCR assay specific for the genes encoding the histamine-producing histidine decarboxylase
enzyme found in some lactic acid bacteria (Le-Jeune et al., 1995). Some humans are sensitive to histamine, which can
cause headaches in humans when consumed in fermented foods. Another example of an enzyme-specific PCR-based
assay is the one reported for the aminopeptidase associated with L. monocytogenes (Winters et al., 1996a). The
problem of both viable and nonviable cells of a pathogen giving a positive PCR test has been addressed by a novel
strategy using the luciferase reporter bacteriophage A511::luxAB, which allowed differentiation of viable and
nonviable cells of Listeria (Loessner et al., 1996).

Clostridium tyrobutyricum cells and spores in raw milk and cheese have long been thought to be the main cause of
extensive cheese losses due to late gas production, called blown curd. PCR assays specific for cells and spores of this
organism recently published prove this hypothesis (Herman et al., 1995: Klijn et al., 1995b).

PCR for Yeasts and Molds Associated with Foods

The usefulness of PCR for detection of molds and yeasts involved in food spoilage include assays for filamentous
ascomycetes and for deuteromycetes (Glass and Donaldson, 1995); Neosartorya fischeri, a spoilage agent of heated
fruit products (Girardin et al., 1995); Aspergillus sojae, a koji mold (Yuan et al., 1995); yeasts and Saccharomyces
cerevisiae (Baleiras-Couto et al., 1995,

 Page 188

1996); Dekkera-Brettanomyces (Ibeas et al., 1996); Zygosaccharomyces bailii, a spoilage agent of salad dressings and
mayonnaise (Stubbs et al., 1994).

Effects of Cellular Stresses on PCR Results

The effects of cellular stress with physical and chemical treatments prior to PCR analysis have been reported for only a
few model systems and a few food systems. E. coli O157:H7 cells cold-stressed in raw ground beef samples
incriminated in human foodborne illnesses were detected in 15 of 17 samples if enrichment and antigen capture with
latex beads was used prior to PCR assay, whereas only 6 of 17 of the same samples were positive for this organism if
conventional FDA methods were used (Weagant et al., 1995). Cells of Shigella dysenteriae (Islam et al., 1993), L.
monocytogenes and enterotoxigenic E. coli strains (Masters et al., 1994), and Salmonella (Josephson et al., 1993) all
gave positive PCR test results even after extensive stresses. In fact, Masters et al. (1994) showed positive PCR results
for E. coli and L. monocytogenes cultures even after autoclaving!

Practical Problems and Solutions in PCR Assays of Foods

Recent, shorter thought-provoking reviews covering the practical problems and solutions for general uses of PCR and
uses of PCR in food assays include those of Niederhauser et al. (1994b), Allmann et al. (1995), and Candrian (1995).
Niederhauser et al. (1994b), citing disturbing results for lack of repeatability of PCR assays for multiple site tests in
three large studies, listed the following problems:

Study 1the false-positive and false-negative rates were 9.3 and 7.4%, respectively.

Study 2the false-negative rates were as high as 1098%.

Study 3some laboratories failed to get PCR-positive results when provided with 10,000 copies of the target DNA from
the target organism.

Most false-positive PCR results were attributable to carryover of amplicons or partially degraded amplicon fragments
from previous experiments or cross-contamination between samples of amplifiable DNA (e.g., from a Listeria
monocytogenes-positive sample to a L. monocytogenes-negative sample). The chief reasons given for false-negative or
reduced-sensitivity PCR results were:

1. Primers lacked sufficient homology to target DNA/ RNA.

 Page 189

2. Inhibitors in foods blocked nucleic acid annealing.

3. There was insufficient PCR amplification due to a shortage of Taq DNA polymerase, nucleotides, Mg ions, or the
wrong buffer, or temperature cycle program was used.

4. Carryover contamination of DNA from primer artifacts or partially degraded PCR amplicons interfered with results.

Niederhauser et al. (1994b) summarized three helpful steps to guard against the generation of false-positive PCR
results. Step one is to use separate rooms and micropipetting devices for preparation of cell lysates and loading the
PCR assay microcentrifuge tubes apart from the PCR thermocycler run per se. Step two involves the use of a pre-PCR
enzyme treatment with uracil DNA glycosylase to degrade stray amplicon fragments from previous PCR runs. A third
step is to use a post-PCR UV treatment to cross-link any stray airborne DNA or RNA so that such molecules cannot
return to the single-stranded state and serve as templates. Other chemical strategies, including use of 8-methoxy
psoralen (MOPS), hydroxylamine, or a commercial product composed of oligomers with 5' fluorescent labels, were not
as effective in blocking generation of false-positive or negative PCR results.
Two major problems mentioned by Allmann et al. (1995) and Candrian (1995) are (1) the failure to achieve positive
PCR results for low numbers of added pathogens in the presence of food materials in direct assays or in enrichment
broth media (presumably, some chemicals in the foods interfere with the ability of the primers to anneal to the target
DNA or with the activity of Taq DNA polymerase in the assay and (2) the relative lack of validation studies, showing
that the proposed PCR assays work properly to detect a pathogen as a natural contaminant in a food by directly
comparing results for PCR and conventional culture tests for the same samples.
Strategies to Remove Food Chemicals Interfering with PCR Assays

Many interesting strategies to overcome food interference problems have been reported, including washing the target
cells by centrifugation prior to PCR assay [poultry products (Wang et al., 1992; Winters et al., 1995, 1996b) and milk
(Cooray et al., 1994; Jinneman et al., 1995)]; the uses of antibody, latex bead antibody, or magnetic immunobead
antibody capture of target cells, L. monocytogenes, Salmonella spp., E. coli O157:H7, Y. enterocolitica, or hepatitis A
virus, from foods or food enrichments prior to PCR (Deng et al., 1994; Fluitt et al., 1993a, b; Jinneman et al., 1995;
Weagant et al., 1995; Kapperud et al., 1993; Atmar et al., 1995); and antibody to the RNA-DNA

 Page 190

complex formed from PCR assay for L. monocytogenes (Blais and Phillippe, 1993; Blais, 1994); filter membranes for
viruses and Campylobacter spp. (Gilgen et al., 1995; Kirk and Rowe, 1994; Jothikumar et al., 1993; Oyofo and Rollins,
1995); bioaerosol detection of pathogens (Alvarez et al., 1994, 1995); direct assay of surface swabs from metal
surfaces (Wiedmann et al., 1995).

Successful PCR assays generally require release of the target DNA or RNA from the cell by enzyme and physical lysis
treatments. Some examples include boiling cells of L. monocytogenes in 1% Triton X-100 (Wang et al., 1992);
sonication and freeze thaw treatments for C. jejuni (Kirk and Rowe 1994); treating mixtures of L. monocytogenes, Y.
enterocolitica, and A. viridans with proteinase K and isopropanol or toluene plus mutanolysin and guanidine
thiocyanate (Niederhauser et al., 1994a); treating mixtures of L. monocytogenes, E. coli, and C. jejuni with TRIS,
EDTA, and 0.5% SDS and pronase followed by treatment with lysozyme and proteinase (Allmann et al., 1995).

Interference in the detection of L. monocytogenes in PCR assays by soft cheese samples (Wang et al., 1992) was
reportedly overcome by using a PEG extract (Lantz et al., 1994) or an alcohol precipitation followed by addition of NaI
(Makino et al., 1995); interference by raw ground beef in the detection of C. perfringens was overcome by use of a
commercial DNA extraction kit (Baez aridjuneja, 1995). Inclusion of bovine serum albumin or T4 gene 32 protein was
very recently reported to relieve PCR inhibition (Dreader, 1996).

Since enteroviruses, hepatitis A, and Norwalk viruses have their genes encoded by RNA, a reverse transcriptase PCR
(RT-PCR) must be used for these target microbes. There are apparently several chemicals in shellfish that interfere with
the RT-PCR assay. This can be overcome by the use of several nucleic acid extraction chemicals following virus
particle lysis, including chloroform-butanol, polyethylene glycol, 1,1-trichloro-2,2-trifluorethane, guanidine
isothiocyanate phosphate buffer, cetyl trimethylammonium bromide, and glass powder (Atmar et al., 1993, 1995, 1996;
Jaykus et al., 1995; Lees et al., 1994).
As PCR technology developed, researchers soon discovered that this assay is prone to several problems that lead to
false-positive or false-negative results and noncomparability of PCR results for the same pathogens and assay
parameters for different laboratories using different thermocyclers. Part of the solution to false-negative results has
been addressed above in the need to remove interfering stray amplicons and food chemicals before running the PCR
assay. Three additional recent strategies to guard against false-negative results in tests for pathogens in foods are the
inclusion of (1) an internal probe with fluorescent labels (Miller et al., 1996), (2) primers

 Page 191

for an alternate nonpathogenic gene marker, such as malB of E. coli, or (3) of a heterologous marker gene such as using
an E. coli hemolytic marker in PCR tests where the intended target is the aerolysin gene of A. hydrophilia (Allmann et
al., 1995; Baloda et al., 1995). These techniques, with internal positive controls, help prove during the PCR assay that
the Taq DNA polymerase is active and that the other assay conditions such as concentrations of nucleotides, Mg ion,
and buffer were appropriate.

Multiple Organism PCR Assay Strategies

Allmann et al. (1995) developed a common lysate procedure for use in PCR tests for the presence of natural
contamination in raw milk samples and cheese samples made with such milks (Table 9.1) and then customized the PCR
tests for detection of several different pathogens in lysis samples from these milks or cheeses (Table 9.2). In this lysis
procedure, note that more digestion buffer is used for the cheese than the milk samples and that three enzymespronase,
lysozyme, and proteinase Kare used. Table 9.2 summarizes the variations in Mg ion concentrations and annealing
temperatures used for PCR detection of cells of normal E. coli as well as of pathogenic E. coli, C. jejuni, and C. coli
and L. monocytogenes. Of interest is that when the mal B gene target was used, 46% of the raw milk samples were
found to be positive, whereas the conventional culture method only yielded 23% positive results from the same
samples. This outcome suggested that the other microflora. present suppressed the growth of the wild-type E. coli
strains.
TABLE 9.1 Common Lysate Procedure for PCR Tests of Raw Milk or Cheese for Presence of L. monocytogenes, E. coli, and C.
jejuni
1. 40 mL of milk + 5 mL of digestion buffer (or 2 g of cheese = 45 mL of 0.1 × digestion buffer)
Digestion buffer contains: 100 mM each in Tris & EDTA, 0.5% SDS
2. Pronase, 40°C, 3 hr
3. Centrifuge: 2300 × g, 15 min, 4°C
4. Discard fat and water phase, wash pellet 3× in TE buffer
5. Final wash in PCR buffer with 1.5 mM Mg+2
6. Resuspend pellet with lysozyme in PCR buffer, 15 min
7. Proteinase K, 200 mg/mL: 60°C for 60 min; 95°C for 15 min
8. Centrifuge: 13,000 × g, 5 min, 4°C
9. Analyze 25-mL aliquots, in PCR for specific pathogen
Source: Adapted from Allmann et al., 1995.

 Page 192

TABLE 9.2 PCR Results for Pathogens Naturally Contaminating Milk

Mg+2 Anneal PCR pos. Culture
Organism Target gene (mM) (°C) (%) pos.
E. coli malB 3.0 65 41/90 21/90
(46) (23)
E. coli elt/LTI 1.5 55 11/41 nd
(27)
E. coli est/STI 4.0 50 11/41 nd
(27)
L. monocytogenes hly 2.0 55 1/90a nd
C. jejuni flaA/ 4.0 53 6/90a 0/54b
C. coli flaB (7) (0)
aLower limit of detection by PCR was 50500 CFU/mL.
bOnly raw milk samples were used in C. jejuni assays.
Source: Adapted from Allmann et al., 1995.

Enhanced Detection of PCR Products

Once the PCR assay has been run, one still must detect the amplicons. The conventional method involves separating
the DNA products in an agarose gel by electrophoresis and staining with ethidium bromide (Atlas and Bej, 1994).
Denaturing gel electrophoresis (Muyzer et al., 1993) or use of polyvinylpyrrolidone in electrophoresis gels (Young et
al., 1993) was reported to increase the sensitivity, of detection of amplicons. Recently the use of antibodies directed to
the RNA-DNA complex for L. monocytogenes has been reported by Blais (1994). Another recent development being
exploited commercially by Perkin Elmer in their PakMan® assay is the use of techniques to generate amplicons labeled
with fluorescent probes during amplification (Bassler et al., 1995), which can then be read by fiberoptic sensors or
biosensors (Strachan and Gray, 1995) in a microtiter plate format (Cano et al., 1994). These developments will permit
the automation of PCR assays, an important criterion for rapid methods development set forth by Fung (1995) in his
succinct review.

New Developments

The commercial prepreparation of all reagents needed to lyse cells and prepare templates for amplification in a
prepackaged microcentrifuge tube format is being introduced as the BAXTM system by Qualicon, Inc., a division

 Page 193

of DuPont. This format promises to circumvent some of the labor and manipulation problems that have limited PCR
assays to highly skilled molecular biologists who understand that there are sufficient DNase and RNase enzymes at
one's fingertips to invalidate a PCR assay. Another promising development is the automated ribotyper being developed
by Qualicon, which involves the use of pure cultures, RNA extraction, EcoRI digestion, separation and probing with
known E. coli ribosomal RNA probes, electrophoretic separation of bands, and photographic recording and computer
image analysis of the bands to distinguish species within a genus and to separate strains of a species from one another.
The power of this automated assay technique was recently shown by Wiedmann et al. (1996a, b), who were able to
show allelic diversity of virulence genes in strains of L. monocytogenes from human foodborne illnesses and
similarities and differences in the ribotype patterns for pairs of strains of L. monocytogenes isolated from incriminated
silage and cows eating that silage and showing listeriosis symptoms. A similar though nonautomated ribotyping pattern
method was used recently by Jacquet et al. (1995) to trace the strains of L. monocytogenes that caused human listeriosis
cases in France in 1992 from the consumption of jelled pork tongue product. Another recent example is the use of PCR
fingerprinting, using PCR followed by restriction enzyme analysis, to separate two serovar 4b strains of L.
monocytogenes (Ericsson et al., 1995).

PCR is also very useful for sequencing and cloning DNA. PCR can be used to amplify genes or other DNA fragments.
The PCR amplified gene can be directly cloned by the TA Cloning Kit (Invitrogen, San Diego, CA). If digestion sites
for a particular restriction enzyme are pre-placed at the 5'-ends of the PCR primers, the PCR amplified DNA can be
digested with that restriction enzyme and then cloned into any vector having sites which can be digested with the same
restriction enzyme. The PCR products obtained can be directly sequenced using a PCR-based Cycle Sequencing Kit
(Epicentre Technologies, Madison, WI).

Another exciting possibility is the use of PCR cloning strategies to construct single-chain antibody arms from murine
hybridoma cells to give antibodies with increased specificity for Bacillus cells and spores (Koo et al., 1996). The irony
here is that antibodies will likely long remain a valuable helpmate to PCRand vice versain evolving strategies for
decreasing detection times and increasing sensitivities in assays for pathogens in foods.

Suggested Future Research

To prove in a court of law that a given food or microbe was involved in a foodborne illness outbreak, the current FDA
or USDA-FSIS methods and

 Page 194

procedures require that their researchers be able to provide isolated, pure cultures of the suspected microbe from the
incriminated food. PCR assay techniques classically have been used on mixed cultures where the target cell is
separated out by some extraction technique that destroys cell viability or is increased by preenrichment and selective
enrichment media that result in mixed cultures. The ribotyping mentioned above meets the purity requirement but
involves the use of expensive equipment (the commercial version offered by Qualicon is quoted at $175,000) usually
only within the reach of major hospitals and major highly centralized food analysis laboratories of major companies.

A simple solution may be to make replica membrane overlays of pathogens on the appropriate selective agar media
(Bhunia et al., 1992), remove the membraneyielding the isolated colonies for follow-up pure culture biochemical
characterization and storageand then do a solid-phase PCR assay of the isolated colonies on the membrane by excising
them into microcentrifuge tubes or microtiter plate wells.

A multitude of successful laboratory-based PCR assays have been catalogued in the recent reviews of Hill (1996) and
Olsen et al. (1995). It is necessary to develop more user-friendly PCR assays to take to the field as we attempt to audit
the preharvest side of the food-production chain, a need strongly suggested by the recent report, "Tracking Foodborne
Pathogens from Farm to Table" (Roberts et al., 1995).

The power of computer data bases for analyzing DNA sequences will be reviewed by Dr. Gendel in Chapter 11 of this
volume. Coupling these data bases with computer-modeling programs (Brunk et al., 1996) and the development of
procedures to overcome biases in how templates anneal during amplification (Suzuke and Giovannoni, 1996) should
make it possible to develop PCR assays with increased specificities and sensitivities.

The practical solutions to PCR problems gleaned from the last 2 years of selected research literature references, as cited
hereincoupled with emerging new molecular genetic insightsstrongly suggest that the future for continued use and
refinement of PCR strategies in food-safety assurance programs in industry, academic, consulting, and federal
regulatory laboratories is bright indeed.

Acknowledgments

The authors gratefully acknowledge partial support for the preparation of this summary by grant funds from the USDA
Food Safety Consortium. The use of prepublication materials provided by MaryAnne Drake, Karen M. Harmon, Walter
E. Hill, Kai Koo, Heidi Schraft, and Martin Wiedmann is

 Page 195

gratefully appreciated. The authors thank James Goff for his expert technical assistance in the preparation of the
manuscript.

References
Allman, M., Hofelein, C., Koppel, E., Luthy, J., Meyer, R., Niederhauser, C., Wegmuler, B., and Chandrian, U. 1995.
Polymerase chain reaction (PCR) for detection of pathogenic microorganisms in bacteriological monitoring of dairy
products. Res. Microbiol. 146: 8597.

Alvarez, A.J., Buttner, M.P., Toranzos, G.A., Dvorsky, E.A., Toto, A., Heikes, T.B., Mertikas-Pifer, L.E., and
Stetzenbach, L.D. 1994. Use of solid phase PCR for enhanced detection of airborne microorganisms. Appl. Environ.
Microbiol. 60: 374376.

Alvarez, A.J., Buttner, M.P., and Stetzenbach, L.D. 1995. PCR for bioaerosol monitoring: Sensitivity and
environmental interference. Appl. Environ. Microbiol. 61: 36393644.

Arias, C.R., Garay, E., and Aznar, R. 1995. Nested PCR method for rapid and sensitive detection of Vibrio vulnificus in
fish, sediments and water. Appl. Environ. Microbiol. 61: 34763478.
Atlas, R.M. and Bej, A.K. 1994. Polymerase chain reaction. In: Methods for General and Molecular Bacteriology, P.
Gerhardt, R.G.E. Murray, W.A. Wood, and N.R. Krieg (Ed.), p. 418435. American Society for Microbiology,
Washington, D.C.

Atmar, R.L., Metcalf, T.G., Neill, F.H., and Estes, M.K. 1993. Detection of enteric viruses in oysters by using the
polymerase chain reaction. Appl. Environ. Microbiol. 59: 631635.
Atmar, R.L., Neill, F.H., Romalde, J.L., Le Guyader, R., Woodley, C.M., Metcalf, T.G., and Estes, M.K. 1995.
Detection of Norwalk virus and hepatitis A virus in shellfish tissues with the PCR. Appl. Environ. Microbiol. 61:
30143018.

Atmar, R.L., Neill, F.H., Woodley, C.M., Manger, R., Rout, G.S., Burkhardt, W., Leja, L., McGovern, E.R., Le
Guyader, F., Metcalf, R.G., and Estes, M.K. 1996. Collaborative evaluation of a method for the detection of Norwalk
virus in shellfish tissues by PCR. Appl. Environ. Microbiol. 62: 254258.

Baez, L.A. and Juneja, V.K. 1995. Detection of enterotoxigenic Clostridium perfringens in raw beef by PCR. J. Food
Prot. 58: 154159.

Baleiras-Couto, M.M., Vogels, J.R.W.E., Hofstra, H., Huis in't Veld, J.H.J., and van der Vossen, J.M.B.M. 1995.
Random amplified polymorphic DNA and restriction enzyme analysis of PCR amplified rDNA in taxon-

 Page 196

omy: Two identification techniques for food-borne yeasts. J. Appl. Bacteriol. 79: 525535.

Baleiras-Couto, M.M., Eijsma, B., Hofstra, H., Huis in't Veld, J.H.J., and van der Vossen, J.M.B.M. 1996. Evaluation
of molecular typing techniques to assign genetic diversity among Saccharomyces cerevisiae strains. Appl. Environ.
Microbiol. 62: 4146.

Baloda, S.B., Krovacek, K., Eriksson, L., Linne, T., and Mansson, I. 1995. Detection of aerolysin gene in Aeromonas
strains isolated from water fish and food by PCR. Comp. Immunol. Microbiol. Infect. Dis. 18: 1726 (CAB abstracts).

Bassler, H.A., Flood, S.J.A., Livak, K.J., Marmaro, J., Knorr, R., and Batt, C.A. 1995. Use of a fluorogenic probe in a
PCR-based assay for the detection of Listeria monocytogenes. Appl. Environ. Microbiol. 61: 37243728.

Bej, A.K., Mahbubani, M.H., Boyce, M.J., and Atlas, R.M. 1994. Detection of Salmonella spp. in oysters by PCR.
Appl. Environ. Microbiol. 60: 368373.

Bhunia, A.K., Ball, P.H., and Johnson, M.G. 1992. A 2024 h microcolony-immunoblot technique to detect and
enumerate Listeria monocytogenes inoculated into foods. J. Rapid Methods Autom. Microbiol. 1: 6782.

Blais, B.W. 1994. Transcriptional enhancement of the Listeria monocytogenes PCR and simple immunoenzymatic
assay of the product using anti-RNA:DNA antibodies. Appl. Environ. Microbiol. 60: 348352.
Blais, B.W. and Phillippe, L.M. 1993. A simple RNA probe system for analysis of Listeria monocytogenes polymerase
chain reaction products. Appl. Environ. Microbiol. 59: 27952800.

Borgo, R., Souty-Grosset, C., Bouchon, D., and Gomot, L. 1996. PCR-RFLP analysis of mitochondrial DNA for
identification of snail meat species. J. Food Sci. 61(1): 14.

Bruce, J.L. 1996. Automated system rapidly identifies and characterizes microorganisms in foods. Food Technol.
50(1): 7781.

Bruce, J.L., Hubner, R.J., Cole, E.M., McDowell, C.I., and Webster, J.A. 1995. Sets of EcoRI fragments containing
ribosomal RNA sequences are conserved among different strains of Listeria monocytogenes. PNAS(USA) 92:
52295233.

Brunk, C.F., Avaniss-Aghajani, E., and Brunk, C.A. 1996. A computer analysis of primer and probe hybridization
potential and bacterial small subunit rRNA sequences. 1996. Appl. Environ. Microbiol. 62: 872879.
Candrian, U. 1995. Polymerase chain reaction in food microbiology. J. Microbiol. Methods 23: 89103.

Cano, R.J., Norton, D.M., Inzunza, A.E., Sanchez, J.G., and Oste, C. 1995. Polymerase chain reaction assay coupled
with fluorescence detection on

 Page 197

microwell plates for Listeria monocytogenes in foods. J. Food Prot. 58: 614620.

Cocconcelli, P.S., Porro, D., Galandini, S., and Senini, L. 1995. Development of RAPD protocol for typing of strains of
lactic acid bacteria and enterococci. Lett. Appl. Microbiol. 21: 376379.

Coleman, S.S., Melanson, D.M., Biosca, E.G., and Oliver, J.D. 1996. Detection of Vibrio vulnificus biotypes 1 and 2 in
eels and oysters by PCR amplification. Appl. Environ. Microbiol. 62: 13781382.

Cooray, K.J., Nishibori, T., Xiong, H., Matsuyama, T., Fujita, M., Mitsuyama, M. 1994. Detection of multiple
virulence associated genes of Listeria monocytogenes by PCR in artificially contaminated milk samples. Appl.
Environ. Microbiol. 60: 30233026.

Czajka, J., Bsat, N., Piani, M., Russ, W., Sultana, K., Wiedmann, M., Whitaker, R., and Batt, C.A. 1993.
Differentiation of Listeria moncytogenes and Listeria innocua by 16S rRNA and intraspecies discrimination of L.
monocytogenes strains by random amplified polymorphic DNA polymorphisms. Appl. Environ. Microbiol. 59:
304308.

Deng, M.Y., Day, S.P., and Cliver, D.O. 1994. Detection of hepatitis A virus in environmental samples by antigen
capture PCR. Appl. Environ. Microbiol. 60: 19271933.

Destro, M.T., Leitao, M.F.F., and Farber, J.M. 1996. Use of molecular typing methods to trace the dissemination of
Listeria monocytogenes in a shrimp processing plant. Appl. Environ. Microbiol. 62: 705711. (see Erratum in Appl.
Environ. Microbiol. 62: 18521853).

Dieffenbach, C. and Dveksler, G. (Ed.) 1995. PCR Primer: A Laboratory Manual. CSHL Press, Planeview, N.Y.

Drake, M., Small, C.L., Spence, K.D., and Swanson, B.G. 1996. Rapid detection and identification of Lactobacillus
spp. in dairy products using the polymerase chain reaction. J. Food Prot. 59: 10311036.

Dreader, C.A. 1996. Relief of amplification inhibition in PCR with bovine serum albumin or T4 gene 32 protein. Appl.
Environ. Microbiol. 62: 11021106.

Dussurget, O. and Rulland-Dussoix, D. 1994. Rapid sensitive PCR-based detection of mycoplasmas in simulated
samples of animal sera. Appl. Environ. Microbiol. 60: 953959.

Emond, E., Fliss, I., and Pandian, S. 1993. A ribosomal DNA fragment of Listeria monocytogenes and its use as a
specific probe in an aqueous phase hybridization assay. Appl. Environ. Microbiol. 59: 26902697.
Ericsson, H., Stalhandske, P., Danielsson-Tham, M.L., Bannerman, E., Bille, J., Jacquet, C., Rocourt, J., and Tham, W.
1995. Division of Listeria mono-

 Page 198

cytogenes serovar 4b strains into two groups by PCR and restriction enzyme analysis. Appl. Environ. Microbiol.
61: 38723874.

Fach, P., Gibert, M., Griffais, R., Guillou, J.P., and Popoff, M.R. 1995. PCR and gene probe identification of botulinum
neurotoxin A-, B-, E-, F-, and G producing Clostridium spp. and evaluation in foods samples. Appl. Environ.
Microbiol. 61: 389392.

Farber, J.M. and Addison, C.J. 1994. RAPD typing for distinguishing species and strains in the genus Listeria. J. Appl.
Bact. 77: 242250.
Fluitt, A.C., Torensma, R., Visser, M.J.C., Aaarsman, C.J.M., Poppelier, M.J.J.G., Keller, B.H.I., Klapwijk, P., and
Verhoef, J. 1993a. Detection of Listeria monocytogenes in cheese with the magnetic immuno-polymerase chain
reaction assay. Appl. Environ. Microbiol. 59: 12891293.

Fluit, A.C., Widjojoatmodjo, M.N., Box, A.T.A., Torensma, R., and Verhoef, J. 1993b. Rapid detection of salmonellae
in poultry with the magnetic immuno-polymerase chain reaction assay. Appl. Environ. Microbiol. 59: 13421346.

Fung, D.Y.C. 1995. What's needed in rapid detection of foodborne pathogens. Food Technol. 49(6): 6467.

Gendel, S.M. 1996. Computational analysis of the specificity of 16S rRNA-derived signature sequences for identifying
food-related microbes. Food Microbiol. 13: 115.

Gilgen, M., Wegmuller, B., Burkhalter, P., Buhler, H., Muller, U., Luthy, J., and Candrian, U. 1995. Reverse
transcription PCR to detect entero-viruses in surface waters. Appl. Environ. Microbiol. 61: 12261231.

Girardin, H., Monod, M., and Latge, J.P. 1995. Molecular characterization of the foodborne fungus Neosartorya
fischeri (Malloch and Cain). Appl. Environ. Microbiol. 61: 13781383.

Glass, N.L. and Donaldson, G.C. 1995. Development of primer sets designed for use with the PCR to amplify
conserved genes from filamentous ascomycetes. Appl. Environ. Microbiol. 61: 13231330.

Goswami, B.B., Koch, W.H., and Cebula, T.A. 1993. Detection of hepatitis A virus in Mercenaria mercenaria by
coupled reverse transcription and polymerase chain reaction. Appl. Environ. Microbiol. 59: 27652770.

Gouvea, V., Santos, N., Carmo-Timenetsky, M., and Do-Carmo-Timenetsky, M. 1994. Identification of bovine and
porcine rotavirus G types by PCR. J. Clin. Microbiol. 32: 13381340.
Guyader, F. le, Dubois, E., Menard, D., and Pommepuy, M. 1994. Detection of hepatitis A virus, rotavirus and
enterovirus in naturally contaminated shellfish and sediment by reverse transcription-seminested PCR. Appl. Environ.
Microbiol. 60: 36653671.

 Page 199

Harmon, K.M. and Wesley, I.W. 1995. Identification of Arcobacter isolates by PCR, p. 7072. In: Food Safety
Consortium Annual Research Meeting Progress Reports, Oct. 25, 1995, Kansas City, Mo.

Harvey, J. and Gilmour, A. 1994. Application of multilocus enzyme electrophoresis and restriction fragment length
polymorphism analysis to the typing of Listeria monocytogenes strains isolated from raw milk, nondairy foods, and
clinical and veterinary sources. Appl. Environ. Microbiol. 60: 15471553.

Herman, L.M.F., DeBlock, J.H.G.E., and Waes, G.M.A.V.J. 1995. A direct PCR detection method for Clostridium
tyrobutyricum spores in up to 100 milliliters of raw milk. Appl. Environ. Microbiol. 61: 41414146.
Hill, W.E. 1995. The polymerase chain reaction-applications for the detection of foodborne pathogens (review). Crit.
Rev. Food Sci. Nutr. 36(12): 123173.

Hubner, R.J., Cole, E.M., Bruce, J.L., McDowell, C.I., and Webster, J.A. 1995. Predicted types of Listeria
monocytogenes created by the position of EcoRI cleavage sites relative to rRNA sequences. PNAS(USA) 92:
52345238.

Ibeas, J.I., Lozano, I., Perdigones, F., and Jimenez, J. 1996. Detection of Dekkera-Brettanomyces strains in sherry by a
nested PCR method. Appl. Environ. Microbiol. 62: 9981003.

Innis, M.A., Gelfand, D.H., Sninsky, J.J., and White, T.J. 1990. PCR Strategies. Academic Press, Inc., Orlando, Fla.

Innis, M.A., Gelfand, D.H., and Snisnsky, J.J. (Ed.) 1995. PCR Strategies. Academic Press, Inc., San Diego, Calif.

Islam, M.S., Hasan, M.K., Hiah, M.A., Sur, G.C., Felsenstein, A., Venkatesan, M., Sack, R.B., and Albert, M.J. 1993.
Use of the polymerase chain reaction and fluorescent antibody methods for detecting viable but non-culturable Shigella
dysenteriae type 1 in laboratory microcosms. Appl. Environ. Microbiol. 59: 536540.

Jacquet, C., Catimel, B., Brosch, R., Buchrieser, C., Dehaumont, P., Goulet, V., Lepoutre, A., Veit, P., and Rocourt, J.
1995. Investigation related to the epidemic strain involved in the French listeriosis outbreak in 1992. Appl. Environ.
Microbiol. 61: 22422246.

Jaykus, L-A., de Leon, R., and Sobsey, M.D. 1995. Development of a molecular method for the detection of enteric
viruses in oysters. J. Food Prot. 58(12): 13571362.

Jensen, M.A., Webster, J.A., and Straus, N. 1993. Rapid identification of bacteria on the basis of polymerase chain
reaction-amplified ribosomal DNA spacer polymorphisms. Appl. Environ. Microbiol. 59: 945952.
Jinneman, K.C., Trost, P.A., Hill, W.E., Weagent, S.D., Bryant, J.L., Kaysner,

 Page 200

C.A., and Wekell, M.M. 1995. Comparison of template preparation methods from foods for amplification of E. coli
O157 Shiga-like toxins-I and II DNA by multiplex polymerase chain reaction. J. Food Protect. 58: 722726.

Johnson, E.W., Pieniazek, N.J., Griffin, D.W., Misener, L., and Rose, J.B. 1995. Development of a PCR protocol for
sensitive detection of Cryptosporidium oocysts in water samples. Appl. Environ. Microbiol. 61: 39493855.

Josephson, K.L., Gerba, C.P., and Pepper, I.L. 1993. Polymerase chain reaction detection of nonviable bacterial
pathogens. Appl. Environ. Microbiol. 59: 35133515.

Jothikumar, N., Aparna, I., Kamatchiammal, S., Paulmurugan, R., Saravandevi, S., and Khanna, P. 1993. Detection of
hepatits E virus in raw and treated wastewater with the polymerase chain reaction. Appl. Environ. Microbiol. 59:
25582562.

Kapperud, G., Vardund, T., Skjerve, E., Hornes, E., and Michaelsen, T.E. 1993. Detection of pathogenic Yersinia
enterocolitica in foods and water by immunomagnetic separation, nested polymerase chain reactions and colorimetric
detection of amplified DNA. Appl. Environ. Microbiol. 59: 29382944.

Kerr, K.G., Kite, P., Heritage, J., and Hawkey, P.M. 1995. Typing of epidemiologically associated environmental and
clinical strains of Listeria monocytogenes by random amplification of polymorphic DNA. J. Food Prot. 58: 609613.

Keuth, S. and Bisping, B. 1994. Vitamin B 12 production by Citrobacter freundii or Klebsiella pneumoniae during
tempeh fermentation and proof of enterotoxin absence by PCR. Appl. Environ. Microbiol. 60: 14951499.
Khan, A.A. and Cerniglia, C.E. 1994. Detection of Pseudomonas aeruginosa from clinical and environmental samples
by amplification of the exotoxin A gene using PCR. Appl. Environ. Microbiol. 60: 37393945.

Kirk, R. and Rowe, M.T. 1994. A PCR assay for detection of Campylobacter jejuni and Campylobacter coli in water.
Lett. Appl. Microbiol. 19: 301303.

Klijn, N., Weerkamp, A.H., and de Vos, W.M. 1995a. Detection and characterization of lactose-utilizing Lactococcus
spp. in natural ecosystems. Appl. Environ. Microbiol. 61: 788792.

Klijn, N., Nieuwenhof, F.F.J., Hoolwerf, J.D., van der Waals, C.B., and Weerkamp, A.H. 1995b. Identification of
Clostridium tyrobutyricum as the causative agent of late blowing in cheese by species specific PCR amplification.
Appl. Environ. Microbiol. 61: 29192924.

Koo, K., Swaisgood, H.E., and Foegeding, P.M. 1996. Development of single

 Page 201

chain antibodies for Bacillus cereus spores. Paper P-73, p. 381, presented at the 96th General meeting of American
Society for Microbiology, New Orleans, La., May 1923.

Lantz, P-G., Tjerneld, F., Borch, E., Hahn-Hagerdal, B., and Radstrom, P. 1994. Enhanced sensitivity in PCR detection
of Listeria monocytogenes in soft cheese through use of an aqueous two phase system as a sample preparation method.
Appl. Environ. Microbiol. 60: 34163418.

Lawrence, L.M. and Gilmour, A. 1994. Incidence of Listeria spp. and Listeria monocytogenes in poultry processing
environment and in poultry products and their rapid confirmation by multiplex PCR. Appl. Environ. Microbiol. 60:
46004604.
Lawrence, L.M. and Gilmour, A. 1995. Characterization of Listeria monocytogenes isolated from poultry products and
from the poultry processing environment by random amplification of polymorphic DNA and multilocus enzyme
electrophoresis. Appl. Environ. Microbiol. 61: 21392144.

Lawrence, L.M., Harvey, J., and Gilmour, A. 1993. Development of a random amplification of polymorphic DNA
typing method for Listeria monocytogenes. Appl. Environ. Microbiol. 59: 31173119.

Lee, C.Y., Pan, S.F., and Chen, C.H. 1995. Sequence of a cloned pR 72H fragment and its use for detection of Vibrio
parahaemolyticus in shellfish with the PCR. Appl. Environ. Microbiol. 61: 13111317.

Lees, D.N., Henshilwood, K., and Dore, W.J. 1994. Development of a method for detection of enteroviruses in
shellfish by PCR with poliovirus as a model. Appl. Environ. Microbiol. 60: 29993005.

Le-Jeune, C., Lonvaud-Fanel, A., ten-Brick, B., Hofstra, H., Vossen, J.M.B.M., Ten-Brink, B., and Vander Vossen,
J.M.B.M., 1995. Development of detection systems for histidine decarboxylating lactic acid bacteria based on DNA
probes, PCR, and activity tests. J. Expt. Med. 181 (3): 316326 (CAB abstracts).

Leng, X., Mosier, D.A., and Oberst, R.D. 1996. Simplified method for recovery and PCR detection of Cryptosporidium
DNA from bovine feces. Appl. Environ. Microbiol. 62: 643647.

Loessner, M.J., Rees, C.E.D., Stewart, G.S.A.B., and Scherer, S. 1996. Construction of luciferase reporter
bacteriophage A511::luxAB for rapid and sensitive detection of viable listeria cells. Appl. Environ. Microbiol. 62:
11331140.

Mahon, J., Murphy, C.K., Jones, P.W., and Barrow, P.A. 1994. Comparison of multiplex PCR and standard
bacteriological methods of detecting Salmonella on chicken skin. Lett. Appl. Microbiol. 19: 169172.

Makino, S.-I., Odkada, Y., and Maruyama, T. 1995. A new method for direct

 Page 202

detection of Listeria monocytogenes from foods by PCR. Appl. Environ. Microbiol. 61: 37453747.

Masters, C.I., Shallcross, J.A., and Mackey, B.M. 1994. Effect of stress treatments in the detection of Listeria
monocytogenes and enterotoxigenic E. coli by PCR. J. Appl. Bacteriol. 77: 7379.

Matthews, K.R. and Oliver, S.P. 1994. Differentiation of Staphylococcus species by polymerase chain reaction based
DNA fingerprinting. J. Food Prot. 57: 486489.

Miller, B., Witham, P., Norton, D., and Batt, C.A. 1996. Design of a 16S rRNA fluorogenic probe as an internal control
for 5' nuclease based assays designed to detect bacterial pathogens. Paper P-78, p. 382, presented at the 96th General
meeting of American Society for Microbiology, New Orleans, La., May 1923.

Muyzer, G., de Waal, E.C., and Uitterlinden, A.G., 1993. Profiling of complex microbial populations by denaturing
gradient gel electrophoresis analysis of polymerase chain reaction amplified genes coding for 16S rRNA. Appl.
Environ. Microbiol. 59: 695700.
Nakagawa, T., Shimada, M., Mukai, H., Asada, K., Kato, I., Fujino, K., and Sato, T. 1994. Detection of alcohol tolerant
hiochi bacteria by PCR. Appl. Environ. Microbiol. 60: 637640.

Niederhauser, C., Hofelein, C., Allmann, M., Burkhalter, P., Luthy, J., and Candrian, U. 1994a. Random amplification
of polymorphic bacterial DNA: Evaluation of 11 oligonucleotides and application to food contaminated with Listeria
monocytogenes. J. Appl. Bacteriol. 77: 574582.
Niederhauser, C., Hofelein, C., Wegmuller, B., Luthy, J., and Candrian, U. 1994b. Reliability of PCR decontamination
systems. PCR Methods and Applic. 4(2): 117123.

Olsen, J.E., Aabo, S., Hill, W., Notermans, S., Wernars, K., Granum, P.E., Popovic, T., Rasmussen, H.N., and Losvik,
O. 1995. Probes and polymerase chain reaction for detection of food-borne bacterial pathogens. Int. J. Food Microbiol.
28: 178.

Oyofo, B.A. and Rollins, D.M. 1993. Efficacy of filter types for detection of Campylobacter jejuni and Campylobacter
coli in environmental water samples by polymerase chain reaction. Appl. Environ. Microbiol. 59: 40904095.

Pogue, G.P., Koul, S., Lee, N.S., Swyer, D.M., and Nakhael, H.L. 1995. Identification of intra- and inter-species
Leishmania genetic polymorphisms by arbitrarily primed polymerase chain reaction and use of polymorphic DNA to
identify differentially regulated genes. Parasitol. Res. 81: 282290 (CAB abstracts).

Rafil, F., Holland, M.A., Hell, W.E., and Cerniglia, C.E. 1995. Survival of

 Page 203

Shigella flexneri on vegetables and detection by polymerase chain reaction. J. Food Prot. 58: 727732.

Reynolds, K.A., Gerba, C.P., and Pepper, I.L. 1996. Detection of infectious enteroviruses by an integrated cell culture-
PCR procedure. Appl. Environ. Microbiol. 62: 14241427.

Rijpens, N.P., Jannes, G., Van Asbroeck, M., Rossau, R., and Herman, L.M.F. 1996. Direct detection of Brucella spp.
in raw milk by PCR and reverse hybridization with 16S-23S rRNA spacer probes. Appl. Environ. Microbiol. 62:
16831688.

Rivera, I.G., Chowdhury, M.A.R., Huq, A., Jacobs, D., Martins, M.T., and Colwell, R.R. 1995. Enterobacterial
repetitive intergenic consensus sequences and the PCR to generate fingerprints of genomic DNAs from Vibrio cholerae
01, 0139, and non-01 strains. Appl. Environ. Microbiol. 61: 28982904.

Roberts, T., Jensen, H., and Unnevehr, L. (Ed.). 1995. Tracking Foodborne Pathogens from Farm to Table: Data Needs
to Evaluate Control Options. Miscellaneous Publication N. 1532, USDA Economic Research Service Food and
Consumer Economics Division, Herndon, Va.

Rodriguez, J.M., Cintas, L.M., Casaus, P., Suarez, A., and Hernandez, P.E. 1995. PCR detection of the lactocin S
structural gene in bacteriocin producing lactobacilli from meat. Appl. Environ. Microbiol. 61: 28022805.

Schraft, H. and Griffiths, M.W. 1995. Specific oligonucleotide primers for detection of lecithinase positive Bacillus
spp. by PCR. Appl. Environ. Microbiol. 61: 98102.
Schraft, H., Miller-Pipper, S., Buechin, A., and Griffiths, M.W. 1996. Identification of DNA sequences specific for
emetic Bacillus cereus. Paper P-86, p. 384, presented at the 96th General meeting of American Society for
Microbiology, New Orleans, La., May 1923.

Simon, M.C., Gray, D.I., and Cook, H. 1996. DNA extraction and PCR methods for the detection of Listeria
monocytogenes in cold smoked salmon. Appl. Environ. Microbiol. 62: 822824.

Soumet, C., Ermel, G., Fach, P., and Colin, P. 1994. Evaluation of different DNA extraction procedures for detection of
Salmonella from chicken products by polymerase chain reaction. Lett. Appl. Microbiol. 19: 294298.

Strachan, N.J.C. and Gray, D.I. 1995. A rapid general method for the identification of PCR products using a fiber-optic
biosensor and its application to the detection of Listeria. Lett. Appl. Microbiol. 21: 59.

Stubbs, S., Hutson, R., James, S., and Collins, M.D. 1994. Differentiation of the spoilage yeast Zygosaccharomyces
bailii from other Zygosaccharomyces

 Page 204

species using 16S rRNA target for non-radioactive ligase chain reaction. Lett. Appl. Microbiol. 19: 268272.

Suzuke, M.T. and Giovannoni, S.J. 1996. Bias caused by template annealing in the amplification of mixtures of 16S
rRNA genes by PCR. Appl. Environ. Microbiol. 62: 625630.

Szabo, E.A., Pemberton, J.M., and Desmarchelier, P.M. 1993. Detection of the genes encoding botulinum neurotoxin
types A to E by the polymerase chain reaction. Appl. Environ. Microbiol. 59: 30113020.

Tsen, H.Y., Chen, T.R., and Yu, G.K. 1994. Application of PCR for the specific detection of enterotoxigenic
Staphylococcus aureus in various food samples. J. Food Drug Anal. 2:217224 (CAB abstracts).
Tsen, H.Y., Yu, G.K., and Lin, I.T. 1995. Plasmid profiles and pulsed field electrophoresis for type A enterotoxigenic
Staphylococcus aureus isolated from foods. J. Food Prot. 68: 147153.

Van Camp, G., Van de Peer, Y., Nicolai, S., Neefs, J-M., Vandamme, P., and De Wachter, R. 1993. Structure of 16S and
23S ribosomal RNA genes in Campylobacter species: Phylogenetic analysis of the genus Campylobacter and presence
of internal transcribed spacers. System. Appl. Microbiol. 16: 361368.

Wagner, M., Maderner, A., and Brandl, E. 1996. Random amplification of polymorphic DNA for tracing and molecular
epidemiology of Listeria contamination in a cheese plant. J. Food Prot. 59: 384389.
Wang, R.F., Cao, W.W., and Johnson, M.G. 1992. 16S rRNA based probes and polymerase chain reaction method to
detect Listeria monocytogenes cells added to foods. Appl. Environ. Microbiol. 58: 28272831.

Wang, R.F., Cao, W.W., Wang, H., and Johnson, M.G. 1993. A 16S rRNA-based DNA probe and PCR method specific
for Listeria ivanovii FEMS Microbiol. Lett. 106: 8592.

Wang, R.F., Cao, W.W., Johnson, M.G., and Cerniglia, C.E. 1995. Comparison of PCR cycler machines for rapid and
sensitive detection of pathogens. The Rapid Cyclist 3(1): 1214.

Wang, R.F., Cao, W.W., and Cerniglia, C.E. 1996. PCR detection and quantitation of predominant anaerobic bacteria in
human and animal fecal samples. Appl. Environ. Microbiol. 62: 12421247.

Weagant, S.D., Bryant, J.L., and Jinneman, K.G. 1995. An improved rapid technique for isolation of E. coli O157:H7
from foods. J. Food Prot. 58: 712.
Wegmuller, B., Luthy, J., and Candrian, U. 1993. Direct polymerase chain reaction detection of Campylobacter jejuni
and Campylobacter coli in raw milk and dairy products. Appl. Environ. Microbiol. 59: 21612165.

 Page 205

Wiedmann, M., Stolle, A., and Batt, C.A. 1995. Detection of Listeria monocytogenes in surface swabs using a
nonradioactive polymerase chain reaction-coupled ligase chain reaction assay. Food Microbiol. 12: 151157.

Wiedmann, M., Bruce, J.L., Johnson, A., Cole, E.M., Kating, C., McDowell, C.I., McDonough, P.L., and Batt, C.A.
1996a. Association of pattern types and allelic diversity of virulence genes in Listeria monocytogenes. Paper P-45, p.
376, presented at the 96th General meeting of American Society. for Microbiol., New Orleans, La., May 1923.

Wiedmann, M., Bruce, J L., Knorr, R., Bodis, M., Cole, E.M., McDowell, C.I., McDonough, P.L., and Batt, C.A.
1996b. Ribotype diversity of Listeria monocytogenes strains associated with outbreaks of listeriosis in ruminants. J.
Clin. Microbiol. 34(5): 10861090.
Winters, D.K, and Slavik, M.F. 1995. Evaluation of a PCR based assay for specific detection of Campylobacter jejuni
in chicken washes. Mol. Cell. Probes 9: 307310.

Winters, D.K., Maloney, T., and Johnson, M.G. 1996a. Detection of Listeria monocytogenes by a PCR assay specific
for an aminopeptidase. Paper 68C9, p. 148, presented at the 56th Annual Meeting of Inst of Food Technologists, New
Orleans, La., June 2226.

Winters, D.K., O'Leary, A.E., and Slavik, M.F. 1996b. Rapid detection of Campylobacter jejuni in chicken products by
a nested PCR assay. Paper #61, p. 46. Presented at Ann. Meeting of Int. Assoc. Milk Food and Environ Sanitarians,
Seattle, Wash.

Witham, P.K., Yamashiro, C.T., Livak, K.J., and Batt, C.A. 1996. A PCR-based assay for the detection of Escherichia
coli shiga-like toxin genes in ground beef. Appl. Environ. Microbiol. 62: 13471353.

Wolcott, M.J. 1991. DNA-based rapid methods for the detection of foodborne pathogens. J. Food Prot. 54: 387401.

Young, C.C., Hurghoff, R.L., Keim, L.G., Minak-Bernero, V., Lute, J.R., and Hinton, S.M. 1993.
Polyvinylpyrrolidone-agarose gel electrophoresis purification of polymerase chain reaction amplifiable DNA from
soils. Appl. Environ. Microbiol. 59: 19721974.

Yuan, G.F., Liu, C.S., and Chen, C.C. 1995. Differentiation of Aspergillus parasiticus from Aspergillus sojae by
random amplification of polymorphic DNA. Appl. Environ. Microbiol. 61: 23842387.

 Page 207

10
Detection of Microorganisms in Foods Using DNA Probes Targeted to Ribosomal RNA Sequences
Mark A. Mozola
GENE-TRAK Systems
Hopkinton, Massachusetts

Introduction

With a history of development and practical application now spanning more than a decade, DNA probe-based methods
have been established as powerful tools for use in the microbiological examination of foods. Where such methods were
once considered esoteric and experimental, they are now used routinely in hundreds of laboratories worldwide in the
assessment of raw materials and finished food products for quality and safety. Probe-based methods have proven to be
especially useful in the qualitative determination of pathogenic bacteria in foods, particularly in situations where a high
degree of specificity is required.
Excellent reviews are available summarizing the field of nucleic acid hybridization and DNA probe technology (Hill
and Keasler, 1991; Jinneman and Hill, Chapter 8 this volume; Tenover, 1993, 1988; Wolcott, 1991). The topics of
nucleic acid sequence data bases, nucleic acid amplification technology, and molecular typing methods, including
ribotyping, are covered in detail elsewhere in these proceedings and will not be discussed in this

 Page 208

chapter. Here we will consider probe-based methods that target specific sequences of ribosomal RNA, with emphasis
on probe design and the application of these methods to difficult problems in the detection of pathogenic bacteria in
foods.

Ribosomal RNA as a Target

Specific sequences within chromosomal DNA (or RNA in some viruses), ribosomal RNA (rRNA), messenger RNA, or
extrachromosomal DNA can all be used as targets in diagnostic hybridization assays. Chromosomal nucleic acids and
rRNA are the most commonly targeted, and rRNA is a particularly useful choice for a variety of reasons. First, rRNA is
present in all organisms except viruses, allowing for a consistency of approach in assay design across a wide range of
organisms of interest. Second, as an essential structural and functional component of ribosomes, rRNA is present in
relatively high copy number (about 104 molecules per cell in most bacteria). This natural abundance of target provides
an automatic enhancement of assay sensitivity. Third, and certainly most interesting as a point of discussion,
evolutionary constraints of the divergence of rRNA sequences make them useful as a focal point in studies into the
phylogenetic relationships of microorganisms (Woese, 1987).

The importance of phylogenetics in the design of probe assays targeting rRNA has been reviewed (Lane and Collins,
1991). A few points will be discussed and illustrated here. Bacteria are much more closely related at the level of rRNA
sequence than at the level of DNA sequence. The Enterobacteriaceae serve as a useful example, as illustrated in Fig.
10.1. Comparing Salmonella typhimurium to one of its nearest genetic relatives, Citrobacter freundii, the organisms are
related to an extent of about 4050% as determined by DNA/DNA hybridization experiments. In contrast, when
comparing 16S rRNA sequences, the organisms are about 96% related. Similar close relationships are found among
other members of the Enterobacteriaceae. The 16S rRNAs of two Salmonella serovars, S. typhimurium and S. arizonae
(these organisms belonging to distinct subspecies), are about 97% homologous. Escherichia coli and Shigella flexneri
are even more closely related, with greater than 99% homology in 16S rRNA sequence. In the latter case, homology is
so extensive that a search for probes specific for Shigella spp. was unsuccessful (Lane and Collins, 1991).

This conservation of sequence is what has made rRNA an indispensible tool in genetic studies of microbial evolution
and relatedness. As an integral part of these studies, large data bases of rRNA sequence information have been
generated (see Chapter 11), facilitated by advances in methodology

 Page 209

Fig. 10.1
DNA and rRNA relatedness among the Enterobacteriaceae. Percent relatedness
between any pair of species or genera corresponds to the position of the
rightmost branch point of the line segments of those groups. Ribosomal RNA
relatedness refers to 16S and was compiled by Lane and Collins (1991).
DNA relatedness data are from Brenner (1984).
(From Lane and Collins, 1991.)

for direct sequencing of rRNA (Lane et al., 1985). Computer-assisted comparison of sequences identifies the most
conserved and the most variable regions within and across taxonomic boundaries. This serves as the starting point in
probe design.

Probe Design
Comparative examination of rRNA sequences from data bases has facilitated the construction of probes with an
impressively wide range of specificities, from all-life and kingdom-specific probes to probes specific for a single genus
(e.g., Salmonella), a subset of organisms within a genus (e.g., Campylobacter jejuni plus C. coli, C. lari, and C. fetus
subsp. fetus), or a single species (e.g., Listeria monocytogenes). The subject of designing probes targeted to

 Page 210

rRNA sequences, citing these and other examples, has been reviewed by Lane and Collins (1991). Generally, the more
highly variable regions of rRNA will prove most fruitful in identifying target sequences for differentiation of closely
related organisms. Following the initial design of candidate probes based on analysis of rRNA sequences from target
and nontarget organisms, these candidate probes are tested, usually by dot-blot hybridization, for reactivity with large
panels of organisms both within and outside of the target group. The sequences of candidate probes are then adjusted,
followed by more empirical testing, and this process is continued until probes with the desired specificity are obtained.
Most typically, probes are synthetic oligonucleotides approximately 2040 bases in length.

What becomes apparent at an early stage in the probe-design process is that there are ''sets" of closely related target and
nontarget sequences, this heterogeneity of sequence within both target and nontarget groups complicating the analysis.
We will return to Salmonella as an instructive example. In Fig. 10.2, sequences in a potential 16S rRNA target region
are shown for several strains of Salmonella, Citrobacter, and Enterobacter. Examination of nearly 1000 stains of
Salmonella and other enteric bacteria revealed five basic sequence patterns for Salmonella in this target region. The
vast majority of common Salmonella strains contain pattern 1, while nearly all Citrobacter and Enterobacter strains
examined have sequences very different from pattern 1. It appeared that this region held promise as a probe target for
the majority of salmonellae. Disappointingly, testing of the large number of nontarget strains revealed one strain,
Citrobacter spp. strain 135, solidly classified as a

Fig. 10.2
16S rRNA variation among Salmonella, Citrobacter, and Enterobacter.
The region corresponding to nucleotide positions 443491 is shown. The
sequence found in S. typhimurium is shown in total (pattern 1). For other
sequences, positions with nucleotides different from those found in pattern
1 are indicated. Y = C or U, R = A or G. (From Lane and Collins, 1991.)

 Page 211

Citrobacter on biochemical and antigenic grounds, with total homology to pattern 1 in this target region. This was of
concern since Citrobacter spp. are commonly found in food products. Similarly, a single strain of Enterobacter, strain
126, was found to have a sequence in the target region identical to that of Salmonella pattern 2 (Lane and Collins,
1991). This example stresses the importance of testing large numbers of strains when screening candidate probes.
Fortunately, in this case, other target regions were found, which yielded probes with the desired inclusivity for
Salmonella without problematic cross reactivity with nonsalmonellae (Wilson et al., 1990; Foster et al., 1992). It
should be noted at this point that the probes used in an assay do not necessarily have to be absolutely specific in order
for the method to be completely specific. Selective enrichment of the sample may in some cases prevent the growth of
nontarget, but possibly cross-reactive, organisms to detectable titers. However, in practice this cannot always be relied
upon; the organisms most closely related genetically to the target organism are usually the same ones that are most
likely to grow through selective enrichment. Additionally, the degree of selectivity of culture media can vary from
batch to batch.

Besides sequence homology, the other major factor in controlling the specificity of hybridization is the degree of
stringency of the hybridization reaction. High-stringency conditions (higher temperature, lower salt concentration, the
presence of denaturing agents such as formamide or chaotropic salts, or shorter probes) will have the effect of requiring
a more precise base pairing between probe and target sequences in order for stable duplexes to be formed. Higher
stringency conditions can be used to maximize the specificity of an assay, e.g., to discriminate salmonellae from some
closely related enterics. However, in practice, pushing stringency too far can result in an undesirable loss of inclusivity
for the target group of organisms. The challenge in assay development is to use careful probe design and control of
assay stringency so that specificity and inclusivity are optimally balanced.

Several tools in probe design are available to facilitate maximizing discrimination between sets of target and nontarget
sequences. Often, multiple probes are used in combination to provide inclusivity for a target group exhibiting
considerable sequence heterogeneity. Mismatches between probe and target sequences will have a more destabilizing
effect if they are positioned near the middle of the probe rather than toward the ends. The so-called noncanonical base
pairs (G:U and A:G) are often tolerated depending on their positioning and, in fact, can be put to strategic use. These
and other concepts in probe design are discussed by Lane and Collins (1991). Use of mismatch placement and
noncanonical base pairs are illustrated in a hypothetical example in Fig. 10.3.

 Page 212

Fig. 10.3
Probe design. A hypothetical example of probe
design for sets of related target and nontarget
sequences is presented. Mismatches with the
hypothetical probe sequence are underlined. The
use of mismatch placement and noncanonical
base pairs to accommodate variation among target
sequences is illustrated.
(From Lane and Collins, 1991.)

Another approach that can be used to achieve difficult specificities is a dual probe or sandwich hybridization assay.
Using this design strategy, certain undesirable cross-hybridizations can be reduced or eliminated. In one example,
probes to two different target regions are used in parallel: one as a capture probe and the other as a detector probe. The
probes hybridize to different regions on the same target molecule, with hybridization of both probes to target required
in order to generate an assay signal (Ranki et al., 1983). The end result of the two probes working in parallel is an assay
specificity greater than that which can be achieved using either probe alone. This requires that both probes be fully
inclusive for the target group and that the two probes exhibit nonoverlapping patterns of cross-hybridization to
nontarget sequences. An example of this sandwich hybridization strategy employed in a commercially available test
system is discussed in detail in a later section.

Assay Formats

A variety of formats are possible for nucleic acid hybridization assays, including solid-phase hybridization, solution
hybridization with subsequent hybrid capture, and homogeneous formats requiring no separate steps. Many different
detection systems are also possible, including, but not limited to, radioisotopes, direct fluorescent or chemiluminescent
labels, and enzyme labels which act on substrates to produce colorimetric or chemiluminescent endpoints. Much more
information on assay formats and detection systems can be found in the review by Wolcott (1991) and in Chapter 8.

Probes targeted to rRNA can be used, in principle, in essentially any of these assay formats and with many different
detection systems. In the following sections we will discuss in detail rRNA-targeted hybridization assay

 Page 213

systems developed specifically for detection of pathogenic and indicator bacteria in foods.

The GENE-TRAK® Colorimetric Hybridization Assays

A sandwich hybridization format using probes targeted to rRNA has been used in the development of a line of
diagnostic assays for a variety of foodborne bacteria (GENE-TRAK colorimetric assays; GENE-TRAK Systems,
Hopkinton, Mass.). This test system and several of the commercially available products based on it have been
previously reviewed (Chan et al., 1989; Mozola et al., 1991). The assays are alternatives to conventional culture
procedures and are performed following enrichment of the test sample, usually by modifications of broth culture
enrichments employed in standard methods. The combination of sometimes abbreviated enrichment periods and the
circumvention of the selective/differential plating steps common to standard culture procedures produces methods that
yield presumptive results in significantly less time than that required by conventional methods. In most cases results
are available within 2448 hours from the start of analysis. Positive results can then be confirmed, if necessary, by
standard plating techniques. Besides the time savings realized when using these methods as a negative screen, they
offer other advantages to the microbiologist, which will be discussed in more detail in the following section.
The assay format and test chemistry are described here briefly. The assays are performed in standard 12 × 75 mm glass
test tubes. Bacteria present in the enriched test sample are lysed using alkali or enzymatic digestion, liberating rRNA
from the cells. A cocktail of synthetic oligonucleotide probes is then introduced for the hybridization reaction, which is
performed in a common phosphate/salt matrix, usually at 65°C for one hour. The probe cocktail consists of at least two
probes: minimally one "capture probe" and one "detector probe." The capture probe contains a 3' tail of
polydeoxyadenylic acid (poly dA), which serves to capture probe:target hybrids from solution. The detector probe is
end-labeled with fluorescein, a component of the colorimetric detection system, which will later indicate the presence
of target rRNA in the test sample. Concurrent with the beginning of the hybridization reaction, the solid phase, a plastic
dipstick coated with polydeoxythymidylic acid (poly dT), is introduced. Complementarity between the poly dA
component of the capture probe and the poly dT on the dipstick results in capture probe:rRNA-target:detector probe
complexes being bound to the dipstick. The dipsticks are then washed to remove unbound probe. Next, the dipsticks
are introduced into a solution containing a conjugate of anti-fluorescein polyclonal antibody (anti-FL) and the enzyme
horseradish peroxidase (HRP) and incubated at room temperature

 Page 214

for 20 minutes. The anti-FL-HRP conjugate binds to the fluorescein moiety on the hybridized detector probe. The
dipsticks are washed to remove unbound conjugate, then transferred to tubes containing a substrate-chromogen reagent
and incubated at room temperature for 2030 minutes. The dipsticks are removed and discarded, the reaction is stopped
with sulfuric acid, and A450 determined using a photometer. Absorbance in excess of a threshold value indicates a
positive result.

The first assays developed in this format were for Salmonella (Wilson et al., 1990), Escherichia coli (Hsu et al., 1991),
and Listeria (King et al., 1989, 1990). The Salmonella and Listeria assays are by far the most extensively validated and
the most widely used. The Salmonella method is appropriate for all food types and has been comprehensively validated
in internal studies (Wilson et al., 1990; Chan et al., 1990; Foster et al., 1992) and in an AOAC International
interlaboratory collaborative study (Curiale et al., 1990a). The high degree of inclusivity and specificity of the assay
has also been independently validated (Curiale et al., 1990b; D'Aoust et al., 1995). The Listeria method is inclusive for
all members of the genus and is applicable to the analysis of dairy products, meats, seafoods, and environmental
samples. This method has also been extensively validated in internal studies (King et al., 1989, 1990; Bottari et al.,
1995) and in collaborative studies under the protocols of AOAC International (Curiale et al., 1994) and the Association
Française de Normalisation (AFNOR). An assay for Listeria monocytogenes has been developed (Stanick et al., 1993;
Durbin et al., 1994) for use in situations where specific detection of this pathogenic species in foods and environmental
samples is desired and the presence of other listeriae are not of concern. This method has also been independently
validated in an AFNOR collaborative study. This assay also has utility in identification of isolates in pure culture
(FDA, 1995a). An assay specific for the pathogenic campylobacters C. jejuni, C. coli, C. lari, and C. fetus subsp. fetus
is also available (Chan et al., 1989; Keough et al., 1994). Enrichment procedures for poultry samples for use in
conjunction with the assay have been defined and the utility of the method for detection of these campylobacters in
poultry has been demonstrated (Stern and Mozola, 1992; Keough et al., 1994; Ransom et al., 1994). Methods for
Staphylococcus aureus (Reynolds et al., 1991) and Yersinia enterocolitica (Chan et al., 1988) have also been described.
Several of the assays have undergone refinement from their original versions. The most recent developments in the
evolution of these assays will be discussed in the final section of this chapter.

Other rRNA-Based Systems

A homogeneous hybridization assay system employing probes targeting rRNA has been described (Arnold et al., 1989).
This system, utilizing oligo-

 Page 215

nucleotide probes labeled with a chemiluminescent compound, forms the basis for a collection of test kits developed
for human clinical diagnostic applications (Accuprobe® assays; Gen-Probe, Inc., San Diego, CA). However, at least
two of these assays, those for Campylobacter spp. (C. jejuni, C. coli, and C. lari) and Listeria monocytogenes, have
been described as having utility in applications in food microbiology. The L. monocytogenes assay has been applied to
identification of pure culture isolates (Okwumabua et al., 1992; Ninet et al., 1992; Johnson and Lattuada, 1993; FDA,
1995a) and also to screening of enriched food samples. In the latter application care must be taken as some selective
enrichment media have proven to be inhibitory to the assay (Ninet et al., 1992; Niederhauser et al., 1993; Partis et al.,
1994). The Campylobacter assay has shown promise in the detection of Campylobacter spp. in enriched poultry
samples (Ransom et al., 1994).

Application of DNA Probe Methods in Food Microbiology: Examples

We will consider now some of the practical challenges in detection of pathogenic and indicator bacteria in foods and
how DNA probe-based methods such as those described above can be of assistance to the food microbiologist. First I
offer a few general comments about diagnostic methods offered as alternatives to conventional culture procedures.
Probe assays, immunoassays, and other methods based on new or alternative technologies are commonly referred to as
"rapid methods," reflecting their usual ability to produce results in significantly less time than traditional culture
procedures. It should be kept in mind, however, that very often a shorter time to result is not the only benefit these
methods can offer, and in fact some of their other advantages can be of considerably greater importance in certain
situations. Among these other potential advantages are objective results, ability to be performed with less labor or by
technicians with less extensive training or experience, improved laboratory workflow, and improved method
sensitivity, inclusivity, and/or specificity. Some of these points are illustrated in the following examples.

Salmonella spp.

Several special challenges exist in the detection of Salmonella in food samples. One of the foremost is, of course,
differentiation of salmonellae from the milieu of extremely closely related (genetically and phenotypically) bacteria
that can be present in foods. Most of these organisms are other Enterobacteriaceae such as Citrobacter, Enterobacter,
Escherichia, Klebsiella, and Proteus spp. Many of these organisms are capable of surviving selective enrichment and
forming colonies that mimic salmonellae on selective/

 Page 216

differential agars. Much of the labor associated with detection of salmonellae by conventional plating methods, then,
relates to identification of suspect colonies. This problem is especially acute when testing foods with a heavy
background microflora, such as meats and eggs. Identification of multiple isolated colonies in these cases can resemble
the proverbial "needle in the haystack" and can add considerable time and expense to the analysis. In addition, the
presence of Salmonella can be masked by heavy competitor loads, increasing the chance of missing the presence of
Salmonella in a sample. Obviously, a primary advantage of a very specific screening assay is the ability to screen out
these competitors at the stage of assay from broth culture enrichments. Only those samples producing positive assay
results need be plated to selective/differential agars for confirmation and identification.

A second problem in Salmonella detection is that a significant percentage of salmonellae isolated from certain foods
[up to 4% according to FDA (1995b)] are biochemically atypical, for example, H2S-negative or lactose-positive. Unless
atypical colonies are also considered, the risk of false-negative determinations is increased. For a probe assay, such as
the GENE-TRAK assay, which is inclusive for biochemically atypical strains, the situation is not so problematic. A
positive assay result will direct the analyst to consider atypical, as well as typical, colonies at the confirmation step. It
is noteworthy that in the most recent revision of FDA's standard culture procedure for Salmonella in foods (FDA,
1995b), a fundamentally important change has been made requiring the examination of atypical colonies in the absence
of typical colonies. This development means that more extensive confirmatory work may be required when performing
the culture method and serves to further increase the added value of specific alternative screening methods such as
probe assays.
Listeria spp. and Listeria monocytogenes

Needs vary widely in Listeria testing, from broad screening for listeriae in environmental samples, to specific detection
of L. monocytogenes in food products, to conclusive identification of L. monocytogenes from isolated colonies.
Fortunately, a variety of methods are available to satisfy these different needs.
The presence of non-monocytogenes listeriae is commonly used as an indicator of conditions favorable to the potential
presence of L. monocytogenes. This determination is typically performed on swab or sponge samples taken from food-
processing environments. Rather than isolating and identifying listeriae by standard culture procedures, it is often faster
and simpler to perform a screening assay, such as a genus specific probe assay, on culture

 Page 217

enrichments of these samples. In this case, it may be unnecessary to confirm assay results, due not only to the inherent
specificity of the assay, but also to the fact that in this application one may be using the information to identify patterns
or trends and not to make decisions about the status of finished products.

In the testing of finished products, the presence of non-monocytogenes listeriae may not be of major concern. In this
case, use of a screening method specific for only L. monocytogenes can assist the analysis by providing an early result
indicating the specific presence of the pathogenic species. In addition, and of even greater importance, applying an
assay specific for L. monocytogenes can increase the chances of finding L. monocytogenes in samples co-contaminated
with other listeriae. It is well established that L. innocua can have a competitive advantage over L. monocytogenes in
mixed culture and that the presence of L. innocua can mask the presence of L. monocytogenes and lead to false-
negative results when performing standard culture methods (Petran and Swanson, 1993; Curiale and Lewus, 1994). It is
simply a question of how many suspect colonies to pick and identify. Standard culture methods typically call for
picking "five or more" colonies (FDA, 1995a). In situations where the ratio of L. monocytogenes to other listeriae is
poor following enrichment, five colonies may be insufficient to find the minority population of L. monocytogenes in
the sample. Use of a probe assay specific for L. monocytogenes will alert the microbiologist to the potential need to
apply more rigorous attempts at confirmation.

Finally, L. monocytogenesspecific probe assays can be useful in the confirmation process itself. They can be used in
conjunction with, or in place of, biochemical tests for the identification of isolates in pure culture. Use of tests such as
the GENE-TRAK or Accuprobe L. monocytogenes assays for this purpose is discussed by FDA (1995a).

Escherichia coli
E. coli presents another interesting example of the potential of DNA probe methods. Two basic types of E. coli testing
are performed in the food industry: determination of specific pathogenic strains (especially the enterohemorrhagic
serotype O157:H7) and determination of the presence or enumeration of total E. coli, which is used as an indicator of
fecal contamination (FDA, 1995c). The subject of pathogenic E. coli is very complex and will not be considered in
detail here. Several of the methods described for detection of specific pathogenic varieties involve the use of DNA
probes (FDA, 1995c,d).

Methods used for detection of total E. coli in foods are usually based on direct plating or most probable number (MPN)
procedures. Some of the

 Page 218

primary diagnostic reactions used in these methods are production of gas from lactose, the MUG reaction (a
fluorogenic assay for active b-glucuronidase), and growth at elevated temperature (4344°C) (reviewed by FDA,
1995c). It is important to note that E. coli exhibit considerable strain-to-strain variability in their biochemical profiles
(Ewing, 1986), and, in particular, pathogenic varieties tend to be phenotypically unusual in one or more ways. Perhaps
the most striking example is the characteristically MUG-negative phenotype of E. coli O157:H7 strains (Doyle and
Schoeni, 1984); these strains will, a priori, go undetected in the MUG screening assay. This can have practical
significance if one is concerned with the potential presence of any E. coli, as one might be in testing a sensitive product
such as infant formula. A probe-based assay inclusive for all varieties of E. coli, including the pathogenic types, can
offer greater assurance that such products are truly E. coli free. This principle is illustrated by the results of inclusivity
testing of the GENE-TRAK E. coli assay. Of 207 E. coli isolates examined (including 11 O157:H7 isolates), all 207
were reactive in the hybridization assay, whereas only 195 were positive for gas from lactose and only 177 were
positive in the MUG assay (Mozola et al., 1991). Additionally, in a study with inoculated food samples, the probe assay
demonstrated a significantly greater sensitivity in the detection of E. coli compared with a standard MPN procedure
(Hsu et al., 1991).

Recent Advances and Future Directions

A new format has recently been developed for the colorimetric GENE-TRAK hybridization assays. Significant changes
have been made to the assay chemistry, resulting in a test format with reduced assay time and improved handling
characteristics. Salmonella is the first application to be developed in this new test format, referred to as the direct
labeled probe (DLP) assay. Development and preliminary validation of the Salmonella DLP assay has been previously
described (Moore et al., 1995) and is reviewed here. The most fundamental changes involve hybridization conditions
and the detector probe. The hybridization temperature has been reduced from 65 to 37°C, accomplished by the addition
of formamide to the hybridization matrix. The posthybridization wash temperature has also been reduced, from 65°C,
to room temperature. The fluorescein-labeled detector probe has been replaced with an oligonucleotide covalently
linked to HRP at the 5' end, eliminating the need for the anti-fluorescein-HRP conjugate used in the original GENE-
TRAK assay. The DLP assay chemistry is illustrated in Fig. 10.4.

The entire assay is performed at 37°C and room temperature, with the

 Page 219

Fig. 10.4
Direct labeled probe hybridization assay. HRP = horseradish peroxidase; dA = polydeoxyadenylic acid;
dT = polydeoxythymidylic acid.

exception of a short lysis step, which is performed at 65°C. The DLP assay chemistry is summarized in Fig. 10.5. The
ability to perform the wash step at room temperature and the hybridization at the more easily controlled temperature of
37°C (vs. 65°C) presents the user with the option to use a compact heater block rather than the customary water bath
for the hybridization step. Elimination of the enzyme conjugate incubation step and the postconjugate wash step
shortens the assay time by approximately 22 minutes to a total of 87 minutes. Altogether, these changes result in an
assay that is shorter, easier to perform, less demanding in its requirement for bench space, and much more amenable to
automation.

The Salmonella DLP assay has been validated in internal studies. Results of sensitivity, inclusivity, and specificity
testing have been reported (Moore et al., 1995) and are summarized in Tables 10.1 and 10.2. Sensitivity is comparable
to that of the original colorimetric GENE-TRAK assay, with all strains tested reactive at 107 CFU/assay and most
strains reactive at 106 CFU/assay (Table 10.1). The probes used in the DLP assay are essentially identical to those used
in the original GENE-TRAK assay, with the exception that a

 Page 220

1. Add 0.2 mL enriched test sample to 12 × 75 mm tube.

2. Add 0.1 mL lysis solution, incubate 5 min at 65°C.

3. Transfer tubes to 37°C, add 0.4 mL hybridization solution, add 0.2 mL probe solution, add dipsticks, mix and
incubate 1 hr at 37°C.

4. Wash dipsticks twice for 1 min each at room temperature.

5. Transfer dipsticks to tubes containing 0.75 mL substrate-chromogen solution, incubate 20 min at room temperature.

6. Remove dipsticks and add 0.25 mL stop solution.

7. Measure A450.
Fig. 10.5
Salmonella direct-labeled probe hybridization assay protocol.

TABLE 10.1 Salmonella Direct-Labeled Probe Assay Inclusivity

Number of Number positive at approx. 107 Number positive at approx. 106
Organism strains CFU/assay CFU/assay
S. enterica subsp. enterica 65 65 58
S. enterica subsp. salamae 14 14 13
S. enterica subsp. arizonae 5 5 5
/diarizonae
S. enterica subsp. houtenae 5 5 3
S. enterica subsp. indica 8 8 7
S. bongoria 0
Totals 97 97 86
Strains were grown in trypticase soy broth with yeast extract for 1618 hr at 37°C to final titers of
approximately 109 CFU/mL. Cultures were diluted appropriately in phosphate-buffered saline. Testing
was performed according to the instructions provided with the probe assay test kit (GENE-TRAK
Systems, 1995). Salmonella nomenclature is according to Popoff et al. (1995).
aNot detectable with probe set used in direct-labeled probe assay.


TABLE 10.2 Salmonella Direct-Labeled Probe Assay Specificity
Organism
Aeromonas sobria
Citrobacter amalonaticus
Citrobacter diversus
Citrobacter freundii
Citrobacter spp.
Edwardsiella hoshinae
Enterobacter agglomerans
Enterbacter cloacae
Enterobacter sakazakii
Enterobacter taylorii
Enterobacter spp.
Erwinia herbicola
Escherichia coli
Escherichia hermannii
Hafnia alvei
Klebsiella oxytoca
Klebsiella pneumoniae
Morganella morganii
Proteus vulgaris
Providencia alcalifaciens
Providencia stuartii
Serratia liquefaciens
Serratia marcescens
Serratia rubidae
Shigella boydiz
Yersinia enterocolitica
Total
Strains were grown in trypticase soy broth with yeast extract for 1618 hr at 37°C to final titers of approximately
109 CFU/mL. Testing was performed according to the instructions provided with the probe assay test kit
(GENE-TRAK Systems, 1995).
Page 221
Number of strains Number positive at approx. 108 CFU/assay
1 0
3 0
2 0
3 0
2 0
1 0
1 0
2 0
1 0
1 0
1 0
2 0
28 0
5 1
1 0
1 0
1 0
1 0
2 0
1 0
1 0
1 0
1 0
1 0
1 0
1 0
66 1
 Page 222

probe specific for the S. bongori group has been eliminated from the probe cocktail due to adverse effects on
specificity in the modified hybridization matrix. Therefore, the DLP assay is not inclusive for the 19 rare serovars
belonging to the separate species S. bongori, but would be predicted to be inclusive for the approximately 2400
serovars belonging to the species S. enterica (see Table 10.1) [the reader is referred to Popoff et al. (1995) for a
discussion of the current status of taxonomy and nomenclature of the salmonellae]. Results of specificity testing
showed a complete lack of reactivity with the nonsalmonellae tested with the exception of a single strain of
Escherichia hermannii (Table 10.2).

A large study with inoculated food samples was performed and the results presented (Durbin et al., 1995) (Table 10.3).
In this study, performance of the Salmonella DLP assay was compared to that of the original colorimetric GENE-
TRAK assay and also to that of the FDA culture method. Sensitivity of the DLP assay was nearly identical to that of
the original assay: 97.7 and 98.0%, respectively. Sensitivity of the culture method was 99.7%. There was 98.6%
agreement between the DLP assay and the culture method. Particularly noteworthy is the fact that the DLP assay
produced no false-positive results out of more than 300 Salmonella-negative samples tested. Considering the wide
variety of foods tested and therefore the broad range of indigenous microflora encountered in these samples, this result
is perhaps more significant than that of pure culture testing in establishing the extremely high degree of specificity of
the DLP assay. The Salmonella DLP assay is currently undergoing external validation under the program of the AOAC
Research Institute. Prototype Listeria spp. and Listeria monocytogenes assays in the DLP format have also been
developed and are undergoing internal evaluation (GENE-TRAK Systems, unpublished results).
In closing, let us consider briefly the prospect of future developments in probe-based methods for applications in food
microbiology. Efforts at further improvement of probe methods (including those targeting rRNA) are aimed at making
these methods faster, more sensitive, and more convenient for the user. Manual formats are evolving to automated
alternatives, and nucleic acid amplification techniques are being coupled with detection chemistries and introduced to
the food microbiologist as test systems. The major challenge will be making these amplified methods user friendly and,
at the same time, economical enough for routine use in quality control applications. Opportunities awaiting a test
system with these attributes abound, including more rapid detection of pathogens such as Salmonella and Listeria,
detection of foodborne viruses, and rapid detection of organisms such as Campylobacter spp. where enrichment of the
organism to detectable titers, and not identification, is the major challenge for the microbiologist.

 Page 223

TABLE 10.3 Salmonella Direct-Labeled Probe Assay: Comparative Study with Inoculated Foods
Number of Total samples Samples positive by positive direct-labeled probe Samples positive by standard probe Samples by FDA culture
Category samples positivea assayb assayc methodd
Milk, eggs, 150 103 103 103 103
cheese
Nuts, fruits 150 82 81 81 82
Flours, pasta 200 152 148 149 152
Confectionary 100 71 66 66 71
Meats 300 231 225 226 229
Other 150 103 102 102 103
Totals 1050 742 725 727 740
aNumber of samples positive by one or more methods.
bNumber of samples positive by the direct-labeled probe assay, including confirmation. The method was performed according to the instructions provided with the probe assay
test kit (GENE-TRAK Systems, 1995).
cNumber of samples positive by the standard probe assay, including confirmation. The method was performed according to the instructions provided with the probe assay test
kit (GENE-TRAK Systems, 1994).
dNumber of samples positive by the FDA culture method (FDA, 1995b).

 Page 224

Acknowledgments

Thanks are extended to all of my present and former colleagues at GENE-TRAK Systems. Their efforts over more than
10 years are the basis for most of the material presented in this chapter. Special thanks are given to Jeff Klinger for his
unwavering support and to Dave Lane and Will Weisburg for their patient tutoring on the subject of probe design.

References
Arnold, L.J., Jr., Hammond, P.W., Wiese, W.A., and Nelson, N.C. 1989. Assay formats involving acridinium-ester-
labeled probes. Clin. Chem. 35: 15881594.

Bottari, D.A., Emmett, C.D., Nichols, C.E., Whippie, K.D., Rodriguez, D., Durbin, G.W., Keough, K.M., Groody, E.P.,
Mozola, M.A., and Reynolds, G.N. 1995. Comparative study of a colorimetric DNA hybridization method and
conventional culture procedures for the detection of Listeria spp. in foods. J. Food Prot. 58: 10831090.
Brenner, D.J. 1984. Facultatively anaerobic gram-negative rods. Family 1. Enterobacteriaceae. In: Bergey's Manual of
Systematic Bacteriology, Vol. 1, N.R. Krieg and J.G. Holt (Ed.), p. 408420. Williams & Wilkins, Baltimore.
Chan, S., Pitman, T., Shah, J., King, W., Lane, D., and Lawrie, J. 1988. Nonradioactive DNA assay for detection and
identification of foodborne Yersinia. Abstracts of the General Meeting of the American Society for Microbiology, p.
280.

Chan, S.W., Wilson, S., Hsu, H.-Y., King, W., Halbert, D.N., and Klinger, J.D. 1989. Model non-isotopic hybridization
systems for detection of foodborne bacteria: Preliminary results and future prospects. In: Biotechnology and Food
Quality, S.-D. Kung, D.D. Bills, and R. Quatrano (Ed.), p. 219237. Butterworth Publishers, Boston.

Chan, S.W., Wilson, S.G., Vera-Garcia, M., Whippie, K., Ottaviani, M., Whilby, A., Shah, J., Johnson, A., Mozola,
M.A., and Halbert, D.N. 1990. Comparative study of colorimetric DNA hybridization method and conventional culture
procedure for detection of Salmonella in foods. J. Assoc. Off. Anal. Chem. 73: 419424.

Curiale, M.S. and Lewus, C. 1994. Detection of Listeria monocytogenes in samples containing Listeria innocua. J.
Food Prot. 57: 10481051.

Curiale, M.S., Klatt, M.J., and Mozola, M.A. 1990a. Colorimetric deoxy-

 Page 225

ribonucleic acid hybridization assay for rapid screening of Salmonella in foods: Collaborative study. J. Assoc. Off.
Anal. Chem. 73: 248256.

Curiale, M.S., McIver, D., Weathersby, S., and Planer, C. 1990b. Detection of salmonellae and other
Enterobacteriaceae by commercial deoxyribonucleic acid hybridization and enzyme immunoassay kits. J. Food Prot.
53: 10371046.

Curiale, M.S., Sons, T., Fanning, L., Lepper, W., Mclver, D., Garramone, S., and Mozola, M. 1994. Deoxyribonucleic
acid hybridization method for the detection of Listeria in dairy products, seafoods, and meats: Collaborative study. J.
AOAC Int. 77: 602617.
D'Aoust, J.-Y., Sewell, A.M., Greco, P., Mozola, M.A., and Colvin, R.E. 1995. Performance assessment of the GENE-
TRAK® colorimetric probe assay for the detection of foodborne Salmonella spp. J. Food Prot. 58: 10691076.

Doyle, M.P. and Schoeni, J.L. 1984. Survival and growth characteristics of Escherichia coli associated with
hemorrhagic colitis. Appl. Environ. Microbiol. 48: 855856.

Durbin, G.W., Keough, K.M., and Reynolds, G.N. 1994. Comparative study for detection of Listeria monocytogenes in
foods by a colorimetric DNA method and conventional culture methods. Abstracts of the Annual Meeting of the
International Association of Milk, Food and Environmental Sanitarians, p. 37.

Durbin, G.W. and Mozola, M.A. 1995. Validation of a direct-labeled probe nucleic acid hybridization assay for
detection of Salmonella spp. in foods. Abstracts of the AOAC International Annual Meeting, p. 70.

Ewing, W.H. 1986. Edwards and Ewing's Identification of Enterobacteriaceae, 4th ed. Elsevier Science Publishing
Co., New York.

FDA. 1995a. Listeria monocytogenes. In: Bacteriological Analytical Manual, 8th ed. AOAC International,
Gaithersburg, Md.

FDA, 1995b. Salmonella. In: Bacteriological Analytical Manual, 8th ed. AOAC International, Gaithersburg, Md.

FDA, 1995c. Escherichia coli and the coliform bacteria. In: Bacteriological Analytical Manual, 8th ed. AOAC
International, Gaithersburg, Md.

FDA, 1995d. Identification of foodborne bacterial pathogens by gene probes. In: Bacteriological Analytical Manual,
8th ed. AOAC International, Gaithersburg, Md.
Foster, K., Garramone, S., Ferraro, K., and Groody, E.P. 1992. Modified colorimetric DNA hybridization method and
conventional culture method for detection of Salmonella in foods: Comparison of methods. J. AOAC Intl. 75: 685692.

 Page 226

GENE-TRAK Systems. 1994. Test kit instructions for the GENE-TRAK® Salmonella assay.

GENE-TRAK Systems. 1995. Test kit instructions for the GENE-TRAK® Salmonella DLP assay.

Hill, W.E. and Keasler, S.P. 1991. Identification of foodborne pathogens by nucleic acid hybridization. In: Rapid
Methods and Automation in Microbiology and Immunology, A. Vaheri, R.C. Tilton, and A. Balows (Ed.), p. 470480.
Springer-Verlag, Berlin.

Hsu, H.-Y., Chan, S.W., Sobell, D.I., Halbert, D.N., and Groody, E.P. 1991. A colorimetric DNA hybridization method
for the detection of Escherichia coli in foods. J. Food Prot. 54: 249255.
Johnson, J.L. and Lattuada, C.P. 1993. Comparison of nucleic acid hybridization assays and biochemical
characterization tests for the confirmation of Listeria monocytogenes. J. Food Prot. 56: 834840.

Keough, K.M., Durbin, G.W., Holland, J.A., and Reynolds, G.N. 1994. Detection of Campylobacter sp. in poultry by a
colorimetric DNA hybridization assay. Abstracts of the Annual Meeting of the Southern Poultry Science Society, p. 48.

King, W., Raposa, S., Warshaw, J., Johnson, A., Halbert, D., and Klinger, J.D. 1989. A new colorimetric hybridization
assay for Listeria in foods. Int. J. Food Microbiol. 8: 225232.

King, W., Raposa, S., Warshaw, J.E., Johnson, A., Halbert, D., and Klinger, J.D. 1990. A colorimetric assay for Listeria
using nucleic acid probes. In: Foodborne Listeriosis, A.J. Miller, J.L. Smith, and G.A. Somkuti (Ed.), p. 117124.
Elsevier Science Publishers, Amsterdam.

Lane, D.L. and Collins, M.L. 1991. Current methods for detection of DNA/ribosomal RNA hybrids. In: Rapid Methods
and Automation in Microbiology and Immunology, A. Vaheri, R.C. Tilton, and A. Balows (Ed.), p. 5475. Springer-
Verlag, Berlin.

Lane, D.J., Pace, B., Olsen, G.J., Stahl, D.A., Sogin, M.L., and Pace, N.R. 1985. Rapid determination of 16S rRNA
sequences for phylogenetic analyses. Proc. Natl. Acad. Sci. USA 82: 69556959.

Moore, N., Trnovsky, J., Durbin, G., and Mozola, M. 1995. A modified DNA hybridization assay for the detection of
Salmonella in foods using horse-radish peroxidase-labeled probes. Abstracts of the General Meeting of the American
Society for Microbiology, p. 390.

Mozola, M., Halbert, D., Chan, S., Hsu, H.-Y., Johnson, A., King, W., Wilson, S., Betts, R.P., Bankes, P., and Banks,
J.G. 1991. Detection of foodborne bacterial pathogens by a colorimetric DNA hybridization method. In: Genetic
Manipulation: Techniques and Applications, J.M. Grange, A. Fox, and

 Page 227

N.L. Morgan (Ed.), p. 203216. Blackwell Scientific Publications, Oxford.

Niederhauser, C., Höfelein, C., Lüthy, J., Kaufmann, U., Bühler, H.-P., and Candrian, U. 1993. Comparison of ''Gen-
Probe" DNA probe and PCR for detection of Listeria monocytogenes in naturally contaminated soft cheese and semi-
soft cheese. Res. Microbiol. 144: 4754.

Ninet, B., Bannerman, E., and Bille, J. 1992. Assessment of the Accuprobe Listeria monocytogenes culture
identification reagent kit for rapid colony confirmation and its application in various enrichment broths. Appl. Environ.
Microbiol. 58: 40554059.
Okwumabua, O., Swaminathan, B., Edmonds, P., Wenger, J., Hogan, J., and Alden, M. 1992. Evaluation of a
chemiluminescent DNA probe assay for the rapid confirmation of Listeria monocytogenes. Res. Microbiol. 143:
183189.

Partis, L., Newton, K., Murby, J., and Wells, R.J. 1994. Inhibitory effects of enrichment media on the Accuprobe test
for Listeria monocytogenes. Appl. Environ. Microbiol. 60: 16931694.

Petran, R.L. and Swanson, K.M.J. 1993. Simultaneous growth of Listeria monocytogenes and Listeria innocua. J. Food
Prot. 56: 616618.

Popoff, M.Y., Bockemühl, J., and Hickman-Brenner, F.W. 1995. Supplement 1994 (no. 38) to the Kauffmann-White
scheme. Res. Microbiol. 146: 799803.
Ranki, M., Palva, A., Virtanen, M., Laaksonen, M., and Soderlund, H. 1983. Sandwich hybridization as a convenient
method for the detection of nucleic acids in crude samples. Gene 21: 7785.

Ransom, G.M., Dreesen, D.W., Rose, B.E., and Lattuada, C.P. 1994. Assessment of three nucleic acid hybridization
systems for detection of Campylobacter spp. in poultry products. J. Food Prot. 57: 703709.

Reynolds, G.N., Durbin, G.W., Shah, A., and Tubert, T.A. 1991. Evaluation of an enrichment for detection of
Staphylococrus aureus in raw meats, poultry, and seafood by a DNA probe assay. Abstracts of the Annual Meeting of
the Institute of Food Technologists, p. 64.

Stanick, J.A., Weisburg, W.G., Durbin, G.W., Keough, K.M., and Reynolds, G.N. 1993. Performance of a colorimetric
DNA hybridization assay for specific detection of Listeria monocytogenes in foods. Abstracts of the Annual meeting of
the Institute of Food Technologists, p. 161.

Stern, N.J. and Mozola, M.A. 1992. Methods for selective enrichment of Campylobacter spp. from poultry for use in
conjunction with DNA hybridization. J. Food Prot. 55: 767770.
Tenover, F.C. 1988. Diagnostic deoxyribonucleic acid probes for infectious diseases. Clin. Microbiol. Rev. 1: 82101.

 Page 228

Tenover, F.C. 1993. DNA hybridization techniques and their application to the diagnosis of infectious disease. Infect.
Dis. Clin. North Am. 7: 171181.

Wilson, S.G., Chan, S., DeRoo, M., Vera-Garcia, M., Johnson, A., Lane, D., and Halbert, D.N. 1990. Development of a
colorimetric, second generation nucleic acid hybridization method for detection of Salmonella in foods and comparison
with conventional culture procedure. J. Food Sci. 55: 13941398.

Woese, C.R. 1987. Bacterial evolution. Microbiol. Rev. 51: 221271.

Wolcott, M.J. 1991. DNA-based methods for the detection of foodborne pathogens. J. Food Prot. 54: 387401.

 Page 229

11
Gene Sequence Databases and Food Microbiology
Steven M. Gendel
National Center for Food Safety and Technology
Food and Drug Administration
Summit-Argo, Illinois

Introduction: Why Are Sequence Databases Important?

Food microbiology, like all of the biological sciences, is feeling the impact of advances in DNA-sequencing
technology. The ability to carry out DNA sequencing and sequence analysis is becoming critical to the development of
advanced technologies for the rapid, sensitive, and accurate analysis of food-related microbes. The ability to find and
use genetic sequence information is becoming as important for some aspects of food microbiology as the ability to
carry out biochemical analysis.
The primary tools for locating and using genetic information are sequence databases. These databases have evolved
from simple mechanisms for information storage and exchange to data resources that can be analyzed for insights into
biological processes and relationships. The pharmaceutical industry actively uses these sequence databases and other
bioinformatic resources for identifying product targets. Similar approaches are beginning to be applied to food
microbiology. The first stage of this process

 Page 230

can be seen in the number of genomic assays that have been developed in the last few years for the detection and
characterization of foodborne pathogens (Hill, 1996).

Before looking at genetic sequence databases in detail, it is important to point out that sequence information is just one
aspect of the vast array of bioinformatics. Biochemical pathway databases, enzyme activity databases, protein structure
databases, culture collection databases, and bibliographic databases all exist, or are under development, and are
becoming available online. One major focus of current bioinformatic research is the development of technology for
integrating these different databases and types of data, so that an individual researcher can access any type of
information and easily move to related topics that help put this information in context (Masys, 1989).

All of the databases discussed in this paper are available for public access. Proprietary databases have also been
developed, but none are of major importance for food microbiology or food safety.

Sequence Databases: What's out there?

There are two classes of sequence databases: large, general purpose databases that serve as primary data depositories
and smaller specialized databases that contain sets of related sequences. Because journals restrict the amount of
sequence data that can be printed in a research article, deposit in a large sequence database has become a primary
mechanism for "publishing" sequence data. Many of the smaller databases are more selective, with the database owner
serving as a curator to ensure the quality of the data and annotation and to ensure consistency in the nomenclature used.

Large Databases

The three largest DNA sequence databases are GenBank (National Center for Biotechnology Information), the
European Molecular Biology Laboratory (EMBL) Nucleotide Sequence Database, and the DNA Data Bank of Japan
(Benson et al., 1994; Emmert et al., 1994). Originally these were independent databases, but it became clear that close
coordination was necessary to prevent wasteful redundancies. Therefore, the agencies involved formed the
International Nucleotide Database Collaboration and instituted a system of sequence exchange that ensures that the
three databases are essentially the same.
These databases are huge and growing rapidly. The latest release of GenBank (Release 94, April 1996), contains almost
5 × 108 nucleotides (Fig. 11.1). The bulk of this sequence information has been generated by the

 Page 231

Fig. 11.1
Growth in the number of nucleotide bases in GenBank
over time.
(From GenBank, Release 94 documentation, April 1996.)

various components of the human genome project and is not relevant to food microbiology. GenBank data are divided
into several sublibraries, including one containing bacterial sequences (with about 5 × 106 nucleotides in over 26,000
sequences in Release 94), that make it possible to search only the most relevant sequences.

Because GenBank is growing so rapidly, it is important to consider the potential effect of new data on search or
analysis results obtained by using the database. For example, a hybridization probe sequence that matches only one
sequence today may match a newly deposited gene family tomorrow. To help scientists cope with this flood of new
data, GenBank provides daily updates containing accession information on newly deposited sequences via the Bionet
news group bionet.molbio.genbank.updates. A similar service is available from EMBL. In addition, GenBank
maintains new sequences in a searchable database between major releases.

The two biggest problems encountered in using GenBank are the presence of redundant sequences and inconsistencies
or incompleteness in sequence annotation. The presence of redundant sequences can affect statistical analysis for some
database search and alignment algorithms. Non-redundant subsets of GenBank have been developed for us in situations
where this is important. In most cases, GenBank relies on submitting authors to provide annotation, such as the
identification of coding regions or other important sequence features. In many cases this information is complete and
accurate, but errors and omissions do occur. Figure 11.2 is an

 Page 232

Fig. 11.2
Example of the annotation for a GenBank accession. The dots indicate
material that was removed for clarity.

example of the annotation for a GenBank accession that provides both the location and translation of the protein
reading frame.

There are also three large protein sequence databases: the GenPept database which is derived solely from those
GenBank accessions that contain appropriate information in the annotation, the SwissProt database available through
EMBL, and the Protein Information Resource (PIR) from the National Biomedical Research Foundation (Bairoch and
Boeckmann, 1994; George et al., 1994). Most of the sequences in the SwissProt and PIR databases are translations of
nucleic acid sequences that appear in Gen-

 Page 233

Bank, but both databases also include sequences that were obtained by direct amino acid sequencing. The annotation of
the SwissProt database is particularly complete and is very useful (Fig. 11.3). SwissProt annotation can provide
valuable information about the activity of a protein and relationships to functional families. The amount of effort
required to produce these databases means that neither can be updated as often as the GenPept database.

Smaller Databases

The smaller databases serve as important complements to these large databases for several reasons. First, production of
many of these databases involves effort by an individual researcher (or curator) or research team to ensure
completeness, consistency, and accuracy in both the data and the annotation. Second, these databases can contain
information taken from the primary literature that does not occur in the larger databases, such as the sequences of
hybridization probes or PCR primers or links to genetic maps. Third, the focused nature of these databases makes it
possible to organize the data in a manner useful to the database user rather than the database administrator. Finally, the
relatively small size of these databases can make it possible to manipulate the entire database on a laboratory computer.

There are three types of smaller databases: those that focus on a single class of sequence, those that focus on a single
organism, and those that focus on a single application or research field. An excellent example of the first type is the
Ribosomal Database Project (RDP) from the University of Illinois (Maidak et al., 1994). This database contains
ribosomal RNA sequences from a large number of organisms, both eukaryotic and prokaryotic. Currently there are
about 3000 sequences in the prokaryotic small subunit (16S) database. Because ribosomal sequences are widely used
for the identification of food-related microbes, both pathogens and production strains, this sequence information is of
great relevance to food microbiology. All entries in the RDP database are given as the actual ribosomal RNA sequence.
Therefore, they all have the same orientation (i.e., are the same strand) and are not encumbered with variable lengths of
flanking sequence. In addition, each sequence is fit to a standard model for the structure of ribosomal RNA so that
significant gaps, duplications, or errors in the data can be identified. This simplifies comparison of sequences within
the database, between the database and sequences obtained from other sources, and between the database and short
sequences (such as hybridization probes or PCR primers) derived from defined regions of the ribosomal gene.

 Page 234

Fig. 11.3
Example of the annotation for a SwissProt accession. The dots indicate material
that was removed for clarity. (A) Top of annotation, showing extensive comment
section. (B) Bottom of annotation, showing functional information and indications
of sequence conflicts.

 Page 235

There are several examples of organism-specific databases, the best known of which is the human genome database.
The only currently available organism databases relevant to food microbiology are two Escherichia coli databases (the
E. coli Data Collection and the E. coli Genome Center database) and a Bacillus subtilis sequence database.

There are a number of other databases for nonmicrobes of interest to food scientists, including yeast, soybeans, rice,
maize, plant pathogenic fungi, cattle, and pigs. Many of these databases have been built using a software system (called
the ACEDB system), designed by biologists, that allows linkages between sequences, gene maps, and biochemical
information.

The most relevant example of the third type of database is the Biotechnology Information for Food Safety (BIFS)
database at the National Center for Food Safety and Technology (Gendel, in preparation). The BIFS database contains
an extensive listing of hybridization probes and PCR primers for the detection of food-related microbes (both
pathogens and production strains), as well as a literature database on the use of genomic technology for identification
and characterization of food-related microbes. Each probe or primer sequence has been tested for apparent specificity
by sequence matching by using one of the larger databases, and the result of this validation assay is reported in the
BIFS database. In addition, the location of many of these probe or primer sequences on the target gene has been
determined. This database contains sequence information that is not available in any other database and provides
comparative data not available in the original literature.

Databases On-Line: How do I access them?

The existence of any database would be of little value without tools for accessing the data. Therefore, virtually all
databases relevant to food microbiology can now be accessed on-line or will be accessible in the near future. Although
several mechanisms were used for on-line access in the past, in the future virtually all database access will be achieved
either through e-mail or via the World Wide Web (WWW) (Harper, 1994).

Finding the address of a database is often more difficult than using it! Users with access to the WWW can use several
directories or virtual library pages that contain pointers to the databases. The locations (Universal Resource Locators,
or URLs) of several relevant directory pages are listed in Table 11.1. Figure 11.4 shows part of one of these virtual
library pages. In addition, the individual databases are becoming increasingly cross-linked so that a user of one can
access data in another. There are, however, no comparable e-mail directories or linkages.

E-mail access to sequence data and search tools is the most straightfor-

 Page 236

TABLE 11.1 Directories and Virtual Library Pages

Name Address
Bio-gophers gopher://gopher.gdb.org/11/biopages
Bioweb http://www.bioweb.com/
Biotech Web http://www.bioweb.com/
BioWurld http://www.ebi.ac.uk/htbin/bwurld.pl
Genome Gophers gopher://genome-gopher.stanford.edu:70/1/topic/genome_db
GenomeNet http://www.genome.ad.jp/
Molecular Biology Computational Resources http://mbcr.bcm.tmc.edu/databases.html
Pedros Biopages http://www.public/iasatate.edu/~pedro/rt _1.html
WWW Virtual LibraryBiomolecules http://golgi.harvard.edu/sequences.html
WWW Virtual LibraryBiosciences http://golgi.harvard.edu/biopages.html
WWW Virtual LibraryBiotechnology http://www.cato.com/interweb/cato/biotech/
WWW Virtual LibraryMicrobiology http://golgi/harvard.edu/biopages/micro.html

ward approach for many users (Henikoff, 1993). An increasing number of food microbiologists use e-mail and are
familiar with the basic commands for composing, sending, and receiving messages. Simple e-mail scripts can be used
to retrieve data or carry out sophisticated searches from databases such as GenBank or the RDP. Table 11.2 contains the
e-mail addresses of a number of relevant databases. The first step in using any new database should be to send for the
help files, which will contain detailed instructions for composing and addressing queries. All of the databases listed in
Table 11.2 will send help files in response to an e-mail message containing the command "help" in the body of the
message (leave the subject line blank).
Access to sequence databases through the WWW is increasingly common and provides the most user-friendly
interface. Use of the hypertext markup language (HTML) has greatly simplified the process of linking disperate
databases and allows users to access multiple data views without needing to know the underlying structure of the
databases. Table 3 lists the URLs of several databases relevant to food microbiologists. Most of these can also be

 Page 237

Fig. 11.4
Example of a small part of a virtual library page on the WWW containing pointers
to a variety of other resources.

 Page 238

TABLE 11.2 E-Mail Addresses for Sequence Databases

Database Address
Darwina cbrg@inf.ethz.ch
EMBL quick@ebi.ac.uk
EMBL/FASTA fasta@ebi.ac.uk
GenBank/Retrievea retrieve@ncbi.nlm.nih.gov
GenBank/BLASTa blast@ncbi.nlm.nih.gov
Geniusa netserv@genius.embnet.dkfz-heidelberg.de
Netserva netserv@ebi.ac.uk
PIR/dFLASH dflash@watson.ibm.com
RDP server@rdp.life.uiuc.edu
SwissProt blitz@ebi.ac.uk
aThese servers provide access to multiple databases.

accessed from the directory pages listed in Table 11.1. Figures 11.5 and 11.6 show parts of the home pages of the
National Center for Biotechnology Information (NCBI) and the European Bioinformatics Institute (EBI).

Perhaps the most sophisticated interfaces available on the WWW are the GenBank Entrez (Fig. 11.7) system and the
EMBL Sequence Retrieval
TABLE 11.3 WWW Addresses for Major Sequence Databasesa
Database Address
EMBL/Nucleic Acids http://www.ebi.ac.uk/ebi_docs/embl_db/ebi/toopembl.html
EMBL SRS http://www.ebi.ac.uk/srs/srsc
GenBank http://www.ncbi.nlm.nih.gov/Web/Search/index.html
GenBank Entrez http://www3.ncbi.nlm.nih.gov/Entrez/index.html
PIR http://www.gdb.org/Dan/proteins/pir.html
PUMA http://www.mcs.anl.gov/home/compbio/PUMA/Production/puma.html
RDP http://rdp.life.uiuc.edu/
rRNA Server http://rrna.uia.ac.be/
SwissProt http://www.ebi.ac.uk/ebi_docs/swissprot_db/swisshome.html
aCapitalization, where indicated, is important.

 Page 239

Fig. 11.5
Part of the home page of the National Center for Biotechnology
Information showing the database and search services available on-line
at this site.

 Page 240

Fig. 11.6
Part of the home page of the European Bioinformatics Institute showing the
database and search services available on-line at this site.

 Page 241

Fig. 11.7
Example of a WWW page providing access to GenBank searches at the NCBI.

System (SRS) (Fig. 11.8). Both allow users to search the respective databases by using simple query forms that can
provide on-line help if necessary. Some query fields can support complex Boolean operations. In both cases, the results
of an initial search can be used as links to other data resources. For example, the Entrez system provides links from
GenBank to the molecular biology subset of the bibliographic database Medline. In addition to the direct citation for a
sequence, Medline also identifies related publications (called neighbors), which can be used as Entrez links to other
sequences back in GenBank. The EMBL SRS system provides links from the gene sequence database to over 25 other
databases.

Although most direct access to these databases will take place over the WWW, using the Hypertext Transfer Protocol
(HTTP), there are two important applications that continue to rely on the File Transfer Protocol (FTP). First, several
sequence databases (including GenBank) can be obtained by FTP for use on local computer systems. This is most
useful for those with

 Page 242

Fig. 11.8
Example of a WWW page providing access to searches
at the EBI.

access to a mainframe or workstation system that is running one of the commercial sequence analysis packages, such as
the ones from Genetics Computer Group (GCG) or Intelligenetics. Instructions for FTP access are usually found in the
help documentation for a database or on the WWW site for that database. Second, FTP is widely used to download
sequence analysis software from various archives. Table 11.4 lists several software archives that contain software for
DNA and protein sequence analysis and manipulation.

Searching Databases: How can I use this information?

The primary food safety-related application of sequence databases is in the development of molecular assays for the
identification and detection of foodborne pathogens. Sequence information can be used to identify a

 Page 243

TABLE 11.4 Addresses of Software Archives

Name Address
EMBL Biocatalog http://www.ebi.ac.uk/biocat/biocat.html
Indiana Univ.(IUBIO) Archive http://iubio.bio.indiana.edu/
Johns Hopkins http://www.gdb/org/Dan/softsearch/softsearch.html
NCBI ftp://ncbi.nlm.nih.gov/
Univ. of Houston ftp://ftp.bchs.uh.edu/
Weizmann Inst. Bioinformatics Unit http://bioinformatics.weizmann.ac.il/

target gene, to help develop unique probes or primers, or to help validate the specificity of probes or primers obtained
from other sources (Hill, 1996; Bsat et al. 1994; Swaminathan and Feng, 1994).

Most molecular assays for foodborne pathogens target specific genes that are chosen either because they are
phylogenetically informative (such as the ribosomal sequences), because they reflect some unique biochemical
characteristic of the target organism (such as ability to use certain carbon sources), or because they are relevant to
pathogenesis (such as toxin genes). Simple keyword searches of the various databases can be used to determine which
genes from a species of interest have been sequenced, whether sequences have been obtained from related species or
strains, and whether multiple alleles of the target genes exist in the population.

Often, when multiple sequences have been obtained from related species, sequence alignment programs are used to
determine which parts of the sequence are conserved and which parts vary between species (and therefore represent
good targets for specific identification) (States and Boguski, 1991). These programs function extremely well to align
sequences that are closely related evolutionarily, but care must be used when aligning sequences that are distantly
related or that may have undergone a significant number of deletion/insertion events. Most alignment programs are
designed to optimize alignments across the entire length(s) of the sequences involved and may miss short regions of
exact similarity if these regions are located at different ends of the sequences.

Similar programs can also be used to search databases to identify related sequences (Coulson, 1994; Doolittle, 1994).
The two most popular programs for this type of searching are BLAST and FastA. BLAST searches of GenBank can be
carried out by e-mail or through the National Institutes of

 Page 244

Health Entrez system. FastA searches can be conducted through several of the sites listed in Table 11.1.

When the best region(s) of the target genes have been identified, a number of different approaches can be used to
derive specific probes or PCR primers. Probe and primer design is far from an exact science and will not be discussed
here, except to note that several software packages exist that claim to assist in the design process.

Once appropriate primers or probes have been designed or obtained from other databases or from literature sources, the
apparent specificity of these sequences can be verified using simple pattern matching algorithms. The RDP has
implemented a particularly useful program (Check-Probe) that identifies exact matches and allows identification of
sequences with defined numbers of mismatches (Maidak et al., 1994). This is useful because the location of these
sequence mismatches are critical in determining whether the mismatched sequences will hybridize and potentially
produce false-positive signals.

Precautions: What are the problems?

There are three major precautions that should be considered when using sequence databases, particularly in a field such
as food safety. First, because sequence databases grow and change, the validity of conclusions drawn using information
in earlier versions of a database may change. Therefore, it is important to periodically repeat database searches or retest
probe or primer sequences. Second, the data in these databases have a significant number of errors (Doolittle, 1994;
Domenighini et al., 1995). These errors are introduced in several ways including during PCR amplification, in reading
sequencing gels, and in data transmission. For example, Domenighini et al. (1995) examined multiple database
accessions for E. coli and Vibrio cholerae heat-labile enterotoxins and found errors in more than 20% of these
sequences. The overall error rate in the sequence databases is not known, but may be less in the secondary or smaller
databases that contain data prescreened by the maintainer/curator. Therefore, multiple databases should be used
whenever possible, and the possible effect of data exchange between them should be considered. Third, the on-line
tools for database access are evolving rapidly and should be considered experimental and temporary. Sites and access
tools are created and removed constantly as the technology advances and people move from one institution to another.
Even the structure of the underlying databases is evolving. Therefore, multiple access methods should be considered to
ensure continuity if one of these resources should become unavailable or unsuitable in the future.

 Page 245

Fig. 11.9
Example of a graphical representation of metobolism in the PUMA
database. Clicking on an enzyme name links the user to information
about that enzyme in biochemical and sequence databases.

 Page 246

TABLE 11.5 Other Addresses of Interest

Site name Address
American Society for Microbiology http://www.asmusa.org/
American Type Culture Collection http://www.atcc.org/
Canadian Society of Microbiologist http://gpu.srv.ualberta.ca/~mmid/csm.html
CDC Morbidity and Mortality Weekly Reporter http://www.cdc.gov/epo/mmwr/mmwr.html
CDC Emerging Infectious Diseases http://www.cdc.gov/ncidod/EID/eid.htm
Centers for Disease Control http://www.cdc.gov/
FDA Center for Food Safety and Applied Nutrition http://vm.cfsan.fda.gov/
Food and Drug Administration http://www.fda.gov/
Institute of Food Technologists http://www.ift.org/
Microbiology BBS http://home.unicomp.net/microbbs/
Microbiology Jumpstation http://www.ifrn.bbsrc.ac.uk/gm/lab/docs/micro.html
National Food Safety Database http://www.agen.ufl.edu/~foodsaf/foodsaf.html
Society for General Microbiology http://www.mcc.ac.uk/pwmirror/pw9/sgm/pharmweb93.html
USDA/FDA Foodborne Illness Education Information Center http://www.nalusda.gov/fnic/foodborne/foodborn.htm
UsenetFood Science sci.bio.food-science
UsenetMicrobiology sci.bio.microbiology
World Data Centre for Microorganisms http://www.wdcm.riken.go.jp/

Future Directions: Where are we going?

The structure and content of genetic sequence databases are evolving rapidly. This evolution is driven by the
continuing increase in the rate at which sequence data are being produced, the development of new modes of access,
and the increasing sophistication of the users. These forces will lead to two major developments in the next several
years. First, the number and

 Page 247

value of the secondary databases will increase. These data resources will continue to provide information that does not
fit into the larger databases.

Second, more sophisticated mechanisms will develop for integrating these data resources. The Entrez and SRS systems
described above represent initial steps in this process. The SILS is a particularly interesting example because it
provides links from a gene sequence database to over 25 other databases, including structural, functional, mapping, and
literature databases. Another example of an integrated database system is the PUMA database at Argonne National
Laboratory. This database provides linkages from metabolic pathways to an enzyme database and to databases of
protein and gene sequences (Fig. 11.9). Eventually, this type of integration will make it possible for the individual
scientist to view all of these data resources as a single database. Tools similar to the independent agents and applets
being developed for accessing data on internet home pages will be developed for mining molecular databases.

Finally, because sequence databases are only one part of the bioinformatic tools available for food safety assurance,
Table 11.5 lists a number of other sties of interest to food microbiologists.

References

Bairoch, A. and Boeckmann, B. 1994. The SWISS-PROT Protein Sequence Data Bank: Current Status. Nucleic Acids
Res. 22: 35783580.

Benson, D.A., Boguski, M., Lipman, D.J., and Ostell, J. 1994. GenBank. Nucleic Acids Res. 22: 34413444.

Bsat, N., Wiedmann, M., Czajka, J., Barany, F., Piani, M., and Batt, C. 1994. Food safety applications of nucleic acid-
based assays. Food Technol. 48(6): 142145.

Coulson, A. 1994. High performance searching of biosequence databases. Trends Biotechnol. 12: 7680.
Domenighini, M., Pizza, M., Jobling, M.G., Holmes, R.K., and Rappuoli, R. 1995. Identification of errors among
database sequence entries and comparison of correct amino acid sequences for the heat-labile enterotoxins of
Escherichia coli and Vibrio cholerae. Mol. Microbiol. 15: 11651167.

Doolittle, R.F. 1994. Protein sequence comparisons: Searching databases and aligning sequences. Curr. Opin.
Biotechnol. 5: 2428.

Emmert, D.B., Stoehr, P.J., Stroesser, G., and Cameron, G.N. 1994. The European Bioinformatics Institute (EBI)
Databases. Nucleic Acids Res. 22: 34453449.

George, D.G., Barker, W.C., Mewes, H.W., Pfeiffer, F., and Tsugita, A. The

 Page 248

PIR-International Protein Sequence Database. Nucleic Acids Res. 22: 35693573.

Gendel, S.M. 1996. The biotechnology for food safety database. In preparation.

Harper, R. 1994. Access to DNA and protein databases on the Internet. Curr. Opin. Biotechnol. 5: 418.

Henikoff, S. 1993. Sequence analysis by electronic mail server. Trends Biochem. Sci. 18: 267268.
Hill, W. 1996. The polymerase chain reaction: Applications for the detection of foodborne pathogens. Crit. Rev. Food
Sci. Nutr. 36: 123173.

Maidak, B.L., Larsen, N., McCaughey, M.J. Overbeek, R., Olsen, G.J., Fogel, K., Blandy, J., and Woese, C.R. 1994.
The Ribosomal Database Project. Nucleic Acids Res. 22: 34853487.

Masys, D.R. 1989. Toward global data interfacing. In: Biomolecular Data; A Resource in Transition, R.R. Colwell
(Ed.), p. 245259. Oxford University Press, Oxford.
States, D.J. and Boguski, M.S. 1991. Similarity and homology. In: Sequence Analysis Primer, M. Gribskov and J.
Devereux (Ed.), p. 89157. Stockton Press, New York.

Swaminathan, B. and Feng, P. 1994. Rapid detection of food-borne pathogenic bacteria. Ann. Rev. Microbiol. 48:
401426.

 Page 249

12
Molecular Fingerprinting of Foodborne Pathogenic Bacteria: An Introduction to Methods, Uses, and
Problems
Timothy J. Barrett
National Center for Infectious Diseases
Centers for Disease Control and Prevention
Atlanta, Georgia

Introduction

The identification of bacterial isolates to the species level, or even to the genus level, is sufficient in many situations.
For example, the identification of Shigella species in the stool of a child with bloody diarrhea is sufficient to establish a
diagnosis and suggest appropriate therapy. In many cases, however, bacterial isolates must be grouped at a lower level.
Bacterial subtyping is the differentiation of isolates below the species level. Examples of bacterial subtyping include
traditional methods such as serotyping, phage typing, biotyping, and antimicrobial resistance typing, and newer
methods such as molecular subtyping or DNA fingerprinting. Molecular methods are typically the most sensitive and
do not require any special reagents or skills, and they are thus becoming increasingly popular.

There are at least four reasons for performing subtyping. Determining the genotype may be important if the genotype is
clinically significant. For example, the type of toxin genes present is an important attribute of E. coli O157 or S.
aureus, and the genotype of hepatitis C virus is important in

 Page 250

predicting the clinical course of infection (Lau et al., 1995). Subtyping is also extremely helpful in determining
whether an infection is new or recurring. This is a critical factor in evaluating the effectiveness of antimicrobial therapy
for Helicobacter pylori (Taylor et al., 1995). But the most common, and probably most important, uses for bacterial
subtyping are the laboratory support of epidemiological investigations (both in hospital and community outbreaks) and
tracing the source of contamination of a food or other product.

An ideal subtyping method would be one in which all epidemiologically related isolates were identical and were
different from all epidemiologically unrelated isolates. An ideal system would also work for every organism and would
be 100% reproducible. There is, of course, no single ideal method, but a number of different approaches with their own
strengths and weaknesses.

Molecular Fingerprinting Methods

Common molecular fingerprinting methods include polymerase chain reaction (PCR) targeting specific genes,
arbitrarily primed PCR (which does not target known sequences), repetitive sequence typing, and a variety of
restriction fragment length polymorphism methods. Multilocus enzyme electrophoresis (MEE) is not a molecular
method in the sense that it is not DNA-based, but is often considered along with DNA-based methods because it
detects changes on the molecular rather than phenotypic level.

PCR Targeting Specific Genes

Since specific (and typically highly conserved) genes are being targeted, the PCR reaction can be run at high
stringency, resulting in a high degree of specificity and reproducibility. This type of PCR is most useful for genotyping
or for a first level of strain discrimination. Examples of this approach include determining which types of Shiga toxin
genes are present in an isolate of E. coli O157:H7 (Fig. 12.1) or which enterotoxin genes are present in an isolate of S.
aureus.

As illustrated by Fig. 12.1, the discriminative power of this approach (the number of different types detected) is usually
low. Typically, one is looking only for the presence or absence of a gene, not for genetic variability. If there are any
genetic differences within the gene, they would only be recognized by a difference in the size of the PCR amplicon,
and size differences are usually too small to be detected by standard gel electrophoresis. However, the discriminative
power of this approach depends upon the target sequences. Kostman et al. (1992) observed length polymorphisms in

 Page 251

Fig. 12.1
Detection of Shiga toxin (stx) genes in E. coli O157:H7 (genotyping)
by polymerase chain reaction (PCR). Isolates were tested in a multi-
plex PCR using primers for both stx1 and stx2. The isolate in lane 2 is
positive for stx1, isolates in lanes 3 and 8 are positive for stx2, and the
isolate in lane 4 is positive for both stx1 and stx2. Lanes 1 and 10
contain 123 bp ladder.

amplicons of the 16S-23S rRNA intergenic spacer region from Pseudomonas cepacia isolates, allowing considerable
strain discrimination.

The power of the specific PCR approach can often be greatly enhanced by choosing a target gene known to have
variable regions, then performing restriction digestion of the PCR product. Figure 12.2 shows results obtained by
restriction digestion of PCR-amplified flaA genes from isolates of Campylobacter jejuni. The isolates in lanes 2 and 3
have identical patterns, but each of the other isolates is a different subtype. Restriction digestion of a PCR-amplified
flaA gene is a relatively simple approach to subtyping C. jejuni, which provides adequate strain discrimination
(Nachamkin et al., 1993).

Arbitrarily Primed PCR

Arbitrarily primed PCR (AP-PCR) (Welsh and McClelland, 1990) and random amplified polymorphic DNA (RAPD)
(Williams et al., 1990) were originally described as two very similar but distinct methods. In recent years, the terms
have come to be used interchangeably. In AP-PCR, the primers are

 Page 252

Fig. 12.2
Molecular fingerprinting of Campylobacter jejuni by Ddel
restriction enzyme digestion of polymerase chain reaction-
amplified flaA gene. Lanes 1,7, and 12 are molecular size
markers (123 bp ladder). The isolates in lanes 2 and 3 have
identical RFLP patterns (fingerprints), as do the isolates in
lanes 8 and 9. All other isolates are different subtypes.
(Photo courtesy of Leta O. Helsel, CDC.)

not designed to match specific target sequences. Hybridization is performed at low stringency, allowing imperfect
matches between primer and target DNA. Since the primer sequences are not designed to match target sequences, a
single primer can be used to prime in both directions. Such primers will bind in multiple locations, wherever there is an
approximate match with the target DNA, yielding a more complex pattern than those obtained using PCR with specific
targets. Because of the complex pattern, strain discrimination is usually higher than with conventional PCR. How-

 Page 253

ever, the reproducibility of patterns obtained using arbitrarily primed PCR has been of considerable concern (Table
12.1), and it is usually necessary to perform several PCR reactions using different primers to obtain maximum strain
discrimination (Lin et al., 1996). Despite these problems, AP-PCR has been successfully applied to the molecular
epidemiology of numerous organisms including L. monocytogenes (Czajka and Batt, 1994).

Repetitive Sequence Typing

Repetitive sequence typing can be viewed as something of a compromise between conventional PCR typing and
arbitrarily primed PCR. Repetitive sequences are found in most bacteria. They are typically noncoding regions that
often contribute to the tertiary structure of the nucleic acid molecule. Since these are specific, known sequences, the
PCR conditions can be more stringent than those used for arbitrarily primed PCR, yielding more readily reproducible
results. However, repetitive sequence typing is typically less discriminating. Examples of repetitive sequences that
have been used for subtyping are enterobacterial repetitive intergenic consensus (ERIC) sequences and repetitive
extragenic palindromic (REP) sequences (Versalovic et al., 1991). These sequences have been shown to be useful in
subtyping C. jejuni, C. coli, and C. lari (Giesendorf et al., 1993).
TABLE 12.1 Reported Sources of Variability in Arbitrarily Primed Polymerase Chain Reaction
Source of variability Ref.
Primer concentration McPherson et al., 1993; Muralidharan and Wakeland, 1993
DNA template McPherson et al., 1993; Micheli et al., concentration 1994
DNA template quality McPherson et al., 1993; Micheli et al., 1994
Taq polymerase concentration and source Meunier, 1993; Schierwater and Ender, 1993
MgCl2 concentration Ellsworth et al., 1993
Number of thermal cycles McPherson et al., 1993; Micheli et al., 1994
Model of thermal cycler McPherson et al., 1993; Micheli et al., 1994

 Page 254

Restriction Fragment Length Polymorphism Methods

Probably the most widely used group of molecular fingerprinting methods are those based on restriction fragment
length polymorphism (RFLP). Perhaps the most discriminating method, but almost certainly the most difficult to
analyze, is restriction enzyme analysis of the total cellular DNA. Cellular DNA, both plasmid and chromosomal, is
extracted, purified, and digested with a frequent-cutting restriction enzyme. When the digested DNA is separated by gel
electrophoresis, 200 or more fragments can usually be visualized. This method is highly discriminating, but very
difficult to interpret due to the large number of bands (Fig. 12.3). Hybridization

Fig. 12.3
Restriction enzyme analysis (REA) of total cellular DNA from
isolates of Corynebacterium diphtheriae. Ethidium bromide-
stained gel showing complex patterns typical of REA.
(Photo courtesy of Michele Pinto and L. Mann Graves, CDC.)

 Page 255

methods have been introduced to reduce the complexity of the patterns, thus making interpretation easier. However,
hybridization methods require the blotting or transfer of the DNA to a membrane followed by hybridization with a
specific DNA or RNA probe, thus increasing the labor and time required to complete testing. The most widely used
hybridization method uses ribosomal RNA, or the DNA encoding ribosomal RNA (rDNA), as a probe and is usually
referred to as ribotyping. Figure 12.4 shows a membrane onto which the DNA from the gel shown in Fig. 12.3 has been
transferred and hybridized with rDNA. Only those fragments containing genes coding for rRNA are visible. While the
complexity of the patterns is greatly reduced, the ability to discriminate between isolates remains essentially
unchanged. Because ribosomal genes are highly conserved, one uni-

Fig. 12.4
Ribotyping of C. diphtheriae isolates. The gel shown in
Fig. 12.3 was blotted to a nylon membrane and hybridized
with an rDNA probe. The complexity of the fingerprinting
patterns are greatly reduced compared to Fig. 12.3, but the
level of straindiscrimination is comparable.
(Photo courtesy of Michele Pinto and L. Mann Graves, CDC.)

 Page 256

versal probe can be used for ribotyping many different organisms (Grimont and Grimont, 1986). Many other
hybridization methods have been reported, but most are specific for the organism of interest (Samadpour et al., 1993).

A very powerful and currently popular RFLP method is pulsed-field gel electrophoresis (PFGE). PFGE also utilizes
restriction digestion of total cellular DNA, but overcomes the problem of excessive complexity by using restriction
enzymes that recognize rare restriction sites and thus cut the DNA infrequently. The resulting fragments are very large
(typically 30500 kilobases or larger) and cannot be separated using conventional gel electrophoresis. Fragments are
separated by alternating the direction of the electric field (pulsing). Before the DNA molecules can begin to migrate,
they must orient themselves in the electric field. The larger the DNA molecule, the longer it takes to reorient before
movement through the gel begins. The smaller fragments begin to move sooner, and thus move further during the same
pulse interval. When the angle of the current is changed, the DNA fragments must reorient again before beginning to
move through the gel. As this process is repeated, the overall movement of the fragments is toward the bottom of the
gel and fragments are separated by size in the same way as conventional electrophoresis. PFGE typically gives high
strain discrimination and has proven useful for subtyping E. coli O157 (Barrett et al., 1994), Vibrio cholerae (Cameron
et al., 1994) many serotypes of Salmonella including typhimurium (unpublished data), and other foodborne pathogens.
The utility of PFGE for investigating outbreaks of E. coli O157:H7 became clear in January 1993, when a large
foodborne outbreak of E. coli O157:H7 occurred in the western United States (Bell et al., 1994). More than 600 culture
confirmed cases were recognized, mostly in the state of Washington, but cases also occurred in Nevada, Idaho, and
California. Illness was associated with eating undercooked hamburgers at 95 different locations of the same fast food
restaurant chain.

E. coli O157:H7 isolates from patients in each of the four states were indistinguishable from each other and from
isolates obtained from incriminated lots of raw hamburger (Barrett et al., 1994). E. coli O157:H7 isolates from ground
beef that was not epidemiologically incriminated had different PFGE patterns. The power of PFGE in discriminating
between the outbreak strain and unrelated isolates is illustrated in Fig. 12.5. Lane 2 contains DNA from an outbreak
isolate. Lanes 315 show isolates from Washington that were not epidemiologically related to the outbreak, but were of
the same phage type as the outbreak strain. None of the PFGE patterns matches the outbreak pattern, demonstrating the
ability of PFGE to distinguish outbreak from nonoutbreak isolates, even within the same phage type.

 Page 257

Fig. 12.5
Pulsed-field gel electrophoresis (PFGE) of E. coli O157:H7 phage type 14
isolates from Washington State. The isolate in lane 2 is from a patient
associated with the 1993 multistate outbreak. Isolates in lanes 315 are
unrelated to the outbreak and show different PFGE patterns.

Multilocus Enzyme Electrophoresis

MEE is based on the electrophoretic mobility of different allelic forms of enzymes (isozymes) involved in the standard
metabolic process of bacterial cells (the ''housekeeping" genes). Crude cellular protein extracts are electrophoresed
through a starch gel and detected with specific precipitating enzyme substrates. Since the allelic form of each enzyme
is independent of the other enzymes being tested (unlike RFLP, where the size of each fragment depends on the size of
other fragments), MEE is probably the best available tool for true phylogenetic studies. The utility of MEE for
subtyping varies with each organism. MEE has been very useful for Listeria monocytogenes (Baloga and Harlander,
1991), for example, but most E. coli O157:H7 isolates are the same MEE type (Whittam et al., 1988).

Problems in Interpretation of Molecular Fingerprinting Data

Genomic Diversity

Molecular fingerprinting methods have proven extremely valuable in epidemiological investigations and efforts to trace
the source of foodborne contamination. Despite their widespread use, there are many problems in interpreting the data
generated by these methods. All inherent problem, unrelated to the methodology used, is the problem of genomic
diversity.

 Page 258

Some bacteria, such as Salmonella serotype enteritidis, are relatively homogeneous, and thus most isolates look very
similar by almost any subtyping method (Powell et al., 1994; Stubbs et al., 1994). At the other end of the spectrum is
Helicobacter pylori, of which most isolates appear to be different subtypes unless the isolates came from the same
person (Hurtado and Owen, 1993). At both of these extremes, it is very difficult to determine the epidemiological
relevance of subtyping results. Fortunately, the genetic diversity of most foodborne pathogens, including E. coli O157
and Salmonella serotype typhimurium, falls somewhere in the middle.

Standardization and Reproducibility

To some extent, the issues of standardization and reproducibility are the same. Unless a method is carefully
standardized, reproducibility is likely to be a problem. But even if a method were 100% reproducible, standardization
would still be necessary for laboratories to be able to compare results. To say that two isolates are the same ribotype,
for example, has little meaning if one does not know the conditions under which the testing was performed. Table 12.2
lists some of the steps that must be standardized in order to compare ribotyping results obtained in different
laboratories. The procedure must be standardized from the first step (DNA extraction) to the last (deciding the range of
molecular sizes to include in the analysis). Perhaps the most critical factor is the choice of a restriction enzyme. It is not
at all unlikely that two isolates could be the same ribotype when tested with one enzyme and different ribotypes when
tested with a different enzyme. The choice of DNA or RNA probe is also critical. Two isolates may be the same when
probed with 16S rRNA and different when probed with 16S and
TABLE 12.2 Standardized Steps to
Compare Ribotyping Results from
Different Laboratories
DNA extraction and purification
Restriction digestion
Electrophoresis conditions
Blotting conditions
Choice of DNA or RNA probe
Hybridization conditions
Signal-detection method
Minimum intensity of positive signal
Range of molecular sizes used in analysis

 Page 259

23S rRNA. The other steps listed in Table 12.2 may be less critical, but they can also affect results.

Even if the molecular typing procedure has been carefully standardized, reproducibility, will vary with the methods
used (some methods are more easily reproducible than others) and with the particular organism. PFGE, for example, is
most reproducible with organisms that are easy to lyse and do not produce high levels of DNAses, such as most
members of the family Enterobacteriaceae. Reproducibility is also highly dependent on the skills and experience of a
particular laboratory and even more dependent on the abilities and attention to details demonstrated by the individual
technician performing the test.

Genetic Drift
Another major problem in interpreting molecular fingerprinting data is genetic drift, or the genetic changes that occur
naturally over time. Common sources of genetic drift include point mutations altering restriction sites or primer
binding sites, insertions (such as phages or transposons), deletions (loss of genetic material), rearrangements of the
genome within an organism, and recombination with other organisms resulting in the exchange of genetic material.
An illustration of the results of genetic drift is shown in Fig. 12.6. Each lane represents an isolate from a common
source outbreak of E. coli O157:H7. All of the PFGE patterns are identical, with the exception of an additional band in
lanes 5 and 8. All of the patients ate ground beef prepared at the same outlet of a Mexican food restaurant chain. Since
there were no known differences in patient exposures, these single band differences do not appear to be
epidemiologically significant. These additional bands may represent the acquisition of a phage or plasmid, but no work
was done to confirm this hypothesis (unpublished data).
Since genetic drift occurs naturally and causes changes in fingerprinting patterns, the obvious question is: How
different do isolates have to be to be considered different strains? The key factor is epidemiological relevance. The
mere ability to demonstrate differences does not mean that the differences are important. As in the case shown in Fig.
12.6, differences may even be misleading. The level of difference required to be epidemiologically significant varies
with the natural genetic diversity of the organism and with the discriminatory power of the method. It also varies with
the time frame of a study. During a surveillance project lasting several months, one would expect much more genetic
drift than during a point source outbreak in which all the cases were exposed at the same meal.

Considering the potential problems of genetic drift, how can the labora-

 Page 260

Fig. 12.6
Pulsed-field gel electrophoresis (PFGE) of E. coli O157:H7
isolates from persons eating undercooked ground beef at a fast-
food restaurant. All isolates have identical PFGE patterns except
for those in lanes 5 and 8, which show an additional band. Lanes
1 and 10 contain molecular size markers (lambda ladder).

torian decide which differences are important? One solution is to use more than one subtyping method. During the
western states outbreak of E. coli O157:H7 described above, we found five isolates from persons not known to be
connected with the outbreak that had PFGE patterns differing from the outbreak pattern by a single band (Barrett et al.,
1994). To determine whether these isolates might be outbreak-related in some undetermined way, we phage-typed them
and repeated the PFGE analysis using a second enzyme. Three of the five isolates were the outbreak phage type and
had the same PFGE pattern as the outbreak strain using the second enzyme. We concluded that these isolates probably
were the outbreak strain that had undergone some minor genetic change and probably were related to the outbreak in
some undetermined way such as undocumented secondary transmission.

When the same analysis was done on seven isolates from the Washington area showing a one-band difference from the
outbreak pattern, but collected prior to the beginning of the outbreak, only two isolates were the

 Page 261

same phage type, and none had the same PFGE pattern using the second enzyme. These results suggest that single band
differences may or may not be epidemiologically significant, at least with E. coli O157, and that further testing should
be performed if the distinction is critical to the epidemiological investigation. Even with additional testing, however, it
must be emphasized that molecular fingerprinting in the absence of careful epidemiological investigation is
meaningless.

The Future of Molecular Fingerprinting

Since molecular fingerprinting does not require specialized reagents or skills, it can be performed in virtually any
molecular biology laboratory. Careful standardization of laboratory methods is critical to the meaningful exchange of
data between laboratories. Computer software packages that allow electronic acquisition of gel images and computer-
assisted analysis of fingerprinting patterns will prove invaluable in establishing databases of pathogen fingerprints. The
Foodborne and Diarrheal Diseases Branch at CDC is working with the Association of State and Territorial Public
Health Laboratory Directors to establish a network for subtyping foodborne pathogens. Four state health department
laboratories have been provided equipment and training to perform PFGE typing of E. coli O157:H7 isolates. These
sites will be subtyping E. coli O157:H7 isolates from their regions and sending electronic images to CDC where they
will be analyzed and stored in a national database of E. coli O157 subtypes. This network will eventually be expanded
to include other sites and additional foodborne pathogens such as Salmonella serotype typhimurium.

PFGE and other subtyping methods are most likely temporary measures that will eventually be replaced by subtyping
based on DNA sequencing. DNA sequencing gives precise, quantifiable measures of genetic variability that cannot be
obtained using any other methods. The keys to developing DNA sequence-based subtyping methods will be the
selection of appropriate genes to sequence and the building of a database with which to compare results. The
practicality of DNA sequencing as a molecular fingerprinting method is limited by the currently available technology.
However, new developments, such as long PCR, which allows the amplification of 20 or more kilobases of DNA,
combined with improved sequencing technologies, will revolutionize the way bacterial identification, virulence testing,
and DNA fingerprinting are done. It may even be possible to identify a bacterium, test for virulence genes, serotype (by
detecting gene sequences responsible for antigen expression), and perform DNA fingerprinting without isolating the
organism.

 Page 262

References

Baloga, A.O. and Harlander, S.K 1991. Comparison of methods for discrimination between strains of Listeria
monocytogenes from epidemiological surveys. Appl. Environ. Microbiol. 57: 23242331.
Barrett, T.J., Lior, H., Green, J.H., Khakhria, R., Wells, J.G., Bell, B.P., Greene, K.D., Lewis, J., and Griffin, P.M.
1994. Laboratory investigation of a multistate food-borne outbreak of Escherichia coli O157:H7 by using pulsed-field
gel electrophoresis and phage typing. J. Clin. Microbiol. 32: 30133017.

Bell, B.P., Goldoft, M., Griffin, P.M., Davis, M.A., Gordon, D.C., Tarr, P.I., Bartleson, C.A., Lewis, J.H., Barrett, T.J.,
Wells, J.G., Baron, R., and Kobayashi, J. 1994. A multistate outbreak of Escherichia coli O157:H7-associated bloody
diarrhea and hemolytic uremic syndrome from hamburgersthe Washington experience. JAMA 272: 13491353.

Cameron, D.C., Khambaty, EM., Wachsmuth, I.K., Tauxe, R.V., and Barrett, T.J. 1994. Molecular characterization of
Vibrio cholerae O1 strains by pulsed-field gel electrophoresis. J. Clin. Microbiol. 30: 16851690.

Czajka, J. and Batt, C.A. 1994. Verification of causal relationships between Listeria monocytogenes isolates implicated
in food-borne outbreaks of listeriosis by randomly amplified polymorphic DNA patterns. J. Clin. Microbiol. 32:
12801287.

Ellsworth, D.L., Rittenhouse, K.D., and Honeycutt, R.L. 1993. Artifactual variation in randomly amplified
polymorphic DNA banding patterns. Biotechniques 14: 214217.

Giesendorf, B.A.J., Van Belkum, A., Koeken, A., Stegeman, H., Henkens, M.H.C., Van der Plas, J., Goosens, H.,
Niesters, H.G.M., and Quint, W.G.V. 1993. Development of species-specific DNA probes for Campylobacter jejuni,
Campylobacter coli, and Campylobacter lari by polymerase chain reaction fingerprinting. J. Clin. Microbiol. 31:
15411546.
Grimont, F. and Grimont, P.A.D. 1986. Ribosomal ribonucleic acid gene restriction patterns as potential taxonomic
tools. Ann. Inst. Pasteur/Microbiol. (Paris) 137B: 165175.

Hurtado, A. and Owen, R.J. 1993. Urease gene polymorphisms in Helicobacter pylori from family members. Med.
Microbiol. Lett. 2: 386393.

Kostman, J.R., Edlind, T.D., LiPuma, J.J., and Stull, T.L. 1992. Molecular epidemiology of Pseudomonas cepacia
determined by polymerase chain reaction ribotyping. J. Clin. Microbiol. 30: 20842087.

Lau, J.Y.N., Mizokami, M., Kolberg,J.A., Davis, G.L., Prescott, L.E., Ohno, T., Perrillo, R.P., Lindsay, K.L., Gish,
R.G., Qian, K-P., Kohara, M., Sim-

 Page 263

monds, P., and Urdea, M.S. 1995. Application of six hepatitis C virus genotyping systems to sera from chronic
hepatitis C patients in the Untied States. J. Infect. Dis. 17: 281289.

Lin, A.W., Usera, M.A., Barrett, T.J., and Goldsby, R.A. 1996. Application of random amplified polymorphic DNA
analysis to differentiate strains of Salmonella enteritidis. J. Clin. Microbiol. 34: 870876.

MacPherson, J.M., Eckstein, P.E., Scoles, G.J., and Gajadhar, A.A. 1993. Variability of the random amplified
polymorphic DNA assay among thermal cyclers, and effects of primer and DNA concentration. Mol. Cell. Probes 7:
293299.
Meunier, J.-R. and Grimont, P.A.D. 1993. Factors affecting reproducibility of random amplified polymorphic DNA
fingerprinting. Res. Microbiol. 144: 373379.

Micheli, M.R., Bova, R., Pascale, E., and D'Ambrosio, E. 1994. Reproducible DNA fingerprinting with the random
amplified polymorphic DNA (RAPD) method. Nucleic Acids Res. 22: 19211922.
Muralidharan, K. and Wakeland, E.K. 1993. Concentration of primer and template qualitatively affects products in
random-amplified polymorphic DNA PCR. Biotechniques 14: 362364.

Nachamkin, I., Bohachick, K., and Patton, C.M. 1993. Flagellin gene typing of Campylobacter jejuni by restriction
fragment length polymorphism analysis. J. Clin. Microbiol. 31: 15311536.

Powell, N.G., Threlfall, E.J., and Brown, D.J. 1994. Subdivision of Salmonella enteritidis PT4 by pulsed-field gel
electrophoresis: potential for epidemiologic surveillance. FEMS Microbiol. Lett. 119: 193198.

Samadpour, M., Grimm, L.M., Desai, B., Alfi, D., Ongereth, J.E., and Tarr, P.I. 1993. Molecular epidemiology of
Escherichia coli O157:H7 strains by bacteriophage lambds restriction fragment length polymorphism analysis:
application to a multistate food-borne outbreak and a day-care center cluster. J. Clin. Microbiol. 31: 31793183.

Schierwater, B. and Ender, A. 1993. Different thermostable DNA polymerase may amplify different RAPD products.
Nucleic Acids Res. 21: 46474648.

Stubbs, A.D., Hickman-Brenner, F.W., Cameron, D.C., and Farmer, J.J., III. 1994. Differentiation of Salmonella
enteritidis phage type 8 strains: Evaluation of three additional phage typing schemes, plasmid profiles, antibiotic
susceptibility patterns, and biotyping J. Clin. Microbiol. 32: 199201.

Taylor, N.S., Fox, J.G., Akopyants, N.S., Berg, D.E., Thompson, N., Shames, B., Yan, L., Fontham, E., Janney, F.,
Hunter, F.M., and Correa, P. 1995.

 Page 264

Long-term colonization with single and multiple strains of Helicobacter pylori assessed by DNA fingerprinting. J.
Clin. Microbiol. 33: 918923.

Versalovic, J., Koeuth, T., and Lupski, J.R. 1991. Distribution of repetitive sequences in eubacteria and application to
fingerprinting bacterial genomes. Nucleic Acids Res. 15: 78657876.

Welsh, J. and McClelland, M. 1990. Fingerprinting genomes using PCR with arbitrary primers. Nucleic Acids Res. 18:
72137218.

Whittam, T.S., Wachsmuth, I.K., and Wilson, R.A. 1988. Genetic evidence of clonal descent of Escherichia coli
O157:H7 associated with hemorrhagic colitis and hemolytic uremic syndrome. J. Infect. Dis. 157: 11241133.

Williams, J.G.K, Kubelik, A.R., Livak, K.J., Rafalski, J.A., and Tingey, S.V. 1990. DNA polymorphisms amplified by
arbitrary primers are useful as genetic markers. Nucleic Acids Res. 18: 65316535.

 Page 265

13
Bioluminescence: Lux as an Enabling Tool for the Microbiological Analysis of Food
Gordon S. A. B. Stewart, Timothy G. Aldsworth,
Rachel L. Sharman, Paula T. Gibson,
and Christine E. R. Dodd
The University of Nottingham
Leicestershire, England

Introduction

The application of the bacterial lux gene cluster as a reporter system for enabling bacterial detection, monitoring
bacterial injury and viability, and exploring the gene expression associated with bacterial adaptation to stress and
environmental change now underpins the validity of bioluminescence as a powerful reporter system for studying
diverse aspects of food microbiology. Since the relationship between cell viability (cuhurability) and light production
has been validated in diverse studies using different organisms and many inimical processes, it is possible to step
beyond the need to compare viable count data with bioluminescence data and use bioluminescence alone. This is
particularly important if microbiologists are to have new tools with which to monitor the growth and survival of
bacteria in real environments where they exist as heterogeneous mixtures and are fre-

 Page 266

quently associated with surfaces. This chapter explores the current state of the art in this field.

Bacterial Bioluminescence Biochemistry and Genetics

Luminescent bacteria are ubiquitous in the environment, where they occupy a number of ecological niches. The
bioluminescence of such bacteria results from the flavin-mediated oxidation of a long-chain aliphatic aldehyde
catalyzed by the enzyme luciferase. All bioluminescent bacteria so far studied, with the exception of Photorhabdus
luminescens, are marine organisms being easily isolated from a variety of marine habitats including free living (from
water), saprophytic or parasitic (on the surface of fish, crustacea, and mollusca), commensal (as gut bacteria of fish),
symbiotic (as in specialized light organs on fish), or from detritis. The sole terrestrial bioluminescent bacterium so far
identified, P. luminescens, has been isolated from nematodes infecting moth larvae, decaying meat, and human wounds
(Forst and Nealson, 1996).

The genes that code for the proteins required for a bioluminescent phenotype in bacteria all reside on a single operon,
the lux operon. This single transcriptional unit contains luxA and luxB, the genes coding for the a and b subunits of
bacterial luciferase, which are essential for luciferase activity, as well as luxC, luxD, and luxE, which code for the fatty
acid reductase complex required to produce the long-chain aldehyde substrate (tetradecanal) for the bioluminescent
reaction. Two of the substrates for the bioluminescent reaction, reduced flavin mononucleotide (FMNH2) and molecular
oxygen, are both readily available in aerobic bacteria (Meighen and Dunlap, 1993).

Transformation by lux genes has achieved light emission from a wide diversity of bacterial species which are normally
nonbioluminescent. Two approaches are commonly in use: insertion of just those genes coding for bacterial luciferase
(luxAB), which allows for bioluminescence upon addition of exogenous aldehyde substrate, or insertion of the whole
lux operon with expression of all the structural genes necessary for bioluminescence. Light emission by these
recombinant organisms may be exploited from a number of perspectives. It provides a real-time, noninvasive reporter
of gene expression, a sensitive marker for bacterial detection, and a holistic determinant of bacterial viability. These
applications have been the subject of several recent reviews (Jassim et al., 1990; Stewart, 1990, 1993; Stewart et al.,
1991, 1993, 1996; Baker et al., 1992; Stewart and Williams, 1992; 1993; Hill et al., 1993a,b).

 Page 267

Instrumentation

The most sensitive detector of photon emission available in most laboratories is the scintillation counter. A relatively
unsophisticated instrument, without sample cooling, can be configured for the detection of less than 10 bioluminescent
bacteria per mL of sample. To achieve this the coincidence counting, which is a standard feature of scintillation
counting, must be disengaged, thus enabling the photon detectors to register every photon, not just those that arrive at
the detectors with coincidence. Dedicated luminometer instruments may be less sensitive by an order of magnitude or
so, but represent the most versatile and least costly detection devices (Stanley, 1994). One of the great attractions of
bioluminescence is the ability to visualize light in two or even three dimensions. Reid et al. (1993) demonstrated the
potential of this technique from a food microbiology perspective to obtain nearly instantaneous images of dynamic
events that have previously been impossible to visualise. Bioluminescence images of the inhibition of S. typhimurium
in contact with fermented meat products were obtained after only 20 min contact time and subsequent accumulation of
light by photon counting. A number of photon-counting imaging devices are available (Nicolas, 1994), and these
provide spatial information in real time and in a nondisruptive manner. These instruments are capable of detecting the
photon emission from a single bacterial cell (Waterhouse et al., 1993; Hill et al., 1994).

Monitoring Injury and Recovery

Given that the production of light from recombinant bacteria containing the lux genes depends upon a functional
intracellular biochemistry, it can be established that any substance or environment that impairs that biochemistry, and
thus compromises cellular viability, will lead to a reduction in light emission (Stewart and Williams, 1992). Most Food-
production processes involve the application of physical treatments or chemical agents with the aim of reducing the
microbial load and extending the shelf life of food products. Although most of the cells may be nonviable, a fraction of
the population is often sublethally injured and difficult to detect using conventional methods, as they are unable to
grow without a recovery period (Busta, 1976). Sublethally injured bacteria may retain their pathogenic traits (Ray,
1979), so it is very important for food safety to enumerate such cells in microbial analyses. Additionally, if they remain
undetected it may be wrongly assumed that particular food-production processes produce uncontaminated products
(Hurst, 1977; Ray, 1979). The extent of cellular damage

 Page 268

depends on many factors, including the type and length of treatment, the physiological state of the cell, and the
processing medium. This makes it difficult to know the length of time required for recovery. In response to this
problem, tests for bacteria typically utilize an overnight incubation in an appropriate preenrichment medium before
attempting detection. Although there are current efforts to minimize the length of time required for bacterial
resuscitation, in general this is the limiting step that will compromise efforts to reduce bacterial detection times based
on new rapid biotechnology methods (see Chapters 1 and 6). There is, therefore, still a need to improve understanding
of the mechanisms of injury and recovery so that resuscitation times can be reduced.

Current methods to assess injury and recovery are based on the ability of cells to form colonies as effectively on
selective as on nonselective agar. This is purely an endpoint determination and gives no information regarding the
biochemical and physiological process of recovery. Using lux recombinant bacteria it is possible to explore in real time
the capability of cells to provide high-energy intermediates (FMNH2) following stress. A potential that remains largely
unexplored in terms of food microbiology is for bioluminescence to provide a selective means of detecting the
presence, activity, and recovery of viable but nonculturable cells (Duncan et al., 1994).

Freeze Injury

The extent of survival following freeze damage can be monitored by viable counts or bioluminescence with both
producing equivalent results (Ellison et al., 1991; Ellison et al., 1994a,b). Nonselective agar allows for the growth of
both uninjured and injured cells, whereas only uninjured cells are capable of growth on selective agar. During the first
50 min of recovery after freezing, the differential in counts of S. typhimurium between the two types of agar is clearly
evident, indicating the presence of injured cells. Directly after thawing the proportion of injured cells is very high,
comprising 98% of the total population (Ellison et al., 1991). A rapid increase in counts on selective agar after this
point indicates recovery of damaged cells. During this period the nonselective count is constant, confirming that the
population consists of mainly injured cells. After the 50-min recovery period the cells start to multiply and the counts
on both agars increase at a parallel rate. Bioluminescence data produce a trend similar to that of the selective plates,
which substantiates the fact that sublethally injured bacteria produce significantly less light than obtainable from an
optimally growing culture. The population typically returns to optimum bioluminescent potential at the end of the
recovery period. Sensitivity to selective agents and extended lag on nonselective media have been observed by many
workers (Clark and

 Page 269

Ordal, 1969; Ray and Speck, 1972; Mackey and Derrick, 1982); however, using bioluminescence, recovery can be
observed in real time providing additional information about the state of cells during the recovery process.

Heat Injury

The survival of bacterial pathogens, such as Salmonella species, in food is usually a consequence of process failure.
Heat, for example, is a traditional food-processing method used to reduce or eliminate microbial levels in foods thus
preventing bacterial associated disease and food spoilage. Critical margins for the amount of heat treatment required to
reduce the microbial load of food products to a safe level is routinely derived from models of thermal death kinetics.
Predictions of thermal death kinetics in a range of environmental conditions is becoming important as the nature of the
heating menstruum is known to have a profound effect on the heat resistance of microorganisms (Olson and
Nottingham, 1980). For instance, many research papers have demonstrated that reduction of water activity (aw) by the
presence of NaCl, sucrose, glucose, sorbitol, fructose, or glycerol enhances heat resistance of salmonellae (Baird-
Parker et al., 1970; Goepfert et al., 1970; Corry, 1974; Lee and Goepfert, 1975). The continued development of
predictive models that can accurately describe data across a wide range of food-processing conditions will be of benefit
to the food industry as processing conditions for changed product formulations can be modified easily. For example,
modeling techniques are now being applied to quantify the interaction of two or more factors (such as heat, pH, salt
concentration, etc.) on bacterial survival (Cole et al., 1993). At present the survival of bacteria such as Salmonella to
changes in temperature, pH, and aw is monitored in pure culture using classical plate count techniques. Principally, this
is because monitoring a target (pathogenic) bacterium in a complex mixture of bacteria is experimentally difficult or
even impossible by traditional methods. Bioluminescence offers a novel tool with which to explore the survival of
bacterial pathogens in complex environments that more closely model real food systems.

Temperatures above 45°C are too high to maintain the in vivo stability of the bacterial luciferase. As expected,
therefore, initial experiments comparing the reduction in bioluminescence with viable counts at 56°C demonstrated that
bioluminescence is inactivated at a rate in advance of viability, (Ellison, 1994a). To use bioluminescence as a monitor
of thermal inactivation, it has been necessary to develop an alternative method. Bioluminescence recovery curves for S.
typhimurium following varying lengths of heating at 56°C consist of an initial steep recovery period followed by an
exponential increase that reflects bacterial growth. The recovery period

 Page 270

represents resynthesis of luciferase enzyme rather than cellular recovery as no injury was detected by differential
counts on selective and nonselective agars (Ellison et al., 1994a). By extrapolating the exponential portion of the
recovery curve back to the y axis, a bioluminescence value is obtained that correlates very well with the reduction in
viable count immediately after heating (Ellison et al., 1994a,b).

Recently Mackey et al. (1994) have demonstrated that luciferases from different bacterial species vary in their heat
resistance. Dhir and Dodd (1995) used the Photorhabdus luminescens luciferase cloned into S. enteritidis (Dhir et al.,
1994) to compare the susceptibility of the organism to heating at 52°C when grown planktonically and surface-attached
through the use of bacterial bioluminescence. The thermal stability of this enzyme at the chosen heating temperature
meant that bioluminescence output directly correlated with cellular viability. This provides a new opportunity for the
use of bioluminescence in thermal inactivation studies by using a luciferase that is heat-stable in the temperature range
of interest.

Plate Count versus Bioluminescence

From the above there is a clear equivalence between in vivo bioluminescence and classical plate count data for
monitoring freeze and heat injury of S. typhimurium. Bioluminescence has the advantage of monitoring recovery in real
time and also reduces the laboratory's media preparation and sampling procedures, which encumber plate counts. This
aspect was demonstrated when sampling during freezing where bioluminescence proved to be an invaluable tool as
samples needed to be taken rapidly as well as recording recovery profiles. The data for survival of S. typhimurium from
heat treatment were of sufficient quality to enable computer-generated models to be fitted (by logistic function or
quadratic response surface), and these also demonstrate the equivalence of bioluminescence and viable count data
(Stewart and Williams, 1993; Ellison et al., 1994b). Such methods allow the interaction between two or more factors to
be quantified and can be used for prediction of thermal inactivation under conditions that have not been tested, using
interpolation within the range of the experimental matrix. Such models are a great asset to the food industry, as thermal
processing requirements can be predicted with ease and confidence.
As the equivalence of bioluminescence and viable count data has been proven, it is now possible to use
bioluminescence in conditions where plate counts are impractical. At the present time survival of microorganisms in
response to exposure to changes in temperature, pH, and aw is monitored in pure culture, whereas in the environment
they are part of a heterogeneous microflora and usually associated with surfaces.

 Page 271

Surfaces and Attached Cells

Effective hygiene-control measures coupled with the use of appropriate disinfection regimes are important tools for the
control of bacteria in food production and processing environments. Control measures must take account, however, of
the differential susceptibility of surface-attached versus planktonic cells (Carpentier and Cerf, 1993). The mechanisms
of bacterial attachment and the resulting hygiene and economic problems caused to the food and dairy industries have
recently been reviewed (Brown and Gilbert, 1993; Conner and Bilgili, 1994; Zottola, 1994). By comparison with their
planktonic counterparts, attached bacteria typically exhibit enhanced resistance to inimical processes such as
antimicrobial agents (Brown and Gilbert, 1993), biocides (Blenkinsopp et al., 1992), and heat (Frank and Koffi, 1990).
Recently, novel tools have become available for assessing the physiological status of planktonic, attached, and
detached cells, and these are proving invaluable to the further understanding of how cells acquire their enhanced
resistance to inimical processes (Wimpenny, 1993; Keevil, 1994; Yu and McFeters, 1994). In this respect the use of
bacterial bioluminescence has a significant role to play.
A good correlation between bioluminescence and viability has been demonstrated with biocide efficacy studies using
planktonic Listeria monocytogenes and Salmonella species (Jassim et al., 1990; Walker et al., 1992a, b). Extension of
these studies to surface-attached cells has demonstrated equivalence between colony count-derived and
bioluminescence-derived data for both the biocide concentration exponent (h), which describes how rapidly the
antimicrobial activity of the biocide is lost upon dilution, and thermal inactivation. Thus there can again be confidence
in the use of bioluminescence in circumstances where the direct enumeration of viable counts is difficult or impractical
(Denyer et al., 1991; Jassim et al., 1993; Walker et al., 1993; Dhir et al., 1994; Dhir and Dodd, 1995).

Bioluminescent Photobacterium leiognathi ATCC, 33469 has been evaluated as an indicator organism for cleanability
testing of food-processing equipment. The bacterium was grown as a biofilm for 5 or 24 hr at room temperature on
various surfaces, clean or milk-soiled, by immersing the surfaces in medium inoculated with P. leiognathi. The surfaces
were then rinsed or cleaned using a CIP procedure before the hygienic status was determined using conventional
plating, bioluminescence output, and image analysis. Rinsing did not eliminate the bacterium from the surfaces, nor did
the CIP procedure when 24-hr biofilms were used, as was readily seen after enrichment in a growth-sustaining
medium. In image analysis the biofilm showed a spotlike growth pattern, which was most evident on stainless steel
(Wirtanen et al., 1995).

 Page 272

The advantages of P. leiognathi for cleaning regimen assessment, which was based on the natural light output of the
test bacterium, are that the test was easy to perform and could be carried out without the use of expensive reagents. The
disadvantages of its use lie in the organism's physiological dissimilarity to many food-related bacteria and the inducible
nature of the light output. Recombinant derivatives of relevant gram-positive and gram-negative food pathogens, where
bioluminescence can be manipulated to be expressed constitutively or in relation to a particular gene function, offer an
important added dimension.

Competitive Microflora

Developing the application of bioluminescence as a monitor for the survival of a bacterial target within a complex
microflora, Duffy et al. (1995) compared bioluminescence and classical plate count methods for assessing the thermal
inactivation of S. typhimurium in pure and complex bacterial cultures. These studies were seen as a prelude to the use
of bioluminescence for studying injury in complex systems where parallel plate counts are unobtainable due to the loss
of injured bacteria on selective agars. The guiding principle for this work was to develop methods that have the
potential to enhance the data bases used for models of bacterial survival in complex environments and real food
systems.

Duffy et al. (1995) used the lowest possible concentration of Salmonella cells in order that the effect of increasing
levels of competitors could be studied. As a realistic heating temperature and a reasonable heating period were
required, the level of cells was quite critical: it was important not to inactivate all the cells or to compromise the ability
of the surviving cells to grow to a detectable level. Many experiments were conducted to establish the optimum
sampling conditions so that the cells would be reduced to a low level but not completely eliminated, allowing recovery
to be followed. An important advantage for bioluminescence was observed; samples that received the longest heat
treatment at 55°C were incubated in a Luminoskan luminometer (Labsystems [UK] Ltd., Hants, RG21 2YH) overnight
and the bioluminescence recorded for ail extended period. The extent of survival for the most severely heat-treated
cells could not be detected by viable counts (lower than the detection limit), whereas the extended recovery curves of
bioluminescence enabled the adjusted bioluminescence data to be obtained. Thus heat inactivation of bacteria can be
studied to lower levels of target bacterial cells than is practical by using traditional viable counts.

 Page 273

The effect of increasing numbers of competitors on the heat inactivation profile of S. typhimurium LT2 [pSB230] was
studied by maintaining a constant number of Salmonella cells (105 per mL) and using increasing numbers of
competitors. In the presence of 108 competitor bacteria/mL there was a significant increase in the number of S.
typhimurium cells surviving heat treatment at 55°C (the D value increased from 0.43 to 2.09 min).

The dramatic shift between 107 and 108 competitor CFU/mL at an underlying Salmonella concentration of 105 CFU/mL
is at just the concentration of cells that could signal commencement of entry into the stationary phase. It is conceivable
that the high concentration of competitors is ''read" by the Salmonella as a signal to induce stationary phase gene
expression mediated by RpoS. This observation has initiated a molecular approach to define the role of rpoS in the
"real world" of food microbiology.

The Role of RpoS

What is RpoS?

In the natural environment, growth rates are limited by the availability of specific nutrients and, consequently,
starvation is one of the most prevalent stresses bacteria encounter. When bacteria are starved of nutrients a large
number of physiological changes take place, which lead to an organism that is generally resistant to a range of stresses
and inimical processes. This empirical observation is a central tenet in food microbiology. With the identification of a
central regulator of stationary phase gene expression (Lange and Hengge-Aronis, 1991), there is now a growing
understanding of the molecular basis of the induction of this resistant state. The central regulator for early stationary
phase gene expression is the rpoS gene product ss or RpoS (Lange and Hengge-Aronis, 1991; Hengge-Aronis, 1993),
which, as an RNA polymerase sigma factor, is responsible for the induction of a specific subset of bacterial genes only
expressed under stress conditions. The ss factor is maximally expressed in the stationary phase of growth and, in turn,
controls the expression of a multitude of genes that confer a range of properties on stationary phase cells. For example,
bacteria in the stationary phase of growth are more thermotolerant (Hengge-Aronis et al., 1991), resistant to oxidative
stress (Lange and Hengge-Aronis, 1991), starvation, and osmotic stress (McCann et al. 1991). If the increased
thermotolerance observed by Duffy et al. (1995) is ss mediated it will have profound implications on the validity of
survival models, which fail to consider the physiological and molecular significance of competitive flora on the
adaptive response of food pathogens (McClure et al., 1994).

 Page 274

A Bioluminescent Biosensor for RpoS Expression

Reporter constructs have played a major role in elucidating the regulatory control of gene expression, and in this
context bioluminescence has been especially productive (Hill and Stewart, 1994). The Salmonella SpvR protein is a
transcriptional activator of the structural genes essential for virulence (Fang et al., 1991); although the detailed
mechanism of its action remains obscure, it is known to be regulated by RpoS (Kovarz et al., 1995). Coupling the spvA
promoter to the luxCDABE operon has provided pSB368, a reporter of SpvR-controlled promoter activity. Swift and
Stewart (1994) used this reporter in a model Escherichia coli system to investigate stationary phase expression of the
Salmonella virulence gene spvA. At the transition between exponential and stationary phases, E. coli BJ4(pSB368)
growing in Luria Bertini broth demonstrates an induction of spvA promoter activity, and hence RpoS activity, from a
basal level.

Induction of RpoS in the Presence of a Competitive Microflora

This rpoS biosensor has been used to monitor the induction of rpoS in S. typhimurium mediated by the presence of a
competitive microflora. Exponential cultures of S. typhimurium LT2(pSB367) at an initial density of 105 CFU/mL were
incubated with an exponential-phase mixed competitor microflora (E. coli, Pseudomonas fluorescens, and Citrobacter
freundii) at densities of 105, 107, or 108 CFU/mL. Growth (OD600nm) and rpoS expression (bioluminescence output) were
measured at intervals after mixing. In the absence of competitors (Fig. 13.1) bioluminescence output and hence rpoS
expression increased significantly as cells entered the stationary phase; this was seen by an inflection point in the curve
and the point of inflection termed the induction time. This induction indicates that this strain of S. typhimurium LT2 has
a functional RpoS homolog, unlike some other strains designated LT2, which have been reported as defective in this
gene (Swords and Benjamin, 1996; Wilmes-Riesenberg et al., 1996). When S. typhimurium was incubated with varying
levels of competitors, the induction time for rpoS expression, normalized against that of the control grown without
competitors, steadily decreased as the density of competitive microflora was increased (Fig. 13.2); these changes were
statistically significant (p = 0.007, one-way analysis of variance). Between 107 and 108 competitor CFU/mL the
induction time was most dramatically reduced; this is the cell density that would signal entry into the stationary phase.
When heat-killed competitor cells were used in analogous experiments, there were no statistically significant
differences between the normalized induction times at different cell densities (p = 0.248, one-way analysis of

 Page 275

Fig. 13.1
Inflection point of rpoS in S. typhimurium LT2[pSB367] with 0 competitors.

Fig. 13.2
Bioluminescence induction times for 105 CFU/mL S. typhimurium
LT2[pSB367] with 0 to 108 CFU/mL live or heat-killed competitors
(as a percentage of the axenic control).

 Page 276

variance) but the induction times were statistically different from those produced by the live competitors (p < 0.001,
one-way analysis of variance). Competitors, when present at a sufficiently high density, therefore induced early
expression of rpoS in S. typhimurium but only when live cells were present. Thus premature RpoS accumulation in
exponential stage S. typhimurium is due to alterations in the environment brought about by the activity of living
competitor cells rather than by the presence of the cells per se. However, the rpoS induction times in the presence of
108 CFU/mL of competitors was still longer than the time observed by Duffy et al. (1995), which induced resistance to
heating. This suggests that the interaction between the Salmonella and competitor cultures is complex and its
consequences may be hierarchical in terms of gene/operon expression.

Considering the results of Duffy et al. (1995), it was decided to extend the work of Ellison et al. (1994) and incorporate
a competitive microflora to investigate the resistance of S. typhimurium LT2[pSB100] to freeze inactivation. Figures
13.3 and 13.4 show that the presence of 108 competitor CFU/ ml, substantially increases the survival of the underlying
population of 105 CFU/mL Salmonella (from 0.23 to 12.97%) compared to when no competitors are present. However,
when 108 CFU/ml, heat-killed competitors were incorporated (Fig. 13.5), there was no difference in survival compared
to when competitors were absent (0.28 and 0.23%, respectively). Thus again living competitors cells are necessary, and
it is not just the additional cellular components acting as a thermoprotectant.

The Adaptive Response and Glass Formation

An example of the direct relevance of RpoS to food microbiology has been the discovery that bacteria defective in
RpoS production are highly sensitive to food-processing procedures (Rees et al., 1995). The production of glassy state
products characterized by molecular immobility, a high viscosity, and a correspondingly low molecular diffusivity
(Blanshard, 1988) is a well-recognized process in the food industry and is widely applied. Blissett et al. (1994) showed
that strains of S. typhimurium LT2 and S. senftenberg were able to survive the production and storage of glassy (aw of
0.450.28) and rubbery (aw of 0.930.96) gelatin sheets for up to 28 days at 22°C with only a 24 log reduction in viable
counts. Similar experiments carried out in our laboratory with the E. coli strain W3110, however, resulted in a
reduction of viability from 107 cells/cm2 to below the minimum level of detection. E. coli W3110 is a laboratory
attenuated strain of common use, which contains an amber mutation in rpoS rendering it inactive (Olsén et al., 1993).
By contrast, pressing of a wild-type isolate of E. coli termed BJ4 which is rpoS+ resulted in survival equivalent to that
obtained for Salmonella. Confirmation

 Page 277

Fig. 13.3
The resistance of 105 CFU/mL S. typhimurium LT2[pSB100] to freeze injury.

 Page 278

Fig. 13.4
The effect of 108 CFU/mL live competitors on the resistance of 105 CFU/mL S. typhimurium T2[pSB100]
to freeze injury.

 Page 279

Fig. 13.5
The effect of 108 CFU/mL heat-killed competitors on the resistance of 105 CFU/mL S. typhimurium LT2[pSB100] to
freeze injury.

 Page 280

of the importance of rpoS was obtained by using the BJ4 isogenic strain BJ4L1 which has a transposon insertion in
rpoS. As observed for E. coli W3110, this strain failed to survive the processing treatment. Figure 13.6 shows the
induction of rpoS in BJ4 as measured by the lux reporter construct pSB368 described above and by Swift and Stewart
(1994). The absence of induction from the rpoS mutant is also demonstrated. Survival of BJ4 following pressing in
gelatin sheets is shown to parallel bioluminescence in Fig. 13.7 demonstrating a highly significant link between the
activity of RpoS (as measured by bioluminescence) and survival of bacteria in pressed gelatin and during the formation
of the glassy state.

Although the above may seem rather esoteric, since rpoS mutants are not likely to be significant in real food systems,
the corollary of this is that cells actively expressing RpoS, as in the natural environment where the stationary phase is
the normal state, will be inherently more resistant to many food-processing conditions. Model systems that reflect the
survival of exponentially growing cells will not have accounted adequately for the influence of this adaptive response.
Since, as shown above, RpoS expression can radically alter bacterial survival during food processing, the prior history
of bacteria in raw materials may also influence the level of bacterial contamination in the final product (Archer, 1996).
Of equal concern is the finding that the expression of virulence determinants, such as spv, is potentiated by those very
conditions used to retard bacterial growth. RpoS is clearly significant in enhancing both survival and virulence of
foodborne pathogens in model systems, but is it exerting this global regulatory influence in real systems and

Fig. 13.6
Induction of rpoS (as measured by bioluminescence) as a function of growth
phase in E. coli BJ4[pSB368] and E. coli BJ4L1 [pSB368].

 Page 281

Fig. 13.7
Survival of E. coli BJ4[pSB368] after pressing as a function of rpoS levels.

under real processing and preservation regimens? If it is, there are very real opportunities to secure both commercial
and safety advantages from this knowledge.

There is one final and remarkable synergy between these studies and lux. The regulation of the lux bioluminescence
operon in P. fischeri is dependent upon a small diffusible autoinducer N-(3-oxohexanoyl)-L-homoserine lactone. Recent
research has shown that this autoinducer molecule is in fact one of a family of such molecules involved in cell-cell
signaling (quorum sensing) within the gram-negative bacteria (Swift et al., 1996). Latifi et al. (1996) have
demonstrated that for Pseudomonas aeruginosa the analogue N-butanoyl-L-homoserine lactone regulates rpoS
expression. This, coupled with the recent discovery that in E. coli the regulation of transcription of the cell division
genes ftsQA involves both rpoS and sdiA-mediated autoinduction through cell-cell signaling (García-Lara et al., 1996;
Sitnikov et al., 1996), strongly suggests that quorum sensing is a gene control system that will be found to regulate
rpoS in E. coli and Salmonella. The future therefore holds the possibility of using small molecule antagonists of cell-
cell communication to suppress rpoS expression in industrially relevant contexts.

A Reporter for Bacterial Detection

The concept of engineering bacteriophage to contain the lux bioluminescent gene cluster and to use such engineered
phage as tools for bacterial

 Page 282

detection was first described by Ulitzur and Kuhn (1987). The initial development target for this new approach to
bacterial detection was, not surprisingly, Salmonella. Since the original concept was defined, however, the gene
engineered into the bacteriophage for Salmonella detection has shifted from luciferase to ina (ice nucleation), and a
substantial body of work now underpins the comparative value of this novel bacterial-detection technology (Stewart et
al., 1996). The extension of food-relevant targets for lux engineered bacteriophage to Listeria (Loessner et al., 1996)
suggests that, while a long time in development, this application of lux and its associated bioluminescence may be on
the threshold of realizing its true potential.

Conclusions

Over the past decade the opportunity to apply lux and bacterial bioluminescence for the microbiological analysis of
food has been identified and consolidated. Like many tools in biology its value lies in enabling new observations on
nature which ultimately reflect the scientific perspective of the investigator. Our message is that bacterial colony counts
are no longer an adequate description of the microbiology of food. We are learning to ask new questions concerning the
intimacies of gene expression, phenotypic adaptation, and multicellular communication. The truth is out there, and the
light of bioluminescence can and will help us slowly yet surely discern it.

Acknowledgments

P.T.G. was supported by the University of Nottingham and the ACTIF II MAFF LINK programme. T.G.A. was funded
through a Leverhulme fellowship. R.L.S. was supported by a MAFF (UK) studentship.

References
Archer, D.L. 1996. Preservation microbiology and safety: Evidence that stress enhances virulence and triggers adaptive
mutations. Trends Food Sci. Technol. 7: 9195.

Baker, J.M., Griffiths, M.W., and Collins-Thompson, D.L. 1992. Bacterial bioluminescence: Applications in food
microbiology. J. Food Prot. 55: 6270.

Baird-Parker, J.M., Boothroyd, M., and Jones, E. 1970. The effect of water activity on the heat resistance of heat
sensitive and heat resistant strains of Salmonellae. J. Appl. Bacteriol. 33: 515522.

 Page 283

Blanshard, J.M.V. 1988. Elements of cereal product structure. In: Food Structure: Its Creation and Evaluation, J.M.V.
Blanshard and J.R. Mitchell (Ed.), p. 313330. Butterworths, London.

Blenkinsopp, S.A., Khoury, A.E., and Costerton, J.W. 1992. Electrical enhancement of biocide efficacy against
Pseudomonas aeruginosa biofilms. Appl. Environ. Microbiol. 58: 37703773.

Blissett, S.J., Bolton, K.J., Dodd, C.E.R., Gould, G.W., and Waites, W.M. 1994. Survival of Salmonella senftenberg
and Salmonella typhimurium in glassy and rubbery states of gelatin. J. Appl. Bacteriol. 76: 345349.

Brown, M.R.W. and Gilbert, P. 1993. Sensitivity of biofilms to antimicrobial agents. J. Appl. Bacteriol. 74 (Suppl):
8797.

Busta, F.F. 1976. Practical implications of injured micro-organisms in food. J. Milk Food Technol. 39: 138145.

Carpentier, B. and Cerf, O. 1993. Biofilms and their consequences, with particular reference to hygiene in the food
industry. J. Appl. Bacteriol. 75: 499511.

Clark, C.W. and Ordal, Z.J. 1969. Thermal injury and recovery of Salmonella typhimurium and its effect on
enumeration procedures. Appl. Microbiol. 18: 332336.

Cole, M.B., Davies, K.W., Munro, G., Holyoak, C.D., and Kilsby, D.C. 1993. A vitalistic model to describe the thermal
inactivation of Listeria monocytogenes. J. Indust. Microbiol. 12: 232239.

Conner, D.E. and Bilgili, S.F. 1994. Skin attachment model for improved laboratory evaluation of potential carcass
disinfectants for their efficacy against Salmonella attached to broiler skin. J. Food Prot. 57: 684688.

Corry, J.E.L. 1974. The effect of sugars and polyols on the heat resistance of salmonellae. J. Appl. Bacteriol. 37: 3143.

Denyer, S.P., Jassim, S.A.A., and Stewart, G.S.A.B. 1991. Bioluminescence and other rapid methods for following
adhesion of bacteria. Biofouling 5: 125132.

Dhir, V.K., Dodd, C.E.R., and Waites, W.M. 1994. The use of bioluminescent Salmonella to study freezing and heating
effects upon attached cells. Colloids Surf. B 2: 103110.

Dhir, V. K. and Dodd, C.E.R. 1995. Susceptibility of suspended and surface attached Salmonella enteritidis to biocides
and elevated temperatures. Appl. Environ. Microbiol. 61: 17311738.

Duffy, G., Ellison, A., Anderson, W., Cole, M.B., and Stewart, G.S.A.B. 1995. The use of bioluminescence to model
the thermal inactivation of Salmonella typhimurium in the presence of a competitive microflora. Appl. Environ.
Microbiol. 61: 34633465.

 Page 284

Duncan, S., Glover, L.A., Killham, K., and Prosser, J.I. 1994. Luminescence based detection of activity of starved and
viable but nonculturable bacteria. Appl. Environ. Microbiol. 60: 13081316.

Ellison, A., Perry, S., and Stewart, G.S.A.B. 1991. Bioluminescence as a real time monitor of injury and recovery in
Salmonella typhimurium. Int. J. Food Microbiol. 12: 323332.

Ellison, A., Anderson, W.A., Cole, M.B., and Stewart, G.S.A.B. 1994a. Modelling the thermal inactivation of
Salmonella typhimurium using bioluminescence data. Int. J. Food Microbiol. 23: 467477.

Ellison, A., Cole, M.B., and Stewart, G..S.A.B. 1994b. The application of molecular biology to understanding the
growth and survival of food poisoning and spoilage bacteria. In: Minimal Processing of Foods and Process
Optimisation, R.P. Singh and F.A.R. Oliveira (Ed.), p. 213227. CRC Press Inc., Orlando, Fla.

Fang, F.C., Krause, M., Roudier, C., Fierer, J., and Guiney, D.G. 1991. Growth regulation of a Salmonella plasmid
gene essential for virulence. J. Bacteriol. 173: 67836789.
Forst, S. and Nealson, L. 1996. Molecular biology of the symbiotic pathogenic bacteria Xenorhabdus species and
Photorhabdus species. Microbiol. Rev. 60: 2143.

Frank, J.F. and Koffi, R.A. 1990. Surface-adherent growth of Listeria monocytogenes is associated with increased
resistance to sanitizers and heat. J. Food Prot. 53: 550554.

García-Lara, J., Shang, L.H., and Rothfield, L.I. 1996. An extracellular factor regulates expression of sdiA, a
transcriptional activator of cell division genes in Escherichia coli. J. Bacteriol. 178: 27422748.

Goepfert, J.M., Iskander, I.K., and Amundson, L.H. 1970. Relation of the heat resistance of Salmonellae to water
activity of the environment. Appl. Microbiol. 19: 429433.

Hengge-Aronis, R. 1993. Survival of hunger and stress: the role of rpoS in stationary phase gene regulation in
Escherichia coll. Cell 72: 165168.

Hengge-Aronis, R., Klein, W., Lange, R., Rimmele, M., and Boos, W. 1991. Trehalose synthesis genes are controlled
by the putative sigma factor encoded by rpoS and are involved in stationary phase thermotolerance in Escherichia coli.
J. Bacteriol. 173: 79187924.

Hill, P.J. and Stewart, G.S.A.B. 1994. Use of lux genes in applied biochemistry. J. Biolumin. Chemilumin. 9: 211215.
Hill, P.J. Rees, C.E.D., Winson, M.K., and Stewart, G.S.A.B. 1993a. The application of lux genes. Biotechnol. Appl.
Biochem. 17: 314.

 Page 285

Hill, P.J., Denyer, S.P., and Stewart G.S.A.B. 1993b. Rapid assays based on in vivo bacterial bioluminescence.
Microbiol. Eur. May/June: 1621.

Hill, P.J., Eberl, L., Mølin, S., and Stewart, G.S.A.B. 1994. Microscopic detection of single cell bioluminescence from
Pseudomonas fluorescens using real time photon imaging. In: Bioluminescence and Chemiluminescence:
Fundamentals and Applied Aspects, A.K. Campbell, L.J. Kricka, and P.E. Stanley (Ed.), p. 629632. John Wiley and
Sons, UK.

Hurst, A. 1977. Bacterial injury: A review. Can. J. Microbiol. 23: 936942.

Jassim, S.A.A., Ellison, A., Denyer, S.P., and Stewart, G.S.A.B. 1990. In vivo bioluminescence: A cellular reporter for
research and industry. J. Biolumin. Chemilumin. 5: 115122.

Jassim, S.A.A., Camprubi, S., Tomas, J.M., Williams, P., Stewart, G.S.A.B., and Denyer, S.P. 1993. In vivo
bioluminescence for studying bacterial adhesion and in vitro phagocytosis. In: Bioluminescence and
Chemiluminescence, A.A. Szalay, L.J. Kricka, and P. Stanley (Ed.), p. 491495. John Wiley and Sons, UK.

Keevil, C.W. 1994. Methods for assessing the activity of biofilm microorganisms in situ. In: Bacterial Biofilms and
Their Control in Medicine and Industry. J. Wimpcnny, W. Nichols, D. Stickler, and H. Lappin-Scott (Ed.), p. 4547.
Bioline, Wiltshire.

Kovarz, L., Coynault, C., Robbesaule, V., and Norel, F. 1995. Role of the sigma-factor sigma(s) (rpoS) in Salmonella
typhimurium virulence. Med. Mal. Infect. 25: 10311034.

Lange, R. and Hengge-Aronis, R. 1991. Identification of a central regulator of stationary phase gene expression in
Escherichia coli. Mol. Microbiol. 5: 4959.

Latifi, A., Foglino, M., Williams, P., and Lazdunski, A. 1996. Regulatory cascade between the regulators by
autoinduction LasR and VsmR and the early stationary phase regulator RpoS in Pseudomonas aeruginosa. Mol.
Microbiol. (in press).

Lee, A.C. and Goepfert, J.M. 1975. Influence of selected solutes on thermally induced death and injury of Salmonella
typhimurium. J. Milk Food Technol. 38: 195200.

Loessner, M.J., Rees, C.E.D., Stewart, G.S.A.B., and Scherer, S. 1996. Construction of luciferase reporter
bacteriophage A511::luxAB for rapid and sensitive detection of Listeria cells. Appl. Environ. Bacteriol. 62: 11331140.

Mackey, B.M. and Derrick, C.M. 1982. A comparison of solid and liquid repair media for measuring the sensitivity of
heat injured Salmonella

 Page 286

typhimurium to selenite and tetrathionate media, and the time needed to recover resistance. J. Appl. Bacteriol. 53:
233242.

Mackey, B.M., Cross, D., and Park, S.F. 1994. Thermostability of bacterial luciferase expressed in different microbes.
J. Appl. Bacteriol. 77: 149154.

McCann, M.P., Kidwell, J.P., and Matin, A. 1991. The putative s factor KatF has a central role in the development of
starvation-mediated general resistance in Escherichia coli. J. Bacteriol. 173: 41884194.

McClure, P.J., Blackburn, C.D., Cole, M.B., Curtis, P.S., Jones, J.E., and Legan, J.D. 1994. Modeling the growth,
survival and death of microorganisms in foodsthe UK food micromodel approach. Int. J. Food Microbiol. 23: 265275.

Meighen, E.A. and Dunlap, P.V. 1993. Physiological, biochemical and genetic control of bacterial bioluminescence.
Adv. Microb. Physiol. 34: 167.

Nicolas, J.C. 1994. Applications of low-light imaging to life sciences. J. Biolumin. Chemilumin. 9: 139144.

Olsén, A., Arnqvist, A., Hammer, M., Sukupolvi, S., and Normark, S. 1993. The RpoS sigma factor relieves H-NS-
mediated transcriptional repression of csgA, the subunit gene of fibronectin binding curli in Escherichia coli. Mol.
Microbiol. 7: 523536.
Olson, J.C. and Nottingham, P.M. 1980. Temperature. In: Microbial Ecology of Foods, Factors Affecting Life and
Death of Micro-organisms, Vol. 1 ICMSF. Academic Press, UK.

Ray, B. 1979. Methods to detect stressed micro-organisms. J. Food Prot. 42: 346355.

Ray, B. and Speck, M.L. 1972. Repair of injury induced by freezing E. coli as influenced by recovery medium. Appl.
Microbiol. 24: 258263.

Rees, C.E.D., Dodd, C.E.R., Gibson, P.T., Booth, I.R., and Stewart, G.S.A.B. 1995. The significance of bacteria in
stationary phase to food microbiology. Int. J. Food Microbiol. 28: 263276.

Reid, C.L., Waites, W.M., and Stewart, G.S.A.B. 1993. Microbial detection and two dimensional preservative mapping
using lux-based bioluminescence. Society for Applied Bacteriology Technical Series, R.G. Kroll, A. Gilmour, and M.
Sussman (Ed.) 31: 181189.
Sitnikov, D.M., Schineller, J.B., and Baldwin, T.O. 1996. Control of cell division in Escherichia coli regulation of
transcription of ftsQA involves both rpoS and sdiA-mediated autoinduction. Proc. Natl. Acad. Sci. USA 93: 336341.

Stanley, P.E. 1994. Low light level measurements and kits and reagents

 Page 287

utilizing bioluminescence or chemiluminescencesurvey update-3. J. Biolumin. Chemilumin. 9: 123125.

Stewart, G.S.A.B. 1990. In vivo bioluminescence: New potentials for microbiology. Lett. Appl. Microbiol. 10: 18.

Stewart, G.S.A.B. 1993. BIOSENSORS bacterial luminescence: development and application. Lancet 341: 279280.

Stewart, G.S.A.B. and Williams, P. 1992. lux genes and the application of bacterial bioluminescence. J. Gen.
Microbiol. 138: 12891300.
Stewart, G.S.A.B. and Williams, P. 1993. Shedding new light on food microbiology. ASM News 59: 241246.

Stewart, G.S.A.B., Denyer, S.P., and Lewington, J. 1991. Microbiology illuminated: gene engineering and
bioluminescence. Trends Food Sci. Technol. 2: 710.

Stewart, G.S.A.B., Jassim, S.A.A., and Denyer, S.P. 1993. Engineering microbial bioluminescence and biosensor
applications. In: Molecular Biology in Clinical Research and Diagnosis. M.R. Walker (Ed.), p. 403424. Blackwell
Scientific Publications, UK.

Stewart, G.S.A.B., Loessner, M.J. and Scherer, S. 1996. The bacterial lux gene bioluminescent biosensor revisited.
ASM News 62: 297301.

Swift, S., Throup, J.P., Williams, P., Salmond, G.P.C., and Stewart, G.S.A.B. 1996. Quorum sensing: A population
density component in the determination of bacterial phenotype. Trends in Biochem. Sci. 21: 214219.

Swift, S. and Stewart, G.S.A.B. 1994. Luminescence as a signal of spvA expression. In Bioluminescence and
Chemiluminescence: Fundamentals and Applied Aspects, A.K. Campbell, L.J. Kricka, and P.E. Stanley (Ed.), p. 9396.
John Wiley and Sons, UK.

Swords, W.E. and Benjamin, J.R. 1996. Salmonella typhimurium LT2 is avirulent because of a defect in rpoS. Poster
B-117 96th ASM General Meeting.

Ulitzur, S. and Kuhn, J. 1987. Introduction of lux genes into bacteria, a new approach for specific determination of
bacteria and their antibiotic susceptibility. In: Bioluminescence and Chemiluminescence New Perspectives, J.
Schlömerich, R. Andreesen, A. Kapp, M. Ernst, and W.G. Woods (Ed.), p. 463472. Wiley, Bristol.
Walker, A.J., Holah, J.T., Denyer, S.P., and Stewart, G.S.A.B. 1992a. The antimicrobial activity of Virkon measured by
colony growth and bioluminescence of lux recombinant Listeria monocytogenes. Lett. Appl. Microbiol. 15: 8082.

Walker, A.J., Jassim, S.A.A., Holah, J.T., Denyer, S.P., and Stewart, G.S.A.B. 1992b. Bioluminescent. Listeria
monocytogenes provide a rapid assay for measuring biocide efficacy. FEMS Microbiol. Lett. 91: 251256.

 Page 288

Walker, A.J., Holah, J.T., Denyer, S.P., and Stewart, G.S.A.B. 1993. The use of bioluminescence to study the behaviour
of Listeria monocytogenes when attached to surfaces. Colloids Surf. A-Physiochem. Eng. Aspects 77: 225229.

Waterhouse, R.N., Silcock, D.J., White, H.L., Buhariwalla, H.K., and Glover, L.A. 1993. The cloning and
characterization of phage promoters, directing high expression of luciferase in Pseudomonas syringae pv phaseolicola,
allowing single-cell and microcolony detection. Mol. Ecol. 2: 285294.

Wilmes-Riesenberg, M.R., Foster, J.W., and Curtiss R. 1996. An altered rpoS allele contributes to the avirulence of
Salmonella typhimurium strain LT2. Poster B-120 96th ASM General Meeting.
Wimpenny, J.W.T. 1993. The physiology and biochemistry of biofilms. In: Microbial Biofilms: Formation and Control.
S.P. Denyer, S.P. Gorman, and M. Sussman (Ed.), p. 5194. Blackwell Scientific Publications, Oxford.

Wirtanen, G., Nissinen, V., Tikkanen, L., and Mattilasandholm, T. 1995. Use of Photobacterium leiognathi in studies
of process equipment cleanability. Int. J. Food Sci. Technol. 30: 523533.
Yu, F.P. and McFeters, G.A. 1994. Physiological responses of bacteria in biofilms to disinfection. Appl. Environ.
Microbiol. 60: 24622466.

Zottola, E.A. 1994. Microbial attachment and biofilm formation: a new problem for the food industry. Food Technol.
48(7): 107114.

 Page 289

14
Detection of Viable but Nonculturable and Stressed Microbial Cells
Rita R. Colwell
University of Maryland Biotechnology Institute
College Park, Maryland
Recent years have seen an upsurge of information concerning the phenomenon first described by our laboratory in the
late 1970s and early 1980s, namely, the ability of gram-negative bacteria not otherwise able to form spores to undergo
a dormancy whereby the cells remain viable and actively metabolizing but not able to be cultured by routine
bacteriological methods. Subsequent to the original observations of the ability to detect such cells in the environment
and in food, water, and clinical specimens, we have carried out studies showing that pathogenic organisms, such as
Salmonella and Shigella species, remain potentially pathogenic and, in the case of the latter, continue to produce toxin.
Such cells in large numbers can potentially be a serious public health hazard as evidenced by human volunteer studies
done with viable but nonculturable Vibrio cholerae 01. It is important in detecting these viable but noncuhurable cells
to recognize that the cells remain intact, able to take up substrate and metabolize substrate detectably. Thus, the use of
fluorescent antibody probes, as well as gene probes and polymerase chain reaction (PCR), makes it possible to detect
and quantitate such cells in a variety of samples. This phenomenon and

 Page 290

these techniques render the standard plate counting procedures unreliable in ensuring the safety of food and water for
public consumption. Methods for the detection of pathogenic microorganisms will need to be revised to take into
account the phenomenon of the dormant but non-spore-forming bacteria.

When bacteria are introduced into a new habitat, environmental changes with which they may be confronted include
temperature, nutrient concentration, salinity, osmotic pressure, pH, ion concentration and composition, and many other
factors. Bacterial cells dynamically adapt to these shifts in the environment, employing a variety of genetic
mechanisms. Bacteria can utilize constitutive and inducible enzyme synthesis to accommodate growth-limiting
nutrients, as well as adjust or reroute metabolic pathways to avoid metabolic and/or structural disruption caused by
specific nutrient limitations or changes in nutrient composition. Furthermore, they are able to coordinate their rates of
synthesis to maintain cell structure and function. These adaptive capabilities provide bacteria with mechanisms to
respond to their environment and survive and remain capable of resuming growth and cell division when
environmental conditions return to optimum.

Until recently, the ability to culture microorganisms on routine bacteriological media in the laboratory was considered
sufficient proof of viability. However, depending on efficiency, limitations, and/or selectivity of the medium employed,
interpretation of the viability of bacteria in environmental samples will vary. In the past, cells observed to be present by
microscopic examination but not able to be grown in the laboratory were termed ''dead" cells, "vegetative," "viable,"
"nonviable," "stressed," "injured," or "moribund" cells, these terms being applied ambiguously. In 1984, Roszak and
Colwell coined the term "viable but nonrecoverable" cells for bacteria having detectable metabolic function but not
able to be cultured by available methods (Roszak and Colwell, 1987). The terminology finally settled on was "viable
but nonculturable," and the proposal was made that the typical growth curve would need to be redefinedsince the so-
called "death" or "decline" phase includes viable but nonculturable, i.e., dormant, cells.

This phenomenon is not new. Microbial ecologists have long recognized that one of the major limitations to research in
microbial ecology is the inability to isolate and grow in culture the vast majority of bacteria that occur in nature. Thus,
recognition of "viable but nonculturable" bacteria has revived and revitalized interest in dormancy, survival, and
persistence of microorganisms in the environment. Bacterial cells, "intact and alive," according to selected metabolic
criteria, but not undergoing cell division in, or on, routinely employed bacteriological media are observed in soil and
water but have not been recognizednor has the concept been accepted

 Page 291

by clinical microbiologists. Food microbiologists refer to the situation as "injury" and have published extensively on
"resuscitation" of bacteria in foodstuffs (Ray, 1979).

Problems Encountered in Determining Metabolic State of Microorganisms

Problems confronted in determining the metabolic state of microorganisms observed in direct counts, but not
recoverable by plating or most probable number (MPN) counts, were recognized as early as the late 1800s and
subsequently investigated by many bacteriologists, including Knaysi (1935) and Janinson (1937). These authors did not
propose specific terminology to define the phenomenon, which reflects various mechanisms of bacterial survival. In
fact, individual strains within a species respond differently to environmental conditions, as well as to the length of
exposure to those conditions.

The history of the viable but nonculturable phenomenon is extensive. The subject of anabiosis, "return to life" or, more
appropriately in the case of bacteria, "latent life," intrigued microbiologists and can be traced back to the early
nineteenth century. Leeuwenhoek in 1702 recorded evidence for it. Spallanzani published observations that some
microscopic ''animals" could be treated with temperature changes, vacuum, electricity, or chemical agents and
subsequently be revived (L. Spallanzani, 1769; T. A. Spallanzani, 1803).

The term "anabiosis" was introduced by Wilhelm Preyer (1872, 1891) for the phenomenon now known as
"resuscitation." Valentine and Bradfield (1954) described bacterial viability in terms of multiplying and forming
colonies, but the designation "live" was suggested also for respiring cells unable to divide under the same conditions. A
few years later, Keilin (1959) introduced the term "cryptobiosis," that is, latent life, for an organism with visible signs
of life but slight metabolic activity. Kurath and Morita (1983) recognized the "live" status of respiring cells, as
previously described by Valentine and Bradfield (1954). Postgate and Hunter (1962), however, suggested that such
cells were dead if they did not divide, acknowledging that nondividing bacteria may, in some sense, be alive because
they retain their osmotic barrier. Postgate (1967) further described the transient state between viability and death when
cells exposed to starvation were incapable of multiplication but maintained metabolic function. Thus, according to
Postgate, the major criteria for cell viability were multiplication and formation of colonies. Roszak et al. (1984)
described "viable but nonculturable" Salmonella enteritidis detected by the direct viable count (Kogure et al., 1979)

 Page 292

that were not culturable on bacteriological media. Viable but nonculturable organisms maintain metabolic activity,
indicated by uptake of various metabolic substrates (Chowdhury et al., 1995).

Bacteria with the greatest capacity to survive during starvation have been shown to be small in size and to have lower
metabolic rates. Dawes and Ribbons (1965) have shown that this characteristic low metabolic rate is shared with
spores, which exhibit a capacity for survival by shutting down almost all metabolic activity with the cells remaining
viable. The "rounding-up" phenomenon, with concomitant reduction in cell volume, was reported for Vibrio spp. under
low nutrient conditions, i.e., in an environment relatively free of organic nutrients, at which time bacterial cells are
observed to adjust to substrate-limited stress conditions by adopting necessary physiological changes. Campylobacter
jejuni forms typical coccoidlike cells with intact membranes when induced to the nonculturable state and retaining
viability (Rollins and Colwell, 1986).

It has been suggested that due to morphological changes, the "rounding-up" and reduction in size of the cells may be as
much as 15- to 300-fold, depending on the level of nutritional stress. Anderson and Hoffman (1965) were able to filter
organisms from seawater that passed through a 0.45-mm filter membrane but were retained on 0.22-mm membranes.
These bacteria were later identified as Spirillum, Leucothrix, Flavobacterium, and Vibrio spp. This size reduction has,
subsequently, been observed for many viable but nonculturable gram-negatives, including V. cholerae. Thus, filtration
membranes with no larger than 0.22-mm pore size must be used to collect such cells (Xu et al., 1982).

For Vibrio cholerae, the term "survival" had been viewed as reflecting a high degree of host adaptation, i.e., cholera
vibrios were believed to be able to exist only for very short periods of time outside the human intestine. Now, in fact,
the evidence accumulated over the past decade shows that V. cholerae is an autochthonous inhabitant of brackish water
and estuarine systems (Colwell et al., 1977). Thus, very early studies of V. cholerae (prior to 1970) were aimed at
identifying any environmental conditions that would allow cholera vibrios to remain culturable in the environment,
since any environment would be considered cholera-free if no cholera vibrios could be cultured. In fact, the dogma was
that an environment had to be recontaminated with infected stool or other material for cholera bacteria to be present.
We now know this not to be accurate.
New information imparts a dynamic meaning to the term "survival," with the evidence showing that V. cholerae cells
do not die when discharged into aquatic environments, but instead remain viable and are capable of transforming into a
"culturable state" if environmental conditions again become favorable, as suggested earlier (Huq et al., 1988).

 Page 293

These studies were based on observations using microcosms simulating saline, estuarine, brackish, and freshwater
environments. They have provided new and important information on V. cholerae physiology, namely, the effect of
temperature and salinity, adherence, and colonization of chitincontaining and mucilaginous macrobiota. Studies prior to
1970 were all based on methods for isolation and characterization of V. cholerae that had been originally developed for
clinical diagnosis of cholera in hospital laboratories (Finkelstein, 1973). The many difficulties associated with isolation
of V. cholerae 01 from the aquatic environment can be related to the simple fact that methods for isolating V. cholerae
from clinical specimens containing large numbers of actively growing cells do not recover organisms from
environmental samples. Clinical samples are likely to contain fewer cells adapted to a variety of environmental
conditions, most commonly low nutrient concentration, pH 78 (typical of seawater), fluctuating temperature and pH,
variation in oxygen tension, exposure to UV from sunlight, etc. (Litsky, 1979). With modern molecular biology
techniques, such as PCR and gene probes, combined with immunological direct detection methods, it can be
convincingly demonstrated that viable but nonculturable cells of V. cholerae 01 also occur in clinical specimens from
cholera patients (Colwell et al., 1985, 1990, 1992; Huq et al., 1994).

Organisms in the viable but nonculturable state go through different stages indicated by morphology of the cells.
Kondo et al. (1994), using freeze-fixation and electron microscopy, demonstrated morphological changes occurring in
viable but nonculturable cells of V. cholerae. A sequence of gradual morphological changes in cell structure has
similarly been documented in our laboratory using transmission and scanning electron microscopy (Chowdhury et al.,
1996). Morphological changes can be correlated with altered nutritional and physicochemical conditions in the
environment. An audiographic study of tritium-labeled Vibrio cholerae (Roszak et al., 1984) revealed aggregation of
silver grains associated with uptake of radiolabeled substrate. Thus, viable but nonculturable cells can be demonstrated
in a variety of ways to be metabolically active.

Bacterial cells in the viable but nonculturable state retain virulence. The first demonstration of this potential was
reported in E. coli, demonstrated by accumulation of fluid in the ileal loop of rabbits after inoculation with viable but
nonculturable cells (Colwell et al., 1985). These findings were corroborated by the observation that virulence plasmids
are maintained in E. coli when the cells are in the viable but nonculturable state (Byrd and Colwell, 1990; Grimes and
Colwell, 1986). Viable but nonculturable cells of Vibrio vulnificus have been shown to cause death in mice (Oliver,
1993). Virulence of viable but nonculturable cells of Campylobacter jejuni also has been demonstrated in rats (Shaha et
al., 1991). However, some reports suggest that

 Page 294

virulence may be lost upon entry of certain cells into the viable but nonculturable state (Medema et al., 1992).
Nevertheless, from results of investigations reported to date, it appears that most pathogens maintain virulence,
suggesting that careful analysis of food and water samples beyond bacteriological culture alone is needed to assess
public health safety.

The serotype 01 of V. cholerae has long been identified as the epidemic agent of cholera, while the 0139 serotype was
recognized only recently as having the potential to cause epidemics (Ramamurthy et al., 1993). Both serotypes have
been shown to enter into the viable but nonculturable state. A 3-year study conducted in Bangladesh showed that 63%
of plankton samples collected very 2 weeks from ponds and rivers were positive for V. cholerae 01 using the direct
fluorescent antibody-detection method, yet only 1% were culture-positive (Huq et al., 1990). These natural waters are
frequently used for bathing and as a source of water for cooking and drinking. Recently, coexistence of both V.
cholerae 01 and 0139 was reported in plankton samples collected in Bangladesh and tested using the direct detection
method because culture methods were rarely successful (Huq et al., 1995). A field study was conducted in Bangladesh
in which stool samples were collected from patients who were symptomatically confirmed as suffering from cholera.
Even these samples did not always yield a positive culture. Furthermore, 8% of the stools that were culture-negative
turned out to be positive by Cholera DFA, CholeraScreenTM, a coagglutination test, and CholeraSMARTTM, a
colorimetric antibody test. All three tests are commercially available direct detection tests (Hasan et al., 1991). Some of
the culture-negative, DFA-positive stools were confirmed to be positive by PCR (Hasan et al., 1994a).

A cholera outbreak resulting in one death occurred in California in 1993 among airline passengers arriving in the
United States from Argentina (Abbott and Janda, 1993). Stool samples from these passengers were subjected to culture,
with less than 100% success. However, vibriocidal antibody titer against V. cholerae was significantly high in blood
collected from patients whose stool samples were culture-negative but positive by CholeraScreenTM (Abbott and Janda,
1993).

A study we conducted demonstrated that infections resulting from diving in contaminated waters do not always
manifest clinical symptoms. Infection was determined by detection of raised antibody titers against Pseudomonas
aeruginosa in positive blood samples collected 30 days after a dive and compared with predive blood samples
(Losonsky et al., 1994), taken when the organism was isolated from the diving site. Elevated antibody titer was
detected against V. cholerae 01 in postdive blood samples collected from American and Russian divers after diving in
waters of their respective coun-

 Page 295

tries (Huq et al., 1994). Culture of V. cholerae was not always successful, but the organisms were readily detected in
water samples from the diving sites using the direct fluorescent antibody (DFA) method employing monoclonal
antibody.

During a field trip to Russia, a few of the water and plankton samples collected from the Black Sea were positive for V.
cholerae 01 by culture, but nearly all of the samples revealed the presence of nonculturable cells of V. cholerae 01 by
fluorescent antibody method. Four weeks after this work was completed, an outbreak of cholera occurred in a coastal
area where swimming and bathing were permitted, a report of which appeared in a local Ukrainian newspaper (Huq et
al., 1994).

Because the standard plate count cannot be used to enumerate viable but nonculturable organisms, alternative methods
have been developed to enumerate these bacteria based on direct counting, the most common of which is
epifluorescence microscopy and acridine orange staining (Daley and Hobbie, 1975; Floodgate, 1979).
Staining methods, including acridine orange, were introduced by Strugger in 1949 to differentiate live cells from dead
cells. However, the AODC method is used mainly to enumerate total bacteria in environmental samples (Bowden,
1977; Fry, 1990). In light of current knowledge of viable but nonculturable bacteria, there are several sensitive and
specific methods of direct detection. A fluorochrome-based staining method employing 4'6-diamidino-2-phenyl indole
(DAPI) was introduced by Porter and Fieg (1980). This staining procedure has become more popular during recent
years and is the method preferred by environmental microbiologists since it estimates the proportion of total bacterial
cells within a specific serotype or taxon, when fluorescent DNA probes or fluorochrome-labeled antibodies are
combined with a general fluorochrome, e.g., DAPI (Kepner and Pratt, 1994).
The direct viable count (DVC) developed by Kogure et al. (1979) is perhaps the most convincing and widely used
method of detecting viable cells, irrespective of culturability. In this procedure, viable cells are easily recognized
because of enlarged, elongated morphology. This test has been further optimized for detection of V. cholerae 0l by
combining it with fluorescent monoclonal antibody (FA) staining. Using a monoclonal antibody in the FA method, Huq
et al. (1990) successfully demonstrated the presence of V. cholerae 01 cells in the natural aquatic environment year
round in Bangladesh, where cholera is endemic. The FA method combined with the DVC method (Brayton and
Colwell, 1987) was further developed to permit direct fluorescent antibody-direct viable counting (DFA-DVC), which
is now used as a more convenient method for detection of viable but nonculturable cells of V. cholorae (Chowdhury et
al., 1995). The DFA-DVC method can be used to

 Page 296

detect very small numbers of organisms in food and water samples by concentrating bacteria on filters prior to staining
(Hasan et al., 1995).

The use of soluble p-iodonitrotetrazolium (INT) to demonstrate viable bacterial cells was introduced by Zimmerman et
al. (1978). Combined with the Kogure et al. (1979) DVC method, it is described as the INT-DVC method by Hasan
(1995). This method makes possible the detection of viable but nonculturable bacteria in water samples, including
Aeromonas hydrophila (Hasan, 1995) and Shigella dysenteriae (Rahman et al., 1996).

PCR has been used by various investigators to detect V. cholerae in food and stool specimens (Koch et al., 1993; Field
et al., 1992). This method is especially useful for organisms in the nonculturable stage. In environmental specimens,
the concentration of bacterial cells is usually very low, making them difficult to detect. Koch et al. (1993) have claimed
to detect as few as 1 V. cholerae cell per 10 g of food with amplification reaction from crude bacterial lysates. Shirai et
al. (1991) reported use of PCR for detection of the cholera enterotoxin operon of V. cholerae 01 in stool specimens.
More recently, we have been able to demonstrate the presence of viable but nonculturable cells of V. cholerae 01 using
PCR to confirm VBNC V. cholerae in laboratory microcosms, food samples, and culture-negative stool samples (Hasan
et al., 1994a).

A modified fluorescent antibody method, originally described by Brayton et al. (1987), has been optimized to kit form,
reducing the time required to complete the test to less than 10 minutes (Hasan et al., 1994b,c). This test, optimized for
clinical specimens, is a colorimetric immunoassay, Cholera-SMARTTM (Hasan et al., 1994a), and has proven valuable
in the field. The test kit requires neither culture nor refrigeration, making the tests valuable for detection of V. cholerae
in the nonculturable stage in field work.
Studies of human volunteers to determine whether ingestion of viable but nonculturable V. cholerae 01 would revive in
the human intestine and revert to a culturable state have been done. Attenuated recombinant vaccine strain CVD 101,
which does not express cholera holotoxin, was selected for the test (Levine et al. 1988; Kaper et al. 1984). Recent
studies (Ravel et al., 1994), have yielded transposon mutants of V. cholerae, demonstrating that the altered viable but
nonculturable response is genetically regulated.

The effect of solar irradiation on viability, i.e., platability, of E. coli, was measured in microcosms and in membrane
chambers. E. coli cells exposed to measured quantities of sunlight showed stable DVC and AODC counts, but plate
counts (culturable cells) decreased from 2 × 106 cells/mL to no detectable culturable cells after exposure to sunlight for
26 hr. Plate counts of cells incubated in the dark did not decrease. However, as soon as exposure of these cells to light
occurred, the plate counts also decreased, but in both cases the AODC and DVC did not change significantly. Plate
counts of

 Page 297

samples of surface water in chambers suspended in the Morlaix Estuary, France, showed rapid decrease in plate counts
upon exposure to sunlight, whereas plate counts for samples from chambers incubated at the bottom of the estuary
remained stable, i.e., the number of culturable bacteria did not change significantly with time (Pommepuy et al., 1992).

Fluid in loops of rabbits (R2) was examined by GM1-ELISA. The direct assays yielded positive results in the GM1-
ELISA. That is, E. coli H10407, after exposure to seawater, i.e., both culturable and nonculturable samples, was able to
produce LT in rabbit intestinal loops, confirmed by GM1-ELISA. Thus, E. coli exposed to seawater retains the ability
to produce enterotoxin, whether culturable or nonculturable.

Based on results of these and earlier studies (Grimes et al., 1986), it is concluded that enteropathogenic E. coli
discharged into seawater in estuaries and coastal waters can retain enteropathogenicity. Results of studies employing
Helicobacter pylori also showed that cells of H. pylori exposed to river water in laboratory microcosms for up to 2
years can retain viability and selected properties associated with pathogenesis (Shahamat et al., 1993).
When ingested by shellfish, V. cholerae, E. coli, and related bacteria may retain culturability, while the same bacteria in
the surrounding seawater exposed to sunlight may be viable but nonculturable, leading to a disparity between shellfish
and water counts for the same site. In light of these considerations, the "die-off" calculation employed by sanitary
engineers for measuring the quality of seawater at sewage outfalls may be too simple a measurement and misleading, at
least in the context of public health.

In conclusion, studies of the persistence and survival of V. cholerae, E. coli, and other gram-negative bacteria in the
environment have shown that, in the absence of nutrient, these organisms evolve toward a nonculturable state, while
remaining viable, a result confirmed by animal assay (Colwell et al., 1985).

Under certain conditions, V. cholerae can remain in a culturable state in the marine environment for very long periods
of timemonths or even years. The significance of this finding is obvious in view of recent outbreaks of cholera in
coastal cities of Asia and Latin America.

Seroconversion Studies
The Oag antigenic property of V. cholerae 01 is divided into three antigenic factors: A, B, and C. Only the A antigenic
compound has been found to be present in all three different serovars, i.e., Inaba (A and C), Ogawa (A and B), and
Hikojima (A, B, and C). Seroconversion of V. cholerae 01 serovar Ogawa to Inaba has been described (Sach and
Miller, 1969). Also, this

 Page 298

conversion can be induced by growth in the presence of anti-Ogawa serum, but the reverse was not observed until very
recently (Stroeher et al., 1992). We reported earlier the seroconversion from V. cholerae non-01 to the 01 serogroup
using artificial seawater microcosms and an incubation temperature of 35°C, with detection by direct microscopy of
fluorescent monoclonal antibody-stained cells (Colwell et al., 1996). In a recent study, we detected a similar frequency
of seroconversion, i.e., one 01 cell in ca. 105 non-01 cells (Colwell et al., 1996). It has been reported that V. cholerae
serogroup non-01 may carry a C-antigenic factor, called the Hakata serogroup (Shimada and Sakazaki 1988). An
environmental cuttlefish isolate of a Vibrio species, presumptively identified as V. fluvialis, has also been reported to
carry the C-antigenic factor (Shimada et al., 1987). It is important to note that the fluorescently labeled monoclonal
antibody employed in our studies is anti-A factor (Brayton et al., 1987).

It is clear from the recent attention given the 01 versus non-01 serogroup and the arrival of a non-01 epidemic
serogroup, 0139, that some seroconversion and/or change in the cell surface properties may occur naturally in the
environment (Bik et al., 1995). The mechanism(s) involved in such seroconversion is, however, not yet known.
Significant differences in seroconversion were not observed that could be ascribed to temperature or salinity,
suggesting that seroconversion can occur in fresh, brackish estuarine, or seawater throughout all seasons. However,
seroconversion occurred more rapidly when cells were harvested from stationary phase culture and the yield of
seroconverted cells was higher, i.e., 2 log units greater. Thus, cells exposed to the environment for very long periods of
time, perhaps in the dormant, i.e., viable but nonculturable (VBNC) stage (Colwell et al., 1985; Hasan et al., 1994a; Xu
et al., 1982) may, upon entry to the culturable stage, seroconvert more frequently, perhaps accounting for the apparent
seasonality of V. cholerae 01 in aquatic environments.

In summary, detection of VBNC pathogens in food and/or water for human consumption is important from a public
health perspective. Standard methods of culture are no longer sufficient for estimating food and water safety. Instead,
the appropriate application of newer methods based on molecular biology can be more informative and reliable.

References

Abbott, S.L. and janda, M. 1993. Rapid detection of acute cholera in airline passengers by coagglutination assay. J.
Infect. Dis. 168: 797799.

Anderson, J.I. and Hoffman, W.P. 1965. Isolation and characterization of filterable marine bacteria. J. Bacteriol. 90:
17131718.

 Page 299

Bik, E.M., Bunschoten, A.E., Gouw, R.D., and Mooi, F.R. 1995. Genesis of the novel epidemic V. cholerae 0139
strain: evidence for horizontal transfer of genes involved in polysaccharide synthesis. EMBO J. 14: 209216.

Bowden, W.B. 1977. Comparison of two direct-count techniques for enumerating aquatic bacteria. Appl. Environ.
Microbiol. 33: 12291232.

Brayton, P.R. and Colwell, R.R. 1987. Fluorescent antibody staining method for enumeration of viable environmental
Vibrio cholerae 01. J. Microb. Methods 6: 309314.

Brayton, P., Tamplin, M., Huq, A., and Colwell, R.R. 1987. Enumeration of Vibrio cholerae 01 in Bangladesh waters
by fluorescent-antibody direct viable count. Appl. Environ. Microbiol. 53: 28622865.

Byrd, J.J. and Colwell, R.R. 1990. Maintenance of plasmids pBR322 and pUC8 in noncultivable Escherichia coli in
marine environment. Appl. Environ. Microbiol. 56: 21042107.

Chaudhury, K., Bhadra, R.K., and Das, J. 1992. Cell surface characteristics of environmental and clinical isolates of V.
cholerae non-01. Appl. Environ. Microbiol. 58: 35673573.

Chowdhury, M.A.R., Miyoshi, S., Yamanaka, H., and Shinoda, S. 1992. Ecology and distribution of toxigenic V.
cholerae in aquatic environments of a temperate region. Microbios 72: 203213.
Chowdhury, M.A.R., Ravel, J., Hill, R.T., Huq, A., and Colwell, R.R. 1995. Physiology and molecular genetics of
viable but nonculturable microorganisms. In: Biotechnology and Risk Assessment, M. Levin, C. Grim, and J.S. Angle
(Ed.), p. 105122. University of Maryland Biotechnology Institute, College Park, Md.

Chowdhury, M.A.R., Montilla, R., Quintero, E.J., Huq, A., Maugel, T., Xu, B., Husan, J.A.K., and Colwell, R.R. 1996.
Changes in cellular morphology of Vibrio cholerae with entry into the viable but nonculturable state. Appl. Environ.
Microbiol. (Submitted.)

Colwell, R.R., Kaper, J., and Joseph, S.W. 1977. Vibrio cholerae, Vibri parahaemolyticus and other vibrios: occurrence
and distribution in Chesapeake Bay. Science 198: 394396.

Colwell, R.R., Brayton, P.R., Grimes, D.J., Roszak, D.R., Huq, S.A., and Palmer, L.M. 1985. Viable but nonculturable
V. cholerae and related pathogens in the environment: implication for release of genetically engineered
microorganisms. Bio/Technology 3: 817820.;

Colwell, R.R., Tamplin, M.L., Brayton, P.R., Gauzens, A.L., Tall, B.D., Harrington, D., Levine, M.M., Hall, S., Huq,
A., and Sack, D.A. 1990. Environmental aspects of V. cholerae in transmission of cholera. In: Advances in

 Page 300

Research on Cholera and Related Diarrhoeas. R.B. Sack and Y. Zinnaka (Ed.), 7th ed., p. 327343. K.T.K.
Scientific Publishers, Tokyo.

Colwell, R.R., Hasan, J.A.K., Huq, A., Loomis, L., Siebling, R.J., Torres, M., Galvez, S., Islam, S., and Bernstein, D.
1992. Development and evaluation of a rapid, simple sensitive monoclonal antibody-based co-agglutination test for
direct detection of V. cholerae 01. FEMS Microbiol. Lett. 97: 215220.

Colwell, R.R., Brayton, P., Herrington, D., Tall, B. Huq, A., and Levine, M.M. 1996. Viable but nonculturable Vibrio
cholerae 01 revert to a cultivable state in the human intestine. World J. Microbiol. Biotechnol. 12: 2831.
Daley, R.J. and Hobbie, J.E. 1975. Direct counts of aquatic bacteria by a modified epi-fluorescent technique. Limnol.
Oceanogr. 20: 875882.

Dawes, E.A. and Ribbons, D.W. 1965. Studies on the endogenous metabolism of Escherichia coli. Biochem. J. 95:
332343.
Field, P., Popovic, I., Wachsmuth, K., and Olsick, O. 1992. Use of the ppolymerase chain reaction for detection of
toxigenic Vibrio cholerae 01 strains from the Latin American cholera epidemic. J. Clin. Microbiol. 30: 21182121.

Finkelstein, R.A. 1973. Cholera. CRC Crit. Rev. Microbiol. 2: 553623.

Floodgate, G.D. 1979. The assessment of marine microbial biomass and activity. In: Proceedings of the Conference
Sponsored by the American Society for Microbiology, R.R. Colwell and J. Foster (Ed.), p. 217252. Maryland Sea Grant
Publication UM-SG-TS-80-03. University of Maryland, College Park, Md.

Fry, J.C. 1990. Direct methods and biomass estimation. Methods Microbiol. 22: 4185.

Grimes, J.D. and Colwell, R.R. 1986. Viability and virulence of Escherichia coli suspended by membrane chamber in
semi-tropical ocean water. FEMS Microbiol. Lett. 34: 161165.

Grimes, D.J., Attwell, R.W., Brayton, P.R., Palmer, L.M., Rollins, D.M., Roszak, D.B., Singleton, F.L., Tamplin, M.L.,
and Colwell, R.R. 1986. The fate of enteric pathogenic bacteria in the estuarine and marine environment. Microbiol.
Sci. 3: 326329.

Hasan, J.A.K. 1995. Development and application of rapid test kits for the detection of V. cholerae in water, food and
clinical samples. Doctoral dissertation. University of Maryland at College Park, Md.
Hasan, J.K.A., Huq, A., and Colwell, R.R. 1991. A method of determination of individual respiring microorganisms
and substrate responsive cells in aquatic microcosms. USA-Japan Cholera Conference Abs., p. 212.

 Page 301

Hasan, J.A.K., Loomis, L., Huq, A., Bernstein, D., Tamplin, M.L., Wier, M., Bodenheimer, H., Siebling, R.J., and
Colwell, R.R. 1994a. Development of a fast and sensitive immunoassay to detect Vibrio cholerae 01 from clinical
samples using SMARTTM technology. Abs. C-118, Abstract 92nd Gen. Meeting Am. Soc. Microbiol., p. 440.

Hasan, J.A.K., Wier, M., Huq, A., Bernstein, D., Loomis, L., and Colwell, R.R. 1994b. Development of a rapid
detection fluorescent labeled monoclonal antibody test kit for the detection of V. cholerae 01 from clinical and
environmental samples. J. Clin. Microbiol. 32: 249252.

Hasan, J.A.K., Chowdhury, M.A.R., Shahabuddin, M., Huq, A., Loomis, L., and Colwell, R.R. 1994c. Cholera toxin
gene polymerase chain reaction for detection of nonculturable Vibrio cholerae 01. World J. Microbiol. Biotechnol. 10:
568571.
Hasan, J.A.K., Huq, A., Nair, G.B., Garg, S., Mukhopadhyay, A.K., Loomis, L., Bernstein, D., and Colwell, R.R. 1995.
Development and testing of monoclonal antibody-based rapid immunodiagnostic test kits for direct detection of Vibrio
cholerae 0139 synonym Bengal. J. Clin. Microbiol. 33: 29352939.
Huq, A., Chowdhury, M.A.R., Felsenstein, A., Colwell, R.R., Rahman, R., and Hossain, K.M.B. 1988. Detection of V.
Cholerae from aquatic environments in Bangladesh. In: Biological Monitoring of Environment Pollution, M. Yasuno
and B.A. Whitton, (Ed.), p. 259264. Tokai University Press, Tokyo.

Huq, A., Colwell, R.R., Rahman, R., Ali, A., Chowdhury, M.A.R., Parveen, S., Sack, D..A., and Russek-Cohen, E.
1990. Detection of V. cholerae 01 in the aquatic environment by fluorescent monoclonal antibody and culture method.
Appl. Environ. Microbiol. 56: 23702373.

Huq, A., Hasan, J.A.K., Losonsky, G., and Colwell, R.R. 1994. Occurrence of toxigenic V. cholerae 01 and V. cholerae
non-01 in professional divers and dive sites in the United States, Ukraine, and Russia. FEMS Microbiol. Letters. 120:
137142.

Huq, A., Colwell, R.R., Chowdhury, M.A.R., Xu, B., Moniruzzaman, S.M., Islam, M.S., Yunus, M., and Albert, M.J.
1995. Coexistence of V. cholerae 01 and 0139 Bengal in plankton in Bangladesh. Lancet 345: 1249.

Ito, T., Himamatsu, K., Oshile, Y., and Yokota, T. 1993. Mutations in the r76T gene are responsible for the Ogawa to
Inaba serotype conversion in V. cholerae 01. Microbiol. Immunol. 37: 281288.

Janinson, M.W. 1937. Relation between plate counts and direct microscopic counts of Escherichia coli during
logarithmic period. J. Bacteriol. 33: 461469.

Kaper, J.B., Lockman, H., Baldini, M.M., and Levine, M.M. 1984. Recombinant live oral cholera vaccine.
Biotechnology 2: 345349.

 Page 302

Keilin, D. 1959. The Leeuwenhoek Lecture. The problem of anabiosis or latent life: history and current concept. Proc.
Royal Soc. 150(b): 149191.

Kepner, R.L. Jr. and Pratt, J.R. 1994. Use of fluorochromes for direct enumeration of total bacteria in environmental
samples. Microbiol. Rev. 58: 602615.

Knaysi, G. 1935. A microscopic method of distinguishing dead from living cells. J. Bacteriol. 30: 193206.

Koch, W.H., Payne, W.L., Wentz, B.A., and Cebula, T.A. 1993. Rapid polymerase chain reaction method for detection
of Vibrio cholerae in foods. Appl. Environ. Microbiol. 59: 556560.
Kogure, K., Simidu, U., and Taga, N. 1979. A tentative direct microscopic method for counting living bacteria. Can. J.
Microbiol. 25: 415420.

Kondo, K., Takade, A., and Amako, K. 1994. Morphology of the viable but non-culturable Vibrio cholerae as
determined by the freeze fixation technique. FEMS Microbiol. Lett. 123: 179184.

Kurath, G. and Morita, Y. 1983. Starvation survival and physiological studies of a marine Pseudomonas spp. Appl.
Environ. Microbiol. 45: 12061211.

Van Leeuwenhoek, A. 1702. On certain animalcules found in the sediment in gutters of the roofs of houses, letter 144.
In: The Select Works of Antony van Leeuwenhoek, Samuel Hoole (lr.), Vol. 2, p. 207213. H. Fry (Publisher), London.

Levine, M.M., Kaper, K.B., Herrington, D.A., Losonsky, G.A., Morris, J.G., Clements, M.L., Black, R.E., Tall, B.D.,
and Hall, R. 1988. Volunteer studies of deletion mutants of Vibrio cholerae 01 prepared by recombinant techniques.
Infect. Immun. 56: 161167.

Litsky, W. 1979. Gut critters are stressed in the environment, more stressed by isolation procedures. In: Aquatic
Microbial Ecology, Proceedings of the Conference Sponsored by the American Society for Microbiology, R.R. Colwell
and J. Foster (Ed.), p. 345347. Maryland Sea Grant Publication UM-SG-TS-80-03. University of Maryland, College
Park, Md.

Losonsky, G.A., Hasan, J.A.K., Huq, A., Kaintuch, S., and Colwell, R.R. 1994. Serum antibody responses of diverse to
waterborne pathogens. J. Clin. Diagnos. Lab. Immun. 1: 182185.

Manning, M., Heuzenroeder, M.W., Yeadon, J., Leavesley, D.I., Reeves, P.R., and Rowley, D. 1986. Molecular cloning
and expression of E. coli K-12 of the O-antigens of the Ogawa and Inaba serotypes of the lipopolysaccharides of V.
cholerae 01 and their potential for vaccine development. Infect. Immun. 53: 272277.
Medema, G.J., Schets, F.M., van de Giessen, A.W., and Haveljar, A.H. 1992.

 Page 303

Lack of colonization of 1 day chicks by viable nonculturable Campylobacter jejuni. J. Appl. Bacteriol. 72: 512516.

Oliver, D.J. 1993. Formation of viable but nonculturable cells. In: Starvation in Bacteria, S. Kjelleberg (Ed.). Plenum
Pres, New York.

Pommepuy, M., Guillaud, J.F., Martin, Y., Dupray, E., Derrien, A., L'yavanc, J., and Cormier, M. 1990. Le devenir des
bacteries en zone litorale. La Mer et les Rejets Urbains. Bandor 1315, June 1990. IFREMER, Actes de Colloques 11:
89100.

Porter, K.G. and Fieg, Y.S. 1980. The use of DAPI for identifying and counting aquatic microflora. Limnol. Oceanogr.
25: 943948.

Postgate, J.R. 1967. Viability measurements and the survival of microbes under minimum stress. Adv. Microb. Physiol.
1: 124.

Postgate, J.R. and Hunter, J.R. 1962. The survival of starved bacteria. J. Gen. Microbiol. 29: 233263.

Preyer, W. 1872. Mittheilung aus dem Vortrag des Herrn Prof. Dr. Preyer. Tagebl. 45 Versamml. Dtsch. Naturs. Aertze.
Leipzig. 18: 46.

Preyer, W. 1891. Über die Anabios. Biol. Zbl. 11.1.

Rahman, I., Shahamat, M., Chjowdhury, M.A.R., and Colwell, R.R. 1996. Potential virulence of viable nonculturable
Shigella dysenteriae type I. Environ. Microbiol. 62: 115120.

Ravel, J., Hill, R.T., and Colwell, R.R. 1994. Isolation of a Vibrio cholerae transposon-mutant with an altered viable
but nonculturable response. FEMS Microbiol. Lett. 120: 5762.

Ray, B. 1979. Methods to detect stressed microorganisms. J. Food Prot. 42: 346355.

Rollins, D.M. and Colwell, R.R. 1986. Viable but nonculturable stage of Campylobacter jejuni and its role in survival
in the natural aquatic environment. Appl. Environ. Microbiol. 52: 531538.

Roszak, D.B. and Colwell, R.R. 1987. Survival strategies of bacteria in the natural environment. Microbiol. Rev. 51:
365379.
Roszak, D.B., Grimes, D.J., and Colwell, R.R. 1984. Viable but nonrecoverable stage of Salmonella enteritidis in
aquatic systems. Can. J. Microbiol. 30: 334338.

Sack, R.B. and Miller, C.E. 1969. Progressive changes of vibrio serotypes in germ-free mice infected with V. cholerae.
J. Bacteriol. 99: 688695.

Shaha, S.K., Shaha, S.N., and Sangal, S.C. 1991. Recovery of injured Campylobacter jejuni cells after animal passage.
Appl. Environ. Microbiol. 57: 33883389.

Shahamat, M., Mai, U., Paszko-Kolva, C., Kessel, M., and Colwell, 1993. Use

 Page 304

of autoradiography to assess viability of Helicobacter pylori in water. Appl. Environ. Microbiol. 59(4): 12311235.

Shimada, T. and Sakazaki, R. 1988. A serogroup of non-01 Vibrio cholerae possessing the Inaba antigen of V. cholerae
01. J. Appl. Bacteriol. 64: 141144.

Shimada, T., Sakezaki, R., and Tobita, K. 1987. Vibrio fluvialis: a new serogroup (19) possessing the Inaba factor
antigen of Vibrio cholerae 01. Japan J. Med. Sc. Biol. 40: 153157.

Shirai, H., Nishibuchi, M., Ramamurthy, T., Bhattacharya, S.K., Pal, S.C., and Takeda, Y. 1991. Polymerase chain
reaction for detection of cholera enterotoxin operon of V. cholerae. J. Clin. Microbiol. 29: 25172521.
Spallazani, L. 1769. Nouvelles recherches sur les decouveries microscopiques et sur la generation des corps organises.
In: Nouvelles recherches physiques et metaphysiques sur la nature et la religion, Vol. 1, p. 25, Needham (lr.). Lacombe,
London, Paris.

Spallazani, T.A. 1803. Tracts on the Natural History of Animals and Vegetables, J.G. Dalyrell (lr.). Creech and
Carstable, Edinburgh.

Stroeher, U.H., Karageorgos, L.E., Morona, R., and Manning, P. 1992. Serotype conversion of Vibrio cholerae 01.
Proc. Natl. Acad. Sci. 89: 25662570.
Strugger, S. 1949. Fluoreszensmikroskopie and Mikrobiologie, p. 151173. M.V.H. Schaper, Hanover.

Valentine, R.C. and Bradfield, J.R.G. 1954. The urea method for bacterial viability counts with electron microscope
and its relation to other viability counting methods. J. Gen. Microbiol. 11: 349357.

Xu, H.S., Roberts, N., Singleton, F.L., Atwell, R.W., Grimes, D.J., and Colwell, R.R. 1982. Survival and viability of
nonculturable Escherichia coli and Vibrio cholerae in the estuarine and marine environment. Microb. Ecol. 8: 313-323.

Zimmerman, R., Iturriaga, R., and Becker-Birek, J. 1978. Simultaneous determination of the total number of aquatic
bacteria and the number therof involved in respiration. Appl. Environ. Microbiol. 36: 926935.

 Page 305

15
Measurement of Microbial Activity by Impedance
P. Rule
bioMérieux Vitek, Inc.
Hazelwood, Missouri
The measurement of physiological components or metabolites may be used to rapidly estimate microbial numbers
before bacterial colonies are visible. Traditionally, the microbial quality of food and raw materials has been monitored
by standard plating methods that require 12 days for bacterial cell counts and 45 days for yeast and mold cell counts.
Commercially available prepared media and automated counting methods have increased productivity but have not
offered substantial time savings because an incubation period is still required for colonies to become visible.
Impedance microbiology utilizes metal electrodes to measure compositional changes occurring within the growth
medium long before colonies are able to reach a visible biomass on standard plating media. As bacteria grow, they
metabolize large, weakly charged molecules (polysaccharides, fats, proteins) and produce small, highly charged
molecules (organic acids, fatty acids, amino acids) as metabolic by-products. As metabolites are produced by bacteria,
the ability of the medium to impede the flow of electricity changes. This change in impedance can be detected by the
instrument. The impedance detection time is inversely proportional to the number of microorganisms

 Page 306

present at initial inoculation and can be standardized to a bacterial plate count. Using this method, a single coliform cell
can be detected in 9 hours; a single yeast cell in 19 hours. Media components and temperature can be used to select for
the growth of specific microorganisms (i.e., coliforms, yeasts and molds, lactic acid bacteria, etc.). Low temperatures
can be used to monitor the growth of psychrotrophs, the predominant spoilage bacteria of dairy products and raw
meats. Reducing agents may also be incorporated in the growth medium to monitor bacterial growth under reduced
oxygen conditions. Additionally, impedance instruments can be helpful in evaluating the growth of bacteria in the
presence of antibacterial agents such as biocides used in the sanitation of production lines or preservatives used in
cosmetic production. In a unique application, the capacitance portion of the impedance signal is used to identify
chemical variations in the product relating to product freshness. Recent literature supports the increased interest in
impedance microbiology and the potential for multiple uses beyond the enumeration of bacteria. Impedance detection
offers a sensitive, rapid, and automated method for enumeration of bacteria, yeasts, and molds.

Introduction
Impedance is the opposition of flow of an alternating electrical current in relationship to the frequency. When two
electrodes are immersed in a

Fig. 15.1
Impedance circuits.

 Page 307

Fig. 15.2
Impedance equations.

conductive medium, they become a series of resistors and capacitors (Fig. 15.1). Impedance (Z) is composed of the
resistive component (R), which is inversely related to the conductance (R = 1/G), and the reactive or capacitance
component, which is a function of the frequency (Fig. 15.2). Impedance microbiology is the measurement of bacterial
growth by monitoring the movement of ions between two electrodes (conductance) or the storage of charge at the
electrode surface (capacitance). Bacterial growth results in the breakdown of larger, relatively uncharged molecules
(proteins, fats, sugars) into smaller, highly charged molecules (amino acids, fatty acids, and lactic acid) creating an
increase in both conductance and capacitance, causing a decrease in impedance.

History

Stewart (1899) first reported the increase in conductance due to microbial contamination in blood. That same year,
Warburg (1899) reported on the bioburden relationship between the electrode polarization and bacterial levels in urine
samples. This pioneering work resulted in two avenues for the electrical measurement of bacterial growth: one
recording changes at the electrode surface and the other in the medium solution. Although described almost 100 years
ago, impedance microbiology as a measurement of microbial growth did not receive its merit due and was only
sporadically reported for the next 60 years. It was Schwan (1963) who described at great length the frequency-
dependent relationship of the conductance and capacitance components of the electrical signal: the lower the frequency,
the larger the reactance of the capacitance. For bacteriological samples, Schwan felt that capacitance changes at the
electrode were minor in comparison to the conductance changes occurring in the medium. This work had a most
profound impact on impedance instrument design in the mid-1970s (Ur

 Page 308

and Brown, 1973; Cady et al., 1978; Hobbs et al., 1977). Although commercial impedance instruments have been
available for more than 15 years, efforts continue to biochemically describe what is being measured at the electrode
surface. Felice et al. (1992) suggest that change during bacterial growth is concentrated at the electrode/electrolyte
interface, perhaps resulting in a third physical component of the impedance growth curve. In the same year, Nobel
discussed the influence of buffer on the capacitance in high ionic media, where the major change in capacitance is due
to protons and not ions near the electrode surface.

Instrumentation

Today there are four commercial impedance instruments: Bactometer (bioMérieux Vitek, Inc.), BacTrac (Sy-Lab, Inc.),
Malthus (Malthus Diagnostics, Inc.), and Rabit (Don Whitley Scientific, Inc.). All four instruments are alike in that
they kinetically monitor bacterial growth in a temperature range of 1055°C and have multiple units allowing anywhere
from 32 to 512 tests to be performed. Whether aluminum block (BacTrac, Rabit), water bath (Malthus), or atmospheric
temperature control (Bactometer), each manufacturer has ensured temperature stability of their individual instrument.
This is because a temperature increase of 1°C results in a 0.9% increase in capacitance and a 1.8% increase in
conductance (Eden and Eden, 1984). Differences in the four instruments can be found in the electrode material,
placement, and number. The Bactometer, BacTrac, and Rabit all use stainless steel electrodes at the bottom of the test
cell, while the Malthus uses platinum electrodes at the top of the test cell. The BacTrac is the only system that uses four
electrodes, which, according to Schwan (1963), enhances the stability of the capacitance signal. The most significant
difference among the instruments is operating frequency. As previously described, the lower the frequency, the greater
the reactance of the capacitance. The Malthus and the Rabit operate at 10,000 Hz, which maximizes the conductance
and minimizes the capacitance. The Bactometer and the BacTrac operate below 2000 Hz, which allows for changes to
be recorded in conductance but at the same time can measure changes in the capacitance. Conductance instruments not
capable of recording change by the capacitance signal have compensated for this disadvantage through the use of low
ionic media, which maximize the conductance change. However, in the case of high salt-selective media or in the
detection of yeasts, where changes in conductance can not be measured directly, microbial growth can be measured by
indirect conductance (Owens et al., 1989; Bolton, 1990). Using indirect conductance, microbial CO2 production is
absorbed by a potassium hydroxide

 Page 309

solution, which is in direct contact with the electrodes. This indirect conductance method has been demonstrated to be
equally as sensitive as direct capacitance in the impedance detection of yeasts (Betts, 1993).

Applications
Just as in traditional microbiology, media development is critical to the impedance growth curve. Sources of nitrogen,
carbohydrate, and buffer must not only be adequate to support growth, but must also result in ionic change with
microbial growth. As bacteria grow, they metabolize larger, weakly charged molecules (polysaccharides, fats, proteins)
and produce smaller, charged metabolic by-products (organic acids, fatty acids, amino acids). The accumulation of
lactic acid and other ionic by-products is detectable in the impedance instrument when bacterial levels reach 106, and
yeasts and mold levels reach 104. A single viable coliform can be detected in 9 hours and a single yeast cell in 19
hours, offering a significant advantage over the traditional 2448 hours required for bacteria and 57 days required for
yeasts and mold. The impedance detection time is inversely proportional to the number of microorganisms present at
initial inoculation (Figure 15.3) and can be calibrated to the standard plate count (Nobel et al., 1991; Deack and
Beuchett, 1993). In addition, impedance microbiology offers simple sample preparation, as serial dilution is often not
required due to the high threshold (106) of impedance.

With impedance microbiology offering a significant time advantage over standard plating methods, it has been readily
adopted by the food and cosmetic industries as a rapid method for the enumeration of bacteria. Although widely used
in the industry, there have not been many formally approved impedance methods. The lack of approved impedance
methods is due, in part, to product diversity and rapid formulation changes made by the food and cosmetic industries.
In addition, internal validation by the individual companies takes less time, is less costly, and has been sufficient for
review by appropriate regulatory agencies. Two formally approved impedance methods are for testing total bacteria
and coliforms in raw milk (American Public Health Standard Methods, 1985) and in eggs products (USDA, 1993). A
conductance method using the Malthus impedance system for the detection of Salmonella in foods (Gibson et al.,
1992) is the first impedance method to receive AOAC approval. After an initial 18-hour preenrichment in buffered
peptone water containing lysine and glucose, samples are subcultured into two selenite cystine-based media, one
containing trimethylamine-N-oxide and dulcitol, and the other a lysine decarboxylase medium. Growth is then
monitored by impedance.

 Page 310

Fig. 15.3
Capacitance detection of yeasts.

In addition to the enumeration of bacteria, yeasts, and molds, impedance can be used for preservative challenge
efficacy, shelf-life prediction (Martin and Hearnsberger, 1993), and hygiene monitoring (Holah et al., 1990; Foster,
1996). In the evaluation of biofilms, where it is difficult to release organisms back into solution, impedance instruments
can measure directly biocide-treated discs previously prepared with biofilms (Dhaliwal et al., 1992; Zhou and King,
1995b). In the case of preservative challenge testing, impedance can quickly screen antimicrobials in products by
monitoring microbial growth rate (Connolly et al., 1994; Tempelaars, 1995; Zhou and King, 1995a). The same basic
impedance premises used in biocide or preservative challenge testing can be used to evaluate the effects of antibacterial
additives in foods as well as screen the bacterial inhibitory effects of temperature, pH, and water activity (Kroyer,
1995).

Although most impedance microbiology has focused on total bacteria and coliform counts, work has been done for the
selective review of specific organism groups. Bishop and White (1985) used impedance at lower temperatures to
monitor psychrotrophic growth rate for predicting shelf life in pasteurized milk. A correlation of r = -0.87 was observed
for the impedance

 Page 311

detection time recorded at 18°C and shelf life of fluid milk. Selective media have been used for the impedance
monitoring of Salmonella (Gibson, 1992), Listeria (Hancock et al., 1993), and other pathogens. Russell et al. (1995)
combined temperature and media selectivity to predict temperature abuse of fresh broiler carcasses by impedance.
Sofia et al. (1995) used media containing Oxyrase® for the impedance detection of anaerobes. In a more novel
approach, Wiese-LeHigh and Marshall (1993) reported on the use of capacitance as a monitor of the freshness of
shrimp. The biodegradation of fish results in a chemical change in the sample that is correlated with a single
capacitance reading at 0.5 hours. The reading is not based on growth, but rather on the chemical changes that have
occurred in the sample due to spoilage.

In addition to these above-mentioned uses of impedance microbiology, Silley and Forsythe (1996) have prepared an in-
depth review of impedance microbiology and its multiple applications.

Conclusion

Impedance microbiology is a rapid alternative method for the enumeration of total bacteria, coliforms, and yeasts and
molds. It offers 14 days' advantage over traditional plating methods and eliminates the need for serial dilution of highly
contaminated samples. Impedance microbiology combines kinetic detection with traditional microbial growth media,
allowing enumeration of microbial flora as well as selectively monitoring for coliforms, lactic acid bacteria, anaerobes,
and pathogens. In adapting standard microbial methods to impedance microbiology, increase of carbohydrate
concentration ensures bacterial production of ionically charged by-products. The automatic real-time measurement of
impedance makes it possible to test the effects of biocides, preservatives, and other antibacterial agents in the
development of new food, pharmaceutical, or cosmetic products. Impedance instruments offer today's microbiologist a
tool for use in development of new products and a way to automate product and raw material testing.

Acknowledgments

I thank Dr. Manfred Schinkinger (SYLABBacTrac), Dr. J. Philip Coombs (Bioscience InternationalRabit), and James
LeRoy (Malthus DiagnosticsMalthus) for supporting literature and information.

 Page 312

References

Betts, R.P. 1993. Rapid electrical methods for the detection and enumeration of food spoilage yeasts. Int. Biodeterior.
Biodegration 32: 1932.
Bishop, J.R. and White, C.H. 1985. Estimation of potential shelf-life of pasteurized fluid milk utilizing bacterial
numbers and metabolites. J. Food Prot. 48(8): 663667.

Bolton, F.J. 1990. An investigation of indirect conductimetry for detection of some food-borne bacteria. J. Appl.
Bacteriol. 69: 655661.

Cady, P., Hadley, D., Martin, J.F., Dufour, S. W., and Kraeger, S.J. 1978. Automated impedance measurements for
rapid screening of milk microbial content. J. Food Prot. 41: 277283.

Connolly, P., Bloomfield, S.F., and Denyer, S.P. 1994. The use of impedance for preservative efficacy testing of
pharmaceuticals and cosmetic products. J. Appl. Bacteriol. 76: 6874.
Deack, T. and Beuchat, L.R. 1993. Comparison of conductimetric and traditional plating techniques for detection
yeasts in fruit juices. J. Appl. Bacteriol. 75: 546550.

Dhaliwal, D.S., Cordier, J.L., and Cox, L.J. 1992. Impedimetric evaluation of the efficiency of disinfectants against
biofilms. Lett. Appl. Microbiol. 15: 217221.

Eden, R. and Eden, G. 1984. Impedance Microbiology. Research Studies Press Ltd., Herts, UK.

Felice, C.J., Velentinuzzi, M.E., Vercellone, M.I., and Madrid, R.E. 1992. Impedance bacteriometry: medium and
interface contributions during bacterial growth. IEEE Trans. Biomed. Eng. 39: 13101313.

Foster, A. 1996. Evaluation of impedance method for monitoring microbiological quality of cleaning systems. J. Am.
Soc. Brewers Chem. 54: 7677.

Gibson, D.M., Coombs, P., and Pimply, D.W. 1992. Automated conductance method for the detection of Salmonella in
foods: collaborative study. J. AOAC Int. 75: 293302.

Hancock, I., Bointon, B.M., and McAthey, P. 1993. Rapid detection of Listeria species by selective impedimetric assay.
Lett. Appl. Microbiol. 16: 311314.

Hobbs, G., Gibson, D.M., Christie, R.H., Jasen, A.C., and Richards, J.C.S. 1977. The use of conductance for
measuring microbiological growth rates. J. Appl. Bacter. 43(3): XIII.

Holah, J.T., Higgs, C., Robinson, S., Worthington, D., and Spenceley, H.

 Page 313

1990. A conductance based surface disinfection test for food hygiene. Lett. Appl. Microbiol. 11: 255259.

Kroyer, G., Fuschik, K., and Geschwandtner, A. 1995. Microbiological analysis of food preservatives by a new
impedance method. Proc. Eur. Food Chem. VIII: 555561.

Martin, J.F. and Hearnsberger, J.O. 1993. Evaluation of impedance microbiology for rapid assessment of shelf-life and
quality of processed channel catfish, Ictalurus punctatus. J. Appl. Aquaculture 3: 353362.

Nobel, P.A. 1992. Formulation of culture media for measuring capacitance changes during bacterial growth. Abstract
Gen. Meet. Am. Soc. Microbiol. 92: 379.
Nobel, P.A., Ashton, E., Davidson, C.A., and Albritton, W.L. 1991. Heterotrophic plate counts of surface water
samples by using impedance methods. Appl. Environ. Microbiol. 57: 32873291.

Owens, J.D., Thomas, D.S., Thompson, P.S., and Timmerman, J.W. 1989. Indirect conductimetry: a novel approach to
the conductimetric enumeration of microbial populations. Lett. Appl. Microbiol. 9: 245249.

Richardson, G.H. (Ed.) 1985. Standard Methods for the Examination of Dairy Products, 15th ed. American Public
Health Association, Washington, D.C., pp. 165170.

Russell, S., Fletcher, D., and Cox, N. 1995. Comparison of media for determining temperature abuse of fresh broiler
carcasses using impedance microbiology. J. Food Prot. 58(10): 11241128.

Schwan, P. 1963. Determination of biological impedances. In: Physical Techniques in Biological Research. W.L.
Nastuk (Ed.), Academic Press, Inc., New York, pp. 323407.

Silley, P. and Forsythe, S. 1996. Impedance microbiologya rapid change for microbiologists. J. Appl. Bacteriol. 80:
233243.

Sofia, T., Martinez, J., and Saulnier, M. 1993. Evaluation d'un nouveau protocole de détection des anaéobies par le
systeme d'impédancemétrie Bactometer. Second bioMérieux Conference.

Stewart, G.N. 1899. The changes produced by the growth of bacteria in the molecular concentration and electrical
conductivity of culture media. J. Exp. Medicine 4: 235243.
Tempelaars, C. 1995. The use of conductance microbiology in the cosmetics and toiletries industry. In: Cosmetics and
Toiletries Manufacture Worldwide 1995, p. 253255.

Ur, A. and Brown, D.F.J. 1973. Detection of bacterial growth and antibiotic sensitivity by monitoring changes in
electrical impedance. J. Int. Res. Commun. 1: 37.

 Page 314

USDA. 1993. Laboratory Method for Egg Products. USDA-ARS, Washington, D.C.

Warburg, E. 1899. Über das Verhalten sogenannter unpolarisierbarer Electroden gegen Wechselstrom. Justus Liebigs
Ann. Chem. 67: 493499.

Wiese-Lehigh P. and Marshal, D. 1993. Determination of shrimp freshness using impedance technology. In: ACS
Symposium Series Food Flavor and Safety: Molecular Analysis and Design. A. M. Spanier, H. Okai, and M. Tamura
(Eds.). American Chemical Society, Washington, D.C., pp. 248259.

Zhou, X. and King, V.M. 1995a. An impedimetric method for rapid screening of cosmetic preservatives. J. Ind.
Microbiol. 15: 103107.

Zhou, X. and King, V.M. 1995b. A rapid Bactometer method for screening of biocides against sulfate-reducing
bacteria. J. Appl. Microbiol. Biotechnol. 43: 336340.

 Page 315

16
Special Techniques for Studying Microbial Biofilms in Food Systems
Edmund A. Zottola
University of Minnesota
St. Paul, Minnesota

Introduction

C. E. Zobell is frequently described as the ''Father" of research on microbial biofilms. His initial paper (Zobell and
Allen, 1935) was one of the first to recognize the role of bacterial attachment in the fouling and corrosion of solid
surfaces submerged in sea water. In a later paper, Zobell (1943) described several methods that could be used to study
bacterial attachment, most notably buried or submerged slide techniques. These procedures are based on the ability of
microbes to attach to the slides, which were stained and viewed using a light microscope. The early research of several
others in this area is described in this chapter. Any person interested in studying biofilms should begin by reading this
chapter.

Critical to the understanding of biofilms and carrying out research in this biological system is an understanding of the
terminology associated with this field. The definitions used here for these terms are as follows:

 Page 316

Biofilms are a confluent growth of microorganisms on an inert surface whereby masses of cells are irreversibly
attached to the surface forming a complex microcommunity of organisms existing in a symbiotic environment.
Biofilms can be composed of only one type of microbe or several. In the context used here, this definition excludes
biofilm formation on living tissues, animal, and plant foods.

Sessile cells are those that are attached to the surface and make up the biofilm, whereas planktonic cells are free-living
and not attached to any surface.

Adherence is related to the physicochemical attractive forces between the cell and the surface. The cell is attracted to
the surface by these forces and is deposited on the surface. Usually the surface is preconditioned; that is, nutrients have
deposited on the surface.
Attachment occurs when the microorganism frequently becomes firmly attached to the substratum or surface by the
production of extracellular material, which literally "cements" it to the surface.

Reversible attachment/Adherence relates to the initial attraction to the substratum whereby the cells are easily removed
from the surface.

Irreversible attachment/Adherence is when the organisms are physically attached to the substratum so that mechanical,
physical, or chemical treatments are needed to dislodge the cells.

Substratum refers to the surface on which attachment/adherence occurs.

In this paper, the terms attachment and adherence and substratum and surface will be used interchangeably.

Biofilm Formation

Several theories have been developed concerning the mechanisms involved in the formation of biofilms on inert
surfaces. I think that the one that best describes the sequence of events is the five-step process developed by Characklis
and Cooksey (1983) and Characklis (1984). This suggested mechanism takes into account the physical, chemical, and
biological phenomena involved. I have modified their five-step process into a six-step process, which better fits what
might occur in a food-processing system. The sequence of events that occurs for biofilm formation includes (1) a
suitable surface, (2) conditioning of the surface or adherence of minerals and organic matter, which may serve as food
for the cells, (3) attraction to the surface of the microbes, (4) initial adherence, which is often reversible,

 Page 317

(5) physical, irreversible adherence involving the development of complex polysaccharides, which firmly fix the cell to
the surface forming a matrix that traps other cells, food materials, and debris, and (6) continued growth of the
organisms forming massive colonies, followed by cell disruption and detachment from the biofilm. Figure 16.1 shows a
diagrammatic representation of the stages in biofilm formation proposed by several investigators (Zottola and Sasahara,
1994). Marshall et al. (1971) suggested that cell attachment and biofilm formation occurred in two stages. The first
involves the attraction of the cells to the surface or substratum. This phase was designated as reversible. If the cells
remain associated with the surface, eventually exopolymeric materials are secreted by the cell, which irreversibly bind
the cell to the surface. A three-step model was proposed by Busscher and Weerkamp (1987) in which it was
hypothesized that the cells attracted to the substratum may be prevented from direct contact with the surface by Van der
Waals forces at distances greater than 50 nm. Electrostatic forces are in effect at shorter distances: 1020 nm. In
addition, physical factors such as flow rate, charge, hydrophobicity, and microtopography of the substratum influence
the degree of association possible between the cell and the substratum. In order for the cell to adhere to the conditioned
surface, it must overcome an interaction barrier described by the DLVO (Dejaguin-Landau-Verwey-Overbeck) theory,
which describes a high-energy repulsion barrier that is influenced by the surface area of the cell involved (Van
Loosdrecht et al., 1989). The bacterial cell must have some mechanism, such as surface protrusions, i.e., pili or flagella,
to assist in overcoming this barrier in order to settle in a stable region where growth and biofilm development will
occur. Figure 16.2 illustrates how this occurs. Figure 16.2a shows initial adherence to a conditioned piece of stainless
steel (SS). At this point the adherence is reversible. Figure 16.2b was taken 3 hr later when the organism has initiated
growth and physical attachment; Fig. 16.2c was taken after 6 hr of growth, and the initiation of biofilm development is
apparent; Fig. 16.2d shows 24-hr growth and what might be termed a biotilm (Zoltai et al., 1981).

One element missing from these theories on biofilm development is time. Time is needed for growth of the organism to
occur. True biofilms, those types that are studied in aquatic systems, on medical devices, and in biofouling, may take
several days, weeks, or months to develop. It is a rare occasion when food contact surfaces are left without some type
of cleaning and sanitizing treatment for more than 24 hr. Thus, it is my opinion that what we deal with in food
processing is not a true or classical biofilm but the initial stages of biofilm development. Consequently, the food
industry should be better prepared to deal with microbes in biofilms than other industries that experience problems
with microbial biofilms.

 Page 318

Fig. 16.1
A diagrammatic representation of the stages involved in the formation of biofilms. The diagram summarizes the various
stages and names used to identify the stages as proposed by Marshall et al. (1971), Characklis and Cooksey (1983), and
Busscher and Werkamp (1987).
(From Zottola and Sasahara, 1994.)

 Page 319

It is critical to the study of biofilms in food systems that one appreciate what is involved in biofilm development. It is
these steps or factors that must be dealt with when trying to develop an understanding of biofilms and the influence that
they may have on both food quality and food safety.

Measurement of Biofilms in Food-Processing Systems: Real and Simulated

A significant amount of research on biofilms or bacterial attachment in food systems has taken a simplistic approach,
including research done in my laboratory; i.e., place a glass slide or piece of stainless steel in a culture of actively
growing microbes and look for adherent microorganisms. Research such as this has given us conflicting reports on the
importance of biofilms in food-processing systems. It is my intent to review and comment on this methodology in this
chapter.

I am not aware of any research on microbial biofilms on food contact surfaces where the biofilms have reached the
dimensions of those described in other areas where microbial biofilms are of concern. There has been a great deal of
speculation on the part of researchers and regulatory agencies that microbial biofilms create major problems in food
processing. I am not convinced that this is true. Part of the error inherent in this speculation is the methodology used by
many investigators studying microbial biofilms related to food processing. Thus, I welcome the opportunity to discuss
here the methods for the measurement of microbial biofilms in food-processing systems.

To successfully study attachment, adherence, and biofilm formation, one has to consider the following questions:

1. Are the conditions that have been established for the study representative of what occurs naturally or in a food-
processing environment?

2. Is the surface or substratum being used as the contact surface representative of what would be found in a food-
processing environment?

3. Are the microbes used of concern in the food industry?

4. Will the time allotted for the study allow for biofilm development?

5. Will the system used to evaluate the biofilm measure only sessile cells and eliminate planktonic cells?

6. Will the methods used differentiate between reversibly attached cells and irreversibly attached cells?

7. Is quantification of the biofilm possible with the methods being used?

 Page 320

 Page 321

Fig. 16.2
(a) Initial adherence by P. fragi on a conditioned SS chip (5000×).
(b) Same organism, 3 hr later; growth and physical attachment has
takenplace (10,000×). (c) 6 hr later, initiation of biofilm development
is apparent (5000×). (d) 24-hr growth showing firmly attached cells
and a biofilm (20,000). Panels of different magnification.
(From Zoltai et al., 1981.)

 Page 322

8. Is quantification necessary to successfully carry out the study? What will quantification add to the findings?

9. Where biofilm control is a major goal, will the procedures used for control reflect what might occur in a food-
processing system?

10. Will the procedures used allow for differentiation of different microbes in the biofilm? Is this needed for the
evaluation?

Unfortunately, as many methods are used to study biofilms as there are researchers studying biofilms. To try to
describe in detail each of these methods would be time consuming and perhaps even boring. I have tried to place most
of these methods into groups and will attempt to discuss them in this manner.

Microscopy Methods

Microscopy is the best method to use for visualizing bacterial attachment, growth, and biofilm formation in food-
processing systems. We have barely begun to exploit the multitude of microscopic techniques available for both the
light and electron microscopes. The technique one chooses is dependent on what aspect of biofilm interaction or
formation one wishes to analyze. It is beyond the scope of this chapter to review and comment on all methodologies
available, so the focus here will be on those techniques that have the most potential for studying biofilms in food
systems.
Electron Microscopy

When microbial interaction within the biofilm matrix is the target of study, electron microscopy techniques are the best
to use. A beam of electrons, usually generated from a tungsten filament or lanthanum boride crystal, is used to
illuminate the sample. The electrons are pulled or accelerated down the column and focused into a coherent beam using
electromagnets. This is done in a vacuum as air molecules interfere with the accelerating electrons. Since the biofilm
sample must also be placed in this vacuum, preparative techniques are designed to stabilize and preserve structure
while eliminating water from the sample. Therefore, sample preparation and resultant interpretations of images are two
of the major problems associated with electron microscopy. It is often difficult to determine what is truly representative
of a sample and what is artifact due to fixation or other sample manipulation. However, improvements in sample
preparation methodologies and in the electron microscopes themselves are continually occurring as new technologies
become available and affordable.

In general, sample preparation involves fixing the sample using either

 Page 323

chemical fixatives, such as glutaraldehyde, paraformaldehyde, and osmium, or cryo-fixation, where the sample is
rapidly frozen to prevent ice crystal damage. Cryo-fixed samples can be viewed in their frozen state on a cold-stage in
the microscope or further processed, as will be discussed later. Once fixation has occurred, most samples need to be
dehydrated in either an ethanol or acetone series. At this point, preparation procedures diverge depending on whether
scanning or transmission electron microscopy is used.

Scanning Electron Microscopy

The scanning electron microscope can be likened to a stereo or dissecting light microscope in that one gets a three-
dimensional view of the sample. Once the sample has been dehydrated, it is then critical-point dried and coated with a
thin layer of metal, usually a gold-palladium alloy, in a vacuum evaporator or sputter coater. Samples need to be metal
coated to make them conductive. The sample is placed into the main chamber of the microscope. Electrons, which are
accelerated down the column, strike the sample surface and are absorbed, deflected, and reflected. They leave the
surface at different energy levels, which are picked up by detectors in the main chamber. Most commonly, secondary
and backscattered electrons are used to visualize the sample. One can also detect x-rays, which give an elemental
analysis (energy-dispersive x-ray detectors, or EDS). True surface analysis involves other instrumentation, which can
detect electrons of much lower energy levels, such as Auger and EELS (electron energy loss spectroscopy). The higher
the accelerating voltage of the electrons coming down the column, the deeper the penetration into the sample and the
lower the true surface resolution. Voltages of 12 and 6 kEv are commonly used, and higher voltages are necessary if x-
ray or backscattered electrons are to be detected. Images are stored digitally or on film.

A typical preparative protocol for scanning electron microscopy (SEM) would involve growing the biofilm on a small
solid support (glass, stainless steel, rubber, etc.) that would fit onto an SEM stub. It would then be fixed, dehydrated,
critical-point dried, and metal-coated. All of these steps influence the resultant image. For example, since many
biofilms contain high levels of bacterial exopolymer, the addition of ruthenium red, a heavy metal known to interact
with neutral and acidic polysaccharides, has been shown to enhance the preservation of these materials (Schwach,
1982).
Low-Voltage Field-Emission Scanning Electron Microscopy

As stated above, the higher the accelerating voltage, the further the electrons penetrate into the sample, thus decreasing
surface resolution. The advent of the field-emission electron gun (FEG) has made it practical to use

 Page 324

lower accelerating voltages (3 kEv) to visualize a surface. In most scanning electron microscopes, as one decreases the
accelerating voltage, the brightness of the image also decreases, making it difficult to resolve surface detail. A field-
emission system is 1000 times brighter than a standard tungsten filament system. The tungsten emitter tip is only nm
wide. The small electron source size coupled with higher-energy electrons pulled from the tip gives increased
brightness, requires lower accelerating voltages down through the column, and increases resolution. This increase in
surface resolution can be orders of magnitude, up to 150,000 times greater than with standard electron microscopes.
Samples still need to be coated, usually in an ion sputterer, and only a thickness of a few nm of chromium is needed.
However, a better vacuum is needed in the microscope itself for FEGs to be effective. It is therefore more difficult to
put food samples into a FE microscope as these samples do not completely dehydrate.

Variable-Pressure Scanning Electron Microscopy

Uncoated samples can be placed directly into a variable-pressure (VP) scanning electron microscope. In many
instances, samples do not even need to be dehydrated or fixed. The main problem with uncoated or nonconductive
samples is charge build-up and electron accumulation on the sample surface. By adjusting the vacuum in the main
chamber of the microscope, these electron charges can be effectively dispersed allowing for visualization of the
surface. Although resolution may suffer slightly, these microscopes offer many advantages for biofilm study.
Magnifications of 500010,000 times are achievable.

Cryo-Scanning Electron Microscopy

The best fixative is the water that naturally surrounds all biological specimens. If cellular structure can be preserved
without further manipulations, then the most accurate representation of the biofilm as it truly exists in the environment
can be achieved. In order to accomplish this, freezing rates must be extremely rapid (104 K s-1) and only extremely thin
layers or the first 500 mm of a surface can be adequately frozen when high-pressure freezing is used (Müller, 1988).
For SEM, rapidly frozen samples are transferred into the microscope using a cryo-transfer unit in which they can be
coated if necessary. Cryo-fixed samples are viewed using a cold stage, which is cooled by liquid nitrogen to -160°C. If
the temperature of this stage is increased slightly, the top layer of water will sublime off, revealing structures deeper
down in the biofilm. The advantage of this technique is that all this occurs within the main chamber of the microscope
and the actual sublimation can be watched and controlled.

Freeze-fracture, a procedure mainly seen with transmission electron mi-

 Page 325

croscopy, can also be used in cryo-SEM, and again, specialized apparatus is required. The frozen sample is struck with
a cold razor blade so that it breaks or fractures along planes of least resistance. Freeze-fracture, although simple to
understand, is technically difficult to accomplish and interpret. This technique is especially suitable to the
characterization of microbial membranes since the fracture plane often proceeds through the hydrophobic interior of
these membranes.

Environmental Scanning Electron Microscopy

This microscope is similar to the variable-pressure microscope in that a true vacuum does not need to be maintained in
the main chamber. It is more useful for nonbiological applications as the resolution is low with hydrated samples. It
may have limited usefulness in studies where the biofilm matrix is rather dry and fine-surface detail is not desired.
Transmission Electron Microscopy

The transmission electron microscope can be likened to a compound light microscope in that thin sections are viewed.
After dehydration, samples are infiltrated with a resin, most often epoxy or methacrylate, and cured so the plastic
hardens. The sample is then sectioned on an ultramicrotome using either a diamond or glass knife to a thickness of
about 90 nm. The sections float off into a water trough along the knife edge and are scooped up onto copper
transmission electron microscopy (TEM) grids. These grids are usually coated with a thin formvar polymer film to help
support the sections. Heavy metal salts of uranium and lead are used to stain the ultra-thin sections. The copper grids
are placed into a holder, which is inserted into the column. Electrons transmitted through the sample can be seen on the
phosphorescent viewing screen. Accelerating voltages between 60 and 100 kEv are typically used in TEM.
Magnifications of 200,000 times are achievable, and it is the technique of choice when microbial cell and surface
structures are studied.

Not all TEM methods require a sample to be fixed and infiltrated with resin. Freeze-fracture (as described above) and
negative staining are two techniques in which biofilm samples can be more directly viewed. Negative staining is the
technique of choice when fimbriae, pili, flagellae, or other small surface structures are to be examined. A small drop of
bacterial or biofilm suspension is placed on a copper TEM grid and wicked off with a piece of filter paper. A drop of a
2% solution of phosphotungstic acid (PTA) in water is placed on the grid and rinsed with deionized water after 5 min.
The grid is dried on filter paper and inserted into the microscope for viewing. Figure 16.3 shows a TEM micrograph of
P. fragi using this method. Note the fine structure of the attachment matrix (Schwach, 1982).

 Page 326

Light Microscopy

Light microscopy, bright field illumination, while perhaps the easiest to use, has some limitations. Magnification and
resolution are not as good as with other instruments now available for use. A transparent substratum, such as glass,
must be used. Cells adhered to the substratum must be stained to visualize and the depth of field that can be visualized
is minimal. A significant amount of the early research on microbial biofilms was accomplished using this instrument.
For the purposes of this chapter, I will acknowledge that it is a useful instrument but not the best to use for studying
biofilms. Other forms of light microscopy, epifluorescence and confocal, are the preferred methods to use for this
research.

Phase-Contrast Microscopy
Phase-contrast microscopy can be used when attempting to follow biofilm development in real time. It works by
converting phase differences between light passing through the specimen and the surrounding medium, which is not
usually discerned by the human eye, into brightness differences that the

Fig. 16.3
TEM micrograph of a negative stain of P. fragi on formvar grid.
(From Schwach, 1982.)

 Page 327

eye can comprehend. The specimen will appear darker than the surrounding medium (Rawlins, 1992). Coupling phase-
contrast microscopy with a computerized imaging system is a useful method for following biofilm development on a
transparent surface.

Epifluorescence Microscopy

Epifluorescence microscopy is a powerful tool to use in quantifying microbial biofilms. It has been used by many
investigators to study biofilm activity in either a simulated or actual food-processing plant. This system utilizes the
property of fluorescent molecules to visualize cells, either directly, as in fluorescent dyes, such as acridine orange, or
indirectly, when coupled in specific tags, such as fluorescent antibodies. These fluorochromes emit light of a given
wavelength when excited by incident light of shorter wavelength. Filters are used to eliminate light of wavelengths
different than that being emitted by the fluorochrome to give specific and sensitive detection of the specimen (Rawlins,
1992).
Confocal Scanning Laser Microscopy
When a laser is used as a light source, specific excitation wavelengths are used. In this way, only probes directly
illuminated by the light beam fluoresce and this beam can be focused at specific depths within the sample. This
epifluorescent technique is enhanced by the confocal scanning unit which allows for examination in the z plane as well
as the x-y planes. Images from the various planes can be stored digitally and then reconstructed into a three-
dimensional view of the sample (Rawlins, 1992).

Nonvisual Methods

There are a number of methods available for the determination of the number of viable cells in a microcolony or a
biofilm. Several are described in the next section. The usual procedure involves obtaining a sample, diluting the sample
in a sterile diluent, adding the diluted sample to a petri dish, adding a nutrient medium, incubating at an appropriate
temperature for a selected time interval, and then counting the number of colonies that are formed. Several of the
procedures are described in some detail in the Compendium of Methods for the Microbiological Examination of Foods
(Vanderzant and Splittstoesser, 1992). The critical points to resolve in using these procedures are:

1. Have all of the sessile cells been removed?

2. Are there planktonic cells in the sample?

 Page 328

3. Will all of the different cells in the mass grow under the cultural conditions used?

4. Will the results reflect what is happening in the biofilm?

Impedance Measurement

As microorganisms metabolize foodstuffs, they convert large molecules into smaller, highly charged ions. This changes
the resistance or impedance in the medium, which in turn changes the rate at which an electrical current will pass
through the medium. The increase in conductivity can be measured, and these changes can be related to numbers of
microbes present and thus to the level of contamination in the area examined (Giese, 1995).
Bioluminescence Measurement

The content of nucleosidephosphate ATP in bacterial cells is fairly constant and can be measured relatively easily. The
amount of ATP present in a sample can be determined by a bioluminescent reaction between luciferin and the enzyme
luciferase. If ATP is present in excess, the maximum intensity of the light given off is proportional to the amount of
ATP present in the sample (Giese, 1995). Since food samples will also contain ATP, this procedure cannot be used to
evaluate food samples or other materials containing biological matter for bacterial numbers, but it can be used to
evaluate the food contact surfaces to determine how much organic material (dirt) remains on the surface. This could
then be related to the potential for microbial biofilms being present on the surface. At the present time, the best use of
this procedure is to determine if cleaning and sanitization has been done correctly. Many food-processing companies
are using this methodology to determine the efficacy of their cleaning and sanitizing practices. It has become a very
useful tool. There are several portable test kits available that are relatively easy to use and give results within a few
minutes of swabbing the surface (Giese, 1995). If calibrated properly it should be possible to use such a system to
identify potential biofilm problem spots in a food system.

Brief Overview of Methods Used to Study Biofilms

Dip and Look

In this procedure a substratum, glass microscopic slide, glass cover slip, or stainless steel chips is put into a culture of
growing bacteria, left there for some time, and then removed and examined microscopically. Light microscopy,
epifluorescence microscopy, and electron microscopy have been the methods selected to ''look at" the adherent
microbes.

Zoltai et al. (1981), in their initial study on microbial adherence, used

 Page 329

round glass cover slips and 8 mm × 8 mm stainless steel chips as the substratum. The glass cover slips were put into
sterile petri dishes and covered with a growing culture of organisms. After 12 hr the cover slips were removed and
fixed for viewing with SEM. The stainless steel chips were put into sterile petri dishes and covered with a milk culture
of the test organism. Fixation for SEM viewing involved fixation in glutaraldehyde in 0.1 M sodium phosphate buffer
for 24 hr at pH 7.0, at 4°C. This was followed by dehydration in a series of graded ethanol solutions and then critical-
point dried using CO2 as the transitional fluid. Samples were attached to specimen studs with silver paint and coated
with a layer of gold palladium by vacuum evaporation. A Phillips 500 SEM operated at 12 kV was used to examine the
samples. Figure 16.4 shows Pseudomonas fragi adhered to a glass cover slip (Zoltai et al., 1981), and Fig. 16.5 shows
the same organism attached to stainless steel (Schwach and Zottola, 1982).

Stanley (1983) studied the irreversible attachment of Pseudomonas aeruginosa to stainless steel using epifluorescence
microscopy. Cells were grown in Peptone yeast extract acetate broth at 2530°C with shaking for 24 hr or until
stationary phase was reached. Cells were harvested by centrifugation and then diluted in 10 mM NaCl. Stainless steel
coupons (2.5 × 7.5 mm)

Fig. 16.4
SEM micrograph of P. fragi on glass.
(From Zoltai et al., 1981.)

 Page 330

Fig. 16.5
SEM micrograph of P. fragi on stainless steel. Note the better resolution of
cell structure compared to Fig. 16.4, when a different and improved fixation
procedure was used. The arrow points to the development of several additional
strands of exopolysaccharide from a single strand.
(From Schwach and Zottola, 1984.)

were added to the cell suspensions, and adherence of the organisms to the stainless steel occurred within 1 min. After
15 min the coupons were removed and rinsed twice with deionized water. The attached cells were enumerated by
staining with acridine orange and viewing in an epifluorescence microscope. This technique was then used to evaluate
the effect of heat, pH, formaldehyde, and the role of the flagella on adherence.

Schwach and Zottola (1982) placed 6 × 6 mm stainless steel (SS) chips on the cut surface of fresh beef for 4 and 24 hr
to allow for microbial adherence. The SS chips and the beef surface were then examined for adherent microorganisms
using SEM. The fixation procedure for SEM described above (Zoltai et al., 1981) was used in this study. In a later
study, Schwach and Zottola (1984) utilized similar techniques to evaluate the effect of sodium hypochlorite on
attachment. The SEM procedure of Zoltai et al. (1981) was modified in this study by the addition of a fixative step
using a

 Page 331

1:1:1 mixture of 0.15% ruthenium red in water to 8% EM grade glutaraldehyde to 0.1 M sodium cacodylate buffer, pH
7.2, for 4 hr. This modification greatly enhanced the visualization of the attachment matrix.

Speers et al. (1984) used SEM to evaluate the surfaces of dairy equipment for attachment of two milkborne
microorganisms. Speers and Gilmour (1985) exposed glass, stainless steel, and rubber surfaces to 15 different cultures
of bacteria in the presence of milk and milk components. These different surfaces were placed into the organism-
containing media and incubated. At the end of the incubation period, the surfaces were removed and stained with
acridine orange. The adhered cells were then quantified by counting the cells in 20 fields on each surface using
epifluorescence microscopy. They concluded that the results were related to the bacterial type, the nature of the
substratum, and the media in which the bacteria were suspended.

In their study on the relationship between growth phase and attachment to stainless steel of P. fragi, Stone and Zottola
(1985b) utilized SS chips in culture flasks containing the organism in milk. The chips were removed from the flasks
and fixed for SEM viewing as described by Schwach and Zottola (1984). Results of this study demonstrated that P.
fragi adhered to the SS within minutes of coming in contact with the surface. The conditions, of growth, temperature,
and stationary or agitated growth did not affect the time of initial attachment or subsequent growth on the surface.

The use of cryo-SEM was investigated by Fraser and Gilmour (1986) in a study involving the use of P. fragi ATCC
4973. They proposed that this method preserved the cells in a more native form than did currently used fixation and
dehydration methods for SEM. They concluded that the dehydration process used in SEM preparations caused collapse
of the polysaccharide on the cell surface giving a false impression that fibrils were present. In 1985, an outbreak of
listeriosis associated with the consumption of soft, white cheese increased interest in the presence of Listeria
monocytogenes in food-processing systems (Zottola and Smith, 1990). This, coupled with the publication of a paper by
Herald and Zottola (1988b) on the attachment of L. monocytogenes to stainless steel, resulted in considerable interest in
the ability of this organism to colonize food-contact surfaces and form biofilms.
Frank and Koffi (1990) developed surface-adherent microcolonies of L. monocytogenes on glass slides and determined
the resistance of these adherent cells to treatment with a sanitizer and heat. Populations of adherent cells before and
after treatments were determined by scraping, swabbing, and rinsing with phosphate buffer and plating with Tryptic
soy agar and incubation at 32°C for 48 hr. They concluded that the adherent cells were more resistant to these
treatments than were nonadherent cells.

Mafu et al. (1990a) used cylinders made from SS, glass, polypropylene,

 Page 332

and rubber in their study on the resistance of adherent L. monocytogenes to sanitizers commonly used in the dairy
industry. The cylinders were sterilized, cooled, and dipped into a broth suspension of the test microbe. The cylinders
were removed, rinsed, dried, and then the efficacy of the sanitizers tested using the AOAC use dilution method
(AOAC, 1984). The results suggested that L. monocytogenes was more susceptible to the action of the sanitizer when
attached to a nonporous surface than a porous surface. They did not visualize the adherent cells in this particular study.

In a second study, Mafu et al. (1990b) used SEM to visualize the cells of L. monocytogenes attached to stainless steel,
glass, polypropylene, and rubber. They also used contact angle measurements to determine the surface energy values of
each of the surfaces. The results indicated that L. monocytogenes was able to attach to all of the test surfaces.

Lee and Frank (1991) developed biofilms of L. monocytogenes on stainless steel coupons by placing the coupons in an
actively growing culture for up to 8 days. These biofilms were then treated with sodium hypochlorite solution or
heated. The data obtained suggested that the longer the cells adhered to the SS surface, the more resistant they were to
these treatments. Rubber and Teflon® were used by Mosteller and Bishop (1993) as the substratum in another study to
evaluate sanitizer efficacy in controlling microbial biofilms. They used P. fluorescens and Yersina enterocolitica in a
milk suspension in which the test materials were placed and the adherence occurred. Adherent and nonadherent cells
were removed from the test surfaces and populations determined using impedance measurement and epifluorescence
microscopy as methods for quantifying results. The data obtained suggested that the different sanitizers used in the
study varied in their ability to control the adherent organisms.

In another study involving L. monocytogenes, Helke et al. (1993) used stainless steel and Buna-N, a material used for
gaskets in the dairy industry, as the substratum. In addition they used Salmonella typhimurium as a second colonizing
organism in the study. They tested the effect of milk and individual milk components, that is, protein and lactose. Chips
of stainless steel and Buna-N were glued to glass microscope slides and put into the attachment menstruum made up of
either milk or the various milk components. The organisms were allowed to attach for up to 1 hr and then removed
from the system, rinsed, and stained with acridine orange. Adherent cells were then enumerated using an
epifluorescence microscope. The results of this study illustrated that attachment involves interaction between the
bacterial cell, the substratum, and the microenvironment. They suggest that experiments dealing with the attachment of
microorganisms in food-processing systems should be carried out under conditions representative of the environments
found in food processing.

 Page 333

Simulated Food-Processing Methods

It is very difficult to simulate food-processing conditions in a laboratory, thus there is a lack of research on true
biofilms in food systems. As many of us are aware, what happens in the food plant is rarely duplicated in a laboratory.
Yet several papers have been published that have tried to duplicate these conditions. One of the first papers published
on bacterial adherence in a simulated food-processing system was that of Stone and Zottola (1985a). In this study, a
stainless steel piping system consisting of 10 ft of pipe, a positive pump, and a surge tank was constructed. The system
was portable so that it could be move into a refrigerated room. The pump was equipped with a timing device so that it
could be turned off and on at various times. Flow rate was 30 L/min. In the pipe line, two sampling points were
available, a T fitting, and an elbow. Stainless steel chips, 6 × 6 mm, were positioned in each of these sample points.
The T fitting was in a position such that it provided a dead space and the elbow was at the point of maximum flow.
Milk inoculated with P. fragi was added to the surge tank and circulated through the system for 30 min every 4 hr for
24, 48, and 96 hr. The 24-hr studies were carried out at 25°C, whereas the 48- and 96-hr studies were done at 4°C.
Chips were removed from the system at various times to determine if adherent cells were present. The system was
cleaned after each experiment using two different cleaning-in-place (CIP) methods. Several SS chips remained in the
system during the cleaning cycles and were removed between cleaning steps to determine the presence of adherent and
viable cells. When the chips were removed they were fixed for SEM viewing as described by Schwach and Zottola
(1984). Duplicate chips were cultured to determine if the organism survived the cleaning and sanitizing process. The
results of the study demonstrated that proper cleaning and sanitizing procedures will kill adherent microorganisms.
When less than optimal cleaning and sanitizing conditions were used, the adherent organisms remained attached and
were viable. The data also suggested that if the intervals between cleaning of equipment exceed 8 hr, then the
probability is quite high that biofilms will form and will not be removed by less than optimal cleaning conditions.

Krysinski et al. (1992) reported on a study in which they tested the effect of cleaners and sanitizers on L.
monocytogenes adhered to food product contact surfaces, stainless steel, and plastic conveyor belts. A three-strain
cocktail of the test organism was used. The food contact surfaces were cut into 1 × 1 cm squares and immersed into an
actively growing culture of the three Listeria. The chips were left in contact with the culture for 24 hr, removed, and
rinsed by shaking the chips for 10 sec in sterile buffer. These chips were then exposed to various cleaners and
sanitizers. The chips were

 Page 334

then fixed for SEM viewing by fixation in 3% glutaraldehyde in 0.1 M phosphate buffer, pH 7.2, for 30 min and rinsed
in phosphate buffer. Dehydration was carried out in a graded ethanol series, followed by critical-point drying in CO2.
The chips were mounted on an aluminum stub, sputter-coated with gold and viewed on with an Amray SEM model
#1600 turbo operated at 30 kv. The number of adherent cells was determined by neutralizing the sanitizer used, then
removing the adherent cells by adding 0.5 g of sterile glass beads, 90210 mm diameter, and vortexing the tubes for 2
min. Enumeration was done by plating on Brain heart infusion agar and incubating at 35°C for 24 hr. Their results
indicated that complete biofilm removal and inactivation was obtained where the surface was first cleaned and then
sanitized.

A similar study on the effect of cleaners and sanitizers on microbial biofilms was done by Humm (1992). Biofilms of
L. monocytogenes, Escherichia

Fig. 16.6
(a) Diagram of the continuous flow slide
chamber (CFSC) system used by Sasahara
and Zottola (1993).
(b) Assembled and unassembled model of
the continuous flow slide chamber. Ruler is
for sizecomparison.

 Page 335

coli, Salmonella enteritidis, and S. typhimurium were developed on 1/2-in. squares of stainless steel by incubating the
chips in broth cultures containing the test organisms at 30°C for 1-, 5-, and 10-day intervals. The culture medium was
replaced every 48 hr and additional cells added. This study used impedance measurement to determine the number of
cells on the SS chips and to determine if the organisms survived the cleaning and sanitizing procedures. The data
obtained using this methodology suggested that increasing intervals between cleaning and sanitizing promoted
increased resistance of the biofilm to cleaning and sanitizing, thus it becomes more difficult for the food processors to
control the organisms. It was demonstrated that proper cleaning and sanitization methods controlled these biofilms.
Sasahara and Zottola (1993) used a continuous flow chamber under a light microscope and phase-contrast microscopy
to follow biofilm formation by L. monoytogenes and P. fragi under constant flow conditions. A diagram of the
apparatus is shown in Fig. 16.6a, and Fig. 16.6b shows a photograph of the continuous flow chamber used. Biofilms of
these two organisms were also formed on glass microscope slides incubated in screw-cap jars containing

 Page 336

the organisms in broth in a shaking water bath. Adherence and biofilm formation by the two organisms singly and
when grown together were followed using both of these methods. To differentiate the two genera in the mixed culture,
a modified Gram stain was used in which the secondary stain, safranin, was replaced with acridine orange and the
biofilm observed using epifluorescence microscopy using a combination of UV and transmitted light. Under these
conditions the gram-positive organisms were purple with an orange outline and the gram-negative organisms were
orange. A comparable study was carried out by Jeong and Frank (1994), in which they used microbes isolated from
meat and dairy-processing facilities and L. monocytogenes grown together to form biofilms. Stainless steel coupons, 2
× 5 cm, cleaned, and sterilized, were incubated with the test microbes in the growth medium. After 5-day incubation at
10°C, the SS coupons were removed from the growth medium, rinsed, and transferred to fresh medium. This was done
again at 9, 13, 17, 21, and 25 days. Adherent microbes on the stainless steel were enumerated by rinsing with sterile
phosphate buffer and then scraping both sides of the SS with a sterile Teflon scraper and rinsed with sterile buffer. This
was done oil each coupon approximately 120 times. By this procedure they showed removal of 97% of the adherent
cells. Numbers of the different organisms were determined using selective plating techniques. The results showed that
L. monocytogenes grew with the other organisms and became a part of the resultant biofilm. In a mixture of the
medium containing the four isolates and L. monocytogenes, Listeria constituted approximately 1% of the total
population.

Kim and Frank (1995) used computerized image analysis to quantify biofilms in another study. Again, L.
monocytogenes was the test organism used. A semi-continuous culture system was used in the study, and a defined,
minimal medium was used to evaluate growth. With this system they were able to speculate on those nutrients that
might affect biofilm formation by this organism.

In a recent study by Dorn and Zottola (1996), a continuous culture system was developed to develop a mathematical
model to predict biofilm formation. The apparatus was similar to the one used by Sasahara and Zottola (1993) (see Fig.
16.6a) except that the microscope was attached to an image-analysis system and real-time images were obtained. This
system was used in the study. The data obtained implied that adherence of P. fragi could be predicted using a
mathematical model.
Holah et al. (1988) used direct epifluorescent microscopy (DEM), direct epifluorescent filter technique (DEFT), and
swabbing followed by plate count enumeration to evaluate microbial numbers on a variety of food contact surfaces.
Biofilms were formed on stainless steel plates, 100 mm × 40 mm × 1 mm. To develop initial attachment, the plates
were covered with

 Page 337

buffered bacterial suspension and left for a selected time period. To allow for biofilm formation or growth, the buffered
suspension was removed from the SS plates and replaced with growth medium. The time for attachment varied
between 10 and 60 min, and the time for biofilm formation (growth) varied between 120 and 240 min. With this
technique they were able to attain populations between 103 and 107 colonies/cm2. The SS plates were removed from the
growth vessel and gently rinsed in distilled water to remove unattached cells. The plates or slides were then enumerated
by all three methods. The data obtained indicated that the DEM technique estimated the populations the best, followed
by DEFT and direct viable count as determined by swabbing and plating. This study significantly identifies direct
epifluorescence microscopy as a method to consider when attempting to quantify biofilm populations.

Holah et al. (1989) carried out a second study in which seven different genera of bacteria isolated from food-processing
environments were used to prepare biofilms on SS 316 with a 2B finish, high-density nylon, polyvinyl chloride belting,
and polypropylene. As in the previous study, biofilm development was carried out in two steps: adherence and growth.
Test surfaces, 13 mm in diameter, cleaned, and sterilized, were added to 5-mL cultures of each of the test organisms.
These attachment cultures were incubated at 25°C for 1 hr. The test substratum was removed, rinsed, and put into
sterile, 12.5-mm internal diameter silicone tubing 150 mm in length. Sterile aerated growth medium was drawn
through the tubing by a peristaltic pump at a flow rate of 0.5 mL/min per tube. After 4 hr the test surfaces were
removed from the tubing, rinsed in distilled water, and drained. The test pieces were "spray-cleaned" by spraying with
distilled water, at ambient temperature, from 20 cm with an impact angle of 90°, through a 0.5-mm nozzle at a flow
rate of 100 mL/min and a nozzle pressure of 28.5 psi. The test surfaces were stained with acridine orange for 2 min,
washed gently in distilled water, and allowed to dry thoroughly. The test materials were examined with an
epifluorescence microscope attached to an image-analysis system. With all the surfaces tested, no difficulty was
apparent in enumerating the adhered organisms. A minimum detection limit of 103 colonies/cm2 was used when 20
microscopic fields were counted. The results showed a wide range in numbers of adherent organisms, from 1.3 × 105 to
4.2 × 107 colonies/cm2, a variety of removal rates during the "spray cleaning," and considerable differences in final
cleanability. The data suggested that type of microbial biofilm on the test materials had a greater effect on removal of
adherent organisms than the surface types that were studied.

These researchers (Holah et al., 1989) also reported on a series of plant trials where biofilm development was studied.
Stainless steel plates (100 × 40

 Page 338

mm) were placed in strategic areas in several food-processing plants. The plates were removed after varying lengths of
contact time and stained with acridine orange. These plates were enumerated in the same manner as given above. The
results of the plant studies demonstrated that this technique could be used to rapidly evaluate product contact surfaces
(<15 min) for thorough cleaning and sanitizing practices.

Scrape and Look

In this technique, biofilms are grown on an inert surface such as glass or stainless steel for a specific time period, the
adherent organisms are scraped from the surface with a device and then enumerated either by doing an enumeration
such as total viable numbers or using a microscopy technique. Wirtanen and Mattila-Sandholm (1992a,b) used this
technique in an extensive study on biofilm formation on stainless steel. In the first study (Wirtanen and Mattila-
Sandholm, 1992a) six different genera were used, and autoclaved meat soup or minced meat medium were used as the
suspending media. The biofilms were formed by placing the stainless steel in the inoculated suspending media for 24
or 48 hr depending on the organism tested. After that time the media were changed to allow only the adherent
organisms to grow in the replaced fresh media. An additional 96-hr biofilm growth period was used. The SS was
removed from the media, rinsed in sterile saline, and treated with various sanitizers to determine resistance of the
biofilm cells to the sanitizers. Biofilms were also formed on SS studs in a Robbins device. [The Robbins device is a
multiport sampling device with evenly spaced sampling ports that can be used to follow biofilm development under a
variety of conditions (Ladd and Costerton, 1990).] This system allows circulation of the test menstrum through a glass
cylinder with the SS studs level with the inside of the glass. Biofilms on the SS studs were treated in the same manner
as mentioned above. The adherent cells were scraped from the test surfaces, suspended in an inactivation solution to
stop the activity of the chemicals, and then diluted in saline and populations of the cells determined using TSA plates
and incubation at 48 hr at 30 or 37°C depending on the test organism. The data obtained implied that the adherent cells
were somewhat more resistant to the action of the sanitizers than were the planktonic cells. This paper presented a
rather elaborate study on biofilms, but the results were inconclusive as to the differences in resistance to the action of
chemical sanitizers of sessile and planktonic cells.
In the second study, Wirtanen and Mattila-Sandholm (1992b) utilized the same organisms and techniques to determine
the influence of age of the culture (biofilm) on resistance to chemical sanitizers. The data obtained indicated that at
least 24 hr of growth is necessary for the development of

 Page 339

biofilms. The organisms were more resistant to the sanitizers when grown in milk compared to the same organisms
grown in a meat medium. They were not able to relate resistance to sanitizers with age of the sessile cells as biofilms
144 hr old were no more resistant to the sanitizers than the same organisms in a 24-hr biofilm. In a third study,
Wirtanen and Matilla-Sandholm (1993) used the same organisms and methods except that the biofilms were allowed to
develop for 10 days on stainless steel coupons (2.5 cm × 7.5 cm) at 25°C, in a medium referred to as autoclaved slime
medium. The SS coupons were scraped and the number of cells in the residue determined by plate count methods.
Duplicate coupons were removed from the slime broth, rinsed with sterile water, and stained with acridine orange, and
the number of adherent cells counted by using epifluorescence microscopy. The area occupied by the 10-day biofilms
was determined using an image-analysis system. In this study four microbes were used: Bacillus subtilis, L.
monocytogenes, Pediococcus pentosaceus, and P. fragi. L. monocytogenes showed a lower area of coverage in the
biofilm mass than the other three organisms. It is an interesting idea to relate the area of the biofilm mass to the total
number of adherent cells on the surface of the SS, which should be considered by others investigating biofilms on
food-processing systems.

In-Plant Studies
Research on microbial biofilms carried out in operating food-processing systems probably gives us a better idea of
what really occurs in a real-time situation than does simulated processing in a laboratory. Unfortunately, many food
processors are reluctant to allow such studies in their plants. Nevertheless, I am aware of at least three that have been
done, and these will be briefly mentioned here.

Frank et al. (1990) reported on a study that involved 15 dairy-processing plants. It was apparent that biofilm analysis
was not the primary focus of this study, but the results imply that biofilms may have been detected in the study. A total
of 409 swabs of environmental surfaces were examined. The swabs were analyzed a number of different ways to
determine the presence of a variety of organism types. A significant number of different genera of bacteria were found,
most notably, L. monocytogenes. Forty-seven of the samples were positive for this organism. The intent of the research
was to find an indicator organism that might predict the likelihood of Listeria presence. The authors suggested from the
data obtained that the presence of staphylocci in the sample was the best indicator of the presence of Listeria. While
not a study involving biofilms per se, the techniques that were used are similar to those that would be used in a study of
biofilms in the same environments. Nelson (1990) reported the results of a survey done in dairy

 Page 340

plants on the incidence of Listeria in the environment. Like the Frank et al. (1990) study biofilms were not the primary
thrust of the survey but were implied in the description of the results. The survey involved 62 dairy plants and reported
the results of 8800 environmental samples. The samples were taken specifically to detect environmental Listeria
contamination. The overall incidence of Listeria contamination was 10%, whereas incidence in the various types of
dairy-processing systems ranged from 12% in fluid milk processing to 1% in dry processing systems. It would be
interesting to repeat these two studies looking for biofilm formation and the presence of other nonpathogenic
organisms that might be primary biofilm formers.

Hood and Zottola (1996) worked with four meat-processing plants to evaluate biofilm formation and presence. The
four plants included a slaughter operation, a hamburger manufacturer, and producers of beef jerky and sausage.
Stainless steel chips were attached to non-food-contact surfaces directly adjacent to areas where food products were
handled. The chips were located in areas where the probability of their becoming dislodged and penetrating into the
food was minimized. Cast iron chips were suspended in floor drains using fishing line in areas where there was water
flowing and in areas where flowing water was at a minimum. The chips were removed at predetermined intervals
ranging from 4 hr to several days and rinsed with sterile 0.1% peptone water. After rinsing, one chip was stained with
acridine orange using the procedure as described by Stanley (1983) and enumerated by counting 10 microscopic fields
at 1000× (field area = 2.01 × 10-4 cm2) using an epifluorescence microscope. Microbes on a second chip were
evaluated by placing the chip in Tryptic soy broth (Difco, Detroit) and mixing for 10 sec, 1 mL of the broth was
transferred to two other tubes and the tubes incubated at 37, 21, and 13°C for 24 hr, 48 hr, and 7 days. After incubation,
the broth cultures were streaked for isolation and identification. To test for the presence of Listeria and Salmonella,
chips were also put into lactose broth to examine for salmonella and UMV broth to look for Listeria. The isolation and
identification of these organisms was carried out as described by Vanderzant and Splittstoesser (1992).

The results of this study were inconclusive, as the plant employees were aware of what was going on. It became
apparent during the study that normal practices were not being followed and special care was being taken to assure that
the areas where the chips had been placed were cleaned (including the drains). But the results did confirm that there are
biofilms in meat-processing plants in these selected locations. The most commonly found organisms in this study were
Pseudomonas and Klebsiella. The finding of these organisms implies that biofilms could develop as both of these
microbes are capable of producing copious amounts of extracellular mate-

 Page 341

rial that could trap other organisms and develop into a mixed species biofilm.

Other Studies on Biofilms

Humm (1992) used impedance measurement to determine the numbers of bacteria associated with the biofilms
developed in her study. The SS chips used in the study were of the same dimensions as the reaction vessels of the
instrument used to measure impedance; the chips were simply placed in the bottom of the vessel, growth medium
added, and changes in conductivity monitored. This method seems to me to be a better method for quantifying cells in
a biofilm than a scrape-and-look procedure. Conductivity changes were used by Holah et al. (1990) to determine the
effect of disinfectants (sanitizers) on three different organismsPseudomonas, Staphylococcus, and Proteusin biofilms
developed on stainless steel. The authors suggest that the conductivity procedure to determine biofilm populations
before and after treatment with the sanitizer allows assessment of the cell viability while the cells are still adhered to
the surface. In this way the physiological conditions of the bacteria are maintained and there are no problems
associated with trying to remove the cells from the surface.

In a similar study, Dhaliwal et al. (1992) evaluated the efficiency of several disinfectants (sanitizers) on biofilms
formed on polyvinyl chloride, Teflon, Plexiglass, wood, rubber, and stainless steel. Impedance measurement was the
method used to determine the numbers of organisms in the biofilms before and after treatment with the sanitizers. The
results obtained were similar to those reported by Holah et al. (1990).

An in situ procedure for determining numbers of cells in a biofilm was described by Yu et al. (1993). This procedure
was used without disturbing the integrity of the interfacial community. Results were obtained in 4 hr. Biofilms were
formed on SS coupons and the coupons placed in petri dishes containing phosphate buffered water, 0.3% casamino
acids, 0.03% yeast extract, and different concentrations of nalidixic acid. The coupons were incubated for 4 hr at 35°C.
After incubation the coupons were withdrawn, fixed with formalin, and stained for 2 min in a 0.02% solution of
acridine orange. The coupons were air-dried, fixed on glass slides, and, viewed using epifluorescence microscopy. The
results were compared with viable numbers obtained using a drop plate procedure. The data were statistically analyzed
and suggested that the two methods gave comparable results.

Yu et al. (1994) described a novel method for quantifying biofilms by using cryosectioning and epifluorensce
microscopy. Biofilms consisting of cells of K. pneumoniae and P. aeruginosa were developed on SS coupons in a

 Page 342

continuous flow reactor. The resultant biofilms were frozen and removed from the SS by gently flexing the surface.
They were then examined by light or epifluorescence microscopy. The procedure preserved biofilm structure features,
which visualized an irregular surface, water channels, local protrusions up to 500 mm thick, and a well-defined
substratum surface. This is an interesting procedure which others might find useful in studying biofilms.

Summary and Conclusions

If one challenges each of the methods described above with the questions posed at the beginning of this paper, it soon
becomes apparent that no one method will respond positively to these concerns. To successfully study biofilms in food
systems requires the use of a number of these procedures. The ideal system for carrying out research on biofilms
should include the use of one or more of the newer SEM methods, and if quantification is required then impedance
measurement should be the method of choice. Epifluorescent microscopy should be used for real-time evaluations of
biofilm formation. There are many challenges awaiting those of us interested in biofilm formation in food systems. We
have only scratched the surface as to what is to be learned about the microcolonies that inhabit the nooks and crannies
of our food-processing systems. To resolve these potential problems will require creativity and perseverance, but
eventually problems with biofilms will be a problem of the past. We have to overcome some of the traditional thinking
in food microbiology and develop concepts and ideas consistent with what is now known and what can be learned with
the new techniques available.

Food microbiologists seem to be obsessed with the need to know numbers of microbes present, when in many
situations actual numbers are insignificant. What is important is whether the microorganisms are present or absent.
Hood and Zottola (1996) suggested that we be more concerned with the biotransfer potential, that is, the potential for
those organisms associated with the biofilm to contribute to the contamination of food. This term could be used to
describe those areas harboring adherent microbes that have not yet developed into a biofilm but that could be dislodged
to contaminate the food products passing over the surface.

I would like to close with the Harvard Law: ''Under the most rigorously controlled conditions of pressure, temperature,
volume, humidity and other variables, the organism will do as it damn well pleases" (Bloch, 1979). We should all keep
this is mind as we try to unravel the mysteries of microbial biofilms.

 Page 343

References

AOAC. 1984. Official Methods of Analysis. Association of Official Analytical Chemists, Inc., Arlington, VA.
Bloch, A. 1979. Murphy's Law and Other Reasons Why Things Go Wrong. Price, Stern, Sloan Publishers, Inc., Los
Angeles.

Busscher, H.J. and Weerkamp, A.H. 1987. Specific and nonspecific interactions in bacterial adhesion to solid substrata.
FEMS Microbiol. Rev. 46: 165173.

Characklis, W.G. 1984. Biofilm Development: A Process Analysis. Springer-Verlag, New York.

Characklis, W.G. and Cooksey, K.E. 1983. Biofilms and microbial fouling. Adv. Appl. Microbiol. 29: 93137.

Dhaliwal, D.S., Cordier, J.L., and Cox, L.J. 1992. Impedimetric evaluation of the efficiency of disinfectants against
biofilms. Lett. Appl. Microbiol. 15: 217221.
Dorn, E.M. and Zottola, E.A. 1996. Predicting microbial biofilm accumulation: a mathematical model. J. Food Proc.
Eng. (In press.) Frank, J.F. and Koffi, R.A. 1990. Surface-adherent growth of Listeria monocytogenes is associated
with increased resistance to sanitizers and heat. J. Food Prot. 53: 550554.

Frank, J.F., Gillett, R.A.N., and Ware, G.O. 1990. Association of Listeria spp. contamination in the dairy processing
plant environment with the presence of staphylococci. J. Food Prot. 53: 928932.
Frazer, T.W. and Gilmour, A. 1986. Scanning electron microscopy preparation methods: their influence on the
morphology and fibril formation in Pseudomonas fragi (ATCC 4973). J. Appl. Microbiol. 60: 527533.

Giese, J. 1995. Rapid microbiological testing kits and instruments. Food Technol. 49(7): 6470.

Helke, D.M., Sommers, E.B., and Wong, A.C.L. 1993. Attachment of Listeria monocytogenes and Salmonella
typhimurium to stainless steel and Buna-N in the presence of milk and individual components. J. Food Prot. 56:
479484.

Herald, P.J. and Zottola, E.A. 1988a. The use of transmission electron microscopy to study the composition of
Pseudomonas fragi attachment material. Food Microstruct. 7: 5357.

Herald, P.J. and Zottola, E.A. 1988b. Attachment of Listeria monocytogenes to stainless steel surfaces at various
temperatures and pH values. J. Food Sci. 53(3): 15491552, 1562.

 Page 344

Holah, J.T., Betts, R.P., and Thorpe, R.H. 1988. The use of direct epifluorescent microscopy (DEM) and the direct
epifluorescent filter (DEFT) to assess microbial populations on food contact surfaces. J. Appl. Microbiol. 65: 215221.

Holah, J.T., Betts, R.P., and Thorpe, R.H. 1989. The use of epifluorescence microscopy to determine surface hygiene.
Int. Biodeter. 25: 147153.

Holah, J.T., Higgs, C., Robinson, S., Worthington, D., and Spenceley, H. 1990. A conductance-based surface
disinfection test for food hygiene. Lett. Appl. Microbiol. 11: 255259.

Hood, S.K. and Zottola, E.A. 1996. The isolation and identification of adherent gram-negative microorganisms from
four meat processing facilities. Dairy Food Environ. Sanit. (In press.)

Humm, B.J. 1992. A research update on the effects of cleaners and sanitizers on food processing biofilms. Food Prot.
Rep. 8(2): 5.

Jeong, D.K. and Frank, J.F. 1994. Growth of Listeria monocytogenes at 10°C in biofilms with microorganisms isolated
from meat and dairy processing environments. J. Food Prot. 57: 576586.

Kim, K.Y. and Frank, J.F. 1995. Effect of nutrients on biofilm formation by Listeria monocytogenes on stainless steel.
J. Food Prot. 58: 2428.
Krysinski, E.P., Brown, L.J., and Marchisello, T.J. 1992. Effect of cleaners and sanitizers on Listeria monocytogenes
attached to product contact surfaces. J. Food Prot. 55(4): 246251.

Ladd, T.I. and Costerton, J.W. 1990. Methods for studying biofilm bacteria. Methods Microbiol. 22: 285307.

Lee, S.H. and Frank, J.F. 1991. Inactivation of surface-adherent Listeria monocytogenes hypochlorite and heat. J. Food
Prot. 54: 46.

Mafu, A.A., Roy, D., Goulet, J., Savoie, L., and Roy, R. 1990a. Efficiency of sanitizing agents for destroying Listeria
monocytogenes on contaminated surfaces. J. Dairy Sci. 73: 34283432.

Mafu, A.A., Roy, D., Goulet, J., and Magny, P. 1990b. Attachment of Listeria monocytogenes to stainless steel, glass,
polypropylene and rubber surfaces after short contact time. J. Food Prot. 53: 742746.
Marshall, K.C., Stout, R., and Mitchell, R. 1971. Mechanism of initial events in the sorption of marine bacteria to
surfaces. J. Gen. Microbiol. 68: 337348.

Mosteller, T.M. and Bishop, J.R. 1993. Sanitizer efficacy against attached bacteria in a milk biofilm. J. Food Prot. 56:
3441.

Müller, M. 1988. Cryopreparation of microorganisms for electron microscopy. Methods Microbiol. 20: 127.

 Page 345

Nelson, J.H. 1990. Where are Listeria likely to be found in dairy plants? Dairy Food Environ. Sanit. 10(6): 344345.

Rawlins, D.J. 1992. Light Microscopy. Alden Press Ltd., Oxford, U.K.

Sasahara, K.C. and Zottola, E.A. 1993. Biofilm formation by Listeria monocytogenes using a primary colonizing
microorganism in flowing systems. J. Food Prot. 56: 10221028.

Schwach, T.S. 1982. Use of electron microscopy to study bacterial attachment to food and food contact surfaces. M.S.
thesis, Univ. of Minnesota, St. Paul.
Schwach, T.S. 1996. Unpublished data. Dept. of Food Science & Nutrition, Univ. of Minnesota, St. Paul.

Schwach, T.S. and Zottola, E.A. 1982. Use of scanning electron microscopy to demonstrate microbial attachment to
beef and beef contact surfaces. J. Food Sci. 47(5): 14011405.

Schwach, T.S. and Zottola, E.A. 1984. Scanning electron microscopic study on some effects of sodium hypochlorite on
attachment of bacteria to stainless steel. J. Food Prot. 47(10): 756759.

Speers, J.G.S. and Gilmour, A. 1985. The influence of milk and milk components on the attachment of bacteria to farm
dairy equipment surfaces. J. Appl. Microbiol. 59: 325332.

Speers, J.G.S., Gilmour, A., Fraser, T.W., and McCall, R.D. 1984. Scanning electron microscopy of dairy equipment
surfaces contaminated by two milk-borne microorganisms. J. Appl. Bacteriol. 57: 139145.

Stanley, P.M. 1983. Factors affecting the irreversible attachment of Pseudomonas aeruginosa to stainless steel. Can. J.
Microbiol. 29: 14931499.

Stone, L.S. and Zottola, E.A. 1985a. Effect of cleaning and sanitizing on the attachment of Pseudomonas fragi to
stainless steel. J. Food Sci. 50 (4): 951956.

Stone, L.S. and Zottola, E.A. 1985b. Relationship between the growth phase of Pseudomonas fragi and its attachment
to stainless steel. J. Food Sci. 59(4): 957960.

Vanderzant, C. and Splittstoesser, D.F. (Ed.). 1992. Compendium of Methods for the Microbiological Examination of
Foods. American Public Health Assoc., Washington, D.C.
Van Loosdrecht, M.C.M.V., Lyklema, J., Norde, W., and Zehnder, A.J.B. 1989. Bacterial adhesion: a physicochemical
approach. Microbiol. Econ. 17: 115.

Wirtanen, G. and Mattila-Sandholm, T. 1992a. Removal of foodborne biofilmscomparison of surface and suspension
tests. Part I. Lebensm. Wiss. Technol. 25: 4349.

 Page 346

Wirtanen, G. and Mattila-Sandholm, T. 1992b. Effect of the growth phase of foodborne biofilms on their resistance to a
chlorine sanitizer. Part II. Lebensm. Wiss. Technol. 25: 5054.

Wirtanen, G. and Mattila-Sandholm, T. 1993. Epifluorescence image analysis and cultivation of foodborne bacteria
grown on stainless steel surfaces. J. Food Prot. 56: 678683.

Yu, F.P., Pyle, B.H., and McFetters, G.A. 1993. A direct viable count method for the enumeration of attached bacteria
and assessment of biofilm disinfection. Microbiol. Methods 17: 167180.

Yu, F.P., Callis, G.M., Stewart, P.S., Griebe, T., and McFetters, G.A. 1994. Cryosectioning of biofilms for microscopic
examination. Biofouling 8: 8591.

Zobell, C.E. 1943. The effect of solid surfaces upon bacterial activity. J. Bacteriol. 46: 3956.

Zobell, C.E. and Allen, E.C. 1935. Attachment of marine bacteria to submerged surfaces. Proc. Soc. Exp. Biol. Med.
30: 14091411.

Zoltai, P.T., Zottola, E.A., and McKay, L.L. 1981. Scanning electron microscopy of microbial attachment to milk
contact surfaces. J. Food Prot. 44(3): 204208.

Zottola, E.A. and Smith, L.B. 1990. The microbiology of foodborne disease outbreaks: an update. J. Food Safety 11:
1329.

Zottola, E.A. and Sasahara, K.C. 1994. Microbial biofilms in the food processing industryShould they be a concern?
Int. J. Food Microbiol. 23: 125148.

 Page 347

INDEX

AccuProbe, for detection of Listeria, 139

Acridine orange, 47, 295

Adherence:

definition, 316

of bacteria to surfaces, 316319

Aeromonas:

gene probes for detection, 128129

identification of, 106

Agar plates, 6

Agglutination, latex, 8284

Alamar, 104105

Amplicons, 192
Anabiosis, 291

Anaerobes, PCR for, 185

Antibodies, 48, 7778

detection systems, 78

to detect pathogens, 78

Antigens, 48, 7778

AOAC, 1
Arbitrarily primed PCR (AP-PCR), 251252

ATP, 9

instruments and kits, 910

to estimate biomass, 9

Attachment:

definition, 316

of bacteria to surfaces, 271272, 316319

Automation:

Alamar, 104105

Biolog, 108109

for detection of Micrococcus, 109

comparison of systems, 9293

 Page 348

microbial identification systems, 91110

MIDI, 109110

for detection of Fusobacterium, 110

Sensititre, 14, 96, 98, 103104

to identify microbes, 91110

Vitek

accuracy, 95
comparison with other methods, 9496, 98

for detection of respiratory bacteria, 95

for detection of urinary bacteria, 95

GNI, 9495, 100

GPI, 100102

uses, 94
WalkAway, 96, 98, 105108

Bacillus cereus, gene probes for detection, 129130

Bacteria, adaptations, 290

Bacteriocin, 187

Bactometer, 308

to measure impedance, 1011

uses, 11

BacTrac, 308

Base pairs, noncanonical, 211212

BAX system, 1516, 192193

Beef, detection of E. coli in, 4950, 6163

Biochemistry, of bioluminescent bacteria, 266

Biofilm:

DEM, 336337

evaluation by impedance, 310

for Listeria, 331336

formation, 316319

history, 315

image analysis, 336

measurement, 319322

methods to visualize, microscopy, 322327

models, 333338
 nonvisual methods of determination, 327328

bioluminescence, 328

impedance, 328

review of methods studying, 328341

dip and look, 328332

in-plant studies, 339341

scrape and look, 338

simulated, 333338

terminology, 315316

Biolog, 14, 108109

for detection of Micrococcus, 109

Bioluminescence, 265282

autoinducer, 281

biochemistry, 266

biosensor for rpoS, 274

comparison to plate count, 270

competitive microflora, 272273

correlation to plate count, 270

engineering bacteriophages, 281282

for monitoring injury, 267268

freeze, 268269
heat, 269270

genetics, 266
instrumentation, 267

surface contamination, 271272

to study biofilms, 328

Biomass:
limits of detection, 9

methods for estimating, 914

methods of measurement:
capacitance, 10

conductance, 10
impedance, 10

BioRad Bryte HS, 64

Biosensors, 8788

definition, 6970
fiber optic fluorescence sensors, 71

optical, 6976
 fiberoptic fluorescence probes, 75

grating-based, 7172

 Page 349

integrated optic interferometer, 7375

resonant mirror, 7273

surface plasmon resonance (SPR), 7172

Biotechnology Information for Food Safety (BIFS), 235

Biotin, 125

Biotransfer potential, 342

Biphasic partitioning, to separate bacteria, 38

BLAST, 243244

Burkholderia cepacia, 95

Campylobacter:
colorimetric hybridization assay for, 214

gene probes for detection, 130131

restriction digestion, 251252

Capacitance, for measuring biomass, 10

Capture element, 30

CASBA, 7
Catalase test: 1213

for psychrotrophs, 12

to estimate biomass, 12

uses, 12

Cells:

nonculturable, viable, 289298

characteristics, 292

history, 291
methods of detection, 295297

morphology, 293

problems in determining metabolism, 291294

terminology, 290
virulence, 293

stress, effect on PCR, 188

viable, sorting of, 6263

Centrifugation, to separate bacteria, 3637

Chemicals, interfering with PCR, 189191

Chemiluminescence, for detection of Salmonella, 142

Cholera, 144147, 289, 292297

 commercial tests, 294

direct viable count (DVC), 295296

DFA-DVC, 295296

isolation, 293

Oag antigenic property, 297298

outbreaks, 294295

seroconversion, 297298

serotypes, 294

survival, 292

CholeraSMART, 296

Chromatography, column, to separate bacteria, 37

Citrobacter, RNA sequence, 210211

Cleanliness, catalase testing of, 1213

Clostridium:

botulinum, gene probes for detection, 131132

perfringens, gene probes for detection, 132133

Colony counter:

CASBA, 7

Countermat, 7
Protos, 7

Colorimeter, tri-stimulus reflectance, to measure dye pigmentation change, 1112

Colorimetry, 126

for detection of Salmonella, 142

Commercial kits, 14

applications, 15
ELISA, comparison to conventional methods, 16

Concentration:
single stage, 30

two stage, 3031

Conductance:
for measuring biomass, 10

Malthus system, 11
Conductivity, to study biofilms, 341

Confocal scanning laser microscopy, 327

Countermat, 7

 Page 350

Cryptobiosis, 291

Cystic fibrosis, bacterial identification in, 95

Dairy equipment, surface bacterial attachment, 331

DAPI (4'6-diamidino-2-phenyl indole), 295

Databases:

Biotechnology Information for Food Safety (BIFS), 235

DNA Data Bank of Japan, 230

e-mail access, 235236

E. coli Data Collection, 235

E. coli Genome Center, 235

EMBL (European Molecular Biology Laboratory), 230

GenBank, 121, 230233

gene sequence, 229247

GenPept, 232233

importance, 229230, 233

Medline, 241
on-line access, 235242

PIR, 232233

precautions, 244245

programs

BLAST, 243244

FastA, 243244

Ribosomal Database Project (RDP), 233

SwissProt, 232233
uses, 242244

WWW access, 236242

DEFT (direct epifluorescent filter technique), 8, 4748

to detect E. coli, 5051, 5962

limits of detection, 60

to study biofilms, 336337

DEM (direct epifluorescent microscopy)

to study biofilms, 336337

Detection:

efficiency, 2931

methods, 2931
 of gene probes, 122

of pathogens in foods, 2829

Dielectrophoresis, to separate bacteria, 38

Digoxigenin, 125, 144

Direct fluorescent antibody (DFA), for Vibrio cholerae, 295

Direct viable count (DVC), 295296

DLP (direct label probe), 218223

for Salmonella

comparison in foods, 222223

inclusivity, 220

sensitivity, 222

specificity, 221

principles, 218219

DNA:

ail region, 147148

Data Bank of Japan, 230

fingerprinting, 249261

hybridization, 117119

for detection of foodborne bacteria, 15

inv region, 147148

labeling probes, 124

nick translation, 123124

PCR for sequencing, 193

probes, 207223

pulsed-field gel electrophoresis (PFGE), 256

random primed labeling, 123

DNA DataBank of Japan, 230

Dysentery, 133134

E-mail, access to databases, 235236, 238

Ecology, microbial, limitations, 290291

Efficiency:
of detection of pathogens, 2931

of separation of pathogens, 3032

of swabs, 33

EHEC (enterohemorrhagic E. coli), 134137

EIEC (enteroinvasive E. coli), 133134

Electrical detachment, to separate bacteria, 34


 Page 351

Electron microscopy, 322323

Electrophoresis:

multilocus enzyme, 257

pulsed-field gel, 256257, 261

ELFA (enzyme linked fluorescent immunoassay), 81

VIDAS, 1617

ELISA (enzyme linked immunosorbent assay), 297

advantages, 81

automation, 17

comparison to conventional methods, 16

principles, 7982

test kits, 7981

VIDAS, 8182
EMBL Sequence Retrieval System, 238, 241, 247

End point heating method, catalase testing, 13

Enrichment cultures, detection of E. coli in, 6163

Enterobacter:

DNA probes for, 215216

identification of, 107109

Enterobacteriaceae, identification of, 94, 96, 98

Enterobacterial repetitive intergenic consensus (ERIC), 253

Enterococcus, identification of, 100, 102

Enterotoxins, 143144

heat-labile, 134

heat-stable, 134

Enterovirus, 152

Enzyme:

based PCR, 187

digestion, 32

EPEC (enteropathogenic E. coli), 133

Epifluorescence microscopy, 50, 327

Escherichia coli:

colorimetric hybridization assays for, 214

databases, 235

detection by flow cytometry, 5767

detection by immunoblotting, 85
 detection using biosensors, 8788

Data Collection, 235

DNA probes for, 217218

effect of irradiation on, 296297

EHEC (enterohemorrhagic E. coli), 134137

plasmid, 135

Shiga toxin, 134135

EIEC (enteroinvasive E. coli), 133134

EPEC (enteropathogenic E. coli), 133

ETEC (enterotoxigenic E. coli), 134

gene probes for detection, 133137

genetic drift in fingerprinting, 260261

Genome Center, 235

method for determination, 4950

microscopy of, 5152

PFGE for detection, 256257, 261

types of testing, 217218

ETEC (enterotoxigenic E. coli), 134

European Bioinformatics Institute (EBI), 238, 240

European Molecular Biology Laboratory (EMBL), 230

Evanescent field sensing, 7071

F
False-negatives, in PCR, 190191

False-positives, in PCR, 188189

FAM (16-carboxyfluorescein), 140

FastA, 243244
Fatty acid reductase, 266

Fiber optic fluorescence sensors, 71, 75

File transfer protocol (FTP), 241242
Fingerprinting, 249261

future directions, 261

genetic drift, 259261

genomic diversity, 257258

methods, 250257

PCR, 250253
RFLP, 254256

 Page 352

problems with interpretation, 257261

reproducibility, 258259

standardization, 258259

FISH (fluorescent in situ hybridization), for flow cytometry, 65

Flagellin, 130

Flow cytometry, 5767

for E. coli detection, 5767

detection by fluorescence, 5961

detection by light scatter, 59

in beef, 6163

limits of detection, 6061

sorting viable cells, 6263

light scatter resolution, 6465

multiparameter analysis, 6465

principles, 5859

Fluorescence:

fiber optic sensors, 71

to detect E. coli, 5961

Fluorescent monoclonal antibody (FA) staining, 295

Foam flotation, to separate bacteria, 3839

Food processing, equipment hygiene, 271272

Formats:

for gene probes

liquid-liquid, 127

liquid-solid, 126127

Freeze-fracture, 324325

Freezing, injury to bacteria, 268269

Fusobacterium, MIDI for detection of, 110

G
Gas bubbles, method to separate bacteria, 35

GenBank, 230233

Entrez, 238, 241, 247

problems, 231232

size, 230

GenPept, 232233

Genes:
 lux, 265282

rpoS, 273276, 280281

sequence databases, 229247

Gene probes, 115153

applications, 128153
for detecting Aeromonas, 128129

for detecting bacteria, 128149

for detecting Bacillus cereus, 129130

for detecting Campylobacter, 130131

for detecting Clostridium botulinum, 131132

for detecting E. coli, 133137

for detecting Listeria, 137140

for detecting Salmonella, 140142

for detecting Shigella, 142143

for detecting Staphylococcus aureus, 143144

for detecting Vibrio, 144147

for detecting viruses, 149152

for detecting Yersinia, 147149

detection, 122, 125126

direct, 125

indirect, 125126
methods, 122

format
liquid-liquid, 127

liquid-solid, 126127
labels, 121126

techniques, 123126
principles, 116

sensitivity, 116, 122

target selection, 120
test development, 119127

to detect nonculturable bacteria, 289

types of probes

natural, 120121
synthetic, 121

GENE-TRAK:
advances, 218223

DLP assay, 218223

 for detection of Listeria, 138139

for detection of Salmonella, 140141

 Page 353

principles, 213214

uses, 214

Genetic drift, in DNA fingerprinting, 255261

Genetics, of bioluminescent bacteria, 266

Genomic diversity, 257258

Glassy state products, 276281

Glucose, fermenting bacteria, 95, 100

GNI (gram negative identification card), 9495, 100

GPI (gram positive identification card), 100102

comparison to other methods, 100

Gram negatives:

identification of, 94100

nonculturable, 289298
Grating-based sensors, 7172

Gravimetric diluter, 56

Heating, injury to bacteria, 269270

Helicobacter pylori, 297

Hemolytic uremia syndrome (HUS), 134135

Hepatitis A, 151152

History:

of impedance, 307308

of methodology, 4

Hybridization:

chemiluminescent assay, 214215

colorimetric assay, 213214

for Campylobacter, 214

for E. coli, 214

for Listeria, 214

for Salmonella, 214

DLP assay, 218227

GENE-TRAK, 213214

advances, 218223

DLP assay, 218223

in RFLP, 254255

kinetics, 117118
 methods for detection of foodborne bacteria, 15

mismatches in, 211212

of DNA, 117119

principles, 117119

protection assay, 127

reaction factors, 118119

sensitivity, 119

specificity, 119

stringency, 118119

sequence homology, 210211

stringency, 211

temperature, 118119

Hygiene:

catalase testing for, 1213

control, 271272

impedance measurement for monitoring, 310

hylA gene, 146

Hypertext markup language (HTML), 236

Hypertext transfer protocol (HTTP), 241

Image analysis, 49
to study biofilms, 336

Immunoblotting, for detection of E. coli, 85

Immunochromatography, 8384
Immunodiagnostics, 7788

antibody-coated paramagnetic beads, 8587

biosensors, 8788

direct immunostaining, 85
ELFA, 81

ELISA, 7982
advantages, 81

principles, 79
test kits, 7981

VIDAS, 8182
particle agglutination assays, 8284

for detection of Listeria, 8384

precipitation tests, 8485

to detect Salmonella, 8485

Immunology, history of methods, 78

Immunomagnetic capture, 17
Immunostaining, 85

 Page 354

for E. coli detection, 85

Impedance, 305311

advantages, 309

applications, 309311
Bactometer, 1011

for measuring biomass, 10

history, 307308
instrumentation, 308

measurement in food, 1011

principle, 305307

RABIT, 11

to study biofilms, 328, 341

uses, 11
Injury, 268270

freezing, 268269

heat, 269270

Instrumentation:

for bioluminescence, 267

of impedance measurement, 308309

BacTrac, 308

Bactometer, 308
Malthus, 308

RABIT, 308

Interferometer, integrated optic, 7375

Internet, addresses, 246

Irradiation, effect on E. coli, 296297

ISOGRID, 7

Kinetics, of DNA hybridization, 117118

Labels:

cDNA probes, 124

5' end labeling, 124

3' end oligonucleotide, 124

3' end tailing, 124

gene probe, 121126

 in vitro transcription of RNA, 124

nonenzymatic, 124

techniques, 123126

nick translation, 123124

random primed DNA labeling, 123

Lactic acid bacteria, PCR for, 187

Light:

microscopy, 326

scatter

resolution, 64

to detect E. coli, 59

Listeria:

AccuProbe for detection of, 139

biofilms, 331336

colorimetric hybridization assays for, 214

commercial kits, 138139

detection by particle agglutination systems, 8384

DNA probes for, 216217

gene probes for detection, 137140

GENE-TRAK, 138139

interference in detection by PCR, 190

methods for detection, 18

ribotyping, 140
Listeriolysin O, 138

Luciferase, 266, 269270

lux gene, 265282

bacterial luciferase, 266

engineering bacteriophage to contain, 281282

Macrodialysis, to separate bacteria, 37

Magnification, of microscopy, 54

Malthus, 308
uses, 11

Marine bacteria, 266

Mass action effect, 3233

Medline, 241
Membrane filtration, to separate bacteria, 3536

Metabolism, problems in determining, 291294

Methods:

automated, attributes, 19
biphasic partitioning, 38

 Page 355

catalase test

for psychrotrophs, 12

to estimate biomass, 12

uses, 12
centrifugation, 3637

colony counter

CASBA, 7
Countermat, 7

Protos, 7

column chromatography, 37

commercial kits, 14

dielectrophoresis, 38

electrical detachment, 34
flow cytometry, 5767

foam flotation, 3839

for biomass, 914

for estimating microbial populations, 914

gas bubbles, 35

history, 4

immunomagnetic capture, 17

key issues, 5
macrodialysis, 37

membrane filtration, 3536

microscopy, 4555

miniaturized

microtiter plates, 1314

uses, 1314

motility enrichment, 1718

of detection of pathogens, 2931

of microbiological analysis, 120

PCR, 1516, 116, 128, 130133, 136, 139, 141142, 144147, 183194, 251253

publications, 12, 19

spiral plating, 67
spray, to separate bacteria, 3334

standing wave ultrasound, 3738

sticky tapes, 3435

 Stomacher, 3, 3132

ultrasound, 34

viable cell count

ISOGRID, 7

Petrifilm, 7

Redigel, 78

spiral plating method, 67

Micrococcus, identification of, 109

Microscopy, 4555

calculation of concentration, 52
computer controlled image analysis, 49

confocal scanning laser, 327

determination of cell viability, 49

electron, 322323

cryo-scanning, 324325

environmental scanning, 325

low voltage field emission scanning, 323324

scanning, 323
transmission, 325

variable pressure scanning, 324

epifluorescence, 327, 336337

filter microscope factor, 52

for quantitation of bacteria, 5053

history, 4546
light, 326

magnification, 54
microscope fields, 54

phase-contrast, 326327
sensitivity, 47, 5354

specificity, 4748
Microtiter plates, 1314

MIDI, for detection of Fusobacterium, 109110

Milk:

attachment of bacteria in, 331332

PCR for pathogens in, 191192
Miniaturization, of microbiological techniques, 1315

Mismatches, in hybridization, 211212

Models, for biofilms in food systems, 333338
Mold, PCR for, 187188
Molecular probes, 48

Morphology, of viable, nonculturable cells, 293

Most probable number counts (MPN), 291

Motility enrichment, 1718

 Page 356

Multilocus enzyme electrophoresis (MEE), 257

Multiparameter analysis, 6465

National Center for Biotechnology Information (NCBI), 238239, 241

Nick translation, 123124

Norwalk virus, 150

Oligonucleotide ligation assay (OLA), 142

Oligonucleotides, 48

probes, 145
Oligoprobes, 151

Omnispec bioactivity monitor system, 1112

Optics:

biosensors, 6976

description of technologies, 7071

Oxyrase, 18

Oysters, hepatitis A virus in, 151152

Paramagnetic beads, 8587

Parasites, PCR for, 186

Particle agglutination assays, 8284

for detection of Listeria, 8384

Pathogens:
concentration, 3032

from foods, 2739

single stage, 30

two stage, 3031

detection

efficiency, 2931

methods, 2931

efficiency of separation, 3032

enumeration, 2930

extraction from foods, 2829

foodborne, needed information, 3

separation, mass action effect, 3233

separation from foods, 2739
 separation methods, 29

PCR (polymerase chain reaction), 116, 130133, 136, 139, 141142, 144, 147, 183194

arbitrarily primed (AP-PCR), 251253

BAX screening system, 1516

effect of cellular stresses, 188

enhanced detection, 192

false-negatives, 190191

false-positives, 188189

for anaerobes, 185

for lactic acid bacteria, 187

for mold, 187188

for parasites, 186

for sequencing DNA, 193

for yeasts, 187188

future research, 193194

gene probe applications, 128

in food microbiology, 1516

interfering chemicals, 189191

labeling DNA by, 124

long, 261
modifications, 16

multiple organism assays, 191192

new developments, 192193
of Vibrio cholerae, 293, 296

principles, 183184
problems with, 188191

random amplified polymorphic DNA (RAPD), 185186, 251

repetitive sequence typing, 253

reverse transcriptase-PCR, 150151, 190

reviews, 184

RiboPrinter, 16
targeting specific genes, 250251

to detect nonculturable bacteria, 289290

Petrifilm, 7

Phage, engineered to contain lux, 281282

Phase-contrast microscopy, 326327

Photoactivation, 124
Photobacterium leiognathi, 271272
Photorhabdus luminescens, 266
Phylogenetics, 208209, 243

 Page 357

PIR (Protein Information Resource), 232233

Plate counting, 295

comparison to bioluminescence, 270

Polio, 152
Precipitation tests, 8485

for detection of Salmonella, 8485

Preservative challenge efficacy, impedance measurement of, 310

Primers, for PCR, 184185

Probes:

DNA, 207223

applications, 215218

assays, 212215

design, 209212
direct label probe (DLP), 218223

for detection of E. coli, 217218

for detection of Listeria, 216217

for detection of Salmonella, 215216

for gene testing, 119153

types

natural, 120121

synthetic, 121
Problems, with PCR, 188191

Processing plants, RAPD assay for detection of bacteria in, 186

Protos colony counter, 7

Pseudomonas aeruginosa, attachment to stainless steel, 329331

Psychrotrophs, catalase test for, 12

Publications, on methodology, 12, 19

Pulsed-field gel electrophoresis (PFGE), 256257, 261

Q
Quantitation, of bacteria with microscopy, 5054

Quorum sensing, 281

RABIT (rapid automated bacterial impedance technique), 11, 308

RAPD (random amplified polymorphic DNA):

for bacteria and viruses in food, 185186, 251

Reaction factors, of DNA hybridization, 118119

Redigel, 78

Repetitive extragenic palindromic (REP) sequences, 253

Reproducibility, of DNA fingerprinting, 258259

Resonant mirror sensors, 7273

Respiratory system, bacterial identification in, 95

Resuscitation, of bacteria, 291

RFLP (restriction fragment length polymorphism), 136137, 145

hybridization methods, 254255

methods, 254257

pulsed-field gel electrophoresis, 256

R-GNB (rapid gram negative bacilli) panel, 105107

R-GPB (rapid gram positive bacilli) panel, 107108

Ribosomal Database Project, 233

Ribotyping, 194, 258259

RNA, 48, 152

hybridization assays, 208209

in vitro transcription of, 124

sequence of Salmonella and Citrobacter, 210211

Rotavirus, 150151
Rotorinser, 33

RpoS, 273276, 280281

adaptive response, 276281

bioluminescent biosensor, 274

glass formation, 276281

induction in presence of competitive microflora, 274276

RT-PCR, 150151, 190

Salmonella:
chemiluminescence for detection, 142

 Page 358

colorimetric hybridization assays for, 214

colorimetry for detection, 142

detection by IO sensor, 7475

detection by precipitation test, 8485

detection using biosensors, 8788

DLP assay, 218223

comparison of foods, 223

inclusivity, 220

sensitivity, 222

specificity, 221

DNA probes for, 215216

gene probes for detection, 140142

GENE-TRAK, 140141
induction of rpoS, 274276

invA-D genes, 141

methods for detection, 1618

UNIQUE system, 17

ribotyping, 142

RNA sequence, 210211

typhimurium, 272273

Sample preparation:
food, 56

methods, 36

gravimetric diluter, 56

Stomacher, 3

Sampling, 36

Sanitizer:

effect on biofilms, 334336, 338339, 341

Scanning electron microscopy, 323

cryo-, 324325

environmental, 325

low-voltage field-emission, 323324

variable-pressure, 324
Scintillation counter, for measuring bioluminescence, 267

Seafood, hepatitis A virus in, 151152

Sensititre, 14, 96, 98, 103104

Sensitivity:

of DNA hybridization, 119

of gene probes, 116, 122

of microscopy, 47, 53

improvement by concentration, 47

method for increasing, 54

Separation:

biphasic partitioning, 38

centrifugation, 3637

column chromatography, 37
dielectrophoresis, 38

efficiency, 3032

electrical detachment, 34

foam flotation, 3839

gas bubbles method, 35

macrodialysis, 37

mass action effect, 3233

membrane filtration, 3536

methods, 29, 3032

of bacteria in suspension, 35
of pathogens from foods, 2739

spray method, 3334

standing wave ultrasound, 3738

sticky tapes, 3435

swabs, 33

ultrasound, 34
Sequencing, databases, 229247

Sereny invasiveness test:

for detection of Shigella, 143

for detection of Yersinia, 147148

Seroconversion, of Vibrio cholerae, 297298

Serotypes, of Vibrio cholerae, 294

Shelf-life, impedance measurement of, 310

Shellfish, 145, 150, 190, 297

Shiga toxin, 134136
Shiga-like toxin, 134135

Shigella:
gene probes for detection, 142143
 Sereny invasiveness test, 143
Signaling, cell-cell, 281

Software archives, addresses, 243

Sorting, of viable cells, 6263

Southern blot, 127

Specificity:

of DNA hybridization, 119, 211

of microscopy, 4748

Spiral plating method, problems, 6

 Page 359

Spoilage bacteria, PCR for, 187

SPR (surface-enhanced plasmon resonance), 7172

Spray methods, to separate bacteria, 3334

Staining method, for viability, 295

Standardization, of DNA fingerprinting, 258259

Staphylococcus:

aureus, gene probes for detection, 143144

identification of, 100101

Starvation, survival of bacteria during, 292

Sticky tapes, to separate bacteria, 3435

Stomacher, 3, 3132

Streptavidin, 125

Streptococcus, identification of, 102, 107108

Stringency, of DNA hybridization, 118119, 211

Substratum, definition, 316

Subtyping:

DNA fingerprinting, 249261

methods, 249

reasons for, 249250

Survival, of bacteria during stress, 292

Suspension, separation of bacteria in, 35

Swabs, efficiency, 33

SwissProt, 232233

TAMARA (6-carboxy-tetramethylrhodamine), 140

Thermostable direct hemolysin (TDH), 146147

Thrombotic/thrombocytopenic purpura (TTP), 134135

Transmission electron microscopy, 325

Typing:
repetitive sequence, 253

enterobacterial repetitive intergenic consensus (ERIC), 253

repetitive extragenic palindromic (REP) sequences, 253

Ultrasound:

standing wave, to separate bacteria, 3738

to separate bacteria, 34

UNIQUE system, for detection of Salmonella, 17

URL (Universal Resource Locators), for databases, 235236, 238

Urine, bacterial identification in, 95, 107108

Verocytotoxin (VT), 134135

Viability:

correlation to bioluminescence, 271

determination by microscopy, 49

Viable cell count:

comparison of methods, 8

ISOGRID, 7

Petrifilm, 7

Redigel, 78

Viable, nonculturable cells:

characteristics, 292

detection of, 289298

direct viable count (DVC), 295296

p-iodonitrotetrazolium (INT), 296

morphology, 293

problems in determining metabolism, 291294

sorting of, 6263

terminology, 290
virulence, 293

Vibrio:
cholerae, 289, 292297
commercial tests, 294

direct viable count (DVC), 295296

DFA-DVA, 295296

isolation, 293
Oag antigenic property, 297298

outbreaks, 294295
seroconversion, 297298

serotypes, 294
survival, 292