Platelets and Megakaryocytes: Volume 4, Advanced Protocols and Perspectives

Methods in
Molecular Biology 1812
Jonathan M. Gibbins
Martyn Mahaut-Smith
Editors
Platelets and
Megakaryocytes
Volume 4, Advanced Protocols
and Perspectives
Methods in M o l e c u l a r B i o lo g y
Series Editor
John M. Walker
School of Life and Medical Sciences
University of Hertfordshire
Hatfield, Hertfordshire, AL10 9AB, UK
For further volumes:

http://www.springer.com/series/7651
Platelets and Megakaryocytes
Volume 4, Advanced Protocols and Perspectives
Edited by
Jonathan M. Gibbins
Institute for Cardiovascular and Metabolic Research, School of Biological Sciences, University of Reading,
Reading, Berkshire, UK
Martyn Mahaut-Smith
Department of Molecular and Cell Biology, University of Leicester, Leicester, UK
Editors
Jonathan M. Gibbins Martyn Mahaut-Smith
Institute for Cardiovascular and Metabolic Department of Molecular and Cell Biology
Research, School of Biological Sciences University of Leicester
University of Reading Leicester, UK
Reading, Berkshire, UK
ISSN 1064-3745 ISSN 1940-6029 (electronic)

Methods in Molecular Biology
ISBN 978-1-4939-8584-5 ISBN 978-1-4939-8585-2 (eBook)
https://doi.org/10.1007/978-1-4939-8585-2
Library of Congress Control Number: 2018949611
© Springer Science+Business Media, LLC, part of Springer Nature 2018

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction
on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation,
computer software, or by similar or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and
regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed
to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty,
express or implied, with respect to the material contained herein or for any errors or omissions that may have been
made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Printed on acid-free paper
This Humana Press imprint is published by the registered company Springer Science+Business Media, LLC part of
Springer Nature.
The registered company address is: 233 Spring Street, New York, NY 10013, U.S.A.
Preface
Research in the intertwined fields of platelet and megakaryocyte biology has developed at
an astonishing pace in the last few years, leading to fundamental new discoveries that impact
upon our understanding of hemostasis, thrombosis, and thrombopoiesis. Indeed, the func-
tions of these cells in a wide range of additional processes ranging from inflammation to
metastasis indicate their multifaceted roles in physiology and pathology. The Platelets and
Megakaryocytes—Methods in Molecular Biology series began in 2004 with the publication
of two volumes describing experimental techniques and perspectives aimed at researchers
embarking on studies in this area. This included protocols developed by leading researchers
and incorporated tips and tricks to enable their successful use. These volumes quickly
became essential resources in many research groups and led to an additional volume of
protocols and perspectives in 2012 to keep apace with new innovations and developments.
The rapid incorporation of sophisticated new techniques into the study of platelets and
megakaryocytes, including new imaging approaches, new methods for platelet production
in vitro, and systems biology approaches, offers important new opportunities. Hence, the
need for an additional volume in this successful series that incorporates advanced method-
ologies and perspectives—Platelets and Megakaryocytes: Volume 4, Advanced Protocols and
Perspectives.
Reading, UK Jonathan M. Gibbins

Leicester, UK Martyn Mahaut-Smith
v
Contents
Preface.................................................................................................................................. v
Contributors......................................................................................................................... ix
1 Immobilization of Nonactivated Unfixed Platelets
for Real-Time Single-Cell Analysis�� 1
Alexander P. Bye, Zeki Ilkan, Amanda J. Unsworth, and Chris I. Jones
2 Imaging Platelets and Megakaryocytes by High-Resolution
Laser Fluorescence Microscopy�� 13
Fred G. Pluthero and Walter H. A. Kahr
3 Single-Molecule Localization and Structured Illumination
Microscopy of Platelet Proteins�� 33
Natalie S. Poulter, Abdullah O. Khan, Chiara Pallini,
and Steven G. Thomas
4 Electron Tomography and Correlative Approaches in Platelet Studies�� 55
Kasia B. Engberts, Cor Seinen, Willie J. C. Geerts,
and Harry F. G. Heijnen
5 Screening and High-Throughput Platelet Assays�� 81
Alexander P. Bye, Amanda J. Unsworth, and Jonathan M. Gibbins
6 High-Throughput Signaling Profiling in Blood Platelets
by Multiplexed Phosphoflow Cytometry�� 95
Benjamin E. J. Spurgeon and Khalid M. Naseem
7 Precise Quantification of Platelet Proteins
and Their Phosphorylation States�� 113
Francoise Mazet and Michael J. Fry
8 The Study of Platelet Receptors Using Artificial Lipid Bilayers�� 127
Michael L. Dustin and Alice Y. Pollitt
9 Three-Dimensional Culture in a Methylcellulose-Based
Hydrogel to Study the Impact of Stiffness
on Megakaryocyte Differentiation�� 139
Alicia Aguilar, Julie Boscher, Fabien Pertuy, Christian Gachet,
and Catherine Léon
10 Differentiation of Human Pluripotent Stem Cells
to Megakaryocytes by Transcription Factor-Driven
Forward Programming�� 155
Thomas Moreau, Amanda L. Evans, and Cedric J. G. Ghevaert
11 Three-Dimensional Tissue Models for Studying Ex Vivo
Megakaryocytopoiesis and Platelet Production�� 177
Christian A. Di Buduo, Vittorio Abbonante, Lorenzo Tozzi,
David L. Kaplan, and Alessandra Balduini
vii
viii Contents
12 Fluorescence Approaches to Image and Quantify the Demarcation

Membrane System in Living Megakaryocytes�� 195
Sangar Osman, Daniel Dalmay, and Martyn Mahaut-Smith
13 High-Resolution 3D Imaging of Megakaryocytes Using Focused
Ion Beam-Scanning Electron Microscopy�� 217
Anita Eckly, Jean-Yves Rinckel, Fabienne Proamer, and Christian Gachet
14 Optical Clearing of Murine Bones to Study Megakaryocytes
in Intact Bone Marrow Using Light-Sheet Fluorescence Microscopy�� 233
Maximilian G. Gorelashvili, Katrin G. Heinze, and David Stegner
15 Mathematical Techniques for Understanding Platelet Regulation
and the Development of New Pharmacological Approaches�� 255
Joanna L. Dunster, Mikhail A. Panteleev, Jonathan M. Gibbins,
and Anastacia N. Sveshnikova
Index�� 281
Contributors
Vittorio Abbonante • Department of Molecular Medicine, University of Pavia, Pavia,

Italy
Alicia Aguilar • Université de Strasbourg, INSERM, EFS GRAND EST, BPPS UMR_
S949, FMTS, Strasbourg, France
Alessandra Balduini • Department of Molecular Medicine, University of Pavia, Pavia,
Italy; Department of Biomedical Engineering, Tufts University, Medford, MA, USA
Julie Boscher • Université de Strasbourg, INSERM, EFS GRAND EST, BPPS UMR_
Christian A. Di Buduo • Department of Molecular Medicine, University of Pavia,
Pavia, Italy
Alexander P. Bye • Institute for Cardiovascular and Metabolic Research, School of
Biological Sciences, University of Reading, Whiteknights, Reading, Berkshire, UK
Daniel Dalmay • Department of Molecular and Cell Biology, Lancaster Road, University
of Leicester, Leicester, UK
Joanna L. Dunster • Institute for Cardiovascular and Metabolic Research, School of
Michael L. Dustin • Kennedy Institute of Rheumatology, University of Oxford, Roosevelt
Drive, Headington, Oxford, UK
Anita Eckly • Université de Strasbourg, INSERM, EFS GRAND EST, BPPS UMR_S949,
FMTS, Strasbourg, France
Kasia B. Engberts • Bijvoet Center for Biomolecular Research, Utrecht University,
Utrecht, The Netherlands
Amanda L. Evans • Department of Haematology, Wellcome Trust – Medical Research
Council Cambridge Stem Cell Institute, NHS Blood and Transplant, Long Road,
University of Cambridge, Cambridge, UK
Michael J. Fry • Institute for Cardiovascular and Metabolic Research, School of Biological
Sciences, University of Reading, Whiteknights, Reading, Berkshire, UK
Christian Gachet • Université de Strasbourg, INSERM, EFS GRAND EST, BPPS
UMR_S949, FMTS, Strasbourg, France
Willie J. C. Geerts • Bijvoet Center for Biomolecular Research, Utrecht University,
Utrecht, The Netherlands
Cedric J. G. Ghevaert • Department of Haematology, Wellcome Trust – Medical
Research Council Cambridge Stem Cell Institute, NHS Blood and Transplant, Long
Road University of Cambridge, Cambridge, UK
Jonathan M. Gibbins • Institute for Cardiovascular and Metabolic Research, School of
Maximilian G. Gorelashvili • Institute of Experimental Biomedicine, University
Hospital Würzburg, Würzburg, Germany
Harry F. G. Heijnen • Laboratory for Clinical Chemistry and Hematology, University
Medical Center Utrecht, Utrecht, The Netherlands; Department of Cell Biology, Cell
Microscopy Core, University Medical Center Utrecht, Utrecht, The Netherlands
ix
x Contributors
Katrin G. Heinze • Rudolf Virchow Center for Experimental Biomedicine, University

of Würzburg, Würzburg, Germany
Zeki Ilkan • Department of Molecular and Cell Biology, Lancaster Road, University of
Leicester, Leicester, UK
Chris I. Jones • Institute for Cardiovascular and Metabolic Research, School of Biological
Sciences, University of Reading, Whiteknights, Reading, Berkshire, UK
Walter H. A. Kahr • Cell Biology Program, Research Institute, , Hospital for Sick
Children, Toronto, Toronto, Ontario, Canada; Department of Biochemistry, University of
Toronto, Toronto, Ontario, Canada; Department of Paediatrics, Division of
Haematology/Oncology, University of Toronto and The Hospital for Sick Children,
Toronto, Ontario, Canada
David L. Kaplan • Department of Biomedical Engineering, Tufts University, Medford,
MA, USA
Abdullah O. Khan • Institute of Cardiovascular Sciences, College of Medical and Dental
Sciences, University of Birmingham, Birmingham, UK
Catherine Léon • Université de Strasbourg, INSERM, EFS GRAND EST, BPPS UMR_
Martyn Mahaut-Smith • Department of Molecular and Cell Biology, Lancaster Road,
University of Leicester, Leicester, UK
Francoise Mazet • Institute for Cardiovascular and Metabolic Research, School of
Thomas Moreau • Department of Haematology, Wellcome Trust – Medical Research
Council Cambridge Stem Cell Institute, NHS Blood and Transplant, Long Road,
University of Cambridge, Cambridge, UK
Khalid M. Naseem • Leeds Institute of Cardiovascular & Metabolic Medicine,
The LIGHT Laboratories, Clarendon Way, University of Leeds, Leeds, UK
Sangar Osman • Department of Molecular and Cell Biology, Lancaster Road,
University of Leicester, Leicester, UK
Chiara Pallini • Institute of Cardiovascular Sciences, College of Medical and Dental
Sciences, University of Birmingham, Birmingham, UK
Mikhail A. Panteleev • Faculty of Physics, Lomonosov Moscow State University, Moscow,
Russia; Center for Theoretical Problems of Physicochemical Pharmacology, Russian
Academy of Sciences, Moscow, Russia; National Scientific and Practical Centre of
Pediatric Hematology, Oncology and Immunology named after Dmitry Rogachev,
Moscow, Russia; Faculty of Biological and Medical Physics, Moscow Institute of Physics
and Technology, Dolgoprudny, Russia
Fabien Pertuy • Université de Strasbourg, INSERM, EFS GRAND EST, BPPS UMR_
Fred G. Pluthero • Cell Biology Program, Research Institute, Hospital for Sick Children,
Toronto, ON, Canada
Alice Y. Pollitt • Institute for Cardiovascular and Metabolic Research, School of
Natalie S. Poulter • Institute of Cardiovascular Sciences, College of Medical
and Dental Sciences, University of Birmingham, Birmingham, UK; COMPARE,
University of Birmingham and University of Nottingham, Midlands, UK
Fabienne Proamer • Université de Strasbourg, INSERM, EFS GRAND EST, BPPS
Contributors xi
Jean-Yves Rinckel • Université de Strasbourg, INSERM, EFS GRAND EST, BPPS

Cor Seinen • Laboratory for Clinical Chemistry and Hematology, University Medical
Center Utrecht, Utrecht, The Netherlands
Benjamin E. J. Spurgeon • Leeds Institute of Cardiovascular & Metabolic Medicine,
The LIGHT Laboratories, Clarendon Way, University of Leeds, Leeds, UK
David Stegner • Institute of Experimental Biomedicine, University Hospital Würzburg,
Würzburg, Germany
Anastacia N. Sveshnikova • Faculty of Physics, Lomonosov Moscow State University,
Moscow, Russia; Center for Theoretical Problems of Physicochemical Pharmacology,
Russian Academy of Sciences, Moscow, Russia; National Scientific and Practical Centre
of Pediatric Hematology, Oncology and Immunology named after Dmitry Rogachev,
Moscow, Russia
Steven G. Thomas • Institute of Cardiovascular Sciences, College of Medical and Dental
Sciences, University of Birmingham, Birmingham, UK; COMPARE, University of
Birmingham and University of Nottingham, Midlands, UK
Lorenzo Tozzi • Department of Biomedical Engineering, Tufts University, Medford,
MA, USA
Amanda J. Unsworth • Institute for Cardiovascular and Metabolic Research, School
of Biological Sciences, University of Reading, Whiteknights, Reading, Berkshire, UK
Chapter 1
Immobilization of Nonactivated Unfixed Platelets

for Real-Time Single-Cell Analysis
Alexander P. Bye, Zeki Ilkan, Amanda J. Unsworth, and Chris I. Jones
Abstract
Existing methods for measuring the response of individual platelets to stimulation are limited. They either
measure each platelet at one discrete time-point (flow cytometry) or rely on adhesive ligands to immobilize
platelets that concomitantly generate activation signals (microscopy). Such methods of immobilization
make it impossible to assess resting platelets, the changes that occur as platelets transition from resting to
active states, or the signals generated by soluble agonists, such as ADP and thrombin, or by mechanical
stimulus, independently from those generated by the adhesive ligand. Here we describe a microscopy
method that allows the immobilization of platelets to a glass cover slip without triggering platelet activa-
tion. This method makes use of specific antibodies that bind platelet PECAM-1 without activating it.
Platelets can therefore be immobilized to PECAM-1 antibody coated biochips without causing activation
and perfused with agonists or inhibitors. Using this method, platelets can be stimulated by an array of
soluble agonists at any concentration or combination, in the presence or absence of inhibitors or shear
forces. This chapter describes in detail this PECAM-1 mediated immobilized platelet method and its use
for measuring changes in Ca2+ signaling in individual platelets under a number of different conditions.
While we focus on the measurement of Ca2+ dynamics in this chapter, it is important to consider that the
basic method we describe will easily lend its self to other measures of platelet activation (integrin activation,
shape change, actin dynamics, degranulation), and may, therefore, be used to measure almost any facet of
platelet activation.
Key words Immobilized Platelet, Calcium, ADP, Thrombin, Microscopy, Biochip
1 Introduction
It is desirable, in many cases, to monitor the cellular processes in

individual platelets rather than a population of platelets. This is
relatively straight forward and can be done either by flow cytome-
try or by microscopy. Flow cytometry enables analysis of both rest-
ing platelets and platelets activated by an array of agonists either
individually or in combination, at fixed time points or in real-time
assays, but with the limitation of only measuring each platelet at
one discrete time-point [1–3]. Microscopy assays, by contrast,
allow changes in individual platelets to be followed over time but
Jonathan M. Gibbins and Martyn Mahaut-Smith (eds.), Platelets and Megakaryocytes: Volume 4,
Advanced Protocols and Perspectives, Methods in Molecular Biology, vol. 1812,
https://doi.org/10.1007/978-1-4939-8585-2_1, © Springer Science+Business Media, LLC, part of Springer Nature 2018
1
2 Alexander P. Bye et al.
with the limitation that platelets must be adhered to coverslips

coated with adhesive receptor ligands such as fibrinogen, collagen,
von Willebrand factor, or synthetic peptides [4–7]. In this approach
activation signals are, however, generated by the adhesive ligand
making it impossible to assess resting platelets or the changes that
occur as platelets transition from resting to active states. It also
makes it impossible to assess signals generated by soluble agonists,
such as ADP or thrombin, or by mechanical stimulus, indepen-
dently from those generated by the adhesive ligand.
With these limitations in mind we have developed a micros-
copy method that allows the immobilization of platelets to a glass
cover slip without triggering platelet activation. This approach
allows the measurement of calcium transients in individual resting
platelets and the changes in platelets in response to agonist- or
shear-induced mechanical stimulation.
The method we outline in detail below is carried out in five
stages using biochips which provide a convenient small capillary
channel onto which platelets can be immobilized and through
which agonists or inhibitors can be infused over the immobilized
platelets (Fig. 1a). The biochips are initially coated in PECAM-1
(WM59) antibodies (Fig. 1b—Stage 1). Excess antibody is removed
and the channel is blocked with 2% BSA to prevent artifactual
platelet activation (Fig. 1b—Stage 2). Platelets are gently loaded
into the flow channel, allowed to bind to the immobilized anti-
body and nonadherent platelets are removed leaving just immobi-
lized nonactivated platelets (Fig. 1b—Stages 3 and 4). Image
acquisition of the immobilized platelets is started and concomi-
tantly agonists are perfused, or shear is increased to cause platelet
activation which can be imaged (Fig. 1b—Stage 5).
This method makes use of the ability of certain antibodies to
bind platelet PECAM-1 without activating it. On the platelet sur-
face PECAM-1 has been estimated to be expressed at between
5000 and 8800 copies per cell [8–10]. Although the level of plate-
let PECAM-1 varies widely within the human population with lev-
els up to 20,000 molecules per platelet seen in around 20% of the
population [11]. PECAM-1 is made up of a 574-amino acid resi-
due extracellular portion organized into six immunoglobulin (Ig)-
like homology domains (Fig. 1c), a 19-amino acid transmembrane
domain and a cytoplasmic domain which includes an
Immunoreceptor Tyrosine-based Inhibitory Motif (ITIM) ((L/V/
I/S/T)XYXX(L/V)) and an Immunoreceptor Tyrosine-based
Switch Motif (ITSM) (TxYxx(V/I)) contained in a lipid-interacting
α-helical segment [12–14] . The principal ligand for PECAM-1 is
PECAM-1 itself through homophilic interaction between immu-
noglobulin (Ig) domains 1 and 2 of the molecules on nearby plate-
lets [15].
In laboratory studies activation of PECAM-1 is most usually
achieved through the binding of selected antibodies that bind to
Immobilised Platelet Imaging 3
Fig. 1 PECAM-1 immobilized platelet method. Cartoon representations of the PECAM-1 immobilized platelet
method. (a) Nonactivated platelets are immobilized on to glass-bottom biochips allowing for agonists, inhibi-
tors, or buffered to be flowed over them at a range of shear stress while continuously imaging changes in
platelet activation (e.g. calcium flux). (b) The five stages of the method for immobilizing platelets. Stage
1—Biochips are coated in PECAM-1 (WM59) antibodies. Stage 2—Exposed glass is coated with 2% BSA to
prevent artifactual platelet activation. Stage 3—Platelets are loaded into the flow channel and allowed to bind
to the immobilized antibody. Stage 4—Nonadherent platelets are removed leaving just immobilized nonacti-
vated platelets. At which point image acquisition can start. Stage 5—Agonists are infused or shear is increased
to cause platelet activation which can be measured. (c) Closer image of Stage 4, showing a platelet immobi-
lized by PECAM-1 tethers. The PECAM-1 (WM59) antibody binds to Ig domains 1 or 2 of the PECAM-1 molecule
which prevents its activation, thereby tethering the platelets without inducing their activation or inhibitory
PECAM-1 signaling [15, 16]
membrane proximal Ig domain 6 of PECAM-1 [16–18]. These

antibodies bind to PECAM-1 but importantly leave Ig domains 1
and 2 free to undergo homophilic interaction that is enhanced
when cross-linking secondary antibodies are used to cluster the
primary antibody and hence PECAM-1 [16–19]. In response to
this direct stimulation of PECAM-1 tyrosine phosphorylation and
signaling occurs leading to an array of inhibitory (and some activa-
tory) responses [16–18, 20, 21].
In the method described below we use antibodies specifically
directed against the Ig domain 1 and 2 of PECAM-1 which inhibit
the activation of PECAM-1 by blocking homophilic ligation that is
dependent on these domains (Fig. 1c) [15, 16]. Thus detailed
knowledge of PECAM-1 physiology and the availability of well-
characterized specific reagents have allowed us to design a method
that provides a surface onto which platelets can attach, via an abun-
dant receptor, and become immobilized without becoming overtly
activated.
This chapter describes in detail the PECAM-1 mediated immo-
bilized platelet method developed in our laboratory, and its use for
measuring changes in Ca2+ signaling in individual platelets under a
number of different conditions. We endeavour to provide all of the
important technical details needed to successfully use this method.
While we focus on the measurement of Ca2+ dynamics in this chap-
ter it is important to bear in mind that the basic method we describe
(the capture and immobilization of nonactivated platelets on a
PECAM-1 antibody surface in capillary channels) will easily lend
its self to other measures of platelet activation (integrin activation,
shape change, actin dynamics, degranulation), and may, therefore,
be used to measure almost any facet of platelet activation.
2 Materials
2.1 Reagents 1. Acid Citrate Dextrose (ACD): 85 mM sodium citrate 111 mM
glucose, and 78 mM citric acid [pH 6.4].
2. 1 μM ADP, 1 μM TRAP-6, or 10 μg/ml CRP-XL (collagen-
related peptide-cross-linked, monomeric sequence
GCI[GPO]10GCOG) was prepared as described previously
[22] (see Note 1, Fig. 2).
3. 500 μM Fluo-4 AM in DMSO (see Notes 2–5).
2.2 Flow Chip 1. Tyrode’s-HEPES buffer—134 mM NaCl, 0.34 mM Na2HPO4,

Preparation 2.9 mM KCl, 12 mM NaHCO3, 20 mM N-2-
and Imaging hydroxyethylpiperazine-N-2-ethanesulfonic acid, 5 mM
glucose and 1 mM MgCl2, pH 7.3—filter using 0.2 μm, 32 mm
Syringe Filters.
Fig. 2 Intracellular Ca2+ measured in platelets adhered to anti-PECAM-1 coated flow chambers during
perfusion with agonists. Washed human platelets incubated with Fluo-4 AM for 1 h at 30 °C were adhered
to a microfluidic flow cell chamber coated with anti-PECAM-1 antibody for 30 min and then perfused with
agonist. (a) Traces are pseudoratios (F/F0) of Fluo-4 fluorescence measured in single platelets (indicated in
images by dashed circle) during perfusion with (i) 1 μM ADP, (ii) 1 μM TRAP-6, or (iii) 10 μg/ml CRP-XL for
7 min. (b) Fluo-4 fluorescence image (top) and brightfield image (bottom) of platelets during perfusion with
ADP (white arrow indicates direction of flow)
2. Monoclonal mouse anti-human antibody against the first or

second Ig domain of PECAM-1 (CD31), clone: WM59
(Serotec, UK). The antibody binds to PECAM-1 but prevents
its activation [15, 16].
3. 2% bovine serum albumin (BSA) in Tyrode’s-HEPES buffer
filtered using 0.2 μm, 32 mm Syringe Filters.
4. Glass-bottom biochips which are precast disposable perfusion
chambers comprising channels, each 1.6 mm wide, 0.16 mm
high, and 28 mm long, with a volume of 0.8 μl.
5. Nanopump to enable precision microfluidic flow at variable

shear rates.
6. Confocal microscope with Resonant Scanner housed in an
environmental hood to maintain the temperature of the bio-
chips, platelets, and any buffer or agonist perfused over the
immobilized platelets and a constant 37 °C.
7. ImageJ 1.47v52 using the Time Series Analyzer V2.0 plugin.
3 Method
The method outlined below is for imaging of Ca2+ transients in

single immobilized platelets. The method of preparing and immo-
bilizing platelets to a glass biochip without causing their activation
is, however, adaptable to many experimental designs of which this
is one example.
3.1 Biochip 1. Coat glass-bottom biochips with monoclonal mouse anti-

Preparation human PECAM-1 (CD31, WM59) antibody by pipetting 1 μl
(containing 1 μg) of antibody into each channel and incubat-
ing at 37 °C for 1 h.
2. Remove excess antibody by very gently flushing each channel
by directly pipetting 10 μl of Tyrode’s-HEPES buffer into the
channels. The buffer should be prewarmed to 37 °C.
3. To prevent glass-induced platelet activation block each channel
by gently directly pipetting 10 μl of 2% BSA, prewarmed to
37 °C, into each channel. Then incubate for 1 h at 37 °C.
4. Remove excess BSA by gently flushing each channel by directly
pipetting 50 μl of Tyrode’s-HEPES buffer (prewarmed to
37 °C) into each channel (see Note 6).
5. Biochips are now ready to use and can be kept at 37 °C or
stored in the fridge at 4 °C overnight. If storing in at 4 °C
overnight make sure that the biochips are brought up to 37 °C
prior to use.
3.2 Blood Collection 1. Obtain human blood (see Note 7) from consenting healthy
and Platelet volunteers who have not taken anti-platelet medication (for
Preparation example aspirin or ibuprofen) in the previous 10 days, via
venesection of the antecubital vein. Collect the blood into
3.8% (w/v) sodium citrate at a ratio of 9 parts blood to 1 part
sodium citrate, adding acid citrate dextrose (ACD) to a final
concentration of 12.5% (v/v).
2. Transfer the blood into 12 × 75 mm polystyrene test tubes and
prepare platelet rich plasma (PRP) by centrifugation at 100 × g
for 20 min using the slowest centrifuge braking setting.
3. Transfer PRP slowly using a wide bore pipette tip to avoid
artifactual shear-induced activation and maintain at 30 °C.
3.3 Dye Loading 1. Load PRP with 2 μM Fluo-4 for 1 h at 30 °C.
2. Wash the platelets by centrifugation at 350 × g for 20 min and
resuspension in Tyrode’s-HEPEs buffer prewarmed to 30 °C
(see Note 8). Adjust platelet concentration to 4 × 107 cells/ml
to ensure that platelets are spatially separated on the biochip,
thereby minimizing artifactual platelet–platelet interaction.
3. Rest platelet for 10 min at 30 °C prior to introducing them
into the biochips.
3.4 Immobilization 1. Immediately prior to each experiment (see Note 9), introduce
of Platelets the Fluo-4 loaded washed platelets into the biochip channel by
onto Biochips very gently pipetting 1 μl of Fluo-4 loaded washed platelets
directly into the channel ensuring that the platelets are not
exposed to shear stress (see Note 10).
2. Incubate at 37 °C for 10 min with occasional very gentle
shaking.
3. The biochip can then be mounted on the microscope stage
making sure that the environmental chamber and the stage are
at 37 °C.
4. Using the pump slowly (<5 dyne/cm2) draw Tyrode’s-HEPEs
buffer (prewarmed to 37 °C) through the channel to wash off
any nonimmobilized platelets.
3.5 Imaging 1. Set up the confocal microscope to monitor platelet calcium

signaling (see Note 11). Observe fluorescence with excitation
at 488 nm and emission at 525 nm with a 60× magnification
lens.
2. Record fluorescence in unstimulated platelets for 5 min prior
to stimulation. The frequency of image acquisition should be
at least 2 FPS (see Note 12).
3.6 Inhibiting or 1. Pharmacological reagents may be introduced at a very low

Stimulating Platelets shear rate (<200 s−1) prior to stimulation.
2. When the microscope is set up and fluorescence has been mon-
itored in unstimulated platelets for 5 min, perfuse agonists
through the channels at low shear rate (<400 s−1) (see Note
10) whilst continuously recording the calcium signal for 5 min
(Fig. 2).
3.7 Image Analysis 1. Captured time-series images are analyzed on ImageJ 1.47v52
using the Time Series Analyzer V2.0 plugin, to obtain the Ca2+
traces (Fig. 2).
2. Changes in Ca2+ concentration can be plotted as F/F0, where
F is the fluorescence intensity (at t = x s) minus background
and F0 is the fluorescence intensity (at t = 0 s) minus
background.
4 Notes
1. Using this method, platelets can be stimulated by an array of

soluble agonists at any concentration or combination, in the
presence or absence of inhibitors or shear stress. Figure 2 pro-
vides typical Ca2+ traces obtained when stimulating with 1 μM
ADP, 1 μM TRAP-6, or 10 μg/ml CRP-XL.
2. Both Fluo-3 and Fura-2 AM may be used with this method if
the confocal microscope is fitted with appropriate lasers.
3. Fluo-4 AM is generally preferable to Fluo-3. Because of its
greater absorption near 488 nm. Fluo-4 emits significantly
brighter fluorescence than Fluo-3, enabling it to be used at
lower intracellular concentrations, both reducing the cost of
experimentation and the amount of invasive dye loading [23].
4. Fura-2 AM: This ratiometric dye allows the measurement of
Ca2+-dependent fluorescence at 340 nm and 380 nm generat-
ing 340/380 ratios to represent intracellular Ca2+ concentra-
tions. The main advantage of using Fura-2 is that variables
such as differences in dye loading, cell thickness or potential
cell movement can be eliminated. These variables could gener-
ate artifacts when nonratiometric dyes are used to measure
intracellular Ca2+ concentrations. The standard procedure of
Fura-2 dye loading involves the treatment of PRP with 2 μM
Fura-2 AM for 45 min at 37 °C, with gentle inversions
throughout the incubation period to allow mixing.
5. Ensure that platelets are not exposed to light during and after
the dye loading steps to avoid photobleaching of Fluo-4.
6. When removing excess antibody or BSA care must be taken
not to flush too vigorously, otherwise the coating will be
disrupted.
7. The technique may be used to study platelets from nonhuman
species, but dye-loading conditions are likely to differ from
those described and may require optimization.
8. Include 0.32 U/ml apyrase in the buffer used to resuspend
platelets and in the external buffer, if P2X1 receptors are being
studied under shear. This will prevent P2X1 channels from
desensitization by ATP which may be found in the extracellular
milieu [24]. Apyrase can be omitted from these steps depend-
ing on the objectives of the study.
9. In our experience it is advisable to immobilize and then stimu-
late the platelets in one channel at a time to obtain optimal
results.
10. Human platelets have been reported to express several shear
stress-operated mechanosensitive ion channels including
Piezo1 [25–28], whose contribution to function in singly
attached platelets can be studied using this protocol. For fur-

ther details on how this method can be adapted to monitor
fluid-shear induced mechanosensitive Ca2+ entry in singly
attached platelets, see Ilkan et al. [28]. Fluid shear stress can
cause mechanosensitive ion channel activation by inducing
tensions within the lipid bilayer of the membrane, which in
turn operate mechanosensitive ion channels permitting ionic
fluxes from the cell exterior [29, 30]. Evidence indicates that
this mechanism is independent of any links to the cytoskeleton
[31], and thus shear stress is directly sensed and transmitted to
the ion channels by the membrane. Using the in vitro approach
described in this chapter, normal or pathological venous or
arterial flow rates (in ml/min) can be applied to immobilized
platelets in biochips by drawing physiological saline through
the biochip channels.
The conversion between flow rate (ml/min) and shear rate
(s−1) can be performed using the following formula, taking
into account the dimensions of the biochip channels:
wh 2t
Q =
6µ
where Q = flow rate (ml/s), w = microslide lumen width in cm

for biochip (0.08); h = microslide lumen height in cm biochip
(0.008); t = shear stress (Pa); μ = viscosity of water
(0.001002 Pa/s). Finally, to calculate the shear rate (s−1), t is
divided by μ.
11. It is essential that everything in the system is at a stable 37 °C. It
is therefore advisable to turn the microscope incubator on
early (1–2 h prior to the start of the experiment) to ensure that
the stage reaches a uniform 37 °C.
12. If shear-induced Ca2+ responses are being studied (such as
mechanosensitive cation channel responses), image capturing
frequency can be adjusted to optimally to capture Ca2+
responses. Higher frequencies will allow the recording of very
brief Ca2+ elevations which would not be captured using lower
frequencies.
References
1. Jones CI, Garner SF, Angenent W, Bernard A, Haemost 5(8):1756–1765. https://doi.
Berzuini C, Burns P, Farndale RW, Hogwood org/10.1111/j.1538-7836.2007.02632.x
J, Rankin A, Stephens JC, Tom BD, Walton J, 2. Jones CI, Tucker KL, Sasikumar P, Sage T,
Dudbridge F, Ouwehand WH, Goodall AH Kaiser WJ, Moore C, Emerson M, Gibbins JM
(2007) Mapping the platelet profile for func- (2014) Integrin-linked kinase regulates the rate
tional genomic studies and demonstration of of platelet activation and is essential for the for-
the effect size of the GP6 locus. J Thromb mation of stable thrombi. J Thromb Haemost
12(8):1342–1352. https://doi.org/10.1111/ 1) that are required for the cellular association

jth.12620 and activation of the protein-tyrosine phospha-
3. Stefanini L, Boulaftali Y, Ouellette TD, tase, SHP-2. J Biol Chem
Holinstat M, Desire L, Leblond B, Andre P, 272(40):24868–24875
Conley PB, Bergmeier W (2012) Rap1-Rac1 13. Vivier E, Daeron M (1997) Immunoreceptor
circuits potentiate platelet activation. tyrosine-based inhibitory motifs. Immunol
Arterioscler Thromb Vasc Biol 32(2):434–441. Today 18:286–291
https://doi.org/10.1161/ATVBAHA.111 14. Paddock C, Lytle BL, Peterson FC, Holyst T,
.239194 Newman PJ, Volkman BF, Newman DK
4. Nesbitt WS, Harper IS, Schoenwaelder SM, (2011) Residues within a lipid-associated seg-
Yuan Y, Jackson SP (2012) A live cell micro- ment of the PECAM-1 cytoplasmic domain are
imaging technique to examine platelet calcium susceptible to inducible, sequential phosphory-
signaling dynamics under blood flow. Methods lation. Blood 117(22):6012–6023. https://
Mol Biol 788:73–89. https://doi. doi.org/10.1182/blood-2010-11-317867
org/10.1007/978-1-61779-307-3_6 15. Sun QH, DeLisser HM, Zukowski MM,
5. Mazzucato M, Pradella P, Cozzi MR, De Paddock C, Albelda SM, Newman PJ (1996)
Marco L, Ruggeri ZM (2002) Sequential cyto- Individually distinct Ig homology domains in
plasmic calcium signals in a 2-stage platelet PECAM-1 regulate homophilic binding and
activation process induced by the glycoprotein modulate receptor affinity. J Biol Chem
Ibalpha mechanoreceptor. Blood 100(8):2793– 271(19):11090–11098
2800. https://doi.org/10.1182/ 16. Jones CI, Sage T, Moraes LA, Vaiyapuri S,
blood-2002-02-0514 Hussain U, Tucker KL, Barrett NE, Gibbins
6. Nesbitt WS, Kulkarni S, Giuliano S, Goncalves JM (2014) Platelet endothelial cell adhesion
I, Dopheide SM, Yap CL, Harper IS, Salem molecule-1 inhibits platelet response to throm-
HH, Jackson SP (2002) Distinct glycoprotein bin and von Willebrand factor by regulating the
Ib/V/IX and integrin alpha IIbbeta 3-depen- internalization of glycoprotein Ib via AKT/
dent calcium signals cooperatively regulate glycogen synthase kinase-3/dynamin and inte-
platelet adhesion under flow. J Biol Chem grin alphaIIbbeta3. Arterioscler Thromb Vasc
277(4):2965–2972. https://doi. Biol 34(9):1968–1976. https://doi.
org/10.1074/jbc.M110070200 org/10.1161/ATVBAHA.114.304097
7. Yap CL, Anderson KE, Hughan SC, Dopheide 17. Moraes LA, Barrett NE, Jones CI, Holbrook
SM, Salem HH, Jackson SP (2002) Essential LM, Spyridon M, Sage T, Newman DK,
role for phosphoinositide 3-kinase in shear- Gibbins JM (2010) Platelet endothelial cell
dependent signaling between platelet glyco- adhesion molecule-1 regulates collagen-
protein Ib/V/IX and integrin alpha(IIb) stimulated platelet function by modulating the
beta(3). Blood 99(1):151–158 association of phosphatidylinositol 3-kinase
8. Metzelaar MJ, Korteweg J, Sixma JJ, with Grb-2-associated binding protein-1 and
Nieuwenhuis HK (1991) Biochemical charac- linker for activation of T cells. J Thromb
terization of PECAM-1 (CD31 antigen) on Haemost 8(11):2530–2541. https://doi.
human platelets. Thromb Haemost org/10.1111/j.1538-7836.2010.04025.x
66(6):700–707 18. Jones CI, Garner SF, Moraes LA, Kaiser WJ,
9. Wu XW, Lian EC (1997) Binding properties Rankin A, Bloodomics C, Ouwehand WH,
and inhibition of platelet aggregation by a Goodall AH, Gibbins JM (2009) PECAM-1
monoclonal antibody to CD31 (PECAM-1). expression and activity negatively regulate mul-
Arterioscler Thromb Vasc Biol tiple platelet signaling pathways. FEBS Lett
17(11):3154–3158 583(22):3618–3624. https://doi.
10. Newman PJ (1994) The role of PECAM-1 in org/10.1016/j.febslet.2009.10.037
vascular cell biology. Ann N Y Acad Sci 19. Cicmil M, Thomas JM, Leduc M, Bon C,
714:165–174 Gibbins JM (2002) PECAM-1 signalling
11. Novinska MS, Rathoare V, Newman DK, inhibits the activation of human platelets.
Newman PJ (2007) PECAM-1. Blood 99:137–144
Platelets:chapter 11. Elsevier, London, 20. Jackson DE, Ward CM, Wang R, Newman PJ
pp 221–230 (1997) The protein-tyrosine phosphatase
12. Jackson DE, Kupcho KR, Newman PJ (1997) SHP-2 binds platelet/endothelial cell adhesion
Characterization of Phosphotyrosine binding molecule-1 (PECAM-1) and forms a distinct
motifs in the cytoplasmic domain of platelet/ signaling complex during platelet aggregation.
endothelial cell adhesion Molecule-1 (PECAM- J Biol Chem 272(11):6986–6993
21. Varon D, Jackson DE, Shenkman B, Dardik R, megakaryocytic cell lines. Thromb Haemost
Tamarin I, Savion N, Newman PJ (1998) 116(2):272–284. https://doi.org/10.1160/
Platelet/endothelial cell adhesion molecule-1 TH15-11-0891
serves as a costimulatory agonist receptor that 27. Boyanova D, Nilla S, Birschmann I, Dandekar
modulates integrin-dependent adhesion and T, Dittrich M (2012) PlateletWeb: a systems
aggregation of human platelets. Blood biologic analysis of signaling networks in
91(2):500–507 human platelets. Blood 119(3):e22–e34.
22. Morton LF, Hargreaves PG, Farndale RW, https://doi.org/10.1182/blood-2011-10-
Young RD, Barnes MJ (1995) Integrin alpha 2 387308
beta 1-independent activation of platelets by 28. Ilkan Z, Wright JR, Goodall AH, Gibbins JM,
simple collagen-like peptides: collagen tertiary Jones CI, Mahaut-Smith MP (2017) Evidence
(triple-helical) and quaternary (polymeric) for shear-mediated Ca2+ entry through
structures are sufficient alone for alpha 2 beta Mechanosensitive Cation channels in human
1-independent platelet reactivity. Biochem platelets and a megakaryocytic cell line. J Biol
J 306:337–344 Chem 292(22):9204–9217. https://doi.
23. Gee KR, Brown KA, Chen WN, Bishop-Stewart org/10.1074/jbc.M116.766196
J, Gray D, Johnson I (2000) Chemical and 29. Coste B, Mathur J, Schmidt M, Earley TJ,
physiological characterization of fluo-4 Ca(2+)- Ranade S, Petrus MJ, Dubin AE, Patapoutian
indicator dyes. Cell Calcium 27(2):97–106. A (2010) Piezo1 and Piezo2 are essential com-
https://doi.org/10.1054/ceca.1999.0095 ponents of distinct mechanically activated cat-
24. Rolf MG, Brearley CA, Mahaut-Smith MP ion channels. Science 330(6000):55–60.
(2001) Platelet shape change evoked by selec- https://doi.org/10.1126/science.1193270
tive activation of P2X1 purinoceptors with 30. Coste B, Xiao B, Santos JS, Syeda R, Grandl J,
alpha,beta-methylene ATP. Thromb Haemost Spencer KS, Kim SE, Schmidt M, Mathur J,
85(2):303–308 Dubin AE, Montal M, Patapoutian A (2012)
25. Burkhart JM, Vaudel M, Gambaryan S, Radau Piezo proteins are pore-forming subunits of
S, Walter U, Martens L, Geiger J, Sickmann A, mechanically activated channels. Nature
Zahedi RP (2012) The first comprehensive and 483(7388):176–181. https://doi.
quantitative analysis of human platelet protein org/10.1038/nature10812
composition allows the comparative analysis of 31. Cox CD, Bae C, Ziegler L, Hartley S, Nikolova-
structural and functional pathways. Blood Krstevski V, Rohde PR, Ng CA, Sachs F,
120(15):e73–e82. https://doi.org/10.1182/ Gottlieb PA, Martinac B (2016) Removal of
blood-2012-04-416594 the mechanoprotective influence of the cyto-
26. Wright JR, Amisten S, Goodall AH, Mahaut- skeleton reveals PIEZO1 is gated by bilayer
Smith MP (2016) Transcriptomic analysis of tension. Nat Commun 7:10366. https://doi.
the ion channelome of human platelets and org/10.1038/ncomms10366
Chapter 2
Imaging Platelets and Megakaryocytes by High-Resolution

Laser Fluorescence Microscopy
Abstract
Microscopy is central to studies of platelets and their precursor megakaryocytes. Here we describe methods
to rapidly obtain high resolution images of fixed platelets, megakaryocytes and megakaryocytic cells via
immunofluorescence microscopy. Protocols covered include: (1) isolation and preparation of cells suitable
for fluorescence staining; (2) staining with antibodies and other molecules; (3) imaging via spinning-disc
confocal and structured illumination laser fluorescence microscopy; (4) processing and presentation of
images. Also included is a list of primary antibodies we have validated for use in staining specific proteins
and subcellular structures in platelets and megakaryocytes.
Key words Platelet, Megakaryocyte, Immunofluorescence, Microscopy, Confocal, Laser, Cell imaging,
Spinning-disc, Structured illumination
1 Introduction
Microscopy has long played a key role in studies of platelets and

megakaryocytes. In the late nineteenth century, Bizzozero used
microscopic analysis of blood in vivo and in vitro to establish the
existence of platelets and observe their role in blood coagulation
[1]. Early in the twentieth century, Wright developed new meth-
ods for staining and preparing fixed tissues, and in pioneering
microscopic examinations he established megakaryocytes as the
source of platelets and identified key morphological characteristics
including platelet granules and the megakaryocyte pseudopods
that give rise to platelets [2], now known as proplatelets [3].
Subsequent studies have employed electron, optical and fluores-
cence microscopy to extend Wright’s observations by delineating
fine details of megakaryocyte and platelet structure [4], character-
ize disorders [5, 6] and examine platelet formation, cytoskeletal
changes, granule migration and other cellular processes associated
with thrombopoiesis [7].
13
14 Fred G. Pluthero and Walter H. A. Kahr
The application of fluorescence microscopy to studies of plate-

lets and megakaryocytes has been well covered in this series.
Notably, Thon and Italiano provided an extensive discussion of the
culturing and manipulation of megakaryocytes and the imaging of
live and fixed cells [8]. Here we describe methods for rapidly
obtaining high resolution cell images via immunofluorescence
microscopy (IFM) of fixed platelets (Figs. 1 and 2), megakaryo-
cytes (Figs. 3 and 4) and megakaryocytic cells, using laser-
illuminated spinning-disc confocal (SDC; Figs. 1 and 3) or
structured illumination (SIM; Figs. 2, 4, and 5) modalities. These
methods and the images generated with them have proven useful
in examining cellular structures and proteins in several studies,
which have characterized platelet and megakaryocyte contents, [9]
structure, [10, 11] functions, [12, 13] and development [14–18]
in normal cells and those derived from individuals with disease-
associated gene variants [10, 14, 17, 19, 20]. We also present a list
of antibodies suitable for IFM of platelets and megakaryocytes
(Table 1) that have been validated in published reports and/or our
own studies (Fig. 5).
Fig. 1 Human platelet imaged by spinning disc confocal laser fluorescence microscopy. Mid-cell Z-section of
a fixed and permeabilized resting platelet lying flat, shown in ZY and XY profiles for four channels plus all
merged (bottom center). Cell structures stained were the plasma membrane (blue; detected with mouse anti-
CD61/integrin β3 and anti-mouse Alexa Fluor 405), peripheral cytoskeletal ring (magenta; detected with rabbit
anti-α tubulin Alexa Fluor 647), α-granule membranes (green; detected with goat anti-P-selectin and anti-goat
Alexa Fluor 488) and cytoplasm (red; detected with rabbit anti-myosin IIA/MYH9 and anti-rabbit Alexa Fluor
546). Bars = 1 μm
Imaging Platelets and Megakaryocytes by High-Resolution Laser Fluorescence… 15
Fig. 2 Mouse platelets imaged by structured illumination laser fluorescence microscopy. Rendered 3D images
of fixed and permeabilized resting platelets lying flat, showing 2 channels plus 3 merged (right). Cell structures
stained were the plasma membrane (magenta; detected with rat anti-CD41/integrin α2b and anti-rat Alexa
Fluor 488), α-granule membranes (green; detected with goat anti-P-selectin and anti-goat Alexa Fluor 647)
and endocytosed α-granule cargo fibrinogen (red; detected with rabbit anti-fibrinogen and anti-rabbit Alexa
Fluor 555). Bar = 1 μm
Fig. 3 Mouse megakaryocyte imaged by spinning disc confocal laser fluorescence microscopy. Mid-cell
Z-section of a fixed and permeabilized mature cell, shown in ZY and XY profiles for four channels plus all
merged (bottom center). Cell structures stained were the plasma membrane (green; detected with rat anti-
CD41/integrin α2b and anti-rat Alexa Fluor 488), nascent α-granule membranes (magenta; detected with goat
anti-P-selectin and anti-goat Alexa Fluor 647), endoplasmic reticulum/nascent platelet sarcoplasmic reticulum
(red; detected with rabbit anti-ERP57 and anti-rabbit Alexa Fluor 546) and nuclear DNA (blue; detected with
Hoechst 33,342). Bars = 5 μm
Fig. 4 Human megakaryocyte imaged by structured illumination laser fluorescence microscopy. Rendered 3D
images are shown for a permeabilized mature cell lying flat, for 4 channels plus all merged (bottom center).
Cell structures stained were the plasma membrane and demarcation membrane system (magenta; detected
with rabbit anti-CD61/integrin β3 and anti-mouse Alexa Fluor 647), microtubular cytoskeleton (green; detected
with mouse anti-α tubulin and anti-mouse Alexa Fluor 488), megakaryocyte-synthesized α-granule cargo in
multivesicular bodies and nascent granules (red; detected with sheep anti-VWF and anti-sheep Alexa Fluor
555) and nuclear DNA (blue; detected with Hoechst 33,342). Bar = 5 μm
Fig. 5 Validation of two anti-fibrinogen antibodies for IFM of human platelets imaged by structured illumination
laser fluorescence microscopy. Rendered 3D images of fixed and permeabilized resting platelets lying flat,
showing 2 channels plus 3 merged (right). Cells were stained for plasma membrane CD61/integrin β3
(magenta) and with rat (green) and sheep (red) anti-fibrinogen antibodies (Table 1) to confirm their common
staining specificity. Note that the patterns do not overlap perfectly. Bar = 2 μm
2 Materials
2.1 Platelet Isolation 1. Acid citrate dextrose (i.e., ACD-A; blood bank ACD):
and Preparation 29.9 mM trisodium citrate, 113.8 mM d-glucose, 72.6 mM
NaCl, and 2.9 mM citric acid [pH 6.4].
2. Safety winged blood collection set (e.g., 21Gx3/4″ with
30 cm tubing).
3. 1–10 mL sterile polypropylene syringes.
4. 25G needles for murine cardiac puncture.
5. Plasticware, all polypropylene: Centrifuge tubes: 15 mL (e.g.,
conical tubes) to 1.5 mL (e.g., microcentrifuge tubes, prefer-
ably with round bottom). Transfer pipettes: 0.5–5 mL; micro-
pipette tips: 10, 200, and 1000 μL; when using 200 μL tips for
platelet suspensions or hybridization solution it is necessary to
clip the tip to create a larger bore to prevent cell damage and
allow even spreading.
6. Paraformaldehyde (PFA): 16% solution.
7. Immunology-grade bovine serum albumin fraction V (BSA).
8. Phosphate buffered saline solution (PBS).
9. 6-Well tissue culture plates with lids.
10. Cover slips: 1.5 thickness; 25 mm square (platelets) or 18 mm
diameter (megakaryocytes). Optimal thickness may vary with
objectives used.
11. Human fibrinogen 200 μg/mL in PBS.
12. Poly-l-lysine solution (0.1% in water).
13.
Abciximab: glycoprotein IIb/IIIa receptor antagonist
(ReoPro, Eli Lilly Canada, Toronto, ON).
14. U46619: thromboxane A2 receptor agonist.
15. Triton X-100.
2.2 Megakaryocyte 1. Matrigel (BD, Canada) diluted 1:6 with Dulbecco’s modified
Preparation Eagle’s medium.
2. 12-Well tissue culture plates, sterile with lids.
2.3 Cell Labeling 1. Donkey serum.

and Staining 2. IF buffer (IFB): 1% BSA and 2% donkey serum in PBS, filter
sterilized. For preblocking/permeabilization buffer add 0.2%
(v/v) Triton-X100.
3. Validated primary antibodies: see Table 1; prepared in IFB.
4. Secondary antibodies (Available from a variety of suppliers.
The data presented in this chapter were obtained using
reagents from Invitrogen; Fisher Scientific, Ottawa, ON):
donkey or goat antibodies specific for mouse, rabbit, goat or
Table 1
18
Antibodies validated for platelet and megakaryocyte IF staining. Listed by type and showing antigen specificity, working dilution (v/v) for IF,
commercial source, product code, cellular structures stained, claimed/verified applicability to staining human or mouse cells, and references for use
in IFM of platelets and/or megakaryocytes. Note: some antibodies may also work for mouse cells, but have not yet been tested
Type Antigen (Clone) Dilution Source Product Code Structure Stained Human Mouse Reference
Rabbit polyclonal
ARPC1B 100 Sigma-Aldrich/Prestige HPA004832 Arp2/3 complex X X [20]
ARPC5 200 Abcam ab118459 Arp2/3 complex X [11]
CALRETICULIN 200 Abcam ab39897 Endoplasmic reticulum X X [18]
DYNAMIN I/II 50 Cell signaling 2342 Cytoplasm X X [15]
ETV6 100 Sigma-Aldrich/Prestige HPA000264 Nucleus X [17]
FIBRINOGEN 100 Agilent/Dako A0080 Alpha granule cargo X X
NONMUSCLE MYOSIN IIA 500 Biomedical technologies BT-567 Cytoplasm X X [12]

PACSIN2 100 Abcam ab37615 Internal membrane X X [16]
ROBO1 500 Rockland 600-401-692 Cell surface X X [13]
THROMBOSPONDIN 1 2000 RayBiotech MD-01-0027 Alpha granule cargo X
VON WILLEBRAND FACTOR 400 Dako A0082 Alpha granule cargo X X [18]
Rabbit monoclonal
ALPHA TUBULIN (11H10) 300 Cell signaling 5046 (AF 647) Cytoskeletal ring X X [18]
Mouse monoclonal
CALNEXIN 400 Abcam ab31290 Endoplasmic reticulum X X [18]
CD42B-GP1BA 200 BD Pharmingen 562255 Cell membrane X [16]
CD61-INTEGRIN B3 (Y2/51) 200 Dako M 0753 Cell membrane X [10]
CD62P/P-SELECTIN 100 Lifespan biosciences LS-C127810 Alpha granule membrane X [14]
CD63/LAMP3 100 BD Pharmingen 556019 Dense granules/lysosomes X [19]
DYNAMIN I 100 BD transduction labs 610264 Cytoplasm X X [15]
FILAMIN A 200 Millipore MAB1678 Inner cell membrane X X [16]
GM130 50 BD transduction Clone 35 Cis-Golgi X
LAMP1 25 Hybridoma Bank H4A3 Lysosomes X [19]
PDI 50 Abcam ab2792 Endoplasmic reticulum X X [18]
RETINOIC ACID RECEPTOR A 100 Millipore MAB5346 Cell membrane X [11]
SERCA3 (PL/IM430) 100 Santa Cruz biotechnology Sc-81759 Endoplasmic reticulum X [18]
THROMBOSPONDIN 1 100 R&D systems MAB3074 Alpha granules X [20]
ALPHA TUBULIN (B-5-1-2) 2000 Sigma-Aldrich T6074 Cytoskeletal ring X [14]
VON WILLEBRAND FACTOR 50 Agilent/Dako M 0616 Alpha granule cargo X
Goat polyclonal
CD62P/P-SELECTIN 100 Santa Cruz biotechnology Sc-6941 Alpha granule membrane X X [16]
Sheep polyclonal
FIBRINOGEN 200 Affinity Biologicals SAFG-AP Alpha granule cargo X X
TGN46 500 ABD Serotech AHP500 Trans-Golgi X
VON WILLEBRAND FACTOR 1000 ABD Serotech AHP062 Alpha granule cargo X X
Rat anti-mouse polyclonal
CD41 200 eBioscience 14-0411 Cell membrane X [18]
Imaging Platelets and Megakaryocytes by High-Resolution Laser Fluorescence…
19
rat IgG conjugated with Alexa Fluor 647 (far red), 568, 555
or 546 (red), 488 (green) or 405 (ultraviolet); prepared
1:1000 (v/v) in IFB.
5. Phalloidin conjugated with Alexa Fluor (Invitrogen); prepared
1:300 (v/v) in PBS.
6. Wheat germ agglutinin conjugated with Alexa Fluor
(Invitrogen); prepared 10 μg/mL in PBS.
7. Hoechst 33342 dye for nuclear DNA; prepared 1:5000 (v/v)
of stock solution in PBS.
8. Fluorescent mounting media: available from a variety of sup-
pliers. Data presented in this chapter used either ProLong
Gold or Diamond antifade archival mountant (Invitrogen).
2.4 Cell Imaging 1. Image acquisition: A variety of imaging acquisition and pro-
cessing systems can be used for IFM of platelets and mega-
karyocytes. We have achieved good results and reasonable
workflow rates using SDC and SIM laser fluorescence micros-
copy acquisition systems. In our SDC system, image acquisi-
tion is controlled via Volocity software (PerkinElmer, Waltham,
MA); our SIM system utilizes Zen software (Carl Zeiss Canada,
Toronto, ON) for acquisition, reconstruction and channel
registry correction.
2. SDC: A range of suitable systems are available. The system
used here (Figs. 1 and 3) is constructed on an Olympus IX81
inverted microscope core (Quorum Technologies, Puslinch,
ON) fitted with a Hamamatsu C9100-13 back-thinned
EM-CCD camera (512 × 512 pixels), Yokogawa CSU X1
spinning disk confocal scan head with Spectral Borealis
upgrade, 4 diode-pumped solid state laser lines (405 nm,
491 nm, 561 nm, 642 nm), emission filters specific for Alexa
Fluor dyes: 405 (447 nm ±60), 488 (525 nm ±50), 568
(593 nm ±40), and 647 (676 nm ±29), and an ASI motorized
XY stage controlled with an Improvision Piezo Focus Drive.
3. SIM: A range of suitable systems are available. The one used
here (Figs. 2, 4, and 5) is a Zeiss ELYRA PS.1 microscopy
system with an Axio Observer Z1 core using a 63×/1.4 NA
oil-immersion objective with 1.6× optovar at 30 °C. The sys-
tem is equipped with an Andor iXon3 885 detector, 405, 488,
561, and 640 nm laser lines, Zeiss motorized XY stage and
Z-piezo focus. This allows for images to be acquired with five-
fold rotation at stepping required for optimal reconstruction
of all channels (typically 91 nm). Acquisition control, SIM
image reconstruction and channel alignment were performed
with Zen 2012 software, using optimized settings and current
calibration data sets established by multicolor bead imaging.
4. Image processing software: SDC and SIM images can be pro-
cessed with commercially available software such as Volocity,
Imaris or other software depending on the analysis required

(e.g., for Figs. 2, 4 and 5 image files were exported in czi format
to Imaris 8 for 3D rendering and preparation of snapshots).
3 Methods
Platelets are easily obtained from blood, and in most cases large
volumes are not required and processing is rapid (see Note 1). For
example, 10 mL of normal human blood is sufficient to prepare
150 or more platelet preparations suitable for IFM, while 1 mL of
mouse blood can yield 40 or more preparations. When sufficient
care is taken during the harvesting and fixing of platelets, it is usu-
ally possible to obtain preparations containing uniform popula-
tions of resting cells, or of activated/spread platelets if desired.
Given the sensitivity of platelets to a range of environmental fac-
tors, it is important to avoid activating them inadvertently and it is
useful to confirm that preparations contain cells in the desired state
prior to using them for extensive analysis (see Note 2).
Megakaryocytes and megakaryocytic cells must be cultured to
produce sufficient numbers of cells for imaging mature stages of
development [14, 17]. The size of the populations obtained from
cultures and the distribution of cells among different stages can
vary considerably from donor to donor and from batch to batch. As
with platelets, it is useful to assess each batch of megakaryocyte
preparations prior to undertaking a large-scale analysis (see Note 2).
3.1 Isolation When handling platelets it is important to avoid some common

of Platelet-Rich aspects of blood collection and cell processing. These include
Plasma exposure of blood to glass surfaces (e.g., collection tubes), small-
bore needles/pipettes, temperatures outside of the normal work-
ing range (21–37 °C), rapid or repeated changes in temperature,
exposure to vacuum during collection, excessive agitation (e.g.,
transport through pneumatic tube systems) and delayed anticoag-
ulation. Commonly used anticoagulants such as EDTA and hepa-
rin are not desirable for platelet work. For preparing platelets from
human or murine blood it is best to anticoagulate with acid citrate
dextrose (ACD). If clinical samples are being examined, sodium
citrate can be used although it is necessary to watch for signs of
platelet aggregation (e.g., “clumping out”). Hirudin and its deriv-
atives can also be used at concentrations determined by the prepa-
ration. We have found that adding corn trypsin inhibitor (20 μg/
mL) to blood does not influence the quality of platelet prepara-
tions, although we do not use it as the sole anticoagulant.
1. Human blood collection: collect blood without vacuum in
accordance with local ethics guidelines. For adults, we use a
21G (or larger) needle with a butterfly collection set attached
to a removable 10 mL polypropylene syringe. After an initial
discard of 5 mL, new syringes are used to collect blood that is
a liquoted into polystyrene or polypropylene tubes containing

anticoagulant to final concentrations of 14% for ACD (i.e., 6
parts blood to 1 part ACD), or 3.2% for sodium citrate (i.e.,
to the fill line in citrate blood collection tubes). After closure,
each tube is gently inverted 5 times and allowed to sit at room
temperature for at least 30 min to allow complete inactivation
of clotting factors.
2. Mouse blood collection: blood is obtained by cardiac punc-
ture with a 25G needle and rapidly added to polypropylene
tubes containing sufficient ACD to produce a minimum anti-
coagulant concentration of 10% by volume.
3. Preparation of human platelet-rich plasma (PRP): Centrifuge
blood from normal donors (see Note 1) at 150 g for 15 min at
room temperature in a swinging-bucket centrifuge. The upper
2/3 of the plasma fraction represents clean PRP that can be
used to prepare washed platelets for preparing lysates or per-
forming experiments. The entire plasma fraction is suitable for
preparing platelets for imaging, provided there are no con-
cerns that other cell types may be present in the final
preparation.
4. Preparation of mouse platelet-rich plasma: dilute blood 1:1
with PBS + 14%ACD prior to centrifugation and collection of
PRP as described in Subheading 3.1, step 3. It is possible to
use a variable-speed microcentrifuge for this step if care is
taken not to disturb the blood separation when removing the
tubes from the rotor. Dilution of mouse blood is necessary
because of its naturally high hematocrit.
5. The final concentration of platelets obtained in PRP will vary
depending on several factors, including the donor. When
automated cell counting is available the platelet concentration
can be directly determined. When an automated count of the
source blood is available, the PRP count will normally be
higher. When counts are not available, it can be assumed that
normal human PRP contains 200–300 platelets/nL for male
donors, and 300–400/nL for female donors (see Note 1).
Concentrations in mouse samples will vary more widely owing
to differences in strains and the amount of dilution that occurs
during sample preparation. Mouse PRP prepared with the
protocol described above should nevertheless have platelet
concentrations higher than the normal human range of 200–
400 platelets/nL.
3.2 Preparation Some electron microscopy protocols recommend fixation of plate-

of Resting Platelets lets directly in blood, but we have found that high quality IFM
preparations of resting human or mouse platelets can be obtained
by fixing platelets in PRP.
1. Aliquot PRP (human or mouse) to half-fill microcentrifuge

tubes, and add an equal volume of PBS plus 8% PFA to each.
Close, invert 5 times and incubate for 10 min at RT.
2. Centrifuge at 1000 × g for 10 min. Draw off the supernatant.
3. Gently resuspend platelets in the maximum volume of PBS the
centrifuge tube will hold using a 1–2 mL transfer pipette or
1 mL micropipette tip. Freshly fixed platelets are likely to form
a film on the side of the tube rather than a visible pellet, so care
must be taken to gently wash them from the side.
4. Repeat steps 2 and 3; repeat step 2.
5. Resuspend the washed platelet pellet at 100–200 cells/nL
(i.e., half the original PRP concentration for normal human
donor samples) in PBS plus 1% BSA.
6. Put one glass coverslip in the well of a 6-well tissue culture
plate for each 50 μL of platelet suspension. Place wads of tis-
sue in the spaces between the wells and wet with water; keep
these wet for the entire period prior to final mounting of prep-
arations. Replace the lids and put plates on a slide warmer set
to 37 °C.
7. Spot 50 μL aliquots of platelet suspension onto coverslips
using a transfer pipette or 200 μL micropipette tip with the
end clipped.
8. Incubate for 90 min at 37 °C. If a slide or plate warmer is not
available, the plates can be floated in a warm water bath.
Incubation can also be done at room temperature for 3 h, or
at 4 °C overnight. In our experience the best results for nor-
mal platelets are obtained with 37 °C incubation.
9. If platelets are very large and/or few in number (see Note 1)
the incubation time can be extended. The likelihood of plate-
lets adhering can also be improved by spreading 100 μL of
poly-l-lysine solution on the center of the coverslips, leaving it
for 5 min prior to removal by aspiration, and allowing cover-
slips to dry for at least 30 min.
10. After incubation, draw off excess liquid and gently rinse platelet
preparations with PBS by adding buffer along the side of the
wells (never direct a stream at the cells) and drawing it off.
Platelets can be stained immediately (e.g., to assess quality, see
Note 2) or stored in the wells of covered plates sealed in plastic
wrap or bags at 4 °C. Preparations will last for at least 14 days if
not contaminated or allowed to dry out (see Notes 3 and 4).
3.3 Preparation 1. Prepare glass coverslips in 6-well culture plates as in Subheading

of Spread 3.2, step 6 above. Cover the slips with a solution of 200 μg/
and Activated mL human fibrinogen in PBS for 30 min at 37 °C. Draw off
Platelets fibrinogen, rinse sequentially with PBS and water, and allow to
dry (a similar method can be used to coat coverslips with other

platelet agonists such as collagen). Prewarm coverslips at
37 °C. Platelets can be allowed to spread on agonist-treated
surfaces in PRP or after washing [20].
2. Aliquot PRP into microcentrifuge tubes.
3. Centrifuge at 1000 × g for 10 min. Draw off supernatant.
4. Resuspend platelets in PBS adjusted to pH 6.4 with ACD.
5. Repeat step 2. Repeat step 3. Repeat step 2.
6. Resuspend washed platelets in the original PRP volume of
physiological buffer (e.g., Tyrode’s pH 7.4).
7. For platelet spreading: spot 50–100 μL aliquots of PRP or
washed platelet suspension onto agonist-treated coverslips
with a transfer pipette or 200 μL micropipette tip with the end
clipped.
8. Incubate for 10–90 min at 37 °C. After the desired time fix
cells in place by adding an equal volume of 8% PFA/PBS and
incubating for 10 min.
9. Rinse coverslips twice with PBS and use preparations immedi-
ately or stored as described above.
10. To prepare platelets activated in suspension, add abciximab
(20 μg/mL v/v) to washed platelets in Tyrode’s buffer to pre-
vent platelet aggregation and stimulate with a suitable agonist
at a sufficient concentration for enough time to achieve the
desired level of activation. For example U46619 at 2 μM typi-
cally produces maximal platelet activation after 10 min at
37 °C [18]. Then fix the platelet suspension and process as for
PRP in Subheading 3.2, step 1.
3.4 Preparation Human and murine megakaryocytes and megakaryocytic cells can
of Megakaryocytes be cultured and induced to express megakaryocyte phenotypes via
and Megakaryocytic a variety of methods.
Cells
1. Culturing is typically done in medium in suspension until cells
are approaching maturity (e.g., for mouse cells 3 days, for
human cells 8 days) at a concentration of approximately 105
cells/mL. When cells are at that point the following steps
should be followed (see Note 4).
2. Prepare coverslips by coating with Matrigel diluted 1:6 with
DMEM (500 μL) in wells of 12-well culture plates, then over-
lay each with 1.5 mL cell suspension.
3. Incubate for 2 or more days under tissue culture conditions
(e.g., 37 °C, 5% carbon dioxide).
4. Draw off culture medium and fix immobilized cells with 4%

paraformaldehyde in PBS for 30 min.
5. Draw off fixative and rinse cells twice in PBS. Preparations can
be kept under PBS and stored at 4 °C, or used immediately
(see Notes 2 and 3).
3.5 Fluorescence 1. Preblocking and permeabilization: Platelet and megakaryo-

Staining of Platelets cyte preparations can be stained without or with permeabiliza-
and Megakaryocytes tion. Cells are permeabilized and preblocked with 0.2% Triton
X-100 in IFB for 60 min, and then rinsed with PBS. Preblocking
only is done with IFB for 60 min. For platelet spreads these
solutions can be overlain as a surface droplet (approx. 100 μL);
for megakaryocytes it is best to fill the well sufficiently to cover
the entire preparation (approx. 250 μL).
2. Preparation of primary antibodies: add antibodies to IFB at
working concentrations in appropriate amounts (e.g., 50 μL/
platelet preparation; 250 μL/megakaryocyte preparation) in
combinations that allow for specific detection by secondary
antibodies. For example a commonly used combination for
human megakaryocytes is mouse anti-CD61 + rabbit anti-
VWF + goat anti-P-selectin. If monoclonal and/or affinity-
purified antibodies verified for IF are used, nonspecific primary
antibody controls are usually not necessary, although it is use-
ful to perform them when trying a new antibody, cell prepara-
tion method or imaging system. Since mouse megakaryocytes
and platelets lack FcγRII receptors, it is possible to use mouse
monoclonal IgG antibodies to stain them.
3. Primary hybridization: Overlay cell preparations with primary
antibody solution and incubate overnight at
4 °C. Megakaryocyte preparations should be gently agitated
on a rocking platform if possible.
4. Washing: Draw off primary antibody solution and rinse quickly
with PBS. Wash cell preparations at least 3× with ample vol-
umes of PBS (5 min/wash) under gentle agitation (e.g., on a
rotating platform). Thorough washing is especially important
for MK preparations, since it takes time for antibodies to dif-
fuse throughout the cells and culture matrix.
5. Secondary antibody hybridization: prepare species-appropri-
ate secondary antibodies conjugated to high-performance
fluorophores (e.g., Alexa Fluor dyes) in IFB at recommended
working concentrations (typically 1:1000 v/v). Antibodies
commonly used are of donkey or goat origin specific for
mouse, rabbit, rat, goat, or sheep conjugated with Alexa
Fluor 647, 568, 555, 488, or 405. Note that goat and sheep
secondary antibodies often stain primary antibodies from
both species. Incubate platelet preparations for 1–2 h and
megakaryocytes for 3–4 h at RT. After incubation rinse and

wash preparations as for primary antibodies (Subheading 3.5,
step 4; see Notes 5 and 6).
6. Staining with fluorescent antibodies and stains: directly con-
jugated primary antibodies can be used at any step where
appropriate, including a tertiary hybridization step if a sec-
ondary antibody against the species has been employed.
Staining for nuclear DNA is done by suspending DAPI or
Hoechst 33342 1:5000 in PBS and overlaying megakaryocyte
preparations for 30 min. Staining with wheat germ aggluti-
nin, phalloidin or other stains diluted in PBS can be per-
formed in the same manner and at the same time, using
concentrations according to the manufacturer’s instructions
for fixed cells. After incubation, preparations are washed as
above (see Subheading 3.5, step 4).
7. Post-fixation: stained platelet and megakaryocyte preparations
are post-fixed by overlaying with 4% paraformaldehyde in PBS
for 5 min. Preparations are then given 3 final rinses in PBS and
kept overlaid with PBS until mounting.
8. Mounting: draw off PBS using a transfer pipette tip attached
to vacuum placed beside the cell preparation until it just
appears to be dry. Pick up the edge of the coverslip with a
needle and lift with forceps. Drain excess buffer on a tissue and
bring cell-side down next to a droplet of fluorescent medium
placed on a microscope slide. Lower the coverslip without
generating bubbles or transverse motion; do not apply pres-
sure to the upper surface. Given the fragility of coverslips it is
useful to stain duplicate sets, which can both be mounted on
the same slide.
9. Mounted preparations are stored lying flat in the dark under
appropriate conditions for sufficient time for the mounting
medium to set (e.g., 24 h at room temperature for ProLong
media) prior to imaging.
3.6 Cell Imaging Once preparations of suitable quality (see Notes 2 and 7) are in
hand, a typical imaging workflow would begin by acquiring a set
of representative SDC fluorescence images for each stained set.
This allows fairly rapid acquisition of multichannel laser fluores-
cence images that can be used on their own merits (see Notes 8
and 9). For example, the images shown in Fig. 1 were acquired
with 250 nm Z-stepping through a 100×/1.40 NA oil objective
and a 1.5× internal magnification lens (Spectral Applied Research)
for a final magnification of 150×, and for Fig. 3 a 60× oil objective
was used for a final magnification of 90×. Laser intensity, camera
and exposure settings were established to produce minimal/unde-
tectable levels of autofluorescence, channel cross talk, and non-
specific background fluorescence. Acquisition, image
deconvolution, and registry correction (maximum one z-pixel

required) were done with Volocity 6 software. Image files were
exported as tiff files for assembly, or exported in ome format for
further analysis with Imaris 8 software (see Notes 10 and 11).
SDC images can also serve as a quality control to determine
whether further information can be gained via SIM, which places
much greater demands on sample quality. For example capturing a
5-axis, 4-channel, 40–50 μm Z-stack depth laser SIM image of a
megakaryocyte with 91 nm Z-stepping (Fig. 4) exposes the sample
to much greater risks of photobleaching and/or photodamage
than acquiring a four-channel laser SDC image with 250 nm step-
ping (Fig. 3). SIM images also require reconstruction and channel
alignment, which can take as much time as the initial acquisition
(although images can be batch-processed during off hours). Thus
while it may seem desirable to always try and get the highest-
resolution images possible, there are usually practical trade-offs to
consider (see Notes 12–14).
4 Notes
1. When macrothrombocytes are present in the sample, the cen-

trifugation force can be lowered to 100 × g and the centrifuga-
tion time extended to 30 min. It is also advisable to collect the
interphase fraction, which can be used to image platelets and
leukocytes (e.g., for examination in MYH9 syndrome). If
platelets are extremely large, blood samples can be left upright
for 2–3 h, then “settled PRP” can be harvested without
centrifugation.
2. It is useful to check the quality of a batch of cell preparations
to ensure that they are worth the time and expense of IFM
analysis. For human platelet or megakaryocyte preparations
this can be quickly done by incubating test preparations for
60 min at room temperature in IFB + 0.2% Triton X-100 con-
taining fluorescently labeled wheat germ agglutinin (10 μg/
mL; stains cell surface and internal structures), phalloidin
(1:300 v/v; stains internal F-actin) and rabbit anti-tubulin
(1:300 v/v; stains cytoskeletal ring in platelets and proplatelet
extensions), plus for megakaryocytes, Hoechst 33342
(1:5000 v/v). After incubation the cell preparations are
washed (above) and mounted using fast-setting mountant
(e.g., SlowFade Gold). The preparations can then be immedi-
ately checked via standard fluorescence or SDC microscopy
for cell abundance and the presence of desired morphological
characteristics, such as intact cytoskeletal tubulin rings for
resting platelets (Fig. 1) or the presence of large multilobed
nuclei in mature megakaryocytes (Figs. 3 and 4). Another
quick stain for human platelets is fluorescently labeled
anti-human IgG. Mouse cells do not stain well with either

wheat germ agglutinin or anti-mouse IgG.
3. The tissue culture plates used for making and staining platelet
preparations can be reused after washing. Fixed megakaryo-
cyte preparations can also be transferred to nonsterile/reused
plates for staining.
4. The platelet and megakaryocyte preparation protocols
described here do not involve cytospinning, because we have
found that this treatment introduces structural distortions,
especially in mature megakaryocytes. We also do not use the
“hanging drop” hybridization method, since it involves exces-
sive manipulation of preparations and does not allow for over-
night hybridization with 100% humidity.
5. If secondary antibody staining is weak, it is better to extend
hybridization time rather than increase secondary antibody
concentration, because this can greatly increase the nonspe-
cific background signal. Secondary antibody-only controls
should be done for each set of preparations.
6. Most endogenously expressed fluorescent tags (e.g., GFP) are
insufficiently bright and/or persistent for high-quality IFM
imaging. It is necessary to enhance their signal using an anti-
body with a stronger fluorophore (e.g., anti-GFP conjugated
with Alexa Fluor 488).
7. If stained samples do not appear extremely bright when exam-
ined through the eyepiece on a fluorescence microscope (e.g.,
they cannot be easily seen at the lowest light setting), imaging
them by laser fluorescence is unlikely to yield useful images.
Only samples that can be imaged by SDC with short exposure
times, low laser intensity, and neutral gain settings are worth
imaging via SIM.
8. Obtaining high-quality IFM images requires well-prepared
samples, careful use of a properly aligned and calibrated acqui-
sition system, and appropriate image analysis. It is also essen-
tial to be confident in the specificity of the antibodies and
stains used to detect particular biomolecules. Well-
characterized antibodies can be validated from publications of
their originators and/or databases such as the Human Protein
Atlas (http://www.proteinatlas.org). Manufacturer descrip-
tions are insufficient for validation, but publications citing the
antibody and online user reviews can confirm specificity.
Regardless of the source, it is always useful to confirm the
specificity of an antibody under the conditions where it is to be
used. This can be done (1) via immunoblot detection of
appropriate-sized protein bands in lysates from the cells to be
stained, and (2) by comparing the IF staining pattern of the
tested antibody to the pattern obtained with a validated
antibody. For example, the rabbit anti-fibrinogen antibody

used in Fig. 2 was validated by immunoblotting of megakaryo-
cytic cells lysates and by costaining human platelets with an
IF-
validated sheep antibody (Fig. 5) and observing strong
overlap of staining patterns. If a protein is of central interest to
a study, it should be imaged with at least two different vali-
dated antibodies to be certain of its subcellular localization.
For the benefit of researchers in the field, Table 1 lists IFM-
validated antibodies we have used for detecting platelet and
megakaryocyte proteins and structures in various studies. The
sets of anti-P-selectin, thrombospondin 1, tubulin and VWF
antibodies listed all produce the same IFM staining patterns.
9. In our experience it is seldom worth the time and risk of pho-
tobleaching to acquire differential interference or phase con-
trast images of fixed platelets or megakaryocytes.
10. A common failing of platelet and megakaryocyte IFM studies
is the failure to show appropriate controls for intracellular
colocalization of molecules, granules and other structures. If
two different proteins show identical IFM patterns, it is impor-
tant to show images of cells stained and imaged separately for
each protein to confirm their distribution patterns under the
conditions used for imaging costained cells. In our experience
“perfect” colocalization where all double-stained spots match
in size and intensity is a highly suspicious result. It is possible
to come close to producing this using two different secondary
antibodies to stain the same primary antibody, or two different
primary antibodies to stain the same protein, but there are
usually a few unmatched spots and visible variations in signal
intensity (Fig. 5). Perfect colocalization is usually an artifact of
the same fluorophore being detected in both channels, which
is relatively easy to do using laser fluorescence. If the exposure
settings used for two channels are strikingly different, then
cross talk is the most likely explanation for apparent
colocalization.
11. Another common deficiency of platelet and megakaryocyte
IFM studies is the failure to consider the three-dimensional
aspects of images. Most confocal microscopy techniques have
lower resolution in the Z-plane relative to XY, but that infor-
mation can still be useful. For example normal resting platelets
are flat, about 1 μm thick and have intact peripheral tubulin
rings (Fig. 1). If Z-sections show cells that are thick or round,
they were probably activated. If platelets appear to be extremely
thin, they have likely become distorted or dried, or they may
be membrane remnants stuck to the glass, which like other cell
“ghosts” will bind many primary/secondary antibodies
nonspecifically.
12. Presenting high-quality IFM images in publications can be

challenging. The standard approach of putting all the channels
in a row followed by the merged image may work for two-
channel images, but for a four-channel image this reduces each
panel to a tiny and visually uninformative square (especially for
megakaryocytes). Unfortunately, many journals, editors, and
reviewers are uncomfortable with alternative modes of presen-
tation such as double row images (e.g., Figs. 1, 3, and 4). A
useful approach to dealing with such shortcomings, as well as
the challenge of showing a three-dimensional image using a
2-dimensional format, is to include online videos of key images
that show 3D renders and various combinations of channels,
as we have done in several studies [17, 18].
13. If scale designations, channel labels and other text on figures
are not converted to graphical elements in figures, be prepared
to spend considerable time inspecting and repairing unwanted
alterations introduced by publishers during manuscript
preparation.
14. We have found that deconvolved multi-axis images of

Z-sections (Figs. 1 and 3) can be rapidly generated with
Volocity, while Imaris does an excellent job of generating ren-
dered multichannel 3D volume images (Figs. 2, 4, and 5) and
rotation videos. Single images can be exported at the highest
resolution available for final assembly and labeling using
Adobe Photoshop or other software. Image analysis can be
performed with a variety of software. For example most image
analysis packages support voxel colocalization using appropri-
ate thresholding to generate co-occurrence/correlation coef-
ficients (e.g., Pearson’s correlation coefficient, Manders
overlap coefficients).
Acknowledgments
This work was supported in part by operating grants from the

Canadian Institutes of Health Research MOP-81208 and MOP-
119450 (W.H.A.K.). The authors also thank Michael Woodside
and Paul Paroutis of the Imaging Facility at the Hospital for Sick
Children (Toronto, ON) for their invaluable assistance.
References
1. Brewer DB (2006) Max Schultze (1865), 3. Italiano JE Jr, ., Lecine P, Shivdasani RA,
G. Bizzozero (1882) and the discovery of the Hartwig JH. Blood platelets are assembled
platelet. Br J Haematol 133(3):251–258 principally at the ends of proplatelet processes
2. Wright JH (1910) The histogenesis of the produced by differentiated megakaryocytes.
blood platelets. J Morphol 21(2):263–278 J Cell Biol 1999;147(6):1299-1312
4. White JG (1987) An overview of plate- 13. Patel S, Huang YW, Reheman A, Pluthero FG,
let structural physiology. Scanning Microsc Chaturvedi S, Mukovozov IM et al (2012)
1(4):1677–1700 The cell motility modulator Slit2 is a potent
5. White JG (1998) Use of the electron micro- inhibitor of platelet function. Circulation
scope for diagnosis of platelet disorders. Semin 126(11):1385–1395
Thromb Hemost 24(2):163–168 14. Kahr WH, Lo RW, Li L, Pluthero FG,
6. Westmoreland D, Shaw M, Grimes W, Metcalf Christensen H, Ni R et al (2013) Abnormal
DJ, Burden JJ, Gomez K et al (2016) Super- megakaryocyte development and plate-
resolution microscopy as a potential approach let function in Nbeal2(−/−) mice. Blood
to diagnosis of platelet granule disorders. 122(19):3349–3358
J Thromb Haemost 14(4):839–849 15. Bender M, Giannini S, Grozovsky R, Jonsson
7. Machlus KR, Italiano JE Jr (2013) The incred- T, Christensen H, Pluthero FG et al (2015)
ible journey: from megakaryocyte devel- Dynamin 2-dependent endocytosis is required
opment to platelet formation. J Cell Biol for normal megakaryocyte development in
201(6):785–796 mice. Blood 125(6):1014–1024
8. Thon JN, Italiano JE (2012) Visualization 16. Begonja AJ, Pluthero FG, Suphamungmee W,
and manipulation of the platelet and mega- Giannini S, Christensen H, Leung R et al (2015)
karyocyte cytoskeleton. Methods Mol Biol FlnA binding to PACSIN2 F-BAR domain reg-
788:109–125 ulates membrane tubulation in megakaryocytes
9. Licht C, Pluthero FG, Li L, Christensen H, and platelets. Blood 126(1):80–88
Habbig S, Hoppe B et al (2009) Platelet- 17. Noetzli L, Lo RW, Lee-Sherick AB, Callaghan
associated complement factor H in healthy M, Noris P, Savoia A et al.(2015) Germline
persons and patients with atypical HUS. Blood mutations in ETV6 are associated with throm-
114(20):4538–4545 bocytopenia, red cell macrocytosis and pre-
10. Kahr WH, Hinckley J, Li L, Schwertz H, disposition to lymphoblastic leukemia. Nat
Christensen H, Rowley JW et al (2011) Genet. 47(5):535–8
Mutations in NBEAL2, encoding a BEACH 18. Crescente M, Pluthero FG, Li L, Lo RW,
protein, cause gray platelet syndrome. Nat Walsh TG, Louriero S et al (2016) Intracellular
Genet 43(8):738–740 trafficking, localization, and mobilization of
11. Rondina MT, Freitag M, Pluthero FG, Kahr platelet-borne Thiol Isomerases. Arterioscler
WH, Rowley JW, Kraiss LW et al (2016) Thromb Vasc Biol 36(6):1164–1173
Non-genomic activities of retinoic acid recep- 19. Urban D, Li L, Christensen H, Pluthero
tor alpha control actin cytoskeletal events FG, Chen SZ, Puhacz M et al (2012) The
in human platelets. J Thromb Haemost VPS33B-binding protein VPS16B is required
14(5):1082–1094 in megakaryocyte and platelet alpha-granule
12. Kahr WH, Savoia A, Pluthero FG, Li L, biogenesis. Blood 120(25):5032–5040
Christensen H, De Rocco D et al (2009) 20. Kahr WHA, Pluthero FG, Elkadri A, Warner
Megakaryocyte and platelet abnormalities in N, Drobac M, Chen CH et al (2017) Loss of
a patient with a W33C mutation in the con- the Arp2/3 complex component ARPC1B
served SH3-like domain of myosin heavy chain causes platelet abnormalities and predisposes to
IIA. Thromb Haemost 102(6):1241–1250 inflammatory disease. Nat Commun 8:14816
Chapter 3
Single-Molecule Localization and Structured Illumination

Microscopy of Platelet Proteins
Natalie S. Poulter, Abdullah O. Khan, Chiara Pallini, and Steven G. Thomas
Abstract
Superresolution microscopy has become increasingly widespread over the past 5 years and allows users to
image biological processes below the diffraction limit of traditional fluorescence microscopy where resolu-
tion is restricted to approximately 250 nm. Superresolution refers to a wide range of techniques which
employ different approaches to circumvent the diffraction limit. Two of these approaches, structured
illumination microscopy (SIM) and single-molecule localization microscopy (SMLM), which provide a
doubling and tenfold increase in resolution respectively, are dominating the field. This is partly because of
the insights into biology they offer and partly because of their commercialization by the main microscope
manufacturers. This chapter provides background to the two techniques, practical considerations for
their use, and protocols for their application to platelet biology.
Key words Platelets, Superresolution, Single-molecule localization microscopy, SMLM, Direct

stochastic optical reconstruction microscopy, dSTORM, Structured illumination microscopy, SIM
1 Introduction
Platelets are small cells, densely packed with granules which con-
tain a dynamic cytoskeleton that is key to their function. Historically,
good quality imaging of platelets has been restricted to the use of
electron microscopy (EM) which is able to give high-resolution
information of structures such as the cytoskeleton [1, 2] and alpha
and dense granules [3, 4]. Indeed, EM has been used clinically in
the diagnosis of certain granule related disorders such as Grey
Platelet Syndrome (alpha granule) and Hermansky–Pudlak syn-
drome (dense granule) [5]. While EM allows for high-quality
imaging of the ultrastructure of platelets, it is restricted to fixed
and highly processed samples. Visualization of specific proteins via
immuno-gold EM is possible, but again it is technically more chal-
lenging than the traditional cell preparation for fluorescence
microscopy. Fluorescence imaging offers many distinct advantages
33
34 Natalie S. Poulter et al.
over EM, including the ability to easily label specific proteins of

interest, perform multicolor imaging and to follow dynamic pro-
cesses in living cells [6].
Fluorescence microscopy has allowed for significant advances
in our understanding of the microscopic structure of cells, and in
particular the dynamics and relationships between multiple pro-
teins or structures of interest. However, a fundamental limitation
of fluorescence for high-resolution imaging is that, until recently,
the resolution of the image was determined by the diffraction limit
of light. First defined by Ernst Abbe in 1873, the diffraction limit
of light was described as a physical restriction on the final resolu-
tion of an image as a consequence of spherical waveforms passing
through a circular aperture [7]. Steadily improving optical config-
urations, including high numerical aperture objectives, which are
able to capture more of the photons emitted from a sample, have
brought the resolution limit down to ~250 nm in xy and ~500 nm
in z. This means that any two points of light that are closer together
than 250 nm in xy will be seen as one object instead of two when
viewed using a light microscope [8]. While widefield and confocal
imaging have provided many unique insights into cellular struc-
tures and their localization, a fundamental caveat has always existed
when studying structures beneath this limit of spatial resolution.
For example, when we consider that the size of vesicles and gran-
ules are in the region of 50–500 nm it becomes clear that colocal-
ization of two different proteins in a diffraction-limited image does
not necessarily mean that the two proteins are found within the
same compartment. These issues are compounded in platelets,
which are small, densely packaged cells where spatial relationships
more often lie beneath the diffraction limit. As such, diffraction
limited imaging of platelets often leads to false colocalizations and
overlap of fluorescent labels, making conclusions about their exact
location and function more difficult to draw. Platelets also lack a
nucleus, which makes them intractable to traditional cell biology
approaches for fluorescently tagging proteins of interest.
Within the last 15 years subdiffraction imaging, more com-
monly referred to as superresolution microscopy, has been devel-
oped. Superresolution microscopy is a term which encompasses a
number of modalities, each of which has overcome the diffraction
barrier using a different approach and therefore each one offers its
own unique advantages and disadvantages. A comparison of the
various techniques and their applications are beyond the scope of
this chapter but various recent reviews [9, 10] have described this
in detail and are worth reading before embarking on any super-
resolution experiments. In this chapter we will introduce two
superresolution approaches: single-molecule localization micros-
copy (SMLM) and structured illumination microscopy (SIM),
with a particular focus on their relevance to the study of platelets.
We will provide protocols and hardware information on how to
Super-Resolution Imaging of Platelets 35
apply these techniques to the study of the platelet cytoskeleton and

platelet signaling.
1.1 Single-Molecule SMLM methods allow for single-molecule detection and the sub-
Localization sequent compilation of these localizations to form a superresolved
Microscopy (SMLM) image [11]. SMLM techniques offer the highest spatial resolution
currently available, down to as little as 10–20 nm in xy. This
approach also offers unique quantitative benefits, particularly when
investigating protein–protein interactions. Where single molecules
are accurately detected and fitted, complex cluster analysis can be
applied to study the density of individual emitters and their spatial
relationship to one another [12, 13]. SMLM techniques also pose
a number of unique methodical challenges due to the sensitivity
and nature of the detection method.
SMLM approaches first emerged as two principally similar but
technically distinct imaging modalities. The first, stochastic optical
reconstruction microscopy (STORM), was published in 2006 [14]
and relies on single molecule detections in fixed samples through
the use of coupled dyes or antibodies. The second, photoactivated
localization microscopy (PALM), generates similar data sets
through the use of unique photoactivatable or photoswitchable
fluorescent proteins [15].
While a number of different STORM approaches have been
developed, direct STORM (dSTORM) [16] is the most widely
used due to its relative simplicity, the ability to utilize the optical
sectioning capability of total internal reflection fluorescence (TIRF)
and the option of performing 3D imaging by incorporating a cylin-
drical or astigmatic lens into the light path [17]. In dSTORM, a
sample labeled with conventional antibodies is imaged at high laser
intensities in the presence of an appropriate redox buffer. Under
these conditions, the fluorophores are driven into a dark state, and
only a small fraction of these become fluorescent in any one time
frame. Therefore, a labeled sample is reduced into what is described
as a “blinking” data set, where a series of frames (in the region of
10,000–30,000) are acquired as the basis for computational recon-
struction with each “blink” representing a single fluorophore.
Over the course of a large number of rapidly acquired frames, fit-
ting algorithms can be applied to compile the probabilistic position
of each emitter/molecule, thus rendering a superresolved image.
While dSTORM can undoubtedly generate stunning images at
tens of nanometers of resolution, there are a number of technical
requirements which must be met in order to achieve high-quality
data sets [11, 18]. Firstly, an imaging system with a high NA
(numerical aperture) objective, high powered lasers, and efficient
camera is required. Modern EMCCD and CMOS cameras have
evolved in leaps and bounds both in terms of speed and sensitivity,
allowing for the acquisition of 100 s of frames per second, thereby
satisfying the need for the rapid acquisition of blinking
fluorophores. In addition, the use of z drift correction systems such

as Nikon’s Perfect Focus System (PFS) helps improve the quality
of the final images displayed.
Beyond the necessary hardware, sample preparation and image
reconstruction are two key components of successful SMLM imag-
ing. With the appropriate camera, illumination, and acquisition
parameters, a sequence of several thousand frames composed of
“blinks” will be generated. This pointillistic data set is ultimately
the source of a SMLM image, and as such appropriate computa-
tional modeling of these blinks is necessary for an accurate final
superresolved image. Early SMLM methods applied a relatively
simple Gaussian fitting approach, whereby each individual point is
fitted to a Gaussian curve, and the resulting compilation of fits
assembled to generate the final image [11]. While effective, this
approach suffered from a number of distinct limitations which has
since led to the development of a host of alternative calculations
for the derivation of SMLM images from blinking data sets. These
are detailed in an excellent comparative review [19], and thus it is
advised that fitting algorithm selection be determined based on the
characteristics of your sample. Regardless of the fitting method
used, the number of photons detected from a single fluorophore is
a major factor in determining the precise location of that molecule
and experimental setups should therefore try to maximize this
parameter to achieve the image with the highest possible resolu-
tion [20].
Finally, sample preparation is another critical component of
quality SMLM imaging. For dSTORM, where antibodies are
applied to a fixed sample, generating a sample which attains a high
labeling density and low background is essential. Firstly, an appro-
priate method of fixation and permeabilization must be selected,
particularly when imaging the cytoskeleton. For example, clearing
with microtubule stabilizing buffer (MTSB) and subsequent meth-
anol fixation can effectively remove nonpolymerized tubulin mol-
ecules, thereby removing a significant amount of background from
tubulin present in the samples. However, such a buffer may not be
appropriate for imaging the fine actin filament network as this can
be destroyed by methanol based fixation methods [21].
Quality antibodies are needed to maintain adequate labeling
while minimizing background. In dSTORM imaging individual
points are assigned to the final reconstructed image, as such extra-
cellular background can quickly clutter a reconstruction. This not
only impacts the qualitative value of the image, but can have a
significant effect on quantification. Labeling density is of equal
importance at nanoscale resolutions, where poor sampling can sig-
nificantly impact the final results [22].
Where the above criteria are satisfied, SMLM approaches, and
in particular dSTORM, are powerful qualitative and quantitative
tools with significant biological applications. Single molecule
resolution allows for a robust interrogation of the spatial arrange-

ment of individual proteins within the cell. Where receptor dynam-
ics are concerned, molecule clustering can provide a unique insight
into how proteins are recruited in response to particular stimuli
[23–25]. Examples of how this can be applied to the study of
platelets are provided below. Similarly this approach can provide an
unprecedented understanding of cytoskeletal structures, all of
which lie significantly below the diffraction limit. For example,
actin exists in filaments 7–9 nm in diameter and microtubules are
in the order of 25 nm, both of which play an integral role in plate-
let morphology and function. Therefore SMLM studies of the
cytoskeleton and its relationship with key cellular components can
reveal novel insights at the single molecule level [26].
1.2 Structured In contrast to SMLM methods which detect single molecules and
Illumination reconstructs the image in a pointillistic manner, structured illumi-
Microscopy (SIM) nation microscopy (SIM) is actually an optical sectioning tech-
nique which can increase axial resolution to around 100–120 nm
[27, 28]. In a fluorescence image, the subdiffraction limit detail is
known as high frequency information and is beyond the range of
frequencies that can be collected by the objective lens. During SIM
image acquisition, a low frequency grid pattern is projected onto
the sample (usually via a diffraction grating in the illumination
pathway) which generates interference patterns (known as Moiré
fringes) due to interactions with the high frequency details of the
sample. The interference patterns are of a lower frequency than the
original detail and thus can be collected by the objective lens and
importantly these patterns contain information regarding the sub-
diffraction limit fine sample detail. During imaging, numerous ori-
entations of the grid pattern are taken at each focal plane (usually
5 phase shifts at 3 rotations) and the resulting interference patterns
are mathematically reconstructed in Fourier space to extract sub-
diffraction limit details of the sample. The technique can be per-
formed in 2D and in 3D and furthermore has the advantage that it
can be performed on live samples [29, 30]. Assuming the appro-
priate lasers, diffraction gratings and filter sets are in place, SIM can
image a range of standard fluorophores, including organic dyes
(FITC, TRITC, Alexa, Atto, etc.) and fluorescent proteins (GFP,
RFP, mCherry, etc.). As for SMLM imaging, high quality optics,
cameras and lasers are required and the use of focal drift correction
mechanisms will improve image quality. However, SIM is
technically less challenging to perform than SMLM and is there-
fore more flexible in its application and can be more easily per-
formed on a range of sample types (e.g., fixed and mounted, live
and unmounted). As for all microscope techniques and discussed
above, good fixation and labeling protocols are required, as is the
need for good quality, validated antibodies against your protein of
interest. The range of reconstruction algorithms for SIM is more
limited than SMLM techniques, but Denmerle et al. [27] describe

important considerations when image processing and spotting
common artifacts.
So while SIM may not provide quantitative information on the
10 nm scale, excellent images can be generated due to the dou-
bling of resolution, optical sectioning, and increased contrast it
provides. This can reveal details about the sample which are not
clear in diffraction limited imaging. Furthermore, because a SIM
image is made up of pixels (and not coordinates as in SMLM),
these can be treated as “standard” images and thus a full suite of
commonly used image analysis techniques are available (e.g., colo-
calization by Pearson’s correlation) for image processing.
1.3 Superresolution Recently a number of platelet studies have employed superresolu-

Imaging of Platelets tion microscopy to investigate the spatial distribution of proteins
within platelets to try to gain insight into their function. SIM is the
technique which has been most widely used, most likely due to the
ease of which it can be applied to samples that have been prepared
for conventional fluorescence microscopy. Several groups have
used SIM to investigate the colocalization of proteins at higher
resolution, for example the coclustering of proteins into alpha-
granules [31] or to see the spatial relationship between actin and
adhesion related proteins such as vinculin [32, 33]. In the case of
Poulter et al. the increased resolution afforded by the use of SIM
allowed the authors to develop a model for the spatial organization
and the role that a specialized actin structure, called the actin nod-
ule, plays in platelet adhesion. This would not have been possible
with conventional, diffraction limited microscopy [33].
SIM has also been proposed as an alternative to EM in the
diagnosis of platelet granule disorders, with Hermansky–Pudlak
syndrome as the proof of principle [34]. In this paper, the authors
present SIM of CD63 in platelets combined with automated analy-
sis of the images as a quantitative, unbiased, high-throughput
method for diagnosis of this syndrome. Recently SIM has been
used to image binding of GPVI-Fc to collagen fibers [35]. This
study also used an alternative superresolution technique known as
stimulated emission depletion (STED) which is a confocal based
technique offering resolutions in the range of 50–75 nm [36].
dSTORM has also been applied to the analysis of platelet r eceptors,
where its quantitative nature has been exploited. For example,
Poulter et al., in addition to using SIM, applied dSTORM to iden-
tify patterns in the distribution of αIIbβ3 integrin at actin nodules
and to quantify the level of phosphoprotein present at these struc-
tures [33]. Pollitt et al. used dSTORM to show that the podo-
planin receptor CLEC-2 clustered above the level expected of a
random distribution when platelets were spread on podoplanin.
This work went on to show that it was this clustering that was
important for CLEC-2 signaling [37]. Other work looking at
GPVI clustering in response to different substrates has shown that

the density of GPVI clusters is determined by the substrate on
which the platelet is spreading and that this clustering of GPVI
dimers represents a second level of control to regulate GPVI-
mediated signaling and platelet activation [37].
While at present PALM has limited applications in the platelet
field due to the need for genetically expressed fluorescent proteins,
current work on CRISPR-Cas9 engineering of mouse lines and iPS
cells, combined with developments in platelet production from
these progenitor cells [38–40] may allow for the generation of
platelets expressing photoswitchable fusion tags which, in turn,
could permit quantitative superresolution imaging of endogenous
proteins, potentially in live platelets. This is an area that will
undoubtedly flourish in the coming years.
2 Materials
2.1 Biological Prepare all solutions using deionized water, unless otherwise
stated.
1. Concentrated citrate solution: 4% (w/v) sodium citrate.
2. Acid Citrate Dextrose (ACD): 120 mM sodium citrate,
110 mM glucose, and 80 mM citric acid
3. Modified Tyrode’s buffer: 134 mM NaCl, 0.34 mM Na2HPO4,
2.9 mM KCl, 12 mM NaHCO3, 20 mM HEPES, 1 mM
MgCl2, 5 mM glucose, pH 7.3.
4. Phosphate Buffered Saline (PBS): 10 mM phosphate buffer,
2.7 mM potassium chloride and 137 mM sodium chloride,
pH 7.4.
5. Prostacyclin: (0.1 μg/mL) in 50 mM Trizma base, pH 9.1.
6. Biological substrate of interest, e.g., Fibrinogen (VWF and
plasminogen depleted) diluted to 100 μg/mL in PBS.
7. 35 mm glass-bottomed imaging dishes with the 10 mm diam-
eter, no. 1.5 coverslip.
8. 13 mm diameter no. 1.5 coverslips.
9. Fatty acid-free BSA—denatured: 5 mg/mL in PBS.
10. Fixative: 10% formalin.
11. Block buffer: 1% BSA, 2% goat serum in PBS.
12. Phalloidin-Alexa Fluor 488: 6.6 μM stock in methanol.
13. Mouse anti-phosphotyrosine antibody (clone 4G10).
14. Goat anti-mouse IgG (H+L) highly cross adsorbed Alexa

Fluor 647 antibody.
15. Nonfluorescing aqueous mounting medium.

16.
STORM blinking buffer: PBS containing 100 mM
2-
mercaptoethylamine (MEA-HCl), 50 μg/mL glucose
oxidase, and 1 μg/mL catalase.
2.2 Hardware Particle count and size analyzer.

2.2.1 General Hardware
2.2.2 SIM SIM imaging is generally performed on commercial SIM micro-

scopes, however, custom built systems do exist. Regardless of the
source of the system, the following components are required for
performing structured illumination microscopy. Critical to the sys-
tem is the ability to generate the illumination patterns onto the
sample and this can be done using a variety of approaches which
are detailed in Denmerle et al. [27]. The system should be based
around a research quality inverted fluorescence microscope stand
with a high numerical aperture (NA), 100× oil immersion objec-
tive lens and excitation, dichroic and emission filter sets suitable for
the fluorophores to be used. For performing 3D-SIM, a motorized
stage is required to allow different focal planes to be captured.
Illumination of samples can be performed using standard lasers of
the required wavelength with commonly used wavelengths includ-
ing 647 nm, 561 nm, and 488 nm and 405 nm. Both Scientific
Complementary Metal-Oxide Semiconductor (sCMOS) and
Electron Multiplying Charged Coupled Device (EMCCD) cam-
eras can be used for SIM and the choice of camera will depend on
the sensitivity or speed required. Finally, the system requires soft-
ware to control both the acquisition of raw data and the processing
of reconstructed data sets. Both freely available and commercial
software packages are available to perform these tasks.
For the SIM experiments described below, we used a Nikon
N-SIM system including an Eclipse Ti-E inverted microscope,
Perfect Focus system 2, Apo TIRF 100x Oil DIC N2 1.49NA
objective lens, and 488 filter cube. Illumination was from a Nikon
LU5 laser bed (including 488 nm laser line) and images were cap-
tured using an Andor DU-897 X-6005 EMCCD camera. The sys-
tem was driven and images reconstructed using NIS-Elements
Advanced Research software v4.5, with SIM module.
2.2.3 STORM dSTORM imaging can be carried out on commercial STORM/

PALM microscope or on custom-built systems. Regardless of the
source of the system, the following components are required for
performing single molecule localization microscopy. The system
should be based around a research quality inverted fluorescence
microscope stand with a high numerical aperture (NA), 100× oil
immersion objective lens and excitation, dichroic and emission
filter sets suitable for the fluorophores to be used. Furthermore,
focus drift mechanisms can improve the quality of the dSTORM

data sets obtained by reducing drift in z. Illumination of samples
should be performed using high power lasers of the required wave-
length to drive fluorophores into the dark state. Commonly used
wavelengths for dSTORM include 647 nm, 561 nm and 488 nm.
A 405 nm laser is also required for reactivation of 647 nm dyes
during imaging. Capture of single molecules requires a high sensi-
tivity EMCCD camera with high readout speeds. However, recent
advances in sCMOS cameras means that they have started to be
used on dSTORM microscopes. dSTORM can be performed
either in TIRF or in 3D. This requires appropriate TIRF illumina-
tor mechanisms or point spread function (PSF) distorting lenses in
the emission pathway to be installed to allow this additional func-
tionality. The ability to tightly control the temperature of the
microscope and thermal drift of the sample, along with the reduc-
tion in air movements across the sample will improve the stability
of the system for the timescale of the image acquisition and there-
fore improve the overall image quality. Finally, the system requires
software to control both the acquisition of raw data and the pro-
cessing of reconstructed data sets. Both freely available and com-
mercial software packages are available to perform these tasks.
For the dSTORM experiments described below, we used a
Nikon N-STORM system including an Eclipse Ti-E inverted
microscope, TIRF module, Perfect Focus system 3, CFl SR
Apochromat TIRF 100× Oil 1.49NA objective lens, and N-STORM
filter cube (Quad or single). Illumination was from an Agilent
Ultra High Power Dual Output Laser bed equipped with four laser
lines (30 mW 405 nm; 120 mW 488 nm; 120 mW 561 nm;
170 mW 647 nm) and images were captured using an Andor
IXON Ultra 897 EMCCD camera. 3D images were captured using
a 3D cylindrical lens and temperature was controlled by an Okolab
S.r.l incubator set at 28 °C. The system was driven and images
reconstructd using NIS-Elements Advanced Research software
v4.5, with STORM module v4.1.
3 Methods
3.1 Platelet The protocol detailed here gives information on how to isolate,
Preparation wash and spread human and mouse platelets onto fibrinogen and
then label and image the actin cytoskeleton (using fluorescent
phalloidin) or phosphorylated proteins (using the 4G10 antibody).
The antibody concentrations and labeling protocol for your pro-
teins of interest will need to be empirically tested but this gives a
good starting point for your experiments.
3.1.1 Preparation All steps are performed at room temperature and in a swing bucket
of Washed Human Platelets rotor centrifuge, unless otherwise stated.
1. Take human blood by venepuncture into sodium citrate (10%

v:v). Further anticoagulate blood by adding ACD to a final
concentration of 10% v:v.
2. Spin anticoagulated blood at 200 × g for 20 min. Carefully col-
lect the platelet rich plasma (PRP) and centrifuge at 1000 × g
for 10 min in the presence of 0.1 μg/mL prostacyclin.
3. Resuspend the platelet pellet in 28 mL of wash buffer (25 mL
of modified Tyrode’s buffer, 3 mL of ACD (both warmed to
37 °C) and 0.1 μg/mL prostacyclin) and centrifuge the cells at
1000 × g for 10 min. Resuspend the pellet in 1 mL of modified
Tyrode’s buffer.
4. Measure the concentration of platelets using a particle count
and size analyzer and dilute to 2 × 108 platelets/mL with mod-
ified Tyrode’s buffer. Rest platelets for 30 min before starting
the spreading experiments to allow excess prostacyclin to lose
activity.
3.1.2 Preparation All steps are performed at room temperature and in a swing bucket
of Washed Mouse Platelets rotor centrifuge, unless otherwise stated.
1. Draw mouse blood into 10% v:v ACD by either direct cardiac
puncture or from the vena cava following laparotomy. Add
anticoagulated blood to 200 μL warmed modified Tyrode’s
buffer.
2. Centrifuge anticoagulated blood for 5 min at 2000 rpm in a
benchtop microfuge. Decant the PRP and a small amount of
red blood cells (to ensure maximum recovery of platelets) into
a clean 1.5 mL Eppendorf tube and centrifuge at 200 × g for
6 min. Carefully collect the plasma and the layer of platelets
sitting on top of the packed red blood cells and place into a
clean 1.5 mL Eppendorf tube.
3. Centrifuge the PRP (the plasma and platelet layers from step
2) at 1000 × g for 6 min in the presence of 0.1 μg/mL prosta-
cyclin in order to isolate the platelets from the PRP.
4. Resuspend the platelet pellet in approximately 200 μL modi-
fied Tyrode’s buffer and measure the platelet concentration
using a cell counter. Dilute the platelet suspension to 2 × 108
platelets/mL with modified Tyrode’s buffer. Rest the platelets
for 30 min before starting the spreading experiments to
allow excess prostacyclin to decay.
3.2 Sample 1.
For STORM/PALM imaging 35 mm diameter, #1.5
Preparation (0.17 mm) glass bottomed dishes are used. For SIM imaging
either 13 mm diameter, #1.5 (0.17 mm) glass coverslips or
3.2.1 Preparation
35 mm #1.5 (0.17 mm) glass bottomed dishes are used (see
of Coverslips and Dishes
Note 1).
for Platelet Spreading
2. Coat the imaging surface by incubating with 100 μg/mL of

fibrinogen (in PBS) overnight at 4 °C.
3. Block any remaining uncoated glass by incubation with 5 mg/
mL BSA for 1 h at room temperature followed by three washes
with PBS. Store coated surface in PBS until ready for platelet
spreading.
3.2.2 Spreading 1. Dilute washed and rested platelets to 2 × 107 platelets/mL in

and Labeling of Platelet Tyrode’s buffer and add to the glass bottomed dish or cover-
F-Actin and slip (held in a 12-well plate) and allow to spread on the coated
Phosphorylated Proteins surface for 45 min at 37 °C (see Note 2).
2. After spreading, wash nonadhered cells away with PBS pre-
warmed to 37 °C and fix the spread platelets with 10% forma-
lin solution for 10 min (see Note 3).
3. Following fixation, wash coverslips three times in PBS. If
aldehyde-based fixation methods are used then residual fixa-
tive can be quenched by incubating the fixed platelets in
50 mM ammonium chloride for 10 min, followed by three
PBS washes.
4. Permeabilize cells with 0.1% Triton X-100 (v:v) in PBS for
5 min, followed by three PBS washes (see Note 4).
5. If the platelets are going to be labeled with an antibody a
blocking step needs to be performed to reduce nonspecific
antibody binding. Incubate the platelets in block buffer for
1 h at room temperature.
6. For SIM imaging of F-actin, incubate fixed and permeabilized
platelets in a 1:500 dilution of Alexa 488-phalloidin in PBS for
1 h at room temperature (see Note 5). Wash samples three
times in PBS.
7. For dSTORM imaging of tyrosine phosphorylated proteins,
dilute the 4G10 antibody 1:500 into block buffer and incubate
with platelets for 1 h at room temperature (see Note 6).
Following labeling, wash three times with PBS.
8. Dilute the anti-mouse Alexa Fluor 647 secondary antibody
1:300 into block buffer and incubate with platelets for 1 h at
room temperature (see Note 7). Wash samples three times in
PBS.
9. For SIM imaging, mount coverslips onto glass slides using
non-fluorescing aqueous mounting medium and store at
4 °C. Samples for SIM imaging in dishes can be stored in PBS
at 4 °C.
10. For dSTORM/PALM imaging, store dishes in PBS at 4 °C.
3.3 Super-Resolution For optimal imaging results using SIM, the microscope needs to
Imaging of Platelet be set up, aligned, and calibrated correctly. For multiuser/core
Samples facilities, this should be done routinely by the facility, but for single
user or non-facility based systems, this will need to be carried out
3.3.1 SIM Imaging
by the user. The details of this calibration are beyond the scope of
of Platelet Samples
this chapter, but extensive details on the theory of SIM, how to
perform calibrations and how to troubleshoot imaging problems
have been provided by Demmerle et al. [27].
1. Place the dish, or slide, to be imaged on the microscope stage
within the microscope incubator. Using high NA immersion
oil applied to the 100× lens, bring the objective lens up to
engage with the coverslip. Bring the cells into focus on the
camera by using the fine focus and engage focus drift correc-
tion system.
2. In the software, select the required optical configuration and
correct diffraction grating for your fluorophore(s).
3. Adjust the laser power and camera exposure time to achieve an
image intensity signal level (grey levels) of approximately 4000
(for the 14 bit camera setting used) (see Note 8).
4. Select 2D-SIM options in your software ensure that the grat-
ing pattern can be clearly seen in focus in the image (Fig. 1a)
and acquire a single z plane. This will collect a raw image which
contains the 15 frames needed for SIM reconstruction (3 rota-
tions and 5 phases—see Fig. 1a). For 3D-SIM, Z-stacks through
the cell can also be acquired by using the motorized stage of
the microscope and capturing the multirotation, multiphase
images at each focal plane (see Note 9). A diffraction limited
image can be collected for reference (Fig. 1b, c) (see Note 10).
3.3.2 Image Processing It is critical for optimal image reconstruction that the correct num-
of Raw SIM Data ber of grey levels are collected and the grating pattern is seen with
good contrast in the raw images. Further adjustments to the final
reconstructed image quality can be made by adjusting r econstruction
parameters in the software.
1. Open the raw SIM image data file within the respective micro-
scope analysis package for your microscope system.
2. In a z-stack, remove the slices that are well above and below
the area of interest to improve the quality of the
reconstruction.
3. Most commonly used SIM reconstruction software packages
allow for adjustment of certain parameters to improve the final
reconstructed SIM image. These may include noise filtering,
out-of-focus blur suppression, and illumination modulation
contrast ratio. Details of how these parameters will affect image
reconstruction will be found in your software’s help files.
Fig. 1 Structured Illumination Microscopy (SIM) of platelet actin cytoskeleton. (a) Acquisition of 3D-SIM images
on the N-SIM system generates an .nd2 file containing 15 images for each z position. These 15 images differ
in the rotation and phase of the diffraction grating. The 3 images show examples of the 3 rotations (indicated
by lines in bottom right corner). The diffraction grating pattern can be seen as striping within the image. (b)
Example diffraction limited epifluorescence image of a single z slice. (c) Intensity profile from the line shown
in panel B. Note that overall shape of cell is seen, but individual structural detail of the cytoskeleton is not. (d)
Fourier transform of the image shown in panel B showing the frequency space representation. Low frequency
information is located in the center. The size of the central observable region is defined by the maximum fre-
quency the objective can transmit. (e) Reconstructed SIM image of the same z slice as panel B. The increased
contrast and resolution of the image is clear. (f) Intensity line profile from the line in panel E. Note the increased
resolution and contrast of the image results in individual actin bundles being resolved. (g) Fourier transform of
the image shown in panel E showing the characteristic petal shape indicating that the high frequency informa-
tion from outside the diffraction limited observable region was successfully captured. Scale bar in all
images = 5 μm
Additionally, Demmerle et al. [27] describe the theoretical

principles behind these parameters in greater detail to enable
the user to perform optimum, artifact-free reconstructions.
4. Once satisfied with the parameters click reconstruct to apply to
the image and generate a SIM superresolution image (Fig. 1e, f)
5. A quick readout of SIM image quality can be performed using
a Fourier transform of the reconstructed image in Fiji [41, 42].
The characteristic petal shaped pattern of a good SIM image
(Fig. 1d, g) indicates that the reconstructed image contains
effective information from each component in the raw data set.
Deviations from this petal shaped pattern may indicate prob-
lems with one or more components of the reconstruction (see
Note 11).
3.3.3 dSTORM Imaging For optimal imaging results using dSTORM, the microscope needs
of Platelet Samples to be set up, aligned, and calibrated correctly (this includes TIRF
alignment, 3D calibration files and chromatic aberration warp files
for 3D and multicolor STORM respectively). For multiuser/core
facilities, this should be done routinely by the facility, but for single
user or non-facility based systems, this will need to be carried out
by the user. The details of this calibration are beyond the scope of
this chapter, but extensive details on how to perform calibrations
should be provide by the microscope provider and troubleshooting
of STORM imaging problems have been provided by several
groups [18, 43–45].
1. Before imaging, prewarm the microscope incubator to 28 °C
for a number of hours (usually overnight). Place both the
glass-bottomed dish with the sample and the STORM blinking
buffer into the microscope incubator to warm for at least
15 min prior to imaging (see Note 12).
2. Put the blinking buffer into the glass-bottomed dish (see Note
13) and place the dish onto the microscope stage within the
microscope incubator, apply high NA immersion oil to the lens
and bring up the lens to engage with the coverslip. Use fine
focus to bring the sample into focus on the camera and engage
focus drift correction system (see Note 14).
3. Using low laser power (usually less than <0.1% to avoid bleaching
the sample) identify an appropriate field of view (FOV), focus the
sample and take a reference snapshot (Fig. 2a). If a 3D image is
required ensure that the cylindrical lens is in position and
3D-STORM mode is selected before starting acquisition.
4. Start a live image and then increase the power of the 647 nm
laser to 100%. Follow the live image until the sample has
bleached and individual fluorescent blinks are observed (usu-
ally between 1 and 30 s depending on the label used). At this
point start image acquistion and capture 20,000 frames using
an appropriate exposure time and camera setting. (For the
Fig. 2 Direct Stochastic Optical Reconstruction Microscopy (dSTORM) of platelets (a) Diffraction limited
epifluorescence image of a human platelet spread on fibrinogen and immunostained for phosphotyrosine. A
secondary antibody conjugated with Alexa 647 was used. (b) Reconstructed 2D-dSTORM image of cell in panel
A. The increased resolution allows the resolution of diffraction limited structures as indicated by the arrow. (c)
Metrics provided by the reconstruction algorithm include total number of molecules detected, localization
precision (in xy for each molecule detected) and the number of photons counted per molecule. The lower graph
shows the important relationship between photon count and localization precision, with the most highly local-
ized molecules being the ones with the highest photon count. (d) Example of a 3D-dSTORM reconstructed
image showing z position indicated by a color scale. (e) Zoomed in 3D representation of the box in panel D
showing 3D organization of phosphotyrosine within the platelet. Scale bar in all images = 2.5 μm
examples shown here we used a camera exposure time; 1 frame

(~10 ms), Conversion gain: 3, EM gain: 300.) Increase 405 nm
laser power in ~2.5–5% steps every 30 s throughout the acqui-
sition to assist in reactivation of fluorophore from the dark
state to maintain the number of detected blinks per frame.
Numbers of molecules detected and the preview dSTORM
image can be followed by selecting the appropriate software
functions. At the end of acquisition save files into the appropri-
ate file format for your software.
5. Return laser settings to starting values. Select a new FOV for
imaging and repeat the above process.
3.3.4 Image Processing 1. Open the raw STORM image data file within the respective
of Raw dSTORM Data microscope analysis package for your microscope system (see
Note 15).
2. Select a middle frame from the image sequence and use this to
set the threshold for point identification. Different threshold
values should be tested to ensure that all “real” blinks are
detected and that background is not (see Note 16).
3. If the image is 3D (i.e., if it was collected using the cylindrical
lens), then select the appropriate options in your software
package which will maximize the identification of distorted
point spread functions. A valid 3D calibration file is needed for
this processing to assign correct z positions (see Note 17).
4. Once settings have been confirmed, start reconstruction of the
STORM image by clicking the process start button. Once fin-
ished, the software will display a high-resolution image of the
data (Fig. 2b). The display options and any subsequent pro-
cessing (density filtering, rendering, 3D visualization, etc.) can
be performed (Fig. 2c–e).
5. STORM analysis software contains an autocorrelation drift
correction algorithm to recognize and correct drift in the
image due to stage movement. An example of this can be seen
in Fig. 3. Furthermore, if multicolor imaging is used, the soft-
ware can correct for chromatic aberrations assuming that a
valid warp calibration has been performed.
6. A molecule list which contains all the information for each
detected molecule in the image can be exported for further
analysis (see Note 18).
4 Notes
1. Glass-bottomed dishes or multiwell slides can be used for both

SIM and dSTORM, e.g., Labtek multiwell slide and MatTek
glass-bottomed dishes. In addition, alternative materials are
available (e.g., Ibidi slides) which have a similar refractive
index to glass and thus may also be considered for your appli-
cation. If you use a different thickness coverslip (e.g., #
1–0.15 mm) you must remember to adjust the correction
collar of the microscope to ensure that the point spread func-
tion (PSF) is even. This can be tested by imaging a z-stack of
fluorescent beads in the type of imaging dish you will be
using. A good PSF has a symmetrical spread of the fluores-
cent signal as the z-series goes above and below the focal plane
of the bead, when viewed in xz and yz orientation.
2. Platelets can be spread for different time periods depending
on the needs of your experiment. They can also be preincubated
with inhibitors or antibodies to surface proteins of interest
(which do not affect adhesion or spreading) as required.
Ensure that appropriate controls for inhibitors are included in
the experiments.
3. The fixation method used will depend on the proteins/struc-
tures that you are interested in examining and will need to
be determined empirically for your application.
4. If you are using an antibody to an extracellular epitope
permeabilization of the cells may not be necessary. This will
need to be determined for each antibody. One thing to be
aware of if labeling after spreading is the potential for restricted
antibody access to the epitope. Increased labeling at the cell
edges may not be the real protein distribution but may repre-
sent inaccessibility of the antibody to areas that are tightly
Fig. 3 Example of software based drift correction in final dSTORM images. (a) Example of a reconstructed
dSTORM image which exhibited some stage drift during acquisition prior to applying drift correction. (b) The
same image following application of the autocorrelation drift correction algorithm within NIS-Elements. Scale
bar = 2.5 μm
adhered. It is important therefore that antibodies are well

characterized, for example in confocal microscopy first.
5. Mice expressing Lifeact-GFP which labels F-actin [46] are
available and these can be used to image F-actin dynamics in
SIM in live platelets.
6. Dilutions and incubation times will depend upon the antibody
used and should be determined empirically for each antibody.
It is possible to do dual labeling and it is good practice to opti-
mize antibody labeling protocols for each antibody indepen-
dently. Furthermore, samples which will be stored before
imaging may benefit from postlabeling fixation [18].
7. To increase spatial resolution, labeling cells with fluorescently
conjugated antibody Fab fragments (~9 nm), or a camelid
nanobody which is even smaller (~3 nm), is preferable to label-
ing with a primary antibody (~15 nm) followed by a fluores-
cently tagged secondary antibody. This ensures that the
fluorescent dye (which is detected) is as close as possible to the
epitope of interest. Using a primary plus secondary antibody
approach can add up to 30 nm between the epitope and the
detected “blink” which, in turn, can artificially enlarge struc-
tures when viewed on the nanometre scale of
dSTORM. Increasingly, samples with HALO and SNAP tags
are being used for superresolution imaging [47], but this
application has not been extended into megakaryocytes or
platelets to our knowledge
8. It is worth spending time setting the appropriate exposure
time and laser intensities at the beginning as this will affect
subsequent image quality if it is not set correctly. A balance
between laser power and exposure time needs to be obtained
which gives a good dynamic range in the captured image while
minimizing photobleaching in the sample.
9. In 3D-SIM, each z-plane consists of 15 individual images, and
therefore care must be taken to ensure that the cell does not
bleach within the acquisition by limiting the laser power and
the number of z-positions, i.e., do not capture excessive images
above or below the focal plane. Bleaching will negatively affect
image quality.
10. Multicolor and live cell imaging is possible with SIM. Provided
the microscope you use is equipped with the appropriate laser
lines, several different fluorophores can be used to label mul-
tiple proteins of interest in both 2D and 3D. The use of single
color controls to ensure no bleed-through or cross talk is
occurring is recommended. Fluorescent labeling of platelets
for live cell imaging could be in the form of a transgenic mouse
model (protein of interest tagged with a fluorescent protein),
a non-function blocking antibody to a surface receptor, or
other fluorescent cell permeable probe. For these, labeling

protocols will need to be determined by the user.
11. If the characteristic petal shape is not seen when a Fourier
transform is performed on your SIM images, acquisition set-
tings may need to be adjusted to obtain a good raw image data
set in which the grid pattern is clear (e.g., adjusting focal
plane, or ensuring that the exposure time and laser power are
optimized to achieve the correct level of grey values). If the
petal shape is missing one or more “lobes,” this may indicate
that there is a problem with one or more of the diffraction grid
pattern rotations or changes may need to be made to the
reconstruction settings [27]. In addition, a suite of ImageJ
plugins called SIMcheck [48] is available for checking the
quality of SIM data and determining the resolution of SIM
images.
12. Having a stable temperature to image in is very important to
minimize thermal drift which can have a big impact on
STORM image quality. Heating the incubator for at least 12 h
prior to imaging will ensure that drift is minimized. Setting
the temperature to slightly higher than room temperature
(e.g., 28 °C) will allow the incubator to maintain a steady tem-
perature more easily.
13. Depending on the fluorophore being imaged, the buffer con-
ditions may need to be varied. Much work has been done on
identifying the optimal buffers for different fluorophores [18,
43], thus the appropriate combination of buffers for the
experiment should be determined by the user.
14. If your protein of interest is membrane-bound, for example a
cell surface receptor, it is recommended that you image your
spread platelet sample in TIRF mode. This will ensure that
you are only illuminating, and therefore imaging, the pro-
teins that are at the adherent platelet surface. This will also
increase the signal-to-noise ratio which will result in cleaner
images. For a detailed method on performing TIRF see
Poulter et al. [49].
15. Raw data can alternatively be reconstructed using the

Thunderstorm plugin [50] for Fiji [41].
16. Individual software programmes have different tools for per-
forming this task. In the examples given in Fig. 2, the Nikon
“peak intensity” tool can help the user to identify the thresh-
old values for a particular image. When this tool is placed over
an area which you consider to be a blink, making sure that the
center is over the brightest pixel, the intensity over back-
ground is displayed. This value can then be used to select the
threshold level.
17. The DAOSTORM algorithm is commonly used to fit overlap-

ping blinks [51]. This is a useful feature especially for 3D data
but it must be noted that this function will increase recon-
struction processing time considerably.
18. The resolution of the final superresolved image can be deter-
mined using a number of methods. For SMLM data, the user
can compare localization precision (also known as uncertainty)
for each identified molecule in the sample or measure full width
half maximal (FWHM) distances of identified objects. While
these provide a good measure of the quality of the data gener-
ated, the effective resolution of the final image can be calculated
using the Fourier ring correlation (FRC) method [52] which
takes into account uncertainty and labeling density.
References
1. Hartwig JH (1992) Mechanisms of actin rear- pp 1–14. https://doi.
rangements mediating platelet activation. org/10.1007/978-1-4615-7133-9_1
J Cell Biol 118(6):1421–1442 9. Huang B, Bates M, Zhuang X (2009) Super
2. Hartwig JH, Barkalow K, Azim A, Italiano resolution fluorescence microscopy. Annu Rev
J (1999) The elegant platelet: signals control- Biochem 78:993–1016. https://doi.
ling actin assembly. Thromb Haemost org/10.1146/annurev.biochem.77.061906.
82(2):392–398 092014
3. van Nispen tot Pannerden H, de Haas F, 10. Sydor AM, Czymmek KJ, Puchner EM,
Geerts W, Posthuma G, van Dijk S, Heijnen Mennella V (2015) Super-resolution micros-
HF (2010) The platelet interior revisited: elec- copy: from single molecules to Supramolecular
tron tomography reveals tubular alpha-granule assemblies. Trends Cell Biol 25(12):730–748.
subtypes. Blood 116(7):1147–1156. https:// https://doi.org/10.1016/j.tcb.2015.10.004
doi.org/10.1182/blood-2010-02-268680 11. John RA, Stephen TR, Michael WD (2013)
4. van Nispen Tot Pannerden HE, van Dijk SM, Single molecule localization microscopy for
Du V, Heijnen HF (2009) Platelet protein superresolution. J Opt 15(9):094001
disulfide isomerase is localized in the dense 12. Deschout H, Shivanandan A, Annibale P,
tubular system and does not become surface Scarselli M, Radenovic A (2014) Progress in
expressed after activation. Blood quantitative single-molecule localization
114(21):4738–4740. https://doi. microscopy. Histochem Cell Biol 142(1):5–17.
org/10.1182/blood-2009-03-210450 https://doi.org/10.1007/
5. White JG (1998) Use of the electron micro- s00418-014-1217-y
scope for diagnosis of platelet disorders. Semin 13. Nicovich PR, Owen DM, Gaus K (2017)
Thromb Hemost 24(2):163–168. https:// Turning single-molecule localization micros-
doi.org/10.1055/s-2007-995836 copy into a quantitative bioanalytical tool. Nat
6. Combs CA (2010) Fluorescence microscopy: a Protoc 12(3):453–460. https://doi.
concise guide to current imaging methods. org/10.1038/nprot.2016.166
Curr Protoc Neurosci Chapter 2:Unit2.1. 14. Rust MJ, Bates M, Zhuang X (2006) Sub-
https://doi.org/10.1002/0471142301. diffraction-limit imaging by stochastic optical
ns0201s50 reconstruction microscopy (STORM). Nat
7. Abbe E (1873) Beiträge zur Theorie des Methods 3(10):793–796
Mikroskops und der mikroskopischen 15. Betzig E, Patterson GH, Sougrat R, Lindwasser
Wahrnehmung. Arch Mikrosk Anat OW, Olenych S, Bonifacino JS, Davidson MW,
9:413–418 Lippincott-Schwartz J, Hess HF (2006)
8. Inoué S (1990) Foundations of confocal Imaging intracellular fluorescent proteins at
scanned imaging in light microscopy. In: nanometer resolution. Science
Pawley JB (ed) Handbook of biological confo- 313(5793):1642–1645. https://doi.
cal microscopy. Springer US, Boston, MA, org/10.1126/science.1127344
16. Heilemann M, van de Linde S, Schuttpelz M, 26. Ashdown GW, Burn GL, Williamson DJ,
Kasper R, Seefeldt B, Mukherjee A, Tinnefeld Pandžić E, Peters R, Holden M, Ewers H,
P, Sauer M (2008) Subdiffraction- resolution Shao L, Wiseman PW, Owen DM (2017) Live-
fluorescence imaging with conventional fluo- cell super-resolution reveals F-actin and plasma
rescent probes. Angew Chem Int Ed Engl membrane dynamics at the T cell synapse.
47(33):6172–6176. https://doi. Biophys J 112(8):1703–1713. https://doi.
org/10.1002/anie.200802376 org/10.1016/j.bpj.2017.01.038
17. Huang B, Wang W, Bates M, Zhuang X
27. Demmerle J, Innocent C, North AJ, Ball G,
(2008) Three-dimensional super-resolution Muller M, Miron E, Matsuda A, Dobbie IM,
imaging by stochastic optical reconstruction Markaki Y, Schermelleh L (2017) Strategic and
microscopy. Science 319(5864):810–813. practical guidelines for successful structured
https://doi.org/10.1126/ illumination microscopy. Nat Protoc
science.1153529 12(5):988–1010. https://doi.org/10.1038/
18. van de Linde S, Loschberger A, Klein T, nprot.2017.019
Heidbreder M, Wolter S, Heilemann M, Sauer 28. Gustafsson MGL (2000) Surpassing the lateral
M (2011) Direct stochastic optical reconstruc- resolution limit by a factor of two using struc-
tion microscopy with standard fluorescent tured illumination microscopy. J Microsc
probes. Nat Protoc 6(7):991–1009 198(2):82–87. https://doi.
19. Sage D, Kirshner H, Pengo T, Stuurman N, org/10.1046/j.1365-2818.2000.00710.x
Min J, Manley S, Unser M (2015) Quantitative 29. Schermelleh L, Carlton PM, Haase S, Shao L,
evaluation of software packages for single- Winoto L, Kner P, Burke B, Cardoso MC,
molecule localization microscopy. Nat Methods Agard DA, Gustafsson MG, Leonhardt H,
12(8):717–724. https://doi.org/10.1038/ Sedat JW (2008) Subdiffraction multicolor
nmeth.3442 imaging of the nuclear periphery with 3D
20. Thompson RE, Larson DR, Webb WW (2002) structured illumination microscopy. Science
Precise nanometer localization analysis for 320(5881):1332–1336. https://doi.
individual fluorescent probes. Biophys org/10.1126/science.1156947
J 82(5):2775–2783. https://doi. 30. Shao L, Kner P, Rego EH, Gustafsson MGL
org/10.1016/s0006-3495(02)75618-x (2011) Super-resolution 3D microscopy of live
21. Whelan DR, Bell TDM (2015) Image artifacts whole cells using structured illumination. Nat
in single molecule localization microscopy: Methods 8(12):1044–1046
why optimization of sample preparation proto- 31. Kamykowski J, Carlton P, Sehgal S, Storrie B
cols matters. Sci Rep 5:7924. https://doi. (2011) Quantitative immunofluorescence
org/10.1038/srep07924 mapping reveals little functional coclustering of
22. Requejo-Isidro J (2013) Fluorescence nanos- proteins within platelet α-granules. Blood
copy. Methods and applications. J Chem Biol 118(5):1370–1373. https://doi.
6(3):97–120. https://doi.org/10.1007/ org/10.1182/blood-2011-01-330910
s12154-013-0096-3 32. Aslan JE, Baker SM, Loren CP, Haley KM,
23. Owen DM, Rentero C, Rossy J, Magenau A, Itakura A, Pang J, Greenberg DL, David LL,
Williamson D, Rodriguez M, Gaus K (2010) Manser E, Chernoff J, McCarty OJT (2013)
PALM imaging and cluster analysis of protein The PAK system links rho GTPase signaling to
heterogeneity at the cell surface. J Biophotonics thrombin-mediated platelet activation. Am
3(7):446–454. https://doi.org/10.1002/ J Physiol Cell Physiol 305(5):C519–C528.
jbio.200900089 https://doi.org/10.1152/ajpcell.00418.2012
24. Pageon SV, Cordoba SP, Owen DM, Rothery 33. Poulter NS, Pollitt AY, Davies A, Malinova D,
SM, Oszmiana A, Davis DM (2013) Nash GB, Hannon MJ, Pikramenou Z,
Superresolution microscopy reveals nanometer- Rappoport JZ, Hartwig JH, Owen DM,
scale reorganization of inhibitory natural killer Thrasher AJ, Watson SP, Thomas SG (2015)
cell receptors upon activation of NKG2D. Sci Platelet actin nodules are podosome-like struc-
Signal 6(285):ra62. https://doi. tures dependent on Wiskott-Aldrich syndrome
org/10.1126/scisignal.2003947 protein and ARP2/3 complex. Nat Commun
25. Williamson DJ, Owen DM, Rossy J, Magenau 6:7254. https://doi.org/10.1038/
A, Wehrmann M, Gooding JJ, Gaus K (2011) ncomms8254
Pre-existing clusters of the adaptor Lat do not 34. Westmoreland D, Shaw M, Grimes W, Metcalf
participate in early T cell signaling events. Nat DJ, Burden JJ, Gomez K, Knight AE, Cutler
Immunol 12(7):655–662. https://doi. DF (2016) Super-resolution microscopy as a
org/10.1038/ni.2049 potential approach to diagnosis of platelet
granule disorders. J Thromb Haemost 43. Dempsey GT, Vaughan JC, Chen KH, Bates
14(4):839–849. https://doi.org/10.1111/ M, Zhuang X (2011) Evaluation of fluoro-
jth.13269 phores for optimal performance in localization-
35. Jamasbi J, Megens RTA, Bianchini M, Uhland based super-resolution imaging. Nat Methods
K, Münch G, Ungerer M, Sherman S, Faussner 8(12):1027–1036. https://doi.org/10.1038/
A, Brandl R, John C, Buchner J, Weber C, nmeth.1768
Lorenz R, Elia N, Siess W (2016) Cross-linking 44. Metcalf DJ. Localisation microscopy immuno-
GPVI-fc by anti-fc antibodies potentiates its labelling guide. National Physics Laboratory
inhibition of atherosclerotic plaque- and colla- Avalable via http://www.npl.co.uk/upload/
gen-induced platelet activation. JACC Basic pdf/localisation-microscopy-immunolabel-
Transl Sci 1(3):131–142. https://doi. ling-guide.pdf. Accessed 29th May 2018
org/10.1016/j.jacbts.2016.03.008 45. Metcalf DJ, Edwards R, Kumarswami N,
36. Hell SW, Wichmann J (1994) Breaking the dif- Knight AE (2013) Test samples for optimizing
fraction resolution limit by stimulated emission: STORM super-resolution microscopy. J Vis
stimulated-emission-depletion fluorescence Exp 79:50579. https://doi.org/10.
microscopy. Opt Lett 19(11):780–782 3791/50579
37. Pollitt AY, Poulter NS, Gitz E, Navarro-Nunez 46. Riedl J, Flynn KC, Raducanu A, Gartner F,
L, Wang YJ, Hughes CE, Thomas SG, Beck G, Bosl M, Bradke F, Massberg S, Aszodi
Nieswandt B, Douglas MR, Owen DM, A, Sixt M, Wedlich-Soldner R (2010) Lifeact
Jackson DG, Dustin ML, Watson SP (2014) mice for studying F-actin dynamics. Nat
Syk and Src family kinases regulate C-type lec- Methods 7(3):168–169. https://doi.
tin receptor 2 (CLEC-2)-mediated clustering org/10.1038/nmeth0310-168
of podoplanin and platelet adhesion to lym- 47. Stagge F, Mitronova GY, Belov VN, Wurm
phatic endothelial cells. J Biol Chem CA, Jakobs S (2013) Snap-, CLIP- and
289(52):35695–35710. https://doi. Halo-
tag Labelling of budding yeast cells.
org/10.1074/jbc.M114.584284 PLoS One 8(10):e78745. https://doi.
38. Moreau T, Evans AL, Vasquez L, Tijssen MR, org/10.1371/journal.pone.0078745
Yan Y, Trotter MW, Howard D, Colzani M, 48. Ball G, Demmerle J, Kaufmann R, Davis I,
Arumugam M, Wu WH, Dalby A, Lampela R, Dobbie IM, Schermelleh L (2015) SIMcheck:
Bouet G, Hobbs CM, Pask DC, Payne H, a toolbox for successful super-resolution struc-
Ponomaryov T, Brill A, Soranzo N, Ouwehand tured illumination microscopy. Sci Rep
WH, Pedersen RA, Ghevaert C (2016) Large- 5:15915. https://doi.org/10.1038/
scale production of megakaryocytes from human srep15915
pluripotent stem cells by chemically defined for- 49. Poulter NS, Pitkeathly WTE, Smith PJ,
ward programming. Nat Commun 7:11208. Rappoport JZ (2015) The physical basis of
https://doi.org/10.1038/ncomms11208 Total internal reflection fluorescence (TIRF)
39. Thon JN, Dykstra BJ, Beaulieu LM (2017) microscopy and its cellular applications. In:
Platelet bioreactor: accelerated evolution of Verveer PJ (ed) Advanced fluorescence micros-
design and manufacture. Platelets 28(5):472– copy: methods and protocols. Springer
477. https://doi.org/10.1080/09537104.20 New York, New York, NY, pp 1–23. https://
16.1265922 doi.org/10.1007/978-1-4939-2080-8_1
40. Thon JN, Mazutis L, Wu S, Sylman JL, 50. Ovesny M, Krizek P, Borkovec J, Svindrych Z,
Ehrlicher A, Machlus KR, Feng Q, Lu S, Lanza Hagen GM (2014) ThunderSTORM: a com-
R, Neeves KB, Weitz DA, Italiano JE Jr (2014) prehensive ImageJ plug-in for PALM and
Platelet bioreactor-on-a-chip. Blood STORM data analysis and super-resolution
124(12):1857–1867 imaging. Bioinformatics 30(16):2389–2390.
41. Schindelin J, Arganda-Carreras I, Frise E, https://doi.org/10.1093/bioinformatics/
Kaynig V, Longair M, Pietzsch T, Preibisch S, btu202
Rueden C, Saalfeld S, Schmid B, Tinevez JY, 51. Holden SJ, Uphoff S, Kapanidis AN (2011)
White DJ, Hartenstein V, Eliceiri K, Tomancak DAOSTORM: an algorithm for high-density
P, Cardona A (2012) Fiji: an open-source plat- super-resolution microscopy. Nat Methods
form for biological-image analysis. Nat 8(4):279–280
Methods 9(7):676–682. https://doi. 52. Nieuwenhuizen RPJ, Lidke KA, Bates M, Puig
org/10.1038/nmeth.2019 DL, Grunwald D, Stallinga S, Rieger B (2013)
42. Schneider CA, Rasband WS, Eliceiri KW Measuring image resolution in optical nanos-
(2012) NIH image to ImageJ: 25 years of copy. Nat Methods 10(6):557–562. https://
image analysis. Nat Methods 9(7):671–675 doi.org/10.1038/nmeth.2448
Chapter 4
Electron Tomography and Correlative Approaches

in Platelet Studies
Kasia B. Engberts, Cor Seinen, Willie J. C. Geerts, and Harry F. G. Heijnen
Abstract
Blood platelets play a central role in the arrest of bleeding and the development of thrombosis. Unraveling
the complex processes of platelet biogenesis from megakaryocytes, platelet adhesion, aggregation, and
secretory responses are important topics in the field of hemostasis and thrombosis. Analysis of the ultra-
structural changes that occur during these processes is essential for understanding the rapid membrane
dynamics and has contributed substantially to our present knowledge of platelet formation and function-
ing. Recent developments in real-time imaging, correlative light and electron microscopy imaging
(CLEM), and 3D (cryo) electron microscopy and tomography offer exciting opportunities to improve
studies of the platelet adhesive responses and secretion at the ultrastructural level in a close to native envi-
ronment. In this chapter we discuss and illustrate cryo preparation techniques (high-pressure freezing,
vitrification), correlative LM and EM workflows, and 3D cryo-electron tomography that we apply in our
current research projects.
Key words TEM, Immuno-EM, Fixation and cryo-immobilization, High pressure freezing EM
tomography, Cryo-EM, Correlative light and electron microscopy (CLEM)
Abbreviations
EM Electron microscopy
ET Electron tomography
CLEM Correlative light and electron microscopy
FS Freeze substitution
HPF High-pressure freezing
LN2 Liquid nitrogen
CCD Charge coupled device
3D Three-dimensional
vWF von Willebrand factor
SIRT Simultaneous iterative reconstruction technique
Electronic supplementary material: The online version of this chapter (doi:10.1007/978-1-4939-8585-2_4) con-
tains supplementary material, which is available to authorized users.
55
56 Kasia B. Engberts et al.
Terminology
Missing Wedge Missing information due to limited tilt angles
Contouring Manual drawing of contour lines in slices of a tomogram
Tomogram Computed 3D volume reconstruction of a specimen by using multiple
projection images
1 Introduction
Platelet biogenesis, platelet adhesion, aggregation, and the secre-

tory responses are complex processes. The unraveling of these pro-
cesses is an important goal for researchers in the field of hemostasis
and thrombosis. Structural and ultrastructural information of
platelets is essential for the understanding of the rapid membrane
dynamics that occur during these events (see ref. 1 and citations
therein for a recent overview). Transmission electron microscopy
(TEM) methods have contributed enormously to our present
knowledge of the platelet ultrastructure and remain an essential
tool to provide insight into the mechanisms of these membrane
dynamics. Due to the short wavelength of electrons, TEMs have a
much higher resolution, i.e., the ability to observe two closely
positioned organelles as discrete structures, than light microscopes
(LM). For comparison, the x-y resolution of the human eye, the
light microscope, the scanning electron microscope, and the trans-
mission electron microscope are in the order of >0.2 mm; >0.2 μm;
>1 nm; and >0.2 nm respectively.
Conventional TEM approaches use ultrathin sections (60–
70 nm) of usually chemically fixed and plastic-embedded cell and
tissue samples to study the subcellular structure and dynamics of
organelles. These methods have been applied in numerous platelet
and megakaryocyte (MK) studies [2–12]. Although still valuable,
conventional TEM approaches provide only ultrastructural images
of the cells with limited resolution in the Z-axis. Furthermore, the
relative low fixation speed and harsh fixation protocols (e.g., glu-
taraldehyde and OsO4) induce ultrastructural artifacts [13, 14],
and the embedding in epoxy resins limits antibody access and thus
detection of the molecular distribution of proteins [15]. The cryo-
sectioning and immunogold labeling technology (IEM) as devel-
oped by Tokuyasu and the group of Slot and Geuze [16, 17]
overcomes this latter problem and is now an accepted and widely
used technique for locating molecules in their subcellular context
with nanometer resolution. IEM has been perfected since its first
use and now surpasses classical resin TEM techniques, combining
optimal membrane preservation with high labeling sensitivity [18].
Over the past decade, real-time imaging and 3D electron
tomography (ET) have increasingly replaced conventional light
microscopy and immunofluorescence microscopy (LM and IF) and
Cryo-Preparation, Electron Tomography and CLEM 57
2D transmission electron microscopy, respectively. In addition, fast

freezing technologies (vitrification, high-pressure freezing, HPF)
have become available to arrest cellular membrane dynamics within
milliseconds, thereby providing a physiological “snapshot” of the
cell without the artifacts produced by chemical fixation [19–21].
Fast immobilization methods, combined with (cryo) TEM tomog-
raphy and 3D reconstruction allow the snap-frozen structures to be
reconstructed into 3D models [22–24]. This led to the develop-
ment of new approaches that combine light microscopy imaging
with high-resolution (cryo) electron microscopy (correlative light
and electron microscopy, CLEM) [25, 26]. CLEM utilizes the two
complementary visual techniques, enabling visualization of dynamic
cellular processes by LM/IF with direct coupling to high-resolu-
tion ultrastructural imaging representing these cellular events. This
chapter describes the preparation methods for entire platelet vitrifi-
cation, HPF for platelets and whole bone marrow, and methods for
(cryo)correlative imaging and electron tomography. In addition,
the future direction of modern EM technologies is discussed.
2 Materials
The different sample preparation procedures that we currently use

in our research projects are depicted schematically in Figs. 1 and 2.
Figure 1 shows the classical chemical fixation procedures and the
high-pressure freezing technology (HPF). These methods are
applicable for cells in suspension (isolated platelets and cultured
MKs) as well as whole bone marrow. Fixation is followed by either
plastic embedding (lane A), or by cryosectioning and immunogold
labeling (Tokuyasu method, lane B). The preparation method for
HPF-FS is shown in lane C. Figure 2 shows two CLEM proce-
dures to study adherent platelets. The protocols are given below.
2.1 Human Platelets 1. 0.1 M sodium citrate anticoagulant.

2. Acid/citrate/dextrose anticoagulant: 85 mM sodium citrate,
71 mM citric acid, and 111 mM d-glucose.
3. Prostacyclin (Cayman Chemical, USA): stock 25 μL aliquots
of 10 μg/mL stored at −80 °C (see Note 1).
4. Cell analyzer for assessment of mean platelet volume (MPV)
(e.g., Abbott Cell-Dyn 1800).
5. 0.1 M TRIS-buffered saline (TBS, pH 9.0): stored as frozen
225 μL aliquots.
6. Modified Tyrode buffer: 129 mM NaCl, 0.34 mM Na2HPO4,
2.9 mM KCl, 12 mM NaHCO3, 20 mM HEPES, 5 mM glu-
cose, 1 mM MgCl2 (pH 6.5 and pH 7.3).
Fig. 1 Schematic overview of the preparation procedures for resting and activated platelets in suspension. (a)
Classical chemical fixation followed by plastic embedding. (b) Cryosectioning and immunogold labeling
(Tokuyasu method). (c) HPF-FS. Note that chemical fixation and HPF-FS can also be applied to the study of
cells in whole bone-marrow or isolated MKs. For details of the procedures see text
2.2 Megakaryocytes 1. 8- to 10-week-old Balb/c mice (see Note 2).

and Other Marrow 2. Instrumentation for perfusion: fixation (refer to Chapter 14
Cells this volume).
3. Dissection instruments for removal of femurs.
4. 0.1 M sodium cacodylate buffer. Stock solution: 0.4 M (8.56 g
Na-cacodylate in 100 mL Milli-Q® water) (see Note 3).
Fig. 2 Schematic overview of two CLEM procedures used for studying platelet adhesion to a physiological
substrate. (a) whole mount room temperature CLEM. (b) cryo-CLEM procedure. For details of the procedures
see text
2.3 Resin Embedding 1. Graded series of ethanol (70%, 90%, 96%, 100% in Milli-Q®
and Sectioning water).
2. 1,2-propylene oxide.
3. Epon-812 embedding resin.
4. Flat embedding molds (clear silicone, e.g., EMS Cat#70900).
5. Ultramicrotome using a diamond knife.
6. Formvar and carbon-coated 200 mesh copper grids (Agar

Scientific, Essex, UK).
7. Uranyl acetate: 0.5% in Milli-Q® water.
8. Lead acetate: 3% in Milli-Q® water.
2.4 Tokuyasu 1. 0.2 M phosphate buffer (pH 7.4): 9.5 mL 0.2 M NaH2PO4

Method and 40.5 mL 0.2 M Na2HPO4 (pH adjusted to 7.4 by the two
components).
2. Fixative solution: 2% paraformaldehyde (PFA) and 0.2%
monomeric EM-grade glutaraldehyde (GA) in 0.1 M phos-
phate buffer.
3. PHEM buffer: 60 mM PIPES, 25 mM HEPES, 10 mM
EGTA, 4 mM MgSO4, adjusted to pH 7.0 with 5 M KOH.
4. 12% gelatin dissolved in PHEM buffer at 37 °C.
5. Sucrose-based cryoprotectant solution: 2.3 M sucrose in
0.1 M phosphate buffer (pH 7.4).
6. Small pins suitable for securing 1 mm3 gelatin blocks.
7. Any type of Whatman filter paper (to remove excess sucrose).
8. Liquid nitrogen storage system.
9. A cryo-ultramicrotome: e.g., Ultracut-S, Leica Microsystems,
Vienna, Austria.
10. Mixture of 2% methylcellulose and 2.3 M sucrose (Subheading
2.4, item 5).
11. If carrying out simultaneous immunohistochemistry for CLEM:
antibodies tagged with fluorophores and/or 10 nm protein A
gold (Cell Microscopy Core, UMCU, Utrecht, the Netherlands
(http://www.cellbiology-utrecht.nl/products.html).
12. 2% uranyl acetate or 2% uranyl oxalate. Both are dissolved in
Milli-Q® water with the pH adjusted to 7.0 with 25% NH4OH.
13. Mixture of 1.8% methylcellulose and 0.3% uranyl acetate in
Milli-Q® water.
2.5 High Pressure 1. HPF equipment is available from several providers (Leica
Freezing Microsystems, Vienna, Austria; Baltzers, Liechtenstein). We
have used the Leica EMPACT2 high pressure freezer and the
Leica EM AFS2 freeze substitution apparatus.
2. Flat HPF specimen carrier (0.2 mm deep and 1.2 mm diame-
ter carrier).
3. 20% human albumin serum (HAS) solution: 20% w/v in
Hepes-Tyrode solution + glucose 1 mg/mL.
4. Cryo substitution apparatus (AFS, Leica Microsystems).
5. 1.5 mL micro tubes.
6. Acetone-based substitution medium (95, 90, 80, and 70%

(v/v) acetone in Milli-Q® water).
7. Additional fixatives. We prefer a mixture of 0.5% glutaralde-
hyde (GA), 0.25% uranyl acetate (UA), 1% OsO4 (osmium)
and 3% H2O in anhydrous acetone (Merck).
8. PHEM buffer (see Subheading 2.4, item 3).
9. 50% and 30% (v/v) acetone in PHEM buffer.
10. Low melting point agarose (2%) in 0.1 M phosphate buffer.
2.6 Whole Mount 1. EM grids: either 200 mesh carbon-coated formvar grids (Agar
Protocol Scientific, Essex, UK), gold quantifoil grids (R2/2 Cat#
S173-7 or R3.5/1 Cat#S177-7 from PLANO, GmbH or
R2/2 from Ted Pella, USA), or gold lacey carbon grids (Cat#
LC300Au25, van Loenen Instruments, the Netherlands). For
correlative approaches carbon-coated gold finder grids can
also be used (Ted Pella, USA) (see Note 4).
2. Glow discharging unit (e.g., Edwards auto 306 HT).
3. Method of holding the grids for glow discharging: e.g., clamp-
ing in sheets with flexible sluts (Leica AC20 sheets are pre-
ferred), or alternatively they can be glued via their edges to
double-sided tape.
4. Fibrinogen: 100 μg/mL.
5. Hepes Tyrode buffer (129 mM NaCl, 0.34 mM Na2HPO4,
2.9 mM KCl, 12 mM NaHCO3, 20 mM HEPES, 5 mM glu-
cose, 1 mM MgCl2, pH 7.3).
6. Blocking buffer: 1% BSA in Hepes Tyrode buffer with 1 mg/
mL glucose, pH 7.3.
7. Humidifier apparatus such as a glass beaker or petri dish placed
upside-down on wetted filter paper.
8. Whatman filter paper.
9. For vWF surface labeling: primary anti-vWF antibody and
10 nm protein A gold. (In the case of monoclonal anti-vWF,
an intermediate bridging antibody should be used).
2.7 Correlated Light 1. CLEM Microscope system such as the iCorr™ (FEI,
and Electron Eindhoven, The Netherlands).
Microscopy (CLEM) The basic goal in CLEM studies is that regions of inter-
est, identified by their fluorescent signal at low magnification
by fluorescence microscopy (FM), can be subsequently ana-
lyzed at the ultrastructural level with TEM. There are several
ways to perform CLEM. The approach that we describe here
is based on the use of the iCorr™ (Fig. 3a–c, Supplementary
Movie 1) [27, 28]. The iCorr™ (FEI Company, Eindhoven,
The Netherlands) is a prototype that involves a Tecnai 12
Twin transmission electron microscope, equipped with a fully
Fig. 3 RT-CLEM of platelet whole mounts. Platelets spread on fibrinogen-coated EM supports are sequentially
immunolabeled with Alexa 488-conjugated anti-vWF and 10 nm protein A gold. (a) Adherent platelets undergo-
ing vWF secretion (highlighted region with fluorescent dots) are imaged at ambient temperature using a Tecnai
20 with integrated IF microscope (iCorrtm FEI Company). (b) The IF objective is withdrawn from the EM column
and the grid is switched 90° and EM overlays of ROIs representing vWF release are stored on the computer
using the iCorr software package mode. (c) Dual axis tilt series are recorded, aligned, and reconstructed using
the IMOD software package. Arrowhead in the highlighted right panel shows immunogold labeled vWF on the
surface of the adherent platelet. Bars from left to right a, 12.5 μm; b, 2.5 μm and 1.25 μm; c, 200 nm. See
Supplementary Movie 1
integrated LED-based wide field fluorescence microscope

(excitation light ranging from 460 to 500 nm) located at the
normal sample position in the TEM column (Fig. 3a). This
setup allows for consecutive acquisition of fluorescence and
TEM data (Fig. 3b) on the same grid within the same micro-
scope and enables the direct correlation of the fluorescent sig-
nal with ultrastructural features. The iCorr™ microscope is
equipped with an Eagle 4x4k CCD camera. Correlative imag-
ing is performed using the iCorr work flow [27, 28], which
creates a shared coordinate system for IF labeled ROIs and
TEM data recordings. Using the common EM specimen
stage, the sample is tilted 90° followed by insertion of the
objective lens of the optical unit close to the specimen. Light
microscopic images are then recorded in the fluorescence
mode and stored on the computer. After fluorescence imag-
ing (Fig. 3a), the iCorr™ is switched to TEM mode by which
the optical element is retracted and the specimen holder is
rotated back to its 0° position (Fig. 3b). The iCorr™ system
is suited for RT and frozen-hydrated samples. For cryo sam-
pling the specimen is inserted in a cryo specimen holder
(Gatan 626, Gatan, Abingdon, UK), ensuring that the speci-
men remains vitrified. The iCorr™ is also equipped with the

tomography data acquisition software package Xplore 3D
(FEI Company, Eindhoven, the Netherlands). This allows
us to immediately record tomographic data sets from the
fluorescently identified regions of interest.
In our studies platelets are allowed to spread on Au-carbon-
coated grids that have been functionalized with a physiological
substrate (fibrinogen, vWF). The adherent platelets are simulta-
neously labeled with antibody-conjugated fluorescent dyes and
gold particles, and sequentially imaged in cryo FM and TEM
mode in the iCorr electron microscope (Figs. 3 and 4). Correlative
approaches can be performed at room temperature (Fig. 3) or
under cryo conditions (Fig. 4), provided that the samples are thin
enough to be visualized in the TEM. CLEM can also be applied
on semi-thin Tokuyasu sections (Fig. 1, lane B). This can be very
useful in whole bone marrow for example, to identify specific
target cells and/or MK maturation stages in the crowded envi-
ronment of the bone marrow, or to identify MKs that have been
transfected to express specific GFP-tagged proteins. In all these
cases specific ROIs are first selected in IF mode and next analyzed
by electron tomography for ultrastructure analysis. We here
describe the protocol for two CLEM approaches for whole
adherent platelets using iCorr, but other approaches that make
use of finder grids and specific navigation software (MAPS and
CorrSight, FEI Company) can also be used [29].
2. Materials as described in Subheading 2.6, items 1–9.
3. Parafilm.
4. Alexa 488-conjugated anti vWF or anti-fibrinogen antibodies.
5. 10 nm protein A gold.
Fig. 4 Cryo-CLEM of platelets adhering to a fibrinogen substrate. The platelets are immunolabeled with Alexa
488-conjugated anti-vWF and 10 nm protein A gold. The grids are immediately plunge-frozen in liquid ethane,
transferred to a Gatan cryo-holder, and imaged in the cryo-stage of a Tecnai 20 with integrated IF microscope
(iCorrtm FEI Company). Regions of interest are recorded in both IF and EM mode using the iCorr software pack-
age. (a) Low magnification overview recorded in IF mode. (b) Highlighted areas representing spread platelets
with released vWF (IF dots). (c) High magnification with selected EM overlay of a single platelet suitable for
cryo electron tomographic recording. (d) Tomographic slice of area outlined in C taken after the tilt series
recording. The black dots represent immunolabeled vWF on the surface of the spread platelets and are used
as fiducial gold markers to aid alignment of the tilt series. Bars: a, 25 μm; b, 12.5 μm; c, 1.75 μm; d, 200 nm
2.7.1 RT-CLEM 1. 2% PFA and 0.2% GA in 0.1 M phosphate buffer (see

(Fig. 2, Lane A) Subheading 2.4, item 2).
2. 2% uranyl acetate (pH 7) or uranyl oxalate (pH 7), both dis-
solved in AD, (Aqua Dest distilled water by Medicalcorner24®),
pH adjusted with 1 M NaOH.
3. Mixture of 1.8% methylcellulose and 0.3% uranyl acetate in
Milli-Q® water.
4. Whatman filter paper.
2.7.2 Platelet Vitrification Since resting platelets in suspension are too thick for whole cell vitri-
and Cryo CLEM fication we use platelets spread on a fibrinogen substrate (see
(Fig. 2, Lane B) Subheadings 3.3.4 and 3.3.5; Materials as described in Subheading
2.6, items 1–9. Gold holey carbon quantifoil R2/2 (Ted Pella, USA)
or lacey carbon grids (van Loenen, the Netherlands) (see Note 5)).
1. Apparatus for vitrification such as the Vitrobot Mark IV (FEI,
Eindhoven, The Netherlands).
2. Whatman grade 4 qualitative filter paper, pore size 20–25 μm.
3. Liquid ethane and liquid nitrogen.
4. Cryo boxes suitable for storage of grids.
2.8 Equipment The specialized equipment required for cryo ET is discussed in

for Cryo Electron Subheading 3.4. For acquisition this includes a high-tilt cryo
Tomography tomography holder (e.g., Gatan 914 from Gatan Inc., USA) and a
and Analysis TEM (e.g., Tecnai 12 Twin TEM or Tecnai 20 STEM, equipped
with a 4x4K CCD Eagle camera; FEI Company, USA). In addi-
tion, specialized software is required for automated data acquisi-
tion and data analysis such as the Xplorer 3D package (FEI,
Eindhoven, the Netherlands) and the IMOD package [30] from
Colorado University (Boulder, USA).
3 Methods
3.1 Preparation First step in studying platelet adhesion and activation processes is
of Washed Human the isolation of platelets from whole blood (Fig. 1).
Platelets
1. Whole blood is drawn by venipuncture from healthy volun-
teers into 0.105 M sodium citrate anticoagulant (1 part acid
citrate to 9 parts blood) (see Note 6).
2. Centrifuge the whole blood at 160 × g for 15 min and remove
the upper platelet-rich plasma (PRP).
3. Determine the mean platelet volume (MPV) in the PRP using
a cell analyzer.
4. Add 1/10 volume of ACD buffer (containing 85 mM sodium
citrate, 71 mM citric acid and 111 mM d-glucose) and mix
gently.
5. The platelets are isolated from the PRP by centrifugation at

360 × g for 15 min.
6. The pellet is resuspended in modified Tyrode buffer, pH 6.5.
7. Thaw a 25 μL aliquot of 10 μg/mL prostacyclin and add
225 μL 0.1 M TRIS-buffered saline (TBS, pH 9.0) and keep
on ice.
8. Add prostacyclin 0.1 μg/mL (final concentration) to the
suspension.
9. The platelets are washed once via centrifugation (360 × g for
15 min) and suspended in modified Tyrode buffer, pH 7.3.
10. Remeasure the MPV, count the number of platelets and adjust
to the desired concentration (see Note 7).
11. Usually the platelet concentrations required for HPF of rest-
ing cells should be in the range of 600–2000 × 109/L. For the
spreading assays onto grids much lower densities are used
(≈100 × 109/L).
3.2 Preparation Follow local ethical guidelines and regulated procedures for perfu-
of Bone Marrow sion fixation of 8–10 week old BALB/c mice (see Chapter 14, this
Megakaryocytes volume for further details of a perfusion fixation protocol).
Dissect out the femurs, cut the epiphyses and flush with 0.1 M
sodium cacodylate buffer into a petri dish using a syringe (see
Chapters 12 and 13 for further guidance on flushing marrow).
Immediately process small pieces of the flushed marrow for
classical resin embedding (Subheading 3.3.1), the Tokuyasu
method (Subheading 3.3.2), or the HPF-FS method (Subheading
3.3.3).
3.3 Platelet Platelets have been analyzed at the ultrastructural level for many years
Preparation with what we now call “conventional electron microscopical meth-
Procedures ods” (Fig. 1, lane A). The conventional approach is based on chemi-
cal fixation, followed by resin embedding, sectioning of thin sections,
contrasting with heavy metals and then transmission electron micros-
copy. This is still a widely used approach but in the last decade several
alternative approaches and data recording and analysis methods have
been developed and applied by us and others in platelet research (see
Fig. 3 for comparison of the different fixation protocols).
3.3.1 Protocol for Resin 1. Centrifuge the platelets 1–2 min @ 8 K RCF in warm (approx-
Embedding (Fig. 1, Lane A) imately 40 °C) low melting point 2% agarose in 0.1 M
Phosphate buffer. Typically a 1 mL suspension of ≈600 × 109/L
is sufficient to obtain a pellet for resin embedding.
2. Solidify by placing the vial with pelleted platelets on ice.
3. Isolate the platelet pellet and dehydrate at room temperature
in a graded series of ethanol (70%, 90%, 96% for 15 min and
3 × 30 min in 100%, respectively).
4. Transfer the pellets into 1,2-propylene oxide (RT, 2 × 10 min).

5. Infiltrate the pellets with a series of resin-propylene oxide mix-
tures (1:3, 1:1, and 3:1, respectively) for 1 h each. We gener-
ally use Epon-812 as the embedding resin.
6. Finally, transfer the platelet pellets into pure resin for over-
night infiltration.
7. The next day the resin is refreshed and the platelet pellets are
incubated for another 2 h before resin polymerization in flat
embedding molds is performed at 60 °C.
8. After polymerization, section the resin blocks on an ultrami-
crotome using a diamond knife.
9. Collect serial gray sections floating in a water trough on form-
var and carbon-coated 200 mesh copper grids.
10. Stain with heavy metals to provide contrast to the membranes,
usually a combination of uranyl acetate followed by lead citrate
(see Note 8). Conditions: 35 min at 45 °C with 0.5% uranyl
acetate, followed by thorough rinsing with Milli-Q®, 20 min
at 25 °C, with 3% lead citrate, followed by rinsing with
Milli-Q® water and air drying.
3.3.2 Tokuyasu Method The Tokuyasu technique is named after his inventor Kiyoteru
(Fig. 1, Lane B) Tokuyasu and is further optimized in the lab of Slot and Geuze.
Since then the method has become the method of choice for high-
resolution immunogold localization studies of frozen specimens
(see for details of the procedure [18]).
1. Platelets are mildly fixed with a mixture of 2% paraformalde-
hyde and 0.2% glutaraldehyde in 0.1 M phosphate buffer for
60 min followed by centrifugation into a pellet.
2. After washing with 0.1 M phosphate buffer the pellet is
immersed for 10 min at 37 °C in 12% gelatin in PHEM buffer.
3. After gelation at 4 °C, small blocks (1 mm3) are trimmed and
infiltrated with a sucrose-based cryoprotectant solution at 4 °C.
4. Individual blocks are placed on small pins and excessive sucrose
solution is removed with a Whatman filter paper.
5. The samples are frozen in liquid nitrogen and stored until cryo
sectioning.
6. A cryo-ultramicrotome is used for sectioning 60–70 nm thin sec-
tions at temperatures between minus 80 °C and minus 140 °C.
7. The sections are collected on 200 mesh formvar-coated grids
using a wire loop filled with a drop of 1% (w/v) methylcellu-
lose and 1.15 M sucrose in PHEM buffer.
8. Simultaneous immunolabeling (Fig. 1, lane B) is performed by
floating the grids successively on drops containing antibodies
and detection probes (fluorophores and/or protein A gold).
9. Sections are post-stained with 2% uranyl acetate (pH 7) or

uranyl oxalate (pH 7), rinsed on successive drops of AD and
floated for 10 min at 4 °C on drops containing a mixture of
1.8% methylcellulose and 0.3% uranyl acetate (see Note 9).
10. The grids are lifted from the methylcellulose drops using a
small loop and excess methylcellulose is removed by touching
the edge of the loop to well-absorbing filter paper and drag-
ging the grid carefully along the filter paper edge until no
more uranyl acetate/methylcellulose comes off. In this way a
thin film of uranyl acetate/methylcellulose is left on the grid.
11. After air drying, the grids are ready for analysis in the electron
microscope.
3.3.3 High Pressure Membranes display the most obvious distortions when chemical
Freezing (HPF) and Freeze fixation protocols are used (Fig. 5a, d and e). Vesiculation or “bleb-
Substitution (FS) (Fig. 1, bing” of cellular and organelle membranes are frequently observed
Lane C) in aldehyde-fixed cells as a consequence of fast membrane flow and
local breakdown of membrane–cytoskeleton coupling [13]. In
terms of optimal preservation the best choice is imaging platelets
in the frozen-hydrated state (Fig. 5c, g). The freezing conditions
must be sufficiently rapid to prevent serious damage of the mem-
branes by ice crystals. The most widely used procedure at the
moment is high-pressure freezing (HPF) followed by freeze
substitution.
During HPF, platelets are pressurized to about 2000 bar and
then cooled by liquid nitrogen within milliseconds. The goal is to
extract the heat from a sample before cell water can rearrange into
ice crystals. At this level of pressure, the freezing point of water is
lowered down to about −20 °C, and the nucleation of ice crystals
as well as their growth is slowed down. For more details about the
theory behind this method, see the articles by Riehle and Höchli
[19], Studer et al. [20], and Vanhecke et al. [21]. HPF is capable
of freezing cells and tissue samples up to 200 μm (see Note 10).
1. Platelet preparation: Since PRP contains insufficient platelets
for HPF-FS, enriched samples of washed platelets are prepared
by centrifuging for 15 min at 300 × g, and resuspension at
densities >600 × 109/L in 20% human albumin serum.
2. Marrow preparation: For the study of megakaryocytes or other
bone marrow cells by HPF, fresh mouse bone marrow is har-
vested from the femurs of 8- to 10-week-old BALB/c mice by
flushing with 0.1 M sodium cacodylate buffer into a petri dish
using a syringe. Small pieces of the flushed marrow are
then immediately transferred into the carrier of the HPF.
An example of a mouse bone marrow sample prepared accord-
ing the HPF-FS protocol is shown in Fig. 6.
Fig. 5 Effect of different fixation and preparation protocols on membrane morphology. (a–c) Electron tomogra-
phy generated by three different fixation protocols: (a) chemical fixation, dehydration, and embedding in Epon;
(b) high-pressure freezing (HPF) followed by low-temperature freeze substitution and plastic embedding in
Epon; (c) whole vitrified adherent platelet. (d–g) Alpha granule substructure following different fixation and
preparation protocols; (d, f, g) are thin slices (≈5 nm) extracted from tomograms, while (e) is a 60 nm cryosec-
tion: (d) conventional chemical fixation and plastic embedding; (e) Tokuyasu method, immunogold labeling of
vWF (double arrowheads indicate luminal vesicles); (f) HPF/FS and plastic embedding; (g) vitrified platelet.
Note that alpha granules show membrane distortions and shrinkage in (d, e) compared to the more regular
limiting membranes after HPF and vitrification (f, g). The membrane preservation in chemical fixed thin-frozen
sections is close to the conventional resin-embedded sections. The membrane preservation in HPF/FS method
is much closer to the native vitreous state than the chemical-fixed sections. Bars: a, 100 nm; b and c, 200 nm;
d–g, 50 nm
3. Precoat the flat HPF specimen carrier (0.2 mm deep and

1.2 mm diameter carrier) of the HPF apparatus with 20% HAS
solution.
4. Platelets in 20% human albumin solution (HAS), or fresh bone
marrow samples are transferred into the flat specimen carrier
(e.g., 0.8 μL platelet suspension in 0.2 mm deep and 1.2 mm
diameter carrier) of the HPF apparatus.
5. Cryo immobilization is performed at a pressure of 2000 bar
(2 × 108 Pa) according to the manufacturer’s manual within
1 min after transfer of the platelets.
6. The cryo-fixed carrier samples are then transferred to the
acetone-based substitution medium in 1.5 mL micro tubes
placed at minus 90 °C in a cryo substitution apparatus. Here,
frozen water in the sample is replaced by the precooled sub-
stitution fluid. Chemical fixatives (uranyl acetate, osmium
Fig. 6 Electron tomography of erythroblast from mouse bone marrow, prepared according to the HPF-FS pro-
tocol (see Fig. 1, lane C). (a) Series of 6 tomographic slices through a 300 nm thick section taken at different
z-axes. Numbers in the top left indicate different z-positions in nm. The images show an autophagosome (star)
containing a mitochondrion. (b, c) 3D reconstruction and modeling. pm plasma membrane, m mitochondrion,
e endosome (transparent yellow) containing ferritin particles (red). Bars: A, 50 nm; C, 20 nm
tetroxide, glutaraldehyde) in different combinations can be

added [14, 31]. We prefer a mixture of 0.5% glutaraldehyde
(GA), 0.25% uranyl acetate (UA), 1% OsO4 (osmium), and
3% H2O in anhydrous acetone. The platelets are then dehy-
drated and fixed in this solution as follows.
7. Keep at −90 °C for 48 h.
8. The temperature is raised to −60 °C (2 °C/h).
9. Samples are kept at −60 °C for 8 h.
10. Next the temperature is raised to −30 °C (2 °C/h).
11. Keep at −30 °C for 8 h. When the substitution solution con-
tains UA it is washed off by rinsing four times with the same
substitution medium but without UA.
12. The samples are removed from the substitution apparatus and
placed on ice for 1 h.
13. After dehydration and fixation during cryo substitution, the
platelets are transferred to an Epon-acetone mixture for plastic
embedding (as described in Subheading 3.3.1).
14. After step 11 in the HPF-FS procedure the fixed samples can
also be rehydrated on ice in six steps of 10 min each in subse-
quent series of 95, 90, 80, and 70% (v/v) acetone in distilled
water, then 50% (v/v) acetone in PHEM buffer and finally in
30% acetone in PHEM buffer (see Note 11 and ref. 31.
15. Platelets are washed four times for 10 min in PHEM buffer.
16. Finally, the platelets are immersed for 10 min at 37 °C in 12%
gelatin in PHEM buffer.
17. After gelation at 4 °C small blocks are trimmed for 2.3%

sucrose infusion and cryo sectioning (see Note 12) as described
for the Tokuyasu procedure (see Subheading 3.3.2). The only
difference is that in the section fixation procedure, phosphate-
containing buffers are avoided at all stages.
3.3.4 Whole Mount The limited thickness of spread platelets allows the application of a
Procedure (Fig. 2, Lane A) technique that we have named “whole mount electron tomogra-
phy.” To this end, platelets are allowed to adhere to EM supports.
Platelets spread equally well on formvar, quantifoil, or lacey carbon
grids, provided that they have been functionalized with a physiologi-
cal substrate (fibrinogen or vWF, Fig. 7). For tomography p urposes
fiducial markers (i.e., 10 nm colloidal gold particles coupled to pro-
tein A) can be applied to the grids. The fiducial gold markers are
extremely convenient for aligning the recorded data set of projection
images to make a tomogram. Adherent platelets can also be quickly
Fig. 7 Platelets spread equally over formvar carbon-coated grids (a), gold lacey-carbon (b), and gold quantifoil
grids (c), provided that they are glow-discharged and precoated with a physiological substrate (i.e., fibrinogen
or vWF). For correlative purposes carbon-coated Au-finder grids can also be used. Bars: A, 3 μm; B and C, 2 μm
immuno-gold labeled (after step 6) using antibodies that identify a

protein of interest. These specific gold particles can then be simulta-
neously used as fiducial markers.
1. The grids are freshly glow discharged. For this process, the
grids can be clamped in sheets with flexible sluts (Leica AC20
sheets are preferred), or alternatively glued with their edges to
double-sided tape (see Note 13).
2. The grids are coated for 30 min at room temperature (RT)
with fibrinogen (100 μg/mL) in a humidified environment.
3. Excess fibrinogen is removed using a piece of fast absorbing
Whatman paper.
4. The grids are then exposed to blocking buffer for 30 min at
RT, still in a humidified environment.
5. Excess blocking buffer is removed with Whatman paper.
6. Small drops of the washed platelet suspension (100–
150 × 106/L) are added to the grids and platelets allowed to
settle for ~7 min at RT on the fibrinogen substrate.
7. The grids with attached platelets are removed from the flexible
sluts and immediately put (with platelets facing down) on suc-
cessive drops of Hepes Tyrode buffer with glucose. The drops
are placed on Parafilm and covered with petri dishes to prevent
dust contamination. Unbound platelets are washed away over
several drops of buffer.
8. To allow platelets to spread further, the grids are left for an addi-
tional 10–12 min at RT on Hepes Tyrode buffer with glucose.
9. vWF surface labeling can be performed by transferring the
grid to successive drops containing anti-vWF antibody (~2 min
is sufficient) in Hepes Tyrode with glucose and 10 nm protein
A gold (3 min) (see Note 14).
3.3.5 Correlated Light 1. Carbon-coated formvar, gold lacey carbon, or quantifoil grids
and Electron Microscopy are glow-discharged and coated with 100 μg/mL fibrinogen
CLEM (Fig. 2) or 100 μg/mL vWF as described in Subheading 3.3.4, steps
1–4 (see Note 15).
2. After a BSA block, washed platelets (~100 × 109/L) are allowed
to spread on the grids (see Subheading 3.3.4, steps 5–6).
3. After spreading for 20 min on the physiological substrate, the
intact platelets are immunolabeled (2 min) by floating them
on small drops on Parafilm containing Alexa 488-conjugated
anti-
vWF or anti-fibrinogen antibodies diluted in Hepes
Tyrode buffer with glucose.
4. Rinse several times on successive drops of Hepes Tyrode buf-
fer with glucose.
5. Incubate with protein A gold (3 min).
6. Quickly rinse 3× on drops in Hepes Tyrode with glucose.
From here the sample preparation methods for RT-CLEM and
cryo CLEM go separate ways. RT-CLEM requires chemical fixa-
tion whereas for cryo CLEM the adherent platelets are vitrified.
RT-CLEM (Fig. 2, Lane A, 1. For RT-CLEM, platelets are chemically fixed with 2% PFA and
Supplementary Movie 1) 0.2% GA in 0.1 M phosphate buffer.
2. The adherent platelets are rinsed 5 × 1 min on successive drops
of PBS, followed by 5× rinsing on Milli-Q® water drops.
3. Analogous to the Tokuyasu sections (see Subheading 3.3), the
whole mount platelets are stained for 5 min by floating the
grids on 2% uranyl acetate (pH 7) or uranyl oxalate (pH 7).
4. Rinse 3 × 1 min with Milli-Q® water.
5. Transfer the grid to a drop containing a mixture of 1.8% meth-
ylcellulose (MC) and 0.3% uranyl acetate (UA) on ice and
leave for 10 min at 4 °C.
6. Pick up the grid with a loop and touch the edge of the loop to
well-absorbing filter paper and drag the grid carefully along
the filter paper edge until no more UA/MC comes off into
the filter paper. In this way a thin even film of UA/MC is left
on the grid.
7. The grid remains adhered to the loop until it is air-dried.
8. Grids can now be inspected in the i-Corr electron microscope
successively in IF and TEM mode using the iCorr work pack-
age (Fig. 3) [27] (see Note 16).
Cryo CLEM (Fig. 2, Lane B) 1. For cryo CLEM grids are immediately transferred to the
Vitrobot Mark IV and plunged into liquid ethane for vitrifica-
tion (see Subheading 3.3.6).
2. Grids are stored under liquid nitrogen until transfer to the
cryo-stage of the iCorr electron microscope.
3. In the iCorr platelets are inspected successively in IF and EM
mode using the iCorr work package and tomographic datasets
can be recorded of regions of interest (Fig. 4).
3.3.6 Platelet Vitrification Accurate ultrastructural analysis of platelet organelles and simultane-
(Fig. 1, Lane C; Fig. 2, ous visualization of macromolecular complexes in their native state
Lane B) requires rapid freezing procedures in combination with high resolu-
tion cryo tomography of the whole vitrified platelet. Vitrification
immobilizes the membrane dynamics in milliseconds, while preserv-
ing the native molecular structure in the hydrated state.
Two strategies can be applied to vitrify tissue samples or plate-
lets. Whereas HPF is capable of freezing cells and tissue samples up
to 200 μm, plunge freezing is suitable for cells up to 10 μm in
diameter. HPF vitrified samples are generally too thick for tomog-
raphy and must be thinned. This can be achieved by cryo electron
microscopy of vitreous sections (CEMOVIS) or by cryo Focused
Ion Beam (FIB) milling. Both CEMOVIS and cryo FIB milling
are beyond the scope of this chapter. For technical details and pro-
cedures of these techniques we refer to the excellent review papers
on these topics [32, 33].
The relative small thickness of blood platelets, particularly
when spread on a physiological substrate like fibrinogen or vWF,
allows for whole cell vitrification and offers possibilities for whole
cell cryo tomography. Cryo TEM is a demanding “expert” tech-
nique that requires expertise and advanced equipment and
addresses special demands on platelet preparation methods. We
here describe our methods to vitrify resting and substrate-adher-
ent human blood platelets.
1. For plunge vitrification we use a Vitrobot Mark IV operating
at 37 °C and 100% humidity.
2. For preparation of adherent cell specimens, see Subheading 3.3.4.
3. After the final rinsing step (Subheading 3.3.4, step 9) the
grids are immediately transferred to the Vitrobot.
4. Excess liquid is removed by blotting either manually or auto-
mated (see Note 17) from both sides with Whatman paper.
The fluid absorbing speed of Whatman paper, blotting force
and duration are crucial to obtain a thin layer of ice and need
to be adjusted experimentally.
5. Immediately after blotting the grid is plunged into liquid eth-
ane to create the vitrified sample. The ethane should be partly
solidified. Retract the vitrified grid slowly out of the ethane
and blot away remaining ethane with a piece of Whatman
paper precooled with liquid nitrogen.
6. Plunge-frozen grids are transferred to cryo boxes and stored
in a liquid nitrogen container until transfer to the cryo-stage
of the electron microscope.
3.3.7 Labeling Immunogold labeling on thin Tokuyasu cryo sections of chemi-

of Platelets cally fixed cells is one of the most favorable protocols available, and
has been used in many MK and platelet studies [34, 35]. With this
IEM or so-called Tokuyasu method, sensitive immunoreactions are
achieved in non-resin-embedded thin cryo sections of room tem-
perature fixed cells. Fixation is usually with a mixture of formalde-
hyde and/or low concentrations of glutaraldehyde. The advantage
of this method is that the cells do not go through a series of dehy-
dration processes and embedding in resins, and thus maintain epi-
tope access for labeling. Furthermore, the use of different sizes of
gold particles guarantee an optimal resolution, and make double
labeling and colocalization studies a simple option. Many EM labs
can perform the basic steps of fixing (either chemical or the more
advanced high-pressure freezing followed by freeze-substitution),
resin embedding, and sectioning and contrasting of sections. These
approaches usually provide good morphological quality for tomog-
raphy analysis, but are limited in immune localization options
because penetration of antibodies and gold probes into the plastic-
embedded section is not possible. Electron tomography can also
be applied on semi-thin (~150 nm) immunolabeled cryo sections.
Although the labeling is restricted to the section surface, z-axis
information is still obtained in the thin subsequent slices of the
tomogram (Fig. 8). This method can also be applied on HPF-
frozen and rehydrated cryo sections [31].
Fig. 8 Electron tomography of 150-nm semi-thin cryo-section, prepared according to the classical Tokuyasu
method. Platelets were stimulated for 30 s with CRP. Series of sequential tomographic slices show immuno-
gold labeling of KDEL in reticular tubule-vesicular structures in close position to alpha granules representing
the DTS (arrowheads). Note that the gold label is restricted to the top of the section (upper slices). Bar 50 nm
3.4 Automated Data Electron tomography (ET) is a general approach that we apply to
Acquisition Methods obtain three-dimensional (3D) information regardless of the plate-
for Tomography let preparation pathway used. It is based on an old concept, in
which projection images of thin specimens (sections or whole
3.4.1 Tomography
mount platelets in the range of 200 to 400 nm) are acquired with
an electron microscope by tilting the specimen through a range of
tilt angles (typically −60° to +60°) at a predefined interval [36,
37]. However, due to increased sample thickness at higher tilt
angles and mechanical limitations by the holders, the tilting range
is limited. This results in loss of information (the so-called missing
wedge) in the final tomogram. From HPF and chemically fixed
thick sections we can generate a dual axis tilt series. This means
that after the first tilt series, the specimen is rotated 90°, and a sec-
ond tilt series is recorded of the same area. The two tilt series are
then combined into one tomogram.
Imaging vitreous platelets for tomography is an additional chal-
lenging aspect of the technique. It is more complicated due to two
factors—the low inherent contrast of the sample and the high sen-
sitivity to electron radiation damage. The sensitivity for electron
damage places limitations on the cryo-ET data acquisition. These
include the magnification (image pixel size), the tilting scheme (sin-
gle axis vs dual axis), and the amount of signal to noise in the images
[38]. The grids with adherent vitrified platelets are mounted in a
Gatan 914 high-tilt cryo-tomography holder (Gatan Inc., USA),
and transferred to the stage of the Tecnai 12 Twin TEM or Tecnai
20, equipped with a 4x4K CCD Eagle camera (FEI Company,

USA). Due to radiation sensitivity only single-axis tilt series are col-
lected (electron dose <20 e/Å2), using a goniometer tilting range
from −60 to +60, with angular increment of 2°. Total electron dose
for the entire tilt series should not exceed 100 e−/Å2. To increase
contrast we use a defocus range from −8 to −12 μm.
3.4.2 Software Besides hardware, specialized software is required for automated data
acquisition and data analysis. A broad range of software packages is
available for these purposes. Some are commercially available (often
expensive) others are academic (much cheaper or even free available).
The main goal of the automated data acquisition software is to accu-
rately collect a tilt series of digital images from a fixed location on the
specimen. Irregularities in stage movement (x, y) and focusing (z)
during the recording of the tilt series must be corrected for. We use
the Xplorer 3D package (FEI, Eindhoven, The Netherlands) for data
acquisition at our electron microscopes. The 3D reconstruction soft-
ware performs two tasks: alignment and reconstruction. Currently,
we use the IMOD package [30] from Colorado University (Boulder,
USA) to align the recorded projection images to create our platelet
tomograms, preferably by using fiducial markers. The platelet volume
can then be analyzed by visual inspection and segmentation of spe-
cific elements of interest. For example, the autophagosome and mito-
chondrion as depicted in Fig. 6b, c.
The resolution-weighted back projection algorithm [39],
which is implemented in the IMOD package is a widely used
method for computing the tomogram. For a more detailed descrip-
tion of this package we refer to Kremer et al. [30]. In cases of very
low contrast images (i.e., in low-dose cryo-tomograms), we use
the SIRT (Simultaneous Iterative Reconstruction Technique) a
modules that is also available in the IMOD package. Depending on
your computer the weighted back projection algorithm is relatively
fast. The SIRT reconstruction algorithm has the disadvantage that
it is computationally more intensive.
3.5 Future Several of the methods described above, in particular the cellular
Perspectives cryo-electron tomography (cryo ET) are time consuming and
require high technical skills. Yet the future looks bright for cryo
ET. The so-called resolution revolution in structural biology [40],
where cryo EM is slowly replacing the more traditional X-Ray crys-
tallography opens up great opportunities for cellular cryo ET. It is
therefore to be expected that cellular cryo ET will become a domi-
nant technique for studying the structure of dynamic interacting
macromolecules at sub-nanometer resolution in the native context
of the cell. The increasing interest in cryo EM and the combined
forces of many cryo EM laboratories and companies will result in
rapid improvements in the entire workflow of cryo
EM. Improvements are currently being made in automated cryo
specimen preparation, such as cryo focused ion-beam (cryo FIB)

milling of snap-frozen whole cells, automated approaches for cor-
relative cryo LM and TEM, and TEM instrument developments
(phase plates, direct detection cameras). Undoubtedly these
improvements will enter the platelet and MK research field and
result in exciting new discoveries.
4 Notes
1. To study resting platelets we add 1/10 volume of ACD buffer

(containing: 85 mM sodium citrate, 71 mM citric acid, and
111 mM d-glucose) and 1 μg/mL prostacyclin (final
concentration).
2. All animal work must follow local and national ethical guide-
lines and be carried out by licensed, competent staff.
3. All buffer and other aqueous solutions are prepared with
Ultrapure Water obtained from a Milli-Q® Integral Water
Purification System.
4. In the case of vitrification of adherent platelets, carbon-coated
formvar grids and copper lacey carbon or quantifoil grids
should be omitted because of thick ice formation and the
induction of cytotoxic effects during adhesion, respectively.
5. When platelets are allowed to spread on a physiological sub-
strate it is necessary to use golden quantifoil grids because plate-
lets are very sensitive to copper ions and become activated.
6. Extraction of blood must only be taken by a trained and compe-
tent phlebotomist after obtaining informed consent from the
donor and authorisation by the relevant local ethical committee.
7. Normal MPV (mean platelet volume) for resting platelets is
6.0–8 fL, and should not increase by more than 1.5 fL during
the platelet washing procedure.
8. The heavy metal staining can be performed manually, as
described here, or automatically with the Leica EM AC20
automated stainer.
9. To prevent uranyl phosphate precipitation, phosphate-containing
buffers should be avoided in the final rinsing steps before stain-
ing with uranyl acetate and embedding in methyl-cellulose.
10. HPF equipment is available from several providers (Leica

Microsystems, Vienna, Austria; Baltzers, Liechtenstein). We
have used the Leica EMPACT2 high pressure freezer and the
Leica EM AFS2 freeze substitution apparatus.
11. Recently, a novel IEM method has been developed by which
cells are fixed by HPF and then rehydrated and processed for
cryosectioning and immunolabeling according to the Tokuyasu

method (Fig. 1, lane C) [31].
12. Alternatively, direct cryo sectioning of HPF frozen platelets can
also be performed (CEMOVIS) [32]. The frozen platelet sam-
ples are transferred to a cryo microtome specimen holder and
glued with a drop of a viscous mixture of 2-propanol/ethanol
(3/2), at −145 °C that hardens when the temperature is again
lowered to −160 °C. Trimming and sectioning can then be per-
formed at −160 °C using a dry diamond knife (Element Six
B.V., Cuijk, the Netherlands or Diatome AG, Biel, Switzerland).
Sections can be pick-up directly on sandwiched grids for direct
transfer to the cryo-stage of the TEM or picked-up with a fixa-
tive-containing solution for fixation during thawing). This latter
so-called SFM method has only been used in two studies so far,
see for additional details reference [31, 41].
13. Clamping the quantifoil grids into the flexible sheet is a critical
point. Care should be taken that the fragile carbon layer is not
touched. This is best achieved using a binocular microscope or
a well-illuminated loupe.
14. The gold particles are conjugated to Protein A, which is used
here as the secondary labeling step. To improve efficiency of
labeling or in cases that certain moAbs do not recognize
Protein A, an intermediate Ab is used.
15. To facilitate correlation also finder grids (Ted Pella, USA) can
be used.
16. Since the electron beam bleaches the fluorescence signal very rap-
idly, particularly at room temperature, make sure that all IF imag-
ing and recording are done prior to the analysis in TEM mode.
Tomographic data sets can be recorded from regions of interest.
For room temp iCorr imaging and tomography, grids with spread
platelets can also be first fixed after spreading, followed by the
same immunolabeling procedure but then for 30 min.
17. Blotting is a delicate and critical step in cryo sample preparation,
especially when adherent whole cells are used. It determines the
ice film thickness and the final quality of the preparation. Most
devices, e.g., the Vitrobot (FEI), offer automated blotting and
plunging functions. Manual blotting, however, may sometimes
be useful to prevent cell damage. Excess solution is removed by
blotting with a piece of Whatman filter paper, producing a thin
liquid film spanning the holes of the grid. The exact amount of
blotting time and force used, in both manual and automated
devices must be adjusted to the nature of the sample. For both
adherent and resting platelets we have used blotting times
between 5 and 10 s and blotting forces up to −5.
References
1. Heijnen HFG, Korporaal SJA (2017) Platelet in quantitation of a soluble protein in ultra-
morphology and ultrastructure. In: Gresele thin sections. J Histochem Cytochem 35(4):
P et al (eds) Platelets in thrombotic and 405–410
non-thrombotic disorders, pathophysiol- 16. Tokuyasu KT, Singer SJ (1976) Improved pro-
ogy, pharmacology and therapeutics. Springer cedures for immunoferritin labeling of ultra-
International Publishing AG, Basel, pp 21–37 thin frozen sections. J Cell Biol 71(3):894–906
2. Behnke O (1968) An electron microscope 17. Liou W, Geuze HJ, Slot JW (1996) Improving
study of the megacaryocyte of the rat bone structural integrity of cryosections for
marrow. I. The development of the demarca- immunogold labeling. Histochem Cell Biol
tion membrane system and the platelet surface 106(1):41–58
coat. J Ultrastruct Res 24(5):412–433 18. Slot JW, Geuze HJ (2007) Cryosectioning and
3. Behnke O (1968) Electron microscopical obser- immunolabeling. Nat Protoc 2(10):2480
vations on the surface coating of human blood 19. Riehle U, Hochli M (1973) In: Benedetti EL,
platelets. J Ultrastruct Res 24(1–2):51–69 Favard P (eds) In freeze-etching technique and
4. Behnke O (1970) The morphology of blood applications. Society Francaise de Microscopy
platelet membrane systems. Ser Haematol Electronique, Paris, pp 31–66
3(4):3–16 20. Studer D, Graber W, Al-Amoudi A, Eggli P
5. Behnke O, Forer A (1998) From megakaryo- (2001) A new approach for cryofixation by high-
cytes to platelets: platelet morphogenesis takes pressure freezing. J Microsc 203(3):285–294
place in the bloodstream. Eur J Haematol 21. Vanhecke D, Graber W, Studer D (2008)
Suppl 61:3–23 Chapter 9 close-to-native ultrastructural pres-
6. Morgenstern E (1982) Coated membranes in ervation by high pressure freezing. Methods
blood platelets. Eur J Cell Biol 26(2):315 Cell Biol 88:151–164
7. Bentfeld-Barker ME, Bainton DF (1982) 22. Valentijn KM, Valentijn JA, Jansen KA, Koster
Identification of primary lysosomes in AJ (2008) A new look at Weibel-Palade body
human megakaryocytes and platelets. Blood structure in endothelial cells using electron
59(3):472–481 tomography. J Struct Biol 161(3):447
8. Stenberg PE, Shuman MA, Levine SP, Bainton 23. Berriman JA, Li S, Hewlett LJ, Wasilewski S,
DF (1984) Redistribution of alpha-granules Kiskin FN, Carter T, Hannah MJ, Rosenthal
and their contents in thrombin-stimulated PB (2009) Structural organization of Weibel-
platelets. J Cell Biol 98(2):748–760 Palade bodies revealed by cryo-EM of vit-
9. White JG (1968) The substructure of human rified endothelial cells. Proc Natl Acad Sci
platelet microtubules. Blood 32(4):638–648 106(41):17407–17412
10. White JG (1968) Fine structural alterations 24. van Nispen tot Pannerden H, de Haas F,
induced in platelets by adenosine diphosphate. Geerts W, Posthuma G, van Dijk S, Heijnen
Blood 31(5):604–622 HF (2010) The platelet interior revisited: elec-
11. White JG, Clawson CC (1980) The surface- tron tomography reveals tubular alpha-granule
connected canalicular system of blood subtypes. Blood 116(7):1147–1156
platelets–a fenestrated membrane system. Am 25. Valentijn KM, van Driel LF, Mourik MJ,
J Pathol 101(2):353–364 Hendriks GJ, Arends TJ, Koster AJ, Valentijn
12. White JG (1998) Use of the electron micro- JA (2010) Multigranular exocytosis of Weibel –
scope for diagnosis of platelet disorders. Semin Palade bodies in vascular endothelial cells. Blood
Thromb Hemost 24(2):163–168 116(10):1807–1816
13. Morgenstern E (1991) Aldehyde fixation 26. Mourik MJ, Faas FGA, Valentijn KM,
causes membrane vesiculation during plate- Valentijn JA, Eikenboom JC, Koster AJ (2014)
let exocytosis: a freeze-substitution study. Correlative light microscopy and electron
Scanning Microsc Suppl 5(4):S109–S115 tomography to study van Willebrand fac-
14. Murk JLAN, Posthuma G, Koster AJ, Geuze tor exocytosis from vascular endothelial cells.
HJ, Verkleij AJ, Kleijmeer MJ, Humbel BM Methods Cell Biol 124:71–92
(2003) Influence of aldehyde fixation on the 27. Agronskaia AV, Valentijn JA, van Driel LF,
morphology of endosomes and lysosomes: Schneijdenberg CTWM, Humbel BM, van
quantitative analysis and electron tomography. Bergen en Henegouwen PMP, Verkleij AJ,
J Microsc 212(1):81–90 Koster AJ, Gerritsen HC (2008) Integrated
15. Posthuma G, Slot JW, Geuze HJ (1987) fluorescence and transmission electron micros-
Usefulness of the immunogold technique copy. J Struct Biol 164(2):183–189
28. Faas FGA, Bárcena M, Agronskaia AV, Gerritsen transport in human platelets via the transloca-
HC, Moscicka KB, Diebolder CA, van Driel tion of the glucose transporter GLUT-3 from
LF, Limpens RWAL, Bos E, Ravelli RBG, alpha-granules to the cell surface. J Cell Biol
Koning RI, Koster AJ (2013) Localization 138(2):323–330
of fluorescently labeled structures in frozen- 35. Heijnen HF, Debili N, Vainchencker W,
hydrated samples using integrated light elec- Breton-Gorius J, Geuze HJ, Sixma JJ (1998)
tron microscopy. J Struct Biol 181(3):283–290 Multivesicular bodies are an intermediate stage
29. de Boer P, Hoogenboom JP, Giepmans BNG in the formation of platelet alpha-granules.
(2015) Correlated light and electron micros- Blood 91(7):2313–2325
copy: ultrastructure lights up! Nat Methods 36. Hoppe W (1974) Towards three-dimensional
12(6):503–513 "electron microscopy" at atomic resolution.
30. Kremer JR, Mastronarde DN, McIntosh Naturwissenschaften 61(6):239–249
JR (1996) Computer visualization of three- 37. Baumeister W, Grimm R, Walz J (1999)
dimensional image data using IMOD. J Struct Electron tomography of molecules and cells.
Biol 116(1):71 Trends Cell Biol 9(2):81–85
31. van Donselaar EG, Posthuma G, Zeuschner 38. Diebolder CA, Koster AJ, Koning RI (2012)
D, Humbel BM, Slot JW (2007) Immunogold Pushing the resolution limits in cryo electron
labeling of cryosections from high-pressure tomography of biological structures. J Microsc
frozen cells. Traffic 8(5):471–485 248(1):1–5
32. Al-Amoudi A, Chang JJ, Leforestier A, 39. Gilbert PFC (1972) The reconstruction of a
McDowall A, Salamin LM, Norlén LP, Richter three-dimensional structure from projections
K, Blanc NS, Studer D, Dubochet J (2004) and its application to electron microscopy. II
Cryo-electron microscopy of vitreous sections. direct methods. Proc R Soc London B Ser Biol
EMBO J 23(18):3583–3588 Sci 182:89–102
33. Schaffer M, Engel BD, Laugks T, Mahamid 40. Kühlbrandt W (2014) The resolution revolu-
J, Plitzko JM, Baumeister W (2015) Cryo- tion. Science 343(6178):1443–1444
focused ion beam sample preparation for imag- 41. Karreman MA, van Donselaar EG, Gerritsen
ing vitreous cells by cryo-electron tomography. HC, Verrips CT, Verkleij AJ (2011) VIS2FIX:
Bio Protoc 5(17):e1575 a high-speed fixation method for immuno-
34. Heijnen HF, Oorschot V, Sixma JJ, Slot JW, electron microscopy. Traffic 12(7):806–814
James DE (1997) Thrombin stimulates glucose
Chapter 5
Screening and High-Throughput Platelet Assays

Alexander P. Bye, Amanda J. Unsworth, and Jonathan M. Gibbins
Abstract
High-throughput assays are important biological research tools but are rarely utilized for platelet research.
However, screening compounds for efficacy against a physiologically relevant cellular response in primary
cells such as platelets can be an advantageous approach to compound screening and drug development. In
this section we describe a panel of three high-throughput microtiter plate assays designed for platelets that
can be used as the basis for compound screening, or be modified and used individually to increase through-
put in platelet research laboratories. The platelet adhesion assay has the lowest requirement for platelet
numbers and is therefore capable of the greatest throughput and so is suggested as the primary screen used
to identify hits. A secondary screen against the “gold standard” of platelet function, aggregation, is used
to confirm and further characterize hits. Finally, a Ca2+ assay is used for initial mechanistic characterization
to begin the process of elucidating the mode of action of hit compounds.
Key words Platelet, High-throughput, High-content, Screening, Ca2+, Adhesion, Aggregation
1 Introduction
Screening on the basis of clinically relevant functions using primary

cells is referred to as physiological or phenotypic screening [1] and,
by identifying compounds that exert a desired clinical effect on a
target tissue, bypasses the need for initial target identification and
screening using a cell line expressing a target protein [2]. Two key
functions of platelet are their ability to adhere to proteins of the
subendothelial matrix and to subsequently bind each other to form
aggregates. These processes contribute to the physiological role of
platelets in hemostasis and pathophysiological role in thrombosis.
High-throughput functional assays that measure adhesion and
aggregation form the basis of the screening approach described in
this chapter to identify “hit” molecules that modulate platelet
function. Cytosolic Ca2+ is a master regulator of platelet function
and is common to nearly all platelet signaling pathways [3] and is
therefore a versatile signaling output that can be used for initial
mechanistic characterization of “hit” molecules.
81
This chapter describes a panel of high-throughput techniques

to characterize the effects of compounds on three aspects of plate-
let function; adhesion, aggregation and Ca2+ signaling. The minia-
turized platelet assays measure these outputs in microtiter plates
and may be applied in sequence to screen compound libraries and
to give comprehensive characterization of “hits” that modulate
platelet function. Alternatively, the methods may be adapted and
utilized individually to reduce costs, increase throughput, and
facilitate a pharmacological approach to platelet research.
Platelet adhesion assays are commonly performed on glass cov-
erslips coated in matrix proteins such as collagen. Our approach
utilizes 96-well microtiter plates, but the assay could easily be min-
iaturized for use in 384-well plates given the correct setup. The
rationale behind using the platelet adhesion assay as the primary
high-throughput screening test is the very low number of platelets
required per test, at 1e6 platelets per well in a half-area 96-well
plate or just 5e5 per well of a 384-well plate. Such low require-
ments are conducive to screening of compound libraries that may
contain thousands of molecules. The challenges of the approach lie
in data collection and analysis, as every well must be imaged and
analyzed to give a quantitative output of platelet adhesion and
morphology. However, automated imaging platforms and bespoke
analysis packages are now common and so capturing and analyzing
thousands of images is no longer the insurmountable challenge
that it once was.
Platelet aggregometry has remained the gold standard in plate-
let function testing since the method was published by Gustav
Born in 1963 [4]. Our plate-based aggregation assay was adapted
from the method published by Lordkipanidze et al. [5] and gives
only an end-point measurement rather than the real-time aggrega-
tion trace generated by Born aggregometry. This means that only
a single value is generated per condition tested giving a simplistic
but easily interpretable read-out of platelet aggregation. Two
advantages of the plate-based aggregation assay are that both the
assay itself and the analysis are extremely quick.
The elevation of cytosolic [Ca2+] is mediated downstream of
almost all platelet activators and so it is also a point of signal inte-
gration. Measurement of cytosolic Ca2+ is a commonly utilized
assay tool within the pharmaceutical industry and so specialized
plate readers with automated liquid handling are similarly well
established and also achieve good results when used to assay plate-
let responses. Our assay utilizes half-area 96-well microtiter plates,
but systems that enable rapid measurements of 384-well plates are
also compatible with screening platelet responses. Ca2+ measure-
ments are performed in real time and are therefore potentially
more time-consuming than other techniques described in this
chapter. However, real time Ca2+ measurements reveal details of
High-throughput Assays 83
signaling dynamics that may be useful and can be used to inform

quantitative systems pharmacology.
High-throughput methodologies are increasingly being
adopted within academia and while few academic labs have access
to robotics and automated systems, facilities have been established
that can give academic researchers access to high-throughput
screening (HTS) platforms, compound libraries and specialists
with expertise in HTS [6]. In addition to compound screening,
high-throughput methodologies facilitate a pharmacological
approach to the study of platelets by enabling concentration-
response relationships to be defined quickly and easily. Estimation
of IC50, EC50 and Emax values from platelet concentration response
data, as well as other more complex pharmacological analyses have
been outlined by Hourani [7]. Any attempts at compound screen-
ing and medicinal chemistry must include such pharmacological
characterization of compounds. Finally, high-throughput method-
ologies facilitate testing of many different combinations of biologi-
cally active compounds, and therefore have many practical
applications in research.
2 Materials
2.1 Buffers 1. Tyrode’s buffer: 137 mM NaCl, 2.7 mM KCl, 1 mM MgCl2,
and Solutions 1.8 mM CaCl2, 0.2 mM Na2HPO4, 12 mM NaHCO3, 5.5 mM
d-glucose, [pH 7.4].
2. Acid citrate dextrose (ACD): 85 mM sodium citrate, 111 mM

glucose, 78 mM citric acid [pH 6.4].
3. Sodium citrate buffer: 10 mM trisodium citrate.
2.2 Reagents 1. Half-area, clear, flat-bottom 96-well microtiter plates (for

and Equipment aggregation assay).
2. Half-area clear, flat-bottom 96-well microtiter plates (for
adhesion assay).
3. Black, flat-bottom, half-area 96-well microplates (for Ca2+
assay).
4. Heated orbital plate shaker (capable of shaking at 1200 rpm).
5. Fluorescence microplate reader.
6. Microplate reader capable of measuring absorption at 405 nm
(other wavelengths within visible spectrum likely to work
equally well).
7. Essentially fatty acid free bovine serum albumin (BSA) solu-
tion (1% w/v in PBS).
8. 10% formalin solution.
9. 3,3′-Dihexyloxacarbocyanine iodide (DiOC6): 1 mg/ml stock

solution in DMSO.
10. Epifluorescence or Confocal microscope capable of imaging
DiOC6 (excitation 482 nm, emission 504 nm) with 20× objec-
tive lens.
11. Fura-2 AM.
3 Method
3.1 Blood Collection 1. Obtain human blood from consenting healthy volunteers who
are free of antiplatelet medication and who have not taken
painkillers (aspirin, ibuprofen, or paracetamol) within 10 days,
via venesection of the antecubital vein. Collect the blood into
3.8% (w/v) sodium citrate.
3.2 Test Compound 1. Prepare test compounds in 96-well “V” or “U”-bottomed

Preparation plates, deep-well plates may be used if volumes are sufficiently
large
2. Test compounds should be diluted in Tyrode’s buffer to
5 × final assay concentration. For example, for initial screening
it may be appropriate to test compounds at 100 μM, and so an
intermediate dilution to 500 μM should be prepared
3. If a concentration range of test compound is to be tested, pre-
pare half-log dilutions at 5 × final concentrations (500, 150,
50, 15, 5 μM, etc.) in Tyrode’s buffer and include a vehicle
control containing an equal concentration of solvent. The vol-
ume test-compound solution prepared should be equal to the
volume required for all experiments plus a “dead volume” of
at least 20%.
4 High-Throughput Adhesion Assay
4.1 Washed Platelet 1. Add acid citrate dextrose (ACD) to a final concentration of
Preparation 12.5% (v/v) and then centrifuge whole blood at 100 × g for
20 min and collect platelet rich plasma (PRP) into falcon
tubes. Centrifuge PRP at 350 × g for 20 min, discard the
supernatant and gently resuspend platelet pellet in Tyrode’s
buffer at a concentration of 2 × 107 platelets/ml
4.2 Adhesion Assay 1. Using a multichannel pipette, coat wells of 96-well, half-area,
Plate Preparation clear-bottom microtiter plate (referred to as “assay plate”)
with 40 μl of adhesive receptor ligand solution of choice for
1 h at 37 °C. Suitable proteins and peptides include collagen
(100 μg/ml), fibrinogen (100 μg/ml), GPVI-specific agonist
cross-linked collagen related peptide (CRP-XL) (10 μg/ml),
and integrin α2β1-specific ligand GFOGER (10 μg/ml).
2. Aspirate ligand solution and wash once with 50 μl Tyrode’s

buffer per well.
3. Pipette 40 μl of 1% BSA solution per well into assay plate and
incubate for 1 h at 37 °C.
4. Aspirate BSA solution and wash three times with Tyrode’s
buffer, and if plate is not going to be used immediately, add
40 μl of Tyrode’s buffer and use a plate sealer or lid to cover
the plate and store at 4 °C for no more than 24 h.
4.2.1 Adhesion Assay 1. Transfer 10 μl of test compound solution to each well of the
assay plate using a multichannel pipette, ensure that at least
one well contains vehicle only, as all other conditions will be
compared to the vehicle-treated sample.
2. Transfer washed platelet suspension to a reservoir and use a
multichannel pipette to add 40 μl to each well of the assay
plate.
3. Cover and incubate for 45 min at 37 °C.
4. Aspirate and discard unbound platelet suspension and wash
each well once with 50 μl Tyrode’s buffer. Avoid aspirating or
dispensing directly onto the center of the well as this can dis-
lodge adhered platelets.
5. Add 40 μl of 10% formalin solution to each well and incubate
for 10 min at room temperature.
6. Aspirate and discard formalin solution and wash each well
once with 50 μl Tyrode’s buffer.
7. Add 40 μl Tyrode’s buffer with 4 μg/ml DiOC6 and incubate
for 20 min at room temperature (there is no need to wash off
the DiOC6).
8. Either image immediately or store at 4 °C for no more than
24 h before imaging.
4.2.2 Imaging 1. Either a microscope system with an automated, motorized

and Analysis stage or a high-content imaging platform is highly recom-
mended for image acquisition.
2. Fluorescence imaging should be performed using a 20× objec-
tive lens (see Note 1)) using excitation at 488 nm and measur-
ing emission at 550 nm.
3. Capture images with a resolution of at least 1024 × 1024 pix-
els so that morphological characteristics of platelets can be
distinguished.
4. Images should be captured in the center of every well and
exported as an image series to simplify analysis.
5. Many different image analysis packages are available and the

exact method to identify, count and measure individual plate-
lets will vary but usually involves:
(a) Creating a “threshold” that identifies platelets on the basis
of fluorescence intensity (see Note 2)) and can be used to
calculate % surface coverage which is the simplest output
from the spreading assay (Fig. 1).
(b) Using the binary image generated by the “threshold” to
perform object or particle measurements, whereby charac-
teristics of platelets (such as number and spread area) are
quantified (Fig. 2a).
Fig. 1 Measurements of platelet surface coverage on fibrinogen. Washed platelet were treated with inhibitors at
a range of concentrations and then incubated at room temperature for 45 min in fibrinogen coated wells of a
96-well plate and then fixed and stained with DiOC6. (a) Representative images of fibrinogen adhered platelets
treated with integrin αIIbβ3 antagonist, integrilin or vehicle. (b) Platelet adhesion quantified in terms of % surface
coverage and plotted against inhibitor concentration and fitted using nonlinear regression. Only integrilin is able
to ablate adhesion of platelets to fibrinogen while cangrelor, dasatinib, and ibrutinib are partial inhibitors
Fig. 2 High content analysis of spreading on different coatings. Washed platelet were treated with ibrutinib at
a range of concentrations and then incubated at room temperature for 45 min in fibrinogen, CRP-XL, or col-
lagen coated wells of a 96-well plate and then fixed and stained with DiOC6. Wells were imaged and then the
area of individual platelets was estimated and used to “bin” platelets into categories corresponding to the
presence of the following morphological characteristics: Lamellipodia, filopodia, or adhered only. (a)
Representative images of platelets fitting into these categories and binary images used to estimate their area.
Numbers of platelets in each category were plotted against [ibrutinib] for platelets adhered to (b) CRP-XL, (c)
collagen, and (d) fibrinogen. High content analysis highlights important differences in the way that ibrutinib
modulates platelet adhesion and spreading on different adhesive surfaces
(c) The open source image analysis software, ImageJ or Fiji

can be used to analyze spreading data and a macro is
included in the notes for this section (see Note 3); how-
ever, users are encouraged to gain familiarity with this
software to ensure accurate results are achieved.
(d) Once measurements of platelet area have been generated,
scoring based on platelet area can be performed (Fig. 2).
4.3 High-Throughput 1. Either:

Aggregation Assay (a) For PRP, centrifuge whole blood at 100 × g for 20 min
4.3.1 Platelet and collect PRP into falcon tubes (also collect 50 μl of
Preparation platelet poor plasma (PPP) for use as a “blank” per assay
plate by centrifuging 60 μl PRP at 10,000 × g for 2 min
and aspirating 50 μl of supernatant).
(b) For washed platelets, add acid citrate dextrose (ACD) to a

final concentration of 12.5% (v/v) and then centrifuge
whole blood at 100 × g for 20 min and collect PRP into
falcon tubes. Centrifuge PRP at 350 × g for 20 min, dis-
card the supernatant and gently resuspend platelet pellet in
Tyrode’s buffer at a concentration of 2 × 108 platelets/ml.
4.3.2 Aggregation Assay 1. Agonists should be prepared at 10× final concentration is

Tyrode’s buffer. For example, if stimulation with 1 μg/ml col-
lagen (see Note 4)) is required, prepare a sufficient volume of
10 μg/ml collagen solution.
2. Using a multichannel pipette, transfer 5 μl of test compound
to each well of a half-area, flat-bottom, clear 96-well microti-
ter plate (referred to as “assay plate”).
3. Decant washed platelets or PRP into a reservoir and transfer
40 μl into each well of the assay plate, avoid creating bubbles
by using reverse pipetting (see Note 5)) and incubate at room
temperature for 5 min.
4. Add 40 μl of Tyrode’s buffer or PPP plus 10 μl of Tyrode’s
buffer to one well of the assay plate to be used as a “blank.”
5. Add 40 μl of washed platelets or PRP to one well and add
10 μl of Tyrode’s buffer, this will be the “resting” or “non-
stimulated” sample.
6. Preheat plate shaker to 37 °C and then securely clip assay plate
onto the plate shaker.
7. Quickly and carefully add 5 μl of 10× agonist into the remain-
ing wells of the assay plate (avoid bubbles), an electronic mul-
tichannel pipette with multidispense function makes this
process quick and easy.
8. Shake plate at 1200 rpm for 5 min.
9. Measure absorption (any wavelength of visible light should
work) using a plate reader within 10 min.
4.3.3 Analysis 1. Convert absorbance to % light transmittance (% LT) using the

following equation:
% LT = 10− Abs × 100
2. Then convert % LT to % aggregation using this equation:

100
%Aggregation = × x − blank
blank − resting
Where “resting” and “blanks” are the % LT value for the
resting and blank samples respectively and x is the % LT value
of a test sample.
Fig. 3 Test compound titrations and log IC50 values. Washed platelets were treated with a range of inhibitor
concentrations (100 μM to 100 nM at half-log intervals). Platelets were stimulated with 1 μg/ml collagen and
shaken for 5 min. Log IC50 values were estimated from nonlinear regression analysis, although concentration
ranges of some compounds could be adjusted in subsequent experiments. Stimulating with a low concentra-
tion of collagen is useful for screening as aggregation under these conditions is highly dependent on tyrosine
kinases such as SFKs and Syk as well as the contribution of receptors activated by secreted agonists such as
ADP. Consequently inhibition caused by dasatinib (SFK inhibitor), entospletinib (Syk inhibitor) and MRS2179
(P2Y1 antagonist) as well as several other inhibitors with diverse functions is easily detected
3. If a concentration range of each test compound was tested,

plot concentration response curves using curve fitting soft-
ware to estimate IC50 and Emax values (Fig. 3).
5 High-Throughput Ca2+ Assay
5.1 Preparation 1. Add acid citrate dextrose (ACD) to a final concentration of

of Fura-2 Loaded 12.5% (v/v) and then centrifuge whole blood at 100 × g for
Washed Platelets 20 min and collect PRP into falcon tubes.
2. Add Fura-2 AM or other Ca2+ dye (see Note 6)) for a final
concentration of 2 μM and incubate at 30 °C for 1 h.
3. Centrifuge PRP at 350 × g for 20 min, discard the supernatant
and gently resuspend platelet pellet in Tyrode’s buffer at a
concentration of 4 × 108 platelets/ml.
4. Store platelets at RT (see Note 7)) and protect from light by
wrapping falcon tube in aluminum foil.
5.2 Ca2+ 1. Prepare agonist and test compound plates as described in the
Measurements previous section, add CaCl2 or EGTA at 10× final concentra-
tion (for a final concentration of 2 mM or 1 mM respectively)
to test compounds if assay is to be performed with or without
the contribution of store operated Ca2+ entry respectively (see
Note 8) or omit these if the assay is to be performed under
“nominally Ca2+ free” conditions (see Note 9).
2. Turn on fluorescence plate reader 1 h in advance of experiment
to ensure it has reached a temperature of 37 °C prior to use.
3. Store assay plates at 37 °C (in plate reader if possible) to enable
them to equilibrate (see Note 10).
4. Load 10 μl of test compound (containing CaCl2 or EGTA)
into as many wells as can be read simultaneously.
5. Add 80 μl of Fura-2 loaded platelets into the same wells.
6. Incubate at 37 °C for 3 min.
7. Load next column/plate with test compounds and incubate at
37 °C (This will allow these samples to reach 37 °C while the
other samples are measured).
8. Use a measurement protocol with the following settings:
(a) Excitation at 340 nm and 380 nm.
(b) Emission at 510 nm.
(c) Measure each well at least once every 4 s (see Note 11).
(d) 10 μl of agonist injected at highest speed (see Note 12)
20 s after measurement begins (Fig. 4).
(e) Measure for a total duration of at least 2 min (see Note 11).
(f) Data can be displayed as a ratio of the fluorescence emis-
sion intensity excited by 380 nm over 340 nm (or [Ca2+]
can be estimated, see Note 13) and further analyzed by
calculating the difference in 380/340 ratio between rest-
ing (prestimulation) and peak signal (Δ380/340).
Fig. 4 Simultaneous Ca2+ measurements stimulated by a range of CRP-XL con-

centrations. Fura-2 loaded washed platelets were stimulated with CRP-XL
(between 10 μg/ml and 0.01 μg/ml at half-log intervals) in a 96-well plate. Agonist
was injected into 8-wells and fluorescence measured simultaneously for 200 s
Fig. 5 Inhibitor combinations to study relative contributions of secreted secondary mediators to CRP-XL evoked
Ca2+ signaling and regulation by Btk, Syk, and PI3K. Fura-2 loaded washed platelets in the presence or absence
of secondary mediator inhibitor cocktail (100 μM MRS2179, 1 μM Cangrelor and 10 μM indomethacin) were
treated with a range of concentrations of (a) btk inhibitor Ibrutinib, (b) Syk inhibitor PRT062607 or (c) PI3K α
inhibitor Alpelisib (between 10 μM and 0.01 μM at half-log intervals) and stimulated with 1 μg/ml CRP-XL in a
96-well plate. Agonist was injected into 8-wells and fluorescence measured simultaneously for 300 s. Ibrutinib
and PRT062607 inhibit Ca2+ signaling with equal potency in the presence or absence of the inhibitor cocktail,
while Alpelisib is a more potent inhibitor of the portion of Ca2+ signaling contributed by secreted secondary
mediators. Log IC50 values are indicated
Compounds can be assayed using different agonists and in

the presence of inhibitors to gain insight into the mecha-
nism of platelet inhibition (Fig. 5).
6 Notes
1. Large images (acquired with a lower magnification objective

lens) enable more accurate measurements by giving larger
sample area but can compromise resolution and must be opti-
mized for the system used for image acquisition.
2. Thresholding must initially be performed manually but then
should be not altered for to ensure consistency
3. Basic FIJI macro for analysis of spreading data:
run(“Set Measurements...”, “area mean area_fraction stack display add redirect=None decimal=3”);
//run(“Threshold...”);
setAutoThreshold(“Huang dark”);
run(“Convert to Mask”, “method=Huang background=Dark”);
run(“Analyze Particles...”, “size=20-200 display clear summarize stack”);
4. Collagen is a versatile agonist in the context of platelet func-

tion screening because both adhesion to collagen and collagen-
evoked aggregation are highly dependent on kinase signaling
as well as signaling evoked by secreted secondary mediators
such as ADP and TxA2. If a screening compound affects any of
these processes, they will cause measurable inhibition collagen-
mediated platelet function.
5. Reverse pipetting, whereby the volume aspirated into the
pipette tip is greater than the volume to be dispensed, is criti-
cal to avoid creation of bubbles in PRP when performing the
aggregation assay. Bubbles cause major artefacts when light
transmittance is measured. To reverse pipette; push the pipette
plunger to the “second stop” and the carefully aspirate liquid
until the plunger is back in the starting position. Transfer liq-
uid to the recipient well by pressing the plunger down to the
“first stop.” The liquid remaining in the tip can be discarded,
returned to the original well or another transfer of the same
liquid can be performed without dispensing the residual liq-
uid. It is important to position the pipette so that it makes
contact with bottom of the well but without blocking the
opening of the pipette tip (see Fig. 6).
6. The Ca2+ assay described in this chapter utilizes the ratiometric
Ca2+ dye, Fura-2; however many nonratiometric Ca2+ dyes that
are excited by a single wavelength, such as Fluo 3 and Fluo 4,
could also be used and require excitation (∼490 nm) and
emission (∼520 nm) that are more likely to be included in
nonspecialist plate readers. One drawback of nonratiometric
dyes is that agonist addition artefacts cannot be compensated
for and can complicate data analysis.
Fig. 6 Pipetting into microtiter plates. When pipetting into microtiter plates (a) avoid forming drops on the tips
of the pipette which can be retained on the pipette tip and (b) touching the pipette tip at a right-angle to the
bottom of the well which may block the opening. (c) For accurate pipetting ensure to angle pipette tips into the
corner of the well to make contact with one of the walls
7. Storage of Fura-2 loaded platelets at RT is preferable to stor-

age at 37 °C as leakage of Ca2+ dyes is temperature dependent,
therefore warm to 37 °C only immediately prior to
measurement.
8. Perform Ca2+ measurements experiments in the presence of
extracellular Ca2+ or EGTA, by combining with test com-
pounds and preincubate with platelets for no longer than
5–10 min to avoid spontaneous platelet activation or depletion
of intracellular Ca2+ stores respectively.
9. It should be noted that unless a Ca2+ chelator is added to assay
buffer, the buffer cannot be truly free of Ca2+ due to contami-
nation with Ca2+-containing salts and is instead referred to as
“nominally Ca2+ free”.
10. Ca2+ measurements are extremely sensitive to temperature so
it is critical to ensure proper equilibration to 37 °C prior to
agonist stimulation in order to achieve reproducible results.
11. Ca2+ release can occur very rapidly, especially when evoked by
G protein coupled receptors, and so care should be taken to
ensure that a reading is taken from each well at least every 4 s.
However, some responses (e.g., to collagen) occur slowly and
total measurement time may need to be increased to capture
the peak of the response.
12. As Ca2+ assays performed in microtiter plates cannot be per-
formed under stirring conditions, mixing of agonists with the
platelet suspension is dependent on the speed of agonist injec-
tion. Agonist injection speed should therefore be set to a high
setting to achieve complete mixing.
13. To quantify [Ca2+] using Fura-2 [8, 9], the following addi-
tional steps must be taken.
(a) Measure the maximum fluorescence ratio by lysing the
cells with 50 μM digitonin to release Fura-2 into the saline
and saturating by adding CaCl2 to a final concentration of
2 mM.
(b) Measure minimum fluorescence ratio in the same lysed
sample by chelating Ca2+ ions with 10 mM EGTA and
10 mM TRIS base to ensure that the pH remains alkaline
for optimum Ca2+ buffering by EGTA.
(c) Measure autofluorescence (‘background’) using platelets
at the same final concentration which had not been loaded
with Fura-2.

(d) Use the following equation to calculate experimental
[Ca2+]i concentrations using the calibration values acquired
as described above:
S R − Rmin
Ca 2+  = K d × f ×
i Sb Rmax − R
Where Kd is the dissociation constant of Fura-2

(224 nM). Sf and Sb are the background-corrected values
of the fluorescence at 380 nm excitation, with zero or
saturating Ca2+ respectively. R is the background-cor-
rected 340/380 nm fluorescence ratio, and Rmin and Rmax
are the ratio limits at zero or saturating Ca2+ respectively,
adjusted using a viscosity constant of 0.85 which corrects
for the effects of the cellular environment upon Fura-2
fluorescence [10].
References
1. Hughes JP et al (2011) Principles of early 6. Karawajczyk A et al (2016) The European lead
drug discovery. Br J Pharmacol 162(6): factory: a blueprint for public-private part-
1239–1249 nerships in early drug discovery. Front Med
2. Dunne A, Jowett M, Rees S (2009) Use of pri- (Lausanne) 3:75
mary human cells in high-throughput screens. 7. Hourani SM (2004) Pharmacological
Methods Mol Biol 565:239–257 approaches to studying platelet function: an
3. Varga-Szabo D, Braun A, Nieswandt B (2009) overview. Methods Mol Biol 273:73–86
Calcium signaling in platelets. J Thromb 8. Ohlmann P et al (2004) Measurement and
Haemost 7(7):1057–1066 manipulation of [Ca2+]i in suspensions of
4. Born GV (1962) Aggregation of blood plate- platelets and cell cultures. Methods Mol Biol
lets by adenosine diphosphate and its reversal. 273:229–250
Nature 194:927–929 9. Grynkiewicz G, Poenie M, Tsien RY (1985) A
5. Lordkipanidze M et al (2014) Characterization new generation of Ca2+ indicators with greatly
of multiple platelet activation pathways in improved fluorescence properties. J Biol Chem
patients with bleeding as a high-throughput 260(6):3440–3450
screening option: use of 96-well Optimul assay.
10. Poenie M (1990) Alteration of intracellular
Blood 123(8):e11–e22 Fura-2 fluorescence by viscosity: a simple cor-
rection. Cell Calcium 11(2–3):85–91
Chapter 6
High-Throughput Signaling Profiling in Blood Platelets

by Multiplexed Phosphoflow Cytometry
Benjamin E. J. Spurgeon and Khalid M. Naseem
Abstract
Multiplexed phosphoflow cytometry is a novel method that provides rapid and quantitative readouts on
intracellular phosphoprotein signaling. In this approach, flow cytometry is combined with fluorescent cell
barcoding (FCB) to facilitate high-throughput analyses of signaling events. After stimulation, fixed and
permeabilized platelets are labeled with distinct dye intensities to create unique fluorescent signatures for
individual samples. These uniquely labeled samples can be combined for simultaneous antibody staining
and acquisition. During software analysis, multiplexed samples can be differentiated by their distinct fluo-
rescence intensities and analyzed as if they had been acquired individually. Multiplexing eliminates inters-
ample variation, increases statistical robustness, and allows 4–96 samples to be processed with no
appreciable increase in antibody consumption or runtime. The method can be performed on washed
platelets, platelet-rich plasma (PRP), and whole blood. Its inherent versatility can fulfil wide-ranging
experimental requirements from simple dose titrations to complex pharmacologic screens.
Key words Flow cytometry, High-throughput, Multiplexing, Phosphorylation, Signaling
1 Introduction
Phosphorylation is a post-translational modification that regulates

protein function and the cellular response to internal and external
stimuli. Phosphorylation events are transient biochemical signals
that are transmitted through complex intracellular networks by pro-
tein kinases and phosphatases. Kinase signaling can be used to pre-
dict cellular behavior, and phosphorylation-based measurements
can provide information about the biochemical processes that regu-
late cellular function in health and disease. Phosphorylation is com-
monly studied by traditional methods that rely on cell lysis and
protein extraction (e.g., immunoassays, immunoblotting, and mass
spectrometry). While these methods have provided excellent
insights into the platelet phosphoproteome [1], they are incompat-
ible with whole blood, require large cell numbers, and lack dynamic
quantification.
95
96 Benjamin E. J. Spurgeon and Khalid M. Naseem
Phosphoflow is an established flow cytometry method that

can overcome traditional limitations and provide quantitative
readouts on phosphorylation-based signaling events in fixed
and permeabilized platelets that have been stained with fluores-
cent antibodies [2]. Phosphoflow is an attractive alternative to
traditional methods, but routine protocols (one sample—one
tube) are ill-suited to large-scale application as they are limited
by low throughput and susceptible to intersample variation.
Fortunately, these limitations can be alleviated by coupling rou-
tine phosphoflow to an innovative multiplexing method called
fluorescent cell barcoding (FCB). Here individual samples (e.g.,
platelets that have been treated with different agonists) are
labeled or “barcoded” with different concentrations of one or
more amine-reactive dyes. These uniquely labeled samples can
be combined into one tube for simultaneous staining and acqui-
sition, thereby increasing sample throughput, eliminating stain-
ing variation, and reducing reagent consumption [3]. Using
appropriate software, multiplexed samples can be differentiated
by their distinct fluorescence intensities and analyzed as if they
had been acquired individually. FCB protocols are inherently
scalable, and dyes can be combined in different ratios to allow
more or fewer samples to be multiplexed. The method can
therefore fulfil different experimental requirements from simple
dose responses (4–6 samples) to complex pharmacologic screens
(96 samples) with no appreciable increase in antibody consump-
tion [2, 3]. As FCB can be performed with standard hardware
and consumables, it provides an accessible and affordable high-
throughput solution that allows researchers to carry out experi-
ments that would otherwise be deemed infeasible due to cost
and time constraints.
We have previously used vasodilator-stimulated phosphopro-
tein (VASP) to show that platelet signaling can be profiled by mul-
tiplexed flow cytometry [2]. This chapter describes the fundamental
protocols that use (a) one dye to barcode four samples, (b) two
dyes to barcode 24 samples, and (c) three dyes to barcode 96 sam-
ples. These protocols are intrinsically versatile, and they can be per-
formed on washed platelets, platelet-rich plasma (PRP), and whole
blood with comparable ease.
2 Materials
1. Acid citrate dextrose: 2.9 mM citric acid, 113.8 mM dextrose,

29.9 mM trisodium citrate, pH 6.4. Store at room
temperature.
Signaling Analysis by Flow Cytometry 97
2. Sodium citrate: 109 mM trisodium citrate, pH 7.4. Store at

room temperature.
3. Citric acid: 300 mM. Store at room temperature.
4. Wash buffer: 36 mM citric acid, 5 mM dextrose, 10 mM
EDTA, 5 mM potassium chloride, 90 mM sodium chloride,
pH 6.5. Store at room temperature.
5. Modified Tyrode’s buffer: 5.6 mM dextrose, 5 mM HEPES,
0.5 mM magnesium chloride, 0.55 mM monosodium phos-
phate, 2.7 mM potassium chloride, 7 mM sodium bicarbonate,
150 mM sodium chloride, pH 7.4. Store at room temperature.
6. Platelets: Phosphoflow is compatible with washed platelets,
PRP, and freshly drawn blood.
7. Paraformaldehyde (PFA) solution: 4–16% methanol-free solu-
tions are available from vendors that provide microscopy con-
sumables. Store at room temperature (see Note 1).
8. Erythrocyte-lysing fixative. We have successfully used BD
Phosflow (Franklin Lakes, NJ) Lyse/Fix Buffer 5X (see Note 2).
9. Phosphate-buffered saline (PBS) : 1.8 mM monopotassium
phosphate, 10 mM monosodium phosphate, 2.7 mM potas-
sium chloride, 137 mM sodium chloride, pH 7.4. Store at
room temperature.
10. Permeabilization buffer: 0.1% Triton X-100 in PBS. Store at
room temperature (see Note 3).
11. Amine-reactive dyes: Alexa Fluor 488 Succinimidyl Ester,
DyLight 800 NHS Ester, and Pacific Blue Succinimidyl Ester.
Dissolve dyes in anhydrous dimethyl sulfoxide (DMSO) at
1 mg/mL, divide into 10–20 μL aliquots, and store protected
from light at −80 °C (see Note 4).
12. Phosphospecific primary antibody: We have successfully used
Phospho-VASP (Ser157) Antibody from Cell Signaling
Technology (Danvers, MA), but the same clone can be
obtained from alternative vendors.
13. Fluorescent secondary antibody: Alexa Fluor 647 goat anti-
rabbit IgG.
14. Single channel and multichannel pipettes.
15. Polypropylene FACS tubes.
16. 96-well polypropylene deepwell plates with round bottoms.
17. Incubator (37 °C with 5% CO2).
18. Benchtop centrifuge with plate rotor.
19. Flow cytometer with 405 nm, 488 nm, and 640 nm lasers, and
450/50, 530/30, 670/14, and 800/60 filters (see Note 5).
3 Methods
This section describes a number of increasingly complex flow

cytometry protocols. We start with routine phosphoflow (one sam-
ple—one tube) and then describe how it can be combined with
FCB to enhance throughput and robustness. In all protocols,
platelets are stimulated with prostaglandins and stained with
Phospho-VASP (Ser157) Antibody, but the method is compatible
with other agonists (e.g., ADP, CRP-XL, and TRAP) and antibod-
ies (see Notes 6 and 7). Each protocol uses distinct dyes that have
been optimized for robust performance, but they can be substi-
tuted with alternate reagents to provide analytical flexibility (see
Note 8).
3.1 Preparation 1. For experiments that require whole blood, the sample should
of Whole Blood be drawn into sodium citrate (1:9). Anticoagulants should be
for Phosphoflow selected with caution as the choice of solution can affect the
Cytometry signaling response (see Note 9).
3.2 Preparation 1. PRP should be obtained from citrated blood by centrifugation

of Platelet Rich at 120 × g for 12 min at room temperature.
Plasma
for Phosphoflow
Cytometry
3.3 Preparation 1. Washed platelets should be prepared from blood that is drawn
of Washed Platelets into acid citrate dextrose (1:5) and centrifuged at 200 × g for
for Phosphoflow 20 min at room temperature.
Cytometry 2. Resultant PRP should be treated with citric acid (1:100) and
centrifuged at 800 × g for 12 min at room temperature.
3. Platelet pellets should be suspended in wash buffer, and centri-
fuged at 800 × g for 12 min at room temperature.
4. Washed platelets should be resuspended in modified Tyrode’s
buffer to a density of 5 × 108 cells/mL (see Note 10).
3.4 Routine 1. Add washed platelets (20 μL), PRP (10 μL), or whole blood
Phosphoflow (15 μL) to an assay plate that contains the desired agonists
(10–20 μL), and incubate for an appropriate time at 37 °C in
the presence of 5% CO2.
2. Fix washed platelets and PRP by adding PFA solution (1.5%
final concentration), mix by pipetting, and incubate at room
temperature for 10 min. Fix whole blood by adding erythrocyte-
lysing fixative (300 μL), mix by pipetting, and incubate at
room temperature for 10 min. Fixed samples can be processed
immediately or stored for later analysis (see Note 11).
3. Centrifuge at 1000 × g for 10 min at 4 °C, and aspirate or

decant the supernatant.
4.
Resuspend platelets in ice-cold permeabilization buffer
(200 μL), and incubate on ice for 10 min.
5. Wash with ice-cold PBS (300 μL), centrifuge at 1000 × g for
10 min at 4 °C, and aspirate or decant the supernatant.
6. Resuspend platelets in ice-cold PBS (100–200 μL). Platelets
can now be stained for routine analysis (see step 7) or mixed
with FCB dyes for multiplexed flow cytometry (see Subheadings
3.5, 3.6, and 3.7).
7. Add Phospho-VASP (Ser157) Antibody (2 μg/mL final con-
centration), mix by pipetting, and incubate on ice for 30 min.
8. Wash (see step 5).
9. Resuspend platelets in PBS (100 μL) that contains Alexa Fluor
647 goat anti-rabbit IgG (1 μg/mL), and incubate on ice for
30 min.
11. Resuspend platelets in ice-cold PBS (200 μL), and analyze by
flow cytometry.
3.5 Dose Response Here we describe how one FCB dye can be incorporated into rou-
with One FCB Dye tine phosphoflow (see Subheading 3.4) to increase throughput and
reduce antibody consumption by fourfold. We describe fixed and
permeabilized platelets that have been stimulated with increasing
concentrations of prostacyclin (PGI2), labeled with Pacific Blue
Succinimidyl Ester, and then combined for simultaneous staining
and acquisition (Fig. 1a). Dye solutions should be prepared freshly
and mixed with platelets that have been suspended in PBS (see
Note 12).
1. Dilute Pacific Blue Succinimidyl Ester stock solution (1 mg/
mL) in DMSO to create dye solutions at 0, 1.6, 8, and 40 μg/
mL (40× final concentration).
2. Add dye solutions (5 μL) to an assay plate so that well A1 con-
tains 0 μg/mL (DMSO only), A2 contains 1.6 μg/mL, A3
contains 8 μg/mL, and A4 contains 40 μg/mL.
3. Add fixed and permeabilized platelets (195 μL) to the assay
plate so that well A1 contains unstimulated platelets, A2 con-
tains PGI2 (1 nM)-stimulated platelets, A3 contains PGI2
(10 nM)-stimulated platelets, and A4 contains PGI2 (100 nM)-
stimulated platelets.
4. Mix thoroughly by pipetting, and incubate on ice for 30 min
(see Note 13).
5. Add ice-cold PBS (300 μL), centrifuge at 1000 × g for 10 min
at 4 °C, and aspirate or decant the supernatant.
Fig. 1 Dose response with one FCB dye. (ai) PGI2 (0–100 nM)-stimulated platelets were fixed and permeabi-
lized, (aii) labeled with Pacific Blue (0–1 μg/mL), and then (aiii, iv) combined for simultaneous staining and
analysis. (bi) Platelets were gated on FSC versus SSC, and (bii) barcodes were gated on FSC versus Pacific
Blue (0–100 nM). (c) Barcodes were exported as individual FCS files and analyzed for Phospho-VASP (Ser157).
Data were presented as color-encoded plots where lighter shades indicate more phosphorylation
6. Repeat step 5 to remove unbound dye.

7. Resuspend well A1 in ice-cold PBS (100 μL) and then move
rightward to resuspend A2, A3, and A4 in the same medium so
that all samples are pooled together in well A4 (100 μL final
volume).
8. Add Phospho-VASP (Ser157) Antibody (2 μg/mL final con-
centration), mix by pipetting, and incubate on ice for 30 min.
30 min.
flow cytometry.
3.5.1 Deconvolution Deconvolution must be performed so that phosphorylation-

and Analysis based data can be obtained for individual samples that have been
multiplexed. Deconvolution describes the process by which bar-
coded populations are separated by their unique dye intensities
(e.g., Pacific Blue) and exported as individual files (Fig. 1bi, ii).
Exported files can be analyzed by conventional methods for anti-
body-related fluorescence (e.g., Alexa Fluor 647), and data can
be presented as histograms or heatmaps where brighter colors

indicate more p hosphorylation (see Note 14) (Fig. 1c).
Deconvolution can be performed with traditional software such
as FlowJo (Ashland, OR) or specialist applications such as
Cytobank (Mountain View, CA) [4].
1. Import FCS files into flow cytometry software, and apply com-
pensation (see Note 15).
2. Gate platelets on forward (FSC) versus side scatter (SSC)
(Fig. 1bi).
3. Display FSC versus Pacific Blue on dot or density plots. Four
populations (barcodes) should be visible with distinct Pacific
Blue intensities. Barcoded populations represent the individual
samples that were multiplexed. Fluorescence intensity corre-
lates with dye concentration so the dimmest population repre-
sents unstimulated platelets and the brightest population
represents PGI2 (100 nM)-stimulated platelets (Fig. 1bii).
4. Gate the barcodes (see Note 16), and name them by their
treatment (e.g., 0, 1, 10, and 100 nM).
5. Export the gated populations as individual FCS files, and ana-
lyze for Phospho-VASP (Ser157)-related fluorescence (Alexa
Fluor 647). Exported files should be opened in new work-
spaces and analyzed as if they had been acquired as individual
samples (Fig. 1c).
3.6 Dose-Time Having shown how one dye can benefit small-scale experiments (see
Response with Two Subheading 3.5), we now demonstrate how an additional dye can
FCB Dyes increase analytical throughput by sixfold. Here we supplemented
the primary dye (Pacific Blue Succinimidyl Ester) with an additional
label (Alexa Fluor 488 Succinimidyl Ester) to perform the afore-
mentioned dose response (PGI2 0–100 nM) at six different time-
points (0–60 min). Dyes were combined in different ratios to create
unique barcodes for 24 individual samples (Fig. 2a). Large-scale
experiments are best performed with multichannel pipettes as many
samples can be processed simultaneously.
1. Dilute Alexa Fluor 488 Succinimidyl Ester stock solution
(1 mg/mL) in DMSO to create dye solutions at 0, 0.74, 2.22,
6.66, 20, and 60 μg/mL (40× final concentration).
2. Add dye solutions (5 μL) to an assay plate so that column 1
contains 0 μg/mL (DMSO only), column 2 contains 0.74 μg/
mL, column 3 contains 2.22 μg/mL, column 4 contains
6.66 μg/mL, column 5 contains 20 μg/mL, and column 6
contains 60 μg/mL (Fig. 2b).
Fig. 2 Sample preparation for dose-time response with two FCB dyes. (ai) Platelets were stimulated with
PGI2 (0-100 nM) at six timepoints (0-60 min) prior to fixation and permeabilization. (aii) Platelets were then
mixed with unique dye combinations and (aiii, iv) pooled for simultaneous staining and analysis. (b) Platelets
were labeled with Alexa Fluor 488 (0-1.5 µg/mL) along timepoints and Pacific Blue (0-1 µg/mL) across doses.
Each well in the 6 × 4 array contained a unique ratio of Alexa Fluor 488 to Pacific Blue to create unique bar-
codes for 24 individual samples
4. Add dye solutions (5 μL) so that wells A1–6 contain 0 μg/mL

(DMSO only), wells B1–6 contain 1.6 μg/mL, wells C1–6
contain 8 μg/mL, and wells D1–6 contain 40 μg/mL (Fig. 2b).
Each well now contains a unique ratio of Alexa Fluor 488 to
Pacific Blue that can be mixed with platelets to create 24
uniquely labeled samples.
5. Add fixed and permeabilized platelets (190 μL) that have
undergone an initial treatment (see Subheading 3.4) so that
wells A1–6 contain unstimulated platelets (0–60 min), wells
B1–6 contain PGI2 (1 nM)-stimulated platelets (0–60 min),
wells C1–6 contain PGI2 (10 nM)-stimulated platelets
(0–60 min), and wells D1–6 contain PGI2 (100 nM)-stimu-
lated platelets (0–60 min).
6. Mix thoroughly by pipetting, and incubate on ice for 30 min.

7. Add ice-cold PBS (300 μL), centrifuge at 1000 × g for 10 min
at 4 °C, and aspirate or decant the supernatant.
8. Repeat step 7.
9. Resuspend column 1 in ice-cold PBS (100 μL per well with an
accurate and precise multichannel pipette), move rightward to
resuspend columns 2–6 in the same medium, and then com-
bine A6–D6 so that all samples are pooled together in one well
(400 μL final volume).
10. Centrifuge (1000 × g for 10 min at 4 °C), and aspirate or
11. Resuspend platelets in PBS (100 μL) that contains Phospho-
VASP (Ser157) Antibody (2 μg/mL), mix by pipetting, and
incubate on ice for 30 min.
30 min.
flow cytometry.
3.6.1 Deconvolution Deconvolution with two dyes is similar to that with one dye (see
and Analysis Subheading 3.5.1) as barcoded populations are gated against
scatter.
2. Gate platelets on FSC versus SSC (Fig. 3a).
3. Display FSC versus Pacific Blue on dot or density plots. Four
barcodes should be visible with distinct Pacific Blue intensities.
Fluorescence intensity correlates with dye concentration so the
dimmest population represents the platelets in row A and the
brightest population represents the platelets in row D (Fig. 3b).
4. Gate the barcodes, and name them by their row (e.g., A, B, C,
and D).
5. Display FSC versus Alexa Fluor 488 for population A. Six bar-
codes should be visible with distinct Alexa Fluor 488 intensi-
ties. Fluorescence intensity correlates with dye concentration
so the dimmest population represents the platelets in well A1
and the brightest population represents the platelets in well A6
(Fig. 3ci).
6. Gate the barcodes, and name them by their well (e.g., A1, A2,
A3, A4, A5, and A6).
Fig. 3 Deconvolution for dose–time response with two FCB dyes. (a) The platelet convolute was gated on FSC
versus SSC, and (b) rows A-D were gated on FSC versus Pacific Blue. (c) Individual samples were deconvoluted
as barcodes (Alexa Fluor 488) were gated on rows A-D. (d) Barcodes were exported as individual FCS files and
analyzed for Phospho-VASP (Ser157). The entire dataset was presented as an all-inclusive heatmap while
discrete dose (row B) and time (column 2) responses were presented as histogram overlays. Red borders
denote the samples in row B while blue borders denote the samples in column 2
7. Repeat steps 5 and 6 for populations B, C, and D (Fig. 3cii,

iii, iv).
8. Export populations A1–D6 as individual FCS files, and analyze
for Phospho-VASP (Ser157)-related fluorescence (Alexa Fluor
647). Exported files should be opened in new workspaces and
analyzed as if they had been acquired as individual samples.
Individual dose-time responses can be presented as histogram
overlays or the entire dataset can be presented as an all-inclu-

sive heatmap (Fig. 3d).
3.7 High-Throughput FCB can be applied to high-throughput screening, and small com-
Screening with Three pound libraries (≤96 candidates) can be combined with three dyes.
FCB Dyes Here we supplemented the aforementioned barcode scheme (see
Subheading 3.6) with an additional dye (DyLight 800 NHS Ester)
to multiplex an entire 96-well plate that contained 24 controls
(rows A and H) and 72 prostaglandins (rows B–G). Prostaglandins
were tested for their ability to induce VASP phosphorylation
(Fig. 4a).
1. Dilute DyLight 800 NHS Ester stock solution (1 mg/mL) in
DMSO to create dye solutions at 0, 1.6, 8, and 40 μg/mL
(40× final concentration).
2. Add dye solutions (5 μL) to an assay plate to delineate quad-
rants that contain 0, 1.6, 8, and 40 μg/mL (Fig. 4bi).
4. Add dye solutions (5 μL) so that rows A and E contain 0 μg/
mL (DMSO only), rows B and F contain 1.6 μg/mL, rows C
and G contain 8 μg/mL, and rows D and H contain 40 μg/
mL (Fig. 4bii).
5. Dilute Alexa Fluor 488 Succinimidyl Ester stock solution
(1 mg/mL) in DMSO to create dye solutions at 0, 0.74, 2.22,
6.66, 20, and 60 μg/mL (40× final concentration).
6. Add dye solutions (5 μL) so that columns 1 and 7 contain
0 μg/mL (DMSO only), columns 2 and 8 contain 0.74 μg/
mL, columns 3 and 9 contain 2.22 μg/mL, columns 4 and 10
contain 6.66 μg/mL, columns 5 and 11 contain 20 μg/mL,
and columns 6 and 12 contain 60 μg/mL (Fig. 4biii). Each
well now contains a unique ratio of dyes that can be mixed
with platelets to create 96 uniquely labeled samples.
7. Add fixed and permeabilized platelets (185 μL) that have been
incubated with controls and prostaglandins, mix thoroughly
by pipetting, and incubate on ice for 30 min.
8. Wash with ice-cold PBS (300 μL), centrifuge at 1000 × g for
10 min at 4 °C, and aspirate or decant the supernatant.
9. Repeat step 8.
10. Resuspend column 1 in ice-cold PBS (100 μL per well with an
accurate and precise multichannel pipette), resuspend columns
2–12 in the same medium, and then combine A12–H12 so
that all samples are pooled together in one well (800 μL final
volume).
Fig. 4 High-throughput screening with three FCB dyes. (ai) Platelets were incubated with buffer-only negative
controls (row A), prostaglandin (1 µM) test compounds (rows B-G), and PGI2 (1 µM) positive controls (row
H) prior to fixation and permeabilization. (aii) Platelets were then mixed with FCB dyes and (aiii, iv) pooled for
simultaneous staining and analysis. Each well in the 12 × 8 array contained unique ratios of DyLight 800,
Pacific Blue, and Alexa Fluor 488 to create unique barcodes for 96 individual samples. (bi) The platelet convo-
lute was gated on scatter (not shown), and quadrants (Q1-4) were gated on DyLight 800. (bii) Rows (Pacific
Blue) were gated on Q1-4, and (biii) barcodes (Alexa Fluor 488) were gated on rows A-H. Barcodes
were exported as individual files and analyzed for Phospho-VASP (Ser157). Data were presented as one com-
pact heatmap to allow the rapid visualization of screening hits (e.g., strong hit E5)
11. Centrifuge (1000 × g for 10 min at 4 °C), and aspirate or

12. Resuspend platelets in PBS (100 μL) that contains Phospho-
VASP (Ser157) Antibody (2 μg/mL), mix by pipetting, and
incubate on ice for 30 min.
30 min.
flow cytometry.
3.7.1 Deconvolution Deconvolution with three dyes is similar to that with two dyes (see
and Analysis Subheading 3.6.1), but barcodes are gated quadrant by quadrant.
Deconvolution can be expedited with advanced gating tools (e.g.,
Cytobank Population Manager) that allow populations to be
defined with one gate set.
2. Gate platelets on FSC versus SSC.
3. Display FSC versus DyLight 800 on dot or density plots. Four
barcodes should be visible with distinct DyLight 800 intensi-
so the dimmest population represents the platelets in the upper
left quadrant and the brightest population represents the plate-
lets in lower right quadrant (Fig. 4bi).
4. Gate the barcodes, and name them by their quadrant (e.g.,
Q1, Q2, Q3, and Q4).
5. Display FSC versus Pacific Blue for Q1. Four barcodes should
be visible with distinct Pacific Blue intensities. Fluorescence
intensity correlates with dye concentration so the dimmest
population represents the platelets in row A and the brightest
population represents the platelets in row D (Fig. 4bii).
6. Gate the barcodes, and name them by their row (e.g., A, B, C,
and D).
7. Display FSC versus Alexa Fluor 488 for population A. Six bar-
codes should be visible with distinct Alexa Fluor 488 intensi-
so the dimmest population represents the platelets in well A1
while the brightest population represents the platelets in well
A6 (Fig. 4biii).
8. Gate the barcodes, and name them by their well (e.g., A1, A2,
A3, A4, A5, and A6).
9. Repeat steps 7 and 8 for rows B, C, and D.

10. Repeat steps 5–8 for Q2, Q3, and Q4.
11. Export populations A1–H12 as individual FCS files, and ana-
lyze for Phospho-VASP (Ser157)-related fluorescence (Alexa
Fluor 647). Exported files should be opened in new work-
spaces and analyzed as if they had been acquired as individual
samples (Fig. 4aiv).
4 Notes
1. Platelets should be fixed with the lowest effective PFA concen-

tration (1.5%) for the shortest effective time (10 min) since
excessive fixation can increase autofluorescence, inhibit anti-
body binding, and decrease signal resolution [2]. PFA solu-
tions should be free from methanol to prevent additional
increases in autofluorescence. Methanol-free solutions should
be used within 1 week to limit formaldehyde degradation.
2. Whole blood should be treated with PFA/diethylene glycol-
based solutions such as BD Phosflow Lyse/Fix Buffer to
simultaneously lyse erythrocytes and fix the remaining cells
(e.g., platelets and leukocytes). Erythrocyte lysis eliminates
scatter overlap and allows platelets to be gated without cell-
type specific antibodies [2].
3. Permeabilization buffers remove lipids from biological mem-
branes and render them permeable to antibodies. Platelets can
be permeabilized with low molecular weight alcohols (e.g.,
ethanol and methanol) or mild detergents (e.g., saponin and
Triton X-100). Alcohol permeabilization is likely to benefit
antibodies that have been raised against linear peptides. While
alcohols can disrupt protein–protein interactions and unmask
target epitopes, they can denature proteins, alter antigen con-
formation, and inhibit reactivity with antibodies that have been
raised against native proteins. Mild detergents do not disrupt
or denature proteins so they are likely to benefit antibodies
that have been raised against native proteins or long peptides
that mimic their conformation. Excellent results can be
achieved with Triton X-100 [2].
4. Succinimidyl (NHS) esters are hydrophobic molecules that
should be dissolved in anhydrous DMSO. Hydrated DMSO
freezes at lower temperatures and hydrolyzes active esters. As
DMSO is hygroscopic and absorbs water over time,
amine-reactive dyes should be aliquoted and stored at −80 °C
for long-term stability.
5. Instruments should be calibrated for optimal performance [5].

FCB samples should be run at low flow rates (≤1000 cells per
second) for best results.
6. Phosphoflow should be performed with highly specific antibod-
ies because nonspecific signals cannot be excluded by protein
standards as they can in Western blotting. Phosphospecificity can
be confirmed by checking for (a) positive signals in response to
agonists that are known to induce phosphorylation, (b) negative
signals in response to agonists that are known not to induce
phosphorylation, and (c) negative signals in response to kinase
inhibition [2]. Phosphospecificity can also be tested in fixed and
permeabilized platelets that have been incubated with phospho-
peptide solutions (e.g., phosphoserine solution should inhibit
Phospho-VASP (Ser157) Antibody reactivity and prevent posi-
tive fluorescence). Phosphoflow is compatible with monoclonal
and polyclonal antibodies, but polyclonals are preferred because
they can recognize multiple epitopes on the same antigen and
negate complications that relate to abundance, conformation,
and accessibility. Direct staining can be performed with fluores-
cent phosphospecific antibodies, but appropriate conjugates are
not always available. We have found that Alexa Fluor 488
Phospho-Zap-70 (Tyr319)/Syk (Tyr352) (Cell Signaling
Technology), Alexa Fluor 647 Anti-PLC-γ2 (pY759) (BD
Phosflow), and BV421 Anti-ERK1/2 (pT202/pY204) (BD
Phosflow) are compatible with phosphoflow (unpublished data),
and we continue to test additional conjugates.
7. Fluorescent secondary antibodies can be used when fluores-
cent primary antibodies are not available or do not provide
adequate signal resolution. Indirect staining can benefit scarce
antigens because multiple secondary antibodies can bind to
each primary antibody and amplify signal intensity. Excellent
resolution can be achieved with bright small-molecule fluoro-
phores such as Alexa Fluor 647 [6]. Large fluorophores such as
APC and PE should be avoided since their passage into the cell
can be hindered by their size.
8. FCB has been optimized with Alexa Fluor 488, Alexa Fluor
647, DyLight 800, and Pacific Blue [2]. Alternate dyes can be
used but should be selected with caution. Best results can be
achieved with spectrally distinct dyes that undergo optimal
excitation and detection. Fluorophores that are illuminated at
suboptimal wavelengths should be used at high concentrations
to ensure adequate resolution. High purity can be achieved
with barcodes that are separated by threefold dilutions (e.g., 0,
0.02, 0.06, 0.18, 0.5 1.5 μg/mL).
9. The choice of anticoagulant should be tailored to the phos-
phoprotein under investigation. While some phosphoproteins
are best resolved in citrated blood, others are better resolved in

heparinized blood [7].
10. Phosphoflow can also be performed on platelets that have been
isolated by gel filtration or prostaglandin treatment, but the
latter should be avoided in experiments that are conducted on
cyclic nucleotide signaling.
11. Fixed (unlabeled) samples can be frozen (−20 °C) or refriger-
ated (4 °C) overnight, but signal resolution will deteriorate due
to prolonged fixation and residual phosphatase activity [2].
12. NHS esters form covalent bonds with deprotonated amines.
Platelets should be labeled in slightly basic solutions (pH 7.2–
9.0) that are free from primary amines such as glycine and tri-
saminomethane as they can quench active esters and inhibit
labeling. PBS is an optimal solution where amines are deprot-
onated and amenable to labeling.
13. Dye solutions are initially immiscible with platelet suspensions
and require thorough mixing to ensure homogeneous distri-
bution and uniform labeling. Insufficient mixing can cause
overlapping populations that cannot be differentiated.
14. Phosphorylation-based data is commonly expressed as the fold
change in median fluorescence relative to an unstimulated con-
trol. Histograms allow phosphorylation to be examined in
detail, and peak distribution can indicate uniform (Gaussian)
or variable (bimodal) signaling responses. Large datasets (e.g.,
dose-time responses) are best interpreted with compact heat-
map visualizations that encode quantitative values with color.
15. Automatic compensation should be performed to ensure accu-
rate and reproducible results. Unstained platelets can be used
as negative controls, and single-stained platelets can be used as
positive controls. Cell-based controls are adequate, but anti-
body-capture beads and amine-reactive beads can also be used.
Bead-based controls preserve sample material and provide
extremely bright signals.
16. Barcoded populations should be displayed on density plots and
gated on their central (dense) regions when adjacent popula-
tions overlap or come into close contact with one another.
Density gating ensures high purity with minimal cell loss [2].
Deconvolution can be performed on dot plots when barcoded
populations are uniformly distributed with few outlier events.
References
1. Dittrich M, Birschmann I, Mietner S et al 2. Spurgeon BEJ, Aburima A, Oberprieler N et al
(2008) Platelet protein interactions: map, sig- (2014) Multiplexed phosphospecific flow
naling components, and phosphorylation cytometry enables large-scale signaling profil-
groundstate. Arterioscler Thromb Vasc Biol ing and drug screening in blood platelets.
28:1326–1331 J Thromb Haemost 12:1733–1743
3. Krutzik PO, Nolan GP (2006) Fluorescent cell matic flow cytometry using a suite of calibration
barcoding in flow cytometry allows high- beads. Nature Protocols 7(12):2067–2079
throughput drug screening and signaling profil- 6. Krutzik PO, Nolan GP (2003) Intracellular
ing. Nat Methods 3:361–368 phosphoprotein staining techniques for flow
4. Kotecha N, Krutzik PO, Irish JM (2010) Web- cytometry: monitoring single cell signaling
based analysis and publication of flow cytome- events. Cytometry A 55:61–70
try experiments. Curr Protoc Cytom. https:// 7. Davies R, Vogelsang P, Jonsson R et al (2016)
doi.org/10.1002/0471142956.cy1017s53 An optimized multiplex flow cytometry proto-
5. Stephen P Perfetto, David Ambrozak, Richard col for the analysis of intracellular signaling in
Nguyen, Pratip K Chattopadhyay, Mario peripheral blood mononuclear cells. J Immunol
Roederer, (2012) Quality assurance for polychro- Methods 436:58–63
Chapter 7
Precise Quantification of Platelet Proteins and Their

Phosphorylation States
Francoise Mazet and Michael J. Fry
Abstract
Systems biology and modeling approaches require quantitative data-rich temporal experiments to better
understand the dynamics and regulation of the components of the signaling pathways that governs cell
biology and physiology. Here we present a modified Western blotting method to rapidly analyze and
accurately quantify protein copy number, and their respective phosphorylation states at specific sites over
detailed time courses.
Key words High-density, Western, Immunoblotting, Protein quantification
1 Introduction
Signaling pathways control cell biology and physiology in response

to external stimuli. Deciphering cell regulation is central to under-
stand and manage diseases, including the development of drugs
which target key molecules. Historically, this type of analysis has
involved looking at changes in the phosphorylation of a particular
protein over a small number of time points to infer its involvement
in a pathway or process. Typically, this was not studied in a quanti-
fied manner. To enable us to understand how the signaling is regu-
lated dynamically in time and space, quantitative data on protein
posttranslational modifications (PTMs) over detailed time courses
are required. The advance of our current understanding, however,
is hampered by technological difficulties to analyze and quantify
proteins and their PTM states in a standardized manner [1]. These
analyses are routinely performed in most laboratories by Western
blotting [2], but these studies are usually only semiquantitative,
and can only analyze a limited number of data points at a time.
Other antibody based techniques such as flow cytometry [3, 4] can
be used for high-density screens, but specificity of protein detec-
tion is not guaranteed as the proteins are not individually resolved
113
114 Francoise Mazet and Michael J. Fry
and visualized. Finally, Mass Spectrometry can provide large

proteomic and phospho-proteomic data sets [5, 6] but requires
specialized equipment and knowledge, the costs to obtain detailed
time courses are high and, in effect, cannot be performed by most
laboratories for high-density screens. Thus the routine analysis of
large numbers of proteins and their PTMs over many time points
that are necessary to study signaling dynamics remains technically
challenging for most laboratories.
In this chapter we describe a modified Western blotting method
that allows simultaneous, rapid and reproducible quantification
and analysis of multiple proteins and their phosphorylation states
at individual phosphorylation sites over detailed time courses. Our
quantification method establishes the relationship between the
amount of protein loaded on a SDS-PAGE gel and the quantitative
signal emitted by fluorescently labeled antibodies that recognize
either the total or the phosphorylated form of the protein. The
simple standardized workflow that we have developed can be per-
formed in most laboratories and produces reproducible results
with sample numbers equivalent to those performed in 384-well
plate assays. We have used this method to study protein phosphor-
ylation in order to produce mathematical models of platelet regula-
tion [7, 8], but this technique may be adapted to study other
PTMs where suitable antibodies are available.
2 Materials
1. Modified Tyrode-Hepes buffer (MTH): 134 mM NaCl,

2.9 mM KCl, 0.34 mM Na2HPO4-12H2O, 12 mM NaHCO3,
1 mM MgCl2, 20 mM Hepes pH 7.3.
2. ACD: 85 mM sodium citrate, 111 mM glucose, 78 mM citric
acid.
3. 1 mM prostacyclin (PGI2) stock solution, in ethanol.
4. 500 U/mL apyrase stock solution, in MTH.
5. Pervanadate solution: 1 mM H2O2, 0.1 mM Na3VO4.
6. 2× RIPA buffer: 100 mM Tris pH 8.0, 300 mM NaCl, 0.2%
(w/v) SDS, 1% (w/v) sodium deoxycholate, 2% (v/v) NP40).
7. 100× Protease inhibitors cocktail: 100 mM PMSF, 1 mM leu-
peptin, 100 μM pepstatin, 100 μM aprotinin.
8. 2× denaturing loading buffer: 4% (w/v) SDS, 10% (v/v)
β-mercaptoethanol, 20% (v/v) glycerol, 0.05 M Tris pH 6.8
(see Note 1).
9. SDS-PAGE gels and electrophoresis buffer (see Note 2).
10. Recombinant protein corresponding to the analyte to be
measured diluted in MTH buffer, 1× denaturing loading
buffer and heat-denatured.
Precise Quantification of Platelet Proteins and Their Phosphorylation States 115
11. Unlabeled phospho-specific antibody for the protein of interest

and a corresponding fluorescently labeled secondary
antibody.
12. Unlabeled antibody recognizing the total (modified and

unmodified) protein of interest and a corresponding fluores-
cently labeled secondary antibody (see Note 3).
13. Protein A/G Sepharose magnetic beads for immuno-

precipitation.
14. Unlabeled isotype control immunoglobulins (IgG), diluted in
MTH and 1× denaturing loading buffer (see Note 4).
15. 0.45 μm Polyvinylidene difluoride (PVDF) membrane, opti-
mized for fluorescence-based immunodetection.
16. 1× Tris Buffered Saline solution: 2 mM Tris pH 7.6, 14 mM
NaCl, 0.1% (v/v) Tween 20 (see Note 5).
17. 5% BSA solution: 5% (w/v) Fraction V BSA molecular grade,
diluted in 1× TBST. Use immediately or freeze in 20–50 mL
batches for later use.
18. Prestained molecular mass marker. Ideally, these markers also
need to be detectable under UV light used for the detection
of the fluorescently labeled secondary antibody.
19. Semidry Western blotting system.
20. Fluorescence imager suitable for use with the chosen fluores-
cently labeled antibodies.
3 Methods
3.1 Preparation The following protocol describes the preparation of platelet sam-
of the Quantification ples and immunoprecipitation of a protein using phospho-specific
Reference Samples antibody. However, these steps are independent of the subsequent
experimental samples, and hence any cell lysate from the same spe-
cies, using optimal conditions and ligands to recover the highest
possible amount of the phosphorylated protein of interest, can
be used.
3.1.1 Preparation 1. Add 45 mg of glucose to 50 mL of MTH in a 50 mL container

of Washed Platelets and warm up the solution in a 30 °C waterbath.
2. Collect 50 mL of blood using a syringe.
3. After collecting, draw 7.5 mL of ACD into the syringe and
mix 3 times by gentle inversion.
4. Dispense up to 8 mL of blood into 15 mL tubes (see Note 6).
5. Centrifuge the tubes for 20 min at 100 × g, 20 °C, brake off.
6. Once the centrifugation is finished, transfer the platelet-rich
plasma (PRP) into a clean 50 mL tube using a plastic Pasteur
pipette, avoiding drawing any red blood cells.
7. Add PGI2 and apyrase to the PRP at final concentration of

0.5 μM and 2 U/mL respectively. Rock the tube gently 3
times.
8. Centrifuge for 10 min at 1400 × g, 20 °C, brake off.
9. Remove the supernatant and dispose into a beaker containing
a chlorine disinfection solution.
10. Resuspend the platelet pellet with 20 mL of MTH-Glucose by
pipetting gently with a plastic Pasteur pipette.
11. Add 4 mL of ACD, 0.5 μM final of PGI2 and 2 U/mL final of
apyrase and mix the platelets gently by inverting the tube a
couple of times.
12. Transfer a small quantity of the resuspended platelets into a
small eppendorf tube for cell counting.
13. Centrifuge for 10 min at 1400 × g, 20 °C, brake off.
14. While centrifuging, perform a platelet count following the cell
counter manufacturer recommendations.
15. Once the centrifuge is finished, remove the supernatant.
16. Immediately resuspend the platelet pellet in MTH-Glucose
buffer at a density of 1 × 109 platelets per mL.
17. Leave the platelets to rest for 30–45 min in a 30 °C water
bath. The platelets can then be used for the next 3 h. Platelet
response decreases after this time and the results may not be
reliable.
3.1.2 Preparation 1. Prepare 50 μL of the pervanadate solution containing the

of the Pervanadate- platelet-activating ligand of choice and mix with 450 μL of
Activated Platelet Lysate washed platelets. Incubate for 5 min at 37 °C.
2. Stop the reaction with 2× RIPA buffer containing 2× protease
inhibitors cocktail.
3.1.3 Immuno- 1. Prewash the Protein A/G magnetic beads by adding 10 μL of
precipitation bead slurry to a clean tube containing 500 μL of 1× RIPA
of the Phosphorylated buffer.
Protein 2. Using a magnetic separation rack, separate the beads from the
buffer and discard the buffer.
3. Take the 500 μL of pervanadate-activated platelet lysate and
add to prewashed Protein A/G magnetic beads.
4. Incubate at 4 °C for 1 h with gentle shaking.
5. Using a magnetic separation rack, separate the beads from the
platelet lysate, transfer the precleared lysate to a clean tube and
discard the magnetic beads pellet.
6. Add 1 μL of the phospho-specific primary antibody of interest
to the precleared platelet lysate.
7. Incubate at 4 °C for 1 h with gentle shaking.

8. Transfer the lysate and antibody solution to the tube contain-
ing 10 μL of washed magnetic beads pellet as described in step
1. Incubate with gentle rocking for 15 minutes at room
temperature.
9. Vortex the stock tube briefly to resuspend the magnetic beads.
10. Place the tubes in the magnetic separation rack and wait a few
seconds for the solution to clear before discarding the
supernatant.
11. Wash the pellet 3 times with 500 μL of 0.1 mM Na3VO4/1×
RIPA buffer.
12. Resuspend the pellet with 100 μL of 2× denaturing loading
buffer.
13. Heat the sample to 85 °C for 5 min (see Note 7) and micro-
centrifuge for 1 min at 10,000 × g to pellet the debris.
14. Transfer the supernatant to a fresh tube. The samples should
be used immediately or stored at −20 °C.
3.2 Quantification The total protein immunoblot allows to estimate the number of
Reference Set: Total molecules present in the immunoprecipitated sample. The primary
Protein Immunoblot phospho-specific and total antibodies used in our methodology
can be raised in different hosts, allowing for the use of different
secondary antibodies as well as different fluorescence emitting
dyes.
1. Prepare two identical dilution series of the immunoprecipi-
tated phosphorylated protein lysate using 50 μL of lysate for
each series. Use MTH and 2× denaturing loading buffer as
diluent solution.
2. As schematized in Fig. 1a, load one of the prepared dilution
series on a SDS-PAGE gel, together with a standard made of a
dilution series of a recombinant protein and an appropriate
dilution series of IgGs raised in the same host species as the
antibody of choice recognizing the “total” protein (i.e., phos-
phorylated and nonphosphorylated) (see Note 8).
3. Run the gel and transfer using standard methodology, as rec-
ommended by the SDS-PAGE and Western blotting apparatus
manufacturer (see Note 9).
4. Remove the membrane from the blotting apparatus and rinse
quickly with 1× TBST. Transfer into a clean plastic container.
5. Block the blot by incubating with 5% BSA for 30 min at room
temperature with constant shaking (see Note 10).
6. Add the primary antibody at the desired concentration and
leave for 1 h at room temperature or overnight at 4 °C, with
constant shaking (see Note 11).
Fig. 1 Schematic diagram of the preparation of the quantification sets. (a) The
first gel is loaded with a dilution series of the immunoprecipitated (IP) phos-
phorylated protein along with a dilution series of the corresponding recombinant
protein (RP) and a dilution series of IgG. The resulting membrane is probed with
an antibody recognizing total (both phosphorylated and nonphosphorylated)
forms of the protein. (b) The recombinant protein dilutions are used to plot a
standard curve of fluorescent signals which is used to determine the total amount
of protein in each IP dilution. (c) A second gel is loaded with an identical IP dilu-
tion series and a dilution series of IgG. The membrane obtained after transfer is
then probed with the phospho-specific antibody used for the IP. This blot allows
the relationship between the known amount of protein in each lane and the fluo-
rescence emitted on this blot to be established. (d) The IgG is necessary for the
normalization of the quantification sets and the experimental samples by allow-
ing for the day to day variability in the blotting and detection system to be account
for in the quantification calculations. It is thus crucial to ensure that the dilution
series signals are linear in every western blot performed
7. Wash 3 times with 1× TBST for 10 min each.

8. Add the secondary antibody, diluted in 5% BSA solution, and
leave for 1 h at room temperature with constant shaking.
9. Wash 3 times with 1× TBST for 10 min each.
10. Scan the resulting blot (Figs. 1a, 3a) with an appropriate
fluorescence imaging system (see Note 12).
11. Using an appropriate imaging software, measure the fluores-
cence intensity of the bands corresponding to the immuno-
precipitated protein and the IgG samples dilution series.
12. Plot the fluorescence values obtained for the IgG samples on a
graph and check for linearity (Fig. 1d). If the dilution series
does not show a linear trend, either the proteins were not fully
transferred or the secondary antibody was not used at a high
enough concentration, and the gel should be repeated.
13. If the IgG values are linear, plot a standard curve of the fluo-
rescence values obtained for the recombinant protein dilution
series.
14. Plot the values obtained for the immunoprecipitated protein
dilutions series to determine their concentration (Xip),
expressed in μg/μL (Fig. 1b).
3.3 Quantification The role of the phospho-specific immunoblot is to detect the level
Reference Set: of protein phosphorylation on a specific site. The comparison with
Phospho-Specific the total protein blot will establish the relationship between the
Immunoblot total amount of protein loaded and the level of fluorescence signal
obtained with the phospho-specific antibody. To allow a proper
comparison, it is imperative to use the same scanning method than
in Subheading 3.2.
1. Load the second dilution series prepared in Subheading 3.2,
step 1, on a second gel, together with a dilution of IgG raised
in the same host than the phospho-specific antibody used for
the immunoprecipitation (Fig. 1c).
2. Transfer the resulting gel as recommended by the SDS-PAGE
and Western blotting apparatus manufacturer. The resulting
membrane is then immunoblotted as described in Subheading
3.2, steps 4–9, using the phospho-specific antibody previously
used for the immunoprecipitation, and a fluorescently labeled
secondary antibody that binds to both the primary antibody
and the IgG standard.
3. The resulting blot (Figs. 1c, 3b) should then be scanned with
an appropriate fluorescence imaging system (see Note 12).
4. Using an appropriate imaging software, measure the fluores-
cence intensity of the bands corresponding to the immunopre-
cipitated protein and the IgG samples dilution series.
5. The values obtained for the protein and IgG dilution series
should then be plotted on graphs to insure that the signal
intensity is linear when plotted against the volume loaded.
6. The intensity of the signal obtained with the phospho-specific
antibody (Yip) is expressed in Arbitrary Fluorescence Units/
μL.
7. Similarly, calculate the intensity of the IgG fluorescence signal
per μL. The value obtained will be necessary to calibrate the
results obtained with further experimental samples.
3.4 High-Density The sections above enable the quantification of the level of signal
Protein Transfer obtained on any further experimental samples without the need of
of Experimental an experiment-by-experiment calibration, as long as the same
Samples approach and reagents are being used, including the detection sys-
tems. The following method describes how to quantify large
amounts of experimental samples by sectioning horizontal strips
Fig. 2 Schematic diagram of the multistrip technique: Multiple SDS-PAGE gels are loaded with experimental
samples. After migration, horizontal strips are cut from the gels and lay on different membranes. After transfer,
each membrane is probed with an antibody targeting a specific protein. IgG dilutions are necessary to normal-
ize the intensity of the signal emitted by the fluorescently labeled secondary antibody and must be present on
each membrane
from the gel prior to the transfer procedure [9]. Using this method,
proteins of different molecular weight can also be analyzed
simultaneously as schematized in Fig. 2.
1. Prepare washed platelets as described in Subheading 3.1.1.
The density of platelets should be between 4 × 108 and 1 × 109
platelets per mL.
2. Activate the platelet samples as desired without using pervana-
date or protein inhibitors.
3. Resuspend the platelet samples in 2× denaturing loading buffer
and heat-inactivate for 3–5 min (see Note 7).
4. Load the samples on SDS-PAGE gels together with a molecu-
lar marker (see Note 13).
5. Load the IgG dilution series on the gels containing the experi-
mental samples or on a separate gel.
6. When the protein migration is finished, prepare the PVDF
membranes and 3 mM paper needed for the transfer as recom-
mended by the manufacturer.
7. Open the first gel cassette and lay the gel still attached to one
plate on the bench or a laminated piece of white paper.
8. Cut a horizontal strip of gel containing your first protein of
interest as shown in Fig. 2, using the molecular weight
markers as a guide (see Note 14). The height of the strips can
be as little as 0.5 cm.
9. Lift each strip gently and transfer onto the piece of membrane,
laying the strip horizontally (see Note 15).
10. If several proteins are analyzed, cut another strip on the gel at
the next desired level and transfer on a different piece of mem-
brane (Fig. 2).
11. Repeat the strip cutting for each gel (see Notes 16 and 17).
12. Cut and transfer a set of IgG dilutions onto each membrane
(Fig. 2).
13. Complete the assembly as recommended by the transfer sys-
tem manufacturer and transfer for 1–2 h (see Note 9).
14. Immunoblot the membrane(s) using the appropriate phospho-
specific and secondary antibodies (see Note 18) and scan the
different pieces of membrane as described in Subheading 3.2,
steps 3–10 (see Note 19, Fig. 3c).
15. A loading control protein (e.g., β-Actin, Tubulin) should be
detected either on the same blot or independently to check for
protein amount variations between the different samples. After
scanning as described in Subheading 3.2, steps 3–10 and
measure the fluorescence intensity as in Subheading 3.3, step
4, any variation should be normalized to allow to compare the
fluorescence levels in the following section.
Fig. 3 Examples of the blots obtained by our laboratory for the quantification of Syk Y348 phosphorylation. (a)
An example blot containing the IP dilutions together with a dilution series of the corresponding recombinant
protein and the IgG and probed with an antibody recognizing the modified and nonmodified protein (ab155187,
Abcam). (b) An example blot with the same dilution series and IgG dilutions probed with the phospho-specific
antibody (ab52212, Abcam). (c) Experimental blot containing sample strips from three different donors and the
IgG dilutions, probed with the phospho-specific antibody
3.5 Quantification 1. Using the same fluorescence imaging software as in Subheading

of the Phosphorylated 3.3, step 4, measure the fluorescence intensity of the bands
Molecules corresponding to the protein analyzed and the IgG samples
in Experimental dilution series.
Samples 2. The IgG samples dilution series fluorescence values should
then be plotted on a graph to insure that the signal intensities
are linear when plotted against the quantity loaded.
3. Calculate the intensity of the IgG fluorescence signal per μL and
compare with the value obtained for the reference set in
Subheading 3.3, step 7. This step allows one to directly compare
independent immunoblots and calibrate the blots containing
experimental samples to the reference set by eliminating techni-
cal variations due to transfer and immunoblotting efficiency.
4. Normalize the fluorescence intensity values obtained for each
experimental sample using the comparison of the IgG signals
from the reference blot and the experimental blot.
The calculated values (“Z” values) for each experimental sample
are finally transformed using the quantification reference values
Xip (Subheading 3.2, step 14) and Yip (Subheading 3.3,
step 6) to calculate the amount of phosphorylated molecules
(“S,” expressed in μg) and applying the following formula:
S = (Z/Yip) ∗ Xip
4 Notes
1. Denaturing loading buffer can be used at concentrations up

to 6×.
2. We use commercial SDS-PAGE gels and buffer to ensure
standardization of the results. Acrylamide percentage should
be selected to allow for maximum separation around the
protein of interest. We recommend to use SDS-PAGE gels
with 15–20 wells for the preparation of the reference dataset.
3. Experimental design considerations: Our method for quantifi-
cation of phosphorylation makes the assumption that there is
only a single phosphorylation site recognized by the phospho-
specific antibody within the given protein of interest. Other
limitations include the availability of recombinant protein and
suitable site-specific antibodies.
4. We noted that commercial IgG batches could be highly vari-
able in their composition and their resulting interaction
with secondary antibodies. To overcome this issue, we rec-
ommend preparing large batches with 2× denaturing load-
ing buffer, aliquoting in small quantities, and keeping them
at −80 °C. We also find that the stability of the signals from
the IgG dilutions are improved by not heat-denaturing

these samples.
5. Add the Tween after adjusting the pH of the solution.
6. Ensure that the end of the syringe is against the wall of the
tube and gently depress the plunger.
7. Phosphorylation is an unstable protein modification, and
intensive heat denaturation should not be attempted. We rou-
tinely denature our sample by heating them at 85 °C for
3–5 min depending on the volume of the sample.
8. The range of the recombinant protein standard has to be tested
and adjusted beforehand, to cover the expected values of the
immunoprecipitate dilution series. Similarly, the IgG dilution
series concentrations have to be within the same fluorescence
detection range for later scanning procedure. In our experi-
ments, the amount of IgG loaded is usually in a range of 1–50 ng.
9. Partial transfer of the proteins would result in inaccurate
results. We recommend to optimize the transfer technique to
ensure a full transfer.
10. Low concentrations of BSA in blocking buffer are not recom-
mended. Alternatively commercial blocking solutions can be
used.
11. It is imperative to saturate the binding of the antibodies to
the phosphorylated sites to obtain accurate results. The
amount of primary and secondary antibodies should thus
always be in-excess. The exact dilution ratio is specific to
each antibody used and should be tested beforehand. We
routinely use dilutions in the range of 1/500–1/1000 for
primary antibodies and 1/1000–1/3000 for secondary anti-
bodies. Higher dilution ratios tend to produce unreliable
results, whilst lower dilution ratios give high background
interfering with the readings.
12. The settings for the detection of all the fluorescent signals on
the gel has to be within linear detection range of the scanner
and show a linear relationship with the volume loaded (Fig. 1d).
13. Load a minimal amount of the marker, usually up to 2 μL
when using 15–20 wells gels. This amount will allow to visual-
ize them by eye on the gel in order to use them as a guide
when cutting the strips, but will not show as a saturated lane
when imaging the immunoblotted membrane.
14. We recommend to use a plastic wedge to cut the strips. The
wedge and gel need to be wet at all time to prevent tearing of
the gel strips. A scalpel can be used instead but tends to stick
to the gel, increasing the risk of tearing the strip.
15. The strips once placed on the membrane should not be allowed
to dry out. To ensure that the strips remain wet, we use a
spraying bottle containing gel running buffer and spray the

strips at regular intervals during the preparation of the
transfer.
16. We recommend to lay all the strips on the membrane in the
same orientation (e.g., from sample 1 to sample10) to simplify
the reading and analysis of the fluorescent signals after
scanning.
17. Theoretically it is possible to cut up to 8–9 strips from the
resolving section of the SDS-PAGE gels. The limitation is
down to the molecular weight of the proteins of interest being
sufficiently separated. This potentially can be improved by
using gels of different percentages or gradient gels.
Alternatively, multiplex detection can be attempted with suf-
ficiently separated emission wavelengths for the secondary
antibodies.
18. It is essential to use the same antibodies than those used for
the phospho-specific reference set. Additionally, it may be
necessary to request specific lot numbers from the antibody
manufacturer.
19. As both the reference data-set and the experimental samples
will be directly compared in the next section, it is imperative to
use identical settings for the fluorescence imager to those used
for the phospho-specific reference blot performed in
Subheading 3.3, step 3.
Acknowledgments
This work was supported by grants from the BHF

(NH/10/2/28425; PG/16/20/32074) and the University of
Reading. We would like to thank Dr. Chris I. Jones, Dr. Sakthivel
Vaiyapuri, and Dr. Neline Kriek for discussions and support during
the development of this work. We also would like to thank the
Undergraduate and MSc students at the University of Reading
who tested, improved, and successfully used this method.
References
1. Solari FA, Dell’Aica M, Sickmann A, Zahedi RP signaling network with phosphospecific flow
(2015) Why phosphoproteomics is still a chal- cytometry. J Immunol 175:2366–2373
lenge. Mol BioSyst 11(6):1487–1493 4. Olive M (2004) Quantitative methods for the
2. Degasperi A, Birtwistle MR, Volinsky N, Rauch analysis of protein phosphorylation in drug devel-
J, Kolch W, Kholodenko BN (2014) Evaluating opment. Expert Rev Proteomics 1(3):327–341
strategies to normalize biological replicates of 5. Burkhart JM, Vaudel M, Gambaryan S, Radau
western blot data. PLoS One 9:e87293 S, Walter U, Martens L, Geiger J, Sickmann A,
3. Krutzik PO, Hale MB, Nolan GP (2005) Zahedi RP (2012) The first comprehensive and
Characterisation of the murine immunological quantitative analysis of human platelet protein
composition allows the comparative analysis of phosphorylation modifications. Sci Rep

structural and functional pathways. Blood 5:16995
120(15):e73–e82 8. Dunster JL, Mazet F, Fry MJ, Gibbins JM,
6. Larance M, Lamond AI (2015) Tindall MJ (2015) Regulation of early steps of
Multidimensional proteomics for cell biology. GPVI signal transduction by phosphatases: a
Nat Rev Mol Cell Biol 16:269–280 systems biology approach. PLoS Comput Biol
7. Mazet F, Dunster JL, Jones CI, Vaiyapuri S, 11(11):e1004589
Tindall MJ, Fry MJ, Gibbins JM (2015) A 9. Kiyatkin A, Aksamitiene E (2009) Multistrip
high-density immunoblotting methodology western blotting to increase quantitative data
for quantification of total protein levels and output. Methods Mol Biol 536:149–161
Chapter 8
The Study of Platelet Receptors Using Artificial Lipid

Bilayers
Michael L. Dustin and Alice Y. Pollitt
Abstract
Artificial lipid bilayers are powerful tools that can be used to model the interactions between platelets and
membrane-bound ligands. To mimic the interaction of platelets with membrane-bound ligands, biotinyl-
ated lipids can be used to couple monobiotinylated recombinant ligands to the upper leaflet of an artificial
lipid bilayer using streptavidin to bridge the two. Artificial lipid bilayers are generated by preparing lipo-
somes, treating glass coverslips to make them hydrophilic and by assembling the bilayer in a specialized
flow chamber. Finally platelets can be added to the flow chamber and the localization of fluorescently
labeled molecules followed using microscopy.
Key words Artificial lipid bilayer, Receptor–ligand interactions, CLEC-2, Podoplanin
1 Introduction
Artificial lipid bilayers, also referred to as planar lipid bilayers, have

made a tremendous impact in the field of immunology to study
the immunological synapse formed by T, B, and natural killer cells
[1–3] and more recently in the study of platelet receptors [4].
This protocol describes the preparation of artificial lipid
bilayers containing dioleoyl phosphatidylcholine and biotinyl-
ated lipids. Commercially available biotinylated lipids enable the
incorporation of monobiotinylated proteins and antibodies to
the upper leaflet of an artificial lipid bilayer using streptavidin to
bridge the two [5]. The lipid dioleoyl phosphatidylcholine is
used to adjust the density of the biotinylated lipids and alone
acts as a negative control. By adjusting the density of biotinyl-
ated lipids the density of proteins incorporated into the upper
leaflet of the bilayer can be carefully controlled to mimic those of
endogenously expressed proteins.
Artificial lipid bilayers can model the interaction of platelets
with membrane-bound ligands. One such interaction studied using
this method is between the platelet receptor CLEC-2 and the
127
128 Michael L. Dustin and Alice Y. Pollitt
platelet ligand podoplanin, a transmembrane protein expressed on

a variety of cell types [6]. Artificial lipid bilayers containing mono-
biotinylated recombinant podoplanin extracellular domain have
enabled the dynamics of platelet–podoplanin interactions to be
visualized [4].
The general principles described here can also be used to
produce artificial lipid bilayers containing other lipids. Suitable
lipids need to have a transition temperature below room tem-
perature to ensure that the lipids remain in their fluid state.
Liposomes containing commercially available nickel chelating
lipids can be used to generate artificial lipid bilayers where histi-
dine tagged proteins can be tethered to the upper leaflet of the
bilayer [5, 7]. The method also opens up the possibility of gen-
erating a procoagulant surface containing phosphatidylserine
lipids to investigate the formation of cell surface coagulation
complexes.
2 Materials
Prepare all solutions using ultrapure water and analytical grade

reagents. Carefully comply with local standard operating proce-
dures when disposing of waste materials.
1. Degassed Tris Buffered Saline (TBS): 25 mM Tris–HCl,
150 mM NaCl, pH 8. Prepare 6 × 1 l in conical flasks.
Deoxygenate the buffer by bubbling through Nitrogen, using
a 70% ethanol treated glass diffuser, for 15 min per flask (see
Note 1). Once degassed, purge with argon, seal the flask with
Parafilm and store at 4 °C. This buffer will be used to dilute
liposomes and to dialyze away detergent.
2. 10% Octyl β-d-glucopyranoside in water. Filter through
0.22 μm filter and keep as frozen stocks.
3. 2% Octyl β-d-glucopyranoside: 20 ml freshly prepared by dilu-
tion of 10% Octyl β-d-glucopyranoside in degassed TBS. Store
on ice until ready to use.
4. Piranha solution: In a fume hood and wearing appropriate per-
sonal protective equipment (see Note 2) prepare in a clean,
100 ml dry Pyrex beaker placed in the center of a Pyrex dish.
The Pyrex dish will contain any spillages should they occur.
Slowly add 15 ml of 30% w/w hydrogen peroxide (H2O2) to
35 ml of 18 M concentrated sulphuric acid (H2SO4) while stir-
ring with a clean dry glass rod.
5. Dialysis tubing preparation: Cut 8–10 cm pieces of dialysis
tubing, Molecular Weight cutoff: 12–14 kDa, 10 mm diame-
ter. Place in a beaker with ultrapure water and sterilize the tub-
ing by microwaving until the water boils.
Artificial Lipid Bilayers 129
6. Phosphate buffered saline (PBS): 137 mM NaCl, 2.7 mM

KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4, adjust to pH 7.4
with HCl.
7. Blocking buffer: 5 mg/ml bovine serum albumin (BSA) in
PBS solution. Weigh 0.2 g of fatty acid-free BSA into a 50 ml
polypropylene tube and make up to 40 ml with PBS. Place into
a boiling water bath for 5 min. Transfer on to ice to cool. Once
cool pass through a 0.22 μm filter and aliquot into 5 ml vol-
umes. Aliquots can be kept frozen and brought up to room
temperature when needed.
8. 10 μg/ml streptavidin diluted in blocking buffer.
9. Imaging chamber: Bioptechs FCS2 chamber with integrated
heating.
3 Methods
3.1 Preparing These protocol notes describe the preparation of bilayers containing
Liposomes 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) and biotinyl-
ated lipids (1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(cap
biotinyl)) (CAP-BIOTIN-PE). Biotinylated lipids can couple mono-
biotinylated proteins (see Note 3) to the upper leaflet of an artificial
lipid bilayer. Other lipids can be prepared similarly (see Note 4).
3.1.1 Evaporation 1. In a fume hood add approximately 2 ml of high grade chloro-
and Lyophilization form to a 16 × 100 mm Borosilicate glass test tube using a
of Chloroform from Lipids sterile glass Pasteur pipette or glass serological pipette (see
Note 5).
2. Lipids are supplied in chloroform in glass ampoules. Break the
cap off the ampoule and transfer lipids, using a glass Pasteur
pipette, into a stock glass vial. Gently purge with argon for
approximately 10 s, to displace any oxygen in the air space,
before sealing the glass vial to remove any oxygen (see Note 6).
Lipids can be stored in glass vials at −20 °C for 6 months.
3. From the stock vial add the volume equivalent to 6.29 mg (see
Note 7) of DOPC to the test tube containing the 2 ml of
chloroform. Layer argon over the remaining lipid in the stock
glass vial.
4. Proceed to evaporation: clamp the test tube in a universal sup-
port clamp. Submerge the bottom 2 cm of the test tube con-
taining the lipid mixture in to a warm, approximately 37 °C,
water bath. Gently stream nitrogen over the chloroform–lipid
mixture avoiding agitation of the liquid surface. The nitrogen
stream can be directed using a glass Pasteur pipette.
5. Stream nitrogen over the mixture until the chloroform evapo-

rates completely (approximately 10 min). This creates a lipid
film on the bottom of the glass test tube.
6. Once evaporated prevent oxygen from entering the test tube
by sealing the opening with Parafilm.
7. Repeat steps 1–4 to prepare a 2% biotinylated DOPC lipid
mixture by adding the equivalent of 6.16 mg of DOPC and
0.177 mg of CAP-BIOTIN-PE.
8. Transfer test tubes to a lyophilizer: Poke holes in the Parafilm
with a hypodermic needle and immediately place inside the
vacuum chamber and close the chamber. Turn on the vacuum
and lyophilize for a minimum of 90 min to remove any residual
chloroform.
9. While the samples are in the vacuum, prepare the 20 ml of 2%
Octyl β-d-glucopyranoside and dialysis tubing. The following
steps are performed in a laminar flow cabinet using aseptic
technique. Use 70% ethanol to clean the flow cabinet work-
space and gloves. Periodically clean gloves with 70% ethanol.
Place dialysis tubing beaker in the flow cabinet. Clean dialysis
closures (small clips) in 70% ethanol and set on tissue to dry in
the flow cabinet.
3.1.2 Solubilization After lyophilization, lipids are solubilized into detergent by sonica-
and Sterilization of Lipid tion and filter-sterilized.
Micelles
1. Remove the test tubes from the lyophilizer. Remove punctured
Parafilm and add 2 ml of 2% octyl β-d-glucopyranoside to each
lipid containing test tube. Once solubilized this will make
4 mM lipid–detergent mixtures. Purge with argon and seal
with Parafilm.
2. Swirl each tube in a bath sonicator to assist solubilization of
the lipids (approximately 1 min). The solution initially becomes
turbid before becoming clear. Once clear, remove and place on
ice and return to laminar flow cabinet.
3. Filter-sterilize the 4 mM lipid–detergent mixtures into poly-
propylene tubes. Purge with argon and immediately seal the
tubes (see Note 8). Add 900 μl of 2% octyl β-d-glucopyranoside
to sterile Eppendorf tubes and add 100 μl of appropriate 4 mM
lipid–detergent mixture to each tube. These will make 0.4 mM
working solutions. Resuspend by pipetting up and down,
being careful not to introduce air bubbles. Purge with argon
(see Note 9)
3.1.3 Removal 1. Transfer 0.4 mM lipid–detergent mix to the dialysis tubing

of Detergent to Form using a wide-tipped sterile disposable Pasteur pipette. Avoid
Liposomes introducing air bubbles during the transfer. Before sealing the
dialysis tubing, sacrifice some lipid–detergent mix to remove

any air.
2. Place the dialysis tubing in 1 l of degassed TBS, add sterile stir-
rer bar. Seal the conical flask with Parafilm and remove any air
by purging with argon. Dialyze overnight on a stirrer plate at
4 °C. Ensure that the stirring bar does not form a vortex,
which might damage the dialysis membrane.
3. Perform two further buffer exchanges by transferring the dial-
ysis tubing to fresh conical flask containing 1 l degassed TBS,
seal with Parafilm and purge with argon.
4. In laminar flow hood, cleaned with 70% EtOH, remove lipo-
somes from the dialysis tubing using a wide tipped sterile dis-
posable Pasteur pipette avoiding air bubbles.
5. Aliquot liposomes into 1.5 ml sterile Eppendorf tubes and
layer each tube with argon. Aliquots can be stored at 4 °C for
3 months (see Note 10).
3.1.4 Determine How Flow cytometry can be used to estimate the number of monobio-
Much Recombinant Protein tinylated recombinant protein molecules incorporated into the
Is Deposited on the Bilayer upper leaflet of the bilayer. The number of molecules incorporated
into the bilayer can be adjusted to mimic the number of molecules
endogenously expressed on the surface of a cell. Adjustments can
be made by diluting the CAP-BIOTIN-PE containing liposomes
with the DOPC liposomes.
1. In a non-tissue culture treated V-bottom 96 well plate add 5 μl
of DOPC liposomes as a negative control or 5 μl of CAP-
BIOTIN-PE containing liposomes.
2. Add 2.5 μl of 5 μm diameter hydrophilic silica bead suspension
to each well. Mix the beads and liposomes together by pulsing
on a vortex mixer.
3. Add 150 μl blocking buffer and centrifuge the plate at 600 × g
for 2 min at room temperature. Remove the supernatant with-
out disturbing the bead pellet. Wash the beads a further two
times with blocking buffer.
4. Add 50 μl of 10 μg/ml streptavidin. Incubate for 20 min at
room temperature with periodic pulsing on a vortex mixer.
5. Add 150 μl blocking buffer and centrifuge plate at 600 × g for
2 min at room temperature. Remove the supernatant without
disturbing the bead pellet. Wash the beads a further 2 times
with blocking buffer.
6. Add 50 μl of 10 μg/ml of fluorescently labeled monobiotinyl-
ated recombinant protein of known fluorescent dye–protein
labeling ratio. Incubate for 20 min at room temperature with
periodic pulsing on a vortex mixer.
7. Add 150 μl blocking buffer and centrifuge plate at 600 × g for

2 min at room temperature. Remove the supernatant without
disturbing the bead pellet. Wash the beads a further 2 times
with blocking buffer.
8. Resuspend beads in 150 μl blocking buffer and transfer con-
tents of each well to flow cytometer tubes. Add a further 0.3 ml
of blocking buffer to each tube and read samples on a flow
cytometer followed by Alexa Fluor 488 standard beads labeled
with known amounts of Alexa Fluor 488.
9. Generate a linear standard curve by plotting the number of
Alexa Fluor 488 molecules on the x axis against the median
fluorescence intensity (MFI) on the y axis.
10. Use the equation of the line defined in step 9 to determine the
number of fluorescent dye molecules associated with the
bilayer coated beads.
11. Use the equation below to calculate the density of monobioti-
nylated molecules (molecules/μm2) tethered to the bilayers
using the fluorescent dye–protein labeling ratio and the surface
area of the silica beads (78.54 μm2 for 5 μm diameter beads).

( Sample MFI ) / ( fluorescent dye : protein ratio )
= density molecules / µm 2
Bead surface area

3.1.5 Preparation 1. A lipid bilayer will spontaneously form when liposomes fuse to
of Glass Coverslips ultraclean glass. Remove any visible debris from 40 mm diam-
for Forming Artificial Lipid eter glass coverslips with a tissue. Clamp at one edge of the
Bilayers coverslip with polypropylene scissor type forceps.
2. Submerge coverslips in freshly prepared piranha solution while
remaining clamped with the forceps, leaving them submerged
for 15–20 min (see Note 11).
3. Carefully remove the forceps from the piranha solution and
rinse the coverslip extensively with ultrapure water.
4. While still holding the coverslip with forceps, dry using a vac-
uum line attached to a clean pipette tip (see Note 12). Keep the
coverslip clamped in the forcep until ready to use.
3.1.6 Forming Artificial To form artificial lipid bilayers we use an FCS2 chamber with inte-
Lipid Bilayers in a Flow grated heating.
Cell
1. The upper half of the chamber, which contains the perfusion
tubes, is placed on a flat surface with the perfusion tubes point-
ing upward. Position the circular gasket with the holes over the
perfusion tubes. Align the holes and position the microaque-
duct slide over the gasket with the fluidic channels facing
Fig. 1 Layout of the 1 μl liposome droplets on the microaqueduct slide. Up to five droplets can be positioned
at one time and can include a mixture of DOPC (negative control) and CAP-BIOTIN-PE containing liposomes
upward (see Note 13). Layer the 0.25 mm gasket with rectan-
gular cutout over the microaqueduct slide.
2. Position up to 5 × 1 μl spots of liposome suspension on the
surface of exposed microaqueduct slide surface (Fig. 1). These
liposome droplets will form small domes.
3. In a single motion, position a dry piranha treated 40 mm cov-
erslip over the gasket with rectangular cutout. A contact will be
made between the top of the droplet and the coverslip.
Immediately clamp the chamber together with the self-locking
base. Mark the position of the bilayers with permanent marker
pen and leave to stand for 20 min at room temperature.
4. Meanwhile prepare the tubing which will permit the addition
of PBS, blocking buffer and components which will be teth-
ered to the bilayer. Place an unprimed piece of tubing with a
two-way stopcock in the open position to the exit perfusion
tube of the chamber. This feeds into a waste beaker.
5. A three-way stopcock enables components to be added to the
microfluidic chamber without the addition of air bubbles (see
Note 14). Prior to connecting the tubing with the three-way
stopcock to the FCS2 chamber prime all the tubing and stop-
cock with PBS to remove any air trapped in the tubing, (the
setup up can be viewed in Fig. 2).
6. Connect the tubing to the chamber and in single motion, gen-
tly but firmly, push through 5 ml of PBS while visually con-
firming no air bubbles pass over the bilayers. Close off the
chamber.
Fig. 2 Layout of the perfusion tubing including the positioning of the two way and three-way stopcocks
7. Block the bilayers with 1 ml of blocking buffer: Take up the

blocking buffer into a 1 ml syringe. Remove any air bubbles by
gently tapping the syringe. Expel some blocking buffer to pro-
duce a convex meniscus. Prime the stopcock port with PBS
and leave a convex meniscus. Attached the syringe containing
blocking buffer by bringing the convex meniscus of the syringe
to the convex meniscus of the stopcock. This method will pre-
vent the introduction of air to the system.
8. Change the stopcock position to close off the 20 ml PBS con-
taining syringe. Inject the blocking buffer into the chamber
(see Note 15). Close off the chamber and incubate for 20 min
at room temperature. When incubating use the two and three
way stopcocks to seal the chamber.
9. Unseal the chamber and gently remove unbound blocking
buffer by passing approximately 5 ml of PBS through the
chamber.
10. Using the same method described in steps 7 and 8 add 10 μg/ml
streptavidin. Incubate for 20 min at room temperature. Remove
unbound streptavidin by passing approximately 5 ml of PBS
through the chamber.
11. Using the same method described in steps 7 and 8 add mono-
biotinylated platelet ligand Incubate for 20 min at room tem-
perature. Remove unbound monobiotinylated ligand by
passing approximately 5 ml of PBS through the chamber.
3.1.7 Imaging Bilayers The flow chamber can be mounted onto an inverted microscope
and the dynamics of fluorescently labeled ligands can be monitored
using total internal reflection fluorescence microscopy and other
microscopy techniques (Fig. 3). Prior to the addition of platelets
into the chamber the mobility of the bilayer must be confirmed.
This is achieved by fluorescence recovery after photobleaching
(FRAP) where an area of fluorescent molecules are bleached and
Fig. 3 Representative images of platelet mediated clustering of artificial lipid bilayer tethered platelet ligands.
A single platelet is clustering a fluorescently labeled monobiotinylated activating antibody tethered to the
upper leaflet of an artificial lipid bilayer. Frame taken every 45 s. Scale bar 2 μm.
Fig. 4 Prior to the addition of platelets into the chamber the lateral mobility of the bilayer must be confirmed.
This is achieved by fluorescence recovery after photobleaching (FRAP). Here, two areas of fluorescent mole-
cules are bleached and the movement of nonbleached molecules into the bleached area followed
the movement of nonbleached molecules into the bleached area

can be followed (Fig. 4).
Once the mobility of the bilayer has been confirmed, platelets,
at 2–3 × 107 platelets/ml can be introduced into the chamber
using the three-way stopcock using the technique described above
to avoid the introduction of air bubbles into the chamber.
4 Notes
1. The buffers are degassed to deoxygenate them. This step pro-

tects the liposomes from oxidation during long-term storage.
2. Wear a full lab coat and chemical-resistant apron, chemical-
resistant gloves. Preparation of piranha solution must occur in a
fume hood. It is expected that this solution will heat up to over
100 °C and begins to produce small bubbles. Measure compo-
nents carefully as making the solution greater than 50% (v/v)
with respect to the 30% (w/w) hydrogen peroxide (H2O2) can
result in a more violent reaction and possible explosion. Always
add the hydrogen peroxide to the sulfuric acid slowly and in this
order. Never mix piranha solution with organic compounds as
this will results in a violent reaction. Consult local standard
operating procedures regarding the disposal of piranha solution
waste. Some institutions permit cooled piranha waste to be dis-
posed into the drains with copious amounts of water.
3. The ability to purchase biotinylated lipids allows for the incor-

poration of biotinylated proteins into the upper leaflet of the
bilayer using streptavidin to bridge the two. It is important
that proteins have a single biotin per molecule. This can be
achieved by biotinylating recombinant proteins chemically
using conditions where the majority of molecules have a single
biotin. Alternatively proteins can be biotinylated in a directed
manner using a BirA tag. If molecules have multiple biotins the
protein will cross-link and aggregate in the bilayer and will
become immobile.
4. The protocol described here employs a dialysis method to pre-
pare liposomes. This method is the gentlest method for mak-
ing proteoliposomes, liposomes in which one or more proteins
have been inserted. Where proteoliposomes are not being
employed alternative methods can also be employed to prepare
liposomes. These include extrusion, sonication, and freeze–
thaw methods [8–10].
5. All work involving Chloroform must occur in a fume hood.
Chloroform is a solvent and therefore the interaction of chlo-
roform and plastics should be avoided.
6. Purging with argon removes any oxygen, thus preventing oxi-
dation of the lipids.
7. Different concentrations of lipids are available. For example if
a 10 mg/ml stock solution of DOPC is purchased 629 μl of
DOPC in chloroform will be needed. To prepare a 2% CAP-
BIOTIN-PE:DOPC lipid mix 616 μl of 10 mg/ml DOPC and
17.7 μl of 10 mg/ml CAP-BIOTIN-PE.
8. Avoid pushing too strongly on the filter as too much pressure
will cause air bubbles to form.
9. When purging oxygen from Eppendorf tubes using argon is it
important that the argon flow is reduced. If the flow is too
high the liquid will be displaced and lost.
10. To test for sterility add a small volume of each lipid to a petri
dish with media and store in incubator at 37 °C. Monitor for
any media contamination over subsequent days.
11. One 100 ml glass beaker can hold a maximum of four forcep
held coverslips. It is advisable to prepare extra coverslips to
mitigate coverslip breakages.
12. There should be no visible residue or particles remaining on
the glass. After drying, coverslips should be used within a day
and protected from dust if exposed to air for any length of
time. Dust may be blown off with a compressed-air duster.
13. Ensure that the microaqueduct slide is well fitted over the per-
fusion tubes. If the holes are not aligned the microaqueduct
slide will crack when the chamber is clamped together.
14. Air bubbles will destroy the bilayer rendering them unusable.
15. When introducing buffers to the chamber ensure that the exit
two-way stopcock is open. When not passing components
through the chamber turn this stopcock to the off position.
References
1. Fooksman DR, Vardhana S, Vasiliver-Shamis Immunol Chapter 18:Unit 18.13. https://doi.

G, Liese J, Blair DA, Waite J, Sacristan C, org/10.1002/0471142735.im1813s76
Victora GD, Zanin-Zhorov A, Dustin ML 6. Watson SP, Herbert JM, Pollitt AY (2010)
(2010) Functional anatomy of T cell activation GPVI and CLEC-2 in hemostasis and vascular
and synapse formation. Annu Rev Immunol integrity. J Thromb Haemost 8(7):1456–1467.
28:79–105. https://doi.org/10.1146/ https://doi.org/10.1111/j.1538-7836.2010
annurev-immunol-030409-101308 .03875.x
2. Groves JT, Dustin ML (2003) Supported pla- 7. Nye JA, Groves JT (2008) Kinetic control of
nar bilayers in studies on immune cell adhesion histidine-tagged protein surface density on
and communication. J Immunol Methods supported lipid bilayers. Langmuir
278(1-2):19–32 24(8):4145–4149. https://doi.
3. Fleire SJ, Batista FD (2009) Studying cell-to- org/10.1021/la703788h
cell interactions: an easy method of tethering 8. Valvo S, Mayya V, Seraia E, Afrose J, Novak-
ligands on artificial membranes. Methods Mol Kotzer H, Ebner D, Dustin ML (2017)
Biol 462:145–154. https://doi. Comprehensive analysis of immunological
org/10.1007/978-1-60327-115-8_9 synapse phenotypes using supported lipid
4. Pollitt AY, Poulter NS, Gitz E, Navarro-Nunez bilayers. Methods Mol Biol 1584:423–441.
L, Wang YJ, Hughes CE, Thomas SG, https://doi.org/10.1007/978-1-4939-
Nieswandt B, Douglas MR, Owen DM, 6881-7_26
Jackson DG, Dustin ML, Watson SP (2014) 9. Schutz GJ, Huppa JB (2017) Forster reso-
Syk and Src family kinases regulate C-type lec- nance energy transfer to study TCR-pMHC
tin receptor 2 (CLEC-2)-mediated clustering interactions in the immunological synapse.
of podoplanin and platelet adhesion to lym- Methods Mol Biol 1584:207–229. https://
phatic endothelial cells. J Biol Chem doi.org/10.1007/978-1-4939-6881-7_14
289(52):35695–35710. https://doi. 10. Su X, Ditlev JA, Rosen MK, Vale RD (2017)
org/10.1074/jbc.M114.584284 Reconstitution of TCR signaling using sup-
5. Dustin ML, Starr T, Varma R, Thomas VK ported lipid bilayers. Methods Mol Biol
(2007) Supported planar bilayers for study of 1584:65–76. https://doi.
the immunological synapse. Curr Protoc org/10.1007/978-1-4939-6881-7_5
Chapter 9
Three-Dimensional Culture in a Methylcellulose-Based

Hydrogel to Study the Impact of Stiffness
on Megakaryocyte Differentiation
Alicia Aguilar, Julie Boscher, Fabien Pertuy, Christian Gachet,
and Catherine Léon
Abstract
The differentiation and maturation of megakaryocytes (MKs) occurs in a 3D environment where the cells
must constantly adapt to the external physical and mechanical constraints during their development and
migration to sinusoid vessels. In this chapter, we present a method for culture of mouse MKs from bone
marrow hematopoietic progenitor cells in a methylcellulose 3D medium with a stiffness mimicking that of
bone marrow. In addition, we describe how the MKs can be recovered to allow for analysis of their differ-
entiation and maturation state by transmission electron microscopy, immunofluorescence or flow cytometry
techniques and to evaluate their ability to form proplatelets. This approach allows (1) generation of MKs
with a morphology that more closely resembles the MKs that differentiate in vivo, (2) recovery of
megakaryocyte phenotypes sometimes observed in vivo but not found in classical liquid cultures, and (3)
study of mechanotransduction pathways induced by the stiffness of the medium.
Key words Megakaryocyte, Proplatelets, Lin− bone marrow progenitors, 3D hydrogel culture,
Methylcellulose, Stiffness
1 Introduction
The importance of substrate stiffness for cell differentiation has

become increasingly recognized since the seminal work of Engler
et al. [1]. Consequently, the mechanisms whereby cells sense the
three-dimensional (3D) topology of their environment is an
emerging field of study [2]. Like other tissues, bone marrow has a
3D structure and despite its being one of the softest tissues of the
organism, its stiffness has been estimated to lie between 15 and
300 Pa [3, 4]. The cells are in close contact with one another and
most of them are migrating to the sinusoid vessels to enter the
blood circulation. This inevitably generates additional external
constraints on adjacent cells, which have to adjust to these forces.
139
140 Alicia Aguilar et al.
It is now known that cells respond to external forces by m odulating

their cytoskeleton and transcriptional program to adapt to their
physical environment. Thus, growing hematopoietic stem and pro-
genitor cells (HSPCs) in traditional liquid cultures would appear
to be a simplistic approach, as is clearly evidenced by the different
ultrastructural morphology of MKs from in vitro cultures as
compared to MKs differentiated in situ (Fig. 1a, b).
To better mimic the physical constraints of the native environ-
ment, we have developed a 3D culture model using a
methylcellulose-based hydrogel. Hydrogels have been employed in
the hematological field for several decades. Derived from cellulose
by replacement of some hydroxyl residues (–OH) by methoxide
groups (–OCH3), methylcellulose forms a nonreactive physical gel
which tends to solidify with increasing temperature, depending on
its concentration and degree of methyl substitution. Such hydro-
gels have been widely used to grow hematopoietic cells for colony
forming assays to quantify numbers of hematopoietic progenitors.
Despite this historical application, the biological significance of the
mechanical impact of hydrogels in cell culture has only recently
been appreciated and their use to improve and better understand
the differentiation of the hematopoietic lineage is still in its infancy.
In this chapter, we describe a method to grow mouse Lin−
HSPCs in a 3D methylcellulose hydrogel in order to improve their
MK differentiation. We have shown that a gel stiffness with a
Young’s modulus in the range of 30–60 Pa is optimal for mega-
karyocyte growth. Compared with standard liquid culture, this
increases polyploidization, improves the development and struc-
turation of the demarcation membrane system (Fig. 1c), and
enhances the capacity of the cells to extend proplatelets once resus-
pended in a liquid medium [5]. In addition, by recreating the
mechanical constraints which cells may encounter in the bone
marrow, we have shown that this culture system can help to reveal
the abnormal phenotype of Myh9 knockout MKs observed in situ
(“leaky” morphology, decreased proplatelet formation) but not in
classical liquid cultures [5–7].
We present methods for the isolation of mouse bone marrow
Lin− cells, their embedding in a methylcellulose hydrogel for 3D
culture and recovery of the MKs for analysis by confocal fluores-
cence imaging, transmission electron microscopy and flow cytom-
etry, and determination of their proplatelet formation capacity.
2 Materials
2.1 General 1. A laminar flow hood.

Equipment 2. A centrifuge (Megafuge 1.0, Heraeus, or equivalent).
3. Transfer pipettes.
Megakaryocyte Differentiation in 3D Methylcellulose-Based Medium 141
Fig. 1 Ultrastructure of murine megakaryocytes. Megakaryocyte from native marrow (upper panels), liquid
culture (middle), and methylcellulose culture (lower panels). Right panels, close up view of the cytoplasm from
the megakaryocyte in the left panels (white squares). This research was originally published in Blood (Aguilar
A, Pertuy F, Eckly A, Strassel C, Collin D, Gachet C, Lanza F, Leon C. Importance of environmental stiffness for
megakaryocyte differentiation and proplatelet formation. Blood. 2016;128:2022-2032. © the American Society
of Hematology)
4. Micropipettes P2, P10, P200, and P1000.

5. A fume hood and protective equipment for fixatives.
6. A CO2 cell culture incubator.
7. A water bath.
2.2 Collection 1. Sterile dissection instruments (forceps, scissors, scalpels).

and Dissociation 2. 70% ethanol in water.
of Bone Marrow
3. DMEM 1× (Gibco, Life Technologies).
4. Stock penicillin, streptomycin, and glutamine (PSG): supplied
as a 100× concentrated 100 mL stock containing 10,000 units/
mL penicillin, 10,000 μg/mL streptomycin and 29.2 mg/mL
glutamine; (Gibco, Life Technologies).
5. Sterile Dulbecco’s phosphate-buffered saline (DPBS, Gibco,
Life Technologies).
6. 5 mL syringes with 21-gauge needles and 50 mL plastic tubes
for flushing of femurs.
7. 23-gauge needles, 25-gauge needles and round bottomed
10 mL plastic tubes for marrow dissociation.
8. An ADAM automated cell counter, slides (AccuChip 2×,
Digital Bio) and AccuStain solutions T and N (Digital Bio).
Alternatively, a Thoma cell counting chamber and trypan blue
to count viable cells.
2.3 Lin− Magnetic 1. Round bottomed 5 mL polystyrene tubes.

Sorting 2. M Medium: DPBS, 2% FBS (Hyclone, GE Healthcare Life
Sciences), 1 mM EDTA.
3. Normal rat serum (Stem Cell Technologies, Vancouver, BC,
Canada).
4. EasySep™ Mouse Hematopoietic Progenitor Cell isolation Kit
(Stem Cell Technologies) with biotinylated antibodies (CD5,
CD11b, CD19, CD45R/B220, Ly6G/C(Gr-1), TER119,
7–4) and streptavidin-coated magnetic beads. Alternatively,
other negative Lin− selection can be used, provided that mega-
karyocyte progenitors are not removed.
5. Easy EasySep™ magnet (Stem Cell Technologies).
2.4 Methylcellulose 1. DMEM 1× (Gibco, Life Technologies).

Encapsulation 2. Penicillin, streptomycin, and glutamine (PSG) 100× stock
(Gibco, Life Technologies).
3. TPO (rhTPO, Stem Cell Technologies): 5 μg/mL stock.
4. r-hirudin (rHV2-Lys47, Transgène, Strasbourg, France):
10,000 U/mL stock.
5. Fetal bovine serum (FBS; Hyclone, GE Healthcare Life Sciences).
6. Complete culture medium: DMEM, 1% PSG, 10% FBS,

50 ng/mL TPO, 100 U/mL hirudin.
7. Concentrated complete culture medium. For each 500 μL final
volume, prepare 167 μL of concentrated complete medium:
102 μL DMEM, 50 μL FBS, 5 μL TPO stock, 5 μL hirudin
stock, and 5 μL 100× PSG stock. For more than one culture
well, increase the volumes proportionally.
8. 3% methylcellulose (MC) stock solution in IMDM medium
(R&D Minneapolis, Minnesota, USA) (see Note 1). Divide
into 1 mL aliquots in 1 mL cryotubes (ClearLine, Biosigma)
and store at −20 °C.
9. 1 mL Luer lock syringes equipped with 18-gauge needles to
collect the methylcellulose.
10. Luer lock syringe connectors (Fisher Scientific).
11. 4-well dishes or alternatively 24-well dishes (ThermoScientific).
2.5 Cell 1. 15 mL plastic tubes.

Resuspension 2. DPBS to dilute the methylcellulose.
3. Complete DMEM culture medium (see Subheading 2.4,
step 6) for cell resuspension in 4-well dishes to visualize
proplatelets.
2.6 Cell Fixation 1. Glutaraldehyde (25% aqueous solution; Electron Microscopy

for Electron Sciences (EMS), USA) stored at +4 °C.
Microscopy 2. Cacodylate buffer: dissolve 21.4 g dimethylarsinic acid sodium
salt trihydrate and 20 g sucrose in 1 L H2O. Under constant
agitation add 1 mL of 1 M CaCl2 and 1 mL of 1 M MgCl2.
The pH is adjusted to 7.3 and the osmolarity to 306 mOsm/L
with approximately equal amounts (13 mL) of 1 M MgCl2 and
1 M CaCl2. The final solution is clear and once filtered (0.2 μm)
is stable at +4 °C for up to several weeks.
3. Glutaraldehyde fixative solution: 5% glutaraldehyde (final con-
centration) in cacodylate buffer. Warning: fixative vapors and
projections are very harmful to the eyes, nose, and throat and
must be handled under a chemical hood, wearing protective
equipment.
4. Agarose type LM-3 low melting point agar (EMS).
5. A heated centrifuge (2–16 KCH, Sigma, France).
6. Razor blades.
2.7 Cell Fixation 1. A cytospin centrifuge (Cytospin 4, ThermoScientific) equipped

and Cytospinning with cytofunnels (Microm Microtech), filter cards, and
for Immunolabeling cytoclips.
2. Paraformaldehyde (EMS) stored protected from the light.
3. Paraformaldehyde fixative solution: 8% paraformaldehyde in

PBS.
4. Polylysine slides (ThermoScientific).
5. Dakopen (Dako) to delimitate the cell area.
2.8 Cell Recovery 1. 28 kPa ESS-coated dishes (Ibidi).

for Flow-Cytometric
Analysis
3 Methods
3.1 Removal Bone dissection must be performed under aseptic conditions to

of Femurs and Tibias, avoid bacterial contamination, and marrow flushing and cell dis-
Extraction of Bone sociation must be carried out under a laminar flow hood.
Marrow, Cell 1. Before starting, add 15 mL of DMEM medium containing 1%
Dissociation, PSG to a 50 mL tube to collect the bones (see Note 2).
and Immunomagnetic 2. Throughout the procedure, frequently disinfect all instru-
Cell Sorting ments in 70% (v/v) ethanol to avoid microbial contamination.
Use cervical dislocation or other approved methods (in line
with local and national ethical guidelines) to euthanize the
mouse and immerse the body in 70% ethanol before removing
the femurs and tibias under aseptic conditions. Rapidly dissect
out the femurs and tibias and remove the adherent tissue.
Gently cut away the epiphyses using a sharp scalpel and briefly
rinse the bones in 70% ethanol before immersing them in
DMEM medium with 1% PSG (see Note 3).
3. Under the laminar flow hood, rinse the bones twice in sterile
DPBS before starting to remove the marrow cells. Holding the
bone with forceps, flush the marrow into a 50 mL tube using
a 5 mL syringe with a 21-gauge needle, filled with DMEM
medium containing 1% PSG. Introduce the extremity of the
needle (just the bevel) into the opening of the bone and press
the plunger with a quick movement and of sufficient force to
flush out the marrow. Use 1 mL per tibia and reuse the medium
for femurs.
4. Distribute the total volume of medium containing flushed
marrow into round bottomed 10 mL tubes. Dissociate the
marrow cells by passing the cell suspension 3 times through a
21-gauge needle, 3 times through a 23-gauge needle and
finally once through a 25-gauge needle. Transfer the suspen-
sion into 15 mL tubes (see Note 4).
5. Measure the cell concentration and viability using an auto-
mated cell counter (see Note 5).
6. Centrifuge the 15 mL tubes for 7 min at 300 × g and discard

the supernatants. Following the instructions of the EasySep™
Mouse Hematopoietic Cell isolation Kit, resuspend the cellular
pellet in M medium to a concentration of 1 × 108 cells/mL. Put
the suspension in a round bottomed 5 mL polystyrene tube, to
a maximum volume of 2 mL.
7. The stem and progenitor cells are isolated by negative immu-
nomagnetic selection. Add blocking normal rat serum (50 μL/
mL) and label nonhematopoietic stem cells and nonprogenitor
cells by adding biotinylated antibodies (CD5, CD11b, CD19,
CD45R/B220, Ly6G/C(Gr-1), TER119, 7–4) (50 μL/mL).
Incubate the cells on ice for 15 min. Add streptavidin-coated
magnetic beads (75 μL/mL) and incubate again on ice for
10 min. If necessary, adjust the volume to 2.5 mL with M
medium. Place the tube without its cap inside the EasySep™
magnet and incubate for 3 min. Keeping the tube inside the
magnet, transfer its contents into a new round-bottomed 5 mL
polystyrene tube. This is done by inverting both the tube and
the magnet with a gentle but regular movement. Discard the
initial tube and place the new tube containing cells, without its
cap, in the magnet for 3 min more. Transfer the isolated Lin−
cells into a new 15 mL tube by gently and regularly inverting
both the tube and the magnet.
8. Measure the Lin− cell concentration and viability.
9. Lin− cells are grown at a final concentration of 2 × 106 viable
cells/mL under both liquid and methylcellulose conditions.
Calculate the required volume of cell suspension to achieve the
final concentration (1 × 106 cells per 500 μL well). Centrifuge
the cells at 300 × g for 7 min. For liquid cultures, discard the
supernatant and resuspend the cell pellet in complete culture
medium. Incubate the cells at 37 °C under 5% CO2 in 4-well
culture plates (500 μL per well).
3.2 Cell Embedding 1. Thaw two 1 mL aliquots of 3% MC stock solution in IMDM
in the Methylcellulose (1 mL for syringe coating and 1 mL for cell encapsulation)
Hydrogel (see Note 6).
2. Draw 1 mL of MC into a 1 mL Luer lock syringe equipped
with an 18-gauge needle to coat the walls of the syringe
(Fig. 2a). Totally expel the MC.
3. Using the same syringe and needle, aspirate the appropriate
volume of MC (333 μL is required per well for a total volume
of 500 μL to achieve a final concentration of 2% MC).
4. Remove the needle, put a Luer lock connector on the end of
the syringe (Fig. 2b) and attach a second, noncoated, 1 mL
Luer lock syringe to the first one (Fig. 2c, d).
Fig. 2 Illustrative guide to the preparation of methylcellulose-based cultures. Refer to Subheading 3.2 for
descriptions
5. Distribute the volume of MC equally between the two syringes

and disconnect them.
6. Prepare the concentrated cell resuspension medium (item 5,
Subheading 2.4), 167 μL per 500 μL in each well. This volume
of medium is calculated so as to obtain a final MC concentra-
tion of 2% (corresponding to a 1.5-fold dilution of the initial
3% MC solution). For more than one culture well, increase the
volumes proportionally.
7. Prepare the cell suspension. Calculate the required volume of

cell suspension to achieve a final concentration of 2 × 106
cells/mL or 1 × 106 cells per 500 μL well. Centrifuge the cells
at 300 × g for 7 min. Discard the supernatant and resuspend
the cell pellet in concentrated complete DMEM medium so as
to obtain an initial cell concentration of 6 × 106 cells/mL (see
Note 7).
8. For each well, very carefully draw 167 μL of cell suspension
into one syringe directly through the connector (Fig. 2e) and
reconnect the two syringes (see Note 8).
9. Homogenize the cells by mixing the MC medium and the cell
suspension with back-and-forth movements between the two
syringes (see Note 9).
10. Draw the cells and medium into one syringe and disconnect
the two syringes, leaving the connector on the empty syringe.
Carefully expel the contents of the other syringe (MC medium
and cells) into one well of a 4-well plate (Fig. 2f).
11. Incubate at 37 °C under 5% CO2.
3.3 Cell Analyses This procedure uses a fixative which must be handled under a fume
hood, wearing protective equipment.
3.3.1 In Situ Cell Fixation 1. Preheat the glutaraldehyde fixative solution (Subheading 2.6,
and Recovery for Electron item 3) to 37 °C so as to minimize fixation artifacts. Fix the
Microscopy cells directly in the hydrogel by gently layering an equal vol-
ume (500 μL in this protocol) of fixative solution on top of
the gel (2.5% glutaraldehyde final concentration), without dis-
rupting the hydrogel (see Note 10). Incubate for 1 h at room
temperature.
2. Using a transfer pipette, transfer the total content of the well
(500 μL hydrogel plus 500 μL fixative) into 10 mL of cacodyl-
ate buffer (see Note 11) in a 15 mL tube. Carefully mix with
several up-and-down pipettings so as to homogeneously dilute
the methylcellulose gel without harming the cells.
3. Centrifuge the mixture at 300 × g for 7 min.
4. Discard the supernatant, add 1 mL of cacodylate buffer and
transfer the cell suspension to a 1.5 mL Eppendorf tube.
Continue with steps 5–12 to prepare for electron microscopy
or keep the cells at 4 °C.
5. Wash the cell pellet 3 times by centrifugation in cacodylate
buffer and incubate the samples in a water bath at 43 °C.
6. Prepare a 2% (w/v) solution of agarose by dissolving agar
powder in boiling cacodylate buffer and let the agar tempera-
ture equilibrate at 43 °C.
7. Using a warm transfer pipette, resuspend the cells in the warm
agarose.
8. Centrifuge the cell suspension in agarose at 16,000 × g and

37 °C for 2 min.
9. Place the tubes on ice to let the agarose solidify.
10. With a razor blade cut the plastic Eppendorf tube above the
cell pellet and then cut the cell pellet into four blocks.
11. Put the blocks in small bottles filled with cacodylate buffer.
12. Proceed to post fixation and inclusion in an Epon resin using
classical procedures. A detailed description of Epon inclusion
and sectioning may be found within a previous volume in this
series (Platelets and Megakaryocytes Volume 3, Chapter 13
“Characterization of Megakaryocyte Development in the
Native Bone Marrow Environment”) [8].
3.3.2 In Situ Cell Fixation This procedure uses a fixative which must be handled under a fume
and Recovery hood, wearing protective equipment.
for Immunolabeling 1. Preheat the paraformaldehyde fixative solution (Subheading
2.7, item 3) to 37 °C. Fix the cells directly in the hydrogel by
gently layering an equal volume of fixative solution (i.e. 500 μL
in this protocol) on top of the gel (4% paraformaldehyde final
concentration), without disrupting the hydrogel (see Note
10). Incubate for 20 min at room temperature.
2. Using a transfer pipette, transfer both the hydrogel and fixative
(1 mL) from each well into 10 mL of DPBS and put it in a
15 mL tube. Carefully mix with several up-and-down pipet-
tings so as to homogeneously dilute the methylcellulose gel
without harming the cells. Centrifuge the mixture at 300 × g
for 7 min.
3. Discard the supernatant and resuspend the cell pellet in 10 mL
of DPBS (see Note 12). Centrifuge the cells again at 300 × g
for 7 min.
4. Discard the supernatant and resuspend the pellet in a volume
of DPBS equal to that of the initial gel (in this protocol 500 μL
per culture well).
5. Prepare the cytofunnels. Attach the cytofunnels and filter cards
with cytoclips to appropriately annotated slides and place them
in the cytospin centrifuge (Fig. 3a–c).
6. Carefully homogenize the cell suspension. Pipette 100 μL of
cell suspension into each cytofunnel, so that 4–5 slides are
obtained for one 500 μL culture well. Cytospin at 500 rpm for
5 min, discard the cytofunnels and recover the slides (Fig. 3d).
Using a marker, immediately draw a line around the cell spot
on the back of each slide and encircle the cells with Dakopen
for further immunolabeling (Fig. 3e), before immersing the
slide in DPBS (see Notes 13 and 14). Proceed to immunola-
beling or keep the slides at 4 °C immersed in DPBS in a hori-
zontal position for up to a few days.
Fig. 3 Illustrative guide to cytospin the cells from cultures for immunolabelling. Refer to Subheading 3.3.2 for
descriptions
3.3.3 Cell Recovery 1. Using a transfer pipette, transfer the hydrogel from each well
for Flow-Cytometric (500 μL) into 10 mL of PBS and place it in a 15 mL tube.
Analyses Carefully mix with several up-and-down pipettings so as to
homogeneously dilute the methylcellulose gel without harming
the cells.
2. Centrifuge the mixture at 300 × g for 7 min. Discard the
supernatant and resuspend the cell pellet in 10 mL of PBS
(see Note 12). Centrifuge the cells at 300 × g for 7 min.
3. Resuspend the cell pellet in an appropriate volume to perform
immunolabeling for flow-cytometric analyses.
3.3.4 Cell Resuspension The physical constraints exerted by the methylcellulose hydrogel
for Proplatelet Analysis tend to inhibit the extension of proplatelets. In order to analyze
proplatelet formation under conditions comparable to those of
liquid cultures, the cells have to be reseeded in a liquid medium.
The cells are resuspended on day 3 of culture, when the control
cells in liquid culture have not yet extended proplatelets; otherwise

centrifugation could alter already extended proplatelets and bias
the final results. Accordingly, be sure to proceed in exactly the
same manner with all samples, including the cells from any control
liquid culture wells (see Subheading 3.3.5), so as to have truly com-
parable controls. The entire procedure is performed under sterile
conditions using a laminar flow hood.
1. Prewarm complete culture medium (Subheading 2.4, item 6)
to 37 °C.
2. Prepare 10 mL of DPBS at 37 °C in a 15 mL tube for each
500 μL well to reseed.
3. Very gently resuspend the cells from each well in the 10 mL of
DPBS. Several up-and-down movements are necessary to
dilute the methylcellulose.
4. Centrifuge the tubes at 300 × g for 5 min.
5. Discard the supernatants and resuspend each pellet in 1 mL of
culture medium (see Note 12).
6. Reseed 500 μL of cell suspension per well of 4- or 24-well
plates and incubate at 37 °C under 5% CO2 (i.e., from one
initial well, obtain 2 wells for proplatelet visualization in
duplicate).
7. On day 4, i.e., 24 h after reseeding, randomly acquire images
from the bottom of each well using bright field microscopy and
the 20× objective (Fig. 4). Usually 10 fields/well are acquired,
Fig. 4 Typical appearance of cells resuspended after culture in methylcellulose for assessment of proplatelet
formation. Megakaryocytes were grown 3 days in MC hydrogel, resuspended and reseeded for 24 h in liquid
medium before pictures are taken to examine for proplatelet formation. See Subheading 3.3.4 for further detail
representing approximately 150–300 megakaryocytes per

condition (two wells). The number of MKs extending
proplatelets is counted on each photograph and the propor-
tion of MKs extending proplatelets is calculated based on these
two wells.
3.3.5 Comparison Methylcellulose cultures are performed in classical 4- or 24-well

of MKs Grown in Liquid culture dishes in a volume of 500 μL (see Note 15). Conventional
and Methylcellulose liquid cultures can also be performed in classical 4- or 24-well
Cultures culture dishes, in a volume of 500 μL, similarly to methylcellu-
lose cultures. However, for some experiments, it may be impor-
tant to recover all cell types and adherent cells may be lost in
classical culture dishes. To limit cell adhesion so that all the cells
are recovered, including adherent remaining stromal cells (see
Note 16), cell culture in liquid medium can be performed in
28 kPa ESS-coated dishes. These dishes accommodate a total
volume of 2 mL.
4 Notes
1. The viscoelastic properties, and hence the stiffness, of methyl-

cellulose vary according to the degree of substitution of
hydroxyl residues by methyl residues, the length of the poly-
mer chains, the concentration of the polymer and the tem-
perature. We characterized the viscoelastic properties of
methylcellulose from R&D and found that at 37 °C and an
optimal concentration of 2%, the stiffness, evaluated by mea-
suring the elastic modulus (Young’s modulus), lies in the
range of 30–60 Pa. These properties may vary according to
the source of methylcellulose and this should be taken into
consideration as small variations may affect the final stiffness
and hence the differentiation process.
2. It is better to warm the medium to 37 °C before the
experiment.
3. You do not need to cut all the epiphyses. To facilitate marrow
flushing, cut only the epiphyses of the ankle side end for the
tibia and of the hip side end for the femur. This will allow the
needle to be retained by the epiphysis of the knee side end of
each bone and will decrease the risk of ejecting the bone from
the needle during flushing.
4. Place the end of the needle on the tube wall while dissociating
the flushed marrow to avoid the formation of air bubbles and
thereby the creation of shear stress.
5. An ADAM automated cell counter is used with AccuChips and
AccuStain solutions T and N to count viable cells. Alternatively,
another automated cell counter may be used, or a Thoma cell
chamber for manual counting in the presence of trypan blue to

exclude dead cells.
6. Use two different aliquots of methylcellulose, one to coat and
one to fill the syringe, because the MC used for coating will
contain undesirable bubbles after its ejection.
7. Be sure to discard all the supernatant because the volume of
liquid added to the methylcellulose changes its concentration.
To obtain an optimal concentration of 2% starting from an
initial concentration of 3% MC, the total volume of liquid is
167 μL for 333 μL of MC medium. This allows one to prepare
500 μL of MC medium at a concentration of 2%, containing
cells at a final concentration of 2 × 106 cells/mL or 1 × 106
cells/500 μL well.
8. Make sure you do not introduce air bubbles into the syringe
when adding the cell suspension. Carefully reconnect the
syringes without losing any MC medium or cell suspension
in the screw thread.
9. Mix back and forth about ten times to homogenize. Do not
mix too vigorously as this will create undesirable small air
bubbles.
10. Do not disturb the hydrogel so as to maintain its stiffness
intact until the cells are fixed. The fixative diffuses very rapidly
throughout the gel, as revealed by its rapid change in color.
11. From this moment on, it is important to use cacodylate buffer
rather than PBS to avoid adding phosphates, which are unde-
sirable for further processing for electron microscopy.
12. Note that some methylcellulose may not be fully diluted or
may remain on the bottom wall of the tube and be pelleted
during centrifugation. In this case, discard the liquid above
the cell and MC pellet, leaving about 2 mL so as not to lose
cells. Add a further 10 mL of buffer to resuspend the cells and
dilute and eliminate all the MC after a second round of
centrifugation.
13. DPBS is added to prevent dehydration and maintain the cell
structure.
14. Dispose of the cytofunnels and filter cards in sealed disposable
bags to avoid the dispersion of fixative vapors.
15. In classical dishes and liquid cultures, adherent cells stay

attached to the plastic bottom after their sedimentation,
whereas in methylcellulose cultures all the cells are recovered
because they remain in suspension. To recover all the cells
from the control liquid cultures, in order to obtain compara-
ble numbers of cells from the MC and liquid cultures for
flow-cytometric analyses, we use ESS-coated dishes for the
control cultures to prevent adherence of the cells.
16. In ESS-coated dishes, the cells are seeded at 2 × 106 cells/mL

as in classical liquid cultures, but the seeding volume is 2 mL
because the surface of the dishes is larger than that of the wells
of 4-well plates.
Acknowledgments
The authors wish to thank Juliette Mulvihill for reviewing the

language part of the manuscript. This work was supported by
INSERM, EFS Grand-Est and the Société Française d’Hématologie.
Disclosure of conflict of interests: The authors have no conflict of
interests.
References
1. Engler AJ, Sen S, Sweeney HL, Discher DE 5. Aguilar A, Pertuy F, Eckly A, Strassel C, Collin
(2006) Matrix elasticity directs stem cell lineage D, Gachet C, Lanza F, Leon C (2016)
specification. Cell 126(4):677–689. https:// Importance of environmental stiffness for mega-
doi.org/10.1016/j.cell.2006.06.044 karyocyte differentiation and proplatelet forma-
2. Jansen KA, Donato DM, Balcioglu HE, tion. Blood 128(16):2022–2032. https://doi.
Schmidt T, Danen EH, Koenderink GH (2015) org/10.1182/blood-2016-02-699959
A guide to mechanobiology: where biology and 6. Eckly A, Rinckel JY, Laeuffer P, Cazenave JP,
physics meet. Biochim Biophys Acta 1853(11 Lanza F, Gachet C, Leon C (2010) Proplatelet
Pt B):3043–3052. https://doi.org/10.1016/j. formation deficit and megakaryocyte death
bbamcr.2015.05.007 contribute to thrombocytopenia in Myh9
3. Choi JS, Harley BA (2012) The combined knockout mice. J Thromb Haemost
influence of substrate elasticity and ligand 8(10):2243–2251. https://doi.
density on the viability and biophysical prop- org/10.1111/j.1538-7836.2010.04009.x
erties of hematopoietic stem and progenitor 7. Eckly A, Strassel C, Freund M, Cazenave JP,
cells. Biomaterials 33(18):4460–4468. Lanza F, Gachet C, Leon C (2009) Abnormal
https://doi.org/10.1016/j.biomaterials. megakaryocyte morphology and proplatelet
2012.03.010 formation in mice with megakaryocyte-
4. Shin JW, Buxboim A, Spinler KR, Swift J, restricted MYH9 inactivation. Blood
Christian DA, Hunter CA, Leon C, Gachet C, 113(14):3182–3189
Dingal PC, Ivanovska IL, Rehfeldt F, Chasis 8. Eckly A, Strassel C, Cazenave JP, Lanza F, Léon
JA, Discher DE (2014) Contractile forces sus- C, Gachet C (2012) Characterization of mega-
tain and polarize hematopoiesis from stem and karyocyte development in the native bone mar-
progenitor cells. Cell Stem Cell 14(1):81–93. row environment. Methods Mol Biol
https://doi.org/10.1016/j.stem.2013.10.009 788:175–192
Chapter 10
Differentiation of Human Pluripotent Stem Cells

to Megakaryocytes by Transcription Factor-Driven
Forward Programming
Thomas Moreau, Amanda L. Evans, and Cedric J. G. Ghevaert
Abstract
The differentiation of megakaryocytes from human pluripotent stem cells in vitro offers intriguing new
perspectives for research and transfusion medicine. However, applications have been hampered by the low
efficiency of cytokine driven differentiation protocols leading to poor megakaryocyte purity and yield.
Here we describe a novel forward programming approach relying on the combined ectopic expression of
the three transcription factors GATA1, FLI1, and TAL1 in human pluripotent stem cells for large scale
production of mature megakaryocytes using chemically defined culture and minimum cytokines.
Key words Human pluripotent stem cells, Megakaryocytes, Transcription factors, Forward program-
ming, Transfusion medicine
1 Introduction
Human pluripotent stem cells (hPSCs) in vitro are akin to early
postimplantation embryonic epiblast cells and endowed with a
broad differentiation potential toward the three germ layers—
ectoderm, mesoderm, and endoderm—ultimately contributing to
the formation of all organs and tissues [1]. Originally derived from
the embryo intracellular mass (ICM) as embryonic stem cells
(hESCs) the ease of use and range of applications for hPSCs has
significantly increased since the discovery of somatic cell repro-
gramming to induced pluripotent stem cells (hiPSCs) by the use of
ectopic transcription factors (TFs) [2]. Due to their pluripotency
and capacity to self-renew in vitro, their potential for basic research,
disease modeling and therapy is immense and has triggered a wide
interest in the scientific and medical communities [3].
Electronic supplementary material: The online version of this chapter (doi:10.1007/978-1-4939-8585-2_10)

contains supplementary material, which is available to authorized users.
155
156 Thomas Moreau et al.
The differentiation of megakaryocytes (MKs) from hESCs was

first described in 2006, rapidly followed by further publications tak-
ing advantage of this new in vitro model to study various aspects of
MK biology [4–6]. This first wave of differentiation protocols used
a very much undefined directed differentiation approach supported
by mouse feeder cells (OP9, C3H10T1/2), animal serum, and
hematopoietic cytokines including thrombopoietin (TPO). With
increasing knowledge of factors controlling hPSC sequential dif-
ferentiation toward the blood lineage, directed differentiation
methods have been gradually refined to chemically defined proto-
cols using complex cytokine cocktails allowing efficient generation
of MKs [7, 8]. However, as testified by many researchers in the
field, the purity and cell yield of the MK culture remains an issue
when using these approaches, limiting the breadth of applications.
The directed differentiation approaches discussed above rely on
mimicking the embryonic developmental pathways in vitro by pro-
viding environmental cues (stromal cells, cytokines), leading to the
sequential differentiation of hPSCs toward the mesoderm lineage,
hemogenic endothelium, blood progenitors, and eventually MKs.
Alternatively, we have developed a method using TFs as primary cell
intrinsic drivers of MK differentiation [9]. This approach, akin to
the hiPSC reprogramming strategy but aiming to differentiate
hPSCs, is called forward programming. We have demonstrated that
the combination of three TFs—GATA1, FLI1, and TAL1—when
ectopically expressed in hPSCs leads to their differentiation to
mature MKs with unprecedented efficiency in a low cytokine,
chemically defined medium. This highly scalable MK forward pro-
gramming (MKFOP) protocol allows the production in vitro of
large quantities of pure mature MKs from hPSCs opening new ave-
nues for basic research on MK and platelet biology, disease model-
ing and for the clinical application to transfusion medicine [10].
We describe below the MKFOP method for production of
mature MKs from hPSCs in a typical 21-day long culture. While
we have developed variations in the initial cell seeding (see
Subheadings 3.1, 3.8, and 3.9) to adapt to specific needs of rare
hPSC lines, the method relies primarily on the transduction of
hPSCs with three lentiviral vectors expressing separately, GATA1,
FLI1, and TAL1 (see Note 1). The cells are cultivated for 48 h in
a mesoderm inducing medium followed by a MK supportive cul-
ture from day 2 onward with a midculture split on day 9 followed
by further suspension culture up to day 21 (Fig. 1). By this stage,
most hPSC lines are expected to give rise to >95% CD41+ (Integrin-
alpha2b) pure MK culture with levels of maturity ranging 30–90%
as assessed by CD42a/b expression (GP9/GP1BA). The genuine
MK phenotype obtained by MKFOP has been validated by
morphological features (size, polyploidy, cytoplasmic ultrastruc-
ture), whole genome expression arrays and functional assays (plate-
let production) [9]. The method has been applied to an array of
hPSC lines (including hiPSCs and hESCs) showing a range of MK
Megakaryocyte Forward Programming 157
Fig. 1 MKFOP protocol outline. The sequential steps of the MK forward programming protocol are outlined with
reference to the corresponding section in the main text. Morphology (scale bar 200μm). The gradual change
of cell morphology over time is illustrated by representative microscopy pictures with the programming days
indicated in grey circles. Patches of round bright cells will form from the adherent cell layer becoming clearly
visible by day 5. These will progressively detach and form berry-like cell clusters in suspension made of grow-
ing MK progenitors. These clusters are loosely cohesive and are easily dispersed to single-cells by gentle
pipetting. After day 9 split, the suspension culture will have a mixture of such cell-clusters and single cells
representing the proliferating MK progenitor and mature MK fractions respectively. Depending on the effi-
ciency of the programming, the culture may include a significant amount of cell debris after the day 9 split
which will be eventually overtaken by the growing MK population. Phenotype. Representative flow cytometry
dot plots are shown illustrating the four subpopulations of cells identified by the CD41/CD42/CD235 triple
staining through the course of MKFOP (MK progenitors, MKs, mature MKs, and contaminating erythroblasts);
corresponding programming days are indicated in grey circles
yield which correlates with the known variability of differentiation

potential in hPSC lines [11, 12]. Our experience on more than 20
hPSC lines showed 55% differentiate efficiently (i.e., >90% CD41+
MK purity and > one fold increase by day 21), amongst them 55%
giving rise to high MK yield (>tenfold increase). A fraction of the
hPSC lines (30% of total) will progress toward long-term prolifera-
tive MK cultures (proliferating for 60–120 days with >60% CD42+)
while the remaining lines will usually plateau around day 30–40
before the end of the culture (see Note 2). The methods in this
chapter are supported by an online video.
2 Materials
2.1 Human Both hiPSC lines derived from dermal fibroblasts using the four
Pluripotent Stem Cell Yamanaka’s factors (OCT4, SOX2, KLF4, MYC) introduced by a
Lines variety of means (integrative retroviral vectors, episomal plasmid vec-
tors and nonintegrative Sendai vectors) and hESCs have been success-
fully used for MKFOP. The hPSC lines can be maintained using a
range of pluripotency conditions before MKFOP, including different
matrices (e.g., feeder cells, gelatine, vitronectin, laminin, Matrigel),

culture media (e.g., KOSR 20%, mTeSR1, Essential 8, Essential 6)
and passaging techniques (enzymatic or nonenzymatic). The hPSC
culture conditions used prior to MKFOP should be the routine one
achieving the best pluripotency culture for any given cell line.
2.2 Culture Media All culture media are kept at 4 °C, brought to room temperature
and Reagents before use and have a shelf-life of a month after opening. The cyto-
kine stock solutions are kept as frozen aliquots at −20 °C (BMP4,
FGF2, SCF) or −80 °C (TPO) and for up to 5 days at 4 °C after
thawing (except TPO which is maintained at −20 °C after the first
thaw); the cytokines are never freeze-thawed more than twice.
1. AE6 medium: DMEM/F12 (with 2.5 mM L-glutamine,
15 mM HEPES, 3.2 g/L d-glucose, 0.02 mM Phenol Red)
supplemented with 0.054% NaHCO3, 64 mg/L l-ascorbic
acid, 20 mg/L insulin, 11 mg/L transferrin, and 0.0134 mg/L
selenium (see Note 3).
2. Pluripotency medium: use the formulation best suiting the
maintenance of the hPSC line to be forward programmed (see
Subheading 2.1).
3. CellGro-SCGM medium: specialized serum-free medium sold
by CellGenix (Cat. 2001) (see Note 4).
4. Vitronectin Recombinant Human Protein: (truncated frag-
ment (VTN-N) as described in Chen et al. [13]) 5 μg/mL in
DPBS.
5. Recombinant cytokines: human FGF2 (5 μg/mL in DPBS
0.1% BSA); human BMP4 (10 μg/mL in ultrapure water 4 mM
HCl 0.1% BSA); human TPO (10 μg/mL in CellGro-SCGM
0.1% BSA); human SCF (50 μg/mL in DPBS 0.1% BSA).
6. Protamine sulfate: 10 mg/mL in ultrapure water.
7. Rock inhibitor Y-27632: 10 mM in ultrapure water.
8. TrypLE Select: recombinant cell-dissociation enzymes that
replace porcine trypsin (sold by ThermoFisher).
9. Dulbecco’s Phosphate-Buffered Saline (DPBS): without cal-
cium chloride and magnesium chloride.
10. Ficoll-Paque PLUS: density gradient separation medium (sold
by GE Healthcare).
11. Dimethyl sulfoxide (DMSO).
12. KnockOut Serum Replacement (KOSR, 90% in freezing

medium).
13. Transduction medium, mesoderm medium and MK medium:
compositions are described in Table 2.
14. Additional materials required for embryoid body culture

(Subheading 3.8): AggreWell™400 (StemCell Technologies,
Cat. 27945), ultralow binding 6-well plates, Collagenase Type
IV (1 mg/mL in AE6 medium), Dispase II (1 mg/mL in AE6

medium), enzyme-free cell dissociation buffer.
15. Additional materials required for cell clump programming

(Subheading 3.9): PI3 kinase inhibitor LY-294002 (10 mM in
DMSO), DPBS supplemented with 0.5 mM EDTA.
2.3 Lentiviral The plasmids containing the recombinant lentiviral backbones for
Vectors expression of GATA1, FLI1 and TAL1 are available through
Addgene (#92416, #92415 and #92417 respectively). Briefly, the
coding sequences of the variant-1 of each TF (Refseq
NM_002049.3, NM_002017.4, NM_003189.5 respectively)
were introduced in place of eGFP into the replication deficient
second generation self-inactivating pWPT viral backbone from
Professor Trono’s laboratory (Addgene #12255). Pseudotyped
amphotropic viral particles can be produced by standard cotrans-
fection protocols using second generation packaging plasmids
(e.g., psPAX2 and pMD2.G; Addgene #12260 and #12259 respec-
tively) following local health and safety regulations. However, the
FLI1 viral vectors have been notoriously difficult to produce with
sufficient titers for MKFOP and an external provider with a special-
ized production platform has proved more reliable (see Note 5).
2.4 Flow Cytometry MK differentiation is routinely checked by flow cytometry during

Antibodies MKFOP progression using a combination of three fluorochrome con-
jugated antibodies directed against the human CD41 (ITGA2B; APC-
conjugated), human CD42a (GP9; FITC-conjugated), and human
CD235a (GYPA; PE-conjugated) and includes a live/dead discrimina-
tor such as DAPI (4′,6-diamidino-2-phenylindole) used at 1 μg/mL.
2.5 Tissue Culture All tissue culture (TC) plasticware are surface treated for tissue-
Plasticware culture attachment by the manufacturer. A wide range of vessel
types have been used for MKFOP (including 24, 12, 6-well plates,
100 mm dishes and 25, 75, 150 cm2 ventilated flasks) allowing
scalability of the method.
3 Methods
All procedures are carried out at room temperature (RT), unless

otherwise stated, following standard aseptic tissue culture tech-
niques. Cell cultures are maintained in standard humidified TC
incubators at 37 °C and 5% CO2. While compatible with 100 U/
mL penicillin-streptomycin culture medium supplementation,
MKFOP is routinely performed without antibiotic addition as we
found the cell growth was adversely affected. The use of hiPSCs
and lentiviral vectors (Genetically Modified Organisms, GMOs) to
initiate MKFOP requires special precautions and good laboratory
practice which must follow local biological health and safety regu-
lations: the work is routinely performed in Containment Level 2
laboratories with personal protective equipment and inactivation

of waste before disposal to the environment.
We describe below the detailed protocol for routine MKFOP, i.e.,
two-dimensional culture on a vitronectin matrix starting from hPSC
colonies dissociated to single cells. Alternative hPSC seeding tech-
niques to initiate MKFOP for a marginal fraction of hPSC lines that
do not perform well using the primary protocol detailed in Subheading
3.1–3.6 are described in the later Subheadings 3.8 and 3.9.
3.1 hPSC Seeding This seeding phase aims to achieve an even layer of single hPSCs
(Day −1) on vitronectin coated TC vessels (refer to Table 1 for recom-
mended starting cell densities on standard vessels; see Note 6).
1. Prepare a sufficient number of vitronectin coated tissue culture
dishes for the experiment. Typically, a ~0.5 μg/cm2 vitronectin
coating is achieved from a solution of 5 μg/mL in DPBS by
adding 250–500–1000 μL per 24–12–6-well plate respectively
and incubating for 1 h at RT. Plates may be stored up to a week
at 4 °C provided they are sealed to prevent drying out.
2. Prepare the required amount of hPSC medium supplemented
with 10 μM Y-27632 Rock-inhibitor (hPSC medium +Y; see
Note 7) following suggested volumes in Table 1 plus 1 mL
extra per hPSC line to be used.
3. Aspirate medium from 50 to 80% confluent hPSC culture and
wash culture well once with DPBS.
4. Remove DPBS and add RT TrypLE using just enough volume
to cover the well (50% of Table 1 culture volumes).
5. Incubate for 3–4 min at 37 °C checking for efficient cell dis-
sociation of hPSC compact colonies under the microscope.
6. Carefully tilt the culture plate and remove TrypLE from the
well by aspiration: while dissociated, the colonies should
remain loosely adherent on the well surface (see Note 8).
7. Dissociate colonies and collect single cells by repeated pipet-
ting of basal AE6 medium over the surface of the well.
Table 1
hPSC seeding density (day −1)
Culture vessel type Surface area/well hPSC seeding/well Vol. medium/well

24-well plate 1.9 cm2 25,000 0.5 mL
12-well plate 3.8 cm2 50,000 1 mL
6-well plate 9.5 cm2 125,000 2.5 mL
100 mm dish 55 cm 2
700,000 14 mL
Table 2
MKFOP culture media composition
Medium Day Base Cytokines (ng/mL) Supplement (μg/mL)

Transduction 0–1 AE6 FGF2 + BMP4 (20 + 10) Protamine sulfate (10)
Mesoderm 1–2 AE6 FGF2 + BMP4 (20 + 10) –
MK 2–120 CellGro TPO + SCF (20 + 25) –
8. Transfer the cells to a conical tube, add 10 mL basal AE6

medium and centrifuge at 300 × g for 5 min.
9. Collect hPSC pellet in 1 mL hPSC medium+Y.
10. Determine the single cell concentration using a hemocytom-
eter (see Note 9).
11. Prepare a cell solution in a new tube at 50,000 cells per mL in
hPSC medium+Y.
12. Distribute the cells into tissue culture dishes following recom-
mended volumes in Table 1.
13. Finally, mix the culture in the incubator by cross-shape move-
ments to distribute cells evenly and incubate for 16 h.
3.2 hPSC This step aims to achieve transduction of the hPSCs with the three
Transduction (Day 0) programming TFs delivered as three separate lentiviral vectors. We
recommend a multiplicity of infection of 20 for each vector (MOI
20, i.e., 20 functional viral particles per cell) which routinely achieves
over 80% transduction efficiencies in hPSC lines (see Note 10). The
lentiviral vectors should be functionally titered by a validated method
(i.e., proviral copy number integration by qPCR against a standard
reporter vector) to obtain an accurate MOI. The calculation for the
volume of lentiviral vector needed to achieve a target MOI based on
cell number and vector batch titer is as follows:
cell number × MOI
Volume lentiviral vector ( uL ) = ×1, 000
vector titre ( TU / mL )

1. Prepare the required amount of transduction medium (see

Tables 2 and 3). The volume is kept to the minimum for a
given surface area to maximize vector concentration and thus
transduction efficiency.
2. Slowly thaw viral vector aliquots at 4 °C.
3. Prepare the 3-vector transduction mix in a sterile tube by add-
ing the required calculated amount of each vector to the trans-
duction medium to achieve MOI 20 (see Note 11).
Table 3
MKFOP culture volumes
Medium (mL) Mesoderm Megakaryocyte Max cells

Culture vessel type Day0–2 Day2–5 Day5–9 Day > 9 Rem Add (2×) (E + 6)
Day1
24-well plate 0.3 0.4 0.6 0.5 0.2 0.3 1
0.3
12-well plate 0.5 0.7 1 1 0.4 0.5 2
0.5
6-well plate 1.5 2 2.5 2 0.8 1 4
1.5
100 mm dish 8 10 15 12 5 6 24
8
The “Rem” and “Add (2×)” columns indicate respectively the volumes of culture medium to be removed and fresh 2×
cytokine concentrated medium to be added for a 50% medium exchange every 72 h. “Max. cells” shows the maximum
recommended cell amount for a given culture well
4. Gently agitate after addition of each vector to avoid

precipitation.
5. Rinse the hPSC culture wells with DPBS once.
6. Add the required 3-vector mix volume per well (see Table 3).
7. Incubate the cells for 24 h.
3.3 MKFOP Culture After transduction, the cells are kept as adherent culture in the
Phase 1 (Days 1–9) original well for a further 8 days, consisting of an initial period of
24 h in mesoderm commitment medium followed by 7 days in MK
supportive medium. The cells will undergo drastic morphological
changes through this period culminating in the release of suspen-
sion cells detaching from the adherent layer as berry-like clusters of
loosely attached cells (Fig. 1). These clusters contain the growing
MK progenitors and become clearly visible from day 5 onward. On
day 9, the whole cell culture content (floating and adherent popu-
lations) is collected for further culture in suspension allowing posi-
tive selection of MKs and their subsequent maturation.
1. On day 1, prepare the required amount of mesoderm medium
to replace the transduction medium (refer to Tables 2 and 3).
2. Remove the transduction medium from the culture wells by
aspiration.
3. Wash the transduced cells with DPBS once.
4. Add the mesoderm medium to the well (see Table 3 for vol-
ume indications) and incubate the cells for 24 h.
5. On day 2, prepare the required amount of MK medium to

replace the mesoderm medium (refer to Tables 2 and 3).
6. Remove the mesoderm medium from the culture wells by
aspiration.
7. Wash the cells with DPBS once.
8. Add MK medium to the well (see Table 3) and incubate the
cells for 72 h.
9. On day 5 prepare the required amount of MK medium with
two times cytokine concentration for refreshing the MK
medium (refer to Add (2×) column in Table 3).
10. Add directly to the culture well without further handling.
12. On day 7, prepare the required amount of MK medium with
two times the cytokine concentration for exchanging the MK
medium (refer to Tables 2 and 3).
13. Carefully remove 40% of culture medium volume from the
well (refer to Rem. column in Table 3). The culture vessels
should be unperturbed before proceeding with the removal.
To avoid aspiration of cells in suspension, gently tilt the plates
at 30° and aspirate slowly from the surface of the medium with
a micropipette (see Note 12).
14. Add the required volume of MK medium with two times cytokine
concentration to the well as in Table 3 and gently mix by agitation.
3.4 MKFOP On day 9 the whole cell culture content—including the suspension
Midculture Split and adherent cell fractions—is collected, dissociated and reseeded
(Day 9) in larger tissue culture wells for further MK growth. This is also a
good timepoint to monitor differentiation efficiency analysing cell
phenotype by flow cytometry using triple-staining for expression
of CD41 (ITGA2B), CD42a (GP9), and CD235a (GYPA). This
combination allows the discrimination of MK progenitors (MKP),
MKs, mature MKs, and contaminating erythroblasts (Fig. 1).
1. Prepare the required volume of MK medium as in Tables 2
and 3. Plan for a 2.5-fold increase of cell culture surface area
after the day 9 culture dissociation (see Note 13).
2. Gently agitate the culture dish and collect the cells in suspen-
sion into a sterile conical centrifuge tube (typically 15 or 50 mL
tubes).
3. Gently rinse the well with DPBS to collect further remaining
floating cells and pool with the first collection.
4. Add minimum sufficient TrypLE volume to cover the surface
of the well (50% of Table 1 culture volumes) and incubate for
10 min at 37 °C.
5. Add basal CellGro medium to the well (3 times the TrypLE

volume added at step 4) and collect the adherent cell fraction
by pipetting up and down several times over the surface
(see Note 14).
6. Pool the adherent cells with the suspension cell fraction.
7. Fill the collection tube up with DPBS and pellet the cells by
centrifugation at 300 × g for 5 min.
8. Collect the cell pellet in 0.5 mL MK-medium.
9. Put the cells back in culture in a new well providing 2.5 times
the culture surface area using the recommended volume of
MK medium (as Table 1).
10. We recommend cell phenotype analysis by flow cytometry, for
monitoring MK programming efficiency: collect a fraction of
the culture for analysis as described in Note 15.
3.5 MKFOP Culture The second culture phase strongly selects for forward programmed
Phase 2 cells in suspension and allows maturation of the MKs to double
and Phenotyping positive CD41+/CD42+ cells. The purity for CD42+ mature MKs
(Days 9–21) by day 21 will vary from 20 to 80% and their expansion from hPSC
input from 2 to 50 depending on the hPSC line. During this phase,
it is important to keep the cell density in the range of 0.5–2 million
cells per mL to maximize healthy culture growth. This can be car-
ried out by checking the culture cell density every 72 h alongside
feeding. If splitting is required, gently mix the culture by swirling
or gentle pipetting and distribute accordingly into new culture
wells. In addition, the cell handling should be minimized and when
necessary performed with care to avoid MK damage: this is rou-
tinely achieved by gentle pipetting using only large aperture tips
(e.g., 1 mL) and reducing centrifugation force to 120 × g with
slow acceleration and braking.
1. On day 12, 72 h after the midculture split, prepare the required
amount of MK medium with two times cytokine concentra-
tion as in Tables 2 and 3 for MK medium refreshment.
2. Carefully remove 40% of the culture medium from the wells.
The culture vessels should be unperturbed before proceeding
with the removal. To avoid aspiration of cells in suspension,
gently tilt the plates about 30° and aspirate slowly from the top
edge of the liquid column with a micropipette (see Note 16).
3. Add the required volume of MK medium containing two
times cytokine concentration to the well.
5. On day 15, prepare fresh MK medium with two times cytokine
concentration and feed the cells by following the same protocol
as described for day 12 (steps 1–3) and check the cell density
to ensure this is in the range of 0.5–2 million cells per mL.
6. On day 18, proceed as day 15.

7. On day 21, prepare required amount of MK medium with
normal cytokine concentration as in Tables 2 and 3.
8. Carefully collect all the cells after mixing by gentle swirling
and pipet into a sterile tube.
9. Wash the well once with CellGRO medium to collect remain-
ing cells and pool with the first cell collection.
10. Centrifuge at 120 × g for 8 min using slow acceleration and
brake settings and collect the cell pellet in 1 mL of MK
medium.
11. Count the cells using a hemocytometer and adjust cell density
to 0.5 million per mL using MK medium and transfer to new
culture plates following volumes suggested in Table 3.
12. Take a 50 μL aliquot for flow cytometry analysis of CD41/
CD42a/CD235a surface phenotype (see Note 15).
3.6 MKFOP Long- The forward programmed MK culture can be maintained from 30
Term Culture to up to 120 days before exhaustion of the MK progenitor pool; the
(Days>21) limit varies with hPSC line and programming batch (see Note 17).
The cell density should be maintained in the 0.5–2 million cells/
mL range with 50% medium exchange using 2× concentrated MK
medium every 72 h as described in Subheading 3.5 above. The cells
are routinely split into new culture dishes with a 100% medium
exchange every 10 days following procedure Subheading 3.5, steps
7–12. It is common for forward programmed MK cultures to go
through acute “crises” where reduced growth is accompanied by
high fragmentation and cell death during the course of long-term
culture. Cultures will generally recover after 5–7 days marked by
the reappearance of clusters of proliferating progenitors. However,
the accumulation of cell debris and platelets in long-term mature
MK cultures eventually becomes detrimental for healthy cell
growth. When the culture drops below 20% total viable cells
(Fig. 2), we suggest performing a Ficoll-Paque PLUS density gradi-
ent separation to enrich for the live cell fraction (expect 50% live cell
recovery) and continuation of long-term culture as follows:
1. Prepare required number of 15 mL conical tubes with 3 mL
Ficoll solution (1 tube per ≤10 mL of MK culture).
2. Gently collect up to 10 mL of the MK culture and carefully
layer over the Ficoll fraction.
3. Centrifuge at 400 × g for 15 min using the slowest accelera-
tion and no brake.
4. After centrifugation, preserve 50% of the media in the top
fraction as a conditioned medium for further culture.
Fig. 2 Long-term MKFOP culture. (a) The cellular content of long-term MK culture is complex and encompasses
a mix of growing MK progenitor clusters, mature MKs, dead cells/debris (notably coming from MK fragmentation
through platelet release), and platelets as depicted. (b) a representative phase contrast microscope picture
including foci of MK proliferation (white arrows) and large single MKs (red arrows); scale bar 50 μm. (c)
Romanowsky staining of long-term MK culture shows frequent large polyploid MKs (red arrows); scale bar 50 μm.
(d) Dead cells, fragmenting MKs and debris (population circled in red in a typical flow cytometry morphological
dot plot) can be excluded from the whole culture by a Ficoll-Paque density gradient centrifugation (see Subheading
3.6); this procedure restores MK purity and preserves MK progenitors sustaining long-term culture
5. Carefully collect the ring of live MKs at the Ficoll-medium

interface with a 1 mL tip and transfer the cells into a new coni-
cal 15 mL centrifuge tube (see Note 18).
6. Wash the collected live cells by adding threefold the volume of
basal CellGro medium and centrifuging at 120 × g for 8 min
using slow acceleration and slow brake.
7. Remove the supernatant by aspiration and gently collect cell
pellet in 0.5 mL of the preserved conditioned medium.
8. Count the cells with a hemocytometer and adjust their concentra-
tion to 1 million/mL with conditioned medium (see Note 19).
9. Further dilute the cells twofold using MK medium with two
times cytokine concentration to continue long-term culture.
3.7 MKFOP Culture The cells obtained by MKFOP can be cryopreserved at any time
Cryopreservation from day 10. This is best done from a healthy and actively growing
culture, with reduced amount of cell debris, by freezing at 0.5–2
million cells per cryovial in a final concentration of 10% DMSO. The
thawing efficiency will vary significantly depending on the culture
state at freezing (i.e., ratio MK progenitors/mature MKs/cell
debris) and could take up to 7–10 days before the culture is actively

proliferating again.
1. Gently collect MKs directly from FOP culture in a conical cen-
trifuge tube and determine cell concentration using a
hemocytometer.
2. Pellet the cells by centrifugation at 120 × g for 8 min.
3. Aspirate the supernatant and gently resuspend the pellet in
cold 100% Knockout Serum Replacement (KOSR) to achieve
1–4 million cells/mL.
4. Prepare an equal volume of cold KOSR +20% DMSO in a
separate tube.
5. Add the KOSR 20% DMSO dropwise to the cell mix while
constantly agitating the tube.
6. Aliquot 1 mL per vial in 1.5 mL cryovials and transfer to a
cryobox to be placed at −80 °C overnight for freezing. For
long term storage, the cells are then kept either in liquid
Nitrogen or −150 °C freezers.
3.8 First Alternative The original MKFOP method described in [9] used embryoid
Protocol: Embryoid body (EB) culture for the first 9 days of programming. This proto-
Body Programming col requires special techniques of tissue culture handling (Fig. 3).
(Days 0–9) While the routine 2D single cell MKFOP method described above
is efficient for most hPSC lines, the embryoid body approach
described in this Subheadings 3.8.1–3.8.7 still provides better out-
come for a small fraction of the lines and may be a valuable alterna-
tive for such refractory lines (see Note 20).
3.8.1 Day 0, EB 1. Prepare EB transduction medium (1 mL per million hPSCs)

Formation made of AE6 supplemented with BMP4 at 10 ng/mL, prot-
and Transduction amine sulfate at 10 μg/mL and Y-27632 at 10 μM.
2. Prepare the AggreWell-400 plate by adding 0.5 mL of EB
transduction medium per well (1 well needed per million
hPSCs) and centrifuging at 2000 × g for 5 min (see Note 21).
3. Prepare single hPSC suspension as Subheading 3.1, steps 3–8
and collect cell pellet in 200 μL of EB transduction medium
(per 50–80% confluent 10cm2 hPSC well dissociated)
4. Determine cell concentration with a hemocytometer and
adjust it to 5 million/mL with EB transduction medium.
5. Pipette 200 μL of cells per programming (i.e., 1 million
hPSCs) into a new sterile tube.
6. Add required amount of the 3-TF lentiviral vectors to reach
MOI 20 as Subheading 3.2.
7. Gently pipette the cell-vector mix up and down before adding
it to one prepared well of the Aggrewell plate.
Fig. 3 Embryoid body MKFOP. The embryoid bodies (EBs) formed in the Aggrewell plates are homogenous in
size on day 1 (average 830 cells/EB per million hPSC sown). Through MKFOP progression, the EBs will grow
and form vesicular structures (white arrows), eventually bursting out and releasing berry-like MK clusters in
suspension (black arrows). Scale bars 200 μm; MKFOP day indicated
8. Gently pipette up and down to homogenize the cell content in

the AggreWell, as much as possible without reintroducing
bubbles, and centrifuge the plate enclosed in an aerosol-tight
container (e.g., a zip-lock plastic bag) at 100 × g for 3 min
using low acceleration and brake (see Note 22).
3.8.2 Day 1, EB 1. Prepare EB mesoderm medium (2 mL × number of wells

Collection seeded at day 0) from AE6 supplemented with BMP4 at
10 ng/mL and FGF2 at 5 ng/mL.
2. Prepare one 15 mL conical tube filled with 7 mL basal AE6
per Aggrewell to be collected.
3. Collect the EBs from the AggreWell using a 1 mL micropi-
pette by firmly pipetting down 1 mL of basal AE6 in one pro-
gressive spiral movement over the whole well surface (without
actually touching the microwell surface to avoid EB disrup-
tion) (see Note 23).
4. Immediately collect the floating EBs and carefully transfer them
into the prepared 15 mL tube (pipette cells into the media to
avoid EBs sticking to the tube wall and to minimize shear stress).
5. Proceed with two additional collections as above to maximize

EB recovery and check under the microscope that most of
them have been dislodged from the microwells (see Note 24).
6. Pellet EBs by low speed centrifugation at 20 × g for 1 min (see
Note 25).
7. Aspirate supernatant and collect EB pellet in 2 mL EB meso-
derm medium.
8. Seed collected EBs into 1 well of an ultralow binding 6-well
plate and mix by cross-shape movements to distribute EBs
evenly and avoid their aggregation.
9. Incubate the EBs for 24 h.
3.8.3 Day 2, EB Split 1. Prepare MK medium (4 mL × number of wells seeded day 1)
and MK Culture as in Table 2.
2. Prepare one 15 mL conical tube filled with 10 mL DPBS per
well to be collected.
3. Carefully collect EBs using a 1 mL micropipette and transfer
them into the prepared 15 mL tube.
4. Pellet EBs by centrifugation at 20 × g for 1 min (see Note 25).
5. Aspirate supernatant and collect EBs in 4 mL MK medium.
6. Plate collected EBs into 2 wells of an ultralow binding 6-well
plate (see Note 26) and mix by cross-shape movements to dis-
tribute EBs evenly and avoid their aggregation.
7. Incubate the EBs for 72 h.
3.8.4 Day 5, MK Culture Follow the steps in Subheading 3.3, steps 9–11.
Refreshment
3.8.5 Day 7, MK Medium Follow the steps in Subheading 3.3, steps 12–15.
Exchange
3.8.6 Day 9, EB 1. Prepare MK medium (6 mL per paired-wells in culture) as in

Collection and Single Cell Table 2.
Split 2. Prepare one 15 mL conical tube filled with 8 mL DPBS for
each paired-well culture to be collected.
3. Collect all the cells (EBs and any floating cells) and transfer
them into the prepared 15 mL tube.
4. Pellet EBs and cells by centrifugation at 300 × g for 5 min.
5. Aspirate the supernatant and add 1 mL of a 1:1 Collagenase/
Dispase mix to the cell pellet.
6. Gently mix by pipetting up and down with a micropipette.
7. Incubate at 37 °C for 45 min with gentle mixing by tapping

the tube a couple of times over the incubation period.
8. Add 10 mL DPBS and centrifuge at 300 × g for 5 min.
9. Aspirate supernatant and add 0.3 mL of enzyme-free cell dis-
sociation buffer (CDB) to the cell pellet.
10. Firmly pipette up and down using a 200 μL micropipette every
2 min for a total of 10 min to generate a single cell suspension
(see Note 27).
11. Add 10 mL CellGro medium to the cell suspension and cen-
trifuge at 300 × g for 5 min.
12. Collect the cell pellet in 1 mL MK medium and count viable
single cells using a hemocytometer.
13. Adjust cell concentration to 0.5 million/mL and seed in tissue
culture plates as recommended in Table 3.
14. Collect a fraction of the culture for recommended monitoring
of MK programming efficiency by flow cytometry analysis of
cell phenotype (see Note 15).
3.8.7 MK Culture Further MK culture is carried out as Subheadings 3.5–3.6.

(Day>9)
3.9 Second This alternative method of MK forward programming is a minor

Alternative Protocol: variation of the routine 2-D single cell protocol where hPSCs are
Cell-Clump sown as colony clumps (see Note 28). This option may be favoured
Programming for rare hPSC lines showing poor attachment and survival after
(Day −1) single cell plating.
1. On day −1, prepare vitronectin coated plates as described in
Subheading 3.1, step 1.
2. Remove culture medium from 50 to 80% confluent hPSC cul-
ture and wash once with DPBS.
3. Add a sufficient volume of DPBS/EDTA to cover the well
surface and incubate for 5 min at RT (see Note 29).
4. Aspirate DPBS/EDTA and add AE6 basal medium to cover
the well.
5. Generate cell clumps from the loosened hPSC colonies by
scraping over the well surface with a 2 mL pipette.
6. Collect the floating clumps and break them further by pipet-
ting up and down 3–5 times with a 1 mL micropipette to the
bottom of a 15 mL conical tube.
7. Based on original culture density (see Note 30), evaluate the
total cell number collected as clumps and bring cell concentra-
tion to 50,000 per mL in basal AE6.
8. Plate suggested number of hPSC (see Table 1) onto prepared

vitronectin plates using routine pluripotency medium required
for the line (as described in Subheading 2.1).
9. Cross-mix the plate and incubate the cells for 24 h.
10. From day 0 onward, follow the steps described for the routine
single cell MKFOP protocol except for the addition of
LY-294002 at 10 μM to the mesoderm medium (day 0–2) to
increase efficiency of mesoderm commitment when starting
from hPSC clumps [14].
4 Notes
1. The development of alternatives to single lentiviral vectors is

ongoing and includes (1) an optimized polycistronic cassette
expressing the 3TFs from a single RNA embedded in a single
lentiviral vector to obtain a homogenous transduction of the
3TFs and (2) the stable directed targeting into the hPSC
genome of a doxycycline inducible expression cassette express-
ing the 3TFs to generate chemically inducible MKFOP lines
independent from lentiviral transduction.
2. The molecular mechanisms leading to the generation of MK
progenitors are under investigation to increase the success rate
in the generation of sustained long-term MKFOP culture in
more hPSC lines.
3. To make AE6 medium (in house equivalent to commercial
Essential-6 medium), we routinely add to 500 mL of DMEM/
F12 (ThermoFisher #11330032): 3.6 mL 7.5% NaHCO3
(ThermoFisher #25080094), 10 mL Insulin-Transferrin-
Selenium premixed 100× solution (i.e., used as 50× here)
(ThermoFisher #41400045) and 5 mL l-ascorbic acid stock
solution. The latter is made from 1 g of L-ascorbic acid 2-phos-
phate sesquimagnesium salt hydrate (Sigma-Aldrich #A8960)
dissolved in 156 mL sterile tissue culture water, 0.2 μm filtered
and stored at −20 °C as 5 mL single-use aliquots.
4. Alternatives to routine CellGro-SCGM medium are StemSpan-
SFEM or StemPro-34 which have been tested albeit less
extensively.
5. MKFOP is highly dependent on efficient vector transduction
and quality of the lentiviral batches. Notably, their titer should
be calculated accurately using functional transduction assays
(e.g., qPCR for genomic integration) in order to achieve correct
and balanced MOI for the 3TFs at the transduction step.
Moreover, the vectors must be handled with care (slow thawing,
gentle mixing, avoid temperature shock) and not used after
more than two freeze–thaw (correcting for 30% efficiency loss
after each cycle) to achieve optimal transduction efficiencies.
6. The starting hPSC density may need to be adjusted for differ-

ent lines depending on their replating efficiencies and prolif-
eration properties. Optimal cell seeding for programming
should give you around 20% confluence on day 0 with even
cell spreading. Most importantly, cell confluence should not
be reached before day 3/4 of programming as it impairs dif-
ferentiation efficiency.
7. The addition of the Rock-inhibitor Y-27632 significantly
increases hPSC survival and thus replating efficiency after sin-
gle cell dissociation [15]. Typically, add 1 μL of Y-27632
10 mM stock solution per mL of hPSC culture medium.
8. If the hPSC colonies have fully detached from the matrix after
TrypLE treatment, collect the cells in TrypLE after addition of
1 mL basal AE6 medium. Proceed with 2 sequential washes in
10 mL basal AE6 medium collecting the cells by centrifuging
at 300 × g for 5 min to remove all TrypLE activity.
9. Cell viability should exceed 90% after single cell dissociation.
Expect freshly dissociated hPSCs from compact colonies to
show irregular size and membrane shape. Small clumps of cells
may remain after cell dissociation and will not negatively impact
on MKFOP provided they represent only a minor fraction of
the cell population and are no bigger than 3–5 cells. As much as
possible, each individual cell in these remaining clumps should
be counted for accuracy. If cell clumps make up the majority of
the collected cell population, the TrypLE treatment duration
should be increased to achieve better single cell dissociation.
10. We use an eGFP reporter lentiviral vector (in which the pWPT
backbone is identical to that of the 3TF vectors) to separately
assess transduction efficiency of hPSC lines. While a vast major-
ity of lines show over 80% transduction with an MOI of 20 and
the indicated seeding procedures, we recommend to test any
new hPSC line to rule out transduction refractoriness.
11. Gently but thoroughly mix the concentrated vector batches by
pipetting before addition to the medium. We suggest prepar-
ing a master mix of vectors in medium for the total number of
wells to be transduced, e.g., for 10 wells of a 12WP make
5.5 mL (i.e., 10% extra) of transduction medium plus vectors
(+10% extra) and then add 0.5 mL per well.
12. For routine medium exchange throughout MKFOP culture,
we remove 40% of the initial culture volume and add 50% of
cytokine twice concentrated fresh medium (to allow for media
and cytokine activity loss, through evaporation and 37 °C cul-
ture respectively, over the 3-day period). This culture regi-
men—50% medium exchange every 72 h—has proved optimal
for sustained cell growth. The medium carefully removed as
described from unperturbed plates should be virtually cell-free
as cells in suspension tend to collect at the bottom and at the

center of the wells. If cells are inadvertently collected, they
may be recovered by small speed centrifugation at 120 × g for
8 min and added back to the culture.
13. Typically, cell expansion over the first 9 days (20–40×), will
require transfer from a well of 12- or 6-well plate into one well
of a 6-well plate or T25 flask respectively.
14. Patches of strongly adherent cells forming nondispersible

clumps may remain after TrypLE treatment. These belongs to
the nonprogrammed cell fraction which will quickly disappear
by counter selection through the second culture phase.
15. The CD41/42/235 surface phenotyping at day 9 is routinely
performed by immunostaining 10% of the total collected cells
directly in their culture medium for 20 min at room tempera-
ture, followed by a wash with DPBS 0.5% BSA 2 mM EDTA and
centrifugation at 300 × g for 5 min. The cells are subsequently
analysed by flow cytometry, including DAPI 1 μg/mL to exclude
dead cells from analysis and quantification of fluorescent beads
to calculate cell number in each sample. A typical flow analysis is
shown in Fig. 4. At this stage, most CD41+ programmed cells
coexpress CD235 with very low percentage of CD42 expressing
cells detected. An efficient MKFOP will show >50% CD41+ cells
by day 9 while <5% CD41+ indicates failure of programming. By
Day 21 the culture will be from 20 to 80% CD41+/CD42+ and
staining as above is repeated using 50 μL aliquot directly from
the culture. The long-term culture will develop toward highly
CD41+ pure cultures showing >60% CD42+.
16. When MKs are cultivated in flasks, the procedure needs to be
adjusted as any removed medium will contain suspension cells.
The latter should then be centrifuged at 120 × g for 8 min in
conical tubes to pellet the cells which are then collected gently
in fresh medium and returned to the culture flask.
17. Some long-term MKFOP cultures may eventually turn to a
CD41 single-positive population having lost CD42 expres-
sion. This occurrence appears to be linked to particular vector
batches and thus probably to a variation in the early program-
ing mechanisms. While the %CD42+ cells may fluctuate from
50 to 95% through the long-term culture, a stable loss of
CD42 expression below 30% correlated with the loss of MK
clusters and the development of bright single cells homoge-
nous in size visible under the microscope, in addition to the
disappearance of the debris/dead cells population in flow
cytometry plots, are strong indications that the long-term cul-
ture has turned into an immature blast CD41+ culture.
18. Ficoll-Paque interface/“ring” collection. This is the same prin-
ciple used for isolating mononuclear cells from whole blood by
Fig. 4 Flow cytometry analyses. Typical flow cytometry data and analysis is shown for day 9, day 21, and day
30 cultures. The surface phenotype is analysed through a cell/singlet/live triple-gating strategy. After the day
9 split, two morphologically distinct cell populations are routinely observed which correspond respectively to
the programmed and nonprogrammed cells (a/b respectively on top-left dot plot). The latter are selected out
by the cell culture conditions and gradually disappear from the culture
layering over ficoll. Viable MKs are collected from the ring of
cells at the interface between the ficoll layer and the media by
using a 1 mL micropipette or a 2 mL sterile movette (plastic suc-
tion pipette). The MKFOP cells frequently form large visible
clusters just below the interface that should be collected as well.
19. Should insufficient conditioned media be available, use basal
CellGro medium to complete.
20. The EB technique is best used from feeder free hPSC cultures
and can improve programming in lines with poor plating effi-
ciency as single cells which can be sensitive to viral load.
21. Refer to manufacturer for a detailed AggreWell™ protocol and
troubleshooting. Using 1 million cells per AggreWell™400
(where each well contains 1200 microwells) gives 833 cells per
EB: the range 600–850 cells per EB favours hematopoietic dif-
ferentiation. We observed that EB formation could be sub-
optimal in some batches of plates: in this case plates may be
rinsed with AggreWell Rinsing Solution (Cat. 07010) following
the manufacturers protocol to ensure all air bubbles are removed.
22. Check under a microscope the homogeneity of cell distribu-
tion inside the microwells. Equal numbers of cells should be
seen and each microwell approximately 75% filled.
23. Avoid as much as possible the generation of bubbles to limit
the production of aerosols throughout the process of collec-
tion as the culture still contains a high viral vector dose.
24. Check around the periphery of the well in particular for any
remaining EBs. Should sticking to either tips or tubes be a
problem, siliconized or LoBind™ tubes and nonstick tips (e.g.,
RPT Repel Polymer Technology) may be used.
25. For smaller EBs (low starting cell number, suboptimal forma-
tion), centrifugation can be pushed up to 100 × g/2 min/RT.
26. For smaller EBs (low starting cell number, suboptimal forma-
tion), seed into one well only.
27. Cell aggregates may remain after treatment but will not impair
further MK maturation; using a smaller p200 tip rather than a
p1000 at this stage will aid breaking up the EBs. Should stick-
ing to either tips or tubes be a problem, siliconized or LoBind™
tubes and nonstick tips (e.g., RPT Repel Polymer Technology)
may be used.
28. If the cells are routinely cultured feeder free and passaged as
clumps using PBS/EDTA, seeding forward programming as
clumps does not require Rock-inhibitor Y-27632 for the first
24 h. If cells are routinely cultured on another matrix such as
Laminin seeding as clumps is more difficult as TrypLE is usu-
ally used for subculture, using the single cell method is there-
fore recommended for these lines.
29. Use of DPBS/EDTA treatment is for slightly longer than a
routine culture split, i.e., up to 5–8 min to foster small clump
production for increased transduction efficiency and subse-
quent differentiation to MKs. Clumps should not be seen
clearly in the collection tube since they are just below resolu-
tion for average eyesight.
30. For estimation of hPSC culture cell count, each hPSC line
tends to have a characteristic cell count for a confluent well
(for hPSCs this means a well approximately 70–80% covered
in colonies). For average lines this may be 1.5–2 million per
well of a 6-well plate. When working with a new line it may be
worthwhile to TrypLE treat a spare confluent well to obtain
an accurate estimate of the number of cells. With experience,
numbers can be estimated with good accuracy.
Acknowledgments
This work was supported by NIHR, NHSBT, MRC grants and a

core support grant from the Wellcome Trust and MRC to the
Wellcome Trust – Medical Research Council Cambridge Stem Cell
Institute.
References
1. Rossant J, Tam PP (2017) New insights into One 8(2):e55530. https://doi.org/10.1371/

early human development: lessons for stem cell journal.pone.0055530
derivation and differentiation. Cell Stem Cell 9. Moreau T, Evans AL, Vasquez L, Tijssen MR,
20(1):18–28. https://doi.org/10.1016/j. Yan Y, Trotter MW, Howard D, Colzani M,
stem.2016.12.004 Arumugam M, Wu WH, Dalby A, Lampela
2. Takahashi K, Tanabe K, Ohnuki M, Narita M, R, Bouet G, Hobbs CM, Pask DC, Payne H,
Ichisaka T, Tomoda K, Yamanaka S (2007) Ponomaryov T, Brill A, Soranzo N, Ouwehand
Induction of pluripotent stem cells from adult WH, Pedersen RA, Ghevaert C (2016) Large-
human fibroblasts by defined factors. Cell scale production of megakaryocytes from human
131(5):861–872. https://doi.org/10.1016/j. pluripotent stem cells by chemically defined
cell.2007.11.019 forward programming. Nat Commun 7:11208.
3. Wu SM, Hochedlinger K (2011) Harnessing https://doi.org/10.1038/ncomms11208
the potential of induced pluripotent stem 10. Heazlewood SY, Nilsson SK, Cartledge K, Be
cells for regenerative medicine. Nat Cell Biol CL, Vinson A, Gel M, Haylock DN (2017)
13(5):497–505. https://doi.org/10.1038/ Progress in bio-manufacture of platelets for
ncb0511-497 transfusion. Platelets 28(7):649–656. https://
4. Gaur M, Kamata T, Wang S, Moran B, Shattil doi.org/10.1080/09537104.2016.1257783
SJ, Leavitt AD (2006) Megakaryocytes derived 11. Cahan P, Daley GQ (2013) Origins and impli-
from human embryonic stem cells: a geneti- cations of pluripotent stem cell variability and
cally tractable system to study megakaryo- heterogeneity. Nat Rev Mol Cell Biol. https://
cytopoiesis and integrin function. J Thromb doi.org/10.1038/nrm3584
Haemost 4(2):436–442. https://doi. 12. Rouhani F, Kumasaka N, de Brito MC, Bradley
org/10.1111/j.1538-7836.2006.01744.x A, Vallier L, Gaffney D (2014) Genetic back-
5. Takayama N, Eto K (2012) Pluripotent ground drives transcriptional variation in
stem cells reveal the developmental biol- human induced pluripotent stem cells.
ogy of human megakaryocytes and provide PLoS Genet 10(6):e1004432. https://doi.
a source of platelets for clinical application. org/10.1371/journal.pgen.1004432
Cell Mol Life Sci. https://doi.org/10.1007/ 13. Chen G, Gulbranson DR, Hou Z, Bolin JM,
s00018-012-0995-4 Ruotti V, Probasco MD, Smuga-Otto K,
6. Nakamura S, Takayama N, Hirata S, Seo H, Howden SE, Diol NR, Propson NE, Wagner
Endo H, Ochi K, Fujita K, Koike T, Harimoto R, Lee GO, Antosiewicz-Bourget J, Teng JM,
K, Dohda T, Watanabe A, Okita K, Takahashi Thomson JA (2011) Chemically defined con-
N, Sawaguchi A, Yamanaka S, Nakauchi H, ditions for human iPSC derivation and culture.
Nishimura S, Eto K (2014) Expandable mega- Nat Methods 8(5):424–429. https://doi.
karyocyte cell lines enable clinically appli- org/10.1038/nmeth.1593
cable generation of platelets from human 14. Bernardo AS, Faial T, Gardner L, Niakan KK,
induced pluripotent stem cells. Cell Stem Cell Ortmann D, Senner CE, Callery EM, Trotter
14(4):535–548. https://doi.org/10.1016/j. MW, Hemberger M, Smith JC, Bardwell L,
stem.2014.01.011 Moffett A, Pedersen RA (2011) BRACHYURY
7. Feng Q, Shabrani N, Thon JN, Huo H, Thiel and CDX2 mediate BMP-induced differen-
A, Machlus KR, Kim K, Brooks J, Li F, Luo tiation of human and mouse pluripotent stem
C, Kimbrel EA, Wang J, Kim KS, Italiano J, cells into embryonic and extraembryonic lin-
Cho J, Lu SJ, Lanza R (2014) Scalable gen- eages. Cell Stem Cell 9(2):144–155. https://
eration of universal platelets from human doi.org/10.1016/j.stem.2011.06.015
induced pluripotent stem cells. Stem Cell Rep 15. Watanabe K, Ueno M, Kamiya D, Nishiyama
3(5):817–831. https://doi.org/10.1016/j. A, Matsumura M, Wataya T, Takahashi JB,
stemcr.2014.09.010 Nishikawa S, Nishikawa S, Muguruma K, Sasai
8. Pick M, Azzola L, Osborne E, Stanley EG, Y (2007) A ROCK inhibitor permits survival
Elefanty AG (2013) Generation of megakaryo- of dissociated human embryonic stem cells.
cytic progenitors from human embryonic stem Nat Biotechnol 25(6):681–686. https://doi.
cells in a feeder- and serum-free medium. PLoS org/10.1038/nbt1310
Chapter 11
Three-Dimensional Tissue Models for Studying Ex Vivo

Megakaryocytopoiesis and Platelet Production
Christian A. Di Buduo, Vittorio Abbonante, Lorenzo Tozzi, David L. Kaplan,
and Alessandra Balduini
Abstract
Three-dimensional (3D) tissue cultures in vitro enable a more physiological reconstruction of native
tissues and organs. The bone marrow environment, structure and composition regulate megakaryocyte
function and platelet production. Here, we describe the use of silk fibroin protein biomaterials to assemble
3D scaffolds mimicking the bone marrow niche architecture and extracellular matrix composition to sup-
port platelet release from human megakaryocytes. Additionally, we also propose the use of hyaluronan
hydrogels, functionalized with extracellular matrix components, to reproduce the 3D matrix structure of
the bone marrow environment for studying human megakaryocyte function.
Key words Bone marrow, Megakaryocytes, Platelets, Silk, Hydrogel, Hyaluronic acid, Extracellular
matrices, Bioreactors, 3D modeling
1 Introduction
Advanced cell culture techniques have led to the development of a

new generation of tissue models that can recapitulate the three-
dimensional (3D) architecture and organization of native tissues
and organs. These systems are likely to promote a radical change in
biomedical research allowing the study of cell processes that were
previously observable only in in vivo settings. Bone marrow resides
within the spongy bones where hematopoietic stem cells (HSCs),
multipotent self-renewing progenitors, differentiate and release
blood cells [1, 2]. It has been demonstrated that both the physical
and biochemical characteristics of the bone marrow environment
impact HSC function [3]. In this context, megakaryocytes release
platelets into the blood stream, responding to a precise regulatory
process that involves soluble factors, extracellular matrix compo-
nents, environmental elasticity and structure [4–8]. On this basis,
in recent years, researchers have focused efforts in developing
177
178 Christian A. Di Buduo et al.
ex vivo models to reproduce bone marrow structure and functions

in order to extrapolate its function in vivo [9, 10]. We have dem-
onstrated that silk fibroin, from Bombyx mori silkworm cocoons,
represents a promising biomaterial for engineering physiologically
relevant human bone marrow environment tissue models [11–13].
The important features of silk are that it is a naturally derived, bio-
compatible, and nonthrombogenic biomaterial [14]. Additionally,
silk can be prepared in a range of material formats and processed
entirely in aqueous systems, allowing for the incorporation of labile
compounds without loss of bioactivity [15–18]. In our tissue mod-
els, silk sponges are fabricated to mimic the spongy structure of the
bone surrounding the marrow, while porous silk vascular tubes are
designed to mimic the vascular niche where platelets are released
from the megakaryocytes and collected by perfusion [11, 12].
Importantly, silk elasticity, morphological pattern and functional-
ization can be modified to reproduce a range of bone marrow
characteristics and to promote different cell functions [11].
A simpler system to reproduce the bone marrow extracellular
matrix environment is also represented by the use of hyaluronan
hydrogels with entrapped extracellular matrix components [19].
Although human megakaryocytes express hyaluronan receptors,
we have demonstrated that these receptors do not impact mega-
karyocyte differentiation, maturation, and platelet formation [19].
Thus, functionalized hyaluronan hydrogels represent a useful bio-
material format to study the impact of extracellular matrix compo-
nents on megakaryocyte function in a 3D setting.
Here, we suggest the implementation of these culture systems
with megakaryocytes derived either form human umbilical cord
blood or adult peripheral blood, two of the best characterized
sources of hematopoietic progenitors for recapitulating the whole
process of human megakaryopoiesis, from differentiation to func-
tional platelet production, in both physiologic and pathologic con-
ditions [8, 20]. However, based on the desired outcomes, it is
conceivable to obtain successful results by using megakaryocytes
derived also from other sources, including mouse derived-
hematopoietic progenitors or the emerging and promising option
of human pluripotent stem cells (hPSCs) [21–23].
2 Materials
Prepare all solutions using ultrapure water and analytical grade

reagents. Prepare and store all reagents at room temperature
(unless otherwise indicated). Diligently follow all waste disposal
regulations when disposing of waste materials.
2.1 Megakaryocyte 1. Standard equipment for sterile collection of human peripheral

Differentiation blood or human umbilical cord blood (see Subheading 4.1,
Note 1).
3D Tissue Models for Studying Megakaryocytopoiesis 179
2. 50 mL sterile conical tubes.

3. 15 mL sterile conical tubes.
4. Phosphate-buffered saline (PBS).
5. Ficoll-Paque™ for density gradient separation of mononuclear
cells.
6. Roswell Park Memorial Institute medium (RPMI).
7. Serum-free medium for stem cell culture (We suggest the use
of StemSpan™ SFEM).
8. Bovine serum albumin (BSA).
9. Ethylenediaminetetraacetic acid (EDTA) stock.
10. Petri dishes (100 mm).
11. 6-well sterile plates.
12. Cytokines: interleukin (IL)-11, thrombopoietin (TPO).
13. Antibiotics (penicillin/streptomycin, P/S, 100× stock).
14. l-glutamine (100× stock).
15. Anti-CD34 immuno-magnetic bead conjugated antibody (for
human umbilical cord blood processing).
16. Anti-CD45 immuno-magnetic bead conjugated antibody (for
adult peripheral blood processing).
17. Immunomagnetic-selection kit.
2.2 Silk Bone 1. Bombyx mori silkworm cocoons (see Subheading 4.2, Note 1).
Marrow Tissue Model 2. Sodium carbonate (Na2CO3).
2.2.1 Silk Fibroin 3. Lithium bromide (LiBr).
Extraction 4. Titanium scissors.
5. Hot hand protectors.
2.2.2 Dialysis 1. Dissolved silk fibroin/LiBr or silk fibroin solution (7–8%, wt/vol).
and Concentration of Silk 2. 10 mL and 20 mL syringes.
Fibroin
3. 18 gauge hypodermic needles.
4. Dialysis cassette 3500 MWCO, 3–12 mL capacity and dialysis
cassette buoy.
5. 50 mL conical tubes.
2.2.3 Silk Microtube 1. Silk fibroin solution (15% wt/vol).

Fabrication 2. 14, 23, 27 gauge stainless steel needles (see Subheading 4.2,
Note 2).
3. Mandrel rotation system (see Subheading 4.2, Note 3).
4. Polytetrafluoroethylene (PTFE)-coated stainless steel wire of
600 μm diameter.
5. 2 mL vials.
6. 1 mL syringes.
7. Feather scalpel.
8. Tweezers.
9. Styrofoam™.
10. Polyethylene oxide (PEO).
11. Purified human laminin.
12. Purified human type IV collagen.
13. Purified human plasma fibronectin.
14. Recombinant human SDF-1α.
15. Methanol and ethanol.
16. Lyophilizer.
18. Petri dishes (100 mm).
2.2.4 Bioreactor 1. Silk fibroin solution (7–8% wt/vol).

Assembly and Silk Sponge 2. Silk microtubes.
Production
3. Sieves (pore sizes 500–600 μm).
4. Salt (NaCl, ~500–600 μm diameter).
5. Bioreactor chamber (see Subheading 3.2.5 for further discus-
sion of suitable size and preparation).
6. Sterile glass slide for sealing the bioreactor chamber.
7. Syringe pump with syringes for perfusion of medium.
8. Blunt-end stainless steel needles (23 gauge).
9. Gas-permeable tubing for syringe-bioreactor connection.
10. Gas-permeable transfer bag with anticoagulant for platelet

collection.
11. Ethanol.
12. Serum-free medium for stem cell culture (we suggest the use
of StemSpan™ SFEM).
13. TPO.
14. Penicillin/streptomycin (P/S, 100× stock).
2.3 Hyaluronan 1. Hyaluronan.

Hydrogels 2. 50 mL conical tubes.
3. 5 M sodium hydroxide.
4. Methacrylic anhydride.
5. Cold ethanol.
6. Dialysis cassette (6–8 KDa cutoff).

8. 2-methyl-1-[4-(hydroxyethoxy)phenyl]-2-methyl-1-
propanone (photoinitiator).
9. Long-wave ultraviolet lamp.
10. Purified bovine type I collagen.
11. Purified human type IV collagen.
12. Purified human plasma fibronectin.
13. Sterile syringes (1 mL).
14. 24-well sterile culture plates.
15. Serum-free medium for stem cell culture (we suggest the use
of StemSpan™ SFEM.).
16. TPO.
17. Penicillin/streptomycin (P/S, 100× stock).
3 Methods
Carry out all procedures at room temperature, unless otherwise

specified.
3.1 Megakaryocyte 1. With a sterile syringe collect human umbilical cord blood or
Differentiation adult peripheral blood (see Subheading 4.1, Note 1) and put it
in 50 mL sterile conical tubes.
2. Dilute blood samples 1:1 with PBS (see Subheading 4.1, Note 2).
3. Isolate mononuclear cells by density gradient through centri-
fuging for 30 min at 515 × g, at room temperature.
4. Collect the mononuclear cell layer.
5. Wash twice in PBS by centrifuging for 8 min at 395 × g at
room temperature.
6. ONLY FOR CORD BLOOD—seed cells in petri dishes in
RPMI supplemented with 1% P/S and 1% l-glutamine, at 37 °C
and 5% CO2, for 30 min (30 × 106 cells/5 mL/petri dish).
7. Harvest cells and centrifuge for 8 min at 394 × g, at room
temperature.
8. Wash once in PBS for 8 min at 394 × g, at room temperature.
9. Suspend the resulting pellet in PBS containing 2 mM EDTA
and 0.5% BSA (hereafter referred to as buffer) and incubate
with antibody against CD34 (cord blood-derived sample) or
CD45 (peripheral blood derived-sample) (according to the
manufacturer’s instruction), for 15 min (CD45) or 30 min
(CD34), at 4 °C.
10. Wash cells once in buffer for 8 min at 287 × g, at room

temperature. CD45+ or CD34+ cells are separated by the
immunomagnetic bead selection technique.
11. Centrifuge CD45+ or CD34+ cells for 8 min at 287 × g, at
room temperature, and finally culture in serum-free medium
(1 mL/106 cells), supplemented with 10 ng/mL TPO and
IL-11, 1% P/S and 1% l-glutamine, at 37 °C in a 5% CO2 fully
humidified atmosphere (see Subheading 4.1, Notes 3 and 4).
12. Change media at day 5, 8, and 11 for CD45+ (end at day 14)
or at day 3, 7, and 10 for CD34+ (end at day 13), by harvesting
the whole cell culture into a 15 mL sterile conical tube, centri-
fuging for 8 min at 287 × g and resuspending the samples in
fresh medium (see Subheading 4.1, Notes 3–5).
13. ONLY FOR CD45+-DERIVED CULTURE—at the end of
the culture, collect and centrifuge the whole sample for 8 min
at 287 × g, at room temperature. Suspend pellet in 1 mL of
medium and isolate mature megakaryocytes by sedimentation
on a 2 step bovine serum albumin (BSA) gradient (1 mL BSA
4% - 1 mL BSA 3%). After 35–40 min collect cells that
sediment on the bottom of the 15 mL sterile tube.
3.2 Silk Bone The protocol for silk fibroin extraction is designed to produce one
Marrow Tissue Model batch from 5 g of silk cocoons, however, if more material is
required, the volumes can be scaled appropriately:
3.2.1 Silk Fibroin
1. Cut Bombyx mori cocoons with titanium scissors, deworm and
Extraction
chop in order to obtain 5 g of cocoon pieces (see Subheading
4.2, Note 4).
2. Boil 2 L of ultrapure water in a glass beaker covered with alu-
minum foil and add 4.24 g of Na2CO3 until the powder is
completely dissolved in order to obtain a 0.02 M solution (see
Subheading 4.2, Notes 5–7).
3. Add the chopped cocoons to the 0.02 M Na2CO3 solution and
boil for 10 min while stirring with a spatula to promote good
dispersion of fibroin (see Subheading 4.2, Note 8).
4. Remove the silk fibroin with a spatula and discard the Na2CO3
solution (see Subheading 4.2, Note 9). Then, cool the silk
fibroin by rinsing in ultrapure cold water.
5. Rinse the silk fibroin in a beaker containing 1 L ultrapure water
for 20 min while gently stirring on a stir plate. Repeat the rins-
ing an additional two times, for a total of three rinses. Always
squeeze excess of water out of the silk fibroin.
6. After the third wash, squeeze the silk fibroin and spread it out
on a clean piece of aluminum foil in order to let it dry in a fume
hood overnight.
7. Prepare a 9.3 M LiBr solution (see Subheading 4.2, Notes 10
and 11).
8. Pack tightly 3 g of dried fibers at the bottom of a 50 mL glass

beaker and add 12 mL of 9.3 M LiBr on top (see Subheading
4.2, Note 12).
9. Let silk fibroin dissolve at 60 °C for 4 h. Once the silk fibroin
is completely dissolved, it will appear amber in color and will
be transparent. Black bits from the silkworm may be visible but
will be removed later. This solution will be highly viscous but
should not contain any intact fibers, as determined by visual
assessment.
3.2.2 Silk Fibroin Dialysis The solubilized silk solution is dialyzed against ultrapure water.
1. Hydrate dialysis cassettes in ultrapure water for a few minutes.
2. Draw out the 12 mL of the dissolved silk fibroin/LiBr solution
with a 20 mL syringe without a needle.
3. Attach a 18 gauge needle to the syringe filled with the solution.
4. Insert the whole 12 mL of the dissolved silk fibroin/LiBr
solution into the dialysis cassette having a 3500 MW cutoff.
Be sure to hold the cassette at the edges and do not touch the
membrane (see Subheading 4.2, Notes 13 and 14).
5. As the cassette becomes inflated, purge excess air by inserting
another 18 gauge needle into the other top port to allow the
air to escape.
6. Remove the needles/syringe from the cassette.
7. Put the dialysis cassette into a plastic beaker containing 2 L of
ultrapure water for 3 days and changing the water a total of
eight times. To ensure proper mixing, use a large stir bar and
place on a magnetic stir plate.
8. After the 3 days withdraw the silk fibroin solution from the
cassette with a 20 mL syringe and an 18 gauge needle. Place it
into a 50 mL conical tube and centrifuge at 3220 × g for
10 min to remove large particulates.
9. If not immediately used, resulting silk fibroin solution can be
stored at 4 °C (see Subheading 4.2, Notes 15 and 16).
10. To determine the concentration of the silk fibroin solution dry
a known volume of the solution and mass the remaining
solids:
(a) Measure the weight of a small weigh boat.
(b) Add 500 μL of the silk fibroin solution to the boat and
allow it to dry at 60 °C.
(c) Once the silk is dry, determine the weight of the silk and
divide it by 500 μL: this will yield the percentage weight
per volume.
3.2.3 Silk Fibroin By faithfully carrying out the protocol of silk fibroin extraction and
Concentration dialysis one should expect to have a 7–8% (wt/vol) silk fibroin
solution that can either be used as it is for silk sponge preparation
or it can be further concentrated to a 15% (wt/vol) solution for silk
microtube fabrication (see Subheading 4.2, Note 17).
1. Centrifuge the 7–8% silk fibroin solution at 56 °C under vac-
uum for about 3 h or until half of the volume is decreased. As
an alternative to centrifugal evaporation, it is possible to let the
silk fibroin solution dry at room temperature inside a dialysis
cassette.
2. Hydrate a dialysis cassette having a 3500 MW cutoff in ultra-
pure water for a few minutes.
3. Remove the dialysis cassette from the water and tap the bot-
tom of the cassette on a paper towel to dry.
4. With a 10 mL syringe and an 18 gauge needle slowly insert the
7–8% silk fibroin solution into the dialysis cassette.
5. Leave the dialysis cassette at room temperature until half the
volume is lost. The concentration process may take ~24 h (see
Subheading 4.2, Note 18).
6. Remove the concentrated silk fibroin solution from the dialysis
cassette. The resulting silk solution should be very viscous and
will appear slightly cloudy when compared with the starting
solution (see Subheading 4.2, Note 19).
7. Measure the solution concentration by weighing out a small
weigh boat and then adding 100 μL of concentrated silk
solution to the boat. Because of the high viscosity of concen-
trated silk, we suggest to cut off the tip of the pipette. Dry the
solution at 60 °C and then weigh the dried film. The weight of
the silk divided by the volume used will yield the weight per
volume percent.
8. If the silk is collected too early, the concentration may be
still low, but the solution can be placed again into the dialy-
sis cassette.
3.2.4 Silk Microtube The concentrated silk fibroin can now be used to prepare silk
Fabrication microtubes by either a simple dip method to create thin-walled
tubes or by gel spinning, wherein the silk fibroin is extruded onto
a rotating mandrel.
Dip Method 1. In order to create a highly porous silk matrix add 6% (wt/vol)
polyethylene oxide (PEO) to the 15% silk fibroin solution to a
volume ratio of 10:1 silk:PEO and gently stir for at least 2 h to
allow homogeneous mixing of the two solutions.
2. Add fibronectin, type IV collagen, and laminin (25 μg/mL
each final concentration) and SDF-1α (300 ng/mL final con-
centration) to the resulting silk fibroin/PEO solution and mix

thoroughly for a few minutes.
3. Dip a stainless steel 23 gauge needle into a 2 mL vial contain-
ing the resulting solution. Remove the needle from the solu-
tion and flip it up and down to allow the beads of the silk
solution to evenly coat the outside of the needle along its
entire length. Let excess silk beads drip off of the needle tip
back into silk solution.
4. Place the needle in a vial of pure methanol for 5–10 s, then
remove from methanol and allow to dry for 30–60 s.
5. Repeat the dipping/coating process and methanol treatment
2–3 times so that the needle is sufficiently coated with the
silk solution. The silk may appear irregular, but it will shrink
as it dries.
6. Dry the tube by sticking the inverted silk-coated needle into
Styrofoam™ and place it at room temperature for 2 h.
7. Put the dried needle into a solution of PBS and soak for 1 h.
8. After soaking, remove the tubes from the PBS and cut the end
of the silk tube with a scalpel to yield an open tube at the end.
9. Remove the tube from the needle using tweezers. The tube
should slide off easily. If it is difficult to remove without
scrunching the tube, then place the tube back into PBS and
wait before trying again.
10. Fill a plastic Petri dish (100 mm) with PBS. Place the tube in
the Petri dish and put on a shaker for 24 h to remove residual
PEO.
11. Silk microtubes can be sterilized in 70% ethanol.
Gel Spinning 1. Repeat steps 1 and 2 as per Dip method.

2. Place a polytetrafluoroethylene (PTFE)-coated stainless steel
wire of 600 μm diameter in the mandrel rotation system.
3. With a 1 mL syringe and a 14 gauge needle slowly draw up the
concentrated silk fibroin solution, trying not to include air
bubbles. After the syringe is full, replace the 14 gauge needle
with a 27 gauge needle.
4. Start the mandrel rotation system. Set the mandrel rotation
and linear horizontal motion to 200 rpm and 1 mm/s respec-
tively, to obtain a continuous coating in a single pass.
5. Extrude the final solution through the 27 gauge needle onto
the wire. The fiber should be uniform without any beads or
discontinuities. To obtain a uniform fiber, a generous amount
of pressure is required. The combination of high viscosity and
small gauge needle causes shear-induced gelation of the solu-
tion during extrusion, stabilizing the deposited structure.
6. Once the solution has been laid onto the wire, rapidly freeze it
to −20 °C for 24 h and then lyophilize.
7. After lyophilization, apply methanol for 60 min in order to
transform the amorphous structure of silk into its β-form silk
fibroin conformation.
8. Allow the methanol to dry, and then place the mandrel into a
solution of PBS. When the tube has softened, grab the tube
uniformly and gently pull it off. If the tube does not readily
move, continue to soak it in the PBS.
9. Fill a plastic petri dish (100 mm) with PBS. Place the tube in
the petri dish and put on a shaker for 24 h to remove residual
PEO.
10. Silk microtubes can be sterilized in 70% ethanol.
3.2.5 Bioreactor A previously reported bioreactor can be adapted to re-create the

Assembly and Silk Sponge characteristic features of the bone marrow environment [24]. We
Production suggest the use of a bioreactor consisting of two wells
(15 × 20 × 5 mm), having central holes for connection to external
perfusion system, within a polydimethylsiloxane (PDMS) block
(35 × 80 × 5 mm), plasma bonded to a cover glass (see Subheading
4.2, Notes 20 and 21) (Fig. 1). However, bioreactors of different
sizes can be easily adaptable to the same protocol by changing the
length of the silk microtubes inserted within the system.
1. Place two 23 gauge blunt-ended stainless steel needles on
either side of the bioreactor chamber, 50 μm from the bottom
edge of the bioreactor.
2. Trim two silk microtubes to approximately 1.5 cm in length
and secure them over the blunt end needles within the perfu-
sion bioreactor chamber.
3. In order to reduce the volume of the well around the silk
microtubes insert two PDMS blocks (15 × 7 × 5 mm) on
either side of the bioreactor chamber.
4. Prepare the salt (NaCl) with the particle sizes of 500–600 μm
through sifting: metal wire-cloth sieves of the desired mesh
size are stacked, with the larger mesh sieve on top. The salt
particles are filtered by manual shaking, until enough salt is
collected between the two sieves (see Subheading 4.2, Note
22).
5. Repeat until the desired amount of salt is collected.
6. Weigh 1 g of salt per chamber.
7. Dispense 500 μL/chamber of 8% silk fibroin solution around
the silk microtube.
8. Slowly sift 1 g of salt on top of the fibroin solution into each
of the two chambers of the bioreactor.
Fig. 1 Schematic representation of silk-based bone marrow model preparation
9. Tap the bioreactor gently on the bench top to remove air bub-
bles and to allow salt mixing with the whole silk fibroin
solution.
10. Place the scaffold at room temperature for 2 days: the silk
fibroin should gel in 48 h maximum.
11. Once the silk has gelled, soak the bioreactor into a plastic bea-
ker containing 2 L of ultrapure water for 2 days to leach out
the salt particles, while stirring and changing water a total of
six times.
12. Sterilize in 70% ethanol for 24 h and then rinsed five times in
PBS over 24 h at 4 °C (see Subheading 4.2, Notes 23 and 24).
13. Harvest 5 × 105 megakaryocytes at the end of culture and
centrifuge for 8 min at 287 × g, at room temperature.
14. Resuspend cells in 500 μL of fresh culture medium supple-
mented with 10 ng/mL TPO, 1% P/S and 1% l-glutamine.
15. Pipette 250 μL of sample in each sponge surrounding the silk

microtubes and incubate at 37 °C in a 5% CO2 fully humidified
atmosphere.
16. After 24 h seal the bioreactor chamber with a sterile glass slide.
17. Connect inlet needles to tubing and media-filled syringes

placed onto a syringe pump.
18. Connect outlet needles to transfer bags for platelet collection
containing anticoagulant.
19. Place the bioreactor into the incubator (37 °C and 5% CO2)
and start perfusion (see Subheading 4.2, Note 25).
3.3 Hyaluronan 1. Dissolve 20 mg hyaluronan in 2 mL ultrapure water.

Hydrogels 2. Divide the 2 mL into four 50 mL conical tubes (each contain-
3.3.1 Methacrylated ing 0.5 mL of the hyaluronan solution).
Hyaluronan (MeHA) 3. Add 20 equivalents of methacrylic anhydride (corresponding
Synthesis to 10 mL final volume); add in increments of 1 mL, with
addition of enough 5 M NaOH between each increment to
maintain a pH of 8 (see Subheading 4.3, Note 1).
4. Stir the solution for 2 h at room temperature and then place
the tubes for 16 h at 4 °C.
5. Precipitate the MeHA with cold ethanol 3 times (for 16 h
each), at 4 °C, to remove the excess methacrylic anhydride.
6. Lyophilize the final precipitate.
7. Resuspend each pellet with 0.5 mL of ultrapure water.
8. Combine the four dissolved hyaluronan solutions.
9. Dialyze the resulting solution for 48 h at 4 °C, away from light
(see Subheading 4.3, Note 2). Dialysis water should be
changed every 12 h.
10. Lyophilize the final solution of MeHA.
11. Store at −20 °C.
3.3.2 Hydrogel 1. Prepare a 0.5% solution of 2-methyl-1-[4-(hydroxyethoxy)

Fabrication phenyl]-2-methyl-1-propanone (photoinitiator) in PBS
(0.05 g in 10 mL of PBS), filtrate and keep it at 37 °C for
3 days light (see Subheading 4.3, Note 3).
2. In order to prepare five hydrogels, weigh 5 mg of MeHA.
3. Cut off the tips of five 1 mL syringes.
4. Sterilize the MeHA and the syringes under UV for 30 min.
5. Dilute the photoinitiator solution to 0.05% in PBS (0.025 mL
into 0.25 mL of PBS).
6. Resuspend the 5 mg of MeHA into 0.25 mL of 0.05% photo-
initiator solution light (see Subheading 4.3, Note 4).
7. To obtain hydrogel containing extracellular matrix (ECM)

proteins, add 0.005 mL of 1 mg/mL ECM protein solution
to 0.045 mL of MeHA solution in a 1.5 mL sterile tube,
before adding the photoinitiator.
8. Turn the long-wave ultraviolet lamp on and let it warm for
10 min.
9. Pellet megakaryocytes (0.5 × 106/hydrogel).
10. Resuspend megakaryocytes in the 0.05 mL of MeHA or

MeHA plus ECM solution by gently pipetting and avoiding
formation of bubbles (see Subheading 4.3, Note 5). Transfer
to the top of the syringe tip mold held vertically.
11. Expose under the long-wave ultraviolet lamp for 25 min (see
Subheading 4.3, Note 6).
12. Hydrogels are finally plunged into a 24-well culture plate
(1 hydrogel/well) containing 1 mL of culture medium supple-
mented with 10 ng/mL TPO, 1% P/S, and 1% l-glutamine.
4 Notes
4.1 Megakaryocyte 1. Human blood samples should be collected using standard

Differentiation phlebotomy techniques by approved staff into acid-citrate-
dextrose (ACD)-containing tubes/bags. Human cord blood
must be collected following normal pregnancies. All human
samples should be collected after written informed consent
and processed in accordance with the ethical committee of the
local Institution and the principles of the Declaration of
Helsinki.
2. Use PBS without calcium and magnesium.
3. Always keep all media and reagents at 37 °C.
4. Avoid l-glutamine precipitation by intense vortexing before
addition to culture medium.
5. Gently pipette megakaryocytes at the end of differentiation.
4.2 Silk Bone 1. Not-living high-quality cocoons for medical research can be
Marrow Tissue Model purchased by different suppliers worldwide. Be sure that the
cocoons are naturally grown and not treated with any
chemicals.
2. A custom-assembled spinning device can be used [25]. The
system consists of a two-axis stepper motor that allows simul-
taneous rotational and axial drive. Motor rotational and axial
speed can be controlled through a LabView coded software
interface (National Instruments, Austin, TX). Teflon-coated
stainless steel rods of the desired diameter can be coupled
to the motor shaft through custom made adapters and are
stabilized by Teflon guides, which allow free rotation and slid-

ing of the mandrel, while limiting wobbling. The concentrated
silk solution is extruded on the rotating mandrel through a
syringe mounted on a 3-axis micromanipulator, which allows
precise positioning of the needle tip. Silk flow can be adjusted
using different gauge needles (25–30G; we suggest the use of
27G) in order to achieve controlled coating thickness on the
rotating mandrel.
3. The length of the 23 gauge needle should be 25–30 mm.
4. Only cocoons that look undamaged should be used.
5. Do not leave the beaker unattended while heating and boiling.
6. Because of high temperatures, plastic beakers should not
be used.
7. Add sodium carbonate slowly to avoid boiling over.
8. Boiling time must be precise to ensure reproducibility: increas-
ing boiling time will degrade the fibroin. Degummed silk
fibroin can be stored indefinitely at room temperature. For
long-term storage, place it in a clean plastic bag or wrap it in
aluminum foil. Be sure to indicate the length of the boiling
step on the label.
9. After boiling, the silk fibroin and solution will be hot: use
hand protectors.
10. Adding LiBr to water results in an exothermic reaction; there-
fore, when preparing large volumes, we recommend carrying
this out on ice.
11. LiBr has a low density and its volume should be taken into
account while preparing the solution. We suggest adding only
60% of the calculated volume of water and then bringing the
solution up to the final volume.
12. The LiBr must be added to the silk rather than adding silk to
the LiBr so that the silk will eventually be covered and dis-
solved by the LiBr.
13. Be careful not to puncture or touch the dialysis cassette

membrane.
14. After dissolving in LiBr, the silk fibroin solution will be very
viscous: the injection of this solution into the cassette will be
easier if the solution is kept warm. It is important to avoid
shearing the solution whenever possible to avoid the induction
of β-sheet structures within the silk solution.
15. A batch of 5 g of silk cocoons generally yields 25 mL of 7–8%
(wt/vol) silk solution. The solution will be tinted yellow but
should be relatively clear and slightly more viscous than water.
If there are impurities such as white flocculents or dark par-
ticulates, it is best to recentrifuge the solution to remove them.
16. The silk solution can be stored at 4 °C for at least a month.
Depending on the purity, stored silk will eventually gel. Once
the silk has gelled, it cannot be used for protocols that require
solution and therefore another batch will need to be extracted.
17. The amount of time used to concentrate the solution may
need to be altered for each batch.
18. The concentration time is not linear, so care must be taken to
avoid gelling the silk in the cassette.
19. The concentrated silk solution can gel if stored for too long
(approximately 4 weeks), therefore we recommend to concen-
trate silk when it is intended to be used relatively soon.
20. The bioreactor chamber can be autoclaved before usage.
21. The size and shape of the silk sponges can be varied by chang-
ing the bioreactor chamber.
22. The pore size of silk sponges will be slightly smaller than the
salt particles used, as the salt is partially dissolved.
23. If needed, silk sponges can be easily cut to different dimen-
sions with a scalpel.
24. Store the silk-based scaffolds in ultrapure water at 4 °C until
needed.
25. In order to allow platelet collection, we suggest perfusion of
medium for minimum 6 h at a physiologic shear rate of 60 s−1
[11]. However, this can be varied based on the purpose of the
research. Based on the desired share rate, the volume of
medium to be perfused can be calculated based on the follow-
ing equation:
γ = 4Q / π r 3
where
γ = Shear rate measured in reciprocal seconds
Q = Volumetric flow rate (μL)
r = Inner pipe radius (μm)
4.3 Hyaluronan 1. The pH of the methacrylic anhydride must be kept at eight.

Hydrogels 2. The dialysis tubes must be boiled in water before usage.
3. The photoinitiator solution is stable for 2 months if stored at
room temperature away from light.
4. Always keep the photoinitiator solution away from light.
5. No supernatant must remain after megakaryocyte centrifuga-
tion, before mixing with MeHA plus ECM solution, as it may
interfere with hydrogel polymerization.
6. Use UV protection gloves.
Acknowledgments
This work presented in this chapter was supported by Cariplo

Foundation (2010-0807, 2013-0717) and US National Institutes
of Health (R01 EB016041-01). Christian A. Di Buduo fellowship
was funded by Collegio Ghislieri, Pavia, progetto “Progressi in
Biologia e Medicina.” The funders had no role in study design,
data collection and analysis, decision to publish, or preparation of
the manuscript.
References
1. Morrison SJ, Scadden DT (2014) The bone Vercellino M, Visai L, Iurlo A, Moratti R,
marrow niche for haematopoietic stem cells. Barosi G, Rosti V, Balduini A (2016)
Nature 505(7483):327–334. https://doi. Thrombopoietin/TGF-β1 loop regulates
org/10.1038/nature12984 megakaryocyte extracellular matrix component
2. Wang LD, Wagers AJ (2011) Dynamic niches synthesis. Stem Cells 34(4):1123–1133.
in the origination and differentiation of haema- https://doi.org/10.1002/stem.2285
topoietic stem cells. Nat Rev Mol Cell Biol 9. Balduini A, Di Buduo CA, Kaplan DL (2016)
12(10):643–655. https://doi.org/10.1038/ Translational approaches to functional platelet
nrm3184 production ex vivo. Thromb Haemost
3. Ivanovska IL, Shin JW, Swift J, Discher DE 115(2):250–256. https://doi.org/10.1160/
(2015) Stem cell mechanobiology: diverse les- TH15-07-0570
sons from bone marrow. Trends Cell Biol 10. Di Buduo CA, Kaplan DL, Balduini A (2017)
25(9):523–532. https://doi.org/10.1016/j. In vitro generation of platelets: where do we
tcb.2015.04.003 stand? Transfus Clin Biol doi:https://doi.
4. Malara A, Abbonante V, Di Buduo CA, Tozzi L, org/10.1016/j.tracli.2017.06.013
Currao M, Balduini A (2015) The secret life of a 11. Di Buduo CA, Wray LS, Tozzi L, Malara A,
megakaryocyte: emerging roles in bone marrow Chen Y, Ghezzi CE, Smoot D, Sfara C,
homeostasis control. Cell Mol Life Sci. https:// Antonelli A, Spedden E, Bruni G, Staii C, De
doi.org/10.1007/s00018-014-1813-y Marco L, Magnani M, Kaplan DL, Balduini A
5. Aguilar A, Pertuy F, Eckly A, Strassel C, Collin (2015) Programmable 3D silk bone marrow
D, Gachet C, Lanza F, Léon C (2016) niche for platelet generation ex vivo and mod-
Importance of environmental stiffness for eling of megakaryopoiesis pathologies. Blood
megakaryocyte differentiation and proplatelet 125(14):2254–2264. https://doi.
formation. Blood 128(16):2022–2032. org/10.1182/blood-2014-08-595561
https://doi.org/10.1182/ 12. Di Buduo CA, Currao M, Pecci A, Kaplan DL,
blood-2016-02-699959 Balduini CL, Balduini A (2016) Revealing
6. Malara A, Gruppi C, Pallotta I, Spedden E, Eltrombopag's promotion of human mega-
Tenni R, Raspanti M, Kaplan D, Tira ME, Staii karyopoiesis through AKT/ERK-dependent
C, Balduini A (2011) Extracellular matrix pathway activation. Haematologica. https://
structure and nano- mechanics determine doi.org/10.3324/haematol.2016.146746
megakaryocyte function. Blood 118(16):4449– 13. Di Buduo CA, Soprano PM, Tozzi L, Marconi
4453. https://doi.org/10.1182/ S, Auricchio F, Kaplan DL, Balduini A (2017)
blood-2011-04-345876 Modular flow chamber for engineering bone
7. Abbonante V, Di Buduo CA, Gruppi C, De marrow architecture and function. Biomaterials
Maria C, Spedden E, De Acutis A, Staii C, 146:60–71. https://doi.org/10.1016/j.
Raspanti M, Vozzi G, Kaplan D, Moccia F, biomaterials.2017.08.006
Ravid K, Balduini A (2017) A new path to 14. Omenetto FG, Kaplan DL (2010) New oppor-
platelet production through matrix sensing. tunities for an ancient material. Science
Haematologica. https://doi.org/10.3324/ 329(5991):528–531. https://doi.
haematol.2016.161562 org/10.1126/science.1188936
8. Abbonante V, Di Buduo CA, Gruppi C, Malara 15. Kluge JA, Li AB, Kahn BT, Michaud DS,
A, Gianelli U, Celesti G, Anselmo A, Laghi L, Omenetto FG, Kaplan DL (2016) Silk-based
blood stabilization for diagnostics. Proc Natl Nishimura S, Eto K (2014) Expandable mega-
Acad Sci U S A 113(21):5892–5897. https:// karyocyte cell lines enable clinically applicable
doi.org/10.1073/pnas.1602493113 generation of platelets from human induced
16. Wray LS, Tsioris K, Gi ES, Omenetto FG, pluripotent stem cells. Cell Stem Cell
Kaplan DL (2013) Slowly degradable porous 14(4):535–548. https://doi.org/10.1016/j.
silk microfabricated scaffolds for vascularized stem.2014.01.011
tissue formation. Adv Funct Mater 22. Moreau T, Evans AL, Vasquez L, Tijssen MR,
23(27):3404–3412. https://doi. Yan Y, Trotter MW, Howard D, Colzani M,
org/10.1002/adfm.201202926 Arumugam M, Wu WH, Dalby A, Lampela R,
17. Lu Q, Wang X, Hu X, Cebe P, Omenetto F, Bouet G, Hobbs CM, Pask DC, Payne H,
Kaplan DL (2010) Stabilization and release of Ponomaryov T, Brill A, Soranzo N, Ouwehand
enzymes from silk films. Macromol Biosci WH, Pedersen RA, Ghevaert C (2016) Large-
10(4):359–368. https://doi.org/10.1002/ scale production of megakaryocytes from
mabi.200900388 human pluripotent stem cells by chemically
18. Rockwood DN, Preda RC, Yücel T, Wang X, defined forward programming. Nat Commun
Lovett ML, Kaplan DL (2011) Materials fabri- 7:11208. https://doi.org/10.1038/
cation from Bombyx mori silk fibroin. Nat ncomms11208
Protoc 6(10):1612–1631. https://doi. 23. Merico V, Zuccotti M, Carpi D, Baev D,

org/10.1038/nprot.2011.379 Mulas F, Sacchi L, Bellazzi R, Pastorelli R,
19. Currao M, Malara A, Di Buduo CA, Abbonante Redi CA, Moratti R, Garagna S, Balduini A
V, Tozzi L, Balduini A (2015) Hyaluronan (2012) The genomic and proteomic blue-
based hydrogels provide an improved model to print of mouse megakaryocytes derived from
study megakaryocyte-matrix interactions. Exp embryonic stem cells. J Thromb Haemost
Cell Res doi:https://doi.org/10.1016/j. 10(5):907–915. https://doi.org/10.1111/
yexcr.2015.05.014 j.1538-7836.2012.04673.x
20. Di Buduo CA, Alberelli MA, Glembostky AC, 24. Lovett M, Cannizzaro C, Daheron L, Messmer
Podda G, Lev PR, Cattaneo M, Landolfi R, B, Vunjak-Novakovic G, Kaplan DL (2007)
Heller PG, Balduini A, De Candia E (2016) Silk fibroin microtubes for blood vessel engi-
Abnormal proplatelet formation and emperipole- neering. Biomaterials 28(35):5271–5279.
sis in cultured human megakaryocytes from gray https://doi.org/10.1016/j.
platelet syndrome patients. Sci Rep 6:23213. biomaterials.2007.08.008
https://doi.org/10.1038/srep23213 25. Lovett ML, Cannizzaro CM, Vunjak-
21. Nakamura S, Takayama N, Hirata S, Seo H, Novakovic G, Kaplan DL (2008) Gel spinning
Endo H, Ochi K, Fujita K, Koike T, Harimoto of silk tubes for tissue engineering. Biomaterials
K, Dohda T, Watanabe A, Okita K, Takahashi 29(35):4650–4657. https://doi.
N, Sawaguchi A, Yamanaka S, Nakauchi H, org/10.1016/j.biomaterials.2008.08.025
Chapter 12
Fluorescence Approaches to Image and Quantify

the Demarcation Membrane System in Living
Megakaryocytes
Sangar Osman, Daniel Dalmay, and Martyn Mahaut-Smith
Abstract
The demarcation membrane system (DMS) develops to provide additional surface membrane for the
process of platelet production. The DMS is an invagination of the plasma membrane that can extend
throughout the extranuclear volume of mature megakaryocytes and its lumen is continuous with the
extracellular solution. DMS ultrastructure in fixed samples has been extensively studied using transmis-
sion electron microscopy (TEM) and more recently with focused ion beam scanning EM. In addition,
whole cell patch clamp membrane capacitance provides a direct measurement of DMS content in living
megakaryocytes. However, fluorescence methods to image and quantify the DMS in living megakaryo-
cytes provide several advantages. For example, confocal fluorescence microscopy is easier to use compared
to EM or electrophysiological methods and the required equipment is more readily available. In addition,
use of living cells avoids artifacts known to occur during the fixation, dehydration, or embedding steps
used to prepare EM samples. Here we describe the use of styryl dyes such as FM 1–43 or di-8-ANEPPS
and impermeant fluorescent indicators of the extracellular space as simple approaches for imaging and
quantification of the DMS.
Key words Megakaryocytes, Demarcation membrane system, Styryl dyes, Megakaryopoiesis
1 Introduction
In addition to the generation of specialist proteins and organelles

required for platelet function, the megakaryocyte expands its sur-
face membrane throughout the extranuclear volume to generate a
large reserve of membrane in support of thrombopoiesis. In the
mid-1950s two groups first described this as a system of mem-
branes only present in megakaryocytes, within studies of human
marrow [1, 2] and mouse spleen [3, 4]. These original reports and
Electronic supplementary material: The online version of this chapter (doi:10.1007/978-1-4939-8585-2_12)

contains supplementary material, which is available to authorized users.
195
196 Sangar Osman et al.
subsequent studies by a number of groups [5–7] raised the

possibility that this unique membrane complex outlines preformed
areas of future “platelet units,” and hence the term “demarcation
membrane system” (DMS). However, as alluded to in Wright’s
original description of megakaryocytes as the origin of platelets [8,
9], it is now recognized that the main physiological mechanism of
platelet production involves the generation of proplatelets rather
than fragmentation into preformed platelet territories [10–12].
Proplatelets are long pseudopodia-like structures that extend from
the megakaryocyte into the venous sinusoids, sections of which
split off into the circulation under the influence of shear [11–13].
The term DMS is therefore somewhat of a misnomer [14],
although remains the normal phrase used to refer to this extensive
surface-connected membrane system. Some authors have proposed
the alternative the term “invaginated membrane system” (IMS)
[14, 15]. The “fragmentation” theory of platelet production, in
which the DMS marks out areas of platelets, is considered implau-
sible due to the lack of key structural characteristics such as micro-
tubule coils within the future hypothetical platelet zones [14, 16].
However, the generation of platelets without the need for normal
proplatelet formation has recently been proposed during inflam-
mation in response to an enhanced requirement for thrombopoi-
esis [17]. Exactly how the cytoplasmic contents and the DMS are
reorganized during such conditions of emergency platelet produc-
tion is at present unclear.
Transmission electron microscopy (TEM) of thin sections from
resin-embedded samples represents the main technique used to
study organization and development of the DMS [18, 19]. Although
this high resolution imaging approach provides important informa-
tion on megakaryocyte ultrastructure, it is time-consuming and may
be subject to artifacts arising from the chemical fixation, dehydration
or embedding processes used during sample preparation [20, 21]. It
is possible to quantify the level of DMS using image processing of
TEM sections [19, 22]; however, the standard grid bars employed in
TEM studies often obscure part of the megakaryocyte. In addition,
it is difficult and time-consuming to generate and analyse serial thin
sections to cover the entire cell. Recent development of volume EM
approaches (e.g., serial block face EM and focused ion beam scan-
ning EM) are overcoming some of these limitations [23–25],
although these approaches are highly labour-intensive and also reli-
ant upon expensive, rare items of equipment.
One technique that can quantify the DMS in living megakaryo-
cytes is the measurement of whole cell capacitance within patch
clamp recordings [26]. This approach has been described in detail in
a previous chapter within this series of methods books [27], and thus
will not be repeated here. While this probably represents the most
reliable technique to accurately quantify the DMS in a living mega-
karyocyte, it requires specialist electrophysiological equipment and
Fluorescence Imaging and Quantification of the DMS 197
knowledge. In addition, although less time-consuming than EM

approaches, it still has a limited capacity (≈20–30 cells per day in
expert hands).
The present chapter describes the use of two simple fluores-
cence methods for assessing the extent of DMS development in
living megakaryocytes using confocal fluorescence imaging. In the
first, styryl dyes added to the extracellular saline are used to stain
the plasma membrane and connected DMS [26, 28]. Due to their
lipophilic nature, these dyes readily insert into membranes acces-
sible from the external saline and in this lipid environment exhibit
a strong fluorescent signal in contrast to a very low fluorescence
within an aqueous environment [29–31]. Most styryl dyes are
only slowly permeable across a lipid bilayer, thus in an intact cell
with low rates of endocytosis, they serve as selective stains of any
surface membranes and their invaginations. In the second tech-
nique, small molecular weight impermeant fluorescent indicators
are used to image the tubules or sheets formed by the DMS that
are continuous with the extracellular space. Single [28] or 2-pho-
ton [26] confocal imaging can be used to assess the quantity of
dye within the DMS lumen. In theory, either the stryryl dye or
extracellular impermeant indicator techniques could be used to
quantify the amount of DMS, although only the latter approach
has been used to date for this purpose [28]. We also discuss
molecular reporter and antibody-based methods previously
described in the literature to label proteins or lipids located pre-
dominantly in the surface membrane which can serve as further
methodologies to fluorescently label the DMS.
2 Materials
1. Cells from a megakaryocyte culture system (see Note 1) or

marrow from the animal model of choice (see Note 2)
2. Dissection instruments: forceps, small scissors and bone cut-
ters for preparation of ex vivo megakaryocytes
3. Equipment for extracting marrow from long bones. This con-
sists of a 1–5 ml syringe, a yellow tip cut to fit on the end of the
syringe and a short length (e.g., 10 mm) of 1–3 mm diameter
silicone tubing. For rat bone marrow, a dental excavator may
also be used
4. Petri dish (10 mm for mouse; 35 mm for rat)
5. Pseudophysiological saline (PPS): 145 mM NaCl, 5 mM KCl,
1 mM MgCl2, 1 mM CaCl2, 10 mM glucose, and 10 mM
HEPES (pH 7.35, titrated with NaOH). Alternatively, stan-
dard phosphate buffered saline can be used.
6. Microfuge tubes and a device that can gently rotate the tubes
at approx 0.2 Hz.
7. Plastic transfer pipettes.

8. Styryl membrane indicator such as di-8-ANEPPS or FM 1–43:
1–10 mM in appropriate vehicle (see Notes 3, 4 and 5).
9. A low molecular weight extracellular indicator: examples are
the Na+ salts of a Ca2+ indicator such as Oregon Green 488
BAPTA-1, Na+ salt of calcein, HPTS, FITC-tagged low molec-
ular weight dextran (see Note 6).
10. (Optional) Hoechst 33342, a fluorescent indicator of double
stranded (ds) DNA (see Note 7): stock 1 mg/ml in water or
saline.
11. Confocal fluorescence imaging system, with high numerical
aperture, high magnification (e.g., 40× or 60×) objective lenses
on an inverted microscope. Most styryl dyes can be excited with
the standard 488 nm laser line since this is near their peak
absorption (e.g., 479 nm for FM 1–43). However, they vary in
their Stokes shift and thus emission wavelengths, therefore
check the spectrum on the manufacturer’s site. For FM 1–43,
the emission maximum is 598 nm and the emission bandwidth
is quite wide so we use the maximum available bandwidth on
this channel, which on our Olympus FV1000 confocal is
100 nm, thus we set this at 550–650 nm. If using multiple indi-
cators (e.g., Hoechst or GFP-tagged protein and an FM dye,
the emission bandwidth for each channel may need to be
reduced or shifted to avoid crossover of individual dye signals
between channels. This is easier to set on a confocal where the
emission wavelengths are selected with spectral detectors rather
than filters. The confocal must also have the appropriate mirrors
beneath the objective to reflect the excitation wavelength(s)
toward the cell and transmit the longer wavelength emission
signal to the detectors. It is also important to collect the trans-
mitted signal to provide an image of cell morphology alongside
the fluorescence images.
12. Small volume (100–1000 μl) imaging chamber with a base
formed from a no. 1.5 glass coverslip (see Note 8). Suitable
systems are available from a number of companies such as
Warner Instruments (Hamden, CT, USA) or Bioptechs (e.g.,
the Interchangeable Coverslip Dish from Bioptechs Inc., Butler,
PA, USA). This should be held securely on the microscope
stage. A chamber perfusion system is not necessary in most situ-
ations (and the styryl dyes are expensive). For a static bath, dyes,
agonists, inhibitors etc. can be normally added directly by dilu-
tion. However, perfusion can be useful in some experiments for
washing away the dye to assess internalization.
13. (Optional) Microscope temperature regulating system. We

have had good success with the Biotechs Stable Z system
(Bioptechs Inc., Butler, PA, USA) which accommodates their
circular Interchangeable Coverslip Dish. An objective heater
can also be used to reduce the temperature gradient in the

chamber due to heat sinking by an immersion objective.
14. Image analysis software such as Image J (Rasband, W.S.,

ImageJ, US National Institutes of Health, Bethesda, Maryland,
USA, https://imagej.nih.gov/ij/, 1997–2016), Huygens
software (Scientific Volume Imaging, the Netherlands), or
Imaris (Bitplane).
3 Methods
3.1 Cell Preparation 1. Generate megakaryocytes via your selected culture technique
(see Note 1). Alternatively if using marrow from an animal
model, this should be isolated as soon as possible after eutha-
nasia (see Note 2). Dissect out the femoral and tibial bones
intact then use a dry tissue to remove as much of the remain-
ing attached soft tissue as possible and immerse in a Petri dish
containing PPS.
2. After approximately 15 min, wipe the bones with dry tissue to
further remove attached tissues.
3. Remove the epiphyses with a pair of bone cutters and attach a
short length of silicone tubing that fits snugly over the end of
the remaining bone. Attach the shortened yellow tip and
syringe filled with 1–2 ml PPS then force saline through the
bone cavity to flush out the marrow into a clean petri dish (see
Note 9). Use the minimum amount of saline for each bone.
For rat bones, an alternative is to cut the bones longitudinally
into a petri dish containing 1–2 ml saline and scrape the mar-
row off the bones with a dental excavator, then remove as
many bone fragments as possible with forceps.
4. Aliquot clumps of marrow and cell suspension into 1.5 ml
microfuge tubes and place on a rotor at 0.2 Hz. Initially, the
cell suspension normally contains mainly “ghostly” megakary-
ocytes in which the integrity of the plasma membrane appears
to be compromised (see step 6 below for further discussion
and Fig. 1). “Intact” megakaryocytes with healthy plasma
membranes appear 1–3 h later, possibly following migration
from the marrow clumps. Therefore, ensure that each aliquot
contains at least one clump of marrow. If the marrow does not
break up with rotation, this can be facilitated prior to an
experiment with a gentle filliping of the tube. Megakaryocytes
prepared in this way can be used for imaging of demarcation
membranes up to ≈16 h after marrow isolation.
5. Use a pipettor to evenly distribute ≈10–50 μl of marrow sus-
pension (dilution of between 1 in 100 and 1 in 20) over the
base of the imaging chamber filled with PPS. Adjust the vol-
ume of suspension added depending upon chamber volume
and density of cells.
Fig. 1 Morphology and plasma membrane integrity of megakaryocytes extracted from rat marrow. Transmitted
light images (left panels) are shown for three different megakaryocytes isolated from rat bone marrow. “Intact”
megakaryocytes with healthy plasma membranes (a and b) display sharp contrast at their periphery and it is
difficult to see the underlying nucleus. In contrast, the plasma membrane integrity of “ghostly” megakaryo-
cytes (c) is compromised, resulting in diffuse edges and clear appearance of the multilobular nucleus.
Fluorescence images are also shown following 30 min incubation with a dsDNA-specific dye; DRAQ5 in a and
c (excitation 635, emission 700-800nm) and Hoechst 33342 in b (405nm excitation, 430-530nm emission).
The type of megakaryocyte shown in c is often observed immediately after isolation of marrow cells. The fre-
quency of occurrence of the intact cell type then slowly increases with time; see main text for further discus-
sion. Hoechst used as described in the methods; DRAQ5 was made at a stock of 100 μM in PPS and a final
concentration of 1 μM. The scale bar at the bottom (15 μm) applies to all images
6. After allowing the cells to settle, screen across the chamber to

identify a megakaryocyte from its large diameter compared to
other marrow cells. For staining by styryl or external dyes, it is
important to use “intact” megakaryocytes with healthy plasma
membranes. The ability to distinguish between these two
types comes with experience. The edge of an intact mega-
karyocyte with a healthy plasma membrane will be sharply
defined and it will be difficult to distinguish the nucleus within
the cell (Fig. 1a, b). If the plasma membrane integrity is clearly
compromised, one can see the multilobular nucleus very
clearly and the edge of the cell is quite diffuse (Fig. 1c).
Furthermore after applying a styryl dye, the signal from a
megakaryocyte with a leaky plasma membrane is much stron-
ger, presumably because it readily stains all organellar mem-
branes (not shown). We are still exploring the fate of the DMS
membranes in these cells. We have also observed that DRAQ5,
a long wavelength dsDNA-specific dye, readily stains the
nucleus of “ghostly” megakaryocytes, whereas little or no
staining is observed after 30 min in intact megakaryocytes
(Fig. 1).
3.2 Staining 1. Set the confocal pinhole and image resolution (i.e., number of
of the Demarcation pixels) to the level required for your experiment (see Notes 10
Membranes Using and 11 for further discussion). Acquire images of unstained
FM 1–43 cells at several levels of excitation (i.e., laser power on a point
scanning confocal, which on the Olympus FV1000 is con-
trolled by an acousto-optical tunable filter) and a range of
emission gains. Make a note of the excitation/emission levels
at which cellular autofluorescence initially appears. Start with
settings well below this level. Ideally you should use a low level
of laser illumination, since dye excitation can release damaging
free radicals [32]. However, increasing the emission gain will
increase the background and cellular autofluorescence levels,
so the optimal settings will be a trade-off between the two
means of increasing the signal from the dye (see Note 12).
2. If carrying out simultaneous measurements of ploidy, incubate
the cells with 5 μg/ml Hoechst 33342 for approximately
30 min before addition of styryl dye (see Note 13).
3. If carrying out a time series during dye labeling, initiate this
(see Note 14).
4. Add the styryl dye at a final concentration of 5–10 μM for FM
1–43 or di-8-ANEPPS. It may be possible to use lower
concentrations depending upon your confocal. Use the lowest
concentration of dye that generates a good signal when the
excitation and emission gains are at levels that yield minimal
autofluorescence from the cell. Other styryl dyes that are more
water soluble (e.g., FM2–10, often used because it has very
fast on and off rates) may require a higher concentration; our

previous work has used 50–200 μM FM 2–10 [26]. The styryl
dye can be added directly to the chamber after the cells. The
FM dyes stain very rapidly, so it is best to avoid exposure to a
large, localized concentration of the dye prior to mixing. This
can be achieved by removing half the chamber volume after
the cells have settled, mixing the dye in the extracted solution
in a microfuge tube, then adding back to the chamber. Adding
this at the edge of the chamber with gentle mixing can achieve
even distribution of the dye without disturbing the cells.
Alternatively, dye can mixed in the chamber before the cells.
5. Healthy small cells will stain as halos while megakaryocytes stain
with different patterns throughout their extranuclear volume
(Figs. 2, 3, 4 and Supplementary online Video 1). Dead cells
will be very bright as the dye stains all internal membranes.
Staining of peripheral membranes by FM 1–43 is very rapid,
within tens of seconds, but can take a few minutes to stain the
entire DMS. Di-8-ANEPPS can take slightly longer. The advan-
tage of Di-8-ANEPPPS is that, unlike the FM dyes, the staining
is not reversed upon wash-out (see Note 15), so the dye can be
removed from the extracellular saline prior to imaging.
Fig. 2 Staining of plasma membranes and the demarcation membrane system by di-8-ANEPPS. Rat marrow
cells were incubated in 10 μM di-8-ANEPPS (<20 min) and fluorescence images acquired by confocal micros-
copy at two Z-sections, one through the centre of the small cells, and the other through the centre of a large
megakaryocyte. 488 nm excitation; >505 nm emission. Fluorescence emission intensity was pseudocoloured
using a green look up table. Scale bars: 10 μm; confocal section: 1 μm. From [26] with permission
Fig. 3 Staining of plasma membranes and the demarcation membrane system by FM 1-43. Rat marrow cells
were incubated in 10 μM FM 1-43 (15 min) and a Z-series of images collected by confocal microscopy. A
transmitted light image and fluorescence image (488 nm excitation; 550–650 nm emission) are shown at 10
positions, with an interval of 2.96 μm. Fluorescence emission intensity was pseudocoloured using a green look
up table. Confocal section: 1 μm
6. The DMS can be assessed using just an image taken at the

central focal plane of the cell or via a Z- series throughout the
cell (see Fig. 3). See Note 16 for further discussion on Z-step
settings.The biggest problem we have encountered when
using styryl dyes is phototoxic activation of the cell leading
initially to membrane blebbing and ultimately membrane per-
meabilization (see Note 17 and Fig. 5). We have not studied
exactly what causes the blebbing, but it is likely due to release
of free radicals [32]. Another problem is that the dyes are pho-
tobleached by the excitation light. For these reasons there is a
limit to the number of images that can be captured from each
styryl dye- stained cell. In particular, very high resolution
imaging, in which a large number of pixels are used to exam-
ine the DMS in detail, will normally be limited to only a few
focal planes.
Fig. 4 Simultaneous staining with a stryryl dye and Hoechst shows that surface-connected demarcation mem-
branes extend throughout the extranuclear volume of the cell. Images were collected from a rat megakaryo-
cyte following staining of both the nucleus (Hoechst, magenta) and surface-connected membranes
(di-8-ANEPPS, green). The rectangular area indicated in the merged image has been expanded (right panel, in
grayscale) to show the separation of peripheral plasma membrane and underlying membrane invaginations.
This gap was occasionally spanned by a thin band of fluorescence (arrow). Confocal section: 1 μm; Scale bars:
5 μm. From [26], with permission
7. At the end of a Z-series, acquire one or two additional sec-

tions to assess photobleaching, membrane blebbing or sub-
stantial membrane movement. If these have occurred to such
a significant extent and thus will influence quantification,
decrease the imaging resolution further and increase the con-
focal pinhole, to allow more rapid acquisition of the complete
Z-series in further experiments (see Note 16).
8. To compare the quantity of surface-connected membrane
including DMS, use ImageJ or similar software to draw an
ROI (region of interest) around the outer edge of the cell and
measure the average or total fluorescence within the ROI,
then subtract the value from a similar size ROI in the extracel-
lular environment (well away from other cells to avoid out of
focus fluorescence). For the complete Z-series, repeat the
analysis throughout the whole cell Z-series and combine for
each cell. The values can then be plotted against cell size or
volume.
Fig. 5 Membrane blebbing caused by extensive imaging of styryl dye-stained megakaryocytes. Rat marrow
cells were incubated in 10 μM FM 1-43 (15–20 min). (a) shows the typical appearance of a megakaryocyte
during an initial Z-series. After acquisition of approximately six successive Z-series, substantial peripheral
membrane blebbing was observed (b). Fluorescence emission intensity for FM 1-43 was pseudocoloured
using a green look up table in ImageJ. Scale bars: 10 μm
3.3 Staining 1. We normally premix the extracellular dye at the required final
of the Lumen concentration in saline and add to the chamber prior to addi-
of the Demarcation tion of cells (e.g., 5–10 μM Oregon Green 488 BAPTA-1 if
Membrane System using a saline with normal Ca2+ content, or 50–400 μM
with an Extracellular HPTS). Lower concentrations can be used, depending upon
Indicator the sensitivity of the confocal detectors, how close to its wave-
length of maximum absorbance the dye is being excited and
the emission collection bandwidth.
2. Add cells at a density such that when settled on the coverslip
they occupy about half the available area and thus do not
excessively overlap each other (see Fig. 6a).
Fig. 6 Single photon confocal imaging of an impermeant extracellular indicator to assess demarcation mem-
brane system development. (a) Rat marrow cells were immersed in pseudo physiological saline containing the
dye HPTS. The field of view included multiple small cells and a large megakaryocyte. A fluorescence image is
shown at two planes of focus; one through the centre of most of the small marrow cells and one through the
centre of the megakaryocyte. The dye is rejected from the small cells, but enters the lumen of the invaginations
formed by the DMS. Scale bars: 10 μm. (b). Relationship between the percentage fluorescence within the
megakaryocyte (calculated using equation 1) and the cell diameter. The solid line was the result of linear
regression analysis in Graphpad Prism (Graphpad Software, La Jolla, California); regression coefficient
r2 = 0.42, p<0.0001. From [28] with permission
3. Screen through the dish and locate a megakaryocyte whose

plasma membrane integrity is not compromised (see
Subheading 3.1, steps 4 and 6 for further discussion and
Fig. 1). Set the overall gain using a combination of excitation
and emission levels so that the signal from the extracellular
indicator is just below saturation of the detectors. Check that
saturation does not occur in the extracellular space at several
Z-positions across the depth of the cells and reduce the gain if
necessary. For unknown reasons, dead cells tend to accumu-
late dye and the signal may be higher than the extracellular
space so ignore these areas when assessing saturation and avoid
them when drawing the extracellular ROIs in step 5 below.
Fig. 7 Two-photon imaging of an impermeant extracellular indicator. (a) Rat marrow cells were immersed in
pseudo physiological saline (with 1mM Ca2+) containing the dye Oregon Green 488 BAPTA-1. The field of view
contained multiple small cells and a large megakaryocyte. Fluorescence images are shown at two planes of
focus; one through the centre of most of the small marrow cells and one through the centre of the megakaryo-
cyte. The dye is rejected from the small cells, but enters the lumen of the invaginations formed by the
DMS. Excitation: 797–800 nm; emission bandwidth was optimized using a spectrophotometer detector and
was in the range of 500–700 nm. From [26] with permission
4. For a “simple” estimate of DMS content, take an image at a

central Z-position for some small marrow cells and also each
megakaryocyte (see Fig. 6a for images acquired with single
photon excitation and Fig. 7 for images from two-photon
excitation). For an in-depth assessment of the DMS through-
out the entire volume of the megakaryocyte, acquire a Z-series
of images that spans the depth of the cells (see Note 11 for
discussion of settings)
5. Simple estimate: in ImageJ or other analysis package, measure
the average fluorescence inside each megakaryocyte using an
image taken at the mid-cell focal plane and a ROI drawn just
inside the cell circumference. Repeat at the lower focal plane
for a few small cells. Then measure the average extracellular
fluoresecence at each focal plane using an ROI drawn away
from the cells. To avoid heterogeneity in extracellular dye val-
ues, we measure the average of several extracellular ROIs
across each field of view. Finally, measure the background fluo-
rescence using the same settings in a chamber without dye. At
the excitation and emission levels normally used, the cell
autofluorescence is negligible, although it is worth checking
and adjusting gains as necessary.
6. Simple estimate: calculate the percentage dye in several small

cells and each megakaryocyte using the following equation:
FCellROI − FBKROI × 100
%Dye inside thecell =
FECFROI − FBKROI
where FCellROI is the average fluorescence for a ROI around a
megakaryocyte or a small cell, and FECFROI is the average fluo-
rescence for ROIs in the extracellular space at a mid-mega-
karyocyte or mid-small cell focal plane accordingly. FBKROI is
the average fluorescence for a ROI within an image taken
without dye.
7. Plot the percentage dye inside each cell versus cell diameter for
the megakaryocytes (Fig. 6b). In our measurements of pri-
mary megakaryocytes from rat bone marrow, there is a clear
correlation between the two parameters, indicating that the
average volume occupied by the dye inside the whole cell
increases as the cell becomes larger. This agrees with previous
electrophysiological measurements in which the membrane
capacitance normalized for cell size increases 1.4-fold per dou-
bling of spherical surface area [26]. Small cells completely
exclude the dye (Fig. 6a) although have a consistent value of
about 5% internal dye [28] probably due to out of focus
fluorescence.
8. For a more complete estimate of the DMS content throughout
the cell, repeat the measurement at multiple Z-positions and
combine for each cell (see Note 16 for discussion of settings).
The extreme lower and upper edges of the cell may need to be
excluded due to increased out of focus fluorescence.
3.4 Discussion Schulze and colleagues [33] have used a genetically encoded
of Other Fluorescence reporter of the membrane phospholipid phosphatidylinositol
Methods Used to Label 4,5-bisphosphate (PI-4,5-P2) to label the DMS in cultured living
the DMS in Living or murine megakaryocytes. The reporter consists of the pleckstrin
Fixed Megakaryocytes homology domain of PLC∂1 fused to EGFP and shows a pre-
dominant peripheral location at rest in a variety of other cell types
3.4.1 PIP2 Reporter
[34, 35]. This distribution suggests that the majority of cellular
PI-4,5-P2 is located in the plasma membrane. In the mature mega-
karyocyte, the sensor shows the distribution expected for expres-
sion throughout the DMS, a conclusion supported by
colocalization with an anti-CD41 antibody [33]. Furthermore, in
mature megakaryocytes, PH-PLC∂1-EGFP staining of the periph-
eral surface membrane is weak, indicating that the DMS may
become loaded with PI-4,5-P2. It should be noted however that
the PH-PLC∂1-EGFP has a higher affinity for IP3 than PI-4,5-P2,
and thus t ranslocates from the plasma membrane following stimu-
lation of receptors coupled to phospholipase C activation [36].
3.4.2 Fluorescent In theory any transmembrane protein which is predominantly

Tagging of a Plasma expressed in the surface membrane could serve as a target to label
Membrane Protein the DMS. It should however be noted that the extent to which
proteins are evenly distributed throughout the DMS is not clear.
Antibodies raised against an extracellular epitope may circumvent
the need to use fixation/permeabilization and thus allow for use
on living megakaryocytes. However, one unrecognized problem is
limited access to the DMS lumen from the external space. In our
experience, large proteins the size of an antibody can fail to enter
the tubules of the DMS (Osman and Mahaut-Smith unpublished
observations). The exact cut-off size of this restricted access is cur-
rently under investigation in our laboratory. A useful antibody for
murine studies is the FITC (or Alexa Fluor™) tagged rat anti-
mouse anti-CD41 monoclonal antibody (clone MWReg30; BD
Pharmingen™, BD Biosciences) [37] since it binds to an extracel-
lular epitope of CD41 (integrin αIIb) and thus labels live mega-
karyocytes [38]. Although we have observed that this antibody
extensively stains the extranuclear volume of many mature mega-
karyocytes as expected for DMS distribution of CD41, we have
also seen a halo-like staining in some megakaryocytes that may
indicate the antibody fails to enter the DMS tubules ([38] and
unpublished observations of Sangar Osman and Martyn Mahaut-
Smith). Further experiments are required to investigate this issue.
Fluorescently tagged antibodies against GPIbα or GPIbβ have
been used as labels of the DMS in several studies [24, 39–42]. In
pulse chase experiments using an antibody raised against an extra-
cellular epitope, immature megakaryocytic progrenitor cells label
as halos, then in culture the stain becomes progressively more
internal with a pattern that correlates with that expected for the
DMS [24]. Another group has also used GPIbα antibodies and
superresolution to assess DMS distribution [41]. Most of these
experiments with GPIb antibodies have used fixation and permea-
bilization and whether this is related to the restricted access phe-
nomenon described above is unclear.
Zhang and colleagues have generated a transgenic mouse in
which expression of farnesylated YFP occurs in place of CD41 for
one CD41 locus [33, 43]. The farnesylation attaches the fluores-
cent reporter to the plasma membrane and thus can also serve as an
indicator of the DMS, as suggested by its presence throughout the
extranuclear volume of mature megakaryocytes, but lacking colo-
calization with an antibody against a cytoskeletal and cytoplasmic
marker (Arp3) [33]. Although this demonstrates the power of such
an approach, a proportion of megakaryocytes (and platelets) in the
knock-in model are not YFP-positive for unknown reasons [43].
Zhang and colleagues also report that a substantial amount of the
YFP does not colocalize with CD41, which may reflect intracellu-
lar stores of YFP rather than a DMS location [43].
4 Notes
1. A number of culture systems support the development of

large, polyploidic megakaryocytes (e.g., [44–46] or Chapters
9, 10 and 11, this volume). We have used styryl dyes to stain
the DMS of large megakaryocytes following 5 days culture of
murine femoral marrow cells with thrombopoietin. The pat-
terns of staining of these megakaryocytes are heterogeneous
compared to similar size cells freshly isolated from marrow,
with several staining as simple halos (Gwen Tolhurst and
Martyn Mahaut-Smith, unpublished observations). This sug-
gests that the DMS may not be as extensive in some mega-
karyocytes that grow in standard culture systems compared to
those that develop in vivo. Differences between the DMS of
cultured versus marrow-derived megakaryocytes has also been
highlighted by Aguilar and colleagues who ascribe this at least
in part to the greater stiffness of marrow compared to normal
culture medium [39] that can be replicated with methylcellu-
lose (see Chapter 9, this volume).
2. Rat and mouse long bones (femurs or tibia) can be used to
generate freshly isolated megakaryocytes. Any work with ani-
mal models must only be carried out by authorized staff with
appropriate ethical permissions.
3. For all fluorescent indicators, protect from excessive light prior
to imaging
4. Styryl dyes differ in the extent to which they are soluble in
standard solvents. It is always best to check the manufacturer’s
latest guidelines. We generally make di-8-ANEPPS at 10 mM
in DMSO but FM 1–43 in water at up to 5 mM. If DMSO is
used as the initial stock, diluted stocks are made into saline to
limit the level of solvent exposed to the cell. We tend to avoid
ethanol/methanol as the rapid evaporation of these solvents
can easily alter the stock concentration.
5. The FM dyes stain membranes rapidly and the fluorescence
throughout a megakaryocyte can reach a plateau within a few
minutes [26]. The staining of any surface-connected mem-
brane by FM dyes is also reversible (see Supplementary online
Video 1). In contrast, di-8-ANEPPS and related dyes take lon-
ger to equilibrate and are effectively irreversible. FM 1–43 and
di-8 ANEPPS emit at wavelengths that overlap with GFP/
YFP, so if simultaneously using a GFP/YFP-tagged reporter
protein, FM4–64 is an alternative as it has a large Stokes shift
and thus emits at a much longer wavelength (peak 734 nm).
6. Many fluorophores can be used as indicators of the extracel-
lular space and thus the DMS. They should be water-soluble,
not readily diffuse across the plasma membrane and not a sub-
strate for a membrane transporter. We have used the Na+ salt

of Oregon Green 488 BAPTA-1 [26] and also HPTS [28].
There are probably better alternatives as Oregon Green
BAPTA-1 is an intracellular Ca2+ indicator, so would not be
compatible with experiments where the extracellular Ca2+ is
lowered below about 5 μM. HPTS has the advantage of being
inexpensive, but is pH-sensitive and not ideally excited with a
standard 488 nm laser (458 and 440 lines are preferable).
7. We have had good success with staining the nucleus of intact
megakaryocytes using Hoechst dyes, which are specific for
dsDNA. The 33342 version is more membrane permeable
than the 34580 and 33258 variants. Hoechst 33342 has an
excitation peak at about 350 nm and emission peak at 460 nm,
so can be used simultaneously with styryl dyes (Fig. 3).
Although we have used a UV laser (364 nm line) to excite
Hoechst in previous studies [26], such lasers are rare due to
their high cost; however, a strong signal can still be obtained
with a 405 nm diode laser. The commonly used alternative
dsDNA dye DAPI is reported to be far less permeable across
the cell membrane and also binds to RNA, albeit generating a
much smaller fluorescent signal compared to its DNA-bound
form. We recently tried the new dsDNA-specific dye DRAQ5
as a potential alternative to Hoechst, as it can be excited with
a 635-nm laser, but found that it displays inadequate permea-
bility across the cell membrane (see Fig. 1).
8. Most objective lenses on inverted microscopes are optimally
corrected for no 1.5 thickness glass coverslips (0.17 mm
thick). The lens normally has this optimal thickness written on
its side.
9. A short “reasonably forceful” depression of the plunger is
normally enough to expel the entire marrow from the femur
or tibia.
10. Pinhole: most confocals have an “optimal” pinhole setting
(normally equal to 1 Airy Unit) for the best axial resolution.
Increasing the pinhole size will increase the fluorescence signal
as this increases the optical thickness of the slice; however, it
will also decrease the axial resolution. When imaging for detail,
use the optimal pinhole setting. However, when carrying out
a Z-series to quantify the DMS, resolution is less of an issue
and the pinhole can be increased so the power of the excita-
tion light and the number of Z-sections can be reduced to
limit phototoxicity and photobleaching.
11. Sampling and resolution: you will also have to decide on how
many pixels to divide the image into. Obviously, more pixels
means greater lateral (x and y axis) resolution up to the theo-
retical limit (0.61 × λ/NA, where λ is the wavelength of the
excitation light and NA is the lens numerical aperture [47]).
According to Nyquist sampling theory, the size of the pixels

should be at least 2.3-fold smaller than this value in order to
take advantage of the maximum ultimate resolution; in prac-
tice, threefold smaller size is often used. However, on a laser
scanning microscope, the greater the number of pixels the lon-
ger the cell is exposed to excitation light, which can result in
more photobleaching or phototoxcity (something we have
seen, in particular with the styryl dyes, see Fig. 5) or the cell
may move during the image acquisition. These factors will
likely limit the feasible resolution. Run a few trials at for exam-
ple 512 x 512 or 1024 x 1024 pixels to assess these problems.
When quantifying the DMS from the fluorescence within a
Z-series, image resolution is less relevant, so the pixel dimen-
sions can be greatly reduced to speed up acquisition and reduce
phototoxicity or photobleaching. If a range of confocal systems
are available, it may be advantageous to use one based on either
a spinning disc or a 2D array scanning system which capture
the image using a camera while illuminating the whole field of
view. This type of system allows much lower average intensity
excitation levels to be used compared to laser scanning confo-
cals, thereby reducing photobleaching and phototoxicity (for
further discussion see [48]). For a greater treatise of these con-
focal imaging topics, please refer to the classic Handbook of
Biological Confocal Microscopy by James Pawley [49].
12. We have found that some conditions can increase the autoflu-
orescence, including extensive exposure to the light levels
used to excite Hoechst or styryl dyes. Thus, it is important to
repeat any experiment, including exposure to excitation light
and test compounds, in the absence of the fluorescent indica-
tors to assess changes in autofluorescent signal.
13. In tests with Hoechst 33342, 30 min exposure to the dye gen-
erated a clear stain of the polyploidic nucleus of intact mega-
karyocytes, although whether the stain had saturated at this
timepoint was unclear from initial experiments.
14. Although the styryl dyes initially stain only the plasma mem-
brane (and thus the DMS), organellar membranes will eventu-
ally become labeled. This is likely a combination of some
transmembrane diffusion and constitutive membrane turnover
(or regulated endocytosis). In tests using small marrow cells
with no significant membrane invaginations, the signal from
internalized dye is still minimal up to 30–40 min after con-
tinuous exposure at room temperature, after which we would
consider the organellar-derived signal to interfere with the
DMS-derived fluorescence.
15. Note that di-8-ANEPPS sticks to the coverslip and even after
washing with ethanol, a fresh batch of cells will show some
staining, and thus the coverslip must be replaced between
experiments.
16. One straightforward approach to quantify the DMS through-

out the megakaryocyte is to set the interval between slices to
be the same as the confocal slice thickness. The fluorescence
values from ROIs drawn around the cell in each Z-section
can then be added together. This will be a reasonable approx-
imation of DMS-dependent fluorescence throughout the cell
and a good compromise considering that cell movements are
likely to occur while acquiring a Z-series. For a theoretically
more accurate value of total fluorescence, a greater number
of Z-sections should be acquired (with an interval at least 2.3
fold smaller than the axial resolution; see discussion on
Nyquist sampling theory above) and a 3D reconstruction
software such as ImageJ or Imaris used to generate total fluo-
rescence within the cell. At the time of writing this chapter,
we are still assessing whether these more complex 3D quan-
tifications of DMS provide any advantage above the simple
single slice method used previously [28].
17. If membrane blebbing occurs, it will be obvious: in the fluo-
rescence image the peripheral plasma membrane can be seen
to move away from the DMS (see Fig. 5).
Acknowledgments
We are grateful to Dr. Kees Straatman of the Advanced Imaging

Facility, Core Biotechnology Services, University of Leicester for
discussion on imaging and microscopy. Work in the authors’ labo-
ratory that led to the methods described in this chapter was funded
by the Medical Research Council and the British Heart Foundation.
S.O. was supported by the Kurdistan Regional Government
Ministry of Higher Education and Scientific Research.
References
1. De Marsh QB, Kautz J, Motulsky AG (1955) 6. Paulus JM, Bury J, Grosdent JC (1979)
An electron microscope study of platelets and Control of platelet territory development in
megakaryocytes. J Clin Investig 34:929–930 megakaryocytes. Blood Cells 5(1):59–88
2. Kautz J, De Marsh QB (1955) Electron micros- 7. Tavassoli M (1980) Megakaryocyte--platelet
copy of sectioned blood and bone marrow ele- axis and the process of platelet formation and
ments. Revue d'hematologie 10(2):314–323 release. Blood 55(4):537–545
discussion, 324–344 8. Wright JH (1906) The origin and nature
3. Yamada E (1957) The fine structure of the of the blood plates. Boston Med Sur J 154:
megakaryocyte in the mouse spleen. Acta Anat 643–645
29(3):267–290 9. Wright JH (1910) The histogenesis of the
4. Yamada E (1955) Some features of the fine blood platelets. J Morphol 21:263–278
structure of megakaryocytes in mouse spleen. 10. Becker RP, De Bruyn PP (1976) The transmu-
Anat Rec 121:458 ral passage of blood cells into myeloid sinusoids
5. Breton-Gorius J, Reyes F (1976) Ultrastructure and the entry of platelets into the sinusoidal
of human bone marrow cell maturation. Int circulation; a scanning electron microscopic
Rev Cytol 46:251–321 investigation. AmJAnat 145(2):183–205
11. Italiano JE Jr, Lecine P, Shivdasani RA, Hartwig 22. Poujol C, Ware J, Nieswandt B, Nurden AT,
JH (1999) Blood platelets are assembled prin- Nurden P (2002) Absence of GPIbα is respon-
cipally at the ends of proplatelet processes pro- sible for aberrant membrane development dur-
duced by differentiated megakaryocytes. J Cell ing megakaryocyte maturation: ultrastructural
Biol 147(6):1299–1312 study using a transgenic model. Exp Hematol
12. Junt T, Schulze H, Chen Z, Massberg S, Goerge 30(4):352–360
T, Krueger A, Wagner DD, Graf T, Italiano JE 23. Denk W, Horstmann H (2004) Serial block-face
Jr, Shivdasani RA, von Andrian UH (2007) scanning electron microscopy to reconstruct
Dynamic visualization of thrombopoiesis within three-dimensional tissue nanostructure. PLoS
bone marrow. Science 317(5845):1767–1770. Biol 2(11):e329. https://doi.org/10.1371/
https://doi.org/10.1126/science.1146304 journal.pbio.0020329
13. Thon JN, Macleod H, Begonja AJ, Zhu J, Lee 24. Eckly A, Heijnen H, Pertuy F, Geerts W,
KC, Mogilner A, Hartwig JH, Italiano JE Jr Proamer F, Rinckel JY, Leon C, Lanza F,
(2012) Microtubule and cortical forces deter- Gachet C (2014) Biogenesis of the demarca-
mine platelet size during vascular platelet pro- tion membrane system (DMS) in megakaryo-
duction. Nat Commun 3:852. https://doi. cytes. Blood 123(6):921–930. https://doi.
org/10.1038/ncomms1838 org/10.1182/blood-2013-03-492330
14. Radley JM, Haller CJ (1982) The demarcation 25. Titze B, Genoud C (2016) Volume scanning
membrane system of the megakaryocyte: a mis- electron microscopy for imaging biological
nomer? Blood 60(1):213–219 ultrastructure. Biol Cell 108(11):307–323.
15. Machlus KR, Italiano JE Jr (2013) The incred- https://doi.org/10.1111/boc.201600024
ible journey: from megakaryocyte devel- 26. Mahaut-Smith MP, Thomas D, Higham

opment to platelet formation. J Cell Biol AB, Usher-Smith JA, Hussain JF, Martinez-
201(6):785–796. https://doi.org/10.1083/ Pinna J, Skepper JN, Mason MJ (2003)
jcb.201304054 Properties of the demarcation mem-
16. Italiano JE (2017) Megakaryocyte develop- brane system in living rat megakaryocytes.
ment and platelet production. In: Gresele Biophys J 84(4):2646–2654. https://doi.
P, Kleiman NS, Lopez JA, Page CP (eds) org/10.1016/S0006-3495(03)75070-X
Platelets in thrombotic and non-thrombotic 27. Mahaut-Smith MP (2004) Patch-clamp
disorders: pathophysiology, pharmacol- recordings of electrophysiological events in
ogy and therapeutics. Springer International the platelet and megakaryocyte. Methods
Publishing AG Mol Biol 273:277–300. https://doi.
17. Nishimura S, Nagasaki M, Kunishima org/10.1385/1-59259-783-1:277
S, Sawaguchi A, Sakata A, Sakaguchi H, 28. Osman S, Taylor KA, Allcock N, Rainbow
Ohmori T, Manabe I, Italiano JE Jr, Ryu T, RD, Mahaut-Smith MP (2016) Detachment
Takayama N, Komuro I, Kadowaki T, Eto K, of surface membrane invagination systems by
Nagai R (2015) IL-1alpha induces throm- cationic amphiphilic drugs. Sci Rep 6:18536.
bopoiesis through megakaryocyte rupture in https://doi.org/10.1038/srep18536
response to acute platelet needs. J Cell Biol 29. Bedlack RS Jr, Wei M, Loew LM (1992)
209(3):453–466. https://doi.org/10.1083/ Localized membrane depolarizations and local-
jcb.201410052 ized calcium influx during electric field-guided
18. Behnke O (1968) An electron microscope neurite growth. Neuron 9(3):393–403
study of the megacaryocyte of the rat bone 30. Betz WJ, Bewick GS (1992) Optical analysis of
marrow. I. The development of the demarca- synaptic vesicle recycling at the frog neuromus-
tion membrane system and the platelet surface cular junction. Science 255(5041):200–203
coat. J Ultrastruct Res 24(5):412–433 31. Gaffield MA, Betz WJ (2006) Imaging synap-
19. Eckly A, Strassel C, Cazenave JP, Lanza F, tic vesicle exocytosis and endocytosis with FM
Leon C, Gachet C (2012) Characterization dyes. Nat Protoc 1(6):2916–2921. https://
of megakaryocyte development in the doi.org/10.1038/nprot.2006.476
native bone marrow environment. Methods 32. Laissue PP, Alghamdi RA, Tomancak P,
Mol Biol 788:175–192. https://doi. Reynaud EG, Shroff H (2017) Assessing
org/10.1007/978-1-61779-307-3_13 phototoxicity in live fluorescence imaging.
20. MacPherson GG (1972) Origin and develop- Nat Methods 14(7):657–661. https://doi.
ment of the demarcation system in megakary- org/10.1038/nmeth.4344
ocytes of rat bone marrow. J Ultrastruct Res 33. Schulze H, Korpal M, Hurov J, Kim SW,
40(1):167–177 Zhang J, Cantley LC, Graf T, Shivdasani RA
21. White JG (1989) Mechanisms of platelet pro- (2006) Characterization of the megakaryo-
duction. Blood Cells 15(1):48–57 cyte demarcation membrane system and its
role in thrombopoiesis. Blood 107(10):3868– F-BAR domain regulates membrane tubu-

3875. https://doi.org/10.1182/ lation in megakaryocytes and platelets.
blood-2005-07-2755 Blood 126:80. https://doi.org/10.1182/
34. Stauffer TP, Ahn S, Meyer T (1998) Receptor- blood-2014-07-587600
induced transient reduction in plasma mem- 42. Antkowiak A, Viaud J, Severin S, Zanoun M,
brane PtdIns(4,5)P2 concentration monitored Ceccato L, Chicanne G, Strassel C, Eckly A,
in living cells. Current Biol: CB 8(6):343–346 Leon C, Gachet C, Payrastre B, Gaits-Iacovoni
35. Watt SA, Kular G, Fleming IN, Downes CP, F (2016) Cdc42-dependent F-actin dynamics
Lucocq JM (2002) Subcellular localization of drive structuration of the demarcation mem-
phosphatidylinositol 4,5-bisphosphate using brane system in megakaryocytes. J Thromb
the pleckstrin homology domain of phospho- Haemost: JTH 14(6):1268–1284. https://
lipase C delta1. Biochem J 363(Pt 3):657–666 doi.org/10.1111/jth.13318
36. Hirose K, Kadowaki S, Tanabe M, Takeshima 43. Zhang J, Varas F, Stadtfeld M, Heck S,
H, Iino M (1999) Spatiotemporal dynamics Faust N, Graf T (2007) CD41-YFP mice
of inositol 1,4,5-trisphosphate that underlies allow in vivo labeling of megakaryocytic cells
complex Ca2+ mobilization patterns. Science and reveal a subset of platelets hyperreac-
284 (5419):1527–1530 tive to thrombin stimulation. Exp Hematol
37. Nieswandt B, Echtenacher B, Wachs FP, 35(3):490–499. https://doi.org/10.1016/j.
Schroder J, Gessner JE, Schmidt RE, Grau GE, exphem.2006.11.011
Mannel DN (1999) Acute systemic reaction 44. Schulze H (2012) Culture of murine
and lung alterations induced by an antiplatelet megakaryocytes and platelets from
integrin gpIIb/IIIa antibody in mice. Blood fetal liver and bone marrow. Methods
94(2):684–693 Mol Biol 788:193–203. https://doi.
38. Carter RN, Tolhurst G, Walmsley G, Vizuete- org/10.1007/978-1-61779-307-3_14
Forster M, Miller N, Mahaut-Smith MP 45. Takayama N, Eto K (2012) In vitro gen-
(2006) Molecular and electrophysiological eration of megakaryocytes and platelets
characterization of transient receptor potential from human embryonic stem cells and
ion channels in the primary murine megakaryo- induced pluripotent stem cells. Methods
cyte. J Physiol 576(Pt 1):151–162. https:// Mol Biol 788:205–217. https://doi.
doi.org/10.1113/jphysiol.2006.113886 org/10.1007/978-1-61779-307-3_15
39. Aguilar A, Pertuy F, Eckly A, Strassel C, 46. Robert A, Cortin V, Garnier A, Pineault N
Collin D, Gachet C, Lanza F, Leon C (2016) (2012) Megakaryocyte and platelet produc-
Importance of environmental stiffness for mega- tion from human cord blood stem cells.
karyocyte differentiation and proplatelet for- Methods Mol Biol 788:219–247. https://doi.
mation. Blood 128(16):2022–2032. https:// org/10.1007/978-1-61779-307-3_16
doi.org/10.1182/blood-2016-02-699959 47. Inoue S (2006) Foundations of confo-
40. Strassel C, Eckly A, Leon C, Moog S, Cazenave cal scanned imaging in light micros-
JP, Gachet C, Lanza F (2012) Hirudin and copy. In: Pawley J (ed) Handbook of
heparin enable efficient megakaryocyte dif- biological confocal microscopy, Third edn.
ferentiation of mouse bone marrow progeni- Springer, Boston, MA, p 19. https://doi.
tors. Exp Cell Res 318(1):25–32. https://doi. org/10.1007/978-0-387-45524-2_1
org/10.1016/j.yexcr.2011.10.003 48. Walker SA, Cozier GE, Cullen PJ (2004) GFP
41. Jurak Begonja A, Pluthero FG, Suphamungmee fusion proteins to study signaling in live cells.
W, Giannini S, Christensen H, Leung R, Lo Methods Mol Biol 273:407–420. https://
RW, Nakamura F, Lehman W, Plomann M, doi.org/10.1385/1-59259-783-1:407
Hoffmeister KM, Kahr WH, Hartwig JH, 49. Pawley JB (2006) Handbook of biological
Falet H (2015) FlnA binding to PACSIN2 confocal microscopy. Springer, Third Edition
Chapter 13
High-Resolution 3D Imaging of Megakaryocytes Using

Focused Ion Beam-Scanning Electron Microscopy
Anita Eckly, Jean-Yves Rinckel, Fabienne Proamer, and Christian Gachet
Abstract
In this chapter, we describe the study of bone marrow megakaryocytes (MKs) using a high-resolution 3D
imaging approach known as focused ion beam-scanning electron microscopy (FIB-SEM). The apparatus
consists of a scanning electron microscope equipped with a focused gallium ion beam, used to sequentially
mill away the sample surface, and an electron beam, used to image the milled surfaces. This produces a
series of ultrastructural images which can be computationally reconstructed into three-dimensional (3D)
volume images. Using this approach it is possible to characterize the 3D ultrastructure of MKs in their
native bone marrow environment, to study subcellular organelle interactions in the context of a complete
cell and to quantify specific features. This chapter provides protocols for sample preparation, image acquisi-
tion and 3D reconstruction, the whole procedure requiring about 7–8 days. It also describes a method
combining light microscopy (LM) with FIB-SEM, a procedure called correlative light electron microscopy
(CLEM), which allows the site-specific 3D imaging of MKs in tissues.
Key words Megakaryocyte, Bone marrow, Focused ion beam-scanning electron microscopy,
Correlative light electron microscopy, 3D reconstruction
1 Introduction
Current methods for ultrastructural studies of blood platelets and

megakaryocytes (MKs) rely on conventional electron microscopy.
Although this technique provides useful information, the interpre-
tation of the results is based on the observation of thin sections
representing only a small part of the cells (<1% of the volume of a
MK). To obtain further insight into biological processes, we need
to study these cells in their native environment and to determine
their whole three-dimensional (3D) intracellular organization.
Until recently, transmission electron tomography was the only
method for 3D ultrastructural studies in a variety of biological
fields. Although it is ideally suited to study cell organelles and cel-
lular substructures, this method is limited to thin sections and is
therefore not appropriate to examine whole cells or tissues. A
217
218 Anita Eckly et al.
major new advance in 3D imaging has become possible with the

invention of focused ion beam-scanning electron microscopy (FIB-
SEM), which allows visualization of ultrastructure in both tissues
and whole cells with nanometer-scale resolution. This microscope
has been mainly used in the field of semiconductors but in recent
years has also been applied to biological specimens [1–3].
The principle of this approach is that a resin-embedded sample
is serial sectioned, using a gallium ion beam integrated into a scan-
ning electron microscope, and the milled face is then imaged with
the electron beam at high resolution. This process called “slice and
view” can be repeated until the desired volume has been recorded
(Fig. 1). The result is a stack of images which can be used to create
3D reconstructions. The specific advantages in our field of research
are that (1) the serial sectioning can be performed on large regions
(up to 100 μm × 100 μm × 100 μm) so that whole MKs can be
investigated in their native environment and (2) the region of
interest (ROI) can be selected with a high degree of precision, i.e.,
this equipment permits targeted, site-specific imaging which is a
particularly valuable feature for bone marrow samples where MKs
are poorly represented (<0.1% of the total cells). One should note
that the microscope has two electron detectors, a back-scattered
electron detector and a secondary electron detector. The back scat-
tered electron signal is generally of lower resolution and thus used
to image the new face of the sample whereas the higher resolution
signal from the secondary electrons is used to image a ROI on the
surface of the sample. Using FIB-SEM large volume analysis, we
were able to reconstruct in 3D the demarcation membrane system
(DMS) of MKs, thereby showing that it originates at the plasma
membrane and that its further expansion requires vesicular mem-
brane delivery from Golgi complexes [4]. We also investigated the
3D organization of the granules in stimulated platelets and dem-
onstrated that weak stimulation favors the fusion of single granules
with the platelet surface, while stronger stimulation induces
granule-to-granule fusion [5].
In this chapter, methods are presented (1) for the prepara-
tion of bone marrow samples, including a contrasting step to
enhance the back-scattered electron signal, (2) for the specific
targeting of MKs within the bone marrow using correlative
light electron microscopy (CLEM), and (3) for the preparation
and platinum coating of a selected site in the block of embedded
tissue. The imaging conditions, procedures for alternate milling
and data acquisition (slice and view) and techniques for process-
ing the 3D data are also described (Fig. 2). The following pro-
tocol illustrates use of the technique for bone marrow MKs and
can likewise be applied to other hematopoietic tissues and to
washed platelets.
3D Imaging of Megakaryocytes 219
Electron
Ion column
column
52°
Image
face
y
Sample
v
vvv
v
vvv
2D image stack
3D reconstrucon
Fig. 1 Schematic illustrating the principle of the FIB-SEM approach. A focused ion beam (yellow) is used to
expose the interior of the tissue, which is then imaged using the scanning electron microscope (red). The ion
source and electron source are placed at an angle of 52°. Iteration of these steps to progress through the cell
volume results in a stack of 2D surface images which can be combined to generate a 3D representation of the
cell. The pictures illustrate bone marrow MKs
preparation
Sample fixation
Sample
Contrast enhancement -Resin embedding
Specimen mounting for FIB/SEM
Semi-thin section from the upper surface of the block
Metallization (Pt/Pd)
Light microscope: select the ROI

CLEM
FIB-SEM (SE mode): find the ROI on the upper surface

FIB milling: create a mark in front of the ROI
Ultramicrotome trimming: imaging face

Preparation
of the ROI
Protective platinum coating over the ROI

Milling of a trench around the ROI
Polishing the block face
Fiducial marks
SEM imaging
FIB milling
Slice and View
(BSE mode)
2D image stack
Alignment
Image Processing
and Segmentation
Gray level-based segmentation

Interpolation of the data
3D reconstruction
Fig. 2 The sequence of steps for 3D imaging of bone marrow MKs includes (1) sample preparation, (2) the
CLEM protocol, (3) preparation of the ROI, (4) the slice and view procedure, and (5) image processing and
segmentation
2 Materials
2.1 Sample 1. 0.1 M sodium cacodylate buffer. Dissolve 21.4 g dimethylar-

Preparation sinic acid sodium salt trihydrate and 20 g sucrose in 1 L
H2O. Under constant agitation add 1 mL of 1 M CaCl2 and
2.1.1 Fixation
1 mL of 1 M MgCl2. The pH is adjusted to 7.3 and the osmo-
and Agarose Infiltration
larity to 306 mOsm/L with approximately equal amounts
of Mouse Bone Marrow
(13 mL) of 1 M MgCl2 and 1 M CaCl2. The final solution is
clear and once filtered (0.2 μm) is stable at +4 °C for up to sev-

eral weeks.
2. Fixative solution: 2.5% glutaraldehyde (Electron Microscopy
Sciences (EMS), USA) in cacodylate buffer. Prepare the same
day and keep at 37 °C. Caution: prepare in a fume hood and
wear eye and skin protection due to the highly toxic nature of
the fixative and its vapor.
3. Agarose type LM-3 low melting point agar (EMS, USA).
4. A water bath.
5. A heated centrifuge (2–16 KCH, Sigma, France).
6. Dissection instruments.
7. A dissecting microscope.
8. Razor blades.
9. Transfer pipettes.
2.1.2 Contrast 1. A graded series of ethanol solutions (75, 90 and 100%) in

Enhancement and Resin water.
Embedding 2. 2% osmium tetroxide (Merck, Germany) in cacodylate buffer.
Warning: osmium tetroxide is extremely volatile at room tem-
perature and its vapors are very harmful to the eyes, nose, and
throat.
3. 3% potassium ferrocyanide (Sigma, France) in 0.1 M cacodyl-
ate buffer.
4. 2% uranyl acetate (Ladd Research Industries, USA) in water.
5. Propylene oxide (1.2-epoxypropane; Sigma, France).
6. Epon (LX-112 resin 217.4 g, dodecenyl succinic anhydride
104.4 g, nadic methyl anhydride 102.78 g, benzyldimethyl-
amine 7.9 g; Ladd Research Industries).
7. A flat embedding silicone mold (Oxford Instruments, Agar
Scientific, UK).
8. An oven set to 60 °C (PeCon GmbH, Germany).
9. A rotating wheel.
2.1.3 Mounting 1. Aluminum SEM stubs (EMS, France).

of the Blocks for FIB-SEM 2. Conductive carbon adhesive (EMS, France).
3. A jeweler’s saw (EMS, France).
4. A sputter coater (e.g., 208HR Cressington, Ted Pella, USA).
5. A sputter target Pt/Pd.
2.1.4 Preparation 1. An ultramicrotome (e.g., Leica EM, Austria).

of Toluidine Blue-Stained 2. Toluidine blue solution: 1 g toluidine blue (Ladd Research
Semithin Sections Industries, USA) and 1 g sodium borate (Sigma, France) are
dissolved in 100 mL distilled water. Toluidine blue solution is

filtered (0.2 μm) and is kept at room temperature in a 10-mL
syringe fitted with a 0.2 μm filter.
3.
A stereomicroscope (MZ12, Leica Microsystems SA,
Germany).
4. Glass microscope slides.
5. A slide warmer (set to 55 °C).
2.2 Preparation 1. An ultramicrotome (e.g., Leica EM, Austria).

of the ROI and Slice 2. A glass knife.
and View Procedure
3. A diamond knife (Cryotrim 35, Diatome, Switzerland)
(optional).
4. A FIB-SEM microscope (e.g., Helios NanoLAB 600i, FEI,
The Netherlands).
2.3 Image 1. A computer: Microsoft Windows 7/8/10 (64 bits), RAM

Processing 16 GB.
and Segmentation 2. A graphic card: NVIDIA, minimum memory of 2 GB.
3. 3D visualization and analysis software (Amira, Image J).
3 Methods
An overview of the experimental protocols used to generate 3D

images of MKs by FIB-SEM imaging is presented in Fig. 2. Briefly,
FIB-SEM begins with the sample preparation (Step 1), followed
by the CLEM and FIB-SEM procedures (Step 2) and the prepara-
tion of the blocks for 3D imaging (Step 3). The automated slice
and view procedure (Step 4) produces serial 2D section images
which are then computationally aligned and segmented into volu-
metric image stacks (Step 5) to allow 3D image visualization and
analysis.
3.1 Bone Marrow A detailed description of murine bone marrow fixation, isolation,
Preparation and preparation may be found in a previous volume within this
series (Platelets and Megakaryocytes, Volume 3, Chapter 13
3.1.1 Dissection
“Characterization of Megakaryocyte Development in the Native
and Fixation of the Bone
Bone Marrow Environment”) [6]. Briefly, the samples are pro-
Marrow
cessed as follows:
1. Under a fume hood, fix the mouse by whole-body perfusion
and dissect the femurs.
2. Quickly flush the marrow with fixative solution to obtain a
cylinder 4–6 mm long.
3. Fix the marrow again by immersion in fixative solution at room

temperature for 1 h.
4. Carefully transfer the sample to a glass slide using a transfer
pipette, cover with 2% agar gel and place on ice until the agar
has solidified. This procedure maximizes tissue integrity.
5. Under a dissecting microscope, cut the cylinder perpendicular
to the long axis with a razor blade to obtain small pieces
(<2 mm × 2 mm) allowing the adequate penetration of
reagents.
6. Place the pieces in an eppendorf tube and rinse them
(2 × 5 min) in 0.1 M cacodylate buffer.
3.1.2 Contrast This protocol is designed to increase the number of back-scattered

Enhancement for Back- electrons, which improves the image quality (see Note 1).
Scattered Electron Imaging
1. Stain the blocks with 1.5% potassium ferrocyanide/1% osmium
and Resin Embedding
tetroxide (wt/vol) in 0.1 M cacodylate buffer for 1 h at room
temperature (see Notes 2 and 3).
2. Postfix the blocks with 1% osmium tetroxide for 1 h at room
temperature.
3. Wash in water (3 × 5 min).
4. Stain the samples with 2% uranyl acetate in water for 1 h at
4 °C.
5. Rinse the samples (3 × 5 min) in distilled water, dehydrate
them in graded alcohol solutions containing 75% ethanol
(4 × 5 min), 95% ethanol (3 × 20 min) and 100% ethanol
(3 × 20 min) and then incubate them in 100% propylene oxide
(2 × 15 min). Add a 1:1 mixture of epon and propylene oxide
to the samples and incubate them for 1 h. Finally, incubate the
samples overnight in 100% epon on a rotating wheel at room
temperature.
6. An exact orientation of the marrow blocks permits subsequent
transverse sectioning of the entire bone marrow. Each block is
placed into a flat silicone mould for embedding. Under a dis-
section microscope, the block is orientated so that the central
sinus is located in the middle of the transverse section.
7. Carefully fill the molds with fresh epon without moving the
position of the block. Place the samples in a 60 °C oven for
48 h to allow the resin to harden. The blocks are then ready for
cutting.
3.1.3 Specimen 1. Using a razor blade, trim the block to a pyramidal shape around
Mounting the tissue (Fig. 3a).
2. Cut off the top of the block using a jeweler’s saw and glue it
onto an SEM stub with carbon adhesive (Fig. 3a, b).
Fig. 3 Preparation of the resin block. The top of the block is trimmed into a pyramidal shape, removed (a) and
glued (b) onto an aluminum SEM stub. (c) A 300 nm-thick section is cut from the upper surface of the resin-
embedded bone marrow and stained with toluidine blue. Bar: 1 cm
3. Cut one semithin section (~ 300 nm) from the top of the resin
block. Stain it with 1% toluidine blue. This section will be
examined under a light microscope to find the region of inter-
est (ROI) for 3D visualization by FIB-SEM (Fig. 3c).
4. Coat the block with a 15 nm thick layer of platinum/paladium
to reduce charging during image acquisition.
3.2 Correlative Light In the CLEM approach, the toluidine blue-stained section from
Electron Microscopy the upper surface of the resin block is viewed by light microscopy
(CLEM) (LM) and the area of interest is located. The block is then placed
in the FIB-SEM microscope and the surface is examined to find
the exact region of the face to be imaged and milled.
3.2.1 Toluidine 1. Focus on the toluidine blue-stained section using the 10×
Blue-Stained Image objective to obtain an overview of the tissue architecture
Acquisition (Fig. 4a). Megakaryocytes can be easily identified due to their
very large size and nuclear lobulation. Record an image of the
section (10×).
2. With the 40× objective, find the exact area of the sample to be
analysed by FIB-SEM (insert Fig. 4a). Choose a landmark
(e.g., vascular sinuses) which can be used to find the same cell
on the top of the resin-embedded sample in the FIB-SEM
microscope. Collect an image of the region of interest (ROI,
40×).
3.2.2 FIB-SEM Image Under the FIB-SEM microscope, the ROI can be identified by
Acquisition visual inspection of the surface of the specimen using the second-
ary electron mode. A high accelerating voltage (e.g., 10 kV) reveals
details of the cells and sinusoids which allow correlation with the
toluidine blue-stained images (Fig. 4b).
Fig. 4 Correlative light electron microscopy (CLEM) to image murine bone marrow. (a) Light microscope image
of a toluidine blue-stained, ultramicrotome-derived semithin section from the top of the resin block. (b) The
upper surface of the block is viewed by FIB-SEM using the secondary electron mode (10 kV). The arrowhead
points to the superficial milling line, while the red rectangle indicates the position of the ROI. The global shape
of the transverse bone marrow section and the central sinus (S) are internal fiducials used to record the posi-
tion of the selected MK. The insets show an enlargement of the same mature MK under the light and electron
microscopes. Bars: 250 μm
1. Place the block in the FIB-SEM microscope.

2. At a low magnification, using the secondary electron mode
(10 kV, 1.4 nAmp), find the ROI on the upper surface of the
block.
3. Switch to a higher magnification to locate with precision the
MK of interest. The 40× toluidine blue-stained image can be
helpful for comparison.
4. Adjust the eucentric height (see Note 4).
5. Tilt the stage to 52°. The exposed surface now lies parallel to
the plane of the ion beam and at an angle of 52° to the electron
beam (Fig. 1).
6. Mark the position by milling a superficial line in front of the
ROI using the FIB (9.4 nAmp, 30 kV, 5 min) (arrowhead in
Fig. 4b). This mark will be visible under the ultramicrotome
binoculars and will be used to facilitate trimming to reach the
target location.
3.3 Preparation The aim here is to remove the excess resin-embedded sample to
of the Tissue Block ensure that the ROI is located immediately below the upper sur-
Containing the ROI face of the block. This will reduce the amount of milling with the
ion beam during the image acquisition phase. We recommend
3.3.1 Ultramicrotome using a glass knife having 90° corners, so that a perpendicular step
Trimming to Expose is cut through the sample. This step forms the “imaging face”
the Imaging Face orthogonal to the upper surface of the sample. The Cryotrim 35
diamond knife is also very well suited for this purpose.
Fig. 5 Ultramicrotome trimming to expose the imaging face. (a) The resin block is oriented so that the fiducial
milling mark visible on the top is aligned with the right edge of the knife. Bar: 1 cm. (b) One side is trimmed
away until the mark lies at the edge. This creates a step orthogonal to the upper surface which forms the
imaging face. Bar: 0.5 cm
1. Clamp the block into the holder of the ultramicrotome. Make

sure that the knife angle is set to 0°.
2. Using the ultramicrotome binoculars, orient the block so that
the milled mark lies parallel to the left edge of the knife
(Fig. 5a).
3. Trim the block until you reach the milled line (Fig. 5b).
3.3.2 Platinum Coating Platinum coating is used to smooth out the surface topography
of the Selected Site (Fig. 6a). This protective layer ensures that the resin face is milled
in the Tissue Block evenly and remains parallel to the direction of the beam and avoids
the formation of vertical white stripes down the imaging face, an
effect known as curtaining.
1. Place the block in the FIB-SEM microscope.
2. At a low magnification, using the secondary electron mode
(2–3 kV, 0.5 nAmp), find the surface overlying the ROI.
3. Deposit a protective layer (~1 μm thick) of platinum on the
surface of the block immediately above the ROI (maximum
area ~100 μm × 100 μm), using the gas injection system cou-
pled to the ion beam (30 kV, 0.79 nAmp) (see Note 5).
3.3.3 FIB-SEM Image In this step, the chosen area of the sample is “trenched” to expose
Targeting the ROI (Fig. 6a).
1. Mill two trenches (~5 μm × 30 μm) on either side of the desired
location using the ion beam (30 kV, 21 nAmp) (see Note 6).
2. Create two score marks (crosses) on the top and side of the
ROI. These fiducial marks are automatically detected by the
Platinum FIB milling

coating
z FIB
sputtered resin
x
SEM
x
Image imaging
FEG y z
Face y
Side trench
Side Sample
a b
trench
Fig. 6 Preparation of the block face for FIB-SEM imaging. (a) Top view of the sample for targeted reconstruc-
tion. The area above the target is protected by a platinum coating and two side trenches are created to expose
the ROI. Two fiducial crosses are milled to allow correct positioning of the ion (FIB) and electron (FEG) beams.
The imaging face is polished. (b) 3D drawing of the resin block showing the position of the gallium ion beam
(yellow) which scans parallel to the block (as indicated by a double-headed arrow) and mills away a layer of
resin, creating a new imaging face. Milling is carried out over 100 μm along the scanning direction
image recognition software and allow accurate positioning of

the electron and ion beams (see Note 7).
3. Finely polish the imaging face using the FIB (2.5 or 9.4 nAmp).
The isolated ROI is now ready for an iterative cycle of resin
milling with the FIB followed by SEM imaging of the newly
revealed face to produce 2D images (Fig. 6b).
3.4 Slice and View 1. Image acquisition protocol: voltage between 1.5 and 3 kV (see
Procedure Note 8), dwell-time = 10 μs/pixel (see Note 9), image
size = 2048 × 1768 pixels, color depth = 8 bits (256 gray
3.4.1 Optimization
scales), pixel size = 4–20 nm, beam current = 0.69 or 1.4
of the Milling and Imaging
nAmp. An extraction voltage of −250 V is employed for the
Conditions
through the lens detector in BSE mode. The images are
inverted and optimized for contrast and focus between snap-
shot iterations.
2. Microscope parameters for milling: 6.4 nAmp and 30 kV (see
Note 10).
3.4.2 Slice and View Run the “slice and view” module and collect 2D images with a sec-
tion thickness of 7–20 nm (see Note 11). The acquisition and
treatment of the images are fully automated, which saves a large
amount of time (see Note 12). Typically, 2–3 days are necessary to
obtain ~1000 pictures (50 μm × 50 μm) using this new approach.
3.5 Image The stacks of images are reconstructed and analysed using 3D
Processing, imaging software (e.g., Amira, FEI Visualisation Sciences Group,
Segmentation and 3D France).
Reconstruction 1. Load the images into the processing program. The image
stacks can be very large (~2 gigabytes or larger) and may pose
technical challenges (see the computer characteristics in the

Materials Section).
2. Indicate the section thickness (7 nm < z < 20 nm).
3. Align the images in pairs.
4. The images can be cropped and bound twice along the x and y
axes.
5. Segment the various cellular components using either the
automatic threshold option for selected regions, i.e., auto-
mated extraction of the features (e.g., nucleus, plasma mem-
brane, demarcation membrane system) (Fig. 7a, b), or manual
drawing (e.g., Golgi apparatus, granules) (Fig. 7c, e) [4, 5].
6. Visualize the 3D reconstruction using the “surface view” tool
(Fig. 7).
Fig. 7 Examples of 3D reconstructions of whole MKs and their organelles using the FIB-SEM approach. These
figures were originally published in Blood. Eckly A, Heijnen H, Pertuy F, Geerts W, Proamer F, Rinckel JY, Léon
C, Lanza F, Gachet C. Biogenesis of the demarcation membrane system (DMS) in megakaryocytes. Blood. 2014
Feb 6;123 (6):921–30. (a, b) Large volume 3D characterization of the DMS (orange) in a bilobulated immature
MK (a) and a mature MK (b). The arrowheads indicate the connections with the cell surface. Automatic seg-
mentation of the plasma membrane (translucent orange), the nucleus (gray), and the DMS (orange). (c, d) 2D
visualization of the Golgi in a mature MK. Selected 2D image from an image stack before (c) and after (d)
manual segmentation of the DMS (yellow) and the Golgi (red). (e) 3D visualization showing the close apposition
of Golgi-derived vesicles at the boundary of the DMS. Interimage spacing 20 nm; in-plane pixel size 6 nm. The
reconstructions were obtained using Amira software
7. Smooth the segmented volumes.

8. The number of events/whole cell and the proportions of the
total volume occupied by organelles can be estimated using the
“material statistics” tool or the “cell counter” plug-in (Image
J, Amira).
3.6 Conclusion The use of FIB-SEM to image cells and tissues is a relatively recent
and Future Prospects development and many future advances in the fields of platelets
and MKs can be anticipated. For example, the spatial location of
information concerning the 3D cellular architecture (e.g., granule
contents, cytoskeletal proteins) can be correlated with the spatial
location of information derived from molecular probes (e.g., quan-
tum dots, metallothionein labeling). Moreover, when FIB-SEM is
paired with multiphoton imaging, functional studies can be per-
formed whereby the same MK and organelles can be imaged first
in a living state using two-photon microscopy and then after fixa-
tion using high resolution EM, where the precise subcellular local-
ization of specific features can be determined [7]. Combining
dynamic cell imaging with 3D imaging at the ultrastructural level
will allow us to address some old, still unresolved questions as well
as new fundamental questions, in particular concerning the mecha-
nisms leading to normal and pathological platelet formation in the
bone marrow.
4 Notes
1. Back-scattered electron (BSE) imaging does not contain topo-

logical information, but the contrast depends on the atomic
composition of the specimen [8]. Materials with larger atomic
numbers such as heavy metals reflect more BSEs [9] and
accordingly, the block face image displays a “brighter” inten-
sity for heavy metal-stained features such as biological mem-
branes stained with osmium tetroxide/potassium ferrocyanide
and a “darker” intensity for light elements such as the sur-
rounding resin. Contrast inversion provides images similar to
those obtained for thin sections using transmission electron
microscopy (TEM), thereby facilitating the interpretation.
2. We tested three different methods to enhance the membrane
contrast: (1) the osmium-thiocarbohydrazide-osmium ligand-
ing (OTO) procedure [10], (2) tannic acid-mediated osmium
impregnation [11], and (3) ferrocyanide-reduced osmium
tetroxide post-fixation (our present method). FIB-SEM obser-
vations showed that the membranes were best stained using
the third procedure. The contrast obtained with methods 1
and 2 was too strong, thickened the images of the membranes
and obscured some ultrastructural details.
3. Interestingly, the enhanced contrast afforded by this proce-

dure allowed direct observation of the sections using TEM,
without further contrasting with heavy metals.
4. The eucentric height is the height at which the electron beam
and the ion beam focus on the same area. Even if you move
the stage (x, y, tilt, rotate) the two beams will remain
coincident.
5. The interaction of the ion beam with the gas leads to the
deposition of a small amount of platinum, by chemical vapor
deposition, at the junction between the needle and the
specimen.
6. The milled resin is ejected from the block and can redeposit on
the block surface, creating imaging artifacts. The two trenches
on both sides of the ROI reduce redeposition on the fields of
view during the image acquisition phase.
7. The beam is initially directed to image a fiducial and the initial
offset to the desired location is determined. Subsequently, the
beam is periodically directed to image the fiducial and the
position of the beam is corrected by determining the offset
between the observed and original coordinates of the
fiducial.
8. It is important that the image be generated by electrons
emerging only from the first few nanometers of the surface of
the block. This is achieved by minimizing the voltage to below
3 kV to ensure that no electrons penetrate too far.
9. The pixel dwell-time needs to be maintained at about 10 μs, so
that the total time to mill the face is about 2 min, which allows
good resolution.
10. The final milled face must be larger than the face to be imaged
because the milled resin is ejected from the block and can
redeposit on the block surface in the immediate vicinity. This
redeposition can encroach onto the field of view, which is
avoided by milling a much larger area than that to be imaged.
11. Before running “slice and view,” leave the microscope for at
least 2 h to allow any thermal changes to dissipate. This
reduces the risk that the block face will drift during imaging,
resulting in misaligned images.
12. Once the specimen has been loaded into the chamber, the
sample processing and imaging can be automated and con-
trolled using a remote-operated computer, thus enabling sepa-
ration of the physical location of the instrument from the site
of control of image acquisition.
Acknowledgments
This work was supported by INSERM, EFS-Alsace, ARMESA

(Association de Recherche et Développement en Médecine et
Santé Publique) and the European Union through the European
Regional Development Fund (ERDF).
Disclosure of Conflict of Interests
The authors declare that they have no conflict of interests.
References
1. Knott G, Marchman H, Wall D et al (2008) 7. Maco B, Holtmaat A, Cantoni M et al (2013)
Serial section scanning electron microscopy of Correlative in vivo 2 photon and focused ion
adult brain tissue using focused ion beam mill- beam scanning electron microscopy of cortical
ing. J Neurosci 28:2959–2964 neurons. PLoS One 8(2):e57405
2. Knott G, Genoud G (2013) Is EM dead? J Cell 8. Richardson CF, Ashcroft NW (1996)
Sci 126:4545–4552 Superconductivity in electron liquids with and
3. Drobne D (2013) 3D imaging of cells and tis- without intermediaries. Phys Rev B Condens
sues by focused ion beam/scanning electron Matter 54:R764–R767
microscopy (FIB/SEM). Methods Mol Biol 9. El Gomati MM, Walker CG, Assa'd AM et al
950:275–292 (2008) Theory experiment comparison of the
4. Eckly A, Heijnen H, Pertuy F et al (2014) electron backscattering factor from solids at
Biogenesis of the demarcation membrane sys- low electron energy (250-5,000 eV). Scanning
tem (DMS) in megakaryocytes. Blood 30:2–15
123:921–930 10. Friedman PL, Ellisman H (1981) Enhanced visu-
5. Eckly A, Rinckel JY, Proamer F et al (2016) alization of peripheral nerve and sensory receptors
Respective contributions of single and com- in the scanning electron microscope using cryo-
pound granule fusion to secretion by activated fracture and osmium-thiocarbohydrazide-osmium
platelets. Blood 128:2538–2549 impregnation. J Neurocytol 10:111–131
6. Eckly A, Strassel C, Cazenave JP, Lanza F, 11. Jimenez N, Vocking K, van Donselaar EG et al
Léon C, Gachet C (2012) Characterization of (2009) Tannic acid-mediated osmium impreg-
megakaryocyte development in the native bone nation after freeze-substitution: a strategy to
marrow environment. Methods Mol Biol enhance membrane contrast for electron
788:175–192 tomography. J Struct Biol 166:103–106
Chapter 14
Optical Clearing of Murine Bones to Study Megakaryocytes

in Intact Bone Marrow Using Light-Sheet Fluorescence
Microscopy
Maximilian G. Gorelashvili, Katrin G. Heinze, and David Stegner
Abstract
In mammals, the differentiation and maturation of megakaryocytes (MKs) occurs in the bone marrow
(BM). The three-dimensional environment influences megakaryopoiesis and platelet release. Thus, imag-
ing MKs within the intact BM is important to understand megakaryopoiesis. Here, we present an optical
clearing protocol for intact bones and the subsequent microscopic analysis including image processing to
quantitatively assess MK size and distribution. This technique overcomes the limitations of classical sec-
tioning methods as the entire bone can be imaged.
Key words Bone marrow, Light sheet fluorescence microscopy, Megakaryopoiesis, Imaging, Optical
clearing
1 Introduction
Megakaryocytes (MKs), residing in the bone marrow (BM), are

constantly releasing platelets into the vessel lumen to maintain
peripheral platelet counts. MKs originate from hematopoietic stem
cells and undergo a maturation process (megakaryopoiesis), during
which they increase in size and become polyploid [1, 2]. Most
MKs reside directly at the BM sinusoids [3, 4], where they form
long cytoplasmic protrusions—termed proplatelets—into the ves-
sel lumen, which fragment and release platelets into the blood cir-
culation [1, 3, 5]. Proplatelet formation has been studied
extensively in vitro [5–7], which elucidated important aspects of
platelet release. In vivo proplatelet formation is strictly regulated
and occurs only at the BM sinusoids [3]. Many “regulators,” like
extracellular matrix proteins [7, 8] or mechanical constraints [9],
inhibit proplatelet formation and thereby appear to “guide” pro-
platelet formation into the vessel lumen. The complex 3D environ-
ment of the BM can only be partially reproduced within in vitro
233
234 Maximilian G. Gorelashvili et al.
assays. Thus, for a more comprehensive understanding of mega-

karyopoiesis, studies within the intact BM are essential. This is,
however, technically challenging as the optical density of the bone
tissue impedes imaging. The development of cryosection tech-
niques for intact bones [10] and intravital two-photon microscopy
[3, 11–13] have been important steps to overcome these prob-
lems. Nevertheless, these methods also have limitations. Sectioning
bones leads to cutting artifacts and 2D images intrinsically lack
valuable 3D information. Two-photon microscopy provides the
possibility of 3D imaging, but from a small field of view, resulting
in low numbers of MKs being investigated. To overcome these
limitations, we made use of a method which enables fast imaging
of large 3D volumes of intact bone at cellular resolution by light-
sheet fluorescence microscopy (LSFM) [4]. In contrast to conven-
tional confocal microscopy techniques, which scan the sample
point-by-point by a laser beam, LSFM uses a laser light-sheet for
whole-plane excitation. The resulting emission signal is detected
by a separate detection objective placed at a 90° angle relative to
the light-sheet and the intact sample is scanned plane-by-plane
(optical sectioning) [14–16]. This allows imaging of the entire
intact bone. The main requirement for LSFM is transparency of
the sample, which can be achieved by optical clearing. Considering
that opaqueness of biological samples results from light scattering
due to differences in the refractive index of tissue components
(water, proteins, lipids, etc.), optical clearing methods aim to
homogenize light scattering properties of the tissue. Over the last
few years a number of clearing methods have been established
using different approaches to reduce opaqueness [17, 18].
However, the majority of these clearing protocols have been devel-
oped for soft tissues. To achieve optical clearing of bones, we
developed a method based on organic solvent clearing using ben-
zyl alcohol and benzyl benzoate (BABB).
In this chapter, we present a powerful method to investigate
MK distribution in intact bone marrow [4], by combining our
optical clearing protocol for bone with 3D LSFM or confocal
microscopy. In addition, we describe image processing and quanti-
tative analysis of acquired 3D image stacks for precise detection
(segmentation) of geometry and localization of MKs, vasculature,
bone, and bone marrow.
2 Materials
2.1 Immunolabeling 1. Adult wild-type mice (8–16 weeks of age) from either inbred
strains (C57Bl/6, 129Sv, Balb/C) or outbred strains (CD1,
NMRi) are commercially available (Janvier, Charles River,
Jackson Laboratories). The procedure is of course also appli-
cable to any transgenic mouse model and respective control
Optical Clearing for Light Sheet Microscopy 235
animals. All animal experiments have to be carried out in accor-

dance with the respective institutional guidelines and govern-
mental regulations.
2. Sterile saline or phosphate buffered saline (1× PBS) without
Ca2+ or Mg2+ to dissolve the antibodies.
3. Antibodies raised against epitopes of interest (see Note 1).
Here, an anti-mouse CD105 (BioLegend, San Diego, CA,
USA) antibody for endothelial cell staining (0.4 μg/g body
weight) and an anti-GPIX derivative (homemade, [19]) for
megakaryocyte staining (0.6 μg/g body weight) are coupled
with dyes which survive the BABB-clearing protocol and are
suitable for the available microscope (see below) (see Note 2).
We have had good success with conjugating the antibodies
using Alexa Fluor protein labeling kits (Alexa Fluor
546/647/750, Life Technologies, Darmstadt, Germany) (see
Note 3).
4. For in vivo labeling these antibodies are dissolved in sterile 1×
PBS or saline to a total volume of 150 μl and administered
intravenously under short-term isoflurane anesthesia.
For ex vivo labeling (optional):
5. Permeabilization buffer: 0.5% Triton™ X-100 (Sigma-Aldrich)
in PBS.
6. Blocking buffer: 1% FCS or BSA and 0.1% Tween® 20 (Sigma-
Aldrich) in 1× PBS.
2.2 Perfusion 1. Triple anesthetic: Mixture of 0.5 mg/kg bodyweight medeto-

Fixation midin (Pfizer, Karlsruhe, Germany), 5 mg/kg bodyweight
midazolam (Roche, Grenzach-Wyhlen, Germany), and
0.05 mg/kg bodyweight fentanyl (Janssen-Cilag, Neuss,
Germany). Can be kept at 4 °C for a week.
2. Dissection instruments: one skin and soft tissue scissor, one
bone scissor, two forceps.
3. Perfusion pump: Ismatec Reglo Analog 2 channels (IDEX—
ISMATEC, USA) with tubing (TYGON LMT-55, 2.54 mm,
0.80 mm, IDEX—ISMATEC, USA).
4. Ice-cold 1× PBS pH 7.1–7.2 and ice cold freshly prepared 4%
paraformaldehyde (PFA; Sigma-Aldrich) in 1× PBS pH 7.1–
7.2 for cardiac perfusion and fixation.
5. 26-gauge needle of 1.5–2 cm length.
6. 70% ethanol in water.
7. Glass vials of 3 ml volume (Hartenstein, Germany).
8. Petri dish filled with 1× PBS for washing of freshly removed
bones.
2.3 Optical Clearing 1.

10% (w/v) ethylenediaminetetraacetic acid (EDTA;
AppliChem, Darmstadt, Germany) in 1× PBS.
2. 1× PBS pH 7.1–7.2.
3. Methanol (Sigma-Aldrich) solution in ddH2O: 50%, 70%, 95%,
100% by volume.
4. BABB optical clearing solution: benzyl alcohol (Sigma-Aldrich)
and benzyl benzoate (Sigma-Aldrich) 1:2 by volume.
2.4 Imaging In contrast to point-scanning methods such as confocal micros-

copy, LSFM scans samples using a plane of light and a 90° detec-
2.4.1 Light-Sheet
tion angle (Fig. 1) resulting in increased acquisition rates and less
Fluorescence Microscopy
photodamage. Due to these advantages, LSFM set-ups are increas-
(LSFM)
ingly available from different microscope manufacturers, such as
the UltraMicroscope (LaVision BioTec), Leica TCS SP8 DLS
(Leica), MVX10 (Olympus), Marianas LightSheet (3i), and Zeiss
Lightsheet Z.1 (Zeiss). Nevertheless, many other laboratories
including our own have home-built microscopes which are adapted
to their needs [15, 20–23]. Here, we describe our custom-built
LSFM set-up to provide a possibility for developing a microscope
system adjustable to individual requirements of different scientific
problems.
1. Imaging chamber: A custom-built cover glass chamber filled
with BABB into which the optically cleared sample is immersed.
Fig. 1 Light-sheet microscopy. The laser beam is elongated by a galvanometric

scanner at a frequency of 600 Hz, focused by a theta lens and relayed by a tube
lens and an objective lens (1) resulting in a thin light-sheet (2). The imaging
chamber (3) is filled with BABB solution and the cleared sample (4), which is
mounted on the microscope stage, is placed into the light-sheet to enable excita-
tion of a sample plane. Emitted light is detected by a detection objective (5) at a
90° angle to the light-sheet
One side of the chamber is partially open for the detection

objective to allow imaging without change of immersion
medium (see Note 4).
2. Microscope stage: The cleared sample is affixed to a glass
micropipette (100 μl, Brand GmbH) with cyanoacrylate glue
(Weicon, Münster, Germany). The micropipette is mounted
on a motorized stage (8CMA06-25/15 Standa, Vilnius,
Lithuania), which performs translational (Newport, Darmstadt,
Germany) and rotational (Standa) movement.
3. Excitation: For excitation, three different laser lines (491 nm,
642 nm and 730 nm) are combined by a custom fiber coupled
laser combiner (BFI OPTiLAS GmbH, Groebenzell,
Germany). The laser beam is collimated by an objective lens
(A10/0.25 Hund, Wetzlar, Germany), resulting in a beam
diameter of circa 3 mm. A dichroic beamsplitter (DCLP 660,
AHF Analysentechnik, Tübingen, Germany) is used to join all
laser lines.
4. Light-sheet generation: The laser beam is elongated with a fre-
quency of 600 Hz by a galvanometric scanner (6210H,
Cambridge Technology, Bedford, MA, USA). The elongated
laser beam is focused via a theta lens (VISIR f. TCS-MR II,
Leica, Mannheim, Germany) to create a virtual light-sheet,
which is relayed to an objective lens (EC Plan-Neofluar
5x/0.16 M27, Zeiss, Göttingen, Germany) via a tube lens and
subsequently to the sample.
5. Detection objective: a HCX APO L20x/0.95 IMM objective
(Leica, Wetzlar, Germany) is mounted on a translation stage
(Newport, Darmstadt, Germany) perpendicular to the light-
sheet and penetrating the imaging chamber.
6. Emission filters: Emission light collected by the detection
objective is filtered using a motorized filter wheel (MAC 6000
Filter Wheel Emission TV 60 C 1.0× (D)) with a MAC 6000
Controller (Zeiss, Göttingen, Germany). The following filters
are used: BrightLine HC 525/50 (laser line 491 nm, detection
of autofluorescence), BrightLine HQ 697/58 (laser line
642 nm, detection of Alexa Fluor 647), BrightLine HC
785/62 (laser line 730 nm, detection of Alexa Fluor 750).
7. Image detection: An infinity-corrected 1.3× tube lens ∞/240–
340 (098.9001.000, Leica, Wetzlar, Germany) is used for
image generation. The created images are detected by an
sCMOS camera Neo 5.5 (2560 × 2160 pixels,
16.6 mm × 14.0 mm sensor size, 6.5 μm pixel size, Andor,
Belfast, UK).
8. Imaging software: Laser illumination, image acquisition, focus
correction and z scanning are controlled by IQ 2.9 software
(Andor). Multicolor stacks are acquired by sequential imaging
of the three color-channels in each plane. Acquired images

have a voxel size of 0.25×0.25×1 μm3, an intensity depth of 11
bit and are saved as 16 bit.tif files.
2.4.2 Confocal For smaller samples, or a detailed analysis of substacks of the sam-
Microscope ple, confocal fluorescence microscopy can be used. Here, a custom-
built cover glass chamber filled with BABB solution to image the
cleared sample immersed in clearing solution is beneficial. In prin-
ciple, any confocal laser scanning microscope can be used. We are
using a Leica TCS-SP8 (Leica, Mannheim, Germany) with differ-
ent excitation laser lines (405 nm, 458 nm, 476 nm, 488 nm,
496 nm, 514 nm, 594 nm, 561 nm, 633 nm). For cleared bones
we use objectives with different magnifications: HCPL Fluotar
10×/0.30 (working distance 11 mm), HC Fluotar L 25×/0.95 W
(working distance 2.5 mm) and HC PL APO 40×/1.30 Oil CS2
(working distance 0.24 mm). Emitted laser light is filtered by an
adjustable 5-channel spectrometer SP Channel 05 and detected
either by a PMT (Hamamatsu R 9624) or a HyD SP GaAsP detec-
tor. In addition, the microscope is equipped with a True Confocal
Scanning system TCS Tandemscanner 8 kHz which is switchable
to the resonant mode for higher imaging speed.
2.5 Image 1. Huygens 17 for image deconvolution (SVI, The Netherlands).

Processing 2. Fiji or ImageJ for image preprocessing [24].
and Segmentation
3. Ilastik 1.2 for bone segmentation [25].
4. Bitplane Imaris 8.3.1 software for image analysis (Bitplane AG,
Zurich, Switzerland).
5. Microsoft Excel (ver. 2013, Microsoft Corporation, Redmond,
WA, USA) or Matlab software (Mathworks, USA) for further
analysis of created objects (MKs, vessels, bone, bone marrow,
vessel-spots).
3 Methods
3.1 Perfusion Investigation of megakaryocytes and the endothelium in the bone

Fixation and In Vivo marrow by fluorescence microscopy techniques requires specific
Immunolabeling staining of these cells by fluorophore conjugated antibodies. In
general, there are two possibilities to perform immunolabeling: (1)
Ex vivo by incubation and (2) in vivo by intravenous injection.
Due to the discontinuous endothelium of the bone marrow sinu-
soids, in vivo immunostaining labels the whole organ homoge-
nously, while ex vivo staining can result in inhomogeneous
distribution of the antibody and/or unspecific accumulation of
fluorescence in the tissue if the fluorophore concentration is not
adjusted carefully beforehand. Thus, ex vivo staining should only
be performed if (1) antibody treatment is not possible due to regu-

lations of animal care, (2) the used fluorophore conjugated anti-
body affects the investigated cells (e.g., depleting them), or (3)
staining of intracellular components is aimed.
For preserving the native structure of the bone marrow, whole
animal perfusion with phosphate buffered saline and 4% parafor-
maldehyde is performed, enabling rapid chemical fixation of the
organs. In addition, perfusion fixation leads to reduction of unde-
sired light absorption due to the removal of red blood cells and
thus to markedly increased image quality.
Local and national regulations for animal care, anesthesia,
treatment, and euthanasia must be followed at all times.
1. 1× phosphate buffered saline (PBS; ca. 40 ml per mouse) and
4% paraformaldehyde (PFA) in PBS (ca. 50 ml per mouse) are
cooled down on ice and kept there during the whole proce-
dure (Fig. 2a).
Fig. 2 Set-up for the perfusion fixation. (a) Clean dissection instruments (1), glass vials for organ storage (5)
and a petri dish with 1× PBS (6) for washing organs are prepared prior to the experiment. 4 °C cold 1× PBS
and 4% PFA (3) are stored on ice and corresponding tubes are inserted into the bottles. The flow rate of the
pump is set to 6 ml/min and a 26-gauge needle (4) is mounted on the free end of the free tube. (b) Tubes 1 and
2 of ca. 60 cm length, which provide 4% PFA and 1× PBS, respectively, are connected to Tube 3 with a switch-
able T-junction. At the beginning of the experiment, Tube 1 is flooded with 4% PFA followed by opening the
T-junction into the PBS direction (as indicated) to fill Tubes 2 and 3 with 1× PBS
Fig. 3 Perfusion fixation. (a) A 8–12 week old anesthetized mouse is fixed on the operating table. After disinfec-
tion of the fur, skin and peritoneum are opened up to the throat. Before perfusion, liver (big arrow) and kidney
(small arrow) have a dark red color. (b) For perfusion surgery, the rib cage is opened without damaging any
organs and is fixed beside the head. The 26G needle is introduced into the left cardiac ventricle and perfused
with 1× PBS followed by an immediate incision of the right atrium. After 3–5 min, the blood is completely
removed as indicated by visible clearing of liver (big arrow) and kidney (small arrow)
2. 2 tubes of ca. 60 cm length and 1 tube of ca. 25 cm length are
connected with a switchable T-junction as indicated in Fig. 2b.
The free ends of the first two tubes are inserted into the bottles
containing 1× PBS and 4% PFA in PBS, respectively. A short
26 G needle is mounted on the free end of the third tube.
3. The flow rate of the pump is adjusted to 6 ml/min resulting in
a lower than physiological pressure.
4. Tube 1 is flooded with 4% PFA in PBS and tubes 2 and 3 with
1× PBS.
5. A 8–12-week-old mouse is anesthetized by an intraperitoneal
injection of an indicated mixture of medetomidine, midazolam,
and fentanyl. Before proceeding with the surgery, complete
absence of pedal and corneal reflexes has to be confirmed.
6. For in vivo immunostaining, the Alexa Fluor conjugated anti-
body derivatives and sterile 1× PBS are adjusted to a total vol-
ume of 150 μl per mouse and injected intravenously 20 min
before the surgery.
7. The mouse is fixed dorsally on the operating table and the fur
is disinfected with 70% ethanol.
8. Skin and peritoneum are carefully cut along the linea alba up
to the throat without incising any viscera (Fig. 3a)
Fig. 4 Bone samples post perfusion fixation. Mouse femur (a), sternum (b) and skull (c) are carefully removed
and briefly washed in 1× PBS. Subsequently surrounding muscles are removed from femur and sternum to
reduce light absorption. Unit size 0.5 cm
9. The left side of the ribs is cut from the diaphragm to the throat.
First, the diaphragm and then the right side of the rib cage are
carefully dissected without damaging the lungs, the heart and
the liver. The rib cage containing the sternum is tilted up straight
without being bent and is fixed beside the head (Fig. 3b).
10. The short needle (26 G) mounted to the end of tube 3 is care-
fully inserted into the left cardiac ventricle (see Note 5).
Immediately after starting the perfusion of ice-cold 1× PBS at
a rate of 6 ml/min, the right atrium is incised to let the blood
flow out (see Note 6). The visible clearing of liver and of the
liquid flowing out of the right atrium is a good indicator for
sufficient blood removal (Fig. 3b, white arrows). The blood
removal procedure takes ca. 3–5 min
11. Subsequently, ice-cold 4% PFA in PBS is perfused through the
mouse for at least 8 min (see Note 7).
12. Femora, sternum, and skull are removed and briefly washed in
1× PBS. Femora and sternum are manually cleaned to remove
attached soft tissue and thereby reduce light absorption during
microscopy (Fig. 4).
13. The samples are put in glass vials (one sample per vial) and
stored in 4% PFA at 4 °C for 30 min to provide post-fixation.
14. The samples are washed by repeated incubation in 1× PBS for
30 min at room temperature (2–3 times).
3.2 Ex Vivo 1. The bone samples are decalcified by incubation in 10% EDTA
Immunolabeling for 4 days at 4 °C on a shaker.
(Optional) 2. The bones are washed by repeated incubation in 1× PBS for
30 min at room temperature (three times).
3. The samples are incubated in permeabilization buffer for

1 week at 4 °C on a shaker.
4. The bones are washed by repeated incubation in 1× PBS for
5. The bones are incubated with 0.5 μg/ml anti-GPIX and
0.5 μg/ml anti-CD105 derivatives in 1 ml blocking buffer for
1 week at 4 °C.
6. To remove the free antibody, the samples are washed by
repeated incubation in 1× PBS for 2 h at room temperature on
a shaker (four times).
7. Proceed with optical clearing as described in Subheading 3.3.
3.3 Optical Clearing Optical clearing is a chemical procedure that allows a dramatic
increase of tissue transparency for visible light by adjusting the
refractive index of the sample. One possibility to do so is replacing
the water in the tissue of interest by a solution of benzyl alcohol
and benzyl benzoate (BABB), which has a refractive index compa-
rable to proteins and lipid bilayers. For this, consecutive steps of
dehydration and immersion in clearing solution are needed. Optical
clearing of whole bones requires an additional step of decalcifica-
tion at the beginning of the optical clearing procedure.
To avoid undesirable chemical reactions, the samples should be
stored in lidded glass vials during the entire optical clearing
procedure.
1. The bone samples are decalcified by incubation in 10% EDTA
for 4 days at 4 °C on a shaker.
2. The bones are washed by repeated incubation in 1× PBS for
3. The samples are dehydrated using methanol solutions of
increasing concentration. Samples are incubated consecutively
in 50%, 70%, 95% and 100% methanol (dissolved in water) for
30 min at 4 °C each. Subsequently, the bones are stored in
fresh 100% methanol for 12–24 h at 4 °C (see Note 8).
4. Afterward, methanol is replaced by BABB solution and the
glass vial is placed in a slightly tilted horizontal position to
avoid internal air bubbles coming in touch with the bone sam-
ple (Fig. 5a). The sample is incubated for 2 h until the organ
sinks in the BABB solution.
5. To achieve sufficient clearing results, the samples are incubated
in fresh BABB solution for at least 12 h at room temperature.
While all bone types become transparent, femora and sternum
remain visible due to their size and yellow-brown color, in con-
trast to skull samples (Fig. 5b). Optically cleared bones can be
stored for up to 4 weeks. In case of sample storage for longer
than 4 weeks, BABB solution should be refreshed every month.
Fig. 5 Optical clearing of decalcified and dehydrated bones. (a) After dehydration, methanol is removed from
the sample. The glass vial is filled with BABB solution and lidded. Subsequently, the vial is placed horizontally
in a slightly tilted position using a 3–4 mm thick spacer to avoid contact of the sample with the air bubble
(arrow) (b) After second incubation in BABB solution (for 12 h) bone samples are sufficiently transparent. In
contrast to skull samples (iii), femur (i) and sternum (ii) are still visible due to their size and yellow shading. Unit
size 0.5 cm
3.4 Imaging Optically cleared bones can be imaged either by confocal or by

light-sheet microscopy. Confocal microscopy provides higher reso-
lution but requires long measurement times which results in
decreasing fluorescence signal due to bleaching. Thus, this tech-
nique is favorable if small regions of the bone marrow are of inter-
est and higher resolution is required. Light-sheet microscopy
enables fast 3D imaging due to its special geometry. Instead of an
excitation beam, an excitation laser sheet is used and emitted light
is detected in a 90° angle (as described in Subheading 2.4 and
Fig. 1). The resolution depends on light-sheet thickness and the
numerical aperture of the used objective and is, in our case, suffi-
cient to resolve single proplatelets protruded into bone marrow
sinusoids (Fig. 6a, b, arrow).
1. Cleared bone is placed into a BABB filled imaging chamber.
2. To prevent light scattering and/or draining of the sample any
contact of the bone with air should be avoided during imag-
ing. Air bubbles are removed from the sample with a small
needle.
3. For overview images 5× or 10× objectives are used (Fig. 6a).
For higher resolution 20× or 40× objectives are preferable
(Fig. 6b) (see Note 9).
4. A region of interest is selected and z-stack imaging is per-
formed. For a resolution sufficient for MK characterization
and distance analysis, we recommend a step size of less than
2 μm.
5. In addition to the used fluorescent dyes, the autofluorescence
signal for the excitation wavelength of 491 nm is detected with
a corresponding emission filter (500–550 nm). This signal is
Fig. 6 Overview and detailed imaging of cleared bones. (a) Optically cleared mouse skull (bregma) was imaged
with an LSFM using a 5× objective. Single MKs (green, anti-GPIX-Alexa 750), vessels (red, anti-CD105-Alexa
647), and the surrounding bone (grey, segmented from autofluorescence, channel at 491 nm) are detectable
in a large sample volume. Segmentation of single MKs and vessels is possible and sufficient for analysis of MK
number and distribution in the bone marrow. (b) The same mouse skull sample was imaged with a LSFM using
a 20× objective. Single MKs (green, anti-GPIX-Alexa 750), vessels (red, anti-CD105-Alexa 647), bone (grey,
segmented from autofluorescence channel at 491 nm), and bone marrow (yellow, segmented from autofluo-
rescence channel at 491 nm) were segmented. The high resolution data are used for precise analysis of cell
shape and size and complex distance calculations. Unit size: 200 μm
used to identify the locations of the bone and bone marrow

structures (see Figs. 6 and 8 and detailed explanation in
Subheading 3.5, step 4).
6. The images are saved as pseudocolor files without compression
to avoid loss of information. For further image processing we
recommend using.tif format.
3.5 Image Image processing is a crucial step for the quantitative analysis of
Processing acquired images. The workflow strongly varies depending on the
and Segmentation aim of the experiment, microscope type, available software and
computing capacity. Here, we provide a generalized protocol for
image processing and quantitative analysis for higher resolution
image stacks, acquired with a 20× objective, which can be modified
and adjusted to the requirements of individual experiments.
1. Deconvolution of the acquired image stack is performed (by
Huygens software) if no BABB compatible dip-in detection
objective was used. This process corrects for image distortion
caused by changing refractive indexes at the interface between
the two different immersion media. Deconvolution of stack
acquired by confocal microscopy is performed with Huygens

Professional 17 using the “deconvolution express” method of
the software, which automatically imports the metadata of the
images and estimates the optimal deconvolution parameters.
For deconvolution of LSFM image stacks Huygens SPIM/
light-sheet Optical Option is used, where the microscope
parameters are edited for the LSFM. For deconvolution we use
the following settings: Classical MLE algorithm, theoretical
PSF, optimized iteration mode with quality change threshold
of 0.1% and number of iterations of 60, estimated signal–noise
ratio of 30, background mode “In/near object” with radius of
2 and bleaching correction.
2. Image preprocessing, i.e., cross talk removal, noise removal,
and contrast correction, is performed using corresponding
functions or plug-ins of Fiji software and leads to high-contrast
data lacking background noise (Fig. 7a). These corrections are
crucial for object detection in images, as high noise and low
contrast can lead to false-positive and false-negative cell seg-
mentation, respectively.
(a)
Cross talk removal: Cross talk or bleed-through for
detected channels should be minimized during imaging by
proper choice of fluorophores, and by well-adjusted com-
bination of excitation laser lines and emission filters. The
remaining cross talk can be removed in a conventional
manner by marking the affected area with the same polyg-
onal region of interest (ROI) in both channels.
Subsequently, the mean voxel intensities I1 (channel 1)
and I2 (channel 2) of the ROI are measured. The channel
with the smaller mean intensity of the ROI (e.g., channel
2 if I2 < I1) is duplicated and multiplied by the ratio I1/I2.
The resulting image is subtracted from channel 1.
(b) Noise removal: All types of detectors will lead to a random
“salt and pepper”-like noise. Thus, we recommend using a
2D median filter with a radius of 2 pixels to successfully
remove this type of noise without affecting the geometry
of imaged cells or structures.
(c)
Background correction: If the background intensity
changes as a gradient in a systematic way, it can be cor-
rected by using a Gaussian filter. The image stack is first
duplicated and a 3D Gaussian filter with radius of 100 vox-
els is applied. The result will be a highly blurry image fit-
ting the structure of the background intensity. This stack is
inverted and added to the original image, resulting in
homogenous background. For correction of more random
background noise, we recommend using Background
subtraction with a rolling ball radius matching the diame-
ter of the smallest object of interest in the image.
Fig. 7 Image preprocessing and segmentation. (a) Cleared mouse femur diaphysis was imaged with a LSFM
using a 20× objective. The image stack was preprocessed by performing noise removal, background correc-
tion, and contrast enhancement. Single MKs (green, anti-GPIX-Alexa 750) and vessels (red, anti-CD105-Alexa
647) can be clearly distinguished and the subcellular resolution is sufficient for depicting proplatelets (arrow)
protruded into vessels. (b) MK (green) and vessel (red) channels from Fig. 7a were segmented by Imaris 8.3.1
Cell module to identify individual objects. High segmentation quality is of great importance and can be achieved
by careful adjustment of processing parameters. When successful, even small structures of complex geometry,
like proplatelets (arrow), can be segmented at high precision. Unit size 200 μm
(d) Contrast enhancement: For local contrast enhancement,

unsharp mask with a rather large radius of 50 pixels and
weight of 0.9 can be used. This process is beneficial for
object detection in later steps and does not change the
geometry of imaged structures.
3. Segmentation of megakaryocytes and vessels is performed with
Imaris 8.3.1. The preprocessed images are loaded and the MKs,
as well as the vasculature, are detected by “Cell module” of the
software. In addition to using an intensity threshold for segmen-
tation, Cell module also takes the shape of the high-intensity
structures into account and therefore, suits better for MK detec-
tion compared to the “Surface” feature of the software.
Generated cell objects are exported as surface objects to enable
extracting the respective quantitative characteristics (Fig. 7b).
Fig. 8 Bone segmentation. (a) Cleared mouse femur diaphysis was imaged with LSFM using a 20× objective.
The autofluorescence channel (excitation: 491 nm, emission: 500–550 nm) shows individual intensity patterns
for different parts of the organ. Bone (1) is of homogenous intensity distribution higher than the background
intensity of empty volumes such as vessel lumen (2). Bone marrow (3) shows a highly inhomogenous intensity
distribution due to autofluorescence of cells residing in this part of the organ. As the absolute intensities of the
bone and the fainter regions of the bone marrow are comparable, segmentation by intensity threshold cannot
be applied. (b) The autofluorescence channel from Fig. 8a was segmented using the machine learning based
Ilastik 1.2 software, which is able to distinguish between different intensity distribution patterns, when trained
accordingly. A surface object for segmented bone was created using Imaris 8.3.1 enabling further quantitative
analysis. Unit size 200 μm
4. Bone and bone marrow are segmented label-free as they can be

distinguished due to their characteristic structure (i.e., inten-
sity patterns) (Fig. 8a). As Imaris 8.3.1 is not able to perform
segmentation according to intensity patterns but only on
intensity thresholds, we recommend using machine learning or
deep-learning software. For our workflow, we established bone
and bone marrow detection by machine learning with Ilastik
v1.2 software. Briefly, the autofluorescence channel is loaded
and three categories of Ilastik-labels are created: Bone, bone
marrow and background. For pattern recognition, we used all
available pattern radii and parameters. After sufficient training
Fig. 9 Distance analysis. A distance transformation for the segmented vasculature from Fig. 7b was created by
“Distance transformation” in Imaris 8.3.1. The intensity of each voxel in the distance map channel corresponds
to the actual distance to the next vessel. Thus, edge-to-edge MK–vessel distances can be measured by esti-
mating the minimum intensity of a distance map channel within the cell. For vessel–vessel distance analysis,
the local maxima of distance map channels are estimated by creating Spot objects. Per definition, the local
maxima are located exactly between two neighboring vessels. Consequently, spot–vessel distance is the half
vessel–vessel distance. Unit size 200 μm
or learning processes with small substacks of the 3D image, the

software is able to segment the entire stack reliably. The seg-
mentation results are binary images for bone and bone mar-
row, which are further transferred to Imaris 8.3.1 and
quantified. Here, Surface objects for bone and bone marrow
can be generated (Fig. 8b) by setting the threshold to larger
than 0 and respective object characteristics are exported as
Excel files.
5. To calculate edge-to-edge MK–vessel, vessel–vessel, and MK–
bone distances, distance maps for the vessel channel and the
bone channel are created in Imaris 8.3.1 (Fig. 9). In general,
distance maps for the vasculature and the bone are generated
channels, where the intensity of each voxel represents the mini-
mum distance to a vessel or bone, respectively. In Imaris 8.3.1

these channels can be created by applying the “Distance trans-
formation” for the respective Surface object using the “Outside
SurfaceObject” method. If no commercial software is avail-
able, distance map generation is also possible using a Fiji plug-
in on binary channels. MK–vessel or MK–bone distance
corresponds to the minimum intensity of the respective dis-
tance map channel within the individual MKs (MK surface
objects). To calculate vessel–vessel distances, local maxima in
the vessel distance map channel must be found. For this, we
make usage of Spot feature of the Imaris 8.3.1 software, where
spherical objects are generated at local maxima of a channel.
Subsequently, similar to MKs, the spot–vessel distance can be
calculated and this distance represents half vessel–vessel dis-
tance by definition.
6. Quantitative characteristics of all created objects (MKs, vessels,
bone, bone marrow, vessel-spots) are exported as Excel files,
containing the information about the geometry of individual
objects as well as their exact position and distance to other
objects. Further analysis is performed with Excel or Matlab
software.
3.6 Conclusion A more detailed understanding of megakaryopoiesis and proplate-

let formation relies on analyzing the cells in their 3D environment
within the bone marrow. In the present chapter, we described an
optical clearing procedure for whole bones and subsequent 3D
image acquisition and processing methods, allowing a complex
quantitative analysis of MK size and shape, as well as their localiza-
tion within the bone marrow. This technique not only increases
the number of analyzed MKs markedly, but also overcomes the
limitations of 2D imaging of cryosections such as cutting artifacts
and lack of 3D information. Indeed, our data reveal that the num-
ber and size of MKs, as well as their distance to vasculature is sig-
nificantly underestimated in 2D images. We could also show that
the bone marrow is densely packed with vasculature. MKs reside in
the remaining space at a high density of ca. 10,000 cells/mm3
(Fig. 10). Further, there are no regions in the bone marrow lack-
ing MKs (Fig. 10). Investigation of other cell types residing in the
bone marrow can also be addressed easily with the presented
method. The optical clearing procedure and LSFM workflow in
combination with image processing and quantitative analysis can
be modified and adjusted to investigate different types of soft tis-
sue (brain, kidney, liver, spleen, etc.). Finally, the protocols are also
applicable for samples of other animals or humans (e.g., bones dis-
sected during surgeries—provided that the ethical regulations
allow such studies).
Fig. 10 Bone marrow is densely packed with vasculature and megakaryocytes.

Cleared mouse sternum was imaged with LSFM using a 20× objective.
Megakaryocytes (green, anti-GPIX-Alexa 750), vessels (red, anti-CD105-Alexa
647), and bone (grey, segmented from autofluorescence channel at 491 nm)
were segmented and reconstructed in 3D. Bone marrow is densely packed with
vasculature resulting in limited free space for megakaryocytes. No MK-free or
not vascularized regions of the bone marrow are detectable indicating, that
megakaryocytes are equally distributed in all parts of the BM. Unit size 200 μm
4 Notes
1. For a successful labeling approach, the epitope and fluoro-

phore should be chosen carefully. As described above, antibod-
ies which can be injected intravenously into the animal in vivo
are favorable. We recommend using anti-GPIX (anti-CD42a)
antibody derivatives which do not affect megakaryocyte func-
tion or localization and GPIX is abundantly expressed on these
cells [4, 11] . For endothelial cells, we recommend labeling of
CD105 (endoglin) rather than CD31 (PECAM-1), as in our
hands anti-CD31 gives a more prominent signal on megakary-

ocytes and platelets as compared to anti-CD105, hampering
the distinction between MKs and the vasculature.
2. As the optical clearing procedure can quench or destroy fluo-
rophores, we recommend using antibodies conjugated to the
extremely bright Alexa dyes. Due to high autofluorescence of
bone marrow at lower wavelengths, red fluorescent dyes are
favorable. Best results could be achieved with the combination
of anti-CD105-Alexa 647 and anti-GPIX-Alexa 750 deriva-
tives. If detection of Alexa 750 is not possible with the avail-
able microscope, anti-GPIX-Alexa 546 derivatives can be used.
3. We recommend long-wavelength fluorophores for smaller
objects to increase signal–noise ratio (e.g., anti-CD105-A647
and anti-GPIX-A750 derivatives for LSFM).
4. The imaging chamber should ideally be of glass that is suitable
for high resolution imaging. The chamber can be designed
either for only the sample (for inverted confocal microscopes)
or for the sample and a BABB compatible detection objective
(for LSFM or upright microscopes).
5. The needle should be introduced with care to avoid penetra-
tion of the interventricular septum.
6. Any air bubbles in the tube should be avoided to enable proper
perfusion.
7. Markedly increased stiffness of the liver and the tail is an indi-
cator for successful tissue fixation.
8. To avoid insufficient dehydration and undesired chemical reac-
tions the samples should be protected from warming up. Thus,
methanol solutions should be stored at 4 °C and the incuba-
tion solutions should be changed quickly or in a cold room.
9. The detection objective should be corrected for chromatic
aberrations (to avoid a shift of the focal plane, resulting in
blurry images) and have a long working distance of more than
1 mm (to achieve a reasonable stack size in z direction) as well
as a high numerical aperture (to increase the image resolu-
tion). Using BABB compatible objectives prevents image dis-
tortion caused by refraction at the site of change of immersion
medium (i.e., BABB to water or BABB to air).
Acknowledgments
We thank Judith M.M. van Eeuwijk for initially establishing the

clearing protocols, Oguzhan Angay for establishing the image pro-
cessing workflow, and Bernhard Nieswandt for constant support.
This work was supported by the Deutsche Forschungsgemeinschaft
(SFB 688) and the Rudolf Virchow Center.
References
1. Patel S, Hartwig J, Italiano J (2005) The bio- https://doi.org/10.1182/

genesis of platelets from megakaryocyte pro- blood-2016-02-699959
platelets. J Clin Invest 115(12):3348–3354. 10. Kawamoto T (2003) Use of a new adhesive
https://doi.org/10.1172/jci26891 film for the preparation of multi-purpose fresh-
2. Grozovsky R, Giannini S, Falet H, Hoffmeister frozen sections from hard tissues, whole-
KM (2015) Regulating billions of blood plate- animals, insects and plants. Arch Histol Cytol
lets: glycans and beyond. Blood 126(16):1877– 66(2):123–143
1884. https://doi.org/10.1182/ 11. Bender M, Stritt S, Nurden P, van Eeuwijk JM,
blood-2015-01-569129 Zieger B, Kentouche K, Schulze H, Morbach
3. Junt T, Schulze H, Chen Z, Massberg S, H, Stegner D, Heinze K, Dutting S, Gupta S,
Goerge T, Krueger A, Wagner D, Graf T, Witke W, Falet H, Fischer A, Hartwig JH,
Italiano J, Shivdasani R, von Andrian U (2007) Nieswandt B (2014) Megakaryocyte-specific
Dynamic visualization of thrombopoiesis Profilin1-deficiency alters microtubule stability
within bone marrow. Science 317(5845):1767– and causes a Wiskott-Aldrich syndrome-like
1770. https://doi.org/10.1126/ platelet defect. Nat Commun 5:4746. https://
science.1146304 doi.org/10.1038/ncomms5746
4. Stegner D, vanEeuwijk JMM, Angay O, 12. Zhang L, Orban M, Lorenz M, Barocke V,
Gorelashvili MG, Semeniak D, Pinnecker J, Braun D, Urtz N, Schulz C, von Brühl M-L,
Schmithausen P, Meyer I, Friedrich M, Dutting Tirniceriu A, Gaertner F, Proia R, Graf T, Bolz
S, Brede C, Beilhack A, Schulze H, Nieswandt S-S, Montanez E, Prinz M, Müller A, von
B, Heinze KG (2017) Thrombopoiesis is spa- Baumgarten L, Billich A, Sixt M, Fässler R, von
tially regulated by the bone marrow vascula- Andrian U, Junt T, Massberg S (2012) A novel
ture. Nat Commun 8(1):127. https://doi. role of sphingosine 1-phosphate receptor
org/10.1038/s41467-017-00201-7 S1pr1 in mouse thrombopoiesis. J Exp Med
5. Italiano JE Jr., , Lecine P, Shivdasani RA, 209(12):2165–2181. https://doi.
Hartwig JH (1999) Blood platelets are assem- org/10.1084/jem.20121090
bled principally at the ends of proplatelet pro- 13. Nishimura S, Nagasaki M, Kunishima S,
cesses produced by differentiated Sawaguchi A, Sakata A, Sakaguchi H, Ohmori
megakaryocytes. J Cell Biol 147 T, Manabe I, Italiano JE Jr, Ryu T, Takayama
(6):1299–1312 N, Komuro I, Kadowaki T, Eto K, Nagai R
6. Malara A, Gruppi C, Rebuzzini P, Visai L, (2015) IL-1alpha induces thrombopoiesis
Perotti C, Moratti R, Balduini C, Tira ME, through megakaryocyte rupture in response to
Balduini A (2011) Megakaryocyte-matrix acute platelet needs. J Cell Biol 209(3):453–
interaction within bone marrow: new roles for 466. https://doi.org/10.1083/
fibronectin and factor XIII-A. Blood jcb.201410052
117(8):2476–2483. https://doi. 14. Huisken J, Swoger J, Del Bene F, Wittbrodt J,
org/10.1182/blood-2010-06-288795 Stelzer EH (2004) Optical sectioning deep
7. Malara A, Currao M, Gruppi C, Celesti G, inside live embryos by selective plane illumina-
Viarengo G, Buracchi C, Laghi L, Kaplan DL, tion microscopy. Science 305(5686):1007–
Balduini A (2014) Megakaryocytes contribute 1009. https://doi.org/10.1126/
to the bone marrow-matrix environment by science.1100035
expressing fibronectin, type IV collagen, and 15. Huisken J, Stainier DYR (2007) Even fluores-
laminin. Stem cells (Dayton, Ohio) 32(4):926– cence excitation by multidirectional selective
937. https://doi.org/10.1002/stem.1626 plane illumination microscopy (mSPIM). Opt
8. Semeniak D, Kulawig R, Stegner D, Meyer I, Lett 32(17):2608. https://doi.org/10.1364/
Schwiebert S, Bosing H, Eckes B, Nieswandt ol.32.002608
B, Schulze H (2016) Proplatelet formation is 16. Voie AH, Burns DH, Spelman FA (1993)
selectively inhibited by collagen type I through Orthogonal-plane fluorescence optical section-
Syk-independent GPVI signaling. J Cell Sci ing: three-dimensional imaging of macroscopic
129(18):3473–3484. https://doi. biological specimens. J Microsc 170(3):229–
org/10.1242/jcs.187971 236. https://doi.
9. Aguilar A, Pertuy F, Eckly A, Strassel C, Collin org/10.1111/j.1365-2818.1993.tb03346.x
D, Gachet C, Lanza F, Léon C (2016) 17. Richardson Douglas S, Lichtman Jeff W (2015)
Importance of environmental stiffness for Clarifying tissue clearing. Cell 162(2):246–
megakaryocyte differentiation and proplatelet 257. https://doi.org/10.1016/j.
formation. Blood 128(16):2022–2032. cell.2015.06.067
18. Silvestri L, Costantini I, Sacconi L, Pavone FS iew light-sheet microscopy. Nat Methods
(2016) Clearing of fixed tissue: a review from a 9(7):755–763. https://doi.org/10.1038/
microscopist's perspective. J Biomed Opt nmeth.2062
21(8):081205. https://doi.org/10.1117/1. 23. Pitrone PG, Schindelin J, Stuyvenberg L,
jbo.21.8.081205 Preibisch S, Weber M, Eliceiri KW, Huisken J,
19. Bergmeier W, Rackebrandt K, Schroder W, Tomancak P (2013) OpenSPIM: an open-
Zirngibl H, Nieswandt B (2000) Structural access light-sheet microscopy platform. Nat
and functional characterization of the mouse Methods 10(7):598–599. https://doi.
von Willebrand factor receptor GPIb-IX with org/10.1038/nmeth.2507
novel monoclonal antibodies. Blood 24. Schindelin J, Arganda-Carreras I, Frise E,
95(3):886–893 Kaynig V, Longair M, Pietzsch T, Preibisch S,
20. Medeiros G, Norlin N, Gunther S, Albert M, Rueden C, Saalfeld S, Schmid B, Tinevez J-Y,
Panavaite L, Fiuza UM, Peri F, Hiiragi T, Krzic White DJ, Hartenstein V, Eliceiri K, Tomancak
U, Hufnagel L (2015) Confocal multiview P, Cardona A (2012) Fiji: an open-source plat-
light-sheet microscopy. Nat Commun 6:8881. form for biological-image analysis. Nat
https://doi.org/10.1038/ncomms9881 Methods 9(7):676–682. https://doi.
21. Krzic U, Gunther S, Saunders TE, Streichan org/10.1038/nmeth.2019
SJ, Hufnagel L (2012) Multiview light-sheet 25. Kerekes RA, Gleason SS, Martins RAP, Dyer M
microscope for rapid in toto imaging. Nat (2010) Fully automated segmentation and
Methods 9(7):730–733. https://doi. characterization of the dendritic trees of retinal
org/10.1038/nmeth.2064 horizontal neurons. In: 2010 Biomedical sci-
22. Tomer R, Khairy K, Amat F, Keller PJ (2012) ences and engineering conference. IEEE,
Quantitative high-speed imaging of entire p 1–4. https://doi.org/10.1109/
developing embryos with simultaneous multiv- bsec.2010.5510843
Chapter 15
Mathematical Techniques for Understanding Platelet

Regulation and the Development of New Pharmacological
Approaches
Joanna L. Dunster, Mikhail A. Panteleev, Jonathan M. Gibbins,
and Anastacia N. Sveshnikova
Abstract
Mathematical and computational modeling is currently in the process of becoming an accepted tool in the
arsenal of methods utilized for the investigation of complex biological systems. For some problems in the
field, like cellular metabolic regulation, neural impulse propagation, or cell cycle, progress is already unthink-
able without use of such methods. Mathematical models of platelet signaling, function, and metabolism
during the last years have not only been steadily increasing in their number, but have also been providing
more in-depth insights, generating hypotheses, and allowing predictions to be made leading to new experi-
mental designs and data. Here we describe the basic approaches to platelet mathematical model develop-
ment and validation, highlighting the challenges involved. We then review the current theoretical models in
the literature and how these are being utilized to increase our understanding of these complex cells.
Key words Systems biology, Mathematical, Computational, Modeling, Mechanistic, Hypothesis

driven, Data driven
1 Motivation for Utilizing Mathematical Techniques
Platelet activation is stimulated by factors exposed, secreted, or

synthesized at sites of vascular damage or disease. Similarly, factors
released by healthy vessels serve to suppress platelet activation.
Platelets possess a host of receptors that allow them to sense the
presence of these factors and respond by initiating complex bio-
chemical processes leading to either platelet activation or inhibi-
tion. These biochemical reactions are frequently drawn as linear
flow diagrams that describe the way in which one signaling mole-
cule influences another, leading to the common terminology of
“signaling pathways.” This works well as a way of simplifying and
formalizing descriptions that contain inherent complexity, but this
is a dramatic oversimplification of processes that are constantly in
255
256 Joanna L. Dunster et al.
flux within a platelet. Platelet activation may be stimulated by an

initiating signal, such as exposure to collagen, but frequently a host
of activatory (and possibility also inhibitory) molecules are present
simultaneously, and therefore the platelet has to integrate these
various signals together and make an appropriate functional
response. Furthermore, platelet activation results in the release of
various proactivatory factors into the local milieu, triggering mul-
tiple parallel positive (or negative) feedback pathways. Thus, mul-
tiple signaling pathways become engaged simultaneously or in
parallel, although the temporal nature of these is unlikely to be
completely simultaneous. It is at this point that the convention of
visualizing cellular regulation as multiple parallel pathways begins
to become problematic. Indeed, platelet signaling would be better
described as a dynamic network.
The complexities of linear signaling pathways are difficult
enough to conceptualize and remember, but once integrated into
complex networks they are sufficient to challenge the most gifted
polymath. Using the conventional cell biologist’s toolbox is insuf-
ficient to answer seemingly simply questions like the one that fol-
lows: What would be the effect on platelet function of inhibiting a
particular enzyme or process by a given amount? Such questions
are key in the development of more effective and safer antithrom-
botic therapies. In addition there are substantial difficulties in
unraveling the variability present within the human population.
Genetic and epigenetic variation impacts on the levels and func-
tions of many proteins that are important for platelet regulation.
Therefore, the responses of one person’s platelets are unlikely to be
identical to those of another. These concepts underpin personal-
ized medicine and challenge scientists to find new ways to inter-
rogate their data to develop a deeper understanding that can be
translated into new therapeutic avenues. The exploitation of math-
ematical and computational approaches to advance our under-
standing of biological responses has been central to one of the
most exciting burgeoning revolutions in the post-genomic era, the
further development of which requires the integration of research
approaches and cultures.
Mathematical and computational modeling of biological sys-
tems has become an exponentially expanding field in recent years.
As in other areas of biomedical research the use of mathematical
modeling in hemostasis research has increased in line with biologi-
cal development and data generation. Very few models of blood
coagulation (or fibrinolysis, or platelet adhesion/signaling/metab-
olism) were developed prior to 1990 [1, 2], but the following
decades witnessed hundreds of reports dealing with mathematical
models of all aspects of hemostasis and thrombosis, e.g., [3–5].
Some of the leading biological research groups have been success-
fully using mathematical modeling as one of their tools for many
years, and systems biology symposia at major thrombosis and
Mathematical Techniques for Understanding Platelet Regulation and the Development… 257
hemostasis congresses are common place. While the initial focus of

such modeling has been on blood coagulation networks and sim-
ple platelet adhesion, an increasing number of such studies since
2008 have focused on platelet signaling and function. It would not
be true, however, to assert that mathematical models have already
become an accepted tool in a noticeable fraction of platelet labora-
tories. For the majority of experimental biology researchers in the
platelet field (as well as in many other biological fields), mathemat-
ical modeling remains something strange and complex, and beyond
their experience. Understanding how mathematical approaches
may be applied to complex biological questions therefore repre-
sents an important step to allow these powerful approaches to be
fully embraced.
Here we shall endeavor to bridge this gap and provide a platelet
researcher with the vital information required to understand how
and why theoretical models of platelets are developed and employed.
The theoretical details will not be discussed as there are excellent
books on mathematical and computational biology [6–9]; the focus
will rather be on the motivation and a description of the tools used
for modeling in platelet research.
1.1 Why Would One The most succinct answer would be that platelet research, mathe-
Wish or Need to Use matical methods and computational power is now at the stage
Mathematical Models when theoretical models can be utilized to efficiently deal with the
in One’s Research complexity of this system, enabling hypothesis- and data-driven
of Platelets? models to make useful predictions.
When one looks at the platelet signaling network (Fig. 1) for
the first time, one is usually stricken by the complexity of interac-
tions, by the sheer number of receptors belonging to different
classes (G protein-coupled receptors (GPCRs), tyrosine kinases,
gated ion channels, etc.), effectors, messengers, signaling pathways
and compartments. Platelets are controlled by numerous activa-
tory and inhibitory stimuli, which interact intricately. Moreover,
responses of these pathways are usually time-dependent (e.g., the
typical mode of calcium signaling occurs in the form of multiple
transients rather than a sustained increase) and spatially nonuni-
form. For a system like this (or even for a system that is one hun-
dredfold smaller and simpler) it is not possible to understand its
outcomes, the roles played by individual reactions, or the effects of
pharmacological intervention by intuition alone. It is exactly these
types of networks that are most attractive for modeling, to allow
researchers to build a virtual version of the network. With such
models, it becomes possible to generate predictions of the system
response, test hypotheses, direct experimental design and uncover
missing reactions or regulatory pathways. Once validated the
models may be used to identify promising drug targets required to
affect the system, and to understand relationships between
mutations of specific molecules and overall platelet response.
Fig. 1 A schematic demonstrating the complexity of signaling networks, with multiple feedback mechanism
and cross talk, that underlies platelet activation. Platelet activation starts with a ligand binding to a receptor on
the platelet plasma membrane (blue rectangles). Activators can be either soluble (thrombin) or membrane-
bound (podoplanin) or insoluble (collagen). The activated receptor induces either heterotrimeric G-protein or
tyrosine-kinase activation. In both cases phospholipase C (PLC) and phosphoinositol-3-kinase (PI3K) became
activated. PLC activation leads to an increase in cytosolic free Ca2+ concentration (shown in shades of red).
Both Ca2+ and PI3K lead to integrin activation (green rectangles) and an increase in the ability of a platelet to
bind with another (platelet aggregation). All other platelet responses, including granule release and shape
change, depend on Ca2+. Some known Ca2+-sensitive proteins are indicated with red dots. Increasing cytosolic
calcium concentration allows calcium to enter mitochondrial matrix to stimulate energy production. NCLX and
mPTP dump calcium from matrix to avoid mitochondria overload and collapse. But in case of strong platelet
activation and intensive calcium signaling the collapse can occur in a fraction of platelet population leading to
platelet necrosis. [Direct activation is shown by solid green arrows, direct inhibition by solid red arrows.
Indirect interactions are shown by dashed lines.] Abbreviations. PR PGI2 receptor, AR α2A-adrenergic receptor,
PAR protease-activated receptor, P2Y purinergic receptor, TR thromboxane A2 receptor, TRPC transient recep-
tor potential channel, SERCA sarcoplasmic/endoplasmic reticulum calcium ATPase, PMCA plasma membrane
calcium ATPase, CDGEF CalDAGGEFI, OCS open canalicular system, DTS dense tubular system, Mit a mito-
chondrion, IP3R receptor for inositol-1,4,5-trisphosphate (IP3), PIP2 phosphoinositol-4,5-bisphosphate, PIP3
phosphoinositol-3,4,5-trisphosphate, DAG diacylglycerol, UNI mitochondrial uniporter, NCLX mitochondrial
sodium/calcium exchanger, mPTP mitochondrial permeability transition pore
1.2 Why Are We Not Theoretical modeling relies on sufficient knowledge of the compo-
Using Mathematical nents of a system before it can be utilized to aid understanding of
Models More? how these components interact. Indeed, there were almost no
mathematical models of the coagulation network prior to 1989, by
which time most of the major proteins and reactions involved had
been discovered. Likewise, development of metabolic control anal-
ysis [10], an example of mathematical biology approaches that has
entered classic biochemical textbooks, was made possible in the
1970s following elucidation of the red blood cell glycolytic path-

way reactions. Theoretical biology is in this respect like theoretical
physics: it is most successful when one has enough information of
individual components to build a model, but has a limited under-
standing of behavior of the integrated system. This aptly describes
the current situation regarding our knowledge of platelet signaling
pathways and understanding of how they interact to control plate-
let function.
In the last two decades substantial progress has been made in
the discovery and characterisation of the key receptors and associ-
ated signaling processes that platelets employ in the control of
their functions (see Fig. 1). While complex network diagrams can
now be drawn, limitations begin to emerge. A substantial focus has
been on platelet activation, frequently considered a one-way pro-
cess, but this is counteracted by inhibitory signaling, about which
much less is known. Furthermore, almost every step within a sig-
naling process can be reversed, but we know very little about these
reverse reactions. Mathematical biology offers the potential to
explore these aspects that are less well characterized, and guide the
design of experiments that are required to fill these gaps.
There are some major differences between modeling in physics/
engineering and biology. Biological systems are much more com-
plex, noisy, have more redundancy and are less well characterized,
which makes mathematical modeling of biological systems very
demanding. The goal of modeling in mathematical and computa-
tional biology is hypothesis testing: can one’s ideas about the sys-
tem’s arrangement be combined to get a response resembling the
real outcome? What did we miss in the understanding of the system?
What could be predicted with such a model? In most cases, the
model would not survive its first experimental testing, which would
mean success and fulfilment of the model’s role: it would mean that
there is something wrong with our ideas about how the system func-
tions, and the model can be used to identify what might wrong and
how to test this. Constructing such models, that are focused on
physiologically relevant outcomes rather than on a general “descrip-
tion” of the system requires detailed biological knowledge. This
provides some explanation of the failure of researchers coming from
theoretical backgrounds to often meet the expectations of the exper-
imental biology community. The answer is for close collaborations
to be formed between theoreticians and experimentalists and this is
happening more and more. A relevant example being how the use of
mechanistic mathematical models are becoming an integral part of
cardiac drug discovery [11]; the aim of this chapter is to promote
this expansion with regard to platelets.
In the sections below we start by describing the modeling pro-
cess, providing a summary (or protocol) of how this process is
generally carried out in (Fig. 2). We then go on to describe the
modeling that has already been achieved before discussing future
prospects and challenges.
Fig. 2 Scheme of the process for the development of mechanism-driven models. Mathematical models convert
biological hypothesis “what we think is happening inside a platelet” into time-dependent simulations that can
be compared to experimental data. In the example above the process starts with a cartoon demonstrating our
current knowledge of RGS (regulator of G-protein signaling) protein role in signal transduction. This is trans-
lated into a system of mathematical equations, where we have utilized mass action kinetics for each reaction
in the scheme. We could also proceed by looking for some experimental data in the literature (in this case the
aggregation data from [71] or conducting our own experiments). If the equations are simple enough we may
be able to analyze them analytically, in the above this leads to the conclusion that without RGS the system
would not settle to a steady state. Using parameter values from literature and parameter inference from
experimental data allows us to simulate the model on a computer; this allows us to explore the time-dependent
behavior of the system. Comparison of model outputs to data can lead to the formulation of new hypothesis.
For example, if we have data on platelet activation (preferably Gα activity) dependence on RGS activity then we
will be able to estimate parameters of the system (e.g., affinity of proteins for each other)
2 The Modeling Process
The modeling process starts with a schematic representation of a

biological system (such as that shown in Fig. 2), that describes
protein interactions, activities and modification in chronological
order, based on current understanding. These cartoons, though
widely used, are problematic in that they are static representations
of systems that we know change over time but translating these

schematics into dynamic mathematical models allows us to bring
the cartoons to life. Computer simulations of the model provide
predictions that describe how the components interact and change
over time. In this way they can be compared to experimental data,
allowing us to test, validate and adapt or reject hypotheses about
what happens inside a platelet to control its activation.
2.1 Model While there is no standard method for the development of mecha-
Development nistic mathematical models, as stated above, the ideal starting point
is the schematic that represents the key components of a system
and how they are thought to interact. These schematics can be
transformed readily into a system of equations that can be solved
numerically on a computer to produce simulations. Knowledge of
the rates of the reactions captured in the model can often be pro-
vided by biologists and literature (see Table 1 for common values
used when modeling platelets). While these values are often based
on alternative experimental conditions, or even other cell types,
they can provide a guide to the feasible range in which these param-
eters may lie. At this point, experimental data is required that
describes an output or interaction captured in the model. The data
allows a fitting process to be carried out, where computer algo-
rithms modify the parameter values, within the range determined
as described above, to obtain the best possible fit between model-
ing simulations and experimental data. This is a critical step that
requires biologists and mathematicians to work closely in assessing
if the refined model parameters are biologically feasible. The dis-
crepancies between modeling simulations and data allow us to
compare various biological hypotheses about which reactions are
important in allowing the model to describe our data. This in turn
leads to refinement of our hypotheses and the resultant mathematical
models. A protocol that describes the development of mechanistic
models is shown in Fig. 2.
2.2 The Choice Mechanistic mathematical models come in a variety of forms (see
of Modeling Structure Fig. 3). The choice of the model type largely depends on the ques-
tion to be answered and the availability of the appropriate experi-
mental data. The most common approach is to use ordinary
differential equations (ODE). This allows the mathematical model
to show how, on average, components vary over time. To include
how components vary not only in time but in space requires the
use of partial differential equations (PDE), commonly named
reaction diffusion systems. Both approaches are described as con-
tinuous as they do not capture the fluctuations that can occur when
components are in low abundance. In contrast, stochastic
differential equations can include the random events that occur in
such situations. This approach is described as discrete. Agent-based
models also fall within this category; they are computational algo-
rithms that simulate the movement of individual components (such
Table 1
262
Key human platelet numbers
Parameter Number, units Method Ref.

Physical parameters
Dimensions 2.0–5.0 μm in diameter and 0.5 μm in [56]
thickness.
Volume 6–10 fl [56]
2
Plasma membrane area 70–100 μm 3D FIB-SEM [57]
Joanna L. Dunster et al.
DTS volume fraction 0.01–0.2 TEM [12, 56, 57]

Number/dimensions/fraction of plt volume 3–8/150 nm/1.3% TEM, 3D FIB-SEM [56, 57]
of dense granules
Number/dimensions/fraction of plt volume 40–80/200–500 nm/15% TEM, 3D FIB-SEM [56, 57]
of α granules
Number/dimensions of other granules 1–3/200–250 nm TEM [56]
Number/dimensions of mitochondria 9/100–500 nm TEM, 3D FIB-SEM [57]
2
Membrane proteins diffusion coefficient 1 μm /s [58]
Number of key receptors and other plasma membrane glycoproteins and channels
Integrins: α2β1/ αIIbβ3 2000–5000/80000 High affinity monoclonal antibody binding [59–61]
and proteomics
GPIb-V-IX 50,000 High affinity monoclonal antibody binding [60, 61]
and proteomics
Main GPCRs:
PAR1/ 547–2063/ Monoclonal antibody/ [61–64]
PAR4/ 1100/ Proteomics/
P2Y1/ 134 ± 68/ Radioligand binding
P2Y12 425 ± 50
Other GPCRs:
PGH2-R 2206 ± 203 Radioligand binding [65]
PGI2-R 63.0 ± 9.7 fmol/107 plt /proteomics [66]
α2-AR 112 ± 12.6/740 [67]/ [61]
Immunoreceptors:
GPVI 5000 [24]
Receptor-channels:
P2X1/ 1400 (150) Proteomics (electrophysiological assays) [61, 69]
TRPC/ 1000 (1)/
TMEM16f/ 4000 (1100)/
Orai1/ 1700 (40)/
IP3RII 1700
Key protein kinases and phosphatases
Serine/threonine kinases: PKA/PKC/PKG 9800/9700/3500 Proteomics, radioligand [61, 68]
Tyrosine kinases: 2763/2581 Immunoblot [24]
Syk/c-Cbl
Phosphatases: TULA-2 7800 [58]
Other key signaling molecules
G-proteins: Gaq/Gai/ 15000/25000/3000/35000/40000 Proteomics [61]
Gas/Gb/Gg
Phospholipases: 4000/2500/1700/2000 Proteomics, immunoblot [61, 70]
A2, Cβ2, Cβ3, Cγ2
PI3K: p85 1700 Proteomics [61]
Mathematical Techniques for Understanding Platelet Regulation and the Development…
263
Fig. 3 Theoretical model frameworks, the different approaches that can be utilized dependent on the questions
to be asked and the experimental data available to test the model
as cells). Hybrid models that incorporate agent-based and contin-

uum models can be utilized to capture multiscale events. The ODE
approach is the easiest to implement and the least computationally
intensive and therefore a good place to begin. If formulating the
model in this way fails to provide an answer to your questions then
the model structure can be modified. If you have experimental evi-
dence that there are fluctuations in the system that arise because
individual components are low in number then a move to a dis-
crete framework may help, alternatively evidence that a biological
systems outcomes are controlled by the location of cell compo-
nents in space may direct you to use a spatial model in the form of
PDEs (Table 2).
Table 2
A summary of the mathematical models reviewed, the platelet functions and mechanisms that
they capture and the theoretical frameworks that they utilize
The model Pathway(s) and mechanisms Modeling framework Ref.

Mechanism-driven models of platelet signaling
Purvis et al. (2008) P2Y1 signaling to calcium. ODE [12]
Lenoci et al. (2011) PAR1 to calcium ODE [17]
Wangorsch et al. cAMP signaling ODE [20]
(2011)
Dunster et al. Early events in the GPVI ODE [24]
(2015) signaling pathway.
Sveshnikova et al. PAR1 signaling to calcium and ODE [18]
(2015) pro coagulant activity.
Mitochondria.
Shakhidzhanov et al. PAR1 and P2Y12 signaling. ODE [22]
(2015)
Balabin et al. (2016) PAR1 signaling to calcium. ODE [25]
Sveshnikova et al. PAR1 and PAR4 signaling to ODE [23]
(2016) calcium and pro coagulant
activity.
Non-(quite)-mechanism-driven models of platelet signaling
Mischnik et al. Boolean model of the central Boolean [21]
(2013) platelet cascades
Chatterjee et al. Neural network for P2Y1/ Machine learning [50]
(2015) P2Y12 GPVI, PAR1/PAR4,
TP, IP receptors.
Non-signaling models of platelet functions
Varfolomeev et al. Thromboxane A2 synthesis ODE [29]
(1986)
Tomas et al. (2014) Stoichiometric metabolic model Flux balance [30]
Shepelyuk et al. Energy metabolism of a platelet. ODE [31]
(2016)
Models that investigate, or just include, platelet functions
Sorensen et al. Platelet deposition and agonist Continuum convection–diffusion– [33, 34]
(1999) release reaction equations (PDEs)
Guy and Fogelson Describes the binding of two ODE [36]
(2002) platelets.
Mody and King Platelet adhesion to vWF PDE, 3D numerical simulations [32–34]
(2005, 2008) under flow.
(continued)
Table 2
(continued)
The model Pathway(s) and mechanisms Modeling framework Ref.

Chatterjee et al. Model of the coagulation ODE [40]
(2010) cascade that includes active
platelets that support
numerous reactions.
Kuharsky and Models of the coagulation ODE, well-mixed [41, 42]
Fogelson (2001) cascade that include two
and Fogelson populations of platelets
et al. (2012) (inactive and thrombin
activated).
Anand et al. (2002, Models of coagulation under PDE [43, 44]
2003) flow supported by activated
platelets.
Pivkin et al. (2006) A model of 3D thrombus PDE [45]
formation that incorporates
platelets and a continuum of
red blood cells.
Xu et al. (2008, Thrombin production via the Hybrid. A form of agent based [48, 49]
2010) coagulation cascade, ADP and modeling tracks platelets and
TXA2 release, and platelet PDEs are utilized for fluid flow
deposition (change words)
Leidermann and Thrombin (coagulation cascade) PDEs describing the convection– [50]
Fogelson (2011) production, ADP release, and diffusion–reaction for all
platelet deposition (change species, including platelets
words)
Flamm et al. (2012) Platelet motion and binding Hybrid approach. A neural [53, 54]
coupled to a neural network network in place of platelet
describing platelet signaling network and PDEs
intracellular calcium [44]. for other species. Lattice
Boltzmann for fluid flow.
Abbreviations: ODE ordinary differential equations, PDE partial differential equations
2.3 Model When constructing a mathematical model it is important to con-

Complexity sider the level of complexity that should be included as it is well
known that highly complex models can fit almost any experimental
data so incorporating complexity needs to be balanced by the avail-
ability of experimental data. The choice of the level of complexity
to include in a model is another example of a critical step that
requires close collaboration between experimentalists and theoreti-
cians. Starting with a simple model is often a good idea. While this
may mean omitting processes and components that are believed to
have a minor influence on the process being modelled, further
complexity and detail can be added when justified by the mismatch
between a model and the experimental data. While the calculation

of the level of complexity to include based on the amount of data
available is not always a simple one, mathematical methods are
available, such as the Akaike Information Criterion, that can help
us assess this balance. The process of model calibration, where we
infer the model’s parameters from the data, is made more difficult
and computationally intensive as the model becomes more com-
plex. The traditional techniques of calibration (called local param-
eter fitting), while fast and easy to interpret, are limited in the
values of parameters that they select. There are new global
approaches that search all possible outcomes of a model. These
result in models that are much better at making accurate predic-
tions but they can be too computationally intensive, particularly
when spatial or stochastic effects are included in the model. Model
calibration, selection, and validation (uncertainty quantification)
are active areas of research that are evolving rapidly in response to
the need of models to make accurate predictions and they are being
aided by rapid increases in computing power.
3 Models of Platelets
Here we provide a brief overview of the current mathematical and

computational literature that focusses on, or involves, a platelet.
These models are divided into those that utilize modeling tech-
niques to explore a platelet’s subcellular mechanisms and those
that capture a platelet’s functional response in a wider context,
such as platelet adhesion to the vessel wall, and its contribution to
coagulation or thrombus formation. For reasons described in the
next sections, a global model of a platelet that fully bridges these
two distinct scales of subcellular mechanisms and functional
response does not yet exist.
3.1 Subcellular Platelets express a diverse range of cell surface receptors that recog-
Mechanisms nize many different environmental cues. These receptors include
GPCRs that respond to soluble agonists, such as ADP and throm-
bin, adhesion receptors, and integrins such as αIIbβ3. Stimulation
of receptors triggers signal transduction pathways that interact
with multiple feedbacks, both positive and negative, to fine-tune a
platelet’s response to its environment.
The first platelet specific model of subcellular signaling mecha-
nisms was proposed by Purvis and colleagues [12]. The model cap-
tures current knowledge of the signaling mechanisms downstream
of the platelet P2Y1 receptor. The model is large and therefore
presented as four individual modules that capture (1) receptor acti-
vation, (2) phosphoinositide (PI) metabolism, (3) calcium release
from the DTS, and (4) an attenuation module that describes a
negative influence of calcium, through PKC, on PLC-beta.
GPCRs and their associated heterotrimeric G proteins have

been the focus of many mathematical models, where they have
been used to test whether current biological knowledge matches
up to experimental data. A useful introduction to these models and
the forms they take is given in [13], where models are being
extended to investigate how combinations of drugs effect the
dynamics of generic GPCRs. Purvis and colleagues based their
module of platelet P2Y1 receptor signaling on a generic ternary
view of a G-protein receptor, gleaning platelet specific parameters
from literature.
In all cell types the liberation of calcium from internal stores or
influx from the extracellular environment gives rise to transient
elevations in cytosolic calcium concentrations that are key to cel-
lular responses to environmental cues. As such, the mechanisms
behind the release of calcium have also been the subject of much
mathematical interest [14]. The model of Purvis captures calcium
flux from the platelet internal stores (the dense tubular system
(DTS)), through inositol 1,4,5-trisphosphate (IP3) receptors and
return into the store via sarcoplasmic/endoplasmic reticulum Ca2+-
ATPase (SERCA). Various models of IP3 receptors have been pro-
posed, reflecting the uncertainty in the activation and regulation
mechanisms of the receptor [15], but a lack of appropriate experi-
mental data means that these models are not yet fully validated.
Purvis et al. utilize the most convincing model of IP3 receptors
available [16], and use the parameter values provided in the paper
that were obtained by comparing model outcomes to data from
hepatic microsomes. The return of calcium from the cytosol,
through SERCA, is, like in many other mathematical models,
assumed to saturate at high levels of cytosolic calcium.
The PI module captures three key phosphoinositides (phos-
phatidyl inositol, PIP, phosphatidylinositol 4,5-bisphosphate,
PIP2, and phosphatidylinositol 3,4,5-trisphosphate, PIP3), hydro-
lysis of PIP2 into IP3 and the recycling of IP3 to the plasma mem-
brane. To address a discrepancy between the model predictions for
the phosphoinositides and experimental data describing phos-
phoinositide behavior in thrombin-stimulated platelets a negative
feedback from protein kinase C (activated by DAG) to PLCβ (pro-
ducing DAG), was introduced. The complete model of Purvis,
with parameter values held fixed to those gleaned from various
sources, was adjusted in terms of protein copy numbers until the
calcium response explained the authors’ own experimental data
describing calcium flux in platelets. Model simulations were able to
replicate both resting and activated system behavior. In addition,
an estimate of the volume of the dense tubular systems was calcu-
lated and the model predicted high ratios of SERCA to IP3R.
Following the work of Purvis et al. a complex model that cap-
tures the interactions downstream of the thrombin receptor PAR1
was proposed by Lenoci and colleagues [17]. This captures not
only heterotrimeric G-protein stimulation through to calcium flux

but also activation of CalDAGGEFI and Rap1b which lead to acti-
vation of integrin αIIbβ3 and dense granule release (see Fig. 1).
This large model (60 ordinary differential equations) with param-
eter values mainly gleaned from literature is nonetheless able to
reproduce agonist concentration-dependent responses of calcium
flux and a number of intermediate steps.
Based on these two works, Sveshnikova et al. developed a
model that focuses on the origin of platelet calcium oscillations
downstream of the PAR1 receptor [18]. The model captured the
entrance of calcium into the cytosol from the extracellular space as
well as from the dense tubular system. Sensitivity analysis revealed
the nonlinear response of IP3R to be the origin of cytosolic cal-
cium oscillations in platelets (as well as in other nonexcitable cells).
The model also included a description of the mechanisms of mito-
chondrial matrix calcium homeostasis: uniporter, sodium-calcium
exchanger and mitochondrial permeability transition pore.
Together this model demonstrates the mechanisms of agonist-
induced platelet necrosis as a result of accumulation of calcium in
mitochondria during calcium oscillations in the cytosol, and mito-
chondrial membrane potential collapse to explain an experimen-
tally observed synergy between two platelet receptors, PAR1 and
P2Y12, in the induction of platelet agonist-induced necrosis. To
our knowledge this was the first mathematical model of platelets
incorporating mechanisms of receptor cross talk. It implied that
both receptors regulate PLC activation, PAR1 through canonical
Gq signaling and P2Y12 through the inactivation of protein kinase
A, which is able to inhibit PLC activation. The proposed role for
PKA in platelet necrosis has been proved by conducting experi-
ments with delayed addition of PAR1 agonist after ADP, in which
case twofold more platelets undergo necrosis [22].
Another model of platelet receptor crosstalk by the same
research group appeared a year later [23], which describes the
necessity for human platelets to have two receptors for thrombin,
PAR1 and PAR4. The model was constructed on the same princi-
ples as the work by same authors [18] and was validated on flow
cytometry data that describes the dynamics of platelet calcium con-
centration and necrosis. The conclusion of Sveshnikova et al. was
that two receptors are needed to allow responsiveness to a wide
range of thrombin concentrations from 0.1 nM to 100 nM [23].
In a similar approach to [21] Dunster et al. [24] derived a
model of the early steps in signal transduction downstream of the
collagen receptor glycoprotein (GP) VI, and adhesion receptor
that is coupled to the activation of tyrosine kinase-dependent sig-
naling. By utilizing their own quantified high-density phosphory-
lation data alongside model calibration, selection, and validation
enabled differentiation between competing biological hypotheses
surrounding the regulation of these initial steps. The conclusion
being that the protein Syk plays a role in its own regulation via
recruitment of a specific tyrosine phosphatase TULA-2 and this
restricts the ability of the GPVI receptor to signal downstream.
It is important to stress that platelet signaling networks con-
tain many positive and negative feedback loops. While some of
these have been incorporated into the described models, their
effects on model outcomes have not always been investigated,
although they have potential to modulate platelet response. The
work of Balabin and Sveshnikova [25], and indeed Dunster et al.
[24], are exceptions. In the former [25] two feedback mechanisms
were shown to have the potential to change a platelet’s peak-like
calcium response to PAR1 receptor stimulation to step-like. These
papers highlight the need for existing and future platelet models to
fully to assess the contribution of feedbacks to model outcomes.
Prior to 1990, platelet energy metabolism was thoroughly
investigated due to its importance for platelets to store various fac-
tors required for hemostasis in its secretory granules [26]. However,
more recently it has become the subject of renewed interest because
cytoskeleton rearrangements are critically depend on availability of
energy, and the formation of reactive oxygen species are linked to
energy production by mitochondria [27, 28]. Yet, apart from an
occasional inclusion of ATP concentration [19, 20], the theoretical
signaling models have not considered platelet energy availability.
However, outside the signaling field there are several models of
platelet metabolism [29–31]. In 1986, Varfolomeev et al.
constructed a kinetic scheme for the exchange of thromboxane and
related lipids, incorporating the activity of the enzymes phospholi-
pase A2 and PGH synthase [29]. Although the model was con-
structed as a system of differential equations it was analyzed only in
steady state, and the main conclusion was that the system could be
held in equilibrium within a wide range of kinetic parameters.
Two additional models [30, 31] have been constructed and
investigated by means of stoichiometric analysis, the main principle
being that the system is always in a steady state and the stoichiom-
etry of reactions determines the distribution of substances between
possible routes. Tomas et al. [30] constructed a stoichiometric
metabolic model of total platelet metabolism comprising a thou-
sand reactions based on platelet genomic, proteomic and biochem-
ical data. Constraints presented by the limited levels of specific
enzymes in platelets were not applied due to the large number of
reactions. The model was applied to data on the metabolism of
isotopically labelled energy substrates obtained by Guppy et al.
[32]. The effect of proposed aspirin resistance on platelet metabo-
lism was evaluated with the outcome that there should be a redi-
rection of glycolytic, fatty acid, and nucleotide metabolism reaction
fluxes in order to accommodate eicosanoid synthesis and reactive
oxygen species stress.
Shepelyuk et al. [31] focused on platelet energy metabolism

and took into account limitations on the reaction fluxes as a conse-
quence of limited levels of specific metabolic enzymes present in
platelets. This model was applied to describe results from a dozen
experimental papers on platelet metabolism published between
1970 and 1990. The main conclusion was that the resting platelet
does not use its mitochondria as most of the glucose is converted
into lactate. Platelet energy metabolism during activation was not
extensively investigated at that point due to some controversy in
experimental results, but a prediction that mitochondria are not
capable of metabolizing all accumulated glucose was made. To
conclude, the described approaches to the steady-state modeling
of platelet metabolism could be useful once the necessary experi-
mental data are obtained.
3.2 Functional A platelet’s primary role is to sense and respond to vascular damage
Responses to form a hemostatic plug, and to interact with the coagulation
cascade and subsequent fibrin formation to form a blood clot.
Platelets respond to multiple environmental cues by activating and
this results in functional changes such as platelet shape change and
granule secretion. These changes also support adhesion, and gen-
erate further platelet activation via autocrine and paracrine routes.
To date, many mathematical and computational models have been
developed to capture these different aspects of a platelet’s f unctional
response and models are now being constructed that attempt to
place a platelet in the context of a growing thrombus that responds
to blood flow. The following models explore a range of simplifying
assumptions and differing mathematical frameworks that reflect
theoreticians’ attempts to overcome the difficulty in incorporating
such diverse events occurring over multiple scales in space and
time.
Platelet adhesion to a site of blood vessel damage is generally
considered a first step in the formation of a blood clot, but platelets
can also adhere to artificial surfaces such as biomaterials. Sorensen
and colleagues [33, 34], building on earlier work by Fogelson
[35], constructed a two dimensional spatial model that simulates
platelet deposition and adhesion to a biomaterial. The model cap-
tures inert and active platelets and, in addition to platelet deposi-
tion on the biomaterial, the model considers platelet–platelet
aggregation at the surface and the roles of prothrombin, thrombin
and the thrombin inhibitor anti-thrombin III. Simulations of the
model compared well to experimental data describing the flow of
whole blood over a collagen substrate. Guy and Fogelson [36]
explored the probability that a collision between two platelets will
result in an adherent bond to be formed by fibrinogen. Their
model predicts that the probability of forming a bond increases
quickly after activation before declining and they postulate that
this decreased ability to form bonds may help regulate platelet
aggregation. Interplatelet binding, mediated by GPIbα interac-

tions with von Willebrand factor, is the subject of a series of papers
by Mody and King [37–39]. The authors investigate the effect of
blood flow on the mechanisms of tethering to exposed vWF. They
conclude that platelet behavior close to a damaged surface is
affected by the platelet’s shape and that a platelet must be brought
in contact with the surface to initiate its capture by vWF.
Platelet aggregation and blood coagulation are tightly cou-
pled. Thrombin, the major product of the coagulation cascade is a
potent activator of platelets while platelets promote coagulation via
the secretion of coagulation factors and by providing surfaces that
enhance many of reactions of the cascade. While numerous math-
ematical models have been formulated to describe the coagulation
cascade, these models have been traditionally compared to data
collected from in vitro experiments that exclude platelets, although
there are a few exceptions. Chatterjee et al. [40] successfully recti-
fied the deficit of previous models of in vitro coagulation to gener-
ate thrombin in the absence of tissue factor by extending a model of
the coagulation cascade to include the effect of platelet activation,
the generation of coagulation factors XIa and XIIa and thrombin
consumption during fibrin polymerization. While platelets were
not specifically incorporated, a thrombin concentration-dependent
function was included to mimic platelet activation and its effect on
the rates of the reactions of the coagulation cascade.
The first model to explicitly incorporate two distinct popula-
tions of platelets, inert and active, into a model of coagulation was
developed by Kuharsky and Fogelson [41]. Platelets progress to an
active state through interaction with subendothelial proteins,
exposure to thrombin, or contact with other platelets (incorpo-
rated to mimic activation via platelet secretion of ADP or throm-
boxane A2). The reactions of the coagulation cascade are divided
into those that occur in plasma and those that occur on the surface
of activated platelets and the model incorporates chemical and cel-
lular transport by flow and diffusion. To reduce computational
costs, reactions were assumed to occur in a thin, well-mixed region
near the injured surface. Model simulations suggest that for small
injuries platelets adhering to and covering exposed subendothe-
lium control the level of the TF- factor VIIa complex in initiating
coagulation. This work was extended [42] to assess the influence
of a feedback from thrombin to the activation of factor XI and
resulted in the prediction that the effects of a severe factor XI defi-
ciency depends on the platelet count.
Blood flow is critical to the development of a blood clot or
thrombus, and the relationship between the flow field and throm-
bus development is reciprocal—the flow field influencing the shape
of the clot and the clot modifying the flow field surrounding it. A
number of researchers have started to produce models that incor-
porate aspects of this relationship. Anand and colleagues [43, 44]
investigated the introduction of spatial effects into a model of clot

formation in flowing blood that incorporates a subset of reactions
from the extrinsic pathway of coagulation and platelets that are
activated from a resting state via thrombin or shear stress due to
blood flow. The effect of blood flow on thrombus growth was
investigated by Pivkin and colleagues [45]. This three-dimensional
computational model tracks individual platelets. The transition
between inert and active platelets being triggered by an agonist or
prolonged exposure to shear stress can be studied. Simulations pre-
dicted the shapes of thrombi growth observed experimentally.
Computational models of stenotic blood flow have been used to
provide insight into how localized flow changes effect platelet
aggregates [46, 47]. This provides an excellent example of the
integration of the predictions from mechanistic models with exper-
imental approaches and helped the authors to investigate the indi-
vidual contributions of shear rates and soluble agonists on thrombus
growth. The authors were able to show that platelet aggregation
and hence thrombus development is driven by rheology with solu-
ble agonists playing a supporting stabilizing role [46].
Xu et al. [48, 49] investigated the use of a hybrid approach to
model clot growth. The movement and activity of individual
platelets was tracked using a form of agent based modeling while
the flow of blood and coagulation factors were described with con-
tinuous PDEs. In this system, platelets can exist in three states: an
inert, an initial and a final activated state; the platelets in the two
latter states generating molecules that promote coagulation and
activation of neighboring platelets. The model captures platelet–
platelet cohesion and platelet deposition to an area of damage.
Leidermann and Fogelson [50] also focussed on thrombus
formation under flow but took an alternative continuous approach,
with platelets tracked not as individuals but as a concentration.
Platelets were captured as inert or active, the transition being trig-
gered by adhesion to the site of injury, thrombin or a secreted
agonist. The model captured how blood flow can affect the growth
of a thrombus and how the growing thrombus feeds back to affect
blood flow. Model simulations predicted that the growth of the
thrombus is influenced by an excess of near-wall platelets and that
the permeability of a thrombus influences its growth. This work
has been extended [51] to investigate the effect of permeability
inside a thrombus, finding that early clot growth is promoted by
the ability of platelets within a thrombus to shelter enzymes from
blood flow.
In a different approach to those described above, Chatterjee
et al. [52] neglect the mechanistic details of subcellular networks
to use machine learning approaches to construct a computational
neural network that can predict calcium flux in a platelet in response
to a wide range of agonists. The neural network was trained to data
describing calcium flux in response to six different agonists (ADP,
convulxin, U46619, SFLLRN (PAR1 agonist), AYPGKF (PAR4

agonist), and prostaglandin E2) applied individually or as paired
combinations. Following on from this work Flamm et al. [53, 54]
incorporate the neural network into a model of platelet deposition,
the neural network being used to track individual platelet responses
to multiple agonists as platelets are deposit onto a damaged sur-
face. Simulations of the model were compared to microfluidic
experiments of whole blood flowing over collagen and successfully
predicted the rank order of drug sensitivity for indomethacin,
aspirin, a P2Y1 inhibitor, and iloprost.
4 Challenges and Future Prospects
The mathematical and computational approaches reviewed here

show that modeling, particularly when applied to subcellular sig-
naling pathways, is starting to make an impact on our understand-
ing of mechanisms of regulation, thereby directing experiments
and the testing of hypotheses. However, there are many challenges
still to be addressed before these theoretical techniques could be
utilized more widely (Fig. 4). Of particular importance will be the
development of approaches to bring together multiscale modeling
approaches in a logical manner that will enable the researcher to
fully understand the effects of the modulation of platelet signaling
molecules on platelet function, thrombus formation and thrombo-
sis. The incorporation of population variability (while in itself pro-
Fig. 4 Open challenges

viding a useful and endless resource with which to challenge or test

our models) will be vital to enable real-world use of these systems,
but while attractive is a formidable task. Ultimately these chal-
lenges will be overcome, but this will be of little consequence if the
tools that are developed are not accessible. Considerable efforts
will therefore be required to enable biologists, pharmacologists
and clinicians to access and use the models, without having to
understand (or see) the underlying mathematics. This requires the
blending of expertise and the training of the next generation of
researchers in cross-disciplinary skills that are presently very rare.
4.1 Accessibility The accessibility of mathematical and computational models and

and Publishing biological data is a topic under much consideration. To increase
Standards take up of theoretical techniques within the biological community,
models need to be openly published in standards that make them
easy to understand. This will also help researchers to review models
and encourage their reuse and redevelopment. At present there is
no one standard that allows this. A common standard for publica-
tion of theoretical models is SBML (Systems Biology Markup
Language) and this allows models to be imported to tools such as
Copasi (a software package, http://copasi.org) so that model sim-
ulations can be run by nonexperts. But SBML and associated tools
are restricted in the allowable model structure and their ability to
utilize full computational architecture. Methods of translating
SBML into programming languages such as the data science favor-
ites of R, Python, Matlab (Octave), and the newly developed Julia
are not yet reliable. As such models are best published in a variety
of formats, manuscripts should provide a full description of the
equations, parameters and initial conditions. Further, the code to
run models should be placed on repositories (such as Github or
Zenodo) and SBML made available where possible. Biology, sup-
ported by advances in high-throughput experimental technologies
is becoming data intensive, but to fully exploit this data deluge
requires the open publication and curation of biological data in
standardized formats.
4.2 Uncertainty One of the great challenges is to develop models that are as credi-
Quantification ble, predictive, and as useful as those used throughout physics and
engineering. The rise in the availability of large-scale data now
means that mathematical modeling of biological and medical sys-
tems can shift from qualitative to quantitative predictions that are
much more useful for driving experimental plans and the develop-
ment of novel treatments. However, for this to arise the reliability
of model predictions must be optimized. With the clear majority of
a model’s parameters being difficult or impossible to measure
experimentally, a model’s precision depends on its ability to be
inferred from biological data. The quantification of uncertainties
in both the biological data and the models and parameters inferred
from the data is critical to access the accuracy of predictions. This

gives rise to a number of challenges both experimental and theo-
retical. Biological experimental data are often noisy, collected at
few points in time, and not quantitative. To reduce the impact of
noise data needs to be collected under standardized conditions.
Inferring models and parameters is an area of active research as is
the assessment of uncertainty in the resulting models.
4.3 Construction To produce a mathematical model that can be utilized to under-

of a Multi Scale Model stand how subcellular mechanisms support a platelet’s response to,
and influence on, its environment requires the construction of a
model that captures multiple scales (temporal and/or spatial). The
mathematical and computational challenges in formulating and
analyzing such models are immense. Indeed, systematically linking
submodels that have been constructed to describe interactions on
distinct scales poses significant mathematical challenge in its own
right. The complexity of these models, with scales that may vary
over many orders of magnitude, makes even simple simulations
computationally intensive, precluding model inference from data
and therefore the construction of experimentally verifiable models.
These are subjects that are under active research by the mathemati-
cal community with techniques such as singular perturbation the-
ory being utilized to reduce complexity in large complex networks
such as the coagulation cascade by extracting the macro scale
dynamics [55]. Experimental data to support the development of
multiscale models is necessarily likely to come from many different
sources. While construction of these models offers the opportunity
to integrate diverse data sources into a coherent picture that can be
utilized to test biological hypotheses, it relies on the data being
openly published and, as mentioned above, curated in standard-
ized formats.
5 Concluding Remarks
In summary, although great advances have been made in under-

standing the components inside a platelet our current inability to
integrate the array of molecular signals that control the function of
these anuclear blood cells is holding back understanding and the
development of more effective antithrombotic therapies. With the
concomitant and ever growing increase in computational power
and the availability of high-dimensional experimental and clinical
data collected over disparate scales we argue that hypothesis- and
data-driven mathematical and computational modeling can now
come of age.
We are entering a revolutionary new era in which patient-
specific genetic, anatomical, and physiological information needs
to be integrated into a coherent story; mechanistic modeling is
expected to play a large role in this revolution.
References
1. Panteleev MA, Ananyeva NM, Ataullakhanov 17. Lenoci L, Duvernay M, Satchell S et al (2011)
FI et al (2007) Mathematical models Mathematical model of PAR1-mediated
of blood coagulation and platelet adhe- activation of human platelets. Mol BioSyst
sion: clinical applications. Curr Pharm Des 7:1129–1137
13(14):1457–1467 18. Sveshnikova AN, Ataullakhanov FI, Panteleev
2. Ataullakhanov FI, Panteleev MA (2005) MA (2015) Compartmentalized calcium sig-
Mathematical modelling and computer simu- nalling triggers sub population formation upon
lation in blood coagulation. Pathophysiol platelet activation through PAR1. Mol BioSyst
Haemost Thromb 34(2–3):60–70 11:1052–1060
3. Xu Z, Kamocka M, Alber M et al (2011) 19. Obydennyy SI, Sveshnikova AN, Ataullakhanov
Computational approaches to studying throm- FI, Panteleev MA (2016) Dynamics of calcium
bus development. Arterioscler Thromb Vasc spiking, mitochondrial collapse and phospha-
Biol 31(3):500–505 tidylserine exposure in platelet subpopula-
4. Diamond SL (2013) Systems biology of coagu- tions during activation. J Thromb Haemost
lation. J Thromb Haemost 11(s1):224–232 14(9):1867–1881
5. Shibeko AM, Panteleev MA (2015) Untangling 20. Wangorsch G, Butt E, Mark R et al (2011)
the complexity of blood coagulation network: Time-resolved in silico modeling of fine-tuned
use of computational modelling in phar- cAMP signalling in platelets: feedback loops,
macology and diagnostics. Brief Bioinform titrated phosphorylations and pharmacological
17(3):429–439 modulation. BMC Syst Biol 5:178
6. Мurray JD (2002) Mathematical biology: I. An 21. Mischnik M, Boyanova D, Hubertus K et al
introduction. Springer, New York, NY (2013) A Boolean view separates platelet acti-
7. Murray JD (2003) Mathematical biology II: vatory and inhibitory signalling as verified by
spatial models and biochemical applications. phosphorylation monitoring including thresh-
Springer, New York, NY old behaviour and integrin modulation. Mol
BioSyst 9:1326–1339
8. Edelstein-Keshet L (2005) Mathematical
models in biology. Society for Industrial and 22. Shakhidzhanov SS, Shaturny VI, Panteleev MA
Applied Mathematics, Philadelphia et al (2015) Modulation and pre-amplification
of PAR1 signalling by ADP acting via the P2Y12
9. Szallasi Z, Stelling J, Periwal V (2006) System receptor during platelet sub population forma-
modeling in cell biology: from concepts to nuts tion. Biochim Biophys Acta 1850:2518–2529
and bolts. MIT Press, Cambridge
23. Sveshnikova AN, Balatskiy AV, Demianova
10. Wildermuth MC (2000) Metabolic control AS et al (2016) Systems biology insights into
analysis: biological applications and insights. the meaning of the platelet's dual-receptor
Genome Biol 1(6). reviews1031-1 thrombin signalling. J Thromb Haemost
11. Davies MR, Wang K, Mirams GR et al (2016) 14:2045–2057
Recent developments in using mechanistic 24. Dunster JL, Mazet F, Fry MJ et al (2015)
cardiac modelling for drug safety evaluation. Regulation of early steps of GPVI signal trans-
Drug Discov Today 21(6):924–938 duction by phosphatases: a systems biology
12. Purvis JE, Chatterjee MS, Brass LF et al (2008) approach. PLoS Comput Biol 11:e1004589
A molecular signalling model of platelet phos- 25. Balabin FA, Sveshnikova AN (2016)
phoinositide and calcium regulation dur- Computational biology analysis of platelet sig-
ing homeostasis and P2Y1 activation. Blood nalling reveals roles of feedbacks through phos-
112:4069–4079 pholipase C and inositol 1,4,5-trisphosphate
13. Woodroffe PJ, Bridge LJ, King JR et al (2010) 3-kinase in controlling amplitude and duration
Modelling of the activation of G-protein cou- of calcium oscillations. Math Biosci 276:67–74
pled receptors: drug free constitutive receptor 26. Holmsen H (1985) Platelet metabolism and
activity. J Math Biol 60(3):313–346 activation. Semin Hematol 22(3):219–240
14. Thul R (2014) Translating intracellular cal- 27. de la Peña NC, Gutiérrez-Aguilar M,
cium signalling into models. Cold Spring Harb Hernández-Reséndiz I, Marín-Hernández Á,
Protoc 5:pdb-top066266 Rodríguez-Enríquez S (2017) Glycoprotein Ib
15. Sneyd J, Falcke M (2005) Models of the ino- activation by thrombin stimulates the energy
sitol trisphosphate receptor. Prog Biophys Mol metabolism in human platelets. PLoS One
Bio 89(3):207–245 12(8):e0182374
16. Sneyd J, Dufour JF (2002) A dynamic model 28. Kramer PA, Ravi S, Chacko B, Johnson MS,
of the type-2 inositol trisphosphate receptor. Darley-Usmar VM (2014) A review of the
Proc Natl Acad Sci U S A 99(4):2398–2403 mitochondrial and glycolytic metabolism in
human platelets and leukocytes: implications 42. Fogelson AL, Hussain YH, Leiderman K
for their use as bioenergetic biomarkers. Redox (2012) Blood clot formation under flow: the
Biol 2:206–210 importance of factor XI depends strongly on
29. Varfolomeev SD, Gachok VP, Mevkh AT platelet count. Biophys J 102(1):10–18
(1986) Kinetic behavior of the multien- 43. Anand M, Rajagopal KR (2002) A mathemati-
zyme system of blood prostanoid synthesis. cal model to describe the change in the con-
Biosystems 19:45–54 stitutive character of blood due to platelet
30. Thomas A, Rahmanian S, Bordbar A et al activation. CR Mecanique 330(8):557–562
(2014) Network reconstruction of platelet 44. Anand M, Rajagopal K, Rajagopal KR (2003)
metabolism identifies metabolic signature for A model incorporating some of the mechanical
aspirin resistance. Sci Rep 4:3925 and biochemical factors underlying clot forma-
31. Shepelyuk TO, Panteleev MA, Sveshnikova AN tion and dissolution in flowing blood. J Theor
(2016) Computational modeling of quiescent Med 5(3–4):183–218
platelet energy metabolism in the context of 45. Pivkin IV, Richardson PD, Karniadakis G (2006)
whole-body glucose turnover. Math Model Blood flow velocity effects and role of activa-
Nat Pheno 11:91–101 tion delay time on growth and form of platelet
32. Guppy M, Abas L, Neylon C, Whisson ME, thrombi. P Natl A Sci 103(46):17164–17169
Whitham S, Pethick DW, Niu X (1997) Fuel 46. Nesbitt WS, Westein E, Tovar-Lopez FJ et al
choices by human platelets in human plasma. (2009) A shear gradient–dependent platelet
Eur J Biochem 244(1):161–167 aggregation mechanism drives thrombus for-
33. Sorensen EN, Burgreen GW, Antaki JF mation. Nat Med 15(6):665–673
(1999) Computational simulation of plate- 47. Tovar-Lopez FJ, Rosengarten G, Khoshmanesh
let deposition and activation: II. Results for K et al (2011) Structural and hydrodynamic
Poiseuille flow over collagen. Ann Biomed Eng simulation of an acute stenosis- dependent
27:449–458 thrombosis model in mice. J Biomech
34. Sorensen EN, Burgreen GW, Antaki JF (1999) 44(6):1031–1039
Computational simulation of platelet deposi- 48. Xu Z, Chen N, Kamocka MM et al (2008) A
tion and activation: I. Model development and multiscale model of thrombus development.
properties. Ann Biomed Eng 27:436–448 J R Soc Interface 5(24):705–722
35. Fogelson AL (1992) Continuum mod- 49. Xu Z, Lioi J, Mu J et al (2010) A multiscale
els of platelet aggregation: formulation and model of venous thrombus formation with
mechanical properties. SIAM J Appl Math surface-mediated control of blood coagulation
52(4):1089–1110 cascade. Biophys J 98(9):1723–1732
36. Guy RD, Fogelson AL (2002) Probabilistic 50. Leiderman K, Fogelson AL (2011) Grow with
modeling of platelet aggregation: effects the flow: a spatial–temporal model of platelet
of activation time and receptor occupancy. deposition and blood coagulation under flow.
J Theor Biol 219:33–53 Math Med Biol 28(1):47–84
37. Mody NA, Lomakin O, Doggett TA et al 51. Leiderman K, Fogelson AL (2013) The influ-
(2005) Mechanics of transient platelet adhe- ence of hindered transport on the development
sion to von Willebrand factor under flow. of platelet thrombi under flow. Bull Math Biol
Biophys J 88(2):1432–1443 75(8):1255–1283
38. Mody NA, King MR (2008) Platelet adhesive 52. Chatterjee MS, Purvis JE, Brass LF et al (2010)
dynamics. Part I: characterization of plate- Pairwise agonist scanning predicts cellular sig-
let hydrodynamic collisions and wall effects. nalling responses to combinatorial stimuli. Nat
Biophys J 95(5):2539–2555 Biotechnol 28(7):727–732
39. Mody NA, King MR (2008) Platelet adhesive 53. Flamm MH, Diamond SL, Sinno T (2009)
dynamics. Part II: high shear-induced transient Lattice kinetic Monte Carlo simulations of
aggregation via GPIbα-vWF-GPIbα bridging. convective-diffusive systems. J Chem Phys
Biophys J 95(5):2556–2574 130(9):094904
40. Chatterjee MS, Denney WS, Jing H et al 54. Flamm MH, Colace TV, Chatterjee MS et al
(2010) Systems biology of coagulation initia- (2012) Multiscale prediction of patient-
tion: kinetics of thrombin generation in rest- specific platelet function under flow. Blood
ing and activated human blood. PLoS Comput 120(1):190–198
Biol 6(9):e1000950 55. Dunster JL, King JR (2017) Mathematical
41. Kuharsky AL, Fogelson AL (2001) Surface- modelling of thrombin generation: asymptotic
mediated control of blood coagulation: the analysis and pathway characterization. IMA
role of binding site densities and platelet depo- J Appl Math 82(1):60–96
sition. Biophys J 80(3):1050–1074
56. White JG (2007) Platelet structure. Platelets in patients with acute myocardial infarction.
2:45–71 Circulation 81(1):212–218
57. Eckly A, Rinckel J, Proamer F, Ulas N, Joshi S, 66. Arehart E et al (2008) Acceleration of cardio-
Whiteheart SW, Gachet C (2016) Respective vascular disease by a dysfunctional prostacyclin
contributions of single and compound granule receptor mutation. Circ Res 102(8):986–993
fusion to secretion by activated platelets. Blood 67. ABRAHM DR et al (1986) Decreased
128(21):2538–2549 α2-adrenergic receptors on platelet membranes
58. http://book.bionumbers.org/ from diabetic patients with autonomic neu-
59. Wagner CL et al (1996) Analysis of GPIIb/IIIa ropathy and orthostatic hypotension. J Clin
receptor number by quantification of 7E3 bind- Endocrinol Metab 63(4):906–912
ing to human platelets. Blood 88(3):907–914 68. Eigenthaler M, Nolte C, Halbrugge M, Walter
60. Clemetson KJ, Clemetson JM (2002) Platelet U (1992) Concentration and regulation of
receptors. Platelets:117–144 cyclic nucleotides, cyclic-nucleotide-dependent
61. Burkhart JM et al (2012) The first comprehen- protein kinases and one of their major sub-
sive and quantitative analysis of human platelet strates in human platelets. Estimating the
protein composition allows the comparative rate of cAMP-regulated and cGMP-regulated
analysis of structural and functional pathways. protein phosphorylation in intact cells. Eur
Blood 120(15):e73–e82 J Biochem 205(2):471–481
62. Ramström S et al (2008) Platelet PAR1 recep- 69. Wright JR, Amisten S, Goodall AH, Mahaut-
tor density—correlation to platelet activation Smith MP (2016) Transcriptomic analysis of
response and changes in exposure after platelet the ion channelome of human platelets and
activation. Thromb Res 121(5):681–688 megakaryocytic cell lines. Thromb Haemost
116(2):272
63. Baurand A, Raboisson P, Freund M, Leon C,
Cazenave JP, Bourguignon JJ et al (2001) 70. Lee SB, Rao AK, Lee KH, Yang X, Bae YS,
Inhibition of platelet function by administra- Rhee SG (1996) Decreased expression of
tion of MRS2179, a P2Y1 receptor antagonist. phospholipase C-beta 2 isozyme in human
Eur J Pharmacol 412:21321 platelets with impaired function. Blood
88(5):1684–1691
64. Ohlmann P et al (2013) The platelet P2Y12
receptor under normal and pathological condi- 71. Hensch NR, Karim ZA, Druey KM, Tansey
tions. Assessment with the radiolabeled selec- MG, Khasawneh FT (2016) RGS10 negatively
tive antagonist [3H] PSB-0413. Purinergic regulates platelet activation and thrombogene-
Signal 9(1):59–66 sis. PLoS One 11(11):e0165984. https://doi.
org/10.1371/journal.pone.0165984
65. Dorn GW et al (1990) Increased platelet
thromboxane A2/prostaglandin H2 receptors
Index
A anti-fibrinogen�� 15, 16, 29, 63, 71

anti-GPIX��235, 242, 244, 246, 250, 251
α2A-adrenergic receptor (α2-AR)�� 258, 263 anti-myosin IIA/MYH9�� 14, 18
α2β1��262 anti-PECAM-1 (WM59)�� 1, 4, 5
α2β1-specific ligand GFOGER��84 anti-phosphotyrosine�� 39, 47
Abciximab (ReoPro)�� 17, 24 anti-phospho-VASP��96–101, 103, 106–109
Acetone-based substitution�� 61, 68 anti-PLC-γ2��109
Acid-citrate-dextrose (ACD)��4, 6, 17, 21, 22, 24, anti-P-selectin�� 14, 15, 18, 19, 25, 29
29, 39, 42, 64, 76, 83, 84, 88, 90, 114–116, 118 anti-thrombotic therapies�� 256, 276
Actin cytoskeleton�� 41, 45 anti-tubulin�� 27, 29, 36
Actin dynamics��4 anti-vWF�� 29, 61, 71
Actin filament��36 biotinylated�� 142, 145
Actin nodules��38 blocking��50
Adhesion�� 38, 49, 56, 59, 64, 76, 81–89, 92, Fab fragments��50
151, 256, 257, 265, 267, 269, 271, 273 phosphospecific�� 109, 122
Adhesion assay��82–85 phospho-Zap-70��109
Adhesion receptors�� 267, 269 Anticoagulant
Adhesive ligands��2 acid citrate dextrose (ACD)�� 21, 57
ADP�� 2, 4, 5, 8, 89, 92, 98, 266, 267, 269, 272, 273 sodium citrate�� 21, 22, 57, 64, 98
Adult peripheral blood�� 178, 179, 181 Apyrase�� 8, 114, 116
Agent based models��261, 264, 266, 273 Arp2/3��18
Aggregation��21, 24, 56, 81–83, 87–89, Arp3��209
92, 169, 258, 260, 271–273 Artificial lipid bilayers��127–137
αIIbβ3 integrin��38, 86, 262, 267 Ascorbic acid�� 158, 171
Akaike Information Criterion��267 Aspirin�� 6, 84, 274
Alexa Fluor�� 14–16, 20, 25, 28, 39, 43, 97, Aspirin resistance��270
99–109, 132, 209, 235, 237, 240 ATP�� 8, 270
Alexa Fluor beads��132 Autofluorescence�� 26, 94, 108, 201, 207,
Alexa Fluor-conjugated antibody��240 212, 237, 243, 244, 247, 250, 251
Alexa Fluor 488 Succinimidyl Ester�� 97, 101, 105 Autophagosome�� 69, 75
Alexa488-phalloidin�� 20, 39, 43
Alpelisib (PI3K α inhibitor)�� 91, 258 B
Alpha granules (α-granules)�� 14–16, 18, 19, 33,
Back-scattered electron (BSE) detector�� 218, 220,
38, 68, 74, 262
223, 227, 229
Alpha tubulin�� 18, 19
β-actin��121
Ammonium chloride��43
Basal AE6 medium�� 160, 161, 168, 170, 172
Anesthetic��235
Benzyl alcohol and benzyl benzoate (BABB)�� 234–236,
Antibodies
242–244, 251
anti-β tubulin�� 14, 16, 18, 27, 121
Biochips��2, 3, 5–7, 9
anti-CD41�� 15, 208, 209
Bioreactor��180, 186–188, 191
anti-CD61��14, 16, 18, 25
Biotinylated lipids��127, 129, 130, 136
anti-CD105��242, 244, 250, 251
BMP4��158, 161, 167, 168
anti-CD61/integrin β3�� 14, 16
Bombyx mori silkworm cocoons�� 178, 179, 182
anti-ERK1/2��109
https://doi.org/10.1007/978-1-4939-8585-2, © Springer Science+Business Media, LLC, part of Springer Nature 2018
281
Platelets and Megakaryocytes: Volume 4
282 Index

Bone��57, 58, 63, 65, 67–69, 139, 140, 142, C3H10T1/2��156
144–145, 148, 151, 177–180, 182–191, 197, 199, 208, CLEC-2�� 38, 127
210, 218–225, 229, 233–251 Clotting factors��22
Bone marrow (BM)�� 57, 58, 63, 65, 67–69, CMOS camera�� 35, 41, 237
139–140, 142, 144–145, 177–182, 186, 187, 189–191, Coagulation��13, 21, 128, 256–258, 266,
197, 200, 208, 218, 220–225, 229, 233–251 267, 271–273, 276
environment�� 148, 177, 178, 186, 222 Coagulation complexes�� 21, 128
Bovine serum albumin (BSA) gradient�� 2, 3, 5, 6, 8, Coagulation factors�� 272, 273
17, 23, 39, 43, 61, 71, 83, 85, 115, 117, 118, 123, 129, Collagen
158, 173, 179, 181, 182, 235 type I��181
Brightfield image��5 type IV��159, 180, 181, 184
Btk��91 Collagenase�� 158, 169
Collagen related peptide, cross-linked
C (CRP-XL)��4, 5, 8, 84, 87, 91, 98
Complete culture medium�� 143, 145, 150
Ca2+ assay�� 83, 90–93
Compound
Ca2+ calibration��94
libraries�� 82, 83
Cacodylate buffer��58, 65, 67, 143, 147, 148,
screening��83
152, 220, 221, 223
Computational modelling��36, 256, 271, 273, 275, 276
Ca2+ concentration�� 7, 8, 258
Computer algorithms��261
Ca2+ entry�� 9, 90
Conductive carbon adhesive��221
Calcein��198
Confocal (fluorescence) microscopy/microscope��6–8,
Calcium�� 2, 7, 158, 189, 257, 258, 265–270, 273
29, 50, 84, 140, 197, 198, 202, 203, 212, 234, 236, 238,
flux��3, 268, 269, 273
243, 245, 251
homeostasis��269
Confocal imaging��34, 197, 206, 212
oscillations��269
Confocal pinhole�� 201, 204
release��267
Continuum models��264
signalling�� 4, 7, 82, 91, 257, 258
Contouring��56
transients��2
Contrast inversion��229
CalDAGGEFI�� 258, 269
Convulxin��274
Calnexin��18
Copper grids�� 60, 66
Cangrelor�� 86, 91
Copper lacey carbon��76
Cardiac perfusion��235
Corn trypsin inhibitor��21
Ca2+ stores��93
Correlative light and electron microscopy
CD5�� 142, 145
(CLEM)�� 57, 59–64, 71–73, 218, 220, 224, 225
CD11b�� 142, 145
Cover slips (cover glass)�� 2, 17, 186, 236, 238
CD19�� 142, 145
CRISPR-Cas9��39
CD31��5, 6, 250, 251
Cryo-electron microscopy (cryo-EM)�� 57, 72, 75
CD34�� 181, 182
Cryo-immobilization��68
CD41�� 19, 156, 157, 159, 163, 165, 173, 209
Cryoprotectant solution�� 60, 66
CD41 (ITGA2B)��163
Cryo-sectioning��56–58, 66, 70, 77
CD42��157, 164, 165, 173
Cryo-substitution�� 60, 70
CD42a (GP9)��156, 159, 163, 165
Cryo-ultramicrotome�� 60, 66
CD42b��18
Culture
CD42b (GP1Balpha)��18
liquid�� 140, 141, 145, 149–153
CD45�� 181, 182
methylcellulose�� 141, 151, 152
CD45R/B220�� 142, 145
three dimensional��139–153
CD61 (Integrin β3)�� 14, 16, 18
Cytobank Population Manager��107
CD62 (P-selectin)�� 18, 19
Cytoclips�� 143, 148
CD63 (LAMP3)�� 19, 38
Cytofunnels�� 143, 148, 152
CD105�� 235, 250
Cytokine�� 156, 158, 161–166, 172, 179
CD235a�� 159, 163, 165
Cytoskeletal proteins��229
CellGro medium��164–166, 170, 174
Cytoskeleton��9, 16, 33, 35–37, 41, 45, 67, 140, 270
Cell viability��172
Cytosolic Ca2+�� 81, 82
Chloroform�� 129–130, 136
Cytospin�� 143, 148, 149
283
Index

D Embedding��56–60, 65–68, 70, 73, 76, 140,
145–147, 196, 221, 223
Dakopen�� 144, 148 Embryoid body (EB)��158, 167–170, 174
DAOSTORM algorithm��52 Embryonic stem cells (ESCs)��155
DAPI��26, 159, 173, 211 EMCCD (EM-CCD) cameras�� 20, 40, 41
Dasatinib�� 86, 89 Endosome��69
Decalcification��242 Endothelium�� 156, 238, 272
Degranulation��4 Entospletinib (Syk inhibitor)��89
Dehydration�� 68, 70, 73, 152, 196, 242, 243, 251 Environment��63, 71, 94, 139, 140, 148, 160, 178,
Demarcation membranes�� 199, 201–204 186, 197, 204, 217, 218, 222, 233, 249, 267, 268, 276
Demarcation membrane system (DMS) stiffness�� 139, 141
imaging��195–213 Epigenetic variation��256
quantification��195–213 Epiphyses��65, 144, 151, 199
Denaturing loading buffer�� 114, 115, 117, 120, 122 Epon (epon resin)�� 59, 66, 68, 70,
Dense granule release��269 148, 221, 223
Dense granules�� 19, 33, 262 Extracellular matrix (ECM)�� 177, 178, 189, 191, 233
Dense tubular system (DTS)��74, 258, 262, 267–269 proteins��233
Depletion of intracellular Ca2+ stores��93
Diacylglycerol (DAG)�� 258, 268 F
Diamond antifade��20
F-actin�� 27, 43, 50
Diamond knife�� 59, 66, 77, 222, 225
Factor
di-8-ANEPPS��198, 201, 202, 204, 210, 212
VIIa��272
Diffraction limited microscopy��38
XIa��272
3,3'-Dihexyloxacarbocyanine iodide (DiOC6)��84–87
XIIa��272
2-D array scanning��212
Farnesylated��209
3D hydrogel culture��139–153
Fatty acid��39, 83, 129, 270
3D modeling�� 177–192, 273
FcγRII receptors��25
3D reconstruction��57, 69, 75, 213, 218, 227–229
Feeder cells�� 156, 158
Dioleoyl phosphatidylcholine��127
Femoral�� 199, 210
1,2-Dioleoyl-sn-glycero-3-phosphocholine
Femur��58, 65, 67, 142, 144–145, 151,
(DOPC)��129–131, 133, 136
210, 211, 222, 241, 243, 246, 247
1,2-Dioleoyl-sn-glycero-3-phosphoethanolamine-N-(cap
Fentanyl�� 235, 240
biotinyl) (CAP-BIOTIN-PE)��129–131, 133, 136
FGF2�� 158, 161, 168
Direct Stochastic Optical Reconstruction Microscopy
Fibrin�� 271, 272
(dSTORM)��35, 36, 38, 40, 41, 43, 46–50
Fibrinogen��2, 15, 17–19, 23, 39, 41,
Dispase�� 159, 169
43, 47, 61–64, 70, 71, 84, 86, 87, 271
DRAQ5�� 200, 201, 211
Fibrinolysis��256
Drug development��268
Fibronectin�� 180, 181, 184
DyLight 800��97, 105–107, 109
Ficoll-Paque�� 158, 165, 166, 173, 179
DyLight 800 NHS Ester�� 97, 105
Fiducial marks��226
Dynamin I (I/II)�� 18, 19
Fiji (Image J)�� 46, 51, 87, 238, 245, 249
E Filamin A��19
Filopodia��87
EC50��83 Finder grids��61, 63, 70, 77
EGFP�� 159, 172, 208 FITC-dextran�� 37, 159, 209
Eicosanoid��270 Fixation�� 22, 26, 36, 37, 43, 49, 50, 56–58,
Electron microscopy (EM) 65, 67, 68, 70, 71, 73, 77, 106, 110, 143–144, 147–148,
back-scattered electron (BSE) detector�� 218, 223 196, 209, 220–223, 229, 235, 238–241, 251
scanning��56, 196, 212, 217–230 Fixative
secondary electron detector��218 glutaraldehyde��56, 60, 69, 73
transmission�� 56, 57, 61, 65, 140, 196, 217, 229 143, 147, 221
Electron tomography (ET)�� 55–77, 217 paraformaldehyde��25, 26, 60, 143, 144, 148, 239
Electrophysiological/electrophysiology�� 196, 208, 263 Flow cell�� 5, 132–134
Emax2��89 Flow chamber�� 5, 134
284 Index

Flow cytometry (cytometer/cytometric)�� 1, 96–98, GPIb�� 209, 262
100, 107, 113, 131, 132, 140, 144, 149, 152, 157, 159, GPIbα�� 209, 272
163–166, 170, 173, 174, 269 Gp-IX��235, 242, 244, 246, 250, 251
Flow rate�� 9, 20, 109, 191, 239, 240 G protein coupled receptors (GPCRs)��93, 257, 262,
Fluo-3��8 263, 267, 268
Fluo-4��4, 5, 7, 8 GPVI�� 38, 39, 84, 263, 265, 270
Fluorescence recovery after photo bleaching Gq��269
(FRAP)�� 134, 135 Granules�� 13, 16, 29, 33, 38, 218, 228, 229,
Fluorescent beads�� 49, 173 258, 262, 269–271
Fluorescent cell barcoding (FCB)��96, 98–106, 109 Grey platelet syndrome��33
Fluorescent proteins��35, 37, 39, 50 GYPA (CD235a)�� 159, 163
Fluorophore�� 25, 28, 29, 35–37, 40, 41, 44,
50, 51, 60, 66, 109, 210, 238, 239, 245, 250, 251 H
FM 1–43��198, 201–205, 210 HALO tags��50
FM 2–10��202 “Hanging drop” hybridization method��28
FM 4–64��210 Heavy metal contrast�� 66, 230
Focal plane�� 37, 40, 44, 49–51, 203, 207, 208, 251 Heavy metal staining�� 76, 229
Focused Ion beam (FIB) milling��72, 76, 225, 227 Hematopoietic differentiation��174
Focused ion beam-scanning electron microscopy Hematopoietic Progenitor Cell isolation Kit��142
(FIB-SEM)�� 196, 217–231 Hematopoietic progenitors�� 140, 178
Formalin solution�� 43, 83, 85 Hematopoietic stem cells (HSCs)
Formvar grids�� 60, 61, 66, 70, 71, 76 from adult peripheral blood�� 178, 179, 181
Forward programming��155–175 from umbilical cord blood�� 178, 179, 181
Fourier ring correlation (FRC) method��52 Hematopoietic tissues��218
Fourier transform�� 45, 46, 51 Hemocytometer�� 161, 165–167
Freeze substitution (FS)��60, 67–69, 73, 76 Hemostasis�� v, 56, 81, 256, 257, 269, 270
Fura-2�� 8, 84, 90–94 Hepes-Tyrode solution��60
Fusion tags��39 Hepatic microsomes��268
Hermansky-Pudlak syndrome�� 33, 38
G
Heterotrimeric G proteins�� 258, 268, 269
Gai��263 High-content�� 85, 87
Gallium ion beam�� 218, 227 High-pressure freezing (HPF)�� 57, 58, 60, 65,
Gaq��263 67–70, 72–74, 76, 77
Gas��180, 226, 230, 263 High-throughput��38, 81–110, 275
GATA1, FLI1 and TAL1�� 156, 159 Hirudin�� 21, 142, 143
GCI[GPO]10GCOG��4 Hoechst�� 15, 16, 20, 26, 27, 198, 200,
Gelatin(e)�� 60, 66, 70 201, 204, 211, 212
Gelatin blocks��60 Homophilic interaction�� 2, 4
Gelatine��158 HPTS��198, 205, 206, 211
Genetic variation��256 Human pluripotent stem cells (hPSCs)��155–175
Genomic�� 171, 256, 270 Human Protein Atlas��28
Genomic integration��171 Humidifier apparatus��61
GFP�� 28, 37, 50, 63, 198, 210 Hyaluronan hydrogels�� 178, 180–181, 188–189, 191
Github��275 Hyaluronic acid��177
Glass bottomed dishes�� 39, 42, 43, 46, 48 Hybrid models��264
Glass coverslips�� 23, 42, 82, 132, 198, 211 Hydrogel
Glutaraldehyde (GA)��56, 60, 61, 64, 66, 69, fabrication��188–189
72, 73, 143, 147, 221 hyaluronan�� 180–181, 188–189
Glycolytic�� 259, 270 methylcellulose��140, 145–147, 149
GM130��19 polymerization��191
Gold (Au)-carbon-coated grids�� 61, 63, 64
Gold particles�� 63, 70, 71, 73, 77 I
Golgi�� 19, 218, 228 Ibidi slides��48
GP9 (CD42a)�� 156, 159, 163 Ibrutinib (btk inhibitor)�� 86, 87, 91
285
Index

Ibuprofen�� 6, 84 K
IC50�� 83, 89
Iloprost��274 KLF4��157
Image deconvolution��26–27, 238, 244 KnockOut Serum Replacement (KOSR)�� 158, 167
Image J (Fiji)�� 6, 7, 46, 51, 87, 92, 199,
L
204, 205, 207, 213, 238, 245, 249
Image registry correction�� 20, 27 LabView software��189
Image thresholding�� 30, 92 Lactate��271
Imaging�� v, 3–7, 13–30, 33–51, 56, 57, 62, 67, Lamellipodia��87
74, 77, 82, 84–87, 118, 119, 122, 123, 129, 134, 140, Laminin��158, 175, 180, 184
195–213, 217–230, 234, 236–238, 243–245, 249, 251 LAMP1��19
chamber�� 129, 198, 199, 236, 237, 243, 251 Laser�� 8, 35, 37, 40, 41, 44, 46, 48, 50, 51,
Imaris�� 21, 27, 30, 199, 213, 238, 246–249 97, 198, 201, 211, 212, 234, 236–238, 243, 245
Immobilized��3, 5, 7, 9 Laser fluorescence microscopy��13–30
Immunoassay��95 Lens
Immunoblotting�� 29, 95, 119, 122, 123 objective�� 37, 40, 41, 44, 62, 84, 85, 92,
Immuno-electron microscopy 198, 211, 236, 237
(Immuno-EM, IEM)�� 56, 73, 76 theta�� 236, 237
Immunofluorescence��56 tube�� 236, 237
Immunofluorescence microscopy (IFM)�� 14, 56, 63, 72 Lentiviral
Immunoglobulin (Ig) domains��2 transduction��171
Immunoglobulins (IgG)�� 20, 25, 28, 39, 97, vectors��156, 159, 161,
99, 100, 103, 107, 115, 118–123 167, 171, 172
Immuno-gold EM�� 33, 56 Lifeact-GFP��50
Immuno-gold labelling��71 Light sheet��234, 236, 237, 245
Immunolabeling��62, 63, 66, 71, 73, 77, Light-sheet fluorescence microscopy
144–145, 148, 149, 234–235, 238–241 (LSFM)��233–251
Immunological synapse��127 Light transmittance�� 88, 92
Immunomagnetic beads��182 Lin-�� 140, 142, 145
Immunomagnetic separation��116–117 Lin-bone marrow progenitors��141
Immunoprecipitation�� 115, 119 Lipid micelles��130
Immunoreceptor Tyrosine-based Inhibitory Motif Liposomes��128–133, 135, 136
(ITIM)��2 Liquid ethane��63, 64, 72, 73
Immunoreceptor Tyrosine-based Switch Motif Lithium Bromide (LiBr)��179, 182, 183, 190
(ITSM)��2 Low melting point agar�� 61, 143, 221
Induced pluripotent stem cells (iPSCs)�� 39, 155 Ly6G/C(Gr-1)�� 142, 145
Inositol-1,4,5-trisphosphate (IP3)��208, 258, 268, 269 Lyophilizer��130, 180, 186, 188
Inositol-1,4,5-trisphosphate (IP3) receptors
(IP3R)�� 258, 268, 269 M
Insulin-Transferrin-Selenium��171 Macrothrombocytes��27
Integrilin��86 Magnetic beads�� 115–117, 179
Integrin activation 4��258 streptavidin-coated�� 142, 145
Integrins Manders overlap coefficients��30
α2β1�� 15, 262 Mass action kinetics��260
αIIbβ3��38, 262, 267, 269 Mass spectrometry�� 95, 114
Interleukin (IL)-11�� 179, 182 Mathematical modelling��114, 256–266, 268,
Intravital microscopy��234 269, 272, 275, 276
Invaginated membrane system (IMS)��196 Matrigel�� 17, 24, 158
Ion channels�� 8, 9, 257 Matrix proteins�� 82, 233
IP3R��258, 263, 268, 269 mCherry��37
Isoflurane��235 Mean platelet volume (MPV)��57, 64, 65, 76
Isotopically labelled energy substrates��270 Mechanical stimulation��2
Isotype control��115 Mechanosensitive ion channels�� 8, 9
ITGA2B (CD41)�� 15, 19, 156, 157, 159, Medetomidin�� 235, 240
163, 164, 173, 208, 209
286 Index

Megakaryocyte MWReg30��209
culture��197 MYC��157
differentiation�� 139–153, 178, 181, 189 Myh9��14
3D culture��140 Myh9 knockout��140
electron microscopy��140, 143, 147, 152 MYH9 syndrome��27
light microscopy��225
native environment�� 217, 218 N
organelle�� 228, 229 Necrosis�� 258, 269
three dimensional ultrastructure��217 NIS-Elements Advanced Research software�� 40, 41
ultrastructure��196 Non muscle myosin IIA��18
Megakaryopoiesis��178, 233, 234, 249 Nucleotide�� 110, 270
Membrane blebbing�� 203–205, 213 Nucleus�� 18, 34, 200, 201, 204,
Membrane capacitance��208 211, 212, 228
Membrane-cytoskeleton��67 Numerical aperture (NA)�� 20, 26, 34, 35, 40,
Mesoderm�� 155, 156, 158, 161–163, 168, 169, 171 44, 46, 198, 211, 243, 251
Metallothionein labeling��229 Nyquist sampling�� 212, 213
Methacrylated hyaluronan (MeHA)
synthesis�� 188, 189, 191 O
Methacrylic anhydride�� 180, 188, 191
Objective
Methylcellulose (MC)�� 60, 64, 66, 67, 72,
numerical aperture�� 34, 35, 243
139–153, 210
Objective lens
Microscope, confocal�� 6–8, 29, 50, 84, 202,
oil immersion�� 20, 40
203, 212, 234, 236, 238, 243, 245, 251
OCT4��157
Microscopy�� 1, 2, 13–30, 33–52, 56, 57, 60,
Octyl β-d-glucopyranoside�� 128, 130
61, 65, 72, 97, 134, 140, 143, 147, 150, 152, 157, 196,
OP9��156
202, 203, 212, 217–230, 233–251
Open canalicular system (OCS)��258
Microtiter plate assays�� 82–85, 87, 88, 90, 93,
Optical clearing��233–251
98, 99, 101, 105
Optical sectioning��35, 37, 38, 234
Microtubule stabilizing buffer (MTSB)��36
Orai1��263
Midazolam�� 235, 240
Ordinary differential equations (ODE)�� 261,
Milling, ion beam��225
264–266, 269
Missing wedge�� 56, 74
Oregon Green 488 BAPTA-1��198, 205, 207, 211
Mitochondria�� 258, 262, 265, 269–271
Organellar membranes�� 201, 212
Mitochondrial matrix�� 258, 269
Organelles��56, 67, 72, 195, 201, 212, 217, 228, 229
Mitochondrial
Osmium tetroxide (OsO4)�� 56, 61, 69, 221, 223, 229
membrane potential��269
post-fixation��229
uniporter��258
Osmium-thiocarbohydrazide-osmium liganding
Mitochondrial permeability transition pore
(OTO)��229
(mPTP)�� 258, 269
Mitochondrion�� 69, 75, 258 P
Modified Tyrode’s buffer��39, 42, 97, 98
Modified Tyrodes-Hepes buffer (MTH)��114–117 Pacific Blue�� 100–104, 106, 107, 109
Monobiotinylated recombinant ligands��134 Pacific Blue Succinimidyl Ester�� 97, 99, 105
Mononuclear cells�� 173, 179, 181 Pacsin2��18
Morphology�� 13, 27, 37, 73, 82, 85, 87, PAR1�� 262, 265, 268–270, 274
140, 156, 157, 162, 166, 174, 178, 198, 200 Paracetamol��84
Mouse Hematopoietic Progenitor Cell isolation Paraformaldehyde (PFA)��17, 25, 26, 60, 66, 97,
Kit�� 142, 145 143, 144, 148, 235, 239
MRS2179 (P2Y1 antagonist)��89 PAR1 agonist
Multilobular nucleus�� 200, 201 SFLLRN��274
Multiplexed phosphoflow cytometry��95–110 PAR4 agonist
Multiplexing��96 AYPGKF��274
Multistrip technique��120 Partial differential equations (PDE)�� 261, 265, 266
Multivesicular bodies��16 Patch clamp��196
PDI��19
287
Index

Pearson’s correlation��38 Photoswitchable fluorescent proteins��35
Pearson’s correlation coefficient��30 Phototoxic��203
PECAM-1�� 2–6, 250 Piezo1��8
Perfect Focus system (PFS)�� 36, 40, 41 PI3kinase (PI3K)�� 91, 258, 263
Perfusion chamber��5 PI3 kinase inhibitor, LY-294002��159
Permeabilization��17, 25, 36, 49, 97, 99, Piranha solution�� 128, 132, 135
102, 106, 108, 203, 209, 235, 242 Pixel dwell-time��230
Permeabilization buffer�� 17, 97, 99, 108, 235, 242 Plasma membrane�� 14–16, 69, 197, 199–204,
Pervanadate�� 114, 116, 120 206, 208–210, 212, 213, 218, 228, 262, 268
PGH2-R��263 Plasma membrane calcium ATPase (PMCA)��258
PGH synthase��270 Plasminogen��39
PGI2-R��263 Plastic embedding��56–58, 68, 70, 73
Phalloidin�� 20, 26, 27, 39, 43 Platelet aggregometry��82
fluorescent��41 Platelet deposition�� 265, 266, 271, 273, 274
Pharmacological intervention��257 Platelet phosphoproteome��95
Pharmacological reagents��7 Platelet-rich plasma (PRP)�� 6–8, 21–24, 27, 42,
Pharmacologic screen��96 64, 67, 84, 87, 88, 90, 92, 96–98, 115, 116
Pharmacology�� 7, 82, 83, 96, 255–276 Platelets
Phosphate buffer��17, 39, 60, 61, 64–66, 72, activation (activatory factors)�� 2–4, 6, 24, 39,
97, 129, 142, 158, 179–181, 197, 235, 239 93, 255, 256, 258–260, 271, 272
Phosphate buffered saline (PBS)�� 17, 20, 22–26, adhesion�� 38, 56, 59, 64, 82, 86, 87,
39, 40, 43, 72, 83, 97, 99, 100, 103, 105, 107, 110, 256, 257, 265, 267, 271
129, 133, 134, 142–144, 148–150, 152, 158–160, aggregation�� 21, 24, 82, 258, 271–273
162, 163, 169, 170, 173, 175, 179–181, 185–189, computational modelling��36, 256, 271,
197, 235, 239–242 273, 275, 276
Phosphatidyl inositol (PIP)��268 energy metabolism�� 265, 270, 271
Phosphatidylinositol 4,5-bisphosphate function�� 81, 82, 92, 195, 256, 259, 265, 274
(PIP2; PI-4,5-P2)�� 208, 258, 268 immobilisation��1–9
Phosphatidylinositol 3,4,5-trisphosphate, individual�� 1, 4, 86, 87, 273, 274
(PIP3)�� 258, 268 inhibition (inhibitory factors)�� 91, 92, 255
Phosphatidylserine��128 mathematical modelling�� 114, 269
Phosphoflow cytometry��95–110 negative feedback pathways��256
Phosphoinositide (PI) metabolism��267 permeabilized�� 43, 96, 99, 102, 105, 109
Phosphoinositides��268 positive feedback pathways��256
Phosphoinositol-3-kinase (PI3K)�� 91, 258, 264 release into the blood stream��177
Phospholipase A��270 secretion��272
Phospholipase-C (PLC)�� 208, 258, 269 shape��271
Phospholipase-Cβ�� 267, 268 signaling pathways�� 35, 81, 96, 256, 257,
Phosphoprotein�� 38, 109 259, 265, 266, 270, 274
Phosphoprotein signalling��95 ultrastructure��56
Phosphoproteome��95 Platinum coating��218, 220, 226, 227
Phospho-proteomic data��114 Pleckstrin homology domain of PLC∂1
Phosphorylation��4, 95, 96, 100, 101, 105, (EGFP tagged)��208
109, 110, 113–124, 269 Plunge vitrification��73
Phosphorylation state��113–124 Pluripotency��155, 157, 158, 171
Phospho-specific antibody��115, 118, 119, 121 Podoplanin�� 38, 128, 258
Phosphotyrosine�� 39, 47 Point spread function (PSF)��41, 48, 49, 245
Photoactivatable fluorescent proteins��35 Polycistronic cassette��171
Photoactivated localization microscopy Polydimethylsiloxane (PDMS) block��186
(PALM)�� 35, 39, 40, 42, 43 Poly-ethylene oxide (PEO)�� 180, 184–186
Photobleach��204 Poly-L-lysine (Polylysine)�� 17, 23, 144
Photobleaching��8, 27, 29, 50, 134, 135, 204, 211, 212 Polyploidic�� 210, 212
Photodamage�� 27, 236 Polyploidization��140
Photon count�� 47, 206 Polyploidy��156
288 Index

Polyvinylidene difluoride (PVDF) membrane�� 115, 120 Resin�� 56, 65, 66, 68, 73, 148, 196, 218,
Post-fixation��26, 148, 229, 241 220, 221, 223–227, 229, 230
Post-labelling fixation��50 embedding��59–60, 65–66, 73, 220, 221, 223
Post-translational modifications (PTM)�� 95, 113 polymerization��66
Potassium ferrocyanide�� 221, 223, 229 Resolution��13–30, 33–39, 43–52, 56, 57,
Primary antibodies�� 4, 17, 25, 26, 29, 50, 61, 66, 72, 73, 75, 85, 92, 108–110, 175, 196, 201, 203,
97, 109, 116, 117, 119, 123 204, 209, 211–213, 217–231, 234, 243, 244, 246, 251
Procoagulant surface��128 axial�� 37, 211, 213
Progenitor cells��39, 140, 142, 145 lateral��135
ProLong Gold��20 spatial�� 34, 35, 50
Proplatelets�� 13, 27, 140, 141, 143, Resolution-weighted back projection algorithm��75
149–151, 196, 233, 243, 246, 249 Retinoic acid receptor A��19
Propylene oxide��59, 66, 221, 223 Reverse pipetting�� 88, 92
Prostacyclin (PGI2)�� 39, 42, 57, 65, 76, RFP��37
99–103, 106, 114, 116, 258, 264 RIPA�� 114, 116, 117
Prostaglandin��98, 105, 106, 110 ROBO1��18
Prostaglandin E2��274 Rock inhibitor Y-27632��158, 160, 172, 175
Protamine sulphate�� 158, 161, 167
Protease inhibitors cocktail�� 114, 116 S
Protein A gold�� 60–63, 66, 71 Sarco/endoplasmic reticulum Ca2+-ATPase
Protein A/G Sepharose magnetic beads��115 (SERCA)�� 19, 258, 268
Protein kinase A (PKA)�� 263, 269 Scanning electron microscope (SEM)��56, 217–230, 262
Protein kinase C (PKC)�� 263, 267, 268 SCF�� 158, 161
Protein kinase G (PKA)��263 Scientific complementary metal-oxide semiconductor
Protein kinases��95, 263, 268, 269 (sCMOS) camera�� 40, 41, 237
Protein phosphatases��95, 114, 119, 263 SDF-1α�� 180, 184
Protein quantification��113–124 SDS-PAGE gel�� 114, 117, 120, 122, 124
Proteomic��114, 262, 263, 270 Secondary antibodies�� 4, 17, 25, 26, 28, 29, 43,
Proteomic data��114 47, 50, 97, 109, 115, 117–123
Prothrombin��271 Secondary electron detector��218
PRT062607 (Syk inhibitor)��91 Secretory granules��270
P-selectin�� 14, 15, 18, 19, 25, 29 SERCA3��19
P2X1�� 8, 263 Serial block face EM��196
P2X1 receptors��8 Serial sectioning��218
P2Y12�� 262, 265, 269 Serine/threonine kinases��263
P2Y1 inhibitor��274 Serum, rat�� 142, 145
P2Y1 receptor�� 267, 268 Shape change�� 4, 258, 271
Shear��2, 3, 7, 8, 151, 168, 185, 190, 196, 273
Q
Shear rate�� 6, 7, 9, 191, 273
Quantifoil�� 61, 64, 70, 71, 76, 77 Silk fibroin protein��177
Quantum dots��229 Silk micro-tube��179–180, 184–186, 188
Simultaneous Iterative Reconstruction Technique
R (SIRT)��75
Rap1b��269 Single molecule localization microscopy
Reactive oxygen species��270 (SMLM)�� 34–38, 52
Receptor Singular perturbation theory��276
adhesion�� 267, 269 Sinusoid
GPVI�� 38, 39, 84, 263, 265, 270 bone marrow�� 233, 238, 243
PAR1�� 262, 265, 268–270, 274 SNAP��50
P2Y1��89, 262, 267–269, 274 Sodium-calcium exchanger (NCLX)�� 258, 269
P2Y12�� 262, 265, 269 Sodium citrate�� 6, 17, 21, 22, 39, 42, 57, 64,
Regulator of G-protein signalling (RGS)��260 76, 83, 84, 96–98, 114
Repel Polymer Technology��175 SOX2��157
289
Index

Spinning-disc confocal�� 14, 15 Transduction efficiency��161, 171, 172, 175
Spontaneous platelet activation��93 Transduction medium�� 158, 161, 162, 167, 172
Stenotic blood flow��273 Transfusion medicine��156
Stiffness, environmental��141 Transient receptor potential channel
Stimulated emission depletion (STED)��38 (TRPC)�� 258, 263
Stoichiometric analysis��270 Transmission electron microscopy (TEM)�� 56, 57,
Stokes shift�� 198, 210 61–65, 72–74, 76, 77, 140, 196, 229, 230, 262
Streptavidin�� 127, 129, 131, 134, 136, 142, 145 TRAP-6�� 4, 5, 8
Stromal cells��156 Tris Buffered Saline��115
Structured illumination microscopy (SIM)�� 14, 20, 0.5% Triton X-100��235
27, 28, 33–52 Triton X-100�� 17, 25, 27, 43, 97, 108
Styryl dyes��197, 198, 201, 203, 205, 210–212 TrypLE (trypsin replacement)�� 158, 160,
Sub cellular networks��273 163, 164, 172, 173, 175
Subendothelial proteins��272 Tubulin�� 14, 16, 18, 19, 27, 29, 36, 121
Superresolution imaging/microscopy�� 33, 34, 38, 39, TULA-2�� 263, 270
44, 46, 50 Two (2)-photon (multi-photon)
“Blinking” data set�� 35, 36 microscopy�� 207, 229, 234
Syk��89, 91, 109, 121, 263, 270 Tyrode’s buffer��24, 39, 42, 43, 57, 65,
Synthetic peptides��2 83–85, 88, 90, 97, 98
Systems biology�� v, 256, 275 Tyrode’s-Hepes buffer�� 4–7, 60, 61, 71, 114
Systems Biology Markup Language (SBML)��275 Tyrosine kinase-dependent signalling��269
Tyrosine kinases�� 89, 257, 258, 263, 269
T Tyrosine phosphatase��270
Tannic acid��229 Tyrosine phosphorylation�� 4, 43
TER119�� 7–4, 142, 145
U
Tethering�� 3, 272
Three dimensional (3D)�� 15, 16, 21, 29, 30, U46619 (Thromboxane A2 receptor agonist)�� 17, 24,
35, 37, 40, 41, 44–48, 50, 52, 56, 57, 63, 64, 69, 74, 75, 258, 274
139–153, 177–191, 213, 233, 234, 243, 245, 248–250, Ultramicrotome��59, 60, 66,
262, 265, 266, 273 220–222, 225, 226
environment�� 139, 233, 249 Ultrastructure/ultrastructural morphology�� 33, 56,
imaging��35, 217–230, 234, 243 57, 61–63, 65, 72, 140, 141, 156, 196, 217, 218, 229
Thrombin�� 2, 258, 266–269, 271–273 Umbilical cord blood�� 178, 179, 181
Thrombin receptor activatory peptide (TRAP)�� 4, 5, Uniporter�� 258, 269
8, 76, 98 Uranyl acetate (UA)�� 60, 61, 64, 66–70,
Thrombopoiesis�� v, 13, 195, 196 72, 76, 221, 223
Thrombopoietin (TPO)��142, 143, 156, Uranyl oxalate��60, 64, 67, 72
158, 161, 179–182, 187, 189, 210
Thrombosis��v, 56, 81, 256, 274 V
Thrombospondin 1�� 18, 19, 29 Vascular damage�� 255, 271
Thromboxane�� 17, 258, 265, 270, 272 Vasodilator-stimulated phosphoprotein
Thrombus��266, 267, 271–274 (VASP)�� 96–101, 103–109
Tibia�� 144–145, 151, 210, 211 Venous sinusoids��196
Tibial��199 Vinculin��38
Tissue culture��17, 23, 24, 28, 131, 159–161, Vitrification�� 57, 64, 68, 72–73, 76
163, 167, 170, 171 Vitronectin��158, 160, 170, 171
Tissue factor (TF)�� 159, 167, 272 Volocity�� 20, 27, 30
TMEM16f��263 von Willebrand Factor (vWF)�� 2, 16, 18, 19,
Tokuyasu method�� 57, 58, 60, 65, 66, 68, 73, 74, 77 25, 29, 39, 61–63, 68, 70–72, 265, 272
Toluidine blue��221–222, 224, 225
Total internal reflection fluorescence (TIRF)�� 35, 40, W
41, 46, 51, 134
Washed platelet preparation�� 23, 84
Transcription factors��155–175
Western blotting��109, 113–115, 117–119
290 Index

Wheat germ agglutinin�� 20, 26–28 Z
Whole mount electron tomography��70
Widefield imaging��34 Z drift correction��36
Zenodo��275
Y Z-plane�� 29, 44, 50
Z-section�� 14, 15, 29, 30,
Yamanaka’s factors��157
202, 211, 213
YFP�� 209, 210
Z-series�� 49, 203–205, 207, 211–213

Platelets and Megakaryocytes: Volume 4, Advanced Protocols and Perspectives

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Platelets and Megakaryocytes: Volume 4, Advanced Protocols and Perspectives

Uploaded by

Copyright:

Available Formats

Methods in

Molecular Biology 1812

For further volumes:

Volume 4, Advanced Protocols and Perspectives

ISSN 1064-3745 ISSN 1940-6029 (electronic)

Library of Congress Control Number: 2018949611

© Springer Science+Business Media, LLC, part of Springer Nature 2018

Printed on acid-free paper

Reading, UK Jonathan M. Gibbins

12 Fluorescence Approaches to Image and Quantify the Demarcation

Vittorio Abbonante • Department of Molecular Medicine, University of Pavia, Pavia,

Katrin G. Heinze • Rudolf Virchow Center for Experimental Biomedicine, University

Jean-Yves Rinckel • Université de Strasbourg, INSERM, EFS GRAND EST, BPPS

Immobilization of Nonactivated Unfixed Platelets

Key words Immobilized Platelet, Calcium, ADP, Thrombin, Microscopy, Biochip

It is desirable, in many cases, to monitor the cellular processes in

with the limitation that platelets must be adhered to coverslips

membrane proximal Ig domain 6 of PECAM-1 [16–18]. These

2.2 Flow Chip 1. Tyrode’s-HEPES buffer—134 mM NaCl, 0.34 mM Na2HPO4,

2. Monoclonal mouse anti-human antibody against the first or

5. Nanopump to enable precision microfluidic flow at variable

The method outlined below is for imaging of Ca2+ transients in

3.1 Biochip 1. Coat glass-bottom biochips with monoclonal mouse anti-­

3.5 Imaging 1. Set up the confocal microscope to monitor platelet calcium

3.6 Inhibiting or 1. Pharmacological reagents may be introduced at a very low

1. Using this method, platelets can be stimulated by an array of

attached platelets can be studied using this protocol. For fur-

where Q = flow rate (ml/s), w = microslide lumen width in cm

12(8):1342–1352. https://doi.org/10.1111/ 1) that are required for the cellular association

Imaging Platelets and Megakaryocytes by High-Resolution

Microscopy has long played a key role in studies of platelets and

The application of fluorescence microscopy to studies of plate-

2.3 Cell Labeling 1. Donkey serum.

NONMUSCLE MYOSIN IIA 500 Biomedical technologies BT-567 Cytoplasm X X [12]

Imaris or other software depending on the analysis required

3.1 Isolation When handling platelets it is important to avoid some common

a­ liquoted into polystyrene or polypropylene tubes containing

3.2 Preparation Some electron microscopy protocols recommend fixation of plate-

1. Aliquot PRP (human or mouse) to half-fill microcentrifuge

3.3 Preparation 1. Prepare glass coverslips in 6-well culture plates as in Subheading

dry (a similar method can be used to coat coverslips with other

4. Draw off culture medium and fix immobilized cells with 4%

3.5 Fluorescence 1. Preblocking and permeabilization: Platelet and megakaryo-

megakaryocytes for 3–4 h at RT. After incubation rinse and

deconvolution, and registry correction (maximum one z-pixel

1. When macrothrombocytes are present in the sample, the cen-

a­nti-­human IgG. Mouse cells do not stain well with either

a­ntibody. For example, the rabbit anti-fibrinogen antibody

12. Presenting high-quality IFM images in publications can be

This work was supported in part by operating grants from the

Single-Molecule Localization and Structured Illumination

Key words Platelets, Superresolution, Single-molecule localization microscopy, SMLM, Direct

over EM, including the ability to easily label specific proteins of

apply these techniques to the study of the platelet cytoskeleton and

fluorophores. In addition, the use of z drift correction systems such

resolution allows for a robust interrogation of the spatial arrange-

limited than SMLM techniques, but Denmerle et al. [27] describe

1.3 Superresolution Recently a number of platelet studies have employed superresolu-

GPVI clustering in response to different substrates has shown that

15. Nonfluorescing aqueous mounting medium.

2.2 Hardware Particle count and size analyzer.

2.2.2 SIM SIM imaging is generally performed on commercial SIM micro-

2.2.3 STORM dSTORM imaging can be carried out on commercial STORM/

Reading, UK Jonathan M. Gibbins

3.1 Biochip 1. Coat glass-bottom biochips with monoclonal mouse anti-

a liquoted into polystyrene or polypropylene tubes containing

anti-human IgG. Mouse cells do not stain well with either

antibody. For example, the rabbit anti-fibrinogen antibody