Content uploaded by Thomas Joos
Author content
All content in this area was uploaded by Thomas Joos
Content may be subject to copyright.
Clinical Proteomics
METHODS IN MOLECULAR BIOLOGY
TM
John M. Walker,SERIES EDITOR
447. Alcohol: Methods and Protocols,edited by
Laura E. Nagy, 2008
446. Post-translational Modification of Proteins:
Tools for Functional Proteomics, Second Edition,
edited by Christoph Kannicht, 2008
443. Molecular Modeling of Proteins,edited by
Andreas Kukol, 2008
439. Genomics Protocols: Second Edition, edited by
Mike Starkey and Ramnanth Elaswarapu, 2008
438. Neural Stem Cells: Methods and Protocols,
Second Edition, edited by Leslie P. Weiner, 2008
437. Drug Delivery Systems,edited by Kewal K. Jain,
2008
436. Avian Influenza Virus,edited by Erica Spackman,
2008
435. Chromosomal Mutagenesis,edited by Greg Davis
and Kevin J. Kayser, 2008
434. Gene Therapy Protocols: Volume 2: Design and
Characterization of Gene Transfer Vectors edited
by Joseph M. LeDoux, 2008
433. Gene Therapy Protocols: Volume 1: Production
and In Vivo Applications of Gene Transfer Vectors,
edited by Joseph M. LeDoux, 2007
432. Organelle Proteomics,edited by Delphine Pflieger
and Jean Rossier, 2008
431. Bacterial Pathogenesis: Methods and Protocols,
edited by Frank DeLeo and Michael Otto, 2008
430. Hematopoietic Stem Cell Protocols,edited by
Kevin D. Bunting, 2008
429. Molecular Beacons: Signalling Nucleic Acid
Probes, Methods and Protocols, edited by Andreas
Marx and Oliver Seitz, 2008
428. Clinical Proteomics: Methods and Protocols,
edited by Antonia Vlahou, 2008
427. Plant Embryogenesis,edited by Maria Fernanda
Suarez and Peter Bozhkov, 2008
426. Structural Proteomics: High-Throughput Methods,
edited by Bostjan Kobe, Mitchell Guss, and Huber
Thomas, 2008
425. 2D PAGE: Volume 2: Applications and Protocols,
edited by Anton Posch, 2008
424. 2D PAGE: Volume 1:, Sample Preparation and
Pre-Fractionation, edited by Anton Posch, 2008
423. Electroporation Protocols,edited by Shulin Li,
2008
422. Phylogenomics,edited by William J. Murphy, 2008
421. Affinity Chromatography: Methods and
Protocols, Second Edition, edited by Michael
Zachariou, 2008
420. Drosophila: Methods and Protocols, edited by
Christian Dahmann, 2008
419. Post-Transcriptional Gene Regulation,edited by
Jeffrey Wilusz, 2008
418. Avidin-Biotin Interactions: Methods and
Applications, edited by Robert J. McMahon, 2008
417. Tissue Engineering, Second Edition, edited by
Hannsjörg Hauser and Martin Fussenegger, 2007
416. Gene Essentiality: Protocols and Bioinformatics,
edited by Svetlana Gerdes and Andrei L. Osterman,
2008
415. Innate Immunity,edited by Jonathan Ewbank and
Eric Vivier, 2007
414. Apoptosis in Cancer: Methods and Protocols,
edited by Gil Mor and Ayesha Alvero, 2008
413. Protein Structure Prediction, Second Edition,
edited by Mohammed Zaki and Chris Bystroff, 2008
412. Neutrophil Methods and Protocols,edited by
Mark T. Quinn, Frank R. DeLeo, and Gary M.
Bokoch, 2007
411. Reporter Genes for Mammalian Systems,edited
by Don Anson, 2007
410. Environmental Genomics,edited by Cristofre
C. Martin, 2007
409. Immunoinformatics: Predicting Immunogenicity
In Silico, edited by Darren R. Flower, 2007
408. Gene Function Analysis,edited by Michael Ochs,
2007
407. Stem Cell Assays,edited by Vemuri C. Mohan,
2007
406. Plant Bioinformatics: Methods and Protocols,
edited by David Edwards, 2007
405. Telomerase Inhibition: Strategies and Protocols,
edited by Lucy Andrews and Trygve O. Tollefsbol,
2007
404. Topics in Biostatistics,edited by Walter T.
Ambrosius, 2007
403. Patch-Clamp Methods and Protocols,edited by
Peter Molnar and James J. Hickman 2007
402. PCR Primer Design,edited by Anton Yuryev, 2007
401. Neuroinformatics,edited by Chiquito J. Crasto,
2007
400. Methods in Membrane Lipids,edited by Alex
Dopico, 2007
399. Neuroprotection Methods and Protocols,edited
by Tiziana Borsello, 2007
398. Lipid Rafts,edited by Thomas J. McIntosh, 2007
397. Hedgehog Signaling Protocols,edited by Jamila I.
Horabin, 2007
396. Comparative Genomics, Volume 2,edited by
Nicholas H. Bergman, 2007
395. Comparative Genomics, Volume 1,edited by
Nicholas H. Bergman, 2007
394. Salmonella: Methods and Protocols, edited by
Heide Schatten and Abraham Eisenstark, 2007
393. Plant Secondary Metabolites,edited by Harinder
P. S. Makkar, P. Siddhuraju, and Klaus Becker,
2007
392. Molecular Motors: Methods and Protocols, edited
by Ann O. Sperry, 2007
391. MRSA Protocols,edited by Yinduo Ji, 2007
390. Protein Targeting Protocols Second Edition,
edited by Mark van der Giezen, 2007
389. Pichia Protocols, Second Edition, edited by James
M. Cregg, 2007
388. Baculovirus and Insect Cell Expression
Protocols, Second Edition, edited by David W.
Murhammer, 2007
387. Serial Analysis of Gene Expression (SAGE):
Digital Gene Expression Profiling, edited by Kare
Lehmann Nielsen, 2007
386. Peptide Characterization and Application
Protocols, edited by Gregg B. Fields, 2007
385. Microchip-Based Assay Systems: Methods and
Applications, edited by Pierre N. Floriano, 2007
METHODS IN MOLECULAR BIOLOGY
TM
Clinical Proteomics
Methods and Protocols
Edited by
Antonia Vlahou
Biomedical Research Foundation,
Academy of Athens, Athens, Greece
Editor
Antonia Vlahou
Academy of Athens
Biomedical Research Foundation
Athens, Greece
Athens 115 27
e-mail: vlahoua@bioacademy.gr
Series Editor
John M. Walker
School of Life Sciences
University of Hertfordshire
Hatfield, Herts., AL10 9AB
UK
ISBN: 978-1-58829-837-9 e-ISBN: 978-1-59745-117-8
Library of Congress Control Number: 2007939413
©2008 Humana Press, a part of Springer Science+Business Media, LLC
All rights reserved. This work may not be translated or copied in whole or in part without the written
permission of the publisher (Humana Press, 999 Riverview Drive, Suite 208, Totowa, NJ 07512 USA),
except for brief excerpts in connection with reviews or scholarly analysis. Use in connection with any form
of information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed is forbidden.
The use in this publication of trade names, trademarks, service marks, and similar terms, even if they are
not identified as such, is not to be taken as an expression of opinion as to whether or not they are subject to
proprietary rights.
While the advice and information in this book are believed to be true and accurate at the date of going to
press, neither the authors nor the editors nor the publisher can accept any legal responsibility for any errors
or omissions that may be made. The publisher makes no warranty, express or implied, with respect to the
material contained herein.
Printed on acid-free paper
987654321
springer.com
Preface
Clinical proteomics has rapidly evolved over the past few years and is
continuously growing as new methodologies and technologies emerge. In
this volume, leading researchers in the field have contributed their state-
of-the-art methodologies on protein profiling and identification of disease
biomarkers in tissues, microdissected cells, and body fluids. Experimental
approaches involving application of two-dimensional electrophoresis, multi-
dimensional liquid chromatography, SELDI/MALDI mass spectrometry and
protein arrays, as well as the bioinformatics and statistical tools pertinent to
the analysis of proteomics data are described. As stated in the introductory
chapter by Prof. Paik, the Vice President of the Human Proteome Organi-
zation, “clinical proteomics needs the integration of biochemistry,pathology,
analytical technology,bioinformatics,and proteome informatics to develop
highly sensitive diagnostic tools for routine clinical care in the future.” The
multi-disciplinary character of clinical proteomics approaches is evident in the
detailed step-by-step protocols described in this volume, which makes them
of potential use to a wide range of researchers, including clinicians, molecular
biologists, chemists, bioinformaticians, and computational biologists.
Antonia Vlahou
v
Acknowledgments
The editor gratefully acknowledges all contributing authors for their
collaboration, which made this project possible and brought it into fruition; the
series editor, Prof. John Walker, whose help and guidance have been instru-
mental; Mr. Patrick Marton, Mr. David Casey, and the whole production team
at Humana headed by the late Mr. Tom Laningan for making an excellent
production of this book.
vii
Contents
Preface............................................................... v
Acknowledgments .................................................... vii
Contributors.......................................................... xiii
1. Overview and Introduction to Clinical Proteomics ................. 1
Young-Ki Paik, Hoguen Kim, Eun-Young Lee,
Min-Seok Kwon, and Sang Yun Cho
Part I: Specimen Collection for Clinical Proteomics
2. Specimen Collection and Handling: Standardization of Blood
Sample Collection .............................................. 35
Harald Tammen
3. Tissue Sample Collection for Proteomics Analysis.................. 43
Jose I. Diaz, Lisa H. Cazares, and O. John Semmes
Part II: Clinical Proteomics by 2DE and Direct
MALDI/SELDI MS Profiling
4. Protein Profiling of Human Plasma Samples
by Two-Dimensional Electrophoresis ........................... 57
Sang Yun Cho, Eun-Young Lee, Hye-Young Kim, Min-Jung
Kang, Hyoung-Joo Lee, Hoguen Kim, and Young-Ki Paik
5. Analysis of Laser Capture Microdissected Cells
by 2-Dimensional Gel Electrophoresis .......................... 77
Daohai Zhang and Evelyn Siew-Chuan Koay
6. Optimizing the Difference Gel Electrophoresis (DIGE)
Technology .................................................... 93
David B. Friedman and Kathryn S. Lilley
7. MALDI/SELDI Protein Profiling of Serum
for the Identification of Cancer Biomarkers......................125
Lisa H. Cazares, Jose I. Diaz, Rick R. Drake, and O. John Semmes
8. Urine Sample Preparation and Protein Profiling
by Two-Dimensional Electrophoresis and Matrix-Assisted Laser
Desorption Ionization Time of Flight Mass Spectroscopy ........ 141
Panagiotis G. Zerefos and Antonia Vlahou
ix
x Contents
9. Combining Laser Capture Microdissection and Proteomics
Techniques .................................................... 159
Dana Mustafa, Johan M. Kros, and Theo Luider
Part III: Clinical Proteomics by LC-MS Approaches
10. Comparison of Protein Expression by Isotope-Coded Affinity
Tag Labeling ................................................... 181
Zhen Xiao and Timothy D. Veenstra
11. Analysis of Microdissected Cells by Two-Dimensional
LC-MS Approaches.............................................193
Chen Li, Yi-Hong, Ye-Xiong Tan, Jian-Hua Ai,
Hu Zhou, Su-Jun Li, Lei Zhang, Qi-Chang Xia,
Jia-Rui Wu, Hong-Yang Wang, and Rong Zeng
12. Label-Free LC-MS Method for the Identification of Biomarkers ..... 209
Richard E. Higgs, Michael D. Knierman,
Valentina Gelfanova, Jon P. Butler, and John E. Hale
13. Analysis of the Extracellular Matrix and Secreted Vesicle
Proteomes by Mass Spectrometry ............................... 231
Zhen Xiao, Thomas P. Conrads, George R. Beck, Jr.,
and Timothy D. Veenstra
Part IV: Clinical Proteomics and Antibody Arrays
14. Miniaturized Parallelized Sandwich Immunoassays ................ 247
Hsin-Yun Hsu, Silke Wittemann, and Thomas O. Joos
15. Dissecting Cancer Serum Protein Profiles Using
Antibody Arrays................................................263
Marta Sanchez-Carbayo
Part V: Statistics and Bioinformatics in Clinical
Proteomics Data Analysis
16. 2D-PAGE Maps Analysis..........................................291
Emilio Marengo, Elisa Robotti, and Marco Bobba
17. Finding the Significant Markers: Statistical Analysis
of Proteomic Data..............................................327
Sebastien Christian Carpentier, Bart Panis,
Rony Swennen, and Jeroen Lammertyn
18. Web-Based Tools for Protein Classification ........................ 349
Costas D. Paliakasis, Ioannis Michalopoulos, and Sophia Kossida
Contents xi
19. Open-Source Platform for the Analysis of Liquid
Chromatography-Mass Spectrometry (LC-MS) Data .............. 369
Matthew Fitzgibbon, Wendy Law, Damon May,
Andrea Detter, and Martin McIntosh
20. Pattern Recognition Approaches for Classifying Proteomic Mass
Spectra of Biofluids ............................................ 383
Ray L. Somorjai
Index..................................................................... 397
Contributors
Jian-Hua Ai •Eastern Hepatobiliary Surgery Hospital,Shanghai,China
George R. Beck, Jr •Division of Endocrinology,Metabolism and Lipids
Emory University,School of Medicine,Atlanta,GA
Marco Bobba •University of Eastern Piedmont,Department
of Environmental and Life Sciences,Alessandria,Italy
Jon P. Butler •Lilly Corporate Center,Indianapolis,IN
Sebastien Christian Carpentier •Faculty of Bioscience Engineering,
Division of Crop Biotechnics,K.U. Leuven,Leuven,Belgium
Lisa H. Cazares •The George L. Wright Jr. Center for Biomedical
Proteomics Eastern Virginia Medical School,Norfolk,VA
Sang Yun Cho •Yonsei Biomedical Proteome Research Center,Department
of Biochemistry,College of Sciences,Seoul,Korea
Thomas P. Conrads •Laboratory of Proteomics and Analytical
Technologies SAIC-Frederick,Inc.,National Cancer Institute at Frederick,
Frederick,MD
Andrea Detter •Fred Hutchinson Cancer Research Center,Seattle,WA
Jose I. Diaz •Cancer Therapy Research Center’s Institute for Drug
Development,University of Texas,Health Science Center,San Antonio,TX
Rick R. Drake •Eastern Virginia Medical School,Norfolk,VA
Matthew Fitzgibbon •Fred Hutchinson Cancer Research Center,
Seattle,WA
David B. Friedman •Proteomics Laboratory,Mass Spectrometry Research
Center,Department of Biochemistry,Vanderbilt University School
of Medicine,Nashville,TN
Valentina Gelfanova •Lilly Corporate Center,Indianapolis,IN
John E. Hale •Lilly Corporate Center,Indianapolis,IN
Richard E. Higgs •Lilly Corporate Center,Indianapolis,IN
Yi-Hong •Eastern Hepatobiliary Surgery Hospital,Shanghai,China
Hsin-Yun Hsu •Biochemistry Department NMI Natural and Medical
Sciences Institute at the University of Tuebingen,Reutlingen,Germany
Thomas O. Joos •Biochemistry Department,NMI Natural and Medical
Sciences Institute at the University of Tuebingen,Reutlingen,Germany
Min-Jung Kang •Yonsei Biomedical Proteome Research Center,
Department of Biochemistry,College of Sciences,Seoul,Korea
xiii
xiv Contributors
Hoguen Kim •Department of Pathology,College of Medicine,Yonsei
University,Seoul,Korea
Hye-Young Kim •Yonsei Biomedical Proteome Research Center,
Department of Biochemistry,College of Sciences,Seoul,Korea
Michael D. Knierman •Lilly Corporate Center,Indianapolis,IN
Evelyn Siew-Chuan Koay •Department of Pathology,Yong Loo Lin
School of Medicine,National University of Singapore,and Molecular
Diagnosis Center, Department of Laboratory Medicine. National University
Hospital,Singapore
Sophia Kossida •Division of Biotechnology,Biomedical Research
Foundation,Academy of Athens,Athens,Greece
Johan M. Kros •Department of Pathology,Josephine Nefkens Institute
Erasmus Medical Center,Rotterdam,The Netherlands
Min-Seok Kwon •Yonsei Biomedical Proteome Research Center,
Department of Biochemistry,College of Sciences,Seoul,Korea
Jeroen Lammertyn •Faculty of Bioscience Engineering,Division
of Mechatronics,Biostatistics and Sensors,K.U. Leuven,Leuven,Belgium
Wendy Law •Fred Hutchinson Cancer Research Center,Seattle,WA
Eun-Young Lee •Yonsei Biomedical Proteome Research Center,
Department of Biochemistry,College of Sciences,Seoul,Korea
Hyoung-Joo Lee •Yonsei Biomedical Proteome Research Center,
Department of Biochemistry,College of Sciences,Seoul,Korea
Chen Li •Research Center for Proteome Analysis,Institute of Biochemistry
and Cell Biology,Shanghai Institutes for Biological Sciences,Chinese
Academy of Sciences,Shanghai,China
Su-Jun Li •Research Center for Proteome Analysis,Institute of
Biochemistry and Cell Biology,Shanghai Institutes for Biological Sciences,
Chinese Academy of Sciences,Shanghai,China
Kathryn S. Lilley •Cambridge Centre for Proteomics,Department
of Biochemistry,University of Cambridge,United Kingdom
Theo Luider •Laboratories of Neuro-Oncology/Clinical and Cancer
Proteomics,Josephine Nefkens Institute Erasmus Medical Center,
Rotterdam,The Netherlands
Emilio Marengo •Department of Environmental and Life Sciences,
University of Eastern Piedmont,Alessandria,Italy
Damon May •Fred Hutchinson Cancer Research Center,Seattle,WA
Martin McIntosh •Fred Hutchinson Cancer Research Center,Seattle,WA
Ioannis Michalopoulos •Biomedical Research Foundation,Academy
of Athens,Athens,Greece
Dana Mustafa •Department of Pathology,Josephine Nefkens Institute
Erasmus Medical Center,Rotterdam,The Netherlands
Contributors xv
Young-Ki Paik •Department of Biochemistry,Yonsei Proteome Research
Center & Biomedical Proteome Research Center,Seoul,Korea
Costas D. Paliakasis •Biomedical Research Foundation,Academy
of Athens,Athens,Greece
Bart Panis •Faculty of Bioscience Engineering,Division of Crop
Biotechnics,K.U. Leuven,Leuven,Belgium
Elisa Robotti •Department of Environmental and Life Sciences, University
of Eastern Piedmont,Alessandria,Italy
Marta S
.anchez-Carbayo •Tumor Markers Group,Spanish National
Cancer Center (CNI0),Madrid,Spain
O. John Semmes •The George L. Wright Jr. Center for Biomedical
Proteomics,Eastern Virginia Medical School,Norfolk,VA
Ray L. Somorjai •Biomedical Informatics Institute for Biodiagnostics,
National Research Council,Winnipeg, Manitoba,Canada
Rony Swennen •Faculty of Bioscience Engineering,Division of Crop
Biotechnics,K.U. Leuven,Leuven,Belgium
Harald Tammen •Digilab BioVisioN GmbH,Hannover,Germany
Ye-Xiong Tan •Eastern Hepatobiliary Surgery Hospital,Shanghai,China
Timothy D. Veenstra •Laboratory of Proteomics and Analytical
Technologies,SAIC-Frederick,Inc.,National Cancer Institute at Frederick,
Frederick,MD
Antonia Vlahou •Division of Biotechnology,Biomedical Research
Foundation,Academy of Athens,Athens,Greece
Hong-Yang Wang •Eastern Hepatobiliary Surgery Hospital,
Shanghai,China
Silke Wittemann •Biochemistry Department,NMI Natural and Medical
Sciences Institute at the University of Tuebingen,Reutlingen,Germany
Jia-Rui Wu •Research Center for Proteome Analysis,Institute of
Biochemistry and Cell Biology,Shanghai Institutes for Biological Sciences,
Chinese Academy of Sciences,Shanghai,China
Qi-Chang Xia •Research Center for Proteome Analysis,Institute of
Biochemistry and Cell Biology,Shanghai Institutes for Biological Sciences,
Chinese Academy of Sciences,Shanghai,China
Zhen Xiao •Laboratory of Proteomics and Analytical Technologies,
SAIC-Frederick,Inc.,National Cancer Institute at Frederick,
Frederick,MD
Rong Zeng •Research Center for Proteome Analysis,Institute of
Biochemistry and Cell Biology,Shanghai Institutes for Biological Sciences,
Chinese Academy of Sciences,Shanghai,China
Panagiotis G. Zerefos •Division of Biotechnology,Biomedical Research
Foundation,Academy of Athens,Athens,Greece
xvi Contributors
Daohai Zhang •Molecular Diagnosis Center Department of Laboratory
Medicine,National University Hospital, Singapore and Department of
Pathology,Yong Loo Lin School of Medicine,National University of
Singapore,Singapore
Lei Zhang •Research Center for Proteome Analysis,Institute of
Biochemistry and Cell Biology,Shanghai Institutes for Biological Sciences,
Chinese Academy of Sciences,Shanghai,China
Hu Zhou •Research Center for Proteome Analysis,Institute of Biochemistry
and Cell Biology,Shanghai Institutes for Biological Sciences,Chinese
Academy of Sciences,Shanghai,China
1
Overview and Introduction to Clinical Proteomics
Young-Ki Paik, Hoguen Kim, Eun-Young Lee, Min-Seok Kwon,
and Sang Yun Cho
Summary
As the field of clinical proteomics progresses, discovery of disease biomarkers becomes
paramount. However, the immediate challenges are to establish standard operating proce-
dures for both clinical specimen handling and reduction of sample complexity and to
increase the ability to detect proteins and peptides present in low amounts. The tradi-
tional concept of a disease biomarker is shifting toward a new paradigm, namely, that an
ensemble of proteins or peptides would be more efficient than a single protein/peptide
in the diagnosis of disease. Because clinical proteomics usually requires easy access to
well-defined fresh clinical specimens (including morphologically consistent tissue and
properly pretreated body fluids of sufficient quantity), biorepository systems need to be
established. Here, we address these questions and emphasize the necessity of developing
various microdissection techniques for tissue specimens, multidimensional fractionation
for body fluids, and other related techniques (including bioinformatics), tools which could
become integral parts of clinical proteomics for disease biomarker discovery.
Key Words: biomarker; body fluids; clinical proteomics; translational proteomics;
depletion; biorepository; multidimensional fractionation; specimen bank; biomarker panel.
Abbreviations: CSF: Cerebrospinal Fluid, SILAC: Stable Isotope Labeling with
Amino acids in Cell culture, FFE: Free Flow Electrophoresis, IMAC: Immobilized Metal
Affinity Chromatography, 2DE: 2-dimensional Gel electrophoresis, CBB: Coomassie
Brilliant Blue, SELDI: Surface-Enhanced Laser Desorption/Ionization, MALDI: Matrix-
Assisted laser desorption/ionization, MDLC: Multi-dimensional Liquid Chromatography,
LC: Liquid Chromatography, TOF: Time-of-Flight, CID: Collision-induced dissociation,
ETD: Electron Transfer Dissociation, LIT: Linear Ion-Trap, FT: Fourier-Transform, Q:
Quadrupole, ELISA; Enzyme-Linked Immunosorbent Assay, SISCAPA: Stable Isotope
Standards with Capture by Anti-Peptide Antibody, AQUA: Absolute Quantitative
From: Methods in Molecular Biology, vol. 428: Clinical Proteomics: Methods and Protocols
Edited by: A. Vlahou © Humana Press, Totowa, NJ
1
2 Paik et al.
Analysis. Commercial brands are also shown: MARS; Multiple Affinity Removal System,
(Agilent, Palo Alto, CA, USA), EnchantTM: EnchantTM Multi-protein Affinity Separation
Kit (Pall Life Sciences, Ann Arbor, MI, USA), GradiflowTM: GradiflowTM Separation (Life
Bioprocess, Frenchs Forest, Australia), FFETM: BD Free Flow Electrophoresis System
(BD Diagnostics, Martinsried/Planegg, Germany), Zoom®: Zoom®Benchtop Proteomics
System (Invitrogen Corporation, Carlsbad, CA, USA), Rotofor: Bio-Rad Rotofor®Prep
IEF Ccll (Bio-Rad, Hercules, CA, USA), PF2D: ProteomeLabTM PF2D Protein Fraction-
ation System (Beckman Coulter, Inc., Fullerton, CA, USA), DIGE: EttanTM DIGE System
(GE Healthcare Bio-Sciences AB, Uppsala, Sweden), Deep PurpleTM: Deep PurpleTM Total
Pprotein Stain (GE Healthcare Bio-Sciences AB, Uppsala, Sweden), ICATTM: Isotope-
coded affinity tags (Applied Biosystems, Foster City, CA, USA), iTRAQTM: iTRAQTM
Reagents (Applied Biosystems, Foster City, CA, USA), Q-TRAPTM: (Applied Biosystems,
Foster City, CA, USA).
1. Overview and Scope of Clinical Proteomics
Clinical proteomics is defined as comprehensive studies of qualitative and
quantitative profiling of proteins (and peptides) present in clinical specimens
such as body fluids and tissues. The comparison of specimens from healthy and
diseased individuals may lead to the discovery of a disease biomarker (1). The
biomarker serves as a molecular signature reflecting stages of disease before or
after treatment and can also be used for prognostic purposes in monitoring the
response to treatment (2). Clinical proteomics consists of a variety of exper-
imental processes, which include the collection of well-phenotyped clinical
specimens, analysis of proteins or peptides of interest, data interpretation, and
validation of proteomics data in a clinical context (Fig. 1). After successful
identification of a few disease biomarker candidates through extensive profiling,
Fig. 1. Clinical and translational proteomics. The key components of experimental
methods are included in each box.
Overview and Introduction to Clinical Proteomics 3
translational proteomics involving validation with a cohort study follows. Even
after proper identification and verification of a disease biomarker, it takes quite
a long time to prove that this biomarker is applicable to clinical diagnosis or
prognosis (3,4).
There has been a remarkable increase in publication of clinical proteomics
papers within a short period of time [more than 800 papers in 2006 (Fig. 2)],
coinciding with the rapid growth of proteomics. Reflecting this trend in clinical
proteomics, this chapter aims to present a review of core technologies that
are used in the field of clinical proteomics with respect to sample specimen
processing, protein separation platforms (e.g., gel-based system or liquid-based
methods), quantitative labeling, mass spectrometry (MS), and proteome infor-
matics tools. It is noteworthy that despite the advent of new technologies,
there remain several bottlenecks in the proteomics field such as lack of dataset
standardization, quantification of the proteins of interest, verification of protein
or peptides identified, and an overall strategy for tackling biomarker post-
identification. Thus, the pace of biomarker discovery, one of the key agendas of
clinical proteomics, will depend on how well these obstacles or bottlenecks are
resolved by technical advancement (4). The following sections address these
issues in the context of clinical proteomics.
Fig. 2. Recent trends in clinical proteomics publications. The distribution of the
articles related to clinical proteomics listed in PubMed is shown here. The key words
used for searching articles are as follows: query (clinical[All Fields] OR ((“biological
markers”[TIAB] NOT Medline[SB]) OR “biological markers”[MeSH Terms] OR
biomarker[Text Word])) AND (“proteomics”[MeSH Terms] OR proteomics[Text
Word] OR proteomic[All Fields] OR “proteome”[MeSH Terms] OR proteome[Text
Word]).
4 Paik et al.
2. Sample Specimens and Processing Techniques Used for Clinical
Proteomics
2.1. General Considerations
Because clinical proteomics rely heavily on the patient specimens, three
important factors need to be considered before the selection and preparation of
clinical specimens: (1) selection of the correct clinical samples according to the
type of research, (2) isolation of the appropriate component from the clinical
samples, and (3) establishment of optimal experimental conditions for each
sample (5,6,7,8). For the selection of correct clinical samples, the relationship
between clinical samples and the specific disease should also be considered.
For example, although cancer tissue represents a specific cancer, several types
of body fluids from patients may also have a relationship to the cancer. If
the selected clinical samples specifically represent the disease, the next step
is to evaluate what components are related to the specific disease. That is,
tumor cells in cancerous tissues are surrounded by many types of stromal cells,
inflammatory cells, and connective tissues that are directly related to changes
in protein expression in the cancer. If the purpose of proteomic analysis is
to identify characteristic changes of specific proteins in tumor cells, then the
precise identification of tumor cell percentage that can be increased by tissue
microdissection would appear to be necessary (5,6,7). As sample specimen
conditions directly impact the results of biomarker discovery, well-defined
clinical specimens should be used since the discovery of disease biomarkers is
much easier when the samples have clear anatomical and pathophysiological
definitions. Because clinical specimens are heterogeneous, sophisticated patho-
logical discrimination is required for the isolation of specific diseased tissue or
body fluids. Without the expertise of a pathologist at the earliest stage, it may
be difficult to isolate a specifically defined specimen for clinical proteomics.
Generally, clinical samples contain variable factors and components originating
from the microenvironment of specific tissues. For instance, liver tissues usually
contain a large amount of blood in the sinusoid and this amount is increased
in tissues with dilated sinusoids (9). Lung tissues usually contain deposited
exogenous materials and this amount is increased in heavy smokers (10). Note
that the amount of blood present in isolated tissues may directly influence the
relative proportion of proteins found in clinical specimens. Deposited materials
and the other chemicals such as stain dye and fixatives used in the microdis-
section may also influence the experimental conditions (11). In the analysis of
clinical samples, suitable buffer conditions, minimal lysis time, and high-yield
protein precipitation are highly recommended. To avoid substantial variations
between experiments using clinical specimens, a large set of specimens are
also necessary because, unlike cultured cell lines, clinical specimens have high
Overview and Introduction to Clinical Proteomics 5
component variability (12). More details on specific disease types are also
described throughout this volume.
2.2. Body Fluids
Surveying the literature, there appears to be five to six different types of
clinical specimens. Body fluids [e.g., plasma, urine, tear, cerebrospinal fluid,
lymph, and ascites], tissues (e.g., liver, heart, muscle, brain, and lung), cells,
bone, and hair have all been used for clinical proteomics (Table 1)(13,14,15,16,
17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33). Each has its own merits
and limitations for biomarker discovery via proteomic analysis. Among those
sample specimens, the number of publications using body fluids has increased
recently, perhaps because of their convenience and ease of use for noninvasive
diagnosis. Since those proteins secreted in the body fluids during or after disease
may reflect a broad range of pathophysiological conditions, much emphasis has
been given to identification of prominent protein/peptide biomarkers that exhibit
differential expression at different stages. In the literature, the terms “body
fluids” and “biofluids” are being used interchangeably, although the former
indicates a greater likelihood of being obtained directly from the patients, while
the latter is applied more broadly, referring to liquid or liquid-like samples
obtained from living organisms including model animals and plants. Throughout
this chapter we will use “body fluids” for clarity.
Given the large dynamic range of protein and peptide sources, plasma (a
complex liquid interface between tissues) and extra cellular fluids may be the
best body fluid to use for clinical proteomics and biomarker discovery (34,35,
36,37,38). In addition to plasma, more than a dozen additional body fluids are
currently used for biomarker discovery, ranging from urine to peritoneal fluids
(Table 1). However, the biggest challenge in body fluids proteomics may be the
multiple pretreatment processes including depletion of high-abundance proteins
(in the case of plasma) (34,35,36) and/or their enrichment (in the case of urine)
(15,39) prior to analysis (Table 1). Thus, the outcome of clinical proteomics
may depend on proper sample processing since the quality of selection and
handling of the most specific type of specimen will affect the overall pattern of
profiling. Because the details of body fluid proteomics have been well described
by Shen Hu et al.(38), we would like to focus on only a few essential points.
First, standard measures need to be introduced to protect specimens from
nonspecific proteolysis, lysis, and modification during collection and prepa-
ration (11). For the standardization of blood sample collection, Tammen
emphasizes many useful considerations of preanalytical variables in plasma
proteomics, which can be applied to processes involved with blood specimens
[(40) and see Chapter 2]. The more specific problems involved in sample
Table 1
Types of Biological Specimens Used in Clinical Proteomics
Type Disease Reference Characteristics of the
samples Pretreatment required
for proteomics
Fluid Secretions Plasma/serum (13,14) • Routinely accessible
body fluids
• Very important in the
discovery of biomarkers
of diseases (systemic
vs. organ specific/local)
• Important for early
detection, disease
severity, prognosis,
monitoring of response
to therapy
• Considerations for
sample adequacy
– Storage
– Hemolysis
– Influence of
anticoagulants
–Consistent results
• Consider whether to
pool samples or analyze
individual samples
• Depletion of
high-abundance proteins
(Albumin consist of
50% of plasma proteins)
Urine
Nasal discharge Prostate cancer
Seasonal allergic rhinitis (15)
(16)
Tears
Saliva Blepharitis and dry eye
Oral and breast cancer (17,18)
(19)
Amniotic-/cervical
fluid Fetal aneuploidy
and intra-amniotic
inflammation
(20,21)
Follicular fluid Recurrent spontaneous
abortion (22)
Seminal fluid
Nipple aspirate
fluid
Cerebrospinal
fluid
Male infertility
Breast cancer
Brain tumor
(23)
(24)
(25)
Proximal
fluid Synovial fluid
Ascites
Bronchial lavage
fluid
Rheumatoid arthritis
Ovarian cancer
Chronic obstructive
pulmonary disease,
asthmatics and lung
disease
(26)
(13)
(27,28)
• Can reflect disease
perturbations in the
organs or tissues from
which they are secreted
• Procedure of synovial
biopsy is not very
difficult
• Mucosa and salt
have to be removed
necessarily
Body
cavity
fluid
Pleural fluid
Peritoneal fluid Lung cancer
Ovarian cancer (29)
(14)
6
Tissue LCM or
LMPC
isolated
Formalin
fixed
Paraffin
embedded
Any type of disease (30) • Very important for the
development of novel
in situ biomarkers
• Immunofluorescence,
immunocytochemistry,
imaging mass
spectrometry
• Considerations for
sample adequacy
• Integrity, degradation
of protein
• Contamination
(microorganisms,
extraneous material)
Cell Cell lines
or
primary
tissue
culture
Any type of disease (31) • Very important in the
discovery of biomarker
candidates
• Validation should be
performed using
primary tumor samples
(e.g., immunohistologic
methods, imaging MS)
• Desalting and removal
of media component
Bone Cartilage Rheumatoid arthritis (32) • Cartilage consists
mainly of extracellular
matrix, mostly made
of collagens and
proteoglycans
• Cetylpyridinium
chloride effectively
aggregate with
proteoglycan
Hair (33) • Over 300 proteins
were found to constitute
the insoluble complex
formed by
transglutaminase
crosslinking
• Need to sufficient
extraction of protein
from insoluble complex
7
8 Paik et al.
handling are also addressed by Rai et al.(41). Second, to increase the dynamic
range of detection and reduce sample heterogeneity, pretreatments such as
depletion of high-abundance proteins appear to be required (34,35,36).In
addition, many pretreatment steps to remove high-abundance proteins may be
required during initial sample processing. Multiple fractionations of clinical
samples prior to major separation work would reduce the sample complexity.
Note that coremoval of low-abundance proteins during this type of multiple
depletion (36,42) and modification of proteins of interest during or after
isolation (43) should be considered as well. For several problems encountered
with specimen collection, Xiao et al. (Chapter 13) in this volume also describe
different methods to isolate extra cellular matrix (ECM) and analyze the
proteome of secreted vesicles. These methods will be useful for studying ECM
and secreted vesicles in various samples ranging from the primary cultured
cells to tissue specimens. Therefore, one must consider the best options for this
process before doing the main experiment.
2.3. Tissues and Other Samples
Usually tissues are used as primary screening samples to find direct causes
of disease from the lesion present in tissues of the corresponding organ, for
example, liver tissue in hepatocellular carcinoma (HCC) (44,45). Tissues are
widely used for clinical proteomics, although there are no standing operation
procedures in specimen fractionation and the detection limit of current instru-
mentation remains borderline. As listed in Table 1, many cancer tissues can be
prepared in different ways such as laser capture microdissection (LCM) (5,6),
pressures catapulting techniques [laser microdissection and pressure catapulting
(LMPC)] (30,46), and formalin-fixed paraffin-embedded sample preparation
(11). Theses techniques are well described in Chapters 3, 5, 9, and 11 in this
volume. It is desirable, however, that proteomics studies of disease tissues
should also be coupled with parallel analysis of the corresponding body fluids.
For example, for the study of cancer biomarkers, paired cancer tissue sets (tumor
vs. nontumor) and the same patient’s plasma were used, which led to a more
comprehensive analysis (47,48). Experiments on tissue samples may mostly be
suitable for pathophysiological studies rather than biomarker discovery due to
the complexity of the sample.
In specimen processing for proteomics studies, there are usually several
unwanted problems such as artifacts created during sample collection, processing,
and storage. Other matters arise in the handling of patient information regarding
sex, age, and race (49). To minimize those problems associated with systematic
sample handling, it is plausible to establish a specimen bank (50,51,52). In fact,
the collection of many clinical samples in a biorepository would have enormous
Overview and Introduction to Clinical Proteomics 9
benefits for proteomic research. This enables the selection of homogeneous
clinical samples according to the research purposes and isolation of specific
components from clinical samples. Additionally, large scale collection of clinical
specimens in a biorepository is essential for the validation of specific markers
after biomarker candidate discovery. Ideally, the clinical samples stored in the
biorepository should be (1) collected and stored immediately because dead cells
and altered proteins affect proteomic analysis, (2) subjected to accurate quality
control, and (3) catalogued by reliable and secure clinical data. The quality control
of clinical samples includes trimming of specimens and confirmation of diagnosis
by pathologists; information gained (such as the confirmation of tumor cell and
stromal cell ratio, percentage of necrosis, percentage of fibrosis, proportion of
infiltrated inflammatory cells, etc.) should be stored in a database of clinical
samples. It is also essential to store clinical and follow-up data for each sample
and each patient’s written informed consent form in the biorepository network.
This clinical specimen banking network provides convenience, reduced budget,
and reliability for researchers involved in clinical proteomic research (50,51,52).
For representative tissue sample collection for proteomics studies, Diaz et al.
(Chapter 3) address a practical experimental strategy for storage and handling of
sample specimens that are used in surface-enhanced laser desorption/ionization
(SELDI), 2D gel, and liquid chromatography (LC)-based proteomics. Emphasis
should be given to the primary responsibility of pathologists in the whole
process of tissue proteomics in addition to morphological analysis at the
molecular level.
3. Biomarker Discovery and Clinical Proteomics
Given that one of the central issues of clinical proteomics is biomarker
discovery and its application, a brief account of this subject is appropriate
here. An excellent review of the whole arena of biomarker development can be
found elsewhere (53,54,55). Until now, it has been generally accepted that a
conventional concept of a disease biomarker would be a single protein/peptide
with high specificity, which is usually present in low abundance, expressed in
a disease in a stage-specific manner, and serve as a major fingerprint of the
body’s response to drugs or other treatments. Although many examples of broad
biomarkers for various diseases are known (56,57,58,59,60), identification of
more specific and selective biomarkers is urgently needed. Accordingly, we
may also need to change the current biomarker concept and eliminate the
inherent bias toward individual disease biomarkers. Recently, a new idea has
been introduced that an ensemble of different proteins would be more efficient
than a single protein/peptide in the diagnosis of disease (61,62,63). To solve
10 Paik et al.
this problem we propose a general strategy of clinical proteomics leading to
disease biomarker discovery as outlined in Fig. 3.
Since biomarker candidate proteins could come from many different cellular
processes, they could be either in low abundance or high abundance, which
would directly or indirectly reflect the physiological condition of the body.
Perhaps they are present in different concentrations depending on the disease
stage or tissue type. For example, common proteins such as Hsp 27 (64,
65), 14-3-3 proteins (66,67), apoA-I (68,69), and serum amyloid precursor
A(70) appear in most of disease samples from lung cancer, gastric cancer,
pancreatic cancer, prostate cancer, neuroblastoma and, inflammation. A number
of questions then arise: should they be treated as disease-specific or disease
nonspecific proteins? What would be the criterion to make this decision? Is this
due to the fact that the number and type of proteins secreted from a specific
Fig. 3. The concept of the creation of a protein biomarker panel for a specific
disease. Each white, gray, dark-gray, and black circle represents a putative protein
biomarker of a specific disease at that clinical stage. A group of slash-lined circles
symbolizes the biomarker panel of liver disease as an example.
Overview and Introduction to Clinical Proteomics 11
physiological condition of many different types of diseases might be similar?
How one can distinguish one type of disease from another simply by looking
at their protein profiles?
As outlined in Fig. 3, at the beginning of certain disease, signals at earlier
stages may be limited to only a few easily counted molecules. As the disease
progresses, more signal molecules might have been produced, resulting in mixed
types of biomarkers representing multiple disease phenomena. Although this
assumption seems to be oversimplified, more noise is created at a certain stage
where it becomes more difficult to identify those molecules at the molecular
level because of two reasons: (1) they are in amounts too small to be detected
using the current technology and (2) it may be too premature for the molecules
to be specific for a particular disease. Presumably, proteins appearing in stage 3
or 4 may have higher specificity of a particular disease but the sensitivity might
be low. It may be likely that this noise interferes with the signaling pathway of
a certain disease, and we may end up having no decisive marker. To circumvent
this problem, it may be desirable to identify a set of biomarker candidate
proteins, termed a “biomarker panel,” which ideally contains potential candidate
proteins or peptides that represent specific stages of the disease as a group.
Given this panel, extensive validation processes may be sought using large
group cohort. Analogous to this strategy, many biomarker candidates at stage 1
can be included in the panel, which can have more specificity and sensitivity as
compared to a single molecule biomarker. Using this kind of biomarker panel,
one can use not only this molecule as diagnostic marker but also as a prognostic
indicator in monitoring treatment effectiveness. For example, Linkov et al.(61)
reported that both the sensitivity and specificity were improved up to 84.5 and
98%, respectively, when they used a panel containing 25 multimarkers in early
diagnosis of head and neck cancer (squamous cell cancer of the head and neck)
(61). In the diagnosis of prostate cancer, specificity was increased from 5–15
to 84–95% when they used a biomarker panel containing six marker proteins
as compared to a single marker. In HCC, studies have been carried out on a
biomarker panel consisting of a protein array that can be used as a diagnostic
kit (62,63).
A general strategy for biomarker discovery is outlined in Fig. 4. In typical
clinical proteomics, work sample collection is the first step, followed by
pretreatment of the sample in order to reduce sample complexity to enable
searching for low-abundance proteins (e.g., disease biomarkers) using various
fractionation tools. This multidimensional fractionation is well-described
elsewhere (34,35,36), and depends on the properties and concentration of the
sample. Typically the prefractionated samples go either to a two-dimensional
electrophoresis (2DE) or LC-based proteomics separation system, followed by
single or multiple steps of mass spectrometric analysis depending on the sample
Fig. 4.
12
Overview and Introduction to Clinical Proteomics 13
quantity and experimental goal. The data obtained from this series of analyses
will be integrated into the proteome informatics system where protein/peptide
identification, quantification, modification, and verification of peak list are
carried out [(71) and also Chapter 19]. Usually this step becomes rate limiting
since major profiling data are constructed and analyzed at this point. The
clinical relevance of those proteins (and changes in their expression level) in
a specific disease state is mostly determined, which eventually leads to identi-
fication of biomarker candidates. In addition, SELDI, molecular imaging and
protein microarrays can also be applied before or after this step. Once major
biomarker candidates are identified, those proteins are subjected to further
verification via sophisticated analytical arrays and translational proteomics,
which involves cohort studies, pre-evaluation, and a robust analytical system
(4,72). Throughout the process of translational proteomics, one may be able to
judge whether the identified panel or single proteins are suitable for biomarkers
of a specific disease. A recent comprehensive review by Zolg (73) addressed
several considerations in the biomarker development pipeline from discovery
to validation. Three critical challenges within the pipeline are reduction of
clinical sample complexity, the proof of principle of biomarker function, and
the detection limit of unique proteins present in the samples.
In the search for biomarker panels, reliable statistical tools and bioinfor-
matics resources are needed, which are now available on the web (Table 2;
see also Chapters 16 and 17). As the number of biomarker panel candidates
increases, more cases are being examined, which require statistical learning
methods. These methods include neural networks, genetic algorithms, k-means
Fig. 4. A typical experimental strategy for clinical proteomics and transla-
tional proteomics. In clinical proteomics research, various experimental techniques
are included: specimen collection, prefractionation, 2DE, Non2DE (liquid-based
separation), mass spectrometry, informatics, and others. The course of each section as
marked (square, circle in different color) is determined by the investigators, depending
on the experimental goal. At the bottom, experimental procedures for the verification
and validation of biomarker candidates are schematically outlined leading to clinical
screening and applications. The squares indicate the separation system based on the
specific characteristics of proteins and general prefractionation system. The open circles
and open triangle represent analytical modules at the protein and peptide level, respec-
tively. The arrow and junction points indicate an option of each selection. Bottom parts
indicate verification procedure employing multiple reaction monitoring and quantitative
mass analysis. Those biomarker candidates identified from typical clinical proteomics
would be subject to translational proteomics for validation where a large scale cohort
study and evaluation would then proceed.
14 Paik et al.
nearest-neighbor analysis, euclidean distance-based nonlinear methods, fuzzy
pattern matching, selforganizing mapping, and support vector machines
(74,75,76,77,78). They are very useful for classification of proteins according
to the specific disease state (see also Chapters 16 and 20). Once biomarker
candidates are identified, it is necessary to predict in silico the function of
these proteins and validate them in the context of clinical application. Table 3
provides web resources, which can be used for clinical data management, in
silico functional annotation (see Chapter 18), prediction, and identification of
modified forms of proteins. Thus, by combining experimental methods (Fig. 4)
and informatics tools (Tables 2 and 3), one is able to obtain a set of biomarker
candidate proteins (panel) that would be further used for validation through
translational proteomics (Fig. 1).
4. Introduction of the Experimental Strategy Described
in This Volume
For protein profiling and identification, proteomics platform technologies
are moving forward in many areas not only in clinical proteomics but also in
the general biological field. In this section, the leading scientists in the field
of proteomics outline core techniques and their application to the studies of
clinical proteomics. For example, in plasma proteome analysis, it is necessary
to deplete high-abundance proteins using various techniques such as multidi-
mensional fractionation by immunoaffinity column, gel permeation, and beads
(Fig. 4). Cho et al. (Chapter 4) addresses this in relation to 2D gel analysis of
plasma wherein the technical details of sample preparation, gel electrophoresis,
and quantification of proteins on the gel are described. Zhang and Koay
(Chapter 5) describe the methods of 2D gel analysis for cells prepared by
LCM. They describe the application of LCM in dissecting tumor cells in
breast cancer for macromolecular extraction and 2D gels. This can be used
for preparation of samples from paraffin-embedded tissue blocks in microdis-
secting the cells of interest. Further to this procedure, Mustafa et al. (Chapter 9)
review the application of LCM for proteomics analysis and demonstrate that
combining LCM and MS would facilitate identification of specific proteins
for each sample type. For urine sample analysis, Zerefos et al. (Chapter 8)
provide simple protocols for protein analysis by 2D gel or direct matrix-assisted
laser desorption/ionization-time-of-flight mass spectrometry. These techniques
include protein enrichment through protein precipitation and ultrafiltration
means. Combining these methods with the above profiling technologies allows
reproducible and sensitive analysis of one of the most significant and complex
biological samples (77).
Overview and Introduction to Clinical Proteomics 15
Table 2
Clinical Proteomics Initiatives and Resources
Details Websites
Institute
CPTI National Cancer Institute’s Clinical
Proteomics Technologies, initiative for
cancer
http://proteomics.cancer.
gov
ABRF The Association of Biomolecular
Resource Facilities, an international
society dedicated to advancing core and
research biotechnology laboratories
through research, communication, and
education
http://www.abrf.org/
PPI Plasma Proteome Institute, the PPI is
working to facilitate clinical adoption of
advanced diagnostic tests using proteins
in plasma and serum
http://www.plasmaprote
ome.org/plasmaframes.
htm
EDRN The Early Detection Research Network,
the EDRN provide up-to-date
information on biomarker research
through this website and scientific
publications
http://edrn.nci.nih.gov
Web resources
ExPASy Expert Protein Analysis System,
proteomics related information and
database
http://www.expasy.org/
NCBI National Center for Biotechnology
Information, the protein entries in the
Entrez search and retrieval system have
been compiled from a variety of sources,
including SwissProt, PIR, PRF, PDB,
and translations from annotated coding
regions in GenBank and RefSeq
http://www.ncbi.nlm.
nih.gov/entrez/query.
fcgi?db = Protein&
itool = toolbar
CPRMap Clinical Proteomics Research Map,
updated research article for disease and
clinical proteomics
http://www.cprmap.com/
Database
MedGene MedGene can make a list of human
genes associated with a particular human
disease in ranking order
http://hipseq.med.harv
ard.edu/MEDGENE
16 Paik et al.
Table 3
Available Bioinformatic Resources for the Analysis of Proteomics Data
Name Description Website URL PMID
Clinical proteome data management system
Proteus LIMS for proteomics
pipeline
http://www.
genologics.com
CPAS LIMS for identification
and quantification using
by LC-MS/MS data
16396501
Systems biology
experiment analysis
management
system
A management system for
collecting, storing,
and accessing data
produced by microarray,
proteomics, and
immunohistochemistry
http://www.
sbeams.org/
16756676
GPM database Open source system for
analyzing, validating,
and storing protein
identification data
http://www.
thegpm.org/
15595733
SpectrumMill MS/MS data analysis and
management system
http://www.chem.
agilent.com/
Phosphorylation
Group-based
phosphorylation
scoring method
Prediction of
kinase-specific
phosphorylation sites
http://973-
proteinweb.ustc.
edu.cn/gps/
gps_web/
15980451
KinasePhos A web tool for identifying
protein kinase-specific
phosphorylation sites
using by hidden Markov
model
http://kinasePhos.
mbc.nctu.edu.tw
15980458
NetPhos Sequence and
structure-based prediction
of eukaryotic protein
phosphorylation sites
http://www.cbs.
dtu.dk/services/
NetPhos/
10600390
NetPhosK Prediction of
post-translational
glycosylation and
phosphorylation of
proteins from the amino
acid sequence
http://www.cbs.dtu.
dk/services/
NetPhosK/
15174133
Overview and Introduction to Clinical Proteomics 17
PredPhospho Prediction of phosphorylation
sites using support vector
machine
http://pred.ngri.
re.kr/Pred
Phospho.htm
15231530
PREDIKIN A prediction of substrates for
serine/threonine protein
kinases based on the primary
sequence of a protein kinase
catalytic domain
http://florey.biosci.
uq.edu.au/kinsub/
home.htm
16445868
Prosite A prediction of substrates
for protein kinases-based
conserved motif search
http://kr.expasy.
org/prosite
17237102
Scansite Prediction of PK-specific
phosphorylation site with
Bayesian decision theory
http://scansite.
mit.edu
16549034
Phospho.ELM A database of experimentally
verified phosphorylation sites
in eukaryotic proteins
http://phospho.elm.
eu.org/
15212693
Human protein
reference database
(HPRD)
A database of known
kinase/phosphatase substrate as
well as binding motifs that are
curated from the published
literature
http://www.hprd.
org/PhosphoMotif_
finder
PhosphoSite A bioinformatics resource
dedicated to physiological
protein phosphorylation
http://www.
phosphosite.org/
Login.jsp
15174125
Glycosylation
NetOGlyc 2.0 Predicts O-glycosylation sites
in mucin-type proteins
http://www.cbs.
dtu.dk/services/
NetOGlyc/
9557871
DictyOGlyc 1.1 Predicts O-GlcNAc sites in
eukaryotic proteins
http://www.cbs.
dtu.dk/services/
DictyOGlyc/
10521537
YinOYang 1.2 Predicts O-GlcNAc sites in
eukaryotic proteins
http://www.cbs.
dtu.dk/services/
YinOYang/
NetNGlyc 1.0 Predicting N-glycosylation
sites
http://www.cbs.dtu.
dk/services/
NetNGlyc/
16316981
GlycoMod Web software for prediction of
the possible oligosaccharide
structures in glycoproteins
from their experimentally
determined masses
http://www.expasy.
ch/tools/glycomod/
11680880
(Continued)
18 Paik et al.
Table 3
(Continued)
Name Description Website URL PMID
Glyco-fragment A web tool to support
the interpretation of
mass spectra of complex
carbohydrates
http://www.dkfz.
de/spec/projekte/
fragments/
14625865
GlycoSearchMS Compares each peak
of a measured mass
spectrum with the calculated
fragments of all structures
contained in the SweetDB
http://www.dkfz.
de/spec/glycosciences.
de/sweetdb/ms/
15215392
GlycosidIQ Based on the matching of
experimental MS2 data with
the theoretical fragmentation
of glycan structures in
GlycoSuiteDB
https://tmat.
proteomesystems.
com/glyco/glycosuite/
glycodb
15174134
Saccharide
topology
analysis tool
A web-based computational
program that can quickly
extract sequence information
from a set of MSn spectra
for an oligosaccharide of up
to 10 residues
10857602
GlycoX To determine simultaneously
the glycosylation sites
and oligosaccharide
heterogeneity of
glycoproteins using
MATLAB
17022651
MODi A web server for identifying
multiple post-translational
peptide modifications from
tandem mass spectra
http://www.
unimod.org
16845006
SWEET-DB An attempt to create
annotated data collections
for carbohydrates
http://www.dkfz.de/
spec2/sweetdb/
11752350
Protein–protein interaction
Munich
information
center for protein
sequence’s MPPI
The database of mammalian
protein–protein interactions
http://mips.gsf.de 16381839
Overview and Introduction to Clinical Proteomics 19
Database of
interacting proteins
A database that documents
experimentally determined
protein–protein interactions
http://dip.doe-
mbi.ecla.edu/
11752321
Molecular
interaction network
database
A database of storing, in
a structured format,
information about
molecular interactions by
extracting experimental
details from work
published in peer-reviewed
journals
http://mint.bio.
uniroma2.it/mint
17135203
Protein–protein
interactions of
cancer proteins
Predicts interactions, which
are derived from homology
with experimentally known
protein–protein interactions
from various species
http://bmm.
cancerresearchuk.
org/˜pip
16398927
IntAct IntAct provides a freely
available, open source
database system and
analysis tools for protein
interaction data
http://www.ebi.
ac.uk/intact/
17145710
Biomolecular
interaction network
database
A database designed to
store full descriptions of
interactions, molecular
complexes and pathways
http://www.bind.ca 12519993
Metabolic and
signal pathway
BioCarta A pathway database http://www.
biocarta.com
KEGG A pathway database with
genomical, chemical, and
biological network
information
http://www.
genome.jp/kegg
16381885
Cancer cell map The cancer cell map is a
selected set of human
cancer focused pathways
http://cancer.
cellmap.org/cellmap/
HPRD A database with
data pertaining
to post-translational
modifications,
protein–protein
interactions, tissue
expression,
http://www.
hprd.org/
(Continued)
20 Paik et al.
Table 3
(Continued)
Name Description Website URL PMID
subcellular localization,
and enzyme–substrate
relationships
Proteomic data resource
The cancer cell
map
A database of clinical data
from SELDI-TOF
http://home.ccr.
cancer.gov/ncifda
proteomics/
ppatterns.asp
Proteomics
identifications
database
A database of protein and
peptide identifications that
have been described in the
scientific literature
http://www.ebi.
ac.uk/pride/
16381953
PeptideAtlas A multiorganism, publicly
accessible compendium of
peptides identified in a
large set of tandem mass
spectrometry proteomics
experiments
http://www.
peptideatlas.org
16381952
Disease resource
Online
mendelian
inheritance in
man
A database of human genes
and genetic disorders
http://www.ncbi.nlm.
nih.gov/entrez/query.
fcgi?db = OMIM
17170002
GeneCards An integrated database of
human genes that includes
automatically mined
genomic, proteomic, and
transcriptomic information
http://www.genecards.
org/index.shtml
15608261
Cancer gene
census
A catalogue those genes for
which mutations have been
causally implicated in cancer
http://www.sanger.
ac.uk/genetics/CGP/
Census/
14993899
Two-dimensional electrophoresis is perhaps the most popular start-up tool
for proteome analysis. For clinical proteomics, 2DE has been the traditional
workhorse of proteomics used for the analysis of different clinical specimens
ranging from plasma to urine (Table 1). Quantification problems in 2DE are now
solved by employing fluorescent dyes (cy3 and cy5), which allow normalization
Overview and Introduction to Clinical Proteomics 21
of data obtained from two different clinical specimens (79). Freedman and
Lilley (Chapter 6) present general optimization conditions for differential in gel
electrophoresis (DIGE) in the quantitative analysis of clinical samples. They
address the usefulness of differentially labeling dyes (Cy2, Cy3, and Cy5).
The essence of any DIGE system is to minimize any potential human errors
in the process of identification and quantification of proteins spotted in a 2D
gel (79). The difficulties in 2D map analysis are introduced by Marengo et al.
(Chapter 16). They describe methods for comparing protein spots using image
analysis technology and related informatics tools to minimize variations between
measurements of spot volume, a key to successful 2D map construction.
There are many variations of LC in protein profiling, including mass detection
methods, column types, data mining through search engines, mass accuracy,
and running conditions (80,81,82). These are all related to quantification of
proteins or peptides in the sample, one of the major bottlenecks in proteomics
(83,84,85,86,87). Among the several techniques are isotope-coded affinity tags
(ICAT), mass-coded affinity tagging, and nonisotope labeled methods. Xiao and
Veenstra (Chapter 10) present the application of ICAT in the course of COX-2
inhibitor regulated proteins in a colon cancer cell line. With emphasis on sample
preparation, they provide details on ICAT procedures for quantitative proteomics
(88). In addition to this approach, Li et al. (Chapter 11) employ a strategy,
which combines LCM techniques for sample preparation of HCC and cleavable
isotope-coded affinity tags in order to identify those markers quantitatively.
However, it should be mentioned here that some other measures are needed to
increase the efficiency of ICAT since it has drawbacks in the efficiency of sample
recovery during or after labeling steps (87). A label-free serum quantification
method has been recently introduced (48) (See Chapter 12 by Higgs et al.).
The use of antibody arrays in clinical proteomics has increased recently in the
context of high-throughput detection of cancer specimens where the identities
of the proteins of interest are known (89,90). The evaluation of antibody cross-
reactivity and specificity is very crucial in these assays. This matter is addressed
by Sanchez-Carbayo (Chapter 15), where technical aspects and application of
planar antibody arrays in the quantification of serum proteins is described as
well as by Hsu et al. (Chapter 14) where the development and use of bead-
based miniaturized multiplexed sandwich immunoassays for focused protein
profiling in various body fluids is provided. The latter method using bead-
based protein arrays or suspension microarray allows the simultaneous analysis
of a variety of parameters within a single experiment. With the versatility of
suspension microarray in the analysis of proteins of interest present in different
types of body fluids ranging from serum to synovial fluids, this multiplexed
protein profiling technology described by Hsu et al. (Chapter 14) seems to
hold a great promise in clinical proteomics. Similarly, in combination with
22 Paik et al.
tissue microarrays technology (91) it would also be possible to perform parallel
molecular profiling of clinical samples together with immunohistochemistry,
fluorescence in situ hybridization, or RNA in situ hybridization. SELDI is
another arena of high-throughput profiling of clinical samples in the course
of disease marker discovery [(92,93), Chapter 7]. It is expected that profiling
approaches in proteomics, such as SELDI-MS, will be frequently used in disease
marker discovery, but only if the proper identification technologies coupled
with SELDI are improved.
During the course of biomarker discovery, large data sets are usually
generated and deposited in a coordinated fashion (Tables 2 and 3)(94,95).
Indeed, statistical analysis of 2DE proteomics, which produce several hundred
protein spots, is complex. To circumvent some inconsistency in 2D gel
proteomics data, Friedman and Lilley (Chapter 6) and Carpentier et al. (Chapter
17) point out available statistical tools and suggest case-specific guidelines for
2D gel spot analysis. Fitzgibbon et al. (Chapter 19) describe an open source
platform for LC-MS spectra where the msInspector program is used to lower
false positives and guide normalization of the dataset. It is also demonstrated
that msInspect can analyze data from quantitative studies with and without
isotopic labels. Paliakasis et al. (Chapter 18) introduce web-based tools for
protein classification, which lead to prediction of potential protein function
and family clustering of related proteins. They provide some guidelines to
classification of protein data into more meaningful families. Finally, Somorjai
(Chapter 20) addresses important filtering criteria for the application of protein
pattern recognition to biomarker discovery using statistical tools.
5. Concluding Remarks
Although there are several bottlenecks in clinical proteomics (such as lack
of standardization of sample specimen process, quantification, and overall
strategy for tackling post-identification of biomarkers), we believe that the
field holds great promise in biomarker discovery. The success of clinical
proteomics depends on the availability and selection of well-phenotyped
specimens, reduction of sample complexity, development of good informatics
tools, and efficient data management. Therefore, sample handling techniques
including microdissection for tissue sample, multidimensional fractionation for
body fluids, and pretreatment of other clinical specimens (e.g., urine, tears, and
cells) should be developed in this context. Since there is no gold standard for
sample collection and handling, one needs to find the best options available for
sample processing without damage. In addition, establishment of a biorepository
system would systematically minimize some artifacts and variation between
samples during or after identification of biomarkers.
Overview and Introduction to Clinical Proteomics 23
It is now generally accepted that an ensemble (or panel) of different proteins
would be more efficient than a single protein/peptide in the diagnosis of disease,
an idea which is poised to replace the conventional concept of a biomarker.
As a high-throughput way of protein profiling, the use of antibody arrays
in clinical proteomics has recently increased in regard to detection of cancer
specimens. However, in the use of antibody arrays to profile serum autoanti-
bodies, issues of cross-reactivity and specificity have to be resolved. Although
not covered here due to space limitations, with the advent of proteomics
techniques one can further analyze a network of protein–protein interaction
as well as post-translational modifications of those proteins involved in a
specific disease (Table 3). It is now highly recommended that common reagents
such as antibodies and standard proteins, which are very useful for spiking
purposes, quantification work, and sensitivity normalization of one machine to
another be used in worldwide efforts like human proteome organization plasma
proteome project (96,97). Finally, clinical proteomics needs the integration of
biochemistry, pathology, analytical technology, bioinformatics, and proteome
informatics to develop highly sensitive diagnostic tools for routine clinical care
in the future (71,98).
Acknowledgments
This study was supported by a grant from the Korea Health 21 R&D project,
Ministry of Health & Welfare, Republic of Korea (A030003 to YKP).
References
1. Etzioni, R., Urban, N., Ramsey, S., McIntosh, M., Schwartz, S., Reid, B., Radich, J.,
Anderson, G., and Hartwell, L. (2003) The case for early detection. Nat. Rev.
Cancer 3, 1–10.
2. Ludwig, J. A. and Weinstein, J. N. (2005) Biomarkers in cancer staging, prognosis
and treatment selection. Nat. Rev. Cancer 5, 845–856.
3. Xiao, Z., Prieto, D., Conrads, T. P., Veenstra, T. D., and Issaq, H. J. (2005)
Proteomic patterns: their potential for disease diagnosis. Mol. Cell Endocrinol.
230, 95–106.
4. Rifai, N., Gillette, M. A., and Carr, S. A. (2006) Protein biomarker discovery
and validation: the long and uncertain path to clinical utility. Nat. Biotechnol.24,
97–983.
5. Emmert-Buck, M. R., Bonner, R. F., Smith, P. D., Chuaqui, R. F., Zhuang, Z.,
Goldstein, S. R., Weiss, R. A., and Liotta, L. A. (1996) Laser capture microdis-
section. Science 274, 998–1001.
6. Gillespie, J. W., Ahram, M., Best, C. J., Swalwell, J. I., Krizman, D. B.,
Petricoin, E. F., Liotta, L. A., and Emmert-Buck, M. R. (2001) The role of tissue
microdissection in cancer research. Cancer J. 7, 32–39.
24 Paik et al.
7. Craven, R. A. and Banks, R. E. (2002) Use of laser capture microdissection to
selectively obtain distinct populations of cells for proteomic analysis. Methods
Enzymol.356, 33–49.
8. Vincourt, J. B., Lionneton, F., Kratassiouk, G., Guillemin, F., Netter, P.,
Mainard, D., and Magdalou, J. (2006) Establishment of a reliable method for direct
proteome characterization of human articular cartilage. Mol. Cell Proteomics 5,
1984–1995.
9. Platt, M. S., Agamanolis, D. P., Krill, C. E. Jr., Boeckman, C., Potter, J. L.,
Robinson, H., and Lloyd, J. (1983) Occult hepatic sinusoid tumor of infancy
simulating neuroblastoma. Cancer 52, 1183–1189.
10. Mahadevia, P. J., Fleisher, L. A., Frick, K. D., Eng, J., Goodman, S. N., and
Powe, N. R. (2003) Lung cancer screening with helical computed tomography
in older adult smokers: a decision and cost-effectiveness analysis. JAMA 289,
313–322.
11. Hood, B. L., Darfler, M. M., Guiel, T. G., Furusato, B., Lucas, D. A.,
Ringeisen, B. R., Sesterhenn, I. A., Conrads, T. P., Veenstra, T. D., and Krizman,
D. B. (2005) Proteomic analysis of formalin-fixed prostate cancer tissue. Mol. Cell
Proteomics 4, 1741–1753.
12. Alaiya, A., Al-Mohanna, M., and Linder, S. (2005) Clinical cancer proteomics:
promises and pitfalls. J. Proteome Res.4, 1213–1222.
13. Gericke, B., Raila, J., Sehouli, J., Haebel, S., Konsgen, D., Mustea, A., and
Schweigert, F. J. (2005) Microheterogeneity of transthyretin in serum and ascitic
fluid of ovarian cancer patients. BMC Cancer 17, 133–141.
14. Swisher, E. M., Wollan, M., Mahtani, S. M., Willner, J. B., Garcia, R., Goff, B. A.,
and King, M. C. (2005) Tumor-specific p53 sequences in blood and peritoneal fluid
of women with epithelial ovarian cancer. Am. J. Obstet. Gynecol. 193, 662–667.
15. Pisitkun, T., Johnstone, R., and Knepper, M. A. (2006) Discovery of urinary
biomarkers. Mol. Cell Proteomics 5, 1760–1771.
16. Ghafouri, B., Irander, K., Lindbom, J., Tagesson, C., and Lindahl, M. (2006)
Comparative proteomics of nasal fluid in seasonal allergic rhinitis. J. Proteome
Res. 5, 330–338.
17. Koo, B. S., Lee, D. Y., Ha, H. S., Kim, J. C., and Kim, C. W. (2005) Comparative
analysis of the tear protein expression in blepharitis patients using two-dimensional
electrophoresis. J. Proteome Res.4, 719–724.
18. Grus, F. H., Podust, V. N., Bruns, K., Lackner, K., Fu, S., Dalmasso, E. A.,
Wirthlin, A., and Pfeiffer, N. (2005) SELDI-TOF-MS ProteinChip array profiling
of tears from patients with dry eye. Invest. Ophthalmol. Vis. Sci. 46, 863–876.
19. Amado, F. M., Vitorino, R. M., Domingues, P. M., Lobo, M. J., and Duarte, J. A.
(2005) Analysis of the human saliva proteome. Expert Rev. Proteomics 2, 521–539.
20. Wang, T. H., Chang, Y. L., Peng, H. H., Wang, S. T., Lu, H. W., Teng, S. H.,
Chang, S. D., and Wang, H. S. (2005) Rapid detection of fetal aneuploidy using
proteomics approaches on amniotic fluid supernatant. Prenat. Diagn. 25, 559–566.
21. Ruetschi, U., Rosen, A., Karlsson, G., Zetterberg, H., Rymo, L., Hagberg,
H., and Jacobsson, B. (2005) Proteomic analysis using protein chips to detect
Overview and Introduction to Clinical Proteomics 25
biomarkers in cervical and amniotic fluid in women with intra-amniotic inflam-
mation. J. Proteome Res. 4, 2236–2242.
22. Kim, Y. S., Kim, M. S., Lee, S. H., Choi, B. C., Lim, J. M., Cha, K. Y., and
Baek, K. H. (2006) Proteomic analysis of recurrent spontaneous abortion: identi-
fication of an inadequately expressed set of proteins in human follicular fluid.
Proteomics 6, 3445–3454.
23. Pilch, B. and Mann, M. (2006) Large-scale and high-confidence proteomic analysis
of human seminal plasma. Genome Biol.7, R40
24. Varnum, S. M., Covington, C. C., Woodbury, R. L., Petritis, K., Kangas, L. J.,
Abdullah, M. S., Pounds, J. G., Smith, R. D., and Zangar, R. C. (2003) Proteomic
characterization of nipple aspirate fluid: identification of potential biomarkers of
breast cancer. Breast Cancer Res. Treat. 80, 87–97.
25. Zheng, P. P., Luider, T. M., Pieters, R., Avezaat, C. J., van den Bent, M. J., Sillevis
Smitt, P. A., and Kros, J. M. (2003) Identification of tumor-related proteins by
proteomic analysis of cerebrospinal fluid from patients with primary brain tumors.
J. Neuropathol. Exp. Neurol.62, 855–862.
26. Gibson, D. S., Blelock, S., Brockbank, S., Curry, J., Healy, A., McAllister, C.,
and Rooney, M. E. (2006) Proteomic analysis of recurrent joint inflammation in
juvenile idiopathic arthritis. J. Proteome Res. 5, 1988–1995.
27. Merkel, D., Rist, W., Seither, P., Weith, A., and Lenter, M. C. (2005)
Proteomic study of human bronchoalveolar lavage fluids from smokers with
chronic obstructive pulmonary disease by combining surface-enhanced laser
desorption/ionization-mass spectrometry profiling with mass spectrometric protein
identification. Proteomics 5, 2972–2980.
28. Wu, J., Kobayashi, M., Sousa, E. A., Liu, W., Cai, J., Goldman, S. J., Dorner, A. J.,
Projan, S. J., Kavuru, M. S., Qiu, Y., and Thomassen, M. J. (2005) Differ-
ential proteomic analysis of bronchoalveolar lavage fluid in asthmatics following
segmental antigen challenge. Mol. Cell Proteomics 4, 1251–1264.
29. Tyan, Y. C., Wu, H. Y., Lai, W. W., Su, W. C., and Liao, P. C. (2005) Proteomic
profiling of human pleural effusion using two-dimensional nano liquid chromatog-
raphy tandem mass spectrometry. J. Proteome Res.4, 1274–1286.
30. Khalil, A. A. and James, P. (2007) Biomarker discovery: a proteomic approach for
brain cancer profiling. Cancer Sci. 98, 201–213.
31. Khodavirdi, A. C., Song, Z., Yang, S., Zhong, C., Wang, S., Wu, H., Pritchard, C.,
Nelson, P. S., and Roy-Burman, P. (2006) Increased expression of osteopontin
contributes to the progression of prostate cancer. Cancer Res. 66, 883–888.
32. Vincourt, J. B., Lionneton, F., Kratassiouk, G., Guillemin, F., Netter, P., Mainard, D.,
and Magdalou, J. (2006) Establishment of a reliable method for direct proteome
characterization of human articular cartilage. Mol. Cell Proteomics 5, 1984–1995.
33. Lee, Y. J., Rice, R. H., and Lee, Y. M. (2006) Proteome analysis of human
hair shaft: from protein identification to post-translational modification. Mol. Cell
Proteomics 5, 789–800.
34. Cho, S. Y., Lee, E. Y., Lee, J. S., Kim, H. Y., Park, J. M., Kwon, M. S., Park, Y. K.,
Lee, H. J., Kang, M. J., Kim, J. Y., Yoo, J. S., Park, S. J., Cho, J. W., Kim, H. S., and
26 Paik et al.
Paik, Y. K. (2005) Efficient prefractionation of low-abundance proteins in human
plasma and construction of a two-dimensional map. Proteomics 5, 3386–3396.
35. Lathrop, J. T., Hayes, T. K., Carrick, K., and Hammond, D. J. (2005) Rarity gives
a charm: evaluation of trace proteins in plasma and serum. Expert Rev. Proteomics
2, 393–406.
36. Lee, H. J., Lee, E. Y., Kwon, M. S., and Paik, Y. K. (2006) Biomarker discovery
from the plasma proteome using multidimensional fractionation proteomics. Curr.
Opin. Chem. Biol.10, 42–49.
37. Anderson, N. L. and Anderson, N. G. (2002) The human plasma proteome: history,
character, and diagnostic prospects. Mol. Cell Proteomics 1, 845–867.
38. Hu, S., Loo, J. A., and Wong, D. T. (2006) Human body fluid proteome analysis.
Proteomics 6, 6326–6353.
39. Park, M. R., Wang, E. H., Jin, D. C., Cha, J. H., Lee, K. H., Yang, C. W.,
Kang, C. S., and Choi, Y. J. (2006) Establishment of a 2-D human urinary proteomic
map in IgA nephropathy. Proteomics 6, 1066–1076.
40. Tammen, H., Schutle, I., Hess, R., Menzel, C., Kellmann, M., and Schulz-
Knappe, P. (2005) Prerequisites for peptidomic analysis of blood samples: I.
Evaluation of blood specimen qualities and determination of technical performance
characteristics. Comb. Chem. High Trhoughput Screen 8, 725–733.
41. Rai, A. J., Gelfand, C. A., Haywood, B. C., Warunek, D. J., Yi, J., Schuchard, M. D.,
Mehigh, R. J., Cockrill, S. L., Scott, G. B., Tammen, H., Schulz-Knappe, P.,
Speicher, D. W., Vitzthum, F., Haab, B. B., Siest, G., and Chan, D. W.
(2005) HUPO plasma proteome project specimen collection and handling: towards
the standardization of parameters for plasma proteome samples. Proteomics 5,
3262–3277.
42. Zhou, M., Lucas, D. A., Chan, K. C., Issaq, H. J., Petricoin, E. F. 3rd, Liotta, L. A.,
Veenstra, T. D., and Conrads, T. P. (2004) An investigation into the human serum
“interactome”. Electrophoresis 25, 1289–1298.
43. Findeisen, P., Sismanidis, D., Riedl, M., Costina, V., and Neumaier, M. (2005)
Preanalytical impact of sample handling on proteome profiling experiments with
matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. Clin.
Chem.51, 2409–2411.
44. Park, K. S., Kim, H., Kim, N. G., Cho, S. Y., Choi, K. H., Seong, J. K., and Paik,
Y. K. (2002) Proteomic analysis and molecular characterization of tissue ferritin
light chain in hepatocellular carcinoma. Hepatology 35, 1459–1466.
45. Park, K. S., Cho, S. Y., Kim, H., and Paik, Y. K. (2002) Proteomic alterations of the
variants of human aldehyde dehydrogenase isozymes correlate with hepatocellular
carcinoma. Int. J. Cancer 97, 261–265.
46. Marko-Varga, G., Berglund, M., Malmstrom, J., Lindberg, H., and Fehniger, T. E.
(2003) Targeting hepatocytes from liver tissue by laser capture microdissection
and proteomics expression profiling. Electrophoresis 24, 3800–3805.
47. Paradis, V., Degos, F., Dargere, D., Pham, N., Belghiti, J., Degott, C., Janeau,
J. L., Bezeaud, A., Delforge, D., Cubizolles, M., Laurendeau, I., and Bedossa, P.
(2005) Identification of a new biomarker of hepatocellular carcinoma by serum
protein profiling of patients with chronic liver diseases. Hepatology 41, 40–47.
Overview and Introduction to Clinical Proteomics 27
48. Ru, Q. C., Zhu, L. A., Silberman, J., and Shriver, C. D. (2006) Label-free semiquan-
titative peptide feature profiling of human breast cancer and breast disease sera via
two-dimensional liquid chromatography–mass spectrometry. Mol. Cell Proteomics
5, 1095–1104.
49. Azad, N. S., Rasool, N., Annuziata, C. M., Minasian, L., Whiteley, G., and
Kohn, E. C. (2006) Proteomics in clinical trials and practice: present uses and
future promise. Mol. Cell Proteomics 5, 1819–1829.
50. Gunter, E. W. (1997) Biological and environmental specimen banking at the
Centers for Disease Control and Prevention. Chemosphere 34, 1945–1953.
51. Strauss, G. H. and Kelly, S. J. (1990) The development of the U.S. EPA health
effects research laboratory frozen blood cell repository program. Mutat. Res.234,
349–354.
52. Romeo, M. J., Espina, V., Lowenthal, M., Espina, B. H., Petricoin, E. F. 3rd, and
Liotta, L. A. (2005) CSF proteome: a protein repository for potential biomarker
identification. Expert Rev. Proteomics 2, 57–70.
53. Conrads, T. P., Hood, B. L., Petricoin, E. F. 3rd, Liotta, L. A., and Veenstra, T. D.
(2005) Cancer proteomics: many technologies, one goal. Expert Rev. Proteomics
2, 693–703.
54. Schrader, M. and Selle, H. (2006) The process chain for peptidomic biomarker
discovery. Dis. Markers 22, 27–37.
55. Danna, E. A. and Nolan, G. P. (2006) Transcending the biomarker mindset:
deciphering disease mechanisms at the single cell level. Curr. Opin. Chem. Biol.
10, 20–27.
56. De Masi, S., Tosti, M. E., and Mele, A. (2005) Screening for hepatocellular
carcinoma. Dig. Liver Dis. 37, 260–268.
57. Yamaguchi, K., Nagano, M., Torada, N. Hamasaki, N., Kawakita, M., and
Tanaka, M. (2004) Urine diacetylspermine as a novel tumor marker for pancreato-
biliary carcinomas. Rinsho. Byori.52, 336–339
58. Dabrowska, M., Grubek-Jaworska, H., Domagala-Kulawik, J., Bartoszewicz, Z.,
Kondracka, A., Krenke, R., Nejman, P., and Chazan, R. (2004) Diagnostic usefulness
of selected tumor markers (CA125, CEA, CYFRA 21–1) in bronchoalveolar lavage
fluid in patients with non-small cell lung cancer. Pol. Arch. Med. Wewn 111, 659–665.
59. Gann, P. H., Hennekens, C. H., and Stampfer, M. J. (1995) A prospective evaluation
of plasma prostate-specific antigen for detection of prostatic cancer. JAMA 273,
289–294
60. Ciambellotti, E., Coda, C., and Lanza, E. (1993) Determination of CA 15–3 in the
control of primary and metastatic breast carcinoma. Minerva Med.84, 107–112.
61. Linkov, F., Lisovich, A., Yurkovetsky, Z., Marrangoni, A., Velikokhatnaya, L.,
Nolen, B., Winans, M., Bigbee, W., Siegfried, J., Lokshin, A., and Ferris, R. L.
(2007) Early detection of head and neck cancer: development of a novel screening
tool using multiplexed immunobead-based biomarker profiling. Cancer Epidemiol.
Biomarkers Prev.16, 102–107.
62. Casiano, C. A., Mediavilla-Varela, M., and Tan, E. M. (2006) Tumor-associated
antigen arrays for the serological diagnosis of cancer. Mol. Cell Proteomics 5,
1745–1759.
28 Paik et al.
63. Nissom, P. M., Lo, S. L., Lo, J. C., Ong, P. F., Lim, J. W., Ou, K., Liang, R. C.,
Seow, T. K., and Chung, M. C. (2006) Hcc-2, a novel mammalian ER thioredoxin
that is differentially expressed in hepatocellular carcinoma. FEBS Lett. 580, 2216–
2226.
64. Feng, J. T., Liu, Y. K., Song, H. Y., Dai, Z., Qin, L. X., Almofti, M. R., Fang, C. Y.,
Lu, H. J., Yang, P. Y., and Tang, Z. Y. (2005) Heat-shock protein 27: a potential
biomarker for hepatocellular carcinoma identified by serum proteome analysis.
Proteomics 5, 4581–1588.
65. Li, D. Q., Wang, L., Fei, F., Hou, Y. F., Luo, J. M., Wei-Chen, Zeng, R.,
Wu, J., Lu, J. S., Di, G. H., Ou, Z. L., Xia, Q. C., Shen, Z. Z., and
Shao, Z. M. (2006) Identification of breast cancer metastasis-associated proteins
in an isogenic tumor metastasis model using two-dimensional gel electrophoresis
and liquid chromatography-ion trap-mass spectrometry. Proteomics 6,
3352–3368.
66. Lee, I. N., Chen, C. H., Sheu, J. C., Lee, H. S., Huang, G. T., Yu, C. Y.,
Lu, F. J., and Chow, L. P. (2005) Identification of human hepatocellular carcinoma-
related biomarkers by two-dimensional difference gel electrophoresis and mass
spectrometry. J. Proteome Res.4, 2062–2069.
67. Righetti, P. G., Castagna, A., Antonucci, F., Piubelli, C., Cecconi, D.,
Campostrini, N., Rustichelli, C., Antonioli, P., Zanusso, G., Monaco, S., Lomas, L.,
and Boschetti, E. (2005) Proteome analysis in the clinical chemistry laboratory:
myth or reality? Clin. Chim. Acta 357, 123–139.
68. Jang, J. S., Cho, H. Y., Lee, Y. J., Ha, W. S., and Kim, H. W. (2004) The
differential proteome profile of stomach cancer: identification of the biomarker
candidates. Oncol. Res.14, 491–499.
69. Steel, L. F., Shumpert, D., Trotter, M., Seeholzer, S. H., Evans, A. A., London,
W. T., Dwek, R., and Block, T. M. (2003) A strategy for the comparative analysis
of serum proteomes for the discovery of biomarkers for hepatocellular carcinoma.
Proteomics 3, 601–609.
70. Yip, T. T., Chan, J. W., Cho, W. C., Yip, T. T., Wang, Z., Kwan, T. L., Law, S. C.,
Tsang, D. N., Chan, J. K., Lee, K. C., Cheng, W. W., Ma, V. W., Yip, C.,
Lim, C. K., Ngan, R. K., Au, J. S., Chan, A., Lim, W. W., and Ciphergen SARS
Proteomics Study Group (2005) Protein chip array profiling analysis in patients
with severe acute respiratory syndrome identified serum amyloid a protein as a
biomarker potentially useful in monitoring the extent of pneumonia. Clin. Chem. 51,
47–55.
71. Anderson, L. and Hunter, C. L. (2005) Quantitative mass spectrometric multiple
reaction monitoring assays for major plasma proteins. Mol. Cell Proteomics 5,
573–588.
72. Lee, J. W., Figeys, D., and Vasilescu, J. (2007) Biomarker assay translation from
discovery to clinical studies in cancer drug development: quantification of emerging
protein biomarkers. Adv. Cancer Res.96, 269–298.
73. Zolg, W. (2006) The proteomic search for diagnostic biomarkers: lost in trans-
lation? Mol. Cell Proteomics 5, 1720–1726.
Overview and Introduction to Clinical Proteomics 29
74. Bensmail, H., Golek, J., Moody, M. M., Semmes, J. O., and Haoudi, A. (2005)
A novel approach for clustering proteomics data using Bayesian fast Fourier
transform. Bioinformatics 21, 2210–2224.
75. Ward, D. G., Cheng, Y., N’Kontchou, G., Thar, T. T., Barget, N., Wei, W.,
Billingham, L. J., Martin, A., Beaugrand, M., and Johnson, P. J. (2006) Changes in
the serum proteome associated with the development of hepatocellular carcinoma
in hepatitis C-related cirrhosis. Br. J. Cancer 94, 287–292.
76. Lin, N. and Zhao, H. (2005) Are scale-free networks robust to measurement errors?
BMC Bioinformatics 6, 119.
77. Castagna, A., Cecconi, D., Sennels, L., Rappsilber, J., Guerrier, L., Fortis, F.,
Boschetti, E., Lomas, L., and Righetti, P. G. (2005) Exploring the hidden human
urinary proteome via ligand library beads. J. Proteome Res.4, 1917–1930.
78. Rauch, A., Bellew, M., Eng, J., Fitzgibbon, M., Holzman, T., Hussey, P., Igra, M.,
Maclean, B., Lin, C. W., Detter, A., Fang, R., Faca, V., Gafken, P., Zhang, H.,
Whiteaker, J., States, D., Hanash, S., Paulovich, A., and McIntosh, M. W. (2006)
Computational proteomics analysis system (CPAS): an extensible open source
analytic system for evaluating and publishing proteomic data and high throughput
biological experiments. J. Proteome Res.5, 112–121.
79. Lilley, K. S. and Friedman, D. B. (2004) All about DIGE: quantification technology
for differential-display 2D-gel proteomics. Expert Rev. Proteomics 1, 401–409.
80. Qian, W. J., Jacobs, J. M., Liu, T., Camp, D. G. 2nd, and Smith, R. D.
(2006) Advances and challenges in liquid chromatography-mass spectrometry-
based proteomics profiling for clinical applications. Mol. Cell Proteomics 5,
1727–1744.
81. Powell, D. W., Merchant, M. L., and Link, A. J. (2006) Discovery of regulatory
molecular events and biomarkers using 2D capillary chromatography and mass
spectrometry. Expert Rev. Proteomics 3, 63–74.
82. Andre, M., Le Caer, J. P., Greco, C., Planchon, S., El Nemer, W., Boucheix, C.,
Rubinstein, E., Chamot-Rooke, J., and Le Naour, F. (2006) Proteomic analysis of
the tetraspanin web using LC-ESI-MS/MS and MALDI-FTICR-MS. Proteomics
6, 1437–1449.
83. Greengauz-Roberts, O., Stoppler, H., Nomura, S., Yamaguchi, H.,
Goldenring, J. R., Podolsky, R. H., Lee, J. R., and Dynan, W. S. (2005) Saturation
labeling with cysteine-reactive cyanine fluorescent dyes provides increased sensi-
tivity for protein expression profiling of laser-microdissected clinical specimens.
Proteomics 5, 1746–1757.
84. Heck, A. J. and Krijgsveld, J. (2004) Mass spectrometry-based quantitative
proteomics. Expert Rev. Proteomics 1, 317–326.
85. Schneider, L. V. and Hall, M. P. (2005) Stable isotope methods for high-precision
proteomics. Drug Discov. Today 10, 353–363.
86. Zhang, J., Goodlett, D. R., Peskind, E. R., Quinn, J. F., Zhou, Y., Wang, Q.,
Pan, C., Yi, E., Eng, J., Aebersold, R. H., and Montine, T. J. (2005) Quantitative
proteomic analysis of age-related changes in human cerebrospinal fluid. Neurobiol
Aging 26, 207–227.
30 Paik et al.
87. Liu, T., Qian, W. J., Strittmatter, E. F., Camp, D. G. 2nd, Anderson, G. A.,
Thrall. B. D., and Smith, R. D. (2004) High-throughput comparative proteome
analysis using a quantitative cysteinyl-peptide enrichment technology. Anal. Chem.
76, 5345–5353.
88. Li, C., Hong, Y., Tan, Y. X., Zhou, H., Ai, J. H., Li, S. J., Zhang, L., Xia, Q. C.,
Wu, J. R., Wang, H. Y., and Zeng, R. (2004) Accurate qualitative and quanti-
tative proteomic analysis of clinical hepatocellular carcinoma using laser capture
microdissection coupled with isotope-coded affinity tag and two-dimensional liquid
chromatography mass spectrometry. Mol. Cell Proteomics 3, 399–409.
89. Sheehan, K. M., Calvert, V. S., Kay, E. W., Lu, Y., Fishman, D., Espina, V.,
Aquino. J., Speer, R., Araujo, R., Mills, G. B., Liotta, L. A., Petricoin, E. F.
3rd, and Wulfkuhle, J. D. (2005) Use of reverse phase protein microarrays and
reference standard development for molecular network analysis of metastatic
ovarian carcinoma. Mol. Cell Proteomics 4, 346–355.
90. Knezevic, V., Leethanakul, C., Bichsel, V. E., Worth, J. M., Prabhu, V. V., Gutkind,
J. S., Liotta, L. A., Munson, P. J., Petricoin, E. F. 3rd, and Krizman, D. B. (2001)
Proteomic profiling of the cancer microenvironment by antibody arrays. Proteomics
1, 1271–1278.
91. Sharma-Oates, A., Quirke, P., Westhead, D. R. (2005) TmaDB: a repository for
tissue microarray data. BMC Bioinformatics 6, 218.
92. Rai, A. J., Stemmer, P. M., Zhang, Z., Adam, B. L., Morgan, W. T., Caffrey,
R. E., Podust, V. N., Patel, M., Lim, L. Y., Shipulina, N. V., Chan, D. W.,
Semmes, O. J., and Leung, H. C. (2005) Analysis of human proteome organization
plasma proteome project (HUPO PPP) reference specimens using surface enhanced
laser desorption/ionization-time of flight (SELDI-TOF) mass spectrometry: multi-
institution correlation of spectra and identification of biomarkers. Proteomics 5,
3467–3474.
93. Engwegen, J. Y., Gast, M. C., Schellens, J. H., and Beijnen, J. H. (2006)
Clinical proteomics: searching for better tumour markers with SELDI-TOF mass
spectrometry. Trends Pharmacol. Sci. 27, 251–259.
94. Domon, B. and Aebersold, R. (2006) Mass spectrometry and protein analysis.
Science 312, 212–217.
95. Domon, B. and Aebersold, R. (2006) Challenges and opportunities in proteomics
data analysis. Mol. Cell Proteomics 5, 1921–1926.
96. Uhlen, M. and Ponten, F. (2005) Antibody-based proteomics for human tissue
profiling. Mol. Cell Proteomics 4, 384–393.
97. Taussig, M. J., Stoevesandt, O., Borrebaeck, C. A., Bradbury, A. R., Cahill, D.,
Cambillau, C., de Daruvar, A., Dubel, S., Eichler, J., Frank, R., Gibson, T. J.,
Gloriam, D., Gold, L., Herberg, F. W., Hermjakob, H., Hoheisel, J. D., Joos, T. O.,
Kallioniemi, O., Koegll, M., Konthur, Z., Korn, B., Kremmer, E., Krobitsch, S.,
Landegren, U., van der Maarel, S., McCafferty, J., Muyldermans, S., Nygren, P. A.,
Palcy, S., Pluckthun, A., Polic, B., Przybylski, M., Saviranta, P., Sawyer, A.,
Sherman, D. J., Skerra, A., Templin, M., Ueffing, M., and Uhlen, M. (2007)
Overview and Introduction to Clinical Proteomics 31
ProteomeBinders: planning a European resource of affinity reagents for analysis
of the human proteome. Nat. Methods 4, 13–17.
98. Ilyin, S. E., Belkowski, S. M., and Plata-Salaman, C. R. (2004) Biomarker
discovery and validation: technologies and integrative approaches. Trends
Biotechnol.22, 411–416.
I
Specimen Collection for Clinical
Proteomics
2
Specimen Collection and Handling
Standardization of Blood Sample Collection
Harald Tammen
Summary
Preanalytical variables can alter the analysis of blood-derived samples. Prior to the
analysis of a blood sample, multiple steps are necessary to generate the desired specimen.
The choice of blood specimens, its collection, handling, processing, and storage are
important aspects since these characteristics can have a tremendous impact on the results
of the analysis.
The awareness of clinical practices in medical laboratories and the current knowledge
allow for identification of specific variables that affect the results of a proteomic study.
The knowledge of preanalytical variables is a prerequisite to understand and control their
impact.
Key Words: blood; plasma; serum; proteomics; specimen; preanalytical variables.
1. Introduction
Proteomic analysis of blood specimens by semi-quantitative multiplex
techniques offers a valuable approach for discovery of disease or therapy-
related biomarkers (1,2). Based on reproducible separation of proteins by their
physical–chemical properties in combination with semi-quantitative detection
methods and bioinformatic data analysis, proteomics allows for sensitive
measurement of proteins in blood specimens (3). Blood can be regarded as
a complex liquid tissue that comprises cells and extracellular fluid (4). The
choice of a suitable specimen-collection protocol is crucial to minimize artificial
processes (e.g., cell lysis, proteolysis) occurring during specimen collection and
preparation (5). Preanalytic procedures can alter the analysis of blood-derived
From: Methods in Molecular Biology, vol. 428: Clinical Proteomics: Methods and Protocols
Edited by: A. Vlahou © Humana Press, Totowa, NJ
35
36 Tammen
samples. These procedures comprise the processes prior to actual analysis of
the sample and include steps needed to obtain the primary sample (e.g., blood)
and the analytical specimen (e.g., plasma, serum, cells). Legal or ethical issues
(e.g., importance of informed consents) or potential risks of phlebotomy (e.g.,
bleeding) are not covered in this article.
1.1. Collection of Blood Samples
It has been reported that the most frequent faults in the preanalytical phase
are the result of erroneous procedures of sample collection (e.g., drawing blood
from an infusive line resulting in sample dilution) (6). The design of blood
collection devices may aid in correct sampling: evacuated containers sustain
the draw of accurate quantity of blood to ensure the correct concentration of
additives or the correct dilution of the blood, such as in the case of citrated
plasma. The speed of blood draw is also controlled and restricts the mechanical
stress. The favored site of collection is the median cubital vein, which is
generally easily found and accessed. As such, it will be most comfortable to
the patient, and should not evoke additional stress. Preparation of the collection
site includes proper cleaning of the skin with alcohol (2-propanol). The alcohol
must be allowed to evaporate, since commingling of the remaining alcohol
with blood sample may result in hemolysis, raise the levels of distinct analytes,
and cause interferences. The position of the patient (standing, lying, sitting)
can affect the hematocrit (7), and hence may change the concentration of the
analytes. Tourniquet should be applied 3–4 inches above the site of venipuncture
and should be released as soon as blood begins flowing into the collection
device. The duration of venous occlusion (>1 min) can affect the sample
composition. Prolonged occlusion may result in hemoconcentration and subse-
quently increase the miscellaneous analytes, e.g., total protein levels. Blood
should be collected from fasting patients in the morning between 7 and 9 a.m.,
because ingestion or circadian rhythms can alter the concentration of analytes
considerably (e.g., total protein, hemoglobin, myoglobin).
1.2. Characteristics of Serum and Plasma Specimens
Serum is one of the most frequently analyzed blood specimens. The
generation of serum is time consuming and associated with the activation of
coagulation cascade and complement system. These processes influence the
composition of the samples, because they result in cell lysis (e.g., thrombo-
cytes, erythrocytes). As a consequence, the concentration of components in
the extracellular fluid, such as aspartate-aminotransferase, serotonin, neuron-
specific enolase, and lactate-dehydrogenase, are increased (8). On the other
hand, degradation of the analytes (e.g., hormones) may occur faster (9).Onthe
Specimen Collection and Handling 37
proteomic level, more peptides and less proteins are observed in serum when
compared to plasma (10,11).
Consequently, the activation of clotting cascades necessary to generate serum
can lead to artefacts. A reason to use serum as a specimen is based on
the notion that the proteome or peptidome of serum may reflect biological
events (12). Post-sampling proteolytic cleavage products have been proposed
as biomarkers, and it has been further suggested that serum peptidome is of
particular diagnostic value for the detection of cancer (13). However, it has
been reported that more protein changes occur in serum than in plasma (14).
Thus, it can be expected that the reproducibility of such ex vivo proteolytic
events is comparatively low.
In contrast to serum, citrate and EDTA inhibit coagulation and other
enzymatic processes by chelate formation with ions, thereby inhibiting ion-
dependent enzymes. This is in contrast to heparin, which acts through the
activation of antithrombin III. The main concern associated with heparinized
plasma for proteomic studies is that it is a poly-disperse charged molecule that
binds many proteins non-specifically (15,16), and may also influence separation
procedures and mass spectrometric detection of peptides and small proteins due
to its similar molecular weight (17).
The sampling of plasma is less time consuming than the acquisition of serum.
Separation of the cells and the liquid phase can be performed subsequently to
sample collection since no clotting time is required (30–60 min). In comparison
to serum, the amount of plasma generated from blood is approximately 10 to
20% higher. Additionally, the protein content of plasma is also higher than in
serum, because of the presence of clotting factors and associated components.
Furthermore, proteins may be bound to the clot, resulting in a decrease of
protein concentration.
1.3. Processing of Blood Samples
A quick separation of cells from the plasma is favorable, since cellular
constituents may liberate substances that alter the composition of the sample.
Generally, it is recommended that plasma and serum be centrifuged with
1300–2000×gfor 10 min within 30 min from the collection of the sample. The
temperature should generally be 15–24°C (18), unless recommended differently
for distinct analytes like gastrin or A-type natriuretic peptide. Processing at 4°C
appears to be attractive, because enzymatic degradation processes are reduced
at low temperatures. However, platelets become activated at low temperatures
(19) and release intracellular proteins and enzymes, which affect the sample
composition. Thus, processing at low temperatures is safe only after thrombo-
cytes have been removed. Since one centrifugation step may be insufficient for
38 Tammen
depletion of platelets below 10 cells/nL, a second centrifugation step (2500×g
for 15 min at room temperature) or filtration step may be required to obtain
platelet-poor plasma. This procedure is applicable only to plasma since the
platelets in serum are already activated.
1.4. Protease Inhibitors
Protease inhibitors would be attractive, but commonly used protease cocktails
may introduce difficulties due to interference with mass spectrometry and
formation of covalent bonds with proteins, which would result in shifting the
isoform pattern (20). Protease inhibitors have been considered and investigated as
additives in proteome research to prevent or slow down proteolytic processes and
thereby provide a means of more sensitive detection of markers in blood (21).
Even though protein integrity has been shown to be maintained by the
addition of 15 commercially available protease inhibitors, the usefulness of
protease inhibitors in overall protein stabilization of blood samples remains to
be investigated in more detail (22). The presence of certain protease inhibitors
in whole blood is toxic to live cells. Stressed, apoptotic, or necrotic cells release
substances, and it may be argued that this affects the composition of serum or
plasma until the cellular and soluble factions of blood are separated. However,
careful selection of an appropriate protease inhibitor may solve this problem.
2. Materials
1. Twenty gauge needles and an appropriate adapter (e.g., Sarstedt, Nümbrecht,
Germany) or a Vacutainer system (BD Bioscience, Franklin Lakes, USA).
2. Alcohol (2-propanol) in spray flask.
3. Swabs.
4. Examination gloves.
5. Tourniquet or sphygmomanometer.
6. Blood collection tubes (e.g., Sarstedt).
7. Centrifuge with a swinging bucket rotor (e.g., Sigma 4K15, Sigma Laborzen-
trifugen, Osterode, Harz).
8. A 10-mL syringe equipped with a cellulose acetate filter unit with 0.2 μm pore
size and 5 cm2filtration area (e.g., Sartorius Minisart, Sarstedt).
9. 2 mL cryo-vials.
10. Pipette and tips.
3. Methods
1. Venipuncture of a cubital vein is performed using a 20-gauge needle (diameter:
0.9 mm, e.g., butterfly system max. tubing length: 6 cm). If tourniquet is applied,
it should not remain in place for longer than 1 min (risk of falsifying results due to
Specimen Collection and Handling 39
hemoconcentration). As soon as the blood flows into the container, the tourniquet
has to be released at least partially. If more time is required, the tourniquet
has to be released so that circulation resumes and normal skin color returns to
extremity.
•Prior to blood collection for proteomic analysis, blood is aspirated into the
first container (e.g., 2.7 mL S-Monovette, Sarstedt, Nümbrecht, Germany).
This is done to flush the surface and remove initial traces of contact-induced
coagulation. This sample is not useful for analysis.
•Afterward, blood is drawn into a standard EDTA or citrate-containing syringe
(e.g. 9 mL EDTA-Monovette, Sarstedt, Nümbrecht, Germany). Depending on
ease of blood flow, several samples can be collected. Free flow with mild
aspiration should be assured to avoid haemolysis.
2. After venipuncture, plasma is obtained by centrifugation for 10 min at 2000×gat
room temperature. Centrifugation should start within 30 min after blood collection.
The resulting plasma sample may now be separated from red and white blood
cells in an efficient and gentle way. Nevertheless, a significant number of platelets
(∼25%) are still present in the sample. This requires an additional preparation
step.
3. For platelet depletion, one of the following procedures has to be undertaken
directly after step 2:
•Platelet removal by centrifugation: The plasma sample is transferred into a
second vial for another centrifugation for 15 min at 2500×gat room temper-
ature. After centrifugation, the supernatant is transferred in aliquots of 1.5 mL
into cryo vials.
•Platelet removal by filtration: Plasma aliquots of 1.5 mL resulting from step
2are transferred into 2-mL cryo vials using a 10-mL syringe equipped with
a cellulose acetate filter unit with 0.2 μm pore size and 5 cm2filtration area
(e.g., Sartorius Minisart®, Sartorius, Göttingen, Germany). Filtration requires
only gentle pressure.
4. Samples are transferred to an –80°C freezer within 30 min. Storage is at –80°C.
Transport of samples is done on dry ice.
4. Notes
4.1. Frequently Made Mistakes
4.1.1. Blood Withdrawal
•The patient was not fasting (i.e., had taken food prior to sampling).
•The blood was drawn from an infusive line.
•The blood was drawn in a wrong position (e.g., supine, upright).
•The consumables used were different than those recommended.
40 Tammen
•The expiry date of consumables was already reached.
•The tubes were not properly filled.
•The tubes were agitated vigorously (instead of gentle shaking to dissolve the antico-
agulant).
•The blood sample tubes were not consistently kept at room temperature.
•The sample tubes were put on ice or in a refrigerator.
.
4.1.2. Lab Handling
•Centrifugation was delayed more than 30 min after blood withdrawal.
•A cooling centrifuge was adjusted below room temperature.
•The centrifugation speed was wrong (e.g., rounds per minute were set instead of
g-force).
•The centrifugation time was wrong.
•The removal of blood plasma by pipetting was done without proper caution. Conse-
quently, the buffy coat or the red blood cells were churned up.
•The second centrifugation of recovered plasma samples was delayed after first
centrifugation.
4.1.3. Storage of Samples
•The storage of samples was delayed.
•The storage temperatures were above –80°C.
•The labeling of sample containers was unreadable or confusable.
•The attachment of labels to the sample containers was not proper during storage or
handling resulted in loss of labels.
4.1.4. General Recommendations
•A proper first centrifugation should produce a visible white blood cell layer (buffy
coat) between red blood cells and plasma. If not, centrifugation speed or time may
be wrong.
•One should discard plasma that is icteric or exhibits signs of haemolysis. One should
check with an expert if this was due to that particular disease.
References
1. Vitzthum F, Behrens F, Anderson NL, Shaw JH. (2005) Proteomics: from basic
research to diagnostic application. A review of requirements and needs. J. Proteome
Res. 4, 1086–97.
2. Lathrop JT, Anderson NL, Anderson NG, Hammond DJ. (2003) Therapeutic
potential of the plasma proteome. Curr. Opin. Mol. Ther. 5, 250–7.
Specimen Collection and Handling 41
3. Wang W, Zhou H, Lin H, Roy S, Shaler TA, Hill LR et al. (2003) Quantification of
proteins and metabolites by mass spectrometry without isotopic labeling or spiked
standards. Anal. Chem. 75, 4818–26.
4. Anderson NL, Anderson NG. (2002) The human plasma proteome: history,
character, and diagnostic prospects. Mol. Cell. Proteomics 1, 845–67.
5. Omenn GS. (2004) The Human Proteome Organization Plasma Proteome
Project pilot phase: reference specimens, technology platform comparisons, and
standardized data submissions and analyses. Proteomics 4, 1235–40.
6. Plebani M, Carraro P. (1997) Mistakes in a stat laboratory: types and frequency.
Clin. Chem. 43, 1348–51.
7. Burtis CA, Ashwood E. (eds) (2001) Fundamentals of Clinical Chemistry.
Saunders, Philadelphia.
8. Guder WG, Narayanan S, Wisser H, Zawata B. (2003) Samples: From the Patient to
the Laboratory. The Impact of Preanalytical Variables on the Quality of Laboratory
Results. GIT Verlag, Darmstadt, Germany.
9. Evans MJ, Livesey JH, Ellis MJ, Yandle TG. (2001) Effect of anticoagulants and
storage temperatures on stability of plasma and serum hormones. Clin. Biochem
34, 107–12.
10. Omenn GS, States DJ, Adamski M, Blackwell TW, Menon R, Hermjakob H et al.
(2005) Overview of the HUPO Plasma Proteome Project: results from the pilot
phase with 35 collaborating laboratories and multiple analytical groups, generating
a core dataset of 3020 proteins and a publicly-available database. Proteomics 5,
3226–45.
11. Rai AJ, Gelfand CA, Haywood BC, Warunek DJ, Yi J, Schuchard MD et al.
(2005) HUPO Plasma Proteome Project specimen collection and handling: towards
the standardization of parameters for plasma proteome samples. Proteomics 5,
3262–77.
12. Villanueva J, Shaffer DR, Philip J, Chaparro CA, Erdjument-Bromage H,
Olshen AB et al. (2006) Differential exoprotease activities confer tumor-specific
serum peptidome patterns. J. Clin. Invest. 116, 271–84.
13. Liotta LA, Petricoin EF. (2006) Serum peptidome for cancer detection: spinning
biologic trash into diagnostic gold. J. Clin. Invest. 116, 26–30.
14. Tammen H, Schulte I, Hess R, Menzel C, Kellmann M, Schulz-Knappe P. (2005)
Prerequisites for peptidomic analysis of blood samples: I. Evaluation of blood
specimen qualities and determination of technical performance characteristics.
Comb. Chem. High Throughput Screen. 8, 725–33.
15. Holland NT, Smith MT, Eskenazi B, Bastaki M. (2003) Biological sample collection
and processing for molecular epidemiological studies. Mutat. Res. 543, 217–34.
16. Landi MT, Caporaso N. (1997) Sample collection, processing and storage. IARC
Sci. Publ. 223–36.
17. Tammen H, Schulte I, Hess R, Menzel C, Kellmann M, Mohring T,
Schulz-Knappe P. (2005) Peptidomic analysis of human blood specimens:
comparison between plasma specimens and serum by differential peptide display.
Proteomics 13, 3414–22.
42 Tammen
18. Favaloro EJ, Soltani S, McDonald J. (2004) Potential laboratory misdiagnosis of
hemophilia and von Willebrand disorder owing to cold activation of blood samples
for testing. Am. J. Clin. Pathol. 122, 686–92.
19. Mustard JF, Kinlough-Rathbone RL, Packham MA. (1989) Isolation of human
platelets from plasma by centrifugation and washing. Methods Enzymol. 169, 3–11.
20. Schuchard MD, Mehigh RJ, Cockrill SL, Lipscomb GT, Stephan JD, Wildsmith J
et al. (2005) Artifactual isoform profile modification following treatment of
human plasma or serum with protease inhibitor, monitored by 2-dimensional
electrophoresis and mass spectrometry. Biotechniques 39, 239–47.
21. Jeffrey DH, Deidra B, Keith H, Shu-Pang H, Deborah LR, Gregory JO, Stanley AH.
(2004) An Investigation of Plasma Collection,Stabilization,and Storage Proce-
dures for Proteomic Analysis of Clinical Samples. Humana, Totowa, NJ.
22. Rai AJ, Vitzthum F. (2006) Effects of preanalytical variables on peptide and protein
measurements in human serum and plasma: implications for clinical proteomics.
Expert Rev. Proteomics 3, 409–26.
3
Tissue Sample Collection for Proteomics Analysis
Jose I. Diaz, Lisa H. Cazares, and O. John Semmes
Summary
Successful collection of tissue samples for molecular analysis requires critical consid-
erations. We describe here our procedure for tissue specimen collection for proteomic
purposes with emphasis on the most important steps, including timing issues and the proce-
dures for immediate freezing, storage, and microdissection of the cells of interest or “tissue
targets” and the lysates for protein isolation for SELDI, MALDI, and 2DGE applications.
The pathologist is at the cornerstone of this process and is an invaluable collaborator.
In most institutions, pathologists are responsible for “tissue custody,” and they closely
supervise the tissue bank. In addition, they are optimally trained in histopathology in
order to they assist investigators to correlate tissue morphology with molecular findings.
In recent years, the advent of the laser capture microscope, a tool ideally designed for
pathologists, has tremendously facilitated the efficiency of collecting tissue targets for
molecular analysis.
Key Words: tissue bank; frozen section; immunofluorescence; laser capture micro-
scope; proteomics.
1. Introduction
From the completion of surgery and the acquisition of tissue sample to
protein isolation and performing the various proteomic techniques, a number
of challenges must be overcome. The first challenge is time. Surgery is
associated with loss of vascular supply, resulting in progressive increase of
endogenous protease activity, protein degradation, and tissue autolysis. For
this reason, specimens submitted for tissue procurement must be processed
without delay. Formalin fixation, a standard processing procedure in pathology,
From: Methods in Molecular Biology, vol. 428: Clinical Proteomics: Methods and Protocols
Edited by: A. Vlahou © Humana Press, Totowa, NJ
43
44 Diaz et al.
stops protease activity. However, formalin is a cross-linking fixative that
irreversibly alters protein, thus compromising the quality of the extracts for
most proteomic techniques. Recent technical developments appear promising
and may ultimately enable peptide analysis and protein identification (bottom
up proteomics) in formalin-fixed paraffin embedded tissue (1). At present,
however, it is imperative to take a representative “fresh” tissue sample immedi-
ately after surgery when collecting tissue for proteomic studies, including
MALDI TOF MS and 2DGE. The surgical specimen should be transported
quickly to pathology, and a representative tissue sample should be obtained
under the supervision of a pathologist. The sample should be embedded in OCT
and frozen without delay. Ideally, a frozen section should be performed for
quality assurance before archiving the sample. Once the pathologist confirms
that the expected targets are present in the collected tissue (for instance, tumor
and non-tumor tissue), the frozen specimen can be stored in a –80°C freezer for
subsequent use. Overcoming time constraints requires appropriate institutional
policies and dedicated personnel. From our experience, it is better to delegate
the responsibility of transporting the surgical specimen from the operating room
to pathology to dedicated tissue procurement personnel, instead of expecting
the surgical team to deliver the specimens. When collecting and archiving tissue
samples, our policy is to bisect the sample into two halves, one embedded
in OCT and stored permanently at –80°C for future molecular studies, and
one submitted as a “mirror image” processed in formalin after performing a
frozen section for morphologic comparison and cell type mapping after basic
hematoxylin and eosin (H&E) staining. This formalin-processed mirror image
tissue provides optimal morphological detail, which might be necessary in
the future. For instance, it is very difficult to identify prostatic intraepithelial
neoplasia (PIN) on frozen section slides; however, the formalin fixed section,
which closely mimics the frozen section, can be used for guidance.
After archiving the tissue sample, the next challenge is to ensure that the
proteomic findings are representative of the tissue targets under investigation,
given the cellular heterogeneity present in most tissues. For instance, if one
would like to determine the differential protein expression in tumor versus
non-tumor, one must ensure that proteins are separately and reliably extracted
from normal and tumor cells. Certainly, many solid tumors are visible to the
naked eye, and both tumor and non-tumor tissues can be collected by gross
inspection. However, under a microscope, the tumor bed contains not only
tumor cells but many other tumor–associated, non-tumoral elements, such as
supporting stromal cells, blood vessels, infiltrating lymphocytes, etc. Moreover,
microscopic foci of tumor may infiltrate grossly normal tissue. In the past,
various approaches were followed to collect cells from tissue sections, including
manual microdissection with a syringe. In the recent years, the procedure
Tissue Sample Collection for Proteomics Analysis 45
of laser-capture microdissection (2) has tremendously increased the quality,
specificity, and speed of the process, allowing selective capture of cells and
various tissue elements while preserving the molecular integrity (3,4,5).
The LCM is a special microscope that isolates cells from frozen or formalin-
fixed tissues and cytological preparations. Microdissection of single cells or
multicellular structures is accomplished by placing a plastic polymer (cap) over
the tissue while pulsing an infrared laser for the polymer to melt and adhere
to the target cells under the laser ring. When the cap is removed, the cells
that adhered to the polymer detach from the surrounding tissue without any
molecular damage, becoming suitable for the extraction of high-quality nucleic
acids and proteins, and for a wide range of downstream molecular analyses,
AB
DC
Fig. 1. Selective immunofluorescent LCM of prostate gland’s basal cells by immuno-
capture: (A) immunofluorescent staining of basal cells with a mAb against high-
molecular-weight keratins, which are highly expressed on basal cells, (B) selection
of immunofluorescent-positive basal cells for subsequent LCM, (C) captured
immunofluorescent-positive cells after LCM photographed from the plastic cap,
(D) remaining of the gland after removing the basal cell layer by LCM.
46 Diaz et al.
such as gene expression microarrays, or proteomics. The use of a microscope
can be coupled with special immunostaining procedures if one wishes to capture
specific cell types not easily identified by morphology alone, which is the
“so called” immunocapture procedure (6,7), which further enhances the speci-
ficity of tissue procurement for molecular analysis. For example, in a former
study (8), we were able to selectively capture basal cells from benign prostate
glands, which are extremely difficult to recognize morphologically but easily
identifiable after immunostaining for high-molecular-weight cytokeratin (Fig. 1).
We obtained excellent protein quality results and were able to identify several
protein peaks preferentially expressed in these cells using SELDI-TOF-MS.
When we compared the protein spectra from the same tissue sample sections
routinely stained with hematoxilin with those immunostained for high-molecular-
weight cytokeratins, there was no difference in the spectra, militating against
any significant protein deterioration due to the immunostaining procedure.
2. Materials
2.1. Tissue Collection and Storage
1. Tissue-Tek Cryomold-standard (Sakura, Torrance, CA)
2. Tissue-Tek OCT (Sakura)
3. 2methylbutane (Mallinckrodt, St. Louis, MO)
4. Shandon Histobath II (Thermo Electron Corp., Waltham, MA)
5. –80°C freezer
2.2. Frozen Tissue Sectioning and Staining
1. Cryostat
2. HistoGeneTM LCM Frozen Section Staining Kit (Arcturus Biosciences Inc,
Mountain View, CA). The kit contains histogene staining solution, ethanol (75,
95, 100%), xylene, distilled water nuclease free, histogene LCM slides, and
disposable slide staining jars.
3. 1×PBS made from 10×stock (Fisher Scientific)
4. Acetone (high purity grade)
5. Cy3-Strepavidin (Invitrogen, Carlsbad, CA)
6. Biotinylated mAbs: Any antibody can be biotinylated. We routinely have 1.5 mg of
antibody labeled with 0.2 mg biotin (Alpha Diagnostic Intl. Inc. San Antonio, TX).
2.3. LCM
1. PixCell II LCM System (Arcturus Biosciences Inc)
2. AutoPixTM Automated LCM System (Arcturus Biosciences Inc)
3. CapSure®LCM caps (Arcturus Biosciences Inc)
4. Prep Strip (Arcturus Biosciences Inc)
5. Microcentrifuge tubes (0.5 ml) (Eppendorf North America)
Tissue Sample Collection for Proteomics Analysis 47
2.4. LCM Lysate
1. Micropipet capable of delivering 1 μl accurately
2. 20 mM HEPES (pH to 8.0 with NaOH) with 1% Triton X-100
3. Sonicator (optional)
4. 1×PBS
2.5. SELDI Analysis
1. IMAC3 or WCX2 Protein Array Chips (Ciphergen Biosystems Palo Alto, CA)
2. HPLC grade water (Fisher Scientific)
3. 100 mM sodium acetate pH 4.0
4. 100 mM ammonium acetate pH 4.0
5. Sinapinic acid (SPA) (Ciphergen Biosystems, Palo Alto, CA)
6. Optima grade Acetonitile (Fisher Scientific)
7. Trifluoroacetic acid, packaged in 1 ml ampules (Pierce Chemical Company,
Rockford, IL)
2.6. MALDI Analysis
1. Target plate
2. Cinaminic acid (CHCA) (Bruker Daltonics, Palo Alto, CA)
3. SPA (Fluka)
4. Optima grade Acetonitile (Fisher Scientific)
5. Trifluoroacetic acid, packaged in 1 ml ampules (Pierce Chemical Company)
3. Method
3.1. Tissue Collection and Storage
1. The tissue sample is embedded in OCT using a cryomold and is frozen in the
Shandon Histobath, which contains 2methylbutane (see Note 1).
2. Hold the cryomold against the 2methylbutane liquid interface and allow the
tissue to freeze slowly (3–5 min) (see Note 2).
3. After achieving complete freezing, place the frozen cryomold containing the
sample in a plastic bag and transport the sample within a liquid nitrogen container.
Store the sample in a –80°C freezer.
3.2. Frozen Tissue Sectioning and Staining
3.2.1. Regular Hematoxylin Staining
Prior to LCM, cut 8-μm-thick frozen tissue sections from the cryostat (discard
folded or wrinkled sections). Keep slides with sections in cryostat after cutting
and stain as follows (see Notes 3 and 9; slides may also be frozen at –80°C
until stained.):
48 Diaz et al.
1. Remove the slides from the freezer or cryostat and place in 70% ethanol (30 s).
2. Place in purified water (5 s).
3. Add the Histogene staining solution (30 s) (see Note 4).
4. Rinse the slides with purified water.
5. Wash with 70% ethanol (60 s).
6. Wash with 95% ethanol twice (60 s each).
7. Wash with 100% ethanol (60 s).
8. Place the slides in xylene to ensure complete dehydration (10 min) (see Note 5).
9. Shake off and drain carefully by touching the corner with a particle-free tissue
paper.
10. Air dry the slides to allow xylene to evaporate completely (at least 2 min).
11. The slides are now ready for LCM (they should not be coverslipped) (see
Note 12)
3.2.2. Immunofluorescence Staining (see Note 7)
1. Thaw slides (1 min).
2. Place in cold acetone at 4°C (2 min).
3. Air dry (30 s).
4. Wash in filtered pH 7.4 1×PBS.
5. Drain off slides.
6. Add 100 μl of first biotinylated Ab at optimal dilution: recommended concen-
tration 30–100 μg/ml, optimize for best results (3 min).
7. Rinse in PBS.
8. Add 100 μl of Cy3 at dilution 1:100 (user may decide the optimal staining
concentration of the Cy3 Streptavidin conjugate by performing a serial dilution
staining experiment) (1 min).
9. Rinse in PBS.
10. Place slides in 75% ethanol (30 s).
11. Place slides in 95% ethanol (30 s).
12. Place slides in 100% ethanol (30 s).
13. Place slides in xylene (5 min) (see Note 6).
14. Air dry (5 min).
3.3. LCM
The new instruments developed by Arcturus, such as the AutoPixTM and the
VeritasTM are enclosed in automated systems entirely operated by a computer.
We describe here the LCM procedure using the PixCell II instrument, which
is manually operated and the least expensive LCM instrument today and,
therefore, more widely used (see Note 8).
1. Turn on the instrument and enter pertinent data such as slide #, case #, cap lot #,
thickness (always 8 μm), and place the stained slide on the mechanical stage (see
Note 10).
Tissue Sample Collection for Proteomics Analysis 49
2. Turn on the vacuum pump to immobilize the slide (small aperture on the left side
of the stage) and push in the filter bottom for optimal image quality.
3. Place the caps in the rail on the right side of the stage. Unlock the mechanical arm,
move it toward the tissue, and drop it at the top of the tissue. Align the joystick
to move the stage to a centered and perpendicular position before beginning the
microdissection process.
4. Turn on the key on the right side of the power supply to enable the infrared laser.
Focus the laser before beginning microdissection using the smallest ring diameter
and adjust to the desired diameter.
5. Select the appropriate energy (mW) and time of exposure (ms) for the desired
laser ring diameter and ensure its effectiveness in an area of the tissue that lacks
any interest using a cap to be discarded (see Note 11).
6. Fire the laser each time the ring is over the desired tissue target. Move the stage
supporting the glass slide with the aid of the joystick, which allows fine and
precise motion. Check if the tissue is appropriately microdissected and capture
the tissue images before and after LCM as well as the image of the target tissue
that was captured in the cap (see Note 13).
7. When the cap is filled with the desired amount of tissue, remove the cap and use a
0.5-ml microcentrifuge tube to collect the tissue (the cap is designed to perfectly
fit to close the tube) (see Note 14).
8. The microcentrifuge tube can be safely stored in a –80°C freezer without adding any
buffer and without lysing the cells, which may be done at a convenient time later.
3.4. LCM Lysate
1. Lyse a total of 1500–2000 laser shots (about 3000 to 6000 microdissected cells)
in 4 μl of 20 mM Hepes pH 8.0 with 1% Triton X-100. This is sufficient for
one SELDI protein array or one MALDI run. For 2D analysis, a minimum of
approximately 25,000 cells are necessary.
2. Add the above lysing buffer on the cap and place in the microfuge tube holding
the cap. This is usually done with two additions of 2 μl to the LCM cap. Pipet
up and down and scrape the surface of the LCM cap to remove all the cells. A
gentle scraping motion with the pipet tip may be necessary to remove the cells,
but be careful not to rip the polymer film (see Note 15). Transfer the lysate
from the surface of the cap to the microfuge tube. Cells from multiple caps may
be combined by subsequently using 4 μl of LCM lysate to lyse cells on another
cap. In this way the volume will remain small. If 2DGE may be performed,
the lysis procedure is different (see below). Make a 1:10 dilution of each lysate
in PBS (for IMAC3 SELDI chips) or 100 mM ammonium acetate pH 4.0 (for
WCX2 chips) (i.e., 36 μl added to the 4 μl lysate) vortex for at least 1 min (see
Note 16). Spin down briefly.
3. Prepare the arrays of the IMAC chip with CuSO4according to the manufacturer’s
specifications: 20 μl, 100 mM CuSO4for 10 min, wash with HPLC water; 20 μl,
100 mM Na acetate pH 4.0 for 5 min, wash with water. Use the Micromix
shaker for all incubations with the following settings: Form-20, Amplitude-5.
50 Diaz et al.
4. Assemble the bioprocessor with the desired number of chips and add 2×200 μl
PBS to each well, incubate on the shaker for 5 min each time. Pretreat the
WCX2 chip with 100 mM ammonium acetate pH 4.0. This can be done on the
BioMek robot.
5. Add the diluted lysate to the spot on the chip(s) in the bioprocessor.
6. Cover the bioprocessor with a plastic seal and incubate overnight on MicroMix
shaker at room temperature, using the same setting as given above.
7. Remove lysates carefully with a pipet; do not touch the surface of the arrays.
Save if needed for another experiment.
8. Wash the spots in bioprocessor 2×with 200 μl PBS (for IMAC) or 100 mM
ammonium acetate pH 4.0 (for WCX) for 5 min on the shaker.
9. Wash the arrays with HPLC water 2×for 5 min (on shaker).
10. Remove the chip(s) from bioprocessor and give them a final rinse with HPLC
water.
11. Let the chip dry completely, usually overnight.
12. Add 2×0.5 μl saturated SPA dissolved in 50% acetonitrile, 0.5% TFA.
13. Read at instrument settings optimized for resolution and intensity for the m/z
range of 1000–20,000. Higher laser energy will be required to see higher
molecular weight peaks.
One method of MALDI sample preparation that reduces the complexity of cell
lysates while remaining robust and easily amenable to automated high-
throughput applications is sample fractionation using magnetic beads
(MB) combined with pre-structured MALDI sample supports (AnchorChip
Technology). Several magnetic bead types with different surface chemistries can
be used to fractionate serum and increase the number of detectable peaks (see
the chapter on serum protein profiling for details). For MALDI analysis, dilute
the lysate 1:10 with CHCA or SPA matrix (5–10 mg/ml in 50% acetonitrile, 0.1%
TFA). Spot on Anchorplate and read in a MALDI instrument. Further dilution
and/or fractionation of the lysate may be necessary to achieve optimal spectra.
If 2DGE analysis will be performed, the cells should be lysed as follows:
Remove the LCM cap from the tube and add a small volume (10 μl) of 1D
focusing rehydration buffer to the tube. The preferred number of laser shots is
approximately 100 K. Replace the cap and invert the tube to allow the buffer
to come in contact with the cells on the cap and lyse them. Incubate 5 min
at room temperature. Sonicate the samples to ensure lysis. Continue with the
basic protocol for 1D IEF and 2D analysis.
4. Notes
1. In our experience, a time window of 30 min between completion of surgery
and tissue freezing yields good protein quality for most proteomic techniques.
However, if one is studying protein phosphorylation, this begins to significantly
decrease 20 min after completion of surgery (10).
Tissue Sample Collection for Proteomics Analysis 51
2. When freezing the tissue sample in the Histobath, avoid immediate and complete
immersion in 2methylbutane to preserve optimal tissue morphology. Hold the
sample at the liquid interface with minimal immersion and wait until the OCT
and the tissue slowly turn white.
3. Use uncoated glass slides for LCM. Coated or electrically-charged glass slides
will interfere with the detachment process of the plastic polymer and are not
suitable for LCM.
4. Precipitate from Hematoxylin can contaminate the surface of the tissue. Filter
these solutions. Add one tablet of protease inhibitor to each staining bath (we use
Complete, from BMB). Do not add protease inhibitor to alcohol baths. If using
the histogene staining kit (Arcturus) for frozen sections, this is not necessary.
5. Change all the staining and alcohol solutions after staining 20 slides.
6. Poor transfers may result if 100% ethanol has hydrated. Increasing the incubation
time in xylene often improves transfer.
7. When specific cells need to be microdissected and these cannot be identified
morphologically, the cells of interest can be immunostained with specific mAbs
against proteins highly expressed on those cells (immunophenotype). It is critical
to expedite the immunostaining procedure because the shorter the immunos-
taining time, the better the protein quality. One must avoid exceeding 30 min for
the total immunostaining and dehydration procedure. In the past, we have used
the immunoperoxidase technique with DAB labeling (6), but it was difficult
to perform quick enough to preserve optimal protein integrity. Also, manual
microdissection of DAB labeled cells with Pixel II is extremely tedious and non-
practical. The immunofluorescence staining method (7) is faster and easier to
perform. This method coupled with the Autopix microscope, which has dark field
fluorescence and automation capabilities, is the ideal procedure for immuno-
capture. Since Cy3-strepavidin binds to the antibody labeled with biotin, there is
no need for a secondary antibody, thereby decreasing the necessary staining time.
It is recommended to run negative control staining; use a biotinylated control
antibody from the same animal species and of the same isotype as your primary
antibody. Dilute to the same working concentration as the primary antibody.
8. Do not forget to wear gloves every time while performing LCM, including when
handling the plastic caps.
9. The thickness of the tissue section is a critical parameter for effective LCM. In
our experience (using the Pixel II and the Autopix instruments by Arcturus),
8 μm is the optimal thickness for LCM.
10. Smooth out the surface of the tissue section with a Prep-strip before placing the
slide on the LCM instrument, which improves the efficiency and uniformity of
the microdissection process.
11. The main factors affecting the efficiency of LCM include the energy, the time
of exposure, and the diameter of the laser beam. Regarding the diameter, when
using Pixel II, the smallest ring is 7 μm, the medium ring is 15 μm, and the widest
ring is 30 μm. Very often, we have used the medium (15 μm, which lifts up
about three cells with each shot). When trying to microdissect single cells with
52 Diaz et al.
Pixel II, one must use the smallest (7 μm) diameter ring, but our experience was
frustrating. With Autopix, we have observed that microdissection of individual
cells is better achieved setting the laser ring at 10 μm diameter, below which it
becomes very difficult to lift up cells efficiently. A 30-μm diameter laser is very
effective for microdissection of whole glands and other large tissue structures.
Regarding the other two parameters, the optimization depends on the tissue
type. For instance, for prostate tissue, an energy of 80 mW with a duration
of 0.5 ms is usually effective for a medium-size ring (15 μm). The tuning of
these parameters is accomplished by a “fail and try” approach, progressively
adjusting the energy and the time of exposure for the desired diameter, which
obviously depends on the desired microdissection task (single cells vs. medium-
or large-size tissue structures).
12. Another factor that affects the effectiveness of LCM is the time the tissue section
has been dry after the staining and dehydration procedure. Ideally, the tissue
should be stained and microdissected within1hifpossible. One must avoid
having the slide under LCM for more than 4 h. If microdissecting many tissues,
stain only four slides at a time.
13. When capturing images before and after microdissection for documentation
purposes, make sure the image on the monitor is focused because that is the
image that would be captured. Sometimes is focused on the microscope but is
unfocused on the monitor. In a typical experiment, you will capture the image
before and after firing the laser, which provides records of the effectiveness in
removing the cell targets. You can also capture the image of microdissected
cells from the polymer cap.
14. Avoid allowing the LCM caps to become excessively crowded. When using
the 15-μm laser ring, microdissection is about three cells per shot. One should
expect around 3000 cells for each 1000 shots, which is about right per single
cap.
15. LCM caps can be viewed under a dissecting microscope to ensure that all cells
have been removed from the polymer film after the lysing procedure.
16. Depending on the cell type, vigorous vortexing and sonication may be necessary
to completely lyse the cells after they are removed from the cap.
References
1. Prieto, D.A., Hood, B.L., Darfler, M.M., Guiel, T.G., Lucas, D.A., Conrads, T.P.,
Veenstra, D.T., and Krizman, D.B. (2005) Liquid TissueTM: proteomic profiling of
formalin-fixed tissues. Biotechniques 38: 32–5.
2. Emmert-Buck, M.R., Bonner, R.F., Smith, P.D., Chuaqui, R.F., Zhuang, Z.,
Goldstein, S.R., Weiss, R.A., and Liotta, L.A. (1996) Laser capture microdissection.
Science 274: 998–1001.
3. Espina, V., Milia, J., Wu, G., Cowherd, S., Liotta, L.A. (2006) Laser capture
microdissection. Methods Mol Biol 319: 213–29.
Tissue Sample Collection for Proteomics Analysis 53
4. Best, C.J., and Emmert-Buck, M.R. (2001) Molecular profiling of tissue samples
using laser capture microdissection. Expert Rev Mol Diagn. 1: 53–60.
5. Ornstein, D.K., Gillespie, J.W., Paweletz, C.P., Duray, P.H., Herring, J.,
Vocke, C.D., Topalian, S.L., Bostwick, D.G., Linehan, W.M., Petricoin, E.F., III,
and Emmert-Buck, M.R. (2000) Proteomic analysis of laser capture microdis-
sected human prostate cancer and in vitro prostate cell lines. Electrophoresis 21:
2235–42.
6. Fend, F., Emmert-Buck, M.R., Chuaqui, R., Cole, K., Lee, J., Liotta, L.A., and
Raffeld, M. (1999) Immuno-LCM: laser capture microdissection of immunostained
frozen sections for mRNA analysis. Am J Pathol 154: 61–6.
7. Murakami, H., Liotta, L., Star, R.A. (2000) IF-LCM: laser capture microdissection
of immunofluorescently defined cells for mRNA analysis rapid communication.
Kidney Int 58(3):1346–53.
8. Cazares, L.H., Adam, B.L., Ward, M.D., Nasim, S., Schellhammer, P.F.,
Semmes, O.J., and Wright, G.L., Jr (2002) Normal, benign, preneoplastic, and
malignant prostate cells have distinct protein expression profiles resolved by
surface enhanced laser desorption/ionization mass spectrometry. Clin Cancer Res
8: 2541–52.
9. Diaz, J., Cazares, L.H., Corica, A., and Semmes O. (2004) Selective capture
of prostatic basal cells and secretory epithelial cells for proteomic and genomic
analysis. Urol Oncol 22(4):329–36.
10. Mora, L., Buettner, R., Seigne, J., Diaz, J., Hamad, N., Garcia, R., Bowman, T.,
Falcone, R., Faigurth, R., Cantor, A., Muro-Cacho, C., Livistong, S., Levitzki, A.,
Kraker, A., Karras, J., Pow-Sang, J., and Jove, R. (2002) Constitutive activation of
Stat3 in human prostate tumors and cell lines: direct inhibition of stat3 signaling
induces apoptosis of prostate cancer cells. Cancer Research 62: 6659–66.
4
Protein Profiling of Human Plasma Samples
by Two-Dimensional Electrophoresis
Sang Yun Cho, Eun-Young Lee, Hye-Young Kim, Min-Jung Kang,
Hyoung-Joo Lee, Hoguen Kim, and Young-Ki Paik
Summary
Human plasma is regarded the most complex and well-known clinical specimen that
can be easily obtained; alterations in the levels of plasma proteins or their corresponding
enzyme activities may reflect either a healthy or a diseased state. Given that there is
no defined genomic information as to the intact protein components in plasma, protein
profiling could be the first step toward its molecular characterization. Several problems
exist in the analysis of plasma proteins, however. For example, the widest dynamic range
of protein concentrations, the presence of high-abundance proteins, and post-translational
modifications need to be considered before proteomic studies are undertaken. In particular,
efficient depletion or pre-fractionation of high-abundance proteins is crucial for the identifi-
cation of low-abundance proteins that may contain potential biomarkers. After the removal
of high-abundance proteins, protein profiling can be initiated using two-dimensional
electrophoresis (2DE), which has been widely used for displaying the differential proteome
under specific physiological conditions. Here, we describe a typical 2DE procedure for
plasma proteome under either a healthy or a diseased state (e.g., liver cancer) in which
pre-fractionation and depletion are integral steps in the search for disease biomarkers.
Key Words: 2-dimensional gel electrophoresis; plasma; HPPP; immunoaffinity
column.
Abbreviations: IEF: Isoelectric Focusing, IPG; Immobilized pH Gradient, TCA:
Trichloroacetic Acid, FFE: Free Flow Electrophoresis, HPMC: Hydroxypropyl Methyl-
cellulose, TBP: Tributylphosphine, 2DE: 2-dimensional Gel Electrophoresis, BPB:
Bromophenol Blue, CHCA: -cyano-4-hydroxycinnamic acid, LTQ: Linear Iontrap
From: Methods in Molecular Biology, vol. 428: Clinical Proteomics: Methods and Protocols
Edited by: A. Vlahou © Humana Press, Totowa, NJ
57
58 Cho et al.
MALDI-TOF: Matrix-assisted Laser Desorption Ionization - Time of Flight Mass
Spectrometry, HPPP: Human Plasma Proteome Project.
1. Introduction
Human plasma is an intravascular fluid that serves as a liquid medium
for blood proteins that are derived from various cells, tissues, and other
biofluids (1). In fact, the components of plasma are very heterogeneous,
including inorganic ions (e.g., bicarbonate, calcium), metabolic intermediates
(e.g., cholesterol, glucose), and plasma proteins (e.g., albumin, globulin), which
are important in maintaining body fluid balance, immune response, blood
clotting, and other metabolic mechanisms of homeostasis. Plasma contains
many different proteins that are primarily synthesized in the liver and are often
subjected to post-translational modification (PTM) (2).
Since human plasma is the most complex and well-known clinical specimen
that can be easily obtained, it has been a central target for many biomedical
studies (2). Alterations in the levels of plasma proteins or their corresponding
enzyme activities may reflect either a healthy or a diseased state that can
be monitored by various analytical tools, including biochemical assays and
proteomics. Given that there is no defined genomic information as to the
intact protein components in plasma, a proteomic study may be the method of
choice (3,4). Recently, plasma protein profiling was conducted as part of the
plasma proteome project of HUPO, termed HPPP (5). The pilot phase of HPPP
produced 3020 non-redundant proteins that were found to be present in human
plasma and serum (5,6).
However, several points must be addressed before proteomic studies are
undertaken. First, plasma protein is believed to contain the most dynamic
concentration range (more than 10 orders of magnitude) of each constituent
protein, creating many technical obstacles in proteomic detection by mass
spectrometry (MS) (2,3). For example, the removal of high-abundance proteins
(e.g., albumin, IgG, transferrin, fibrinogen, IgA, etc.) that occupy more than
90% of all plasma proteins prior to biochemical analysis may be a big
challenge and perhaps even problematic in light of plasma-derived biomarker
discovery (3,7). Second, since many plasma proteins have many structural
isoforms, more efficient analytical system is needed to facilitate the analysis
of multiple isoforms of plasma proteins (1). Third, since many plasma proteins
are synthesized as pre-proteins that are subjected to various PTMs for cellular
function, more efficient methods to analyze modified proteins (e.g., glycosy-
lated proteins) are required. For example, since glycopeptides are not easily
ionized completely during MS analysis, which leads to inadequate spectral
data and low detection sensitivity due to the attached glycans, a strategy
Protein Profiling by Two-Dimensional Electrophoresis 59
for the removal of glycans must be considered for protein identification.
Taken together, all these factors are important for the proteomic study of
plasma (8).
Of the problems listed above, the first problem that concerns the protein
profiling of plasma may be the depletion or pre-fractionation of high-abundance
plasma proteins (3,4,7). Without this depletion procedure, the identification of
low-abundance proteins (including biomarkers) may not be practical. After the
removal of high-abundance proteins, two-dimensional electrophoresis (2DE)
may be the first step chosen to analyze plasma proteins because it is easy to
perform in the laboratory. Although 2DE has several limitations in terms of
reproducibility, separation of membrane or low-molecular-weight proteins, and
proteins with extreme pIs (<3 or >10), this technique has been widely used
as a first analysis of proteins in a particular physiological state when coupled
with MS (9). Recently, quantitative 2DE was performed with a difference in
gel electrophoresis (DIGE) system (see Chapter by Friedman and Lilley for
detail), where two or three differentially staining dyes can be applied to specific
protein populations to determine their quantitative changes in expression levels
under a specific physiological condition (10). Thus, this chapter is intended
to provide the reader with necessary information on the systematic analysis
of the plasma proteome using 2DE in an attempt to search for disease
biomarkers from the plasma proteins of patients with hepatocellular carcinoma
(HCC) (11,12).
2. Materials
2.1. Preparation of Human Plasma Samples
1. Blood collection tubes: BD Plus Plastic K2EDTA (BD, 367525; 10 mL), BD
Glass Serum with silica clot activator (367820, 10 mL).
2. Protease inhibitor (Complete Protease Inhibitor Cocktail, Roche, 11 697 498 001,
20 tablets): One tablet contains protease inhibitors (antipain, bestatin, chymostatin,
leupeptin, pepstatin, aprotinin, phosphoramidon, and EDTA) sufficient for the
processing of 100 mL plasma samples. Prepare 25× stock solutions in 2 mL
distilled water.
2.2. Depletion of High-Abundance Proteins with an Immunoaffinity
Column
1. HPLC system, such as the HP1100 LC system (Agilent).
2. Multiple affinity removal system (MARS): LC column (Agilent, 5185-5984);
Buffer A for sample loading, washing, and equilibrating (Agilent, 5185-5987);
Buffer B for eluting (Agilent, 5185-5988).
60 Cho et al.
2.3. Isoelectric Focusing (IEF) with Immobilized pH Gradient (IPG)
Strip
1. MultiPhorTM (GE Healthcare) or Protean IEF cell (Bio-Rad): Numerous commer-
cially available isoelectric focusing units exist
2. Re-swelling tray
3. Mineral oil: Immobiline Dry Strip Cover Fluid (GE Healthcare)
4. Power supply, such as the EPS 3501 XL power supply (GE Healthcare)
5. Thermostatic circulator: Multitemp III thermostatic circulator (GE Healthcare)
6. IPG strip: Immobiline Dry Strip, pH 3-10 nonlinear (NL), or pH 4.0-5.0, and pH
5.5-6.7, 18 cm long, 0.5 mm thick (GE Healthcare) or with the same pH ranges
for ReadyStrip IPG strip (Bio-Rad)
7. Carrier ampholyte mixtures: IPG buffer or Pharmalyte, same range as the selected
IPG strip
8. Sample buffer: 7 M urea, 2 M thiourea, 4% (w/v) CHAPS, 0.5% (v/v) ampholyte,
100 mM DTT, 40 mM Tris-HCl, pH 7.5, a trace amount of bromophenol blue
(BPB)
2.4. Microscale Solution Isoelectric Focusing: ZOOM®
1. ZOOM®(IEF Fractionator (Invitrogen, ZF10001)).
2. ZOOM®disks: pHs 3.0, 4.6, 5.4, 6.2, 7.0, and 10.0 [Invitrogen, ZD series (e.g.,
ZD10030 for pH 3.0)]
3. IEF Anode Buffer (50X) (Novex, LC5300, 100 mL)
4. IEF Cathode Buffer (10X) (Novex, LC5310, 125 mL)
5. Anode buffer: 8.4 g urea, 3.0 g thiourea, 3.3 mL Novex®IEF Anode Buffer
(50X). Add water to a final volume of 20 mL.
6. Cathode buffer: 8.4 g urea, 3.0 g thiourea, 3.3 mL Novex®IEF Cathode Buffer
(50X). Add water to a final volume of 20 mL.
2.5. Fractionation of Plasma Samples by Free Flow Electrophoresis
(FFE)
1. ProTeamTM FFE instrument (Tecan)
2. 1% 2-(4-sulfophenylazo)-1,8-dihydroxy-3,6-naphthalenedisulfonic acid (SPAD-
NS) (Tecan, 517074)
3. 0.8% hydroxypropyl methylcellulose (HPMC) (Tecan, 5170709)
4. pI markers: mixture of pI markers that indicate pHs 4.2, 5.1, 6.3, 7.4, 8.7, and
10.1 (Tecan, 5170705)
5. ProlyteTM 1, ProlyteTM 2, and ProlyteTM 3 (Tecan, 0309081, 0309102, and
0309093)
6. Anodic stabilization medium (Inlet I1): 14.5% (w/w) glycerol, 8 M urea, 0.03%
(w/w) HPMC, 100 mM H2SO4
7. Separation medium 1 (Inlet I2): 14.5% (w/w) glycerol, 8 M urea, 0.03% (w/w)
HPMC, 14.5% (w/w) ProlyteTM 1
Protein Profiling by Two-Dimensional Electrophoresis 61
8. Separation medium 2 (Inlet I3−5): 14.5% (w/w) glycerol, 8 M urea, 0.03% (w/w)
HPMC, 14.5% (w/w) ProlyteTM 2
9. Separation medium 3 (Inlet I6): 14.5% (w/w) glycerol, 8 M urea, 0.03% (w/w)
HPMC, 14.5% (w/w) ProlyteTM 3
10. Cathodic stabilization medium (Inlet I7): 14.5% (w/w) glycerol, 8 M urea, 0.03%
(w/w) HPMC, 100 mM NaOH
11. Counter flow medium (Inlet I8): 14.5% (w/w) glycerol, 8 M urea
12. Anodic circuit electrolyte: 100 mM H2SO4
13. Cathodic circuit electrolyte: 100 mM NaOH
2.6. Preparation of 2D Gels
1. Gradient former: One of the two Bio-Rad models can be used in this step: Model
385 (30-100 mL capacity) or Model 395 (100-750 mL capacity).
2. Orbital shaker with speed controller.
3. SDS-PAGE: Protean II xi multicell and multicasting chamber (Bio-Rad) or Ettan
DALT twelve large vertical system (GE Healthcare).
4. 5× Tris-HCl buffer: Dissolve 227 g Tris into 800 mL distilled water and adjust
the buffer to pH 8.8 with HCl (∼30 mL). Add distilled water to a final volume
of1L.
5. 5× Gel buffer: Dissolve 15 g Tris, 72 g glycine, and 5 g sodium dodesyl sulfate
(SDS) into 800 mL distilled water and add distilled water to a final volume
of1L.
6. SDS Equilibration buffer contains 6 M urea, 2% (w/v) SDS, 5× gel buffer (pH
8.8), 50% (v/v) glycerol, and 2.5% (w/v) acrylamide monomer.
7. Acrylamide stock solution: Acrylamide/Bis-acrylamide 37:5.1, 40% (w/v)
solution (Amresco, M157, 500 mL).
8. Fixing solution: 40% (v/v) methanol and 5% (v/v) phosphoric acid in distilled
water.
9. Coomassie blue G-250 staining solution: 17% (w/v) ammonium sulfate, 3% (v/v)
phosphoric acid, 34% (v/v) methanol, and 0.1% (w/v) Coomassie blue G-250 in
distilled water.
2.7. 2D Gel Image Analysis
1. Scanner with transparency unit, such as Bio-Rad GS710 or GS800
2. 2D gel image analysis program: Image Master Platinum 5 (GE Healthcare),
PDQuest 7.3.0 (Bio-Rad), or Progenesis Discovery (NonLinear Dynamics, Ltd.)
2.8. Destaining, In-gel Deglycosylation, and In-gel Tryptic Digestion
1. Speed Vac (Heto)
2. PNGase F stock solution for in-gel deglycosylation PNGase F (Glyko, Inc, GKE-
5010). Dilute 1 μL PNGase F (2 mU) with 2.5 mL 1× N-glycanase incubation
buffer (20 mM sodium phosphate, pH 7.5, and 0.02% (w/v) sodium azide)
62 Cho et al.
3. Sequencing-grade modified trypsin (Promega, V5111, 100 μg, 18,100 U/mg)
4. 50 mM ammonium bicarbonate
2.9. Desalting of Peptides and MALDI Plating
1. GELoader tips (Eppendorf, No. 0030 048.083, 20 μL capacity)
2. Poros 10 R2 resin (PerSeptive Biosystems, 1-1118-02, 0.8 g)
3. Oligo R3 resins (PerSeptive Biosystems, 1-1339-03, 6.3 g)
4. 2% (v/v) formic acid in 70% (v/v) acetonitrile (ACN)
5. 0.1% (v/v) trifluoroacetic acid in 70% (v/v) ACN
6. 1-mL syringe
7. Matrix: -cyano-4-hydroxycinnamic acid (CHCA)
8. Opti-TOFTM 384-well insert (123 × 81 mm, 1016491, Applied Biosystems)
2.10. MALDI-TOF and Peptide Mass Fingerprinting
1. MALDI-TOF and MALDI-TOF/TOF: Voyager DE-Pro and 4800 MALDI
TOF/TOFTM Analyzer (Applied Biosystems) equipped with a 355-nm Nd:YAG
laser. The pressure in the TOF analyzer is approximately 7.6e-07 Torr.
3. Methods
3.1. Human Plasma Sample Preparation
The following protocol is conducted according to the HUPO reference
sample collection protocol (13).
1. Each sample pool consisted of 400 mL blood from one healthy, fasting male and
one healthy, fasting postmenopausal female, and was collected into 10-mL tubes
by two venipunctures, 20 tubes per veni-puncture (see Note 1).
2. Equal numbers of tubes and aliquots were generated with appropriate concentra-
tions of K2-EDTA, lithium heparin, or sodium citrate for plasma or were permitted
to clot at room temperature for 30 min to yield serum (with micronized silica as
the clot activator) (see Note 2).
3. The specimens were centrifuged for 10–15 min under refrigerated conditions at
2–6°C.
4. The resultant serum and plasma from 10 spun tubes of the same type from each
donor were pooled into one secondary 50-mL conical bottom BDTM Falcon tube
for each tube type.
5. The secondary tube was centrifuged at 2400×gfor 15 min to remove residual
cellular material from serum and to prepare platelet-poor plasma from the EDTA,
heparin, and citrate secondary tubes.
6. Equal volumes of either serum or plasma were pooled from each secondary tube
into media bottles (see Note 3).
7. Serum/plasma was mixed gently and kept on ice while distributed as 20-μL
aliquots into cryovials and was then frozen and stored at –70°C.
Protein Profiling by Two-Dimensional Electrophoresis 63
3.2. Depletion of High-abundance Proteins with an Immunoaffinity
Column
For efficient depletion of high-abundance proteins prior to their molecular
analysis, many reports have indicated that it is convenient to use commercially
available immunoaffinity columns, such as the MARS (Agilent) (2,3) or the
prepacked 2-mL SepproTM MIXED12 affinity LC column (GenWay Biotech.)
(14), coupled with an HPLC system. For depletion of the six most abundant
proteins (i.e., albumin, transferrin, IgG, IgA, haptoglobin, and anti-trypsin) in
either serum or plasma, we introduced MARS, which has been used successfully
with a wide variety of sample types, including cerebrospinal fluid (CSF) and
follicular fluid (2,3) (see Fig. 1 ).
1. Dilute human serum or plasma fivefold with Buffer A (for example: 20 μL
human plasma with 80 μL Buffer A) containing the protease inhibitor stock
solution (40 μL per 1 mL plasma) (see Note 4) (adopted from the manufacturer’s
instructions).
2. Remove the particulates with a 0.22-μm spin filter for 1 min at 16,000×g.
3. Inject 75-100 μL of the diluted serum or plasma at a flow rate of 0.5 mL/min.
Fig. 1. The 2DE images of total human plasma proteins that were depleted of the
major six abundant proteins through MARS. Proteins were isoelectrically focused with
pH 3–10 NL IPG strips in the first dimension and then resolved by 9–16% SDS-
PAGE in the second dimension. (A) Whole plasma. (B) Flow through from MARS.
Approximately 800 protein spots are displayed by 2DE and identified by MALDI-TOF
mass spectrometry. The names of the major proteins of each gel are marked on the
image (5) (from (4)with permission)
64 Cho et al.
4. Collect the flow-through fractions that appear between 1.5 and 4.5 min and store
them at –20°C if they were not to be analyzed immediately.
5. Elute bound proteins from the column with Buffer B (elution buffer) at a flow
rate of 1 mL/min for 3.5 min.
6. Regenerate the column by equilibrating with Buffer A for an additional 7.4 min
at a flow rate of 1 mL/min.
3.3. TCA/Acetone Precipitation
During 2DE, interfering compounds, such as proteolytic enzymes, salts,
lipids, nucleic acids, and any residual high-abundance proteins present after
depletion, must be removed or inactivated. In the case of plasma samples, the
two most important parameters are salt and proteolysis. TCA/acetone precip-
itation is the most useful method for desalting the whole plasma and the
flow-through fractions of MARS.
1. Add 50% (w/v) trichloroacetic acid (TCA, Sigma, T9159) to reach a final TCA
concentration of 5-8%. Mix gently by inverting the tube 5 to 6 times and incubate
on ice for 2 h.
2. Centrifuge the sample at 14,000×gfor 15 min and discard the supernatant.
3. Add 200 μL cold acetone and resuspend the protein pellet with a pipette.
4. Incubate on ice for 15 min and centrifuge the sample at 14,000×gfor 20 min,
discard the acetone, and dry the pellet in air (see Note 5).
5. Dissolve the pellet in the sample buffer for 2DE and quantify the protein concen-
tration by the Bradford protein assay.
3.4. Rehydration of the IPG Gel Strip
For analytical purposes, typically 0.3–1.0 mg protein can be loaded onto an
18-cm-long IPG with a wide pH range (e.g., pH 3-10), or 0.5–2.0 mg on an
IPG with a narrow pH range (e.g., pH 5.5–6.7). A narrow-range IPG usually
produces a higher resolution when separate proteins are analyzed by sequential
IEF systems: first, fractionate the proteins over several pI ranges in solution
with ZOOM®disks or FFE (see Subheadings 3.6 and 3.7) and then perform
IEF with IPG strips [one pH unit range strips are also available (e.g., pH 3.0–
4.0 or pH 3.5–4.5 up to pH 6.7)]. Certain proteins appear to be trapped in the
disk membrane; partitions and sample loss should be considered.
1. Dilute 1.0 mg protein with the sample buffer to a final volume of 400 μL for
18-cm-long IPG strips (see Note 6).
2. Transfer the entire protein-containing sample buffer into the re-swelling tray.
3. Peel off the protective cover from the IPG strip and slowly slide the IPG strip (gel
side down) onto the sample solution. Avoid trapping air bubbles and distribute
the sample solution evenly under the strips.
Protein Profiling by Two-Dimensional Electrophoresis 65
4. Overlay the strip with mineral oil and leave for 12-16 h at room temperature (see
Note 7 for cup loading)
3.5. IEF with IPG Strip
1. Remove the rehydrated IPG strips that are carrying the protein samples and place
them (gel side up) on the strip tray.
2. Place the 2.5-cm filter papers, wetted with distilled water, on both sides of the
strips at both cathodic and anodic ends. Place the strip tray on the IEF unit.
3. Cover the strips entirely with mineral oil.
4. Program the instrument (e.g., Multiphor II): Increase the voltage from 100 to
3500 V to reach 80,000 total voltage hours (Vh) (e.g., sequentially, 300 Vh at
100 V, 600 Vh at 300 V, 600 Vh at 600 V, 1000 Vh at 1000 V, and 2000 Vh at
2000 V, for a total of 80,000 Vh at 3500 V) (see Notes 8 and 9).
5. During IEF, the temperature is set to 20°C with a water circulator.
3.6. Microscale Solution IEF: ZOOM®
To reduce typical artifacts that may occur when using narrow-range IPG
strips (e.g., streaking, distortion, and loss of protein spots), one may use
MicroSol-IEF (e.g., ZOOM®, Invitrogen) prior to running 2D gels (3) (see
Fig. 2). MicroSol-IEF is a preparative solution-phase IEF apparatus that
is dissected by a defined pH membrane disc (15,16). Using MicroSol-IEF,
2.5-3.0 mg plasma proteins can be loaded and efficiently fractionated into five
separate chambers by their pI values.
1. Add 2 μL of 99% dimethylamine (DMA) to the 400-μL sample (see Subheading
3.4, Step 2) for alkylation and incubate the sample on a rotary shaker for 30 min
at room temperature (adopted from the manufacturer’s instructions).
2. Add 4 μL of 2 M DTT to quench any excess DMA. Centrifuge at 16,000×gfor
20 min at 4°C.
3. Preparation of protein samples: Dilute 3 mg protein to a 3250-μL volume with
sample buffer. The amount of diluted sample per chamber in the ZOOM®IEF
Fractionator is 650 μL.
4. Assemble the ZOOM®IEF Fractionator according to the manufacturer’s instruc-
tions. Six disks (pHs 3.0, 4.6, 5.4, 6.2, 7.0, and 10.0) are used to create five
fractions that have a range of pH 3.0–10.0.
5. Add each buffer (anode or cathode) to the corresponding blank chamber.
6. Remove the sample chamber cap and add 650 μL of protein sample (step 3)to
each chamber.
7. Fractionation can be carried out under the following conditions: 100 V for 20 min,
200 V for 80 min, and 600 V for 80 min (see Note 10). The starting current is
approximately 0.6 mA, which increases to approximately 1.2 mA at the beginning
of the 200-V step, and the ending current is approximately 0.2 mA.
8. Load the electro-focused samples to the narrow pH IPG strips for 2DE.
66 Cho et al.
Fig. 2. Narrow pH range 2DE images of plasma proteins after depletion of the major
six abundant proteins through MARS. After microscale solution IEF (ZOOM®), the pH
5.5–6.2 fraction was separated on pH 5.5–6.7 IPG strips by second isoelectric focusing
and then resolved on a 9–16% gel. (A) Whole 2DE image of pH 3–10 NL and pH
5.5–6.7. (B) One spot on the pH 3–10 NL gel can be separated into two or more spots
in the narrow pH range 2DE. (C) Many hidden spots on the pH 3–10 NL gel appear
in the narrow pH range 2DE of normal and HCC plasma.
Protein Profiling by Two-Dimensional Electrophoresis 67
3.7. Fractionation of the Plasma Samples by Free Flow Electrophoresis
To identify and isolate biomarker candidates from the plasma of diseased
patients with HCC using 2DE, a higher resolution is critical, and the analysis
can be done by performing narrow pH range IEF. However, for narrow pH range
IEF, higher amounts of proteins (e.g., 10-fold or higher) should be loaded onto
the IPG strip since the proteins present in other pH ranges will be discarded.
Nevertheless, prefractionation or depletion is required prior to running both
IEF and 2D gel. FFE is useful for prefractionation of plasma samples since it
gives rise to a specific fraction of interest (e.g., pI, or density). For example, if
one knows the pI of certain proteins, free fractionation by FFE can be useful
for prefractionation of complex plasma. We describe here one of the several
procedures for prefractionation of plasma samples using FFE.
1. Dissolve the TCA-precipitated, flow-through fractions of MARS (∼2.0 mg) into
the 500-μL separation medium 3 (see below) (adopted from the manufacturer’s
instructions).
2. Add traces of red acidic dye 2-(4-sulfophenylazo)-1,8-dihydroxy-3,6-
naphthalenedisulfonic acid (SPADNS, Aldrich) to ease the optical control of the
migration of sample within the separation chamber.
3. FFE is carried out at 10°C using the following media (solutions marked
at each inlet are applied): Anodic stabilization medium (Inlet I1), separation
medium 1 (Inlet I2), separation medium 2 (Inlet I3−−5), separation medium 3
(Inlet I6), cathodic stabilization medium (Inlet I7), and counter-flow medium
(Inlet I8).
4. To both the anode and the cathode, anodic circuit electrolyte and cathodic circuit
electrolyte are applied, respectively.
5. Assemble the ProTeamTM FFE instrument (Tecan). Use a 0.4-mm spacer for the
separation chamber and a flow rate of approximately 60 mL/h (Inlet I1−7) and a
voltage of 1500 V, which results in a current of 20–24 mA.
6. Perfuse the separation chamber with the sample using the cathodal inlet at approx-
imately 0.7 mL/h (4,17). Residence time in the separation chamber is approxi-
mately 33 min.
7. Collect each fraction into polypropylene, 96 deep-well plates, numbered 1 (anode)
through 44 (cathode) (4).
8. Remove glycerol and HPMC by TCA/acetone precipitation and dissolve the
proteins with sample buffer.
9. Load the electro-focused samples with narrow pH to the IPG strips for 2DE.
3.8. Preparation of 2D Gels
1. Cast the glass plates (separated by two 1.5-mm spacers positioned along the sides)
and thin plastic sheets in the multi-casting chamber (20).
2. Prepare gel solution for making 10 gels (20 × 20 cm, 1.5-mm spacer, 9–16%
gradient): heavy solution (66.7 mL of 5× Tris-HCl buffer, 75 mL of a 40%
68 Cho et al.
acrylamide stock solution, 0.7 mL of 10% ammonium persulfate (APS), 70 μL
TEMED, and 191.7 mL of 50% glycerol), light solution (66.7 mL of 5× Tris-HCl
buffer, 141.7 mL of a 40% acrylamide stock solution, 0.7 mL of 10% APS, 70 μL
TEMED, and 125 mL distilled water).
3. Assemble the gradient maker and peristaltic pump. Pour the light gel solution into
the mixing chamber (close to the casting chamber) and the heavy gel solution
into the reservoir chamber of the gradient maker. Operate the magnetic stirrer in
the mixing chamber. Turn on the peristaltic pump until the gel solution reaches
0.5-1.0 cm below the end of the glass plates (∼5 min). Check the flow rate, which
should be between 100-120 mL/min.
4. After the gel solution is poured, overlay the gel solution with distilled water to
exclude air and to ensure a level surface on the top of the gel.
5. Allow polymerization to occur overnight at room temperature.
3.9. Equilibration of the Sample and Running of the Gel
To solubilize the electro-focused proteins and to allow SDS to polymerize,
it is necessary to soak the IPG strips in SDS equilibration buffer. This step
is analogous to boiling the sample in SDS buffer prior to SDS-PAGE. The
reducing agents, dithiothreitol (DTT) and tributylphosphine (TBP), reduce
disulfide bonds to sulfhydryls (cysteine residues). Alkylating agents and iodoac-
etamide (IAA) prevent reoxidation of the free sulfhydryl groups (21).
1. Prior to use, add approximately 158 μL TBP in 1 mL isopropanol to 100 mL
SDS equilibration buffer and sonicate in a bath-type sonicator until the solution
becomes transparent (see Note 11) (termed TBP equilibration buffer).
2. Add 15 mL TBP equilibration buffer to each strip (gel side up) and gently shake
for 25 min (TBP equilibration) (see Note 12) on an orbital shaker.
3. Briefly rinse the IPG strip with 1× gel buffer and load the IPG strips onto the
top of the gel and pour the agarose embedding solution (molten agarose solution
with trace amounts of BPB) (see Note 13).
4. Perform SDS-PAGE (40 mA/gel) until the BPB dye reaches the bottom of the
gel. Keep the temperature at 10°C. The total run time for 20 × 20 cm gels is
approximately 6 h.
3.10. Coomassie Brilliant Blue G-250 Staining
1. Fix the separated proteins into the gel in a 200-mL fixing solution for 1 h.
2. Decant the fixing solution and stain the gel in Coomassie brilliant blue G-250
overnight.
3. Decant the staining solution.
4. Wash several times (>3 times) in distilled water for more than 4 h.
5. Scan the gel, then wrap the gel in plastic, and store it at 4°C.
Protein Profiling by Two-Dimensional Electrophoresis 69
3.11. 2D Gel Image Analysis
1. Import the gel image (recommended 12–16 bit, tiff format) and convert it into an
ImageMaster file (*.mel).
2. Detect the protein spots and determine the volume and percentage volume of
each spot. The percentage volume is the normalized value that remains relatively
independent of any irrelevant variations between gels, particularly those caused
by varying experimental conditions.
3. Select the differentially displayed protein spots (see Fig. 3).
3.12. Destaining, In-gel Deglycosylation, and In-gel Tryptic Digestion
Most plasma proteins are glycosylated, including clotting factors, lipopro-
teins, and antibodies (22,23). These carbohydrate-containing proteins play
major roles in the normal biological functions in plasma. Since glycopeptides
are not easily completely ionized during MS analysis, which may lead to inade-
quate spectral data and low detection sensitivity due to the attached glycans, a
strategy for the removal of glycans is necessary for protein identification.
1. Pick (or excise) the protein spot with an end-cut yellow tip and transfer the gel
piece into a 1.5-mL Eppendorf tube.
2. Wash the gel piece with 100 μL distilled water.
3. Add 50 μL of 50 mM NH4HCO3 (pH 7.8) and ACN (6:4), and shake for 10 min.
4. Repeat step 3 until the Coomassie blue G250 dye disappears (2 to 5 times).
5. Decant the supernatant and dry the gel piece in a Speed Vac for 10 min (see
Note 14).
6. Add 5 μL trypsin (12.5 ng/μL in 50 mM NH4HCO3) and leave the gel piece on
ice for 45 min.
7. Add 10 μL of 50 mM NH4HCO3 to the gel slice.
8. Incubate the gel piece at 37°C for 12 h.
3.13. Desalting of Peptides and MALDI Plating
1. Resin packing: Twist the column body (GELoader tip, Eppendorf) near the end of
the tip and push the resin solution [Poros R2:Oligo R3 (2:1) in 70% (v/v) ACN,
occasionally in a more efficient ratio of 1:1] with a 1-mL syringe. A packed resin
length of 2-3 mm is suitable (18,19).
2. Equilibration of the column: Add 20 μL of 2% (v/v) formic acid and push the
solution through the column with the 1-mL syringe.
3. Peptide binding: Add the peptide solution (supernatant of step 9 in Subheading
3.12, approximately 10-12 μL) and push this solution through the column with
the syringe.
4. Washing: Add 20 μL of 2% (v/v) formic acid and push this solution through the
column with the syringe.
70 Cho et al.
Fig. 3. Detection of PTMs on the 2DE of plasma proteins. (A) 2DE images of
plasma proteins that were depleted of the major six abundant proteins through MARS,
untreated (left) and alkaline phosphatase (AP)-treated (AP) (right). (B) One of the
differentially displayed proteins after treatment with AP. (C) Data-dependant neutral
loss scan spectrum of sequence KEPCVESLVSpQYFQTVTDYGKD corresponding to
the phosphorylated apolipoprotein A-II precursor.
Protein Profiling by Two-Dimensional Electrophoresis 71
5. MALDI spotting: Add 1 μL matrix solution [10 mg/mL CHCA in 70% (v/v) can
and 2% (v/v) formic acid] and directly spot the eluted peptides and matrix mixture
onto the MALDI plate (Opti-TOFTM 384-well Insert, Applied Biosystems).
6. Reuse the column: Add 20 μL of 100% ACN and push this solution through the
column with the syringe and repeat step 2 for equilibration of the column.
3.14. MALDI-TOF and Peptide Mass Fingerprinting
1. Analyze the peptide mass fingerprinting (PMF) with the Voyager DE-PRO or
4800 MALDI-TOF/TOF mass spectrometer (Applied Biosystems).
2. Obtain the mass spectra in reflectron/delayed extraction mode with an accelerating
voltage of 20 kV and sum data from either 500 laser pulses (4800 MALDI-
TOF/TOF) or 100 laser pulses (Voyager DE-PRO).
3. Calibrate the spectrum with tryptic auto-digested peaks (m/z 842.5090 and
2211.1046) and obtain monoisotopic peptide masses with Data Explorer 3.5
(PerSeptive Biosystems).
4. Search the Swiss-Prot and NCBInr databases with the Matrix Science search
engine (http://www.matrixscience.com).
3.15. Profiling of PTMs on Selected Spots
Although shotgun proteomics that utilize various labeling techniques (e.g.,
SILAC and iTRAQ) are useful for protein identification in a high-throughput
manner, it has many limitations for PTM analysis. However, 2D gels usually
display proteins with PTMs or isoforms of certain proteins on a single gel
as spots in different positions, which can lead to further identification for
their molecular characteristics with the aid of high resolution LC-MS/MS. For
example, in a typical 2D gel of plasma, the phosphorylated forms of certain
protein can be easily detected in a ladder form that results from different
pIs. Figure 3 shows the localization of the exact site of phosphorylated
apolipoprotein A-II precursor. As seen in the figure, there is clear difference
between spots that are alkaline phosphatase (AP)-treated and those that are
untreated in the 2D gel where the treated group has been shifted to a more
basic position. The phosphorylation site of these proteins can be determined
using multidimensional MS (MS2and MS3). Here, we describe the procedure
for identification of phosphorylated proteins by 2DE coupled to MS.
1. Desalting is processed for the MARS-treated (high-abundance proteins depleted)
plasma sample using Amicon Ultra-15 (Molecular Weight Cut Off; 5 kDa,
Millipore).
2. Dephosphorylation is carried out overnight at 37°C in a solution of 0.4%
ammonium carbonate buffer (pH 8.5) with 24 ng/μL calf intestine AP in 0.4%
NH4HCO3.
3. The reaction is stopped by freeze drying for further analysis.
72 Cho et al.
4. Execute 2DE, picking, extraction, and desalting of peptides under the same
conditions (see Subheadings 3.8-3.13).
5. Dissolve the extracted and desalted peptides in 10 μL of LC-MS/MS
solution [0.4% (v/v) acetic acid and 0.005% (v/v) heptafluorobutyric acid
(HFBA)].
6. Nano LC-MS/MS analysis is then performed on an Agilent Nano HPLC system
(Agilent) and LTQ mass spectrometer (Thermo Electron, San Jose, CA).
7. The capillary column used for LC-MS/MS analysis (150 mm × 0.075 mm)
was obtained from Proxeon (Odense M, Denmark), and the slurry was packed
in-house with a 5-μm, 100-Å pore size Magic C18 stationary phase (Michrom
Bioresources, Auburn, CA).
8. The mobile phase A for LC separation was 0.4% acetic acid and 0.005% HFBA
in deionized water (Cascada™, Pall, USA), and the mobile phase B was 0.4%
acetic acid and 0.005% HFBA in ACN.
9. The sample obtained from the Oasis HLB (Waters, USA) desalting step and
Nanosep (Pall, USA) filtering was loaded onto the LC column.
10. The chromatography gradient was designed to provide a linear increase from
5% B to 35% B over 50 min and from 40% B to 60% B over 20 min and from
60% B to 80% B over 5 min. The flow rate was maintained at 300 nL/min.
11. The mass spectra were acquired using data-dependent acquisition with a full mass
scan (400-1800 m/z) followed by MS/MS scans. Each MS/MS scan acquired
was an average of three microscans on LTQ.
12. The temperature of the ion transfer tube was controlled at 200°C, and the spray
was 2.0–3.0 kV. The normalized collision energy was set at 35% for MS2.
13. To determine the exact position of the phosphorylation site, the automated
neutral loss MS3 scan was employed, which relies on the observed behavior
of phosphopeptides subjected to MS/MS analysis in an ion trap. If the MS/MS
scan produces a fragment phosphate group (98 with charge state 1+, 49 with
charge state 2+, and 32.6 with charge state 3+), an MS3 scan of the product ion
is initiated (see Note 15).
4. Notes
1. Donors were tested and determined negative for HIV-1 and HIV-2 antibodies,
HIV-1 antigen (HIV-1), Hepatitis B surface antigen (HBsAg), Hepatitis B core
antigen (anti-HBc), Hepatitis C virus (anti-HCV), HTLV-I/II antibody (anti-
HTLV-I/II), and syphilis.
2. No protease inhibitor cocktails were used. This procedure required2hat2-6°C.
3. Approximately 10% of the sample was left at the bottom of the secondary tube
to ensure that no cellular material was collected.
4. If excess of protease inhibitors are used, the resolving power of protein spots in
the 2D gel will be decreased, and the border of the spots will be unclear.
5. If protein pellets are dried completely in the Speed Vac, they will be not re-
dissolved in sample buffer. Pellets should be air dried for 15–30 min.
Protein Profiling by Two-Dimensional Electrophoresis 73
6. To ensure complete dissolution of the sample buffer, it is usually recommended
to warm the sample buffer at room temperature. The sample buffer that includes
proteins should not be heated to avoid carbamylation of proteins by isocyanate,
which may lead to charge heterogeneities that are formed from the decomposition
of urea.
7. Cup loading: Rehydrate the IPG gel strip with 350 μL sample buffer (proteins
are not included), and load the 100-μL protein sample in sample buffer in the
sample cup. High salt concentrations are better tolerated by cup loading.
8. Apply low voltages (100 V) at the beginning of the run for 3–5 h. Replace the
filter paper (for desalting purposes) at the end of the run.
9. After 1D (first dimension) is run, IPG strips that were not immediately used for
2D (second dimension) run can be preserved at –80°C for several months.
10. If electrical current passes through the system, BPB dye starts to migrate toward
the anode reservoir, which eventually results in a change in the color of the
anode buffer (to yellow).
11. Concentrated TBP reacts violently with organic matter. All procedures for
preparing TBP stock solutions should be done in a fume hood. Store the TBP
stock solution in the dark at 4°C. Do not store it longer than 2 weeks.
12. DTT/IAA equilibration procedure: For reduction and alkylation of proteins,
the DTT/IAA equilibration procedure is also useful to replace the use of TBP
equilibration procedure. Divide the SDS equilibration buffer into two 50-mL
aliquots. Add 1 g DTT to the first aliquot and 1.25 g IAA to the second aliquot.
Add 10 mL of the DTT equilibration buffer to each strip and place on a shaker
for 10 min. Decant the DTT equilibration buffer and shake with 10 mL of the
IAA equilibration buffer for another 10 min.
13. To prepare the agarose embedding solution, dissolve1gofagarose in 100 mL
of small gel buffer and melt in a microwave on medium power. For complete
melting of the agarose solution, heat the agarose solution in short intervals with
occasional swirling to mix the solution.
14. In-gel deglycosylation: After destaining, one may remove the glycan groups
of glycoproteins by trypsin digestion for obtaining peptides of highest purity.
Rehydrate gel spots (see Subheading 3.12, step 5) with 10 μL of PNGase F
stock solution (10 μU) and incubate for3hat37°C. Decant the supernatant
including the glycans. Wash the gel piece with 50 μL 50 mM NH4HCO3 (pH
7.8) and ACN (6:4). Dry the gel piece in a Speed Vac.
15. The SEQUEST software was used to identify the peptide sequences:
DeltaCn ≥0.1 and Rsp ≤4; Xcorr ≥1.9 with charge state 1+, Xcorr ≥2.2 with
charge state 2+, and Xcorr ≥3.75 with charge state 3+ were used as cutoffs for
peptide identification.
Acknowledgments
This study was supported by a grant from the Korean Health 21 R&D project,
Ministry of Health & Welfare, Republic of Korea (A030003 to YKP).
74 Cho et al.
References
1. Putnam, F. W. (ed) (1987) The Plasma Proteins, Academic Press, New York.
2. Anderson, N. L., and Anderson, N. G. (2002) The human plasma proteome: history,
character, and diagnostic prospects. Mol. Cell. Proteomics 1, 845–867.
3. Lee, H. J., Lee, E. Y., Kwon, M. S., and Paik, Y. K. (2006) Biomarker discovery
from the plasma proteome using multidimensional fractionation proteomics. Curr.
Opin. Chem. Biol.10, 42–49.
4. Cho, S. Y., Lee, E. Y., Lee, J. S., Kim, H. Y., Park, J. M., Kwon, M. S., Park, Y. K.,
Lee, H. J., Kang, M. J., Kim, J. Y., Yoo, J. S., Park, S. J., Cho, J. W., Kim, H. S., and
Paik, Y. K. (2005) Efficient prefractionation of low-abundance proteins in human
plasma and construction of a two-dimensional map. Proteomics 5, 3386–396.
5. Omenn, G. S., States, D. J., Adamski, M., and Blackwell, T. W. (2005). Overview
of the HUPO Plasma Proteome Project: results from the pilot phase with 35
collaborating laboratories and multiple analytical groups, generating a core dataset
of 3020 proteins and a publicly-navailable database. Proteomics 5, 3226–3245.
6. States, D. J., Omenn, G. S., Blackwell, T. W., Fermin, D., Eng, J., Speicher, D. W.,
and Hanash, S. M. (2006) Challenges in deriving high-confidence protein identifi-
cations from data gathered by a HUPO plasma proteome collaborative study. Nat.
Biotechnol.24, 333–338.
7. Yang, Z., Hancock, W. S., Chew, T. R., and Bonilla, L. (2005) A study of
glycoproteins in human serum and plasma reference standards (HUPO) using
multilectin affinity chromatography coupled with RPLC-MS/MS. Proteomics 5,
3353–3366.
8. Wang, Y., Wu, S. L., and Hancock, W. S. (2006) Approaches to the study of
N-linked glycoproteins in human plasma using lectin affinity chromatography
and nano-HPLC coupled to electrospray linear ion trap-Fourier transform mass
spectrometry. Glycobiology 16, 514–523.
9. Gorg, A., Boguth, G., Kopf, A., Reil, G., Parlar, H., and Weiss, W. (2002) Sample
prefractionation with Sephadex isoelectric focusing prior to narrow pH range two-
dimensional gels. Proteomics 2, 1652–1657.
10. Wu, T. L. (2006) Two-dimensional difference gel electrophoresis. Methods Mol.
Biol.328, 71–95.
11. Park, K. S., Kim, H., Kim, N. G., Cho, S. Y., Choi, K. H., Seong, J. K., and
Paik, Y. K. (2002) Proteomic analysis and molecular characterization of tissue
ferritin light chain in hepatocellular carcinoma. Hepatology 6, 1459–1466.
12. Park, K. S., Cho, S. Y., Kim, H., and Paik, Y. K. (2002) Proteomic alterations of the
variants of human aldehyde dehydrogenase isozymes correlate with hepatocellular
carcinoma. Int. J. Cancer 2, 261–265.
13. Rai, A. J., Glefand, C. A., Haywood, B. C., Warunek, D. J., Yi, J., Schuchard, M. D.,
Mehigh, R. J., Cockrill, S. L., Scott, G. B., Tammen, H., Schulz-Knappe, P.,
Speicher, D. W., Vitzthum, F., Haab, B. B., Siest, G., and Chan, D. W.
(2005) HUPO plasma proteome project specimen collection and handling: towards
the standardization of parameters for plasma proteome samples. Proteomics 5,
3262–3277.
Protein Profiling by Two-Dimensional Electrophoresis 75
14. Huang, L., Harvie, G., Feitelson, J. S., Gramatikoff, K., Herold, D. A., Allen, D. L.,
Amunngama, R., Hagler, R. A., Pisano, M. R., Zhang, W. W., and Fang, X. (2005)
Immunoaffinity separation of plasma proteins by IgY microbeads: meeting the
needs of proteomic sample preparation and analysis. Proteomics 5, 3314–3328.
15. Herbert, B. and Righetti, P. G. (2000) A turning point in proteome analysis: sample
prefractionation via multicompartment electrolyzers with isoelectric membranes.
Electrophoresis 21, 3639–3648.
16. Miklos, G. L. and Maleszka, R. (2001) Integrating molecular medicine with
functional proteomics: realities and expectations. Proteomics 1, 30–41.
17. Weber, G., Islinger, M., Weber, P., Eckerskorn, C., and Volkl, A. (2004)
Efficient separation and analysis of peroxisomal membrane proteins using free-flow
isoelectric focusing. Electrophoresis 25, 1735–1747.
18. Choi, B. K., Cho, Y. M., Bae, S. H., Zoubaulis, C. C., and Paik, Y. K. (2003)
Single-step perfusion chromatography with a throughput potential for enhanced
peptide detection by matrix-assisted laser desorption/ionization-mass spectrometry.
Proteomics 3, 1955–1961.
19. Gobom, J., Nordhoff, E., Mirgorodskaya, E., Ekman, R., and Roepstorff, P. (1999)
A sample purification and preparation technique based on nano-scale RP-columns
for the sensitive analysis of complex peptide mixtures by MALDI-MS. J. Mass
Spectrom.24, 105–116.
20. Walsh, B. J., and Herbert, B. R. (1999) Casting and running vertical slap-gel
electrophoresis for 2D-PAGE. Methods Mol. Biol.112, 245–253.
21. Newhall, W. J. and Jones, R. B. (1983) Disulfide-linked oligomers of the major
outer membrane protein of chlamydiae. J. Bacteriol.154, 998–1001.
22. Kaufman, R. J. (1998) Post-translational modifications required for coagulation
factor secretion and function. Thromb. Haemost. 79, 1068–1079.
23. Tabas, I. (1999) Nonoxidative modifications of lipoproteins in atherogenesis. Annu.
Rev. Nutr.19, 123–139.
II
Clinical Proteomics by 2DE and Direct
MALDI/SELDI MS Profiling
5
Analysis of Laser Capture Microdissected Cells
by 2-Dimensional Gel Electrophoresis
Daohai Zhang and Evelyn Siew-Chuan Koay
Summary
Laser capture microdissection (LCM) is a powerful tool for procuring near-pure
populations of targeted cell types from specific microscopic regions of tissue sections,
by overcoming problems due to tissue heterogeneity and minimizing intermixture and
contamination by other cell types. The combination of LCM with various proteomic
technologies has enabled high-throughput molecular analysis of human tumors, and
provided critical tools in the search for novel disease markers and therapeutic targets. As
an example, we describe the application of LCM in dissecting the tumor cells in breast
cancer for macromolecular extraction and subsequent protein separation by 2-dimensional
gel electrophoresis (2-D GE). The protocols and the key issues involved in preparing
ethanol-fixed paraffin-embedded tissue blocks and microscopic sections, microdissecting
the cells of interest using the PixCell II LCM system, extracting and separating the cellular
proteins by 2-D GE, and preparing selective proteins for peptide mass analysis by mass
spectrometry, are discussed. The aim is to provide a practical guide in performing high-
throughput microdissection of target cells and gel-based proteomics, which can be adapted
to research in cancer formation and growth.
Key Words: laser capture microdissection; 2-dimensional gel electrophoresis; breast
cancer; proteomics; silver staining.
1. Introduction
Cellular proteins (collectively known as “proteomes”) are less susceptible
than the transcriptome to experimental artifacts arising from the rigors of tissue
collection and processing, and advances in global protein expression analysis
From: Methods in Molecular Biology, vol. 428: Clinical Proteomics: Methods and Protocols
Edited by: A. Vlahou © Humana Press, Totowa, NJ
77
78 Zhang and Koay
(expression proteomics) have been used in mapping cellular pathways, identi-
fying the molecular alterations associated with disease onset and progression
and searching for potential tumor markers or drug targets in human disease,
especially in cancer. However, to obtain cell-specific protein profiles, homoge-
neous or near-pure populations of the cells of interest, free from contamination
by adjacent cell types, are prerequisites. Laser capture microdissection (LCM)
was developed to enable the procurement of near-pure populations of the target
cells with a greater speed and precision than is possible with manual dissection
methods. LCM permits selective transfer of specific cell types, under direct
microscopic visualization, from complex tissues onto a polymer film that is
activated by laser pulses, whilst retaining their morphology. The homogeneity
of encapsulated cells can be verified microscopically. With these inherent
advantages, LCM has become a valuable research tool and has been applied to
cellular and molecular studies of various cancers, including breast (1,2), colon
(3), and liver (4) cancers. It is equally efficacious in procuring cell populations
from both frozen tissues (3,4) and ethanol-fixed, paraffin-embedded tissues
(1,5).
Protein profiles of the LCM-dissected cells can be obtained by two-
dimensional fluorescence difference gel electrophoresis (2-D DIGE) (6),
16O/18 O isotopic labeling (7), differential iodine radioisotope detection (2),
isotope-coded affinity tag (iCAT) coupled with two-dimensional tandem mass
spectrometry (2-D LCMS/MS) (8), and mass spectrometry compatible silver
staining (1,9). Protein samples from LCM-dissected cells can also be applied
to reverse-protein arrays to analyze the key cellular signaling pathways and
metabolic networks (10,11). In this chapter, the in-house protocols used in
the authors’ laboratory for procuring near-pure populations of breast tumor
cells from clinical samples, and for the extraction, isolation, and analysis of
their protein profiles, are described. These include: (1) preparation of ethanol-
fixed paraffin-embedded tissue blocks; (2) microdissection using the Pix II
LCM System and cellular protein extraction; (3) protein separation by 2-D gel
electrophoresis (2-D GE), silver staining, and gel image analysis; and (4) prepa-
ration of targeted proteins of interest for peptide mass analysis by tandem mass
spectrometry and identification of proteins of interest via database search.
2. Materials
2.1. Histology—Tissue Block and Tissue Section Preparation
1. 70% (v/v), 80% (v/v), 95% (v/v), 100% ethanol
2. Deionized or Milli-Q water (Millipore, Bedford, MA, USA)
3. Hematoxylin solution, Mayer’s (Sigma, St. Louis, MO, USA)
4. Eosin Y solution (Sigma)
Combining LCM with 2-D Gel Electrophoresis 79
5. Complete, mini protease inhibitor cocktail tablets (Roche Applied Science,
Pleasanton, CA, USA)
6. Disposable microtome blades (Feather Safety Razor Co., Ltd., Osaka, Japan)
7. Uncharged microscopic glass slides (Paul Marienfeld GmbH & Co, KG, Lauda-
Koenigshofen, Germany)
8. Sakura Tissue-Tek®V.I.P.TM 5 Jr tissue processor (Sakura Finetek, Inc. Japan
Co., Ltd, Tokyo)
9. Paraffin wax—Paraplast®tissue embedding medium; melting point 56-58°C,
store at room temperature (RT) (Structure Probe, Inc., West Chester, PA, USA)
10. Xylenes, Reagent Grade (Sigma)
11. Embedding molds—super metal base molds, 66mm × 54mm×15mm (Surgipath
Medical Industries, Richmond, IL, USA)
2.2. Laser Capture Microdissection and Protein Sample Preparation
1. PixCell II LCM system (Arcturus Engineering, Mountain View, CA, USA)
2. CapSure transparent plastic caps (Arcturus Engineering)
3. Lysis buffer: 7 M urea, 2 M thiourea, 4% (w/v) CHAPS, 1% Nonidet P (NP)-40,
0.5% (v/v) Triton X-100, 50 mM dithiothreitol (DTT), 40 mM Tris-HCl, pH 7.5,
2 mM tributyl phosphine (TBP), and 1% (v/v) IPG buffer (pH 3–10). Store at RT.
4. PlusOne 2-D Clean-up Kit (GE Healthcare, San Francisco, CA, USA)
5. Immobilized pH gradient (IPG) buffer (pH 3–10) (GE Healthcare)
6. PlusOne 2-D Quantitation Kit (GE Healthcare)
2.3. Isoelectric Focusing (IEF) and Sodium Dodecyl
Sulfate-Polyacrylamide Gel Electrophoresis (SDS-PAGE)
1. EttanTM IPGphorTM IEF electrophoresis unit (GE Healthcare)
2. Ceramic strip holders and EttanTM IPGphorTM Strip Holder Cleaning Solution
(GE Healthcare)
3. ImmobilineTM IPG DryStrips (18 cm, pH 3–10, NL) (GE Healthcare)
4. DryStrip Cover Fluid (GE Healthcare)
5. Sample rehydration buffer: 7 M urea, 2 M thiourea, 4% (w/v) CHAPS, 1%
(w/v) NP-40, 1% (v/v) IPG buffer, 50 mM DTT. DTT was added freshly to the
rehydration buffer prior to use. Store at RT.
6. Equilibration buffer A (prepare 10 ml for each strip): 6 M urea, 30% glycerol,
2% SDS, 1% DTT, 50 mM Tris-HCl, pH 8.8. DTT is added to the stock solution
before use.
7. Equilibration buffer B (prepare 10 ml for each use strip): 6 M urea, 30% glycerol,
2% SDS, 250 mg (2.5%, w/v) iodoacetamide (IAA), 50 mM Tris-HCl, pH 8.8.
IAA is added to the stock solution before use.
8. 10% SDS-acrylamide gel: 33 ml acrylamide/bis (30% T, 5% C) (Bio-Rad
Laboratories, Hercules, CA, USA), 25 ml Tris (1.5 M, pH 8.8), 1 ml 10% (w/v)
SDS, 0.5 ml 10% (w/v) ammonium persulfate (freshly prepared on the day of
use), 35 μl TEMED (Bio-Rad). Make up to 100 ml with Milli-Q water.
80 Zhang and Koay
9. Water-saturated isobutanol: Shake equal volumes of Milli-Q water and isobu-
tanol in a glass bottle and allow the mixture to separate. Transfer the top layer
to a new bottle and store at RT.
10. Agarose sealing solution: Dissolve 0.5% low-melting-point agarose and 0.1%
(w/v) bromophenol blue in 1× SDS-PAGE running buffer. Store at RT.
11. SDS-PAGE running buffer: 25 mM Tris, 198 mM glycine, 0.2% (w/v) SDS,
pH 8.3
12. PROTEANTM II xi Cell system (Bio-Rad)
2.4. Silver Staining (see Note 1)
1. Fix solution: 5% acetic acid and 50% ethanol per 100 ml
2. Sensitivity-enhancing solution: 30% (v/v) ethanol, 6.8% (w/v) sodium acetate,
100 μl of 2% (w/v) sodium thiosulphate per 100 ml
3. Silver staining solution: 0.25% (w/v) silver nitrate
4. Development solution: 2.5% (w/v) anhydrous potassium carbonate, 20 μl of 2%
(w/v) sodium thiosulphate per 100 ml, 40 μl of 37% formaldehyde per 100 ml.
5. Stop solution: 4% (w/v) Tris and 2% (v/v) acetic acid per 100 ml
6. Gel store (soak) solution: 1% (w/v) sodium acetate and 10% (v/v) methanol per
100 ml
2.5. Gel Image Analysis
1. Personal Densitometer SI (Molecular Dynamics, Sunnyvale, CA, USA)
2. ImageMaster 2D Elite (Platinum) software (GE Healthcare)
2.6. In-gel Trypsin Digestion and Preparation for MS Analysis
1. Destaining solution: 30 mM potassium ferricyanide and 100 mM sodium
thiosulfate (1:1)
2. 25 mM sodium bicarbonate
3. Dehydrating solution: 50 mM sodium bicarbonate and 50% (v/v) methanol per
100 ml
4. SpeedVac centrifuge (TeleChem International, Inc., Sunnyvale, CA, USA)
5. Digestion solution: 40 ng/μl trypsin sequencing grade (Promega, Madison, WI,
USA) in 20 mM ammonium bicarbonate solution
6. Extraction solution (for hydrophobic peptides): 5% (v/v) trifluoracetic acid
(TFA) and 50% (v/v) acetonitrile (ACN) per 100 ml
7. Peptide reconstitution solution: 0.1% (v/v) TFA
8. ZipTip C18 columns (Millipore)
9. Eluant: 70% (v/v) ACN and 0.1% TFA per 100 ml
10. Stainless steel MALDI-TOF sample target plates (Applied Biosystems,
Framingham, MA, USA)
11. Alpha-cyano-4-hydroxycinnamic acid (-CHCA) matrix, 3 mg/ml (Sigma)
12. Applied Biosystems 4700 MALDI-TOF/TOF mass spectrometer
Combining LCM with 2-D Gel Electrophoresis 81
2.7. Database Search for Protein Identification
1. MASCOT software (Matrix Science, London, England)
2. MS-Fit software (http://prospector.ucsf.edu)
3. Methods
The methods described below have been successfully used in the authors’
laboratory for proteomics studies in human breast cancer specimens (1,9) and
can be applied to other cancer tissues as well. Breast tumors and matched
normal tissues were obtained from the Tissue Repository Unit of the National
University Hospital, Singapore, after approval by our Institutional Review
Board.
3.1. Preparation of Tissue Sections for LCM
In this step, frozen tissues can be directly transferred from the –80°C freezer,
where they had been stored after surgical excision and trimming, to a pre-cooled
tube containing 70% (v/v) ethanol and kept on ice. Ethanol-fixed paraffin-
embedded tissue blocks should be prepared as quickly as possible, and the
completed blocks stored at or below 4°C.
1. Fix the frozen tissue overnight in 70% ethanol at 4°C.
2. Place each ethanol-fixed tissue piece, trimmed to appropriate dimensions, into
a pre-cooled cassette within the tissue processor and dehydrate according to the
following procedure: 30 min each in 70% and 80% ethanol at 40°C; 45 min in
95% ethanol at 40°C (twice); 45 min in 100% ethanol at 40°C (twice), and 45 min
in xylene at 40°C (twice) (see Note 2).
3. Embed the specimen in paraffin using embedding molds, with four changes of
paraffin after every 30-min interval.
4. Store the paraffin blocks at or below 4°C, if they were not to be processed
immediately for sectioning.
5. Put the block in a –20°C freezer for at least 1 h before cutting sections from it.
6. Cut sections of 8 μm thickness using a standard microtome. Blades should be
changed regularly (see Note 3).
7. Collect the tissue sections on uncharged microscopic glass slides, allow tissue
sections to be air dried, and store the cut sections at or below 4°C.
3.2. Staining of Paraffin-embedded Sections
The staining of sections for LCM is similar to that used in most histology
laboratories for morphological assessment. However, using minimal amount of
the stain to visualize the tissue for microdissection will improve macromolecule
recovery (see Note 4). One tablet of protease inhibitor cocktail should be added
82 Zhang and Koay
to every 10 ml of each reagent (except xylene), and all reagents prepared using
double deionized water or Milli-Q®water. Staining should be performed as
close as possible to the scheduled LCM dissection.
1. Deparaffinize the sections in fresh xylene for 5 min, followed by another 5 min
with a fresh change of xylene.
2. Rehydrate for 15 s in each step of the following series: 100% ethanol, 95%
ethanol, 75% ethanol, and deionized water.
3. Stain with Mayer’s Hematoxylin for 30 s.
4. Rinse off excess stain with deionized water for 15 s; repeat rinse a second time.
5. Dehydrate for 15 s in 70% ethanol.
6. Stain with Eosin Y for 5 s.
7. Dehydrate the sections for 15 s (twice) in 95% ethanol, 15 s (twice) in 100%
ethanol, and 60 s in xylene.
8. Air-dry for approximately 2–5 min to allow xylene to evaporate completely (see
Note 5).
9. The tissue is now ready for LCM (see Note 6).
3.3. Laser Capture Microdissection and Protein Sample Preparation
The PixCell II LCM system (Arcturus Engineering, Mountain View, CA,
USA) is used for specific microdissection of tumor cells in our laboratory.
Tissue sections are usually mounted on uncoated glass slides to provide support
for the CapSure cap during microdissection. LCM utilizes an infrared laser
integrated into a standard microscope, and when the desired cells move into
the path of the light source, the investigator activates the laser, which in
turn activates the membrane (a short laser pulse emitted heats the transparent
membrane to ∼90°C for 5 ms). This melts the membrane, with subsequent
binding and encapsulation of the cells of interest, segregating them from the
surrounding cells and connective tissues. Images of the tissues before and after
microdissection and of the captured cells on the cap can be visualized, thus
maintaining an accurate record of each dissection. The laser beam diameter
may be adjusted from 7.5 to 30 μm to procure either single cells or groups of
cells, respectively.
1. Place the slide containing the prepared tissue on the microscope stage. Set the
laser parameters as follows: spot diameter at 15 μm, pulse duration at 5 ms, and
power at 50 mW.
2. Scan the tissue section to locate the desired cells. Dissect out the target cells of
interest and capture all encapsulated cells from each section in quick succession
into one cap. Cells dissected from ∼2500 shots can be captured into one cap (see
Note 7). Figure 1 shows an example of tumor cells before and after microdis-
section.
Combining LCM with 2-D Gel Electrophoresis 83
A B C
Fig. 1. Laser capture microdissection (LCM) of breast tumor cells. The tissue section
on the uncharged glass slide was stained with hematoxylin and eosin and microdissected
with the PixCell II LCM system (Arcturus Engineering). (A) section before LCM; (B)
section after LCM; (C) microdissected cell.
3. Place the LCM cap on an Eppendorf tube containing 100 μl of lysis buffer with
protease inhibitor and invert the tube and vortex vigorously for 1 min.
4. Place the tube on ice for approximately 20 min and sonicate the microdissected
sample in a bath sonicator with 5 s pulses, in between 5-s intervals, for a duration
of 1 min.
5. Replace the sample on ice immediately after 1-min sonication.
6. Centrifuge the sample at 16,000 gfor 20 min at 4°C and transfer the supernatant
to a new Eppendorf tube.
7. Determine the protein concentration using the PlusOne 2D Quantitation kit (GE
Healthcare) and clean up the sample using the PlusOne 2-D cleanup kit (GE
Healthcare), following the manufacturer’s instructions closely.
8. Dissolve the protein pellet in the appropriate volume of sample rehydration buffer
and aliquot according to experimental plans for immediate and later usage. Store
the aliquotted samples at –80°C until analyzed (see Note 8).
3.4. First-dimension Gel Electrophoresis (Isoelectric Focusing)
1. Prepare the strip holder for the 18-cm IPG strip (see Note 9).
2. Squeeze a few drops of Ettan™IPGphor™Strip Holder Cleaning Solution (GE
Healthcare) into the slot and clean thoroughly. Rinse with Milli-Q water and dry
completely.
3. Mix approximately 50 μl of the reconstituted protein samples (∼100–150 μg)
with the appropriate volume of rehydration buffer. The total volume should be
340 μl for one 18-cm IPG strip.
4. Transfer the entire volume of the diluted protein sample into the groove of the
IPG strip holder.
5. Remove the cover from the IPG strip (18 cm, pH 3–10) and place the IPG strip
in the holder such that the gel of the strip is in contact with the sample (i.e., gel
84 Zhang and Koay
side down). Try to remove any trapped air bubbles by lifting the strip up and
down from one side.
6. Overlay the IPG strip with 2–3 ml of DryStrip Cover Fluid to prevent urea
crystallization and evaporation, and replace the cover on the strip holder.
7. Rehydrate the IPG strip at 20 V for 12 h at 20°C.
8. Perform IEF under the following conditions: 500 V for 1 h, 2000 V for 1 h,
4000 V for 1 h, and 8000 V for 6 h.
9. Once focusing is complete, pour off the oil. The strips can be stored at –20°C for
several weeks, or immediately treated as described below (see Subheading 3.5).
3.5. IPG Strip Equilibration
1. Place the focused IPG strips in a container with 10 ml of equilibration buffer A
and shake for 15 min at RT (see Note 10).
2. Transfer the IPG strip to a container with 10 ml of equilibration buffer B and
shake for 15 min at RT (see Note 10).
3. The equilibrated strips can then be processed for second-dimension gel
electrophoresis.
3.6. Second-dimensional SDS-PAGE
Prepare the SDS-polyacrylamide gels in advance, and make sure that the
gels are well polymerized before performing the equilibration of IPG strips.
The proteins have to be charged by equilibration with SDS, and be reduced
and alkylated to avoid the formation of oligomers. In our laboratory, we use
the PROTEAN II xi Cell system (Bio-Rad) for SDS-PAGE.
1. Assemble the gel casting cassette as per the manufacturer’s instructions.
2. Prepare 10% SDS-PAGE (see Note 10) and pour the solution slowly into the
cassette (two 16 cm × 20 cm glass plates sandwiched by 1.5-mm thick spacers)
until the gel height is approximately 1 cm from the top.
3. Overlay the gel solution with 2 ml of water-saturated isobutanol. It is best to
pour 1 ml of water-saturated isobutanol from one side of the gel and 1 ml on
the other side. Do not pour it all along the gel meniscus.
4. Allow the gel to polymerize for at least 2 h.
5. When polymerization is completed, remove the water-saturated isobutanol and
rinse with water again.
6. With a pair of forceps, carefully place the equilibrated strip on top of the PAGE
gel, with the acidic side of the strip at left. Cover the strip with melted agarose
sealing solution (see Note 11).
7. Assemble the electrophoresis unit (Bio-Rad) and perform electrophoresis at 15°C
as follows: 40 V for 15 min or until the blue dye enters the gel and then raise
the voltage to 125 V and run the gel overnight or until the blue dye migrates to
the bottom of the gel.
8. Switch off the main power and disassemble the gel cassette.
Combining LCM with 2-D Gel Electrophoresis 85
9. Place the gel in a glass container and wash the gel with Milli-Q water.
10. Stain the gel using the mass spectrometry-compatible silver staining protocol
(see Subheading 3.7).
3.7. Silver Staining and Image Analysis
1. The silver staining protocol as described below is used in the authors’ laboratory
and is highly compatible with protein identification by MALDI-TOF MS and
MALDI-TOF/TOF MS/MS. It should be noted that adequate washing with Milli-
Q water is essential to reduce the risk of keratin contamination. All the solutions
must be prepared with Milli-Q water, and all the chemical reagents should be
filtered to remove any particles that may cause interference during MS analysis.
All solutions prepared from solid chemicals should be freshly prepared before
performing silver staining. Fix the gel with fixing solution for at least 2 h,
changing the solution afresh at hourly intervals.
2. Briefly wash with Milli-Q water, with constant shaking for about 15 min.
3. Remove the wash and cover the gel with appropriate sensitivity-enhancing
solution and incubate for 1 h, with constant shaking.
4. Wash the gel thoroughly with Milli-Q water for6×15min, with gentle shaking
and replacing with fresh Milli-Q water after each cycle (see Note 12).
5. Stain the gel with silver staining solution for 30 min.
6. Wash off excess stain from the gel with Milli-Q water (twice, for2×1min).
7. Develop the gel for 5–30 min in a developing solution (see Note 13).
8. Add Stop Solution and shake the gel for approximately 20 min to stop the
reaction.
9. Wash the gel using Milli-Q water for 20 min; replace water and repeat the wash.
10. Scan the gel using Personal Densitometer SI, or store the gel in the gel soak
solution for analysis at a later time.
11. Capture the image using ImageMaster 2D Elite software (GE Healthcare). The
image analysis includes spot detection, quantification and normalization of spot
intensity to the background interferences, according to the instructions from the
software. An example of images showing the differences between the protein
profiles of LCM-microdissected HER-2/neu positive and -negative tumor cells
is shown in Fig. 2.
12. Analyze the image using the software and identify spots that show signif-
icant differences in spot intensities (see Note 14), reflecting differential protein
expression in the two subtypes of breast cancer triggered by the presence or
suppression of HER-2/neu oncogene. Only those spots that show either more
than threefold or less than threefold change in signal intensity, consistently
from three replicate sets of gels, are considered as demonstrating differential
protein expression and selected for further analysis by MALDI-TOF MS/MS.
The likelihood of any protein displaying less convincing evidence of differential
protein expression being a potential biomarker for early detection of tumor
growth or a therapeutic target for breast cancer treatment is low.
86 Zhang and Koay
NP000627
P06753-2
AAB49495
P07339
AAH025396 P04075
28
35
50
92
kDa pI3 pI3
HER-2/neu-P HER-2/neu-N
10 10
NP004095
NP001531
Fig. 2. Silver-stained protein profiles of LCM-dissected cells. Protein samples from
HER-2/neu positive and -negative cells are separated by using IPG®( strips (18 cm,
pH 3–10 NL) and homogeneous SDS-PAGE (10%), and then stained with silver
nitrate. Silver-stained gels were scanned using the Personal Densitometer SI (Molecular
Dynamics) and differentially expressed protein spots were analyzed by ImageMaster
2-D Elite software (GE Healthcare). The Accession Numbers indicate the protein
ID identified by MALDI-TOF/TOF tandem mass spectrometry and NCBInr database
search using Mascot software (Matrix Science, London, UK).
3.8. Trypsin Digestion and Preparation of Peptides for Mass
Spectrometric Analysis
1. Excise the silver-stained protein spots showing significant differential protein
expression, as mentioned above, one at a time, taking care not to include adjacent
proteins in vicinity, and transfer to individual tubes.
2. Wash with 100 μl of Milli-Q water for 5 min.
3. Add 50 μl of the destaining solution into the tubes, and about 20 min on a
platform shaker at RT until the gels become clear in color.
4. Remove the solution carefully and wash with 100 μl of Milli-Q water.
5. Incubate the gel pieces with 25 mM sodium bicarbonate for 20 min, and then
cut them into smaller pieces with the tip of the transfer pipette. Avoid carryover
and contamination during repetitive work on consecutive samples.
6. Rinse the gel pieces with Milli-Q water, discard the wash after pulsing down
the gel pieces, and repeat the washing process three times.
7. Add 100 μl of dehydrating solution and incubate for 20 min at RT.
8. Dry the gel pieces in a SpeedVac centrifuge.
9. Re-swell the dried gel pieces with 10–20 μl of Digestion Solution and leave
overnight at 37°C to ensure complete digestion.
10. Extract the resultant hydrophilic peptides first with 10 μl of Milli-Q water for 1 h.
Combining LCM with 2-D Gel Electrophoresis 87
11. Then extract the hydrophobic peptides with Extraction Solution for 2 h.
12. Pool the extracted hydrophilic and hydrophobic peptides and dry the peptide
mixture using the SpeedVac centrifuge.
13. Redissolve the dried peptides in 10 μl of 0.1% (v/v) TFA.
14. Desalt the sample with ZipTip C18 columns (Millipore) and elute the treated
and purified peptides with 2.5 μl of Eluant.
15. Mix 0.5 μl of the sample eluate with 0.5 μl of CHCA matrix (3 mg/ml) and spot
the mixture onto the stainless steel MALDI-TOF sample target plates.
16. The pretreated peptide samples must be stored on ice during transfer to the
core facility for mass spectrometric analysis. In our laboratory, peptide mass
spectra are obtained by the Applied Biosystems 4700 Proteomics Analyzer
MALDI-TOF/TOF mass spectrometer, set in the positive ion reflector mode.
The subsequent MS/MS analyses are performed in a data-dependent manner,
and the 10 most abundant ions fulfilling certain preset criteria are subjected to
high-energy CID analysis. The collision energy is set to 1 keV, and nitrogen is
used as the collision gas.
3.9. Database Search to Match Protein Identities
Database searches were conducted using the MASCOT search engine
(http://www.matrixscience.com). For database search, known contamination
peaks, such as keratin and autoproteolysis peaks, were removed prior to
database search. Protein identification was performed using the MASCOT
software (Matrix Science, London, UK), and all tandem mass spectra were
searched against the NCBInr database, with mass accuracy of within 200 ppm
for mass measurement, and within 0.5 Da for MS/MS tolerance window.
Searches were performed without constraining the protein molecular weight
(Mr) or isoelectric point (pI) and species, and allowing for carbamidomethy-
lation of cysteine and partial oxidation of methionine residues. Up to one missed
tryptic cleavage was considered for all tryptic-mass searches. Protein scores
greater than 75 are considered to be significant (p< 0.05).
3.10. Experimental Example: Differential Protein Profiles
between HER-2/neu Positive and -Negative Breast Tumors
We dissected the tumor cells from two different subtypes of breast tumors
and compared their protein profiles, based on the protocols described above.
Figure 2 shows the LCM-dissected tumor cell protein patterns visualized by
silver staining. It should be noted that pooled protein samples from different
cases of the same tumor subtypes were used for 2-D GE. This gel-based
protein visualization technique requires high amount of proteins, and thus
more sensitive detecting reagents and protein identification strategies had to
be developed to produce meaningful results (see Notes 15 and 16). Using
88 Zhang and Koay
the silver-staining protocol, we identified 500–600 protein spots in the protein
profiles generated by coupling LCM and 2-D GE. Protein spots of interest would
be excavated and digested with trypsin (Promega), desalted with ZipTipc18
(Millipore), and analyzed using MALDI-TOF/TOF tandem mass spectrometry.
Protein identities, as shown in Fig. 2, are obtained by searching the NCBInr
databases using the MASCOT software (Matrix Science).
4. Notes
1. All the chemical solutions should be filtered by passing them through filter paper
(Cat No. 1001 150, Whatman®, Whatman International Limited, Springfield
Mill, Maidstone, Kent, England) to minimize precipitates occurring onto the
gels during silver staining.
2. Tissue processors in standard histopathology laboratories generally include
formalin fixation as the first step in the paraffin infiltration procedure. It is
important to avoid these steps when processing tissues intended for molecular
gene and proteome profiling.
3. Consistent LCM transfers have been demonstrated from 5–10 μm thick paraffin-
embedded tissue sections. For a successful LCM transfer, the strength of the bond
between polymer film and targeted tissue must be stronger than that between the
tissue and the underlying glass slide. Therefore, for most tissue types, sections
should be collected with uncharged glass slides. To prevent cross-contamination
while sectioning, residual paraffin and tissue fragments should be wiped off
from the area of the sectioning blade with xylenes between consecutive slides.
If possible, a fresh microtome blade should be used to section a different block.
4. In our hands, hematoxylin and eosin are best reduced to 10% of their standard
concentrations used for routine histomorphological work, when applied to slides
prepared for LCM. Breast tumor cells can be clearly visualized and identified
from other cell types, without influencing the procurement of tumor cells by
LCM, with this modification. Minimum staining also improves macromolecular
recovery during cellular protein extraction.
5. Complete dehydration and air drying of sections are the main factors influencing
the efficiency of LCM. Prolonged air drying or presence of moisture in the
sections appears to inhibit, at least partially, the transfer of cells to the plastic
firm.
6. If the investigators have less experience in checking cancer tissue sections,
we strongly recommend that investigators consult with the pathologists in their
institutions to get assistance in identifying the target cell types that will be
microdissected using LCM. It is essential to avoid contamination of other cell
types, or dissecting the wrong cells.
7. During microdissection, make sure that there are no irregularities on the tissue
surface in or near the area to be microdissected. It should also be noted that
wrinkles can elevate the LCM cap away from the tissue surface and decrease the
Combining LCM with 2-D Gel Electrophoresis 89
membrane contact during laser activation. Use an adhesive pad after microdis-
section to remove cells that may have attached non-specifically to the LCM
cap. A cap-alone control is recommended for each experiment to ensure that
non-specific transfer is not occurring during microdissection. The cap should be
processed together with other tissue-containing caps and serves as a negative
control. For protein separation by 2-D GE, 20 to 30 sections from each tissue
sample are dissected, depending on the percentage of targets cells in the full
sections. Generally, 2300–2700 laser pulse shots are used for each cup. Cells
from at least 50,000 shots (spot diameter is 15 μm) are required for each
18-cm gel.
8. Up to 15 mg of proteins can be solubilized with 500 μl of the sample rehydration
buffer, but with our breast tumor tissue samples, we usually reconstitute 1–2 mg
of extracted proteins in 500 μl, or 2–4 mg/ml. It is recommended that the
reconstituted proteins be stored in appropriate aliquots, and that only the required
number of aliquots needed for the experiment at hand be removed at any time,
to avoid repeated freezing and thawing the peptides, which will lead to sample
deterioration.
9. IEF is performed using Ettan™IPGphor™IEF electrophoresis unit. Rehydration
loading of protein samples is used in the authors’ laboratory. The IPG strips for
first-dimensional separation are commercially available, and can be procured
from GE Healthcare and other suppliers. IPG strips with various pH gradients and
dimensions are available. They are used for protein separation with appropriate
resolution needed. The strips should be kept frozen at –20°C, and thawed just
before use. The IEF conditions are dependent on the pH range. Reference to the
manufacturer’s protocol is recommended. For alkali pH loading, cup loading
is a must, and DTT in the rehydration buffer should be replaced by other
reducing agents, such as hydroxyethyl-disulfide (HED) reagent (Destreak, GE
Healthcare).
10. It is essential to equilibrate the strips before being applied for the second-
dimension gel electrophoresis (2-D SDS-PAGE). DTT added to buffer A will
reduce the disulfide bonds whereas IAA in buffer B will alkylate the formed
sulfydryl groups of proteins. This is to prevent re-oxidation of sulfydryl groups
and streaking of spots during 2-D SDS-PAGE. Further, the presence of SDS
makes the proteins negatively charged and suitably primed for SDS-PAGE. Use
the best quality SDS available for sample and running buffers that include SDS
in their formulation. We recommend C12 Grade SDS from Pierce (Rockford, IL,
USA).
11. When placing the strips on top of the gel, ensure that the plastic backing of the
strips is in contact with the glass wall. If necessary, the strips can be trimmed
properly. When adding agarose sealing solution, make sure that there are no air
bubbles trapped between the IEF strip and 2-D gel.
12. Wash the gels thoroughly and repeatedly, as recommended, prior to the devel-
opment step and during the development step itself, to get clear stained gels.
During the development of the gels, formaldehyde should be added prior to use,
90 Zhang and Koay
and the suggested concentration should be followed strictly to avoid interference
during MALDI-TOF analysis. During the developing stage, the gel should be
constantly shaken to reduce the background.
13. The developing time depends on the total amount of protein that is used for
2-D separation. With a higher amount of protein, a shorter developing time can
be used, without compromising the aim of visualizing the maximum number of
protein spots.
14. It is important to manually verify spot detection and matching, as the variations
in gel resolution, staining, gel background, and automatic image analysis may
not correctly define the spot contours in every case. This variability and the
complexity of 2-D gel patterns hinder the accurate matching of analogous spots
in different gels.
15. In our experience, approximately 500 to 600 distinct proteins from the dissected
breast tumor cells can be visualized on 2D-PAGE stained with silver. On average,
we can extract approximately 4–6 μg of total cellular proteins from 2500 laser
pulses. Our experience is that silver staining of LCM-dissected cell proteins is a
sufficiently sensitive tool for isolating and identifying the dysregulated cellular
proteins of high or moderate abundance. However, for the dysregulated proteins
of low abundance, the lower detection limit of this technology would have to
be enhanced by other techniques such as 125-iodine labeling or biotinylation
and fluorescent dye labeling. In addition, the use of scanning immunoblotting
with class-specific antibodies, for example, would allow sensitive detection of
specific subsets of proteins, e.g., all known proteins involved with cell-cycle
regulation.
16. Protein identification by MALDI-TOF, LC-MS/MS, or other techniques is also
limited by the requirement of a minimal protein input amount, which is often not
attainable from certain types of biopsy samples. A useful strategy to improve
protein identification is to produce parallel “diagnostic” fingerprints derived
from microdissected cells and “sequencing” the fingerprints generated from the
whole tissue section from each case. Alignment of the diagnostic and sequencing
2D gels permits determination of the proteins of interest for subsequent mass
spectrometry or N-terminal sequence analysis.
Acknowledgments
The Tumor Repository of the National University Hospital, Singapore,
provided the clinical breast cancer frozen tissues for LCM. The use of the
PixCell II LCM system was courtesy of the Department of Pathology, Yong
Loo Lin School of Medicine, National University of Singapore (NUS). This
work was supported by an Academic Research Fund from the NUS (Grant No.
R-179-000-032) to the authors.
Combining LCM with 2-D Gel Electrophoresis 91
References
1. Zhang, D., Tai, L. K., Wong, L. L., Sethi, S. K., Koay, E. S. (2005) Proteomics of
breast cancer: enhanced expression of cytokeratin 19 in human epidermal growth
factor receptor type 2 positive breast tumors. Proteomics 5, 1797–1805.
2. Neubauer, H., Clare, S. E., Kurek, R., Fehm, T., Wallwiener, D., Sotlar, K., et al.
(2006) Breast cancer proteomics by laser capture microdissection, sample pooling,
54-cm IPG IEF, and differential iodine radioisotope detection. Electrophoresis 27,
1840–1852.
3. Lawrie, L. C., Curran, S., McLeod, H. L., Fothergill, J. E., Murray, G. I. (2001)
Application of laser capture microdissection and proteomics in colon cancer. J.
Clin. Pathol: Mol. Pathol.54, 253–258.
4. Ai, J., Tan, Y., Ying, W., Hong, Y., Liu, S., Wu, M., et al. (2006) Proteome
analysis of hepatocellular carcinoma by laser capture microdissection. Proteomics
6, 538–546.
5. Ahram, M., Flaig, M. J., Gillespie, J. W., Duray, P. H., Linehan, W. M.,
Ornstein, D. K., et al. (2003) Evaluation of ethanol-fixed, paraffin-embedded tissues
for proteomic applications. Proteomics 3, 413–421.
6. Greengauz-Roberts, O., Stoppler, H., Nomura, S., Yamaguchi, H., Goldenring,
J. R., Podolskym R. H., et al. (2005) Saturation labeling with cysteine-reactive
cyanine fluorescent dyes provides increased sensitivity for protein expression
profiling of laser-microdissected clinical specimens. Proteomics 5, 1746–1757.
7. Zang, L., Palmer-Toy, D., Hancock, W. S., Sgroi, D. C., Karger, B. L. (2004)
Proteomic analysis of ductal carcinoma of the breast using laser capture microdis-
section, LC-MS, and 16O/18 O isotopic labeling. J. Proteome Res.3, 604–612.
8. Li, C., Hong, Y., Tan, Y. X., Zhou, H., Ai, J. H., Li, S. J., et al. (2004) Accurate
qualitative and quantitative proteomic analysis of clinical hepatocellular carcinoma
using laser capture microdissection coupled with isotope-coded affinity tag and
two-dimensional liquid chromatography mass spectrometry. Mol. Cell. Proteomics
3, 399–409.
9. Zhang, D., Tai, L. K., Wong, L. L., Chiu, L. L., Sethi, S. K., and Koay, E. S. (2005)
Proteomic study reveals that proteins involved in metabolic and detoxification
pathways are highly expressed in HER-2/neu-positive breast cancer. Mol. Cell.
Proteomics 4, 1686–1696.
10. Cowherd, S. M., Espina, V. A., Petricoin, E. F. III, Liotta, L. A. (2004) Proteomic
analysis of human breast cancer tissue with laser-capture microdissection and
reverse-phase protein microarrays. Clin. Breast Cancer 5, 385–392.
11. Gulmann, C., Espina, V., Petricoin, E. III, Longo, D. L., Santi, M., Knutsen, T.,
et al. (2005) Proteomic analysis of apoptotic pathways reveals prognostic factors
in follicular lymphoma. Clin. Cancer Res.11, 5847–5855.
6
Optimizing the Difference Gel Electrophoresis
(DIGE) Technology
David B. Friedman and Kathryn S. Lilley
Summary
Difference gel electrophoresis (DIGE) technology has been used to provide a powerful
quantitative component to proteomics experiments involving 2D gel electrophoresis. DIGE
combines spectrally resolvable fluorescent dyes (Cy2, Cy3, and Cy5) with sample multi-
plexing for low technical variation, and uses an internal standard methodology to analyze
replicate samples from multiple experimental conditions with unsurpassed statistical confi-
dence for 2D gel-based differential display proteomics. DIGE experiments can facilely
accommodate sufficient independent (biological) replicate samples to control for the large
interpersonal variation expected from clinical samples. The use of multivariate statistical
analyses can then be used to assess the global variation in a complex set of independent
samples, filtering out the noise from technical variation and normal biological variation
thereby focusing on the underlying variation that can describe different disease states. This
chapter focuses on the design and implementation of the DIGE methodology employing
the use of a pooled-sample internal standard in conjunction with the minimal CyDye
chemistry. Notes are also provided for the use of the alternative saturation labeling
chemistry.
Key Words: difference gel electrophoresis; two-dimensional gel electrophoresis;
quantification.
1. Introduction
Human disease phenotypes are a direct result of protein expression and
modification. In many cases, such phenotypes cannot be tied directly to a single
alteration in the genome or resulting proteome, but are likely to be the result
From: Methods in Molecular Biology, vol. 428: Clinical Proteomics: Methods and Protocols
Edited by: A. Vlahou © Humana Press, Totowa, NJ
93
94 Friedman and Lilley
of multiple factors. Studying disease at the protein level is challenging, but
as proteins are the mediators of phenotype, the study of protein abundance
on a global scale is required to gain a more complete understanding of the
underlying molecular mechanisms of disease. Proteomics in the clinical setting
is rapidly developing and is having a major impact on the way in which diseases
will be diagnosed, treated, and monitored (1). It has been estimated that there
could be hundreds of thousands of different protein isoforms in a mammalian
cell, but the vast dynamic range of protein abundance results in only the
most abundant species of proteins being observable by quantitative proteomics
approaches unless technically variable biochemical or subcellular fractionation
is employed. The repertoire of techniques and associated hardware, which is
now applied to this field, is expanding exponentially, and although a complete
visualization of the proteome is still beyond reach of any single technique, each
technology platform can provide complementary datasets.
Difference gel electrophoresis (DIGE) has proven to be a powerful quanti-
tative technology for differential display proteomics on a global level, where
the individual abundance changes for thousands of intact proteins can be simul-
taneously monitored in replicate samples over multiple variables with statistical
confidence (see Note 1). This includes quantitative information on protein
isoforms that arise due to post-translational modifications (such as acetylation
or phosphorylation), which result in a change in the isoelectric point of the
protein. This also includes splice variants and the results of protein processing,
all of which are resolved for individual quantification and subsequent analysis
by MS.
DIGE is based on conventional 2D gel technology that is capable of resolving
several thousands of intact proteins first by charge using isoelectric focusing
(IEF) and then by apparent molecular mass using SDS-polyacrylamide gel
electrophoresis (PAGE) (6,7) (see Note 2 and Chapters 4 and 5 by Cho et al.
and Zhang et al., respectively). Importantly, DIGE overcomes many of the
limitations commonly associated with 2D gels such as analytical (gel-to-gel)
variation and limited dynamic range that can severely hamper a quantitative
differential display study. This is accomplished using up to three spectrally
resolvable fluorescent dyes (Cy2, Cy3, and Cy5, referred to as CyDyes) that
enable low- to subnanogram sensitivities with >104linear dynamic range, and
then by multiplexing the prelabeled samples into the same analytical run (2D
gel). Multiplexing in this way allows for direct quantitative measurements
between the samples coresolved in the same gel, and is therefore beyond
the limitations imposed by between-gel comparisons with conventional 2D
gels.
The highest statistical power of this multiplexing approach stems from the
utilization of a pooled-sample internal standard comprised of an equal aliquot
Optimizing DIGE Technology 95
of every sample in the experiment (see Subheading 1.2.1). With this method,
two dyes (Cy3 and Cy5) are used to individually label two independent samples
from a much larger experiment, and the Cy2 dye is used to label an internal
standard, which is comprised of an equal aliquot of proteins from every sample
in the experiment. This pooled-sample internal standard is labeled only once in
bulk to avoid additional technical variation, and enough is made and labeled to
allow for an equal aliquot to be coresolved on each gel. The three differentially
labeled samples are then coresolved on the same 2D gel, after which direct
measurements can be made for each resolved protein using the spectrally
exclusive dye channels without interference from technical variation of the
separation (gel-to-gel variation).
Rather than making direct quantitative measurements between the two
samples in the gel, the measurements are instead made relative to the Cy2 signal
for each resolved protein. The Cy2 signal should be the same for a given protein
across different gels because it came from the same bulk mixture/labeling;
therefore, any difference represents gel-to-gel variation, which can be effec-
tively neutralized by normalizing all Cy2 values for a given protein across all
gels. Using the Cy2 signal to normalize ratios between gels then allows for the
Cy3:Cy2 and Cy5:Cy2 ratios for each protein within each gel to be normalized
to the cognate ratios from the other gels, encompassing all samples. Each gel
may contain different (and/or replicate) samples in the Cy3 and Cy5 channels,
but all samples can be quantified relative to each other because each protein
from each sample is measured to the cognate Cy2 signal from the internal
standard present on each gel. With the use of sufficient replicates, a plethora of
advanced statistical tests can be applied, which can highlight proteins of interest
whose change in expression is related to the disease state under investigation.
Since the technical noise is low,these vital replicates should be independent
(biological) replicates as most of the observed variations will be clinical sample
related rather than technical or experimental related.
In a final step, specific proteins of interest are then identified using standard
mass spectrometry (MS) approaches on gel-resolved proteins that have been
excised and proteolyzed into a discrete set of peptides. Briefly, excised proteins
are subjected to in-gel digestion with trypsin protease (typically), and MS
is used to acquire accurate mass determinations on the resulting peptides,
as well as fragmentation on individual peptides. The mass spectral data are
then used to identify statistically significant candidate protein matches through
sophisticated computer search algorithms that compare the observed MS data
with theoretical peptide masses (using data generated by peptide mass finger-
printing) or collision-induced fragmentation patterns (obtained from tandem
MS) generated in silico from protein sequences present in databases. (see
Chapter 19 by Fitzgibbon et al.).
96 Friedman and Lilley
1.1. Optimizing Sensitivity and Resolution
There are currently two forms of CyDye labeling chemistries available:
minimal labeling involving the use of N-hydroxy succinimidyl (NHS) ester
reagents for low-stoichiometry labeling of proteins largely via lysine residues,
and saturation labeling, which utilize maleimide reagents for the stoichiometric
labeling of cysteine sulfhydryls.
The most established DIGE chemistry is the “minimal labeling” method,
which has been commercially available since July 2002. Here the CyDye DIGE
fluors are supplied as NHS esters, which react with the -amine groups of
lysine side chains. The three fluors are mass matched (ca. 500 Da), and carry
an intrinsic +1 charge to compensate for the loss of each proton-accepting site
that becomes labeled (thereby maintaining the pI of the labeled protein). Each
dye molecule also adds a hydrophobic component to proteins, which along with
MW influences how proteins migrate in SDS-PAGE.
Minimal labeling reactions are optimized such that only 2–5% of the total
number of lysine residues are labeled, such that on average a given labeled
protein would contain only one dye molecule. This is necessary because lysine is
an abundant amino acid, and multiple labeling events may affect the hydropho-
bicity of some proteins such that they may no longer remain soluble under
2DE conditions. Although a given protein form may exhibit specific labeling
efficiencies, these will be the same for labeling with all three dyes, allowing
for direct relative quantification. Minimal labeling with CyDye DIGE fluors is
very sensitive, comparable to silver-staining or postelectrophoretic fluorescent
stains such as Sypro Ruby, Deep Purple or Flamingo Pink (ca. 1 ng), but with
a linear response in protein concentration over five orders of magnitude (8)(see
Note 3).
For maleimide labeling of the cysteine sulfhydryls, the overall lower cysteine
content in proteins allows for labeling of these residues to saturation without
increasing the overall hydrophobicity of the proteins to cause insolubility
problems. Saturation labeling is ultimately more sensitive (150–500 picograms,
and even more so for proteins with high cysteine content). Its use is not as
commonplace, most likely due to the availability of only Cy3 and Cy5 with
this chemistry (see Note 4), the fact that it is blind to the small but significant
population of noncysteine containing proteins, and the additional optimization
of complete cysteine reduction necessary for reproducible labeling. For these
reasons, saturation DIGE is usually reserved for experiments where samples
are limited, where the advantage of the increased sensitivity outweigh these
additional considerations.
To maximize the information that can be gained from DIGE experiments, it is
imperative that resolution of protein species within gels is optimized. Although
single 2DE runs can resolve proteins with pI ranges between pH 3 and 11, and
Optimizing DIGE Technology 97
apparent molecular mass ranges between 10 and 200 kDa, higher resolution and
sensitivity can be obtained by running a series of medium range (e.g., pH 4–7,
7–11) and narrow range (e.g., pH 5–6) IEF gradients with increasing protein
loads, leading to an overall more comprehensive proteomic analysis (6,7,10).
(see Note 5). This is analogous to gaining increased resolution and sensitivity in
an LC/MS-based strategy by using multiple high performance liquid chromatog-
raphy columns with different affinity chemistries [e.g., MuDPIT (12)]. Much of
the sensitivity limitation associated with 2D gels can be attributed to the analysis
of unfractionated, whole-cell and whole-tissue extracts. Additional sensitivity
can be gained via enrichment for the proteins of interest, such as by analyzing
prefractionated or subcellular samples, or immune complexes. However, the
additional experimental manipulations required for prefractionation introduce
more technical variation into the samples and necessitates increased independent
(biological) replicates (which can be accommodated with the DIGE internal
standard methodology).
The identification of proteins of interest using MS can be performed directly
from the DIGE gels when protein amounts have been optimized in this way (see
Subheading 3.5). Alternatively, some experimental approaches perform DIGE
analysis using “analytical” gels with lower protein amounts, followed by protein
excision from a secondary, “preparative” gel with higher protein amounts. This
approach has its advantages when dealing with small sample amounts, such is
often the case using the saturation dye chemistries, but is also prone to uncer-
tainties that arise due to the disproportionate amount of protein loading (see
Note 6). The methods presented in this protocol are for optimization of both
the DIGE data as well as material for subsequent MS using high protein loads.
1.2. Optimizing Statistical Significance
1.2.1. Using the Internal Standard
The ability to coresolve and compare two or three samples in a single gel is
attractive, because it allows for direct relative quantification for a given protein
without any interference from gel-to-gel variations in migration and resolution,
removing the need for running replicate gels for each sample (similar to stable
isotope LC/MS-based strategies, see Chapter 10). This approach has limited
statistical power, however, since confidence intervals are determined based on
the overall variation within a population (see Subheading 3.6.2).
Many researchers new to DIGE technology are not immediately aware of
the increased statistical advantage and multiplexing capabilities of DIGE when
combining this approach with a pooled-sample mixture as an internal standard
for a series of coordinated DIGE gels (13). This design will allow for repet-
itive measurements (vital to any type of experimental investigation), and in
98 Friedman and Lilley
such a way as to control both for gel-to-gel variation and provide increased
statistical confidence. In this way, statistical confidence can be measured for
each individual protein based on the variance of repetitive measurements,
independent of the variation in the population. Incorporating independently
prepared replicate samples into the experimental design also controls for
unexpected variation introduced into the samples during sample preparation.
This more complex and statistically powerful experimental design is accom-
plished by using one of the three dyes (usually Cy2) to label an internal standard,
which is comprised of equal aliquots of protein from all of the samples in an
experiment. The total amount of the Cy2-labeled internal standard is such that
an equal aliquot can be coresolved within each DIGE gel that also contains
an individual Cy3- and Cy5-labeled sample from the experiment. Since this
standard is composed of all of the samples in a coordinated experiment, each
protein in a given sample should be represented in the standard and thus have
its own unique internal standard (see Note 7). Direct quantitative comparisons
are made individually for each resolved protein between the Cy3- or Cy5-
labeled samples and the cognate protein signal from the Cy2-labeled standard
for that gel (without interference from gel-to-gel variation) and results in the
calculation of a standardized abundance for every spot matched across all
gels within a multigel experiment. The individual signals from the internal
standard are also used to normalize and compare between each in-gel direct
quantitative comparison for that particular protein from the other gels. Using
the Cy2-labeled standard in this fashion, therefore, allows for more precise
and complex quantitative comparisons between gels, including independent
(biological) sample repetition (Fig. 1).
Importantly, the internal standard experimental design allows for the identifi-
cation of significant changes that would not have been identified if the analyses
were performed separately, even when using Cy3- and Cy5-labeled samples on
the same DIGE gel (14). This experimental design also allows for multivariable
analyses to be performed in one coordinated experiment, whereby statistically
significant abundance changes can be quantitatively measured simultaneously
between several sample types (e.g., different genotypes, drug treatments, or
disease states), with repetition and without the necessity for every pairwise
comparison to be made within a single DIGE gel (15,16) (see Note 8 and
Chapter 17 by Carpentier et al.).
1.2.2. Assessing Intersample Variation
Clinical proteomics is hampered by the significant variation associated with
patient samples. The largest proportion of this variation comes from biological
diversity, but a significant amount may also come from variable collection
Optimizing DIGE Technology 99
Fig. 1. Illustration of DIGE and experimental design using the mixed-sample internal
standard. (A) Representative gel from a six-gel set containing three differentially
labeled samples: Cy2-labeled internal standard, Cy3-labeled sample #1, and Cy5-
labeled sample #2. The individual protein forms all coresolve in this one gel, but these
three independently labeled populations of proteins can be individually imaged using
mutually exclusive excitation/emission properties of the CyDyes. (B) Schematic of
the sample loading matrix indicating gel number, CyDye labeling and three replicates
(indicated as “1, 2, and 3”) of the four conditions being tested (A, B, C, D). Within
the boxed regions representing each labeled sample is depicted a theoretical protein
that is upregulated in condition D. Dotted lines illustrate how the protein signals from
each sample are directly quantified relative to the Cy2 internal standard signal for
that protein without interference from gel-to-gel variation, and how the Cy3:Cy2 and
Cy5:Cy2 intragel ratios are normalized between the six gels. (C) A graphical represen-
tation of the normalized abundance ratios for this theoretical protein change. Adapted
from (10).
and storage of biological samples. It is of vital importance to identify changes
in protein abundance that are disease specific rather than patient or sample
specific.
In order to gain the more robust data sets necessary to be able to draw
accurate conclusions from clinical proteomics studies, it is, therefore, necessary
to collect and store samples using very stringent and closely adhered to
100 Friedman and Lilley
protocols. It is also necessary to assess the biological variation within the
population being tested and also within a single individual. Interindividual
variation has been the focus of several studies (17,18) and determining a
typical diversity within a single patient (i.e., taking longitudinal samples and
assessing variability in protein abundance) and between patients will determine
the minimum number of patient samples required for an experiment. This is
an essential step before embarking on any large-scale and potentially costly
DIGE experiment. Without this type of pretest, the results of underpowered
experiments run the risk of being peppered with false information (both false
positives and negatives).
As with all complex technologies, the DIGE technique itself is subjected
to technical variation, which will be laboratory specific to a greater or lesser
extent. However, the amplitude of this variation is generally outweighed by the
biological variation associated with a typical sample set (19).
1.2.3. Univariate Statistical Analyses
To date, the majority of published quantitative proteomics studies using the
DIGE technology have applied a univariate test, such as a Student’s t-test
or analysis of variance (ANOVA), to identify protein species with significant
changes in expression [(20) and Chapter 17 by Carpentier et al.]. These tests
calculate the probability (p) that the samples being compared are the same and
therefore any apparent change in expression occurs by chance alone. Typically
an expression change is considered significant if the calculated p-value falls
below a prescribed significance threshold, typically 0.05 (whereby 1 in 20 tests
may give a change in expression by chance). For more stringent analyses, a
p-value of 0.01 is often used as the significance threshold.
When employing these tests on DIGE datasets, there are several factors
that must be considered if correct assumptions are to be made from ensuing
analyses. Student’s t-tests and ANOVA assume that the data achieved is
normally distributed and that any variance is homogeneous. The measurement
and correction of systematic bias within DIGE experiments have been the
subject of several studies, which chart methods to optimize normalization of
data sets (21,22,23).
Another important consideration is that of false discovery rate (FDR), which
could arise as a result of statistical tests such as the ones described above.
These tests involve the simultaneous and independent testing of thousands of
spots. The probability of a false positive being recorded for each test is such
that a substantial number of false positives may accumulate. There are several
approaches to determine the FDR and adjust p-scores to compensate for this,
Optimizing DIGE Technology 101
the most widely used to date being the Benjamini and Hochberg method, whose
use in conjunction with DIGE data has been described by Fodor et al.(21).
1.2.4. Multivariate Statistical Analyses
Discovery phase proteomics often produce large lists of proteins that are
identified as changing significantly in the experiment, many of which may well
be false positives. Another approach to overcome these is the application of
additional multivariate statistical analyses to these datasets, which can help to
filter out false positives that result from whole sample outliers (i.e., sample
misclassification and/or poor sample preparation technique). These analyses,
such as principle components analysis (PCA), partial least squares discriminate
analysis, and unsupervised hierarchical clustering (HC) (see Figs. 2 and 3 and
Chapter 16 by Marengo et al.) have recently been applied to DIGE datasets
[(10,24,25,26,27,28,29,30,31,32)]. Raw and normalized data can be exported
from most DIGE software solutions (e.g., DeCyder, Progenesis), and several
multivariate analyses are now part of an extended data analysis (EDA) software
module as part of the DeCyder suite of software tools (GE Healthcare), which
was specifically developed for DIGE analysis (see Subheading 3.6).
These multivariate analyses work essentially by comparing the expression
patterns of all (or a subset of) proteins across all samples, using the variation
of expression patterns to group or cluster individual samples. Technical noise
(poor sample prep, run-to-run variation) and biological noise (normal differ-
ences between samples, especially present in clinical samples) are almost always
–Fe control
PC1
PC2
Δ
fur Heme
Fig. 2. Illustration of the use of principle component analysis. DIGE was used
to analyze changes in Staphylococcus proteins in response to genetic and chemical
alterations affecting iron utilization. Adapted from (24).
102 Friedman and Lilley
Fig. 3. Hierarchical clustering (by average distance correlation) of representative
novel circadian proteins detected by 2D DIGE of soluble protein extracts from mouse
liver. Pale gray represents low levels of protein expression, black represents interme-
diate levels, and dark gray represents high levels of expression. Adapted from (32).
associated with any analytical dataset of this nature, and may well override
any variation that arises due to actual differences related to the biological
questions being tested. Unsupervised clustering of related samples, therefore,
adds additional confidence that a “list of proteins” changing in a DIGE exper-
iment are not arising stochastically (10).
Optimizing DIGE Technology 103
1.3. DIGE in the Clinical Setting
Although the potential for DIGE to address clinical studies is only beginning
to be addressed [for example, see (29,30)], many studies have been published
demonstrating the feasibility and benefit of DIGE/MS using small patient
cohorts for preliminary studies in colon (14), liver (33,34,35), breast (36,37),
esophageal (38,39), and pancreatic cancers (40), as well as other important
clinical studies such as Severe Acute Respiratory Syndrom (SARS) (41). Many
studies also explore the important benefit of procuring samples using laser
capture microdissection (LCM – see Chapters 3, 5, and 9 by Diaz et al., Zhang
et al., and Mustafa et al., respectively) for a highly enriched population of the
cells under study (16,30,42,43,44). These LCM studies necessitate the use of
the saturation chemistry owing to the increased sensitivity but limited multi-
plexing power, and typically require secondary preparative gels with higher
protein loads to enable protein identification by MS.
The study of Suehara et al.(29) represents the utility of a multivariable
DIGE/MS analysis with an extended sample set pertinent for a clinical study.
Eighty soft tissue sarcoma samples comprising seven different histological
backgrounds were analyzed. Using the saturation DIGE fluors, individual
samples were labeled with Cy5 and multiplexed with a pooled-sample internal
standard (labeled in bulk with Cy3) for each DIGE gel. Using high-resolution
2D gel separations and a combination of multivariate statistical tools (support
vector machines, leave-one-out cross-validation, PCA, and HC), these studies
identified a small subset of proteins including tropomyosin and HSP27 that
were able to discriminate between the different classes of tumors. HSP27 in
particular was part of a subclass of discriminating proteins that could distin-
guish between leiomyosarcoma and malignant fibrous histiocytoma (MFH), as
well as correlate with patient survival between low-risk and high-risk groups.
HSP27 has long been associated with prognosis in MFH as well as in other
human carcinomas (45).
2. Materials
This chapter assumes a solid understanding in 2D gel electrophoresis and
will focus on the design and implementation of the DIGE method using the
pooled-sample internal standard methodology and the minimal dye chemistry
for Cy2, Cy3, and Cy5, with notes provided for saturation labeling chemistry.
2.1. Cell Lysis Buffers
1. TNE: 50 mM Tris–HCl pH 7.6, 150 mM NaCl, 2 mM EDTA pH 8.0, 2 mM
DTT, 1% (v/v) NP-40.
104 Friedman and Lilley
2. RIPA buffer: 50 mM Tris–HCl pH 8.0, 150 mM NaCl, 1% NP-40, 0.5% deoxy-
cholic acid, 0.1% SDS.
3. Two-dimentional gel electrophoresis lysis buffer: 7 M urea, 2 M thiourea, 4%
CHAPS, 2 mg/mL DTT, 50 mM Tris–HCl pH 8.0.
4. ASB14 lysis buffer: 7 M urea, 2 M thiourea, 2% amidosulfobetaine 14, 50 mM
Tris–HCl pH 8.0.
NB: depending on the sample, it may also be necessary to add protease
inhibitors and phosphatase inhibitors [sodium pyrophosphate (1 mM), sodium
orthovanadate (1 mM), beta-glycerophosphate (10 mM) and sodium fluoride
(50 mM)] to the chosen lysis buffer (see Subheading 3.1).
2.2. SDS-Polyacrylamide Gel Electrophoresis
1. Immobilized pH gradient (IPG) strips and accompanying ampholyte mixures can
be purchased from a number of commercial vendors. Strip lengths vary from
7 cm to high-resolution 24 cm strips, and pH ranges vary from wide-range (e.g.,
pH 3–11) to high-resolution narrow-range (e.g., pH 5–6) strips.
2. Bind silane working solution (50 mL): 40 mL ethanol, 1 mL acetic acid, 50 μL
bind silane solution (GE Healthcare), 9 mL water (see Note 9).
3. 4× separating gel buffer. 1.5 M Tris-base pH 8.8.
4. 30% acrylamide:bis-acrylamide (37.5:1), N,N,N,N´-tetramethyl-ethylenedia-
mine, and ammonium persulfate.
5. 10× SDS-PAGE running buffer (1 L): 30.25 g Tris-base, 144.13 g glycine, 10 g
SDS (0.1%).
6. Fixing solution for SyproRuby staining (1 L): 100 mL methanol, 70 mL acetic
acid, 830 mL water. SyproRuby stain is available form several commercial
sources and can be substituted by other total protein stains, such as Deep Purple
(GE Healthcare) or Flamingo Pink (BioRad).
7. Two-dimensional equilibration buffer: 6 M urea, 50 mM Tris-base pH 8.8, 30%
glycerol, 2% SDS, trace bromophenol blue.
8. Water-saturated butanol (see Note 10).
9. Dithiolthreitol (store dessicated).
10. Iodoacetamide (store dessicated, keep in the dark).
2.3. DIGE Labeling Materials
1. N,N-dimethyl formamide (DMF) (see Note 11).
2. Labeling (L) buffer: 7 M urea, 2 M thiourea, 4% CHAPS, 30 mM Tris-base
(do not pH, but ensure that pH of final solution is between 8.0 and 9.0), 5 mM
magnesium acetate (see Note 12). Alternatively, 4% CHAPS can be replaced
with 2% ASB14, especially in cases where membrane rich samples are being
utilized.
3. Rehydration (R) buffer: 7 M urea, 2 M thiourea, 4% CHAPS, 2 mg/mL DTT
(13 mM; 2%).
Optimizing DIGE Technology 105
4. Cyanine dyes with NHS-ester chemistry for minimal labeling (Cy2, Cy3, and
Cy5), and with maleimide chemistry for saturation labeling (Cy3 and Cy5) are
available from GE Healthcare as dry solids.
5. Quenching solution (for minimal labeling): 10 mM lysine.
6. Dithiothreitol reduction stock solution: 200 mg/mL DTT.
3. Methods
The DIGE is a powerful technique for quantitative multivariable differential
display proteomics. However, the quality of the data will only be as good as the
quality of the underlying 2D gel electrophoresis technology upon which it is
based. The main focus of this chapter is to provide detailed notes on the DIGE
technology; however, some key considerations to successful high-resolution 2D
gel electrophoresis are also provided. This section describes methods associated
with labeling using minimal CyDyes.
3.1. Sample Preparation
The key to success for any analytical measurement begins with robust sample
preparation. This not only includes the buffers and materials used, but also the
nature of the samples and the way in which they are procured. The addition
of exogenous materials (such as DNAse, RNAse), or allowing for uncontrolled
manipulation of the sample (such as conditions that may lead to proteolysis) can
severely hamper and sometimes completely prevent an analysis. Care should
be taken to ensure against common laboratory contaminants (e.g., mycoplasma
for tissue culture) that if present may be detected as significant changes using
DIGE, either due to the presence in a subset of samples, or by responding to
the experimental perturbation.
1. Prepare protein extracts using any method of preference.
The appropriate amount of protein can be subsequently precipitated prior to
resuspension in the CyDye labeling buffer (see Subheading 3.2). Ensure against
proteolysis and loss of post-translational modifications (e.g., phosphorylation) as
this is of monumental importance.
Care should be taken not to use reagents that will resolve on the 2D gel, such as
soybean trypsin inhibitor. Small molecule inhibitors such as aprotinin, leupeptin,
pepstatinA, antipain, 4 - (20aminoethyl) benzenesulfonyl fluoride hydrochloride
(AEBSF), sodium orthovanadate, okadaic acid, and microcystin, among others,
are far better choices.
2. Lyse cells using standard lysis buffers such as TNE and RIPA buffers, or even
the buffers used for 2D gel electrophoresis.
106 Friedman and Lilley
All of these buffers have the capability of producing high-resolution samples for
2DE. In most cases, the presence of reagents that would otherwise interfere with
CyDye labeling (such as those that contain primary amines) will be removed prior
to labeling by protein precipitation (see Subheading 3.2).
3. Sonicate cells if necessary to improve sample quality.
Sonication improves sample quality by disrupting nucleic acids, which are subse-
quently removed by sample cleanup (see Subheading 3.2) along with phospho-
lipids. Both of these nonproteinaceous ionic components can obliterate the
resolution during IEF.
Short bursts with a tip-sonicator are suggested. It is important to keep the system
chilled, especially in the presence of urea-containing samples that should never
be heated (see Note 12).
4. Determine the protein concentration of the sample using a system that is
compatible for the buffer that the proteins are extracted in.
CHAPS and thiourea in the buffers used for DIGE, although adequately
chaotropic, interfere with either the Bradford or bicinchoninic acid assays, making
the data inaccurate and unreliable. In these cases, aliquots should be precipitated
prior to quantification in a suitable buffer, or the use of a detergent compatible
assay should be utilized.
5. Aim to use a protein concentration between 1 and 10 mg/mL.
Too dilute and it will be difficult to quantitatively recover proteins following
precipitation cleanup (see Subheading 3.2); too concentrated and it will be
difficult to accurately dispense the appropriate volume for the experiment.
Freeze/thawing should also be kept to a minimum; freezing samples in 1 mL
aliquots or less will usually suffice.
3.2. Sample Cleanup
The desired amount of sample to be used in the experiment should be
precipitated prior to labeling. This removes both nonproteinaceous ions from
the sample (e.g., nucleic acids, phospholipids) that can interfere with IEF,
as well as transfers the proteins into a labeling buffer optimized for CyDye
labeling and subsequent IEF. Determine how much total protein will be on
each gel, and precipitate ½ of that amount for each sample to be run on that
gel. This is straightforward for a two-component separation, but also works out
for the multigel experiments where 1/3 of the total protein amount on each gel
comes from the pooled-sample internal standard (see Table 1.) Precipitate only
what is needed for each sample for the experiment; too much material may
create pellets that are difficult to resolubilize completely.
Table 1
Experimental Design for CyDye Labeling Using a Pooled-Sample Internal Standard
Samples
Gel 1 Gel 2 Gel 3 Pool
Control-1 Treated-1 Control-2 Treated-2 Control-3 Treated-3
Precipitated amount 150 μg 150 μg 150 μg 150 μg 150 μg 150 μg
L-buffer 24 μL 24 μL 24 μL 24 μL 24 μL 24 μL
Aliquot 16 μL 16 μL 16 μL 16 μL 16 μL 16 μL 8 μL (×6)
Cy2 6μL
Cy3 2 μL 2 μL 2 μL
Cy5 2 μL 2 μL 2 μL
30 min on ice in the dark
Lysine (quench) 2 μL 2 μL 2 μL 2 μL 2 μL 2 μL 6 μL
10 min on ice in the dark
Total volume 20 μL 20 μL 20 μL 20 μL 20 μL 20 μL 60 μL
For each gel, combine the quenched Cy3-and Cy5-labeled samples and add 1/3 of the
quenched Cy2-labeled pooled mixture
20+20+20μL 20+20+20μL 20+20+20μL
2× R-buffer 60 μL 60 μL 60 μL
Total 120 μL 120 μL 120 μL
R-buffer to Vfto Vfto Vf
This table illustrates a typical DIGE labeling experiment, as described in Subheadings 3.2 and 3.3.
107
108 Friedman and Lilley
Many precipitation methods are available, the following is a MeOH/CHCl3
protocol that works well for DIGE, and can be easily performed in 1.5 mL
tubes [adapted from (46)]:
1. Bring up predetermined amount of protein extract to 100 μL with water.
2. Add 300 μL (3-volumes) water.
3. Add 400 μL (4-volumes) methanol.
4. Add 100 μL (1 volume) chloroform.
5. Vortex vigorously and centrifuge; the protein precipitate should appear at the
interface.
6. Remove the water/MeOH mix on top of the interface, being careful not to
disturb the interface. Often the precipitated proteins do not make a visibly white
interface, and care should be taken not to disturb the interface.
7. Add another 400 μL methanol to wash the precipitate.
8. Vortex vigorously and centrifuge; the protein precipitate should now pellet to
the bottom of the tube.
9. Remove the supernatant and briefly dry the pellets in a vacuum centrifuge.
10. Resuspend the pellets in a suitable amount of CyDye labeling buffer (L-buffer,
see Table 1).
An alternative widely used precipitation method is as follows:
1. Add 5 volumes of cold 0.1 M ammonium acetate in methanol.
2. Leave at –20°C for 12 h or overnight.
3. Centrifuge at ∼3000 rpm (1400×g) for 10 min at 4°C and remove the supernatant.
4. A pellet of protein should be visible at this stage.
5. To wash the pellet, add 80% 0.1 M ammonium acetate in methanol and mix to
resuspend the protein.
6. Centrifuge at 3000 rpm (1400×g) for ten min at 4°C and remove the supernatant.
7. To dehydrate the pellet add 80% acetone and resuspend the pellet by mixing.
8. Centrifuge at 3000 rpm (1400×g) for ten min at 4°C and remove the supernatant.
9. Dry pellet for 15 min by leaving open tube in a laminar flow cabinet.
3.3. DIGE Experimental Design
1. Start with a preliminary gel. All experiments should start with a preliminary gel
on representative samples to ensure equivocal protein amounts between samples,
and that the highest resolution and sensitivity are obtained before embarking on a
multigel DIGE experiment. (see Notes 13 and 6). The preliminary gel will also show
any problems with the sample preparation that may be corrected by adjusting the
procurement methods (see Subheading 3.1). This step can also be used to optimize
the maximal amount of protein can be loaded without adversely affecting resolution.
The preliminary gel needs only to test one or two of the samples of a much
larger experiment. This gel can simply be stained with a total protein stain (e.g.,
Sypro Ruby or Deep Purple) to visually inspect the resolution and sensitivity.
Optimizing DIGE Technology 109
Alternatively, the gel can contain two different samples prelabeled with Cy3
and Cy5 and coresolved. (see Note 14).
2. Choose a suitable pH gradient for the IEF. Precast IEF strips are commercially
available from several vendors. The widest length is currently 24 cm, providing
the highest resolving power for a given pH range. Medium-range IEF gradients
(e.g., pH 4–7) offer the best trade-off between overall resolution and sensitivity.
Subsequent experiments can then be designed to resolve proteins in the basic
range (pH 7–11) and in narrow pI ranges with commensurate increases in protein
loading to gain access to the lower abundant proteins in a given sample (see
Note 5). In this way a more comprehensive picture of the proteomes under study
can be obtained.
3. Incorporate a pooled-sample mixture internal standard on every DIGE gel in
a coordinated experiment. This internal standard, usually labeled with Cy2, is
composed of an equal aliquot of every sample in the entire experiment, and
therefore represents every protein present across all samples in an experiment. The
use of this pooled-sample internal standard on every DIGE gel in a coordinated
experiment allows for the facile comparison of independent sample replicates
with increased statistical confidence. This experimental design also enables the
simultaneous quantitative comparison between multiple variables in a coordinated
experiment (Fig. 1).
4. Plan out which samples will be labeled with which dyes ahead of time. For
minimal dye labeling chemistry (see Subheading 3.4), each gel will contain two
individual samples labeled with either Cy3 or Cy5, and an equal amount of the
pooled-sample internal standard. The example outlined in Table 1 is for a two-
component comparison repeated in triplicate, with 300 μg total protein loaded
onto each of three gels. In this case, 150 μg of each sample should be precipitated
(see Subheading 3.2), resuspended in L-buffer and then split 2:1. Two-thirds
of each sample (100 μg) will be individually labeled with either Cy3 or Cy5.
The remaining 1/3 of each sample will be pooled together and labeled with Cy2
to serve as an internal standard. By following this, there will be enough of the
Cy2-labeled internal standard to have an equal amount as the Cy3 or Cy5 samples
loaded onto each gel. (see Note 15).
3.4. CyDye Labeling
All steps are performed on ice. The following protocol is for sample loading
via rehydration of IPG strips, and assumes incorporation of a pooled-sample
internal standard to coordinate many samples across multiple DIGE gels simul-
taneously. The steps are summarized in Table 1 (see Note 16).
1. Resuspend precipitated sample in 24 μL labeling (L) buffer. Remove 8 μL (1/3
of sample) and place into a new tube that will contain the pooled-sample internal
standard (8 μL from all of the other individual samples will be pooled into this
tube) (see Note 17).
110 Friedman and Lilley
2. CyDyes are purchased as dry solids and should be reconstituted to 10× stock
solutions (1 nmol/μL) in fresh DMF. Dilute stock solutions of CyDyes 1:10 in
fresh DMF to a final working concentration of 100 pmol/μL (see Note 11).
3. Label each sample (50–250 μg) with 2–4 μL (200–400 pmol) of either Cy3 or Cy5
working dilution for 30 min on ice in the dark. Label the pooled-sample mixture
with 2–4 μL (200–400 pmol) of Cy2 working dilution for every equivalent amount
of sample present in the pooled standard as compared with the individually labeled
samples. That is, if 100 μg of each sample is labeled with 200 pmol of Cy3 or
Cy5, then 50 μg of each of these samples is present in the pooled standard, and
200 pmol of Cy2 is used for every 100 μg of pooled standard. (see Table 1 and
Note 18).
4. Quench reactions with 2 μL of 10 mM lysine for 10 min on ice in the dark.
5. For each gel, combine the quenched Cy3- and Cy5-labeled samples and add 1/3
of the quenched Cy2-labeled pooled mixture.
6. To each tripartite mixture, add an equal volume of 2× R-buffer and incubate on
ice for 10 min. 2× R-buffer is R-buffer supplemented with an additional 2 mg/mL
DTT using the 200 mg/mL DTT stock solution. DTT is omitted from the L-buffer
to prevent unfavorable interaction with the CyDyes. Adding an equal volume of
2× R-buffer to the quenched reactions provides the reducing agents to the total
reaction volume at a 1× final concentration.
7. Add R-buffer (1× DTT concentration) to a final volume suggested by the manufac-
turer for the given IPG strip length (e.g., 450 μL for 24 cm strips). Add the
appropriate volume of IPG buffer ampholines to 0.5% final (v/v) for IEF. Proceed
with rehydration of dehydrated IPG strips for >16 h and proceed with IEF (see
Subheading 3.5.3 and Note 19).
3.5. 2D Gel Electrophoresis and Poststaining
As a result of the minimal labeling, quantification with the CyDyes is carried
out on only 2–5% of the proteins that are labeled, and the labeled portion of
the protein may migrate at a higher apparent molecular mass than the majority
of the unlabeled protein due to the added mass and hydrophobicity of the dyes
(exacerbated in lower Mrspecies). To ensure that the maximum amount of
protein is excised for subsequent in-gel digestion and MS, minimally labeled
2D DIGE gels are poststained with a total protein stain such as SyproRuby or
Deep Purple. Accurate excision is also ensured by preferentially affixing the
second dimension gel to a presilanized glass plate during gel casting so that
the gel dimensions do not change during the analysis (see Notes 20 and 21).
These methods assume the use of the Ettan 2D electrophoresis system (GE
Healthcare), but are easily adaptable to other commercially available systems.
It also assumes usage of high-resolution 24 cm × 20 cm gels.
1. Special gels for second dimension SDS-PAGE. Using low-fluorescence glass
plates, pretreat one plate for each gel with 3–5 mL bind silane working solution,
Optimizing DIGE Technology 111
carefully wiping the entire surface of the plate with a lint-free wipe. Leave treated
plates covered with lint-free wipes for several hours to allow for sufficient out-
gassing of fumes (that may contain bind silane) before assembling gel plates and
casting of second dimensional SDS-PAGE gels (see Note 22).
2. Assemble plates and pour 12% homogeneous SDS-PAGE gel(s) using the appro-
priate amount of 30% stock acrylamide and 4× separating gel buffer for the
volumes needed for the number of gels being poured (see Note 23). Overlay the
gels with water-saturated butanol for several hours to provide a straight and level
surface to place the focused IPG strip (see Note 10).
3. Perform IEF using an IPGphor II IEF unit (GE Healthcare) of the combined
tripartite-labeled samples, brought up to final volume with 1× R-buffer and
passively rehydrated into IPG strips for >16 h (see Subheading 3.4.7)(see
Note 24).
4. Equilibrate the focused IPG strips into the second dimensional equilibration buffer.
During this step, the cysteine sulfhydryls in the focused proteins are reduced
and carbamidomethylated by supplementing the equilibration buffer with 1%
DTT for 20 min at room temperature, followed by 2.5% iodoacetamide in fresh
equilibration buffer for an additional 20 min room temperature incubation (see
Note 25).
5. Place equilibrated IPG strip on top of the SDS-PAGE gels that were precast with
low-fluorescence glass plates. Use a thin card or ruler to carefully tamp down the
IPG strip to the SDS-PAGE gel, removing air bubbles at the interface (see Notes
26 and 27).
6. Perform second dimensional SDS-PAGE at constant wattage, using 1 W/gel
for at least 1 h prior to ramping up to <20 W/gel (see Note 28).
7. CyDye images are acquired using a fluorescence imager, such as the Typhoon
9400 series (GE Healthcare) equipped with lasers and filters that are compatible
with the emission/excitation spectra of the dyes. Imaging is performed through
the glass plates using the intact gel cassette (see Note 29).
8. After imaging the gels, carefully remove the plate that was untreated with bind
silane. The gel will remain stuck to the treated plate and can be stained with
an appropriate total protein dye (such as SyproRuby Deep Purple or Flamingo
Pink) “open-faced” in the fixation/staining solutions. For SyproRuby, fix gels for
at least 2 h with fixation solution sufficient to completely cover the gel. Longer
fixations are possible without adversely affecting subsequent MS. After removing
the fixation solution, stain gels overnight in SyproRuby and acquire images using
a fluorescence imager (see Notes 21 and 30).
3.6. DIGE Analysis
3.6.1. Software Algorithms
Many bioinformatics tools are commercially available for the comparison
of multiple 2D gel-separated protein spot patterns. Some free internet-based
utilities (e.g., www.lecb.ncifcrf.gov/flicker/) provide simple alternation between
112 Friedman and Lilley
two spot patterns, whereas most of the commercial products contain proprietary
algorithms for protein spot detection, intergel matching, protein spot quantifi-
cation, and even utilities for building web-based tools for data dissemination.
Many include the ability to average replicate patterns into a single virtual
pattern to be used in a comparative study. They are all designed to compare
multiple spot patterns and quantify abundance changes for individual proteins
between experimental conditions.
Several software packages allow for the analysis of DIGE data. The DeCyder
suite of software tools was specifically developed to support the DIGE platform
when this technology was first marketed by GE Healthcare and is therefore used
as an example here. The differential in-gel analysis (D I A) module of DeCyder
is used for direct quantification of protein spot volume ratios between the triply
codetected signals emanating from each resolved protein, and can be used for
the simplest form of a DIGE experiment for pairwise comparisons with N=1.
The more advanced DIGE experiments that use the internal standard to cross-
compare replicate samples from pairwise and multivariable analyses (N>3)
are handled by the biological variation analysis (BVA) module of DeCyder. In
a BVA experiment, the signals emanating from the internal standard are used
both for direct quantification within each DIGE gel in a coordinated set (using
Differential In-gel Analysis (DIA) module), as well as for normalization and
protein spot pattern matching between gels (see Note 31). This allows for the
calculation of Student’s t-test and ANOVA statistics for individual abundance
changes (see Subheading 3.6.2, and Table 2). BVA is also used to match
patterns between SyproRuby- and CyDye-stained images to facilitate protein
excision for subsequent MS (see Notes 20,21, and 30).
3.6.2. Experimental Design and Statistical Confidence
In the simplest form of a DIGE experiment, two or three samples are
separately labeled with one of the three dyes and separated in the same gel for
direct pairwise comparisons. In this case, the software first normalizes the entire
signal for each CyDye channel and then calculates the protein spot volume
ratio for each protein pair. A normal distribution is modeled over the actual
distribution of protein pair volume ratios, and two standard deviations of the
mean of this normal distribution represent the 95th percent confidence level for
significant abundance changes.
This N= 1 type of experiment has limited statistical power, since the 95th
percentile confidence interval is determined based on the overall distribution of
changes within the population (see Note 32). Many more changes in abundance
of much lesser magnitude can be detected with much greater statistical confi-
dence (Student’s t-test and ANOVA, Table 2) by incorporating independent
Optimizing DIGE Technology 113
Table 2
Statistical Applications of DeCyder Biological Variation Analysis and Extended
Data Analysis (EDA) Modules
Average ratio Calculated for each protein spot feature between two groups
or experimental conditions. Derived from the log standardized
protein abundance changes that were directly quantified
within each DIGE gel relative to the internal standard for the
protein spot feature.
Student’s t-test Univariate test of statistical significance for an abundance
change between two groups or experimental conditions.
p-values reflect the probability that the observed change has
occurred due to stochastic chance alone. With DIGE, p-values
of <0.01 are often observed, assumes normal distributions
of protein abundance, can be performed either unpaired or
paired.
One-way ANOVA Tests for differences in standardized abundance of a given
protein across all groups of a multicomponent analysis.
Indicates that one group is significantly different from another
in the group.
Two-way ANOVA Tests for differences in standardized abundance of a given
protein between multiple groups with the same condition,
where multiple conditions are analyzed.
Principle component
analysis (EDA only)
Reduces the dimension of the variables in a multidimensional
space. The first principal component (PC1) divides the dataset
along an axis describing the most variance in a system, with
the orthogonal second component (PC2) accounting for the
second greatest source of variation.
Hierarchical
clustering (EDA only)
Compares groups based on similarity of the collective
expression patterns of individual proteins, often displayed in
an expression matrix (heatmap). Similarity between groups
is proportional to the lateral distance depicted as a branched
dendrogram.
K-means (EDA only) Used to classify proteins into a predefined number of bins
based on similarity.
Self organizing maps
(EDA only)
Similar to K-means, but also clusters nearest neighbors (based
on expression patterns) in a two-dimensional map.
Gene shaving (EDA
only)
Used to identify groups of proteins that have similar
expression profiles. Unlike K-means, proteins can belong to
more than one group provided there is high coherence within
each group.
Discriminant analysis
(EDA only)
Identifies proteins that can discriminate between groups based
on a variety of classifier schemes, including cross-validation,
feature selection, partial least squares, K-nearest neighbors.
114 Friedman and Lilley
replicate samples into the experiment (see Note 33). The number of replicates
required in a study depends on the amount of variation in the system being
investigated. Increasing the number of replicates will increase confidence in
smaller changes in expression. The number of gel replicates that are needed for
the experiment to have sufficient sensitivity to detect expression changes can
be determined using power calculations (for example see (19)).
With replicate samples, the Student’s t-test and ANOVA statistics are
measuring the significance of the variation of a specific protein change,
independent of the overall distribution of abundance changes in the population.
Incorporating replicate samples into the experimental design also controls for
unexpected variation introduced into the samples during sample preparation.
This design not only allows for the identification of abundance changes that
are consistent across multiple replicates of an experiment, but can also identify
significant abundance changes that would not have been identified even if the
analyses were performed using Cy3- and Cy5-labeled samples on the same
gels, but without the pooled-sample internal standard to coordinate them (14).
3.6.3. Multivariate Statistical Analysis
Univariate analyses such as the Student’s t-test and ANOVA have tradi-
tionally been used in DIGE experiments to provide a list of statistically signif-
icant changes in protein abundance. The application of multivariate statis-
tical analyses (as outlined in Subheading 1.2.4) allow for the assessment of
changes on a global scale, and can bring added insight to the usual “list of
proteins” generated. Most software packages allow for the export of raw and
normalized protein spot volumes to allow for these additional statistical tests
and data manipulations; in addition, the DeCyder suite of software tools now
provides an Extended Data Analysis (EDA) module, that includes many of these
tools (Table 2). These tools are now becoming more evident in recent DIGE
publications (10,24,28,29,30,32,52). Although these multivariate analyses are
especially beneficial when analyzing a DIGE experiment that contains three or
more conditions, they can also useful in two-condition comparisons to detect
sample outliers, fouled samples or even poor experimental design.
Figure 2 illustrates an example of PCA applied to a DIGE dataset comprised
of four experimental conditions each measured in quadruplicate. PCA simplifies
multidimensional datasets by reducing the variation down to the two or three
most significant sources of variation. In this example, the first principle
component (PC1) accounts for 62.3% of the variation amongst 156 proteins
of interest, with the second principle component (PC2) accounting for an
additional 12.5% of the variation. Each sample datapoint describes the collective
expression profile for the subset of 156 proteins, and PC1 and PC2 orthogonally
Optimizing DIGE Technology 115
divide the samples into quadrants based on these two largest sources of variation
within DIGE dataset. In this case, 75% of the variance between these proteins
clusters the samples into the proper categories (adapted from (24)).
Figure 3 is taken from a 2D DIGE study, which determined the change in
protein abundance in mouse liver over a 24 h period. In this, study proteins
were harvested from groups of mice on a second cycle after transfer from
synchronized (12 h light:12 h dim red light) to free running conditions (constant
dim red light). Proteins were extracted from each liver and pooled from six
mice per 4-h time point. HC (by average distance correlation) was used to
investigate the expression of 49 novel circadian proteins. This gave a range
of phase groups with 10 proteins peaking during the subjective day and 39
proteins distributed between two clusters, which were most abundant during
the subjective night (adapted from (32)).
Finally, additional information may be gleaned by mapping proteins found
to be changing by DIGE to existing biological pathways and networks.
Many software solutions and services are becoming available for this type
of extended analysis (e.g., Kegg pathways, Ingenuity pathways analysis,
WebGestalt, DeCyder EDA). Although additional validation is necessary to
establish biological significance, the mapping of members of a “list of proteins”
to established pathways and networks can provide validating support for the
proteins observed by DIGE alone. In some cases, it can also indicate potential
proteins associated with the biological question that were not accessible in the
DIGE analysis. For example, Friedman et al.(10) recently reported the use of
network/pathway mapping for proteins found by DIGE/MS in MCF10A cells
overexpressing the HER2 receptor after treatment with TGF-. The majority of
proteins identified with DIGE/MS mapped to a network of pathways involving
TGF-as a major hub, but also included an intercalating pathway involving p53
that effected many proteins that were independently identified in the DIGE/MS
experiments. This insight linking new players to those identified with DIGE/MS
led to the further investigation of a direct role for p53 in the expression of the
tumor suppressor maspin (53).
4. Notes
1. 2DE has traditionally been a popular method for differential display proteomics
on a global scale, but until recently, these strategies lacked the ability to directly
quantify abundance changes in the same fashion as in stable isotope LC/MS-
based strategies (2,3,4). This has been mainly due to the inability to directly
correlate migration patterns and protein staining between gel separations (gel-
to-gel variation). Stable isotopes have been used in gel-based proteomics as
well, whereby different proteomes have been separately labeled with different
stable isotopes (e.g., growing cells using 14N vs. 15 N-labeled medium) prior to
116 Friedman and Lilley
mixing and running together through the same 2DE separation (5). In this case,
abundance changes can be monitored during the mass spectrometry (MS) stage
on individual proteins, but requires the in-gel digestion and MS on every protein
present to discover the subset of proteins that is changing.
2. Both hydrophobicity and molecular weight influence how proteins migrate
during SDS-PAGE, yielding information on apparent molecular mass.
3. In comparison, commonly used silver or colloidal coomassie blue (ca. 5–10 ng
sensitivity) stains typically exhibit a dynamic range of less than two orders of
magnitude (8,9). The CyDye labeling system is compatible with the downstream
processing commonly used to identify proteins via MS and database interro-
gation, which involves the generation of tryptic peptides within excised gel
plugs. Trypsin cleaves the peptide bonds the C-terminal side of lysine and
arginine residues, but peptide generation is mostly unhindered as so few lysine
residues are modified by dye labeling.
4. DIGE experiments can still be performed using the internal standard method-
ology with only two CyDyes, but twice as many gels are required to analyze the
same number of samples compared with the three-dye minimal labeling scheme.
With saturation labeling, one dye is used to label the internal standard, and the
other is used to label individual samples. A dye-swap scheme is not necessary
in this case because the individual samples are always labeled with the same
CyDye.
5. The use of hydroxyethyl disulfide (commercially available as “DeStreak
reagent”), combined with anodic cup loading, should be used for enhanced
resolution for IEF above pH 8 (11).
6. Running every DIGE gel with the maximal amount of protein (without adversely
effecting first dimension resolution) not only enables detection of lower
abundance proteins, but also provides more material for subsequent protein
identification using MS. This makes every gel in a coordinated DIGE exper-
iment a “pick-able” gel, without the need to run subsequent preparative gels
with increased protein load that then have to be carefully matched to a lower
abundant, analytical gel. When combined with narrow range IEF, maximizing
the protein amount also allows interrogation of the lower abundant proteins in
a sample.
7. If one sample within a study has very skewed protein distributions compared
with others, then many of the “novel proteins” within this sample will effectively
be diluted out in the pool. Such a sample outlier can be easily identified using
the multivariate statistical analyses described.
8. Repetition not only enables the identification of subtle differences with statistical
confidence, it is also vital to control for nonbiological variation. In most cases
biological variation will outweigh technical variation, therefore, only biological
replicates are necessary. Thus it is important that each replicate sample is derived
from an independent experiment, ideally performed on different occasions as
perhaps using different batches of medium. The independent samples can then be
Optimizing DIGE Technology 117
analyzed coordinately using the pooled-sample internal standard methodology.
See Table 1 for an example of this design.
9. All solutions should be prepared using water that has a resistivity of 18.2 M-
cm; this is referred to as “water” throughout the text.
10. Mix equal parts of butanol and water and shake vigorously. Let the two phases
separate overnight, and use the butanol phase for overlay. Butanol that is not
completely water saturated can extract water from the top of the gel. A more
recent improvement is to use a 0.1% SDS solution in a conventional spray bottle,
used to carefully spray a fine mist over the top of the gels to thoroughly cover
the top of the gel (the gel/overlay interface will not be as obvious).
11. DMF can degrade, producing amines, which can react with the NHS-ester
CyDyes. DMF stocks should be kept fresh (<3 months) and anhydrous to ensure
optimal labeling conditions.
12. For buffers that contain urea, care should be taken to ensure the urea is fresh
and free of the natural break down product isocyanate, which will carbamylate
free amines and thereby neutralizing the protonatable epsilon-amine groups of
lysine residues. This is problematic for several reasons, the foremost being the
fact that this gives rise to artificial charged train isoforms in the first dimension
IEF. Heating samples above 37°C should also be avoided, as this facilitates
the conversion to isocyanate. Any buffer component that contains a primary
amine, such as pharmalytes, ampholytes, HEPES buffers, etc. should be avoided
as these components may react with the CyDyes, thus reducing their affective
concentration.
13. For example, 500 μg of material may be loaded onto a pH 4–7 24 cm IPG
strip, but due to the overall distribution of proteins in the sample, as well as a
sometimes unusually high abundance of a subset of proteins, may result in much
less material actually resolving between the electrodes. A good rule to follow is
to load the desired amount based on the protein concentrations, and then adjust
the load by eye as necessary.
14. This is DIGE in its most simplistic form, and can show differences between
the samples without interference from gel-to-gel variation, but provides limited
statistical power to help distinguish true biological variation from background
such as artificial noise introduced during sample preparation.
15. Employing a dye-swapping approach will control for any dye-specific effect that
may result from preferential labeling or different fluorescence characteristics of
acrylamide at the different wavelengths of excitation for Cy2, Cy3, and Cy5,
especially at low protein spot volumes. This is easily incorporated into any DIGE
analysis where repetitive samples are used (along with the internal standard to
compare across multiple DIGE gels).
16. For saturation chemistry, general methods and considerations are the same as for
the minimal chemistry, but there are several unique features to also consider for
the saturation chemistry. First, careful optimization of the labeling conditions
must be carried out for each new sample set to ensure complete reduction of
cysteine residues. Insufficient labeling will lead to multiple spots in the second
118 Friedman and Lilley
dimension due to MW and hydrophobicity shifts. Overlabeling results in side
reactions with the epsilon-amine groups of lysine side chains, but since the
maleimide dyes do not carry compensatory charge, this results in the overall
loss of a charge, which creates a series of isoelectric forms in the first dimension
(“charge trains”). Labeling buffer should not contain any components with free
thiols, as these will react with the satCyDyes.
17. L-buffer volume can be increased if necessary for complete resolubilization,
although 100–250 μg or more should resolubilize readily in this volume. The
volume of labeling buffer used for resolubilization should not exceed 40 μL per
sample when using cup loading for sample entry to ensure that the final volumes
will not exceed the capacity of the cup loading (ca. 100–150 μL).
18. These methods are provided assuming that all gels to be run will be used both
for analytical (quantification) as well as preparative (providing material for
subsequent MS) purposes. Current recommendations from the manufacturer are
to label 50 μg of sample with 400 pmol CyDye. Sufficient amount of unlabelled
sample can be added to the quenched reactions to achieve final protein amounts
to facilitate subsequent MS. Alternatively, many have found that the ratios can
be adjusted to label increasing amounts of sample (up to 200 μg with 200 pmol
dye) without adversely affecting the overall labeling reaction (presented here).
19. If samples are to be introduced using anodic cup loading, simply bring this
mixture up to 100 μL in R-buffer and proceed with cup loading. R-buffer can
always be supplemented with additional DTT using the 200 mg/mL DTT stock
solution. In the presence of Destreak reagent for focusing in pH ranges above pH
8, the addition of equal volume 1× R-buffer should provide sufficient amount
of DTT without interfering with the Destreak reagent.
20. Comparison of minimally labeled protein 2D maps with unlabeled protein maps
is generally not a problem, as the addition of only one dye molecule does not
generally prevent the facile matching of small alterations in protein mobility
between the 2- and 5%-labeled protein and the remaining unlabeled protein that
will provide enough material for MS.
21. Poststaining is not necessary with saturation DIGE, since an unlabeled population
with potentially different migration characteristics will not exist.
22. This treatment binds the gel to one of the glass plates and therefore prevents
shrinking/swelling during the poststaining and protein excision processes,
thereby facilitating accurate robotic protein excision. Nothing should be placed
on top of wipes that are covering bind silane-treated plates, as this may leave
impressions that are detected during the scanning phase. Assembly and casting
too soon may create a binding surface on the opposite glass plate, preventing
the gel to be subsequently poststained and picked. Automated protein excision
can be facilitated for certain systems by placing fluorescent alignment reference
targets on the plate, which can be performed at this stage.
23. A stacking gel is not required for 2D gel electrophoresis, as the proteins are
effectively “stacked” to the height of the IPG strip. SDS is also not essential in the
separating gel, as the SDS associated with the proteins during the equilibration
Optimizing DIGE Technology 119
step, and present in the running buffer, is sufficient (although many traditionally
use it in the separating gel). Using 2× concentration running buffer in the upper
buffer chamber can produce higher quality separations in some circumstances.
24. Samples of similar nature should always be focused simultaneously for optimal
reproducibility. Focusing programs vary for some pH gradients. A typical
program for many ranges is 500 V for 500 V-h, stepping to 1000 V for 1000 V-h,
followed by a final step to 8000 V until >50 V-h has been reached. Check
recommendations from specific vendors.
25. Volume of equilibration buffer should be large to ensure sufficient removal of
ampholines and other components of the first dimensional run.
26. Carefully wash out any remaining liquid on top of the SDS-PAGE gel. Prewet
the IPG strip with 1× running buffer and place the strip between the gel plates
with the plastic backing adhering to the inside surface of one of the glass plates.
The prewetted running buffer will facilitate the manipulation of the IPG strip
down the inside surface of the plate and on top of the SDS-PAGE gel.
27. An agarose overlay, used by many protocols, is not absolutely necessary to
ensure proper contact between the IPG strip and the second dimensional SDS-
PAGE gel. Using a thin card or ruler to carefully tamp down the IPG strip to
the gel is usually sufficient and removes the added problems associated with the
overlay, such as trapped air bubbles in the solidified agarose.
28. Running gels at less than 1 W/gel can improve resolution in the high molecular
weight regions of the second dimension gel. Use wattage appropriate for the
second dimensional unit being used. Many different gel units can accommodate
increased power by compensating for the increased heat.
29. Absorption/emission maxima in DMF are 491/506 for Cy2, 553/572 for Cy3,
and 648/669 for Cy5; although care must be taken to scan in regions of each
spectrum that do not contain absorbance or emission in the other spectra, which
may mean using a nonmaximal region of a given spectrum.
30. Comparison of the 2D spot maps between saturation-labeled samples and
minimal labeled or unlabeled samples is impossible, as proteins containing
multiple cysteine residues may appear as significantly larger Mrspecies when
labeled with the saturation dyes, which of course cannot be predicted without
first knowing the protein identity.
31. Almost all software packages for 2D electrophoresis involve matching of protein
spot patterns between gels. For DeCyder, it is used in the BVA module to match
the quantitative data obtained from the triply coresolved protein signals from
each gel in the DIA module (where gel-to-gel variation does not come into
play). Manual verification of the matching is almost always required with any
software package.
32. There are many “all-or-none” type of experiments where the single gel
comparison may be valid, and subtle changes are not expected. Nevertheless,
using independent replicates and the pooled-sample internal standard method-
ology is still needed to control for nonbiological sample preparation error.
120 Friedman and Lilley
33. The multigel approach allows many data points to be collected for each group
to be compared. Spots of interest can be selected by looking for significant
change across the groups. Student’s t-test and ANOVA probability scores (p)
indicate the probability that the observed change occurred due to stochastic,
random events (null hypothesis). Probability values <0.05 are traditionally used
to determine a statistically significant difference from the null hypothesis. As
this represents 50 potential false positives for 1000 resolved proteins, confidence
intervals within the 99th percentile (p< 0.01) are arguably more valid, and can
be attained using DIGE (10,14,24,47,48,49,50,51).
References
1. Petricoin, E., Wulfkuhle, J., Espina, V. and Liotta, L.A. (2004) Clinical proteomics:
revolutionizing disease detection and patient tailoring therapy. J Proteome Res
3(2):209–17.
2. Gygi, S.P., Rist, B., Gerber, S.A., Turecek, F., Gelb, M.H. and Aebersold, R. (1999)
Quantitative analysis of complex protein mixtures using isotope-coded affinity tags.
Nat Biotechnol 17(10):994–99.
3. Mason, D.E. and Liebler, D.C. (2003) Quantitative analysis of modified proteins by
LC-MS/MS of peptides labeled with phenyl isocyanate. J Proteome Res 2(3):265–
72.
4. Ross, P.L., Huang, Y.N., Marchese, J.N., Williamson, B., Parker, K., Hattan, S.,
Khainovski, N., Pillai, S., Dey, S., Daniels, S., Purkayastha, S., Juhasz, P.,
Martin, S., Bartlet-Jones, M., He, F., Jacobson, A. and Pappin, D.J. (2004) Multi-
plexed protein quantitation in Saccharomyces cerevisiae using amine-reactive
isobaric tagging reagents. Mol Cell Proteomics 3(12):1154–69. Epub 2004 Sep 22.
5. Vogt, J.A., Schroer, K., Holzer, K., Hunzinger, C., Klemm, M., Biefang-Arndt, K.,
Schillo, S., Cahill, M.A., Schrattenholz, A., Matthies, H. and Stegmann, W. (2003)
Proteinabundancequantificationinembryonicstemcellsusingincompletemetabolic
labelling with 15N amino acids,matrix-assisted laser desorption/ionisation time-
of-flight mass spectrometry,and analysis of relative isotopologue abundances of
peptides. Rapid Commun Mass Spectrom 17(12):1273–82.
6. Gorg, A., Obermaier, C., Boguth, G., Harder, A., Scheibe, B., Wildgruber, R.
and Weiss, W. (2000) The current state of two-dimensional electrophoresis with
immobilized pH gradients. Electrophoresis 21(6):1037–53.
7. Gorg, A., Postel, W., Domscheit, A. and Gunther, S. (1988) Two-dimensional
electrophoresis with immobilized pH gradients of leaf proteins from barley
(Hordeum vulgare): method,reproducibility and genetic aspects. Electrophoresis
9(11):681–92.
8. Tonge, R., Shaw, J., Middleton, B., Rowlinson, R., Rayner, S., Young, J.,
Pognan, F., Hawkins, E., Currie, I. and Davison, M. (2001) Validation and devel-
opment of fluorescence two-dimensional differential gel electrophoresis proteomics
technology. Proteomics 1(3):377–96.
Optimizing DIGE Technology 121
9. Lilley, K.S., Razzaq, A. and Dupree, P. (2002) Two-dimensional gel
electrophoresis: recent advances in sample preparation,detection and quantitation.
Curr Opin Chem Biol 6(1):46–50.
10. Friedman, D.B., Wang, S.E., Whitwell, C.W., Caprioli, R.M. and Arteaga, C.L.
(2007) Multi-variable difference gel electrophoresis and mass spectrometry: A
case study on TGF-beta and ErbB2 signaling. Mol Cell Proteomics 6:150–69.
11. Olsson, I., Larsson, K., Palmgren, R. and Bjellqvist, B. (2002) Organic disulfides
as a means to generate streak-free two-dimensional maps with narrow range basic
immobilized pH gradient strips as first dimension. Proteomics 2(11):1630–32.
12. Wolters, D.A., Washburn, M.P. and Yates, J.R. 3rd (2001) An automated multi-
dimensional protein identification technology for shotgun proteomics. Anal Chem
73(23):5683–90.
13. Alban, A., David, S.O., Bjorkesten, L., Andersson, C., Sloge, E., Lewis, S. and
Currie, I. (2003) A novel experimental design for comparative two-dimensional gel
analysis: two-dimensional difference gel electrophoresis incorporating a pooled
internal standard. Proteomics 3(1):36–44.
14. Friedman, D.B., Hill, S., Keller, J.W., Merchant, N.B., Levy, S.E., Coffey, R.J.
and Caprioli, R.M. (2004) Proteome analysis of human colon cancer by two-
dimensional difference gel electrophoresis and mass spectrometry. Proteomics
4(3):793–811.
15. Gerbasi, V.R., Weaver, C.M., Hill, S., Friedman, D.B. and Link, A.J. (2004) Yeast
Asc1p and mammalian RACK1 are functionally orthologous core 40S ribosomal
proteins that repress gene expression. Mol Cell Biol 24(18):8276–87.
16. Sitek, B., Luttges, J., Marcus, K., Kloppel, G., Schmiegel, W., Meyer, H.E.,
Hahn, S.A. and Stuhler, K. (2005) Application of fluorescence difference gel
electrophoresis saturation labelling for the analysis of microdissected precursor
lesions of pancreatic ductal adenocarcinoma. Proteomics 5(10):2665–79.
17. Hu, Y., Malone, J.P., Fagan, A.M., Townsend, R.R. and Holtzman, D.M. (2005)
Comparative proteomic analysis of intra- and interindividual variation in human
cerebrospinal fluid. Mol Cell Proteomics 4(12):2000–9.
18. Zhang, X., Guo, Y., Song, Y., Sun, W., Yu, C., Zhao, X., Wang, H., Jiang, H.,
Li, Y., Qian, X., Jiang, Y. and He, F. (2006) Proteomic analysis of individual
variation in normal livers of human beings using difference gel electrophoresis.
Proteomics 6(19):5260–68.
19. Karp, N.A., Spencer, M., Lindsay, H., O’Dell, K. and Lilley, K.S. (2005)
Impact of replicate types on proteomic expression analysis. J Proteome Res 4(5):
1867–71.
20. Meunier, B., Dumas, E., Piec, I., Bechet, D., Hebraud, M. and Hocquette, J.F.
(2007) Assessment of hierarchical clustering methodologies for proteomic data
mining. J Proteome Res 6(1):358–66.
21. Fodor, I.K., Nelson, D.O., Alegria-Hartman, M., Robbins, K., Langlois, R.G.,
Turteltaub, K.W., Corzett, T.H. and McCutchen-Maloney, S.L. (2005) Statistical
challenges in the analysis of two-dimensional difference gel electrophoresis exper-
iments using DeCyder. Bioinformatics 21(19):3733–40.
122 Friedman and Lilley
22. Karp, N., Kreil, D. and Lilley, K. (2004) Determining a significant change in
protein expression with DeCyderTM during a pair-wise comparison using two-
dimensional difference gel electrophoresis. Proteomics 4(5):1421–32.
23. Kreil, D., Karp, N. and Lilley, K. (2004) DNA microarray normalization methods
can remove bias from differential protein expression analysis of 2-D difference gel
electrophoresis results. Bioinformatics 20(13):2026–34.
24. Friedman, D.B., Stauff, D.L., Pishchany, G., Whitwell, C.W., Torres, V.J. and
Skaar, E.P. (2006) Staphylococcus aureus redirects central metabolism to increase
iron availability. PLoS Pathog 2(8):e87.
25. Fujii, K., Kondo, T., Yamada, M., Iwatsuki, K. and Hirohashi, S. (2006) Toward
a comprehensive quantitative proteome database: protein expression map of
lymphoid neoplasms by 2-D DIGE and MS. Proteomics 3:3.
26. Fujii, K., Kondo, T., Yokoo, H., Yamada, T., Matsuno, Y., Iwatsuki, K. and
Hirohashi, S. (2005) Protein expression pattern distinguishes different lymphoid
neoplasms. Proteomics 5(16):4274–86.
27. Karp, N.A., Griffin, J.L. and Lilley, K.S. (2005) Application of partial least squares
discriminant analysis to two-dimensional difference gel studies in expression
proteomics. Proteomics 5(1):81–90.
28. Seike, M., Kondo, T., Fujii, K., Yamada, T., Gemma, A., Kudoh, S. and
Hirohashi, S. (2004) Proteomic signature of human cancer cells. Proteomics
4(9):2776–88.
29. Suehara, Y., Kondo, T., Fujii, K., Hasegawa, T., Kawai, A., Seki, K., Beppu, Y.,
Nishimura, T., Kurosawa, H. and Hirohashi, S. (2006) Proteomic signatures
corresponding to histological classification and grading of soft-tissue sarcomas.
Proteomics 6(15):4402–09.
30. Hatakeyama, H., Kondo, T., Fujii, K., Nakanishi, Y., Kato, H., Fukuda, S. and
Hirohashi, S. (2006) Protein clusters associated with carcinogenesis,histological
differentiation and nodal metastasis in esophageal cancer. Proteomics 6(23):
6300–16.
31. Verhoeckx, K.C., Gaspari, M., Bijlsma, S., van der Greef, J., Witkamp, R.F.,
Doornbos, R.P. and Rodenburg, R.J. (2005) In search of secreted protein
biomarkers for the anti-inflammatory effect of beta2-adrenergic receptor agonists:
application of DIGE technology in combination with multivariate and univariate
data analysis tools. J Proteome Res 4(6):2015–23.
32. Reddy, A.B., Karp, N.A., Maywood, E.S., Sage, E.A., Deery, M., O’Neill,
J.S., Wong, G.K., Chesham, J., Odell, M., Lilley, K.S., Kyriacou, C.P. and
Hastings, M.H. (2006) Circadian orchestration of the hepatic proteome. Curr Biol
16(11):1107–15.
33. Lee, I.N., Chen, C.H., Sheu, J.C., Lee, H.S., Huang, G.T., Yu, C.Y., Lu, F.J.
and Chow, L.P. (2005) Identification of human hepatocellular carcinoma-
related biomarkers by two-dimensional difference gel electrophoresis and mass
spectrometry. J Proteome Res 4(6):2062–69.
34. Liang, C.R., Leow, C.K., Neo, J.C., Tan, G.S., Lo, S.L., Lim, J.W., Seow, T.K.,
Lai, P.B. and Chung, M.C. (2005) Proteome analysis of human hepatocellular
Optimizing DIGE Technology 123
carcinoma tissues by two-dimensional difference gel electrophoresis and mass
spectrometry. Proteomics 5(8):2258–71.
35. Nabetani, T., Tabuse, Y., Tsugita, A. and Shoda, J. (2005) Proteomic analysis of
livers of patients with primary hepatolithiasis. Proteomics 5(4):1043–61.
36. Huang, H.L., Stasyk, T., Morandell, S., Dieplinger, H., Falkensammer, G., Gries-
macher, A., Mogg, M., Schreiber, M., Feuerstein, I., Huck, C.W., Stecher, G.,
Bonn, G.K. and Huber, L.A. (2006) Biomarker discovery in breast cancer serum
using 2-D differential gel electrophoresis/ MALDI-TOF/TOF and data validation
by routine clinical assays. Electrophoresis 27(8):1641–50.
37. Somiari, R.I., Sullivan, A., Russell, S., Somiari, S., Hu, H., Jordan, R., George, A.,
Katenhusen, R., Buchowiecka, A., Arciero, C., Brzeski, H., Hooke, J. and
Shriver, C. (2003) High-throughput proteomic analysis of human infiltrating ductal
carcinoma of the breast. Proteomics 3(10):1863–73.
38. Nishimori, T., Tomonaga, T., Matsushita, K., Oh-Ishi, M., Kodera, Y., Maeda, T.,
Nomura, F., Matsubara, H., Shimada, H. and Ochiai, T. (2006) Proteomic analysis
of primary esophageal squamous cell carcinoma reveals downregulation of a cell
adhesion protein,periplakin. Proteomics 6(3):1011–18.
39. Zhou, G., Li, H., DeCamp, D., Chen, S., Shu, H., Gong, Y., Flaig, M.,
Gillespie, J.W., Hu, N., Taylor, P.R., Emmert-Buck, M.R., Liotta, L.A.,
Petricoin, E.F. 3rd and Zhao, Y. (2002) 2D differential in-gel electrophoresis for
the identification of esophageal scans cell cancer-specific protein markers. Mol
Cell Proteomics 1(2):117–24.
40. Yu, K.H., Rustgi, A.K. and Blair, I.A. (2005) Characterization of proteins in
human pancreatic cancer serum using differential gel electrophoresis and tandem
mass spectrometry. J Proteome Res 4(5):1742–51.
41. Wan, J., Sun, W., Li, X., Ying, W., Dai, J., Kuai, X., Wei, H., Gao, X., Zhu, Y.,
Jiang, Y., Qian, X. and He, F. (2006) Inflammation inhibitors were remarkably up-
regulated in plasma of severe acute respiratory syndrome patients at progressive
phase. Proteomics 6(9):2886–94.
42. Greengauz-Roberts, O., Stoppler, H., Nomura, S., Yamaguchi, H., Goldenring, J.R.,
Podolsky, R.H., Lee, J.R. and Dynan, W.S. (2005) Saturation labeling with
cysteine-reactive cyanine fluorescent dyes provides increased sensitivity for
protein expression profiling of laser-microdissected clinical specimens. Proteomics
5(7):1746–57.
43. Kondo, T., Seike, M., Mori, Y., Fujii, K., Yamada, T. and Hirohashi, S. (2003)
Application of sensitive fluorescent dyes in linkage of laser microdissection and
two-dimensional gel electrophoresis as a cancer proteomic study tool. Proteomics
3(9):1758–66.
44. Sitek, B., Potthoff, S., Schulenborg, T., Stegbauer, J., Vinke, T., Rump, L.C.,
Meyer, H.E., Vonend, O. and Stuhler, K. (2006) Novel approaches to analyse
glomerular proteins from smallest scale murine and human samples using DIGE
saturation labelling. Proteomics 3:3.
45. Tetu, B., Lacasse, B., Bouchard, H.L., Lagace, R., Huot, J. and Landry, J. (1992)
Prognostic influence of HSP-27 expression in malignant fibrous histiocytoma:
124 Friedman and Lilley
a clinicopathological and immunohistochemical study. Cancer Res 52(8):
2325–28.
46. Wessel, D. and Flugge, U.I. (1984) A method for the quantitative recovery of
protein in dilute solution in the presence of detergents and lipids. Anal Biochem
138(1):141–43.
47. Knowles, M.R., Cervino, S., Skynner, H.A., Hunt, S.P., de Felipe, C., Salim, K.,
Meneses-Lorente, G., McAllister, G. and Guest, P.C. (2003) Multiplex proteomic
analysis by two-dimensional differential in-gel electrophoresis. Proteomics
3:1162–71.
48. Prabakaran, S., Swatton, J.E., Ryan, M.M., Huffaker, S.J., Huang, J.J., Griffin, J.L.,
Wayland, M., Freeman, T., Dudbridge, F., Lilley, K.S., Karp, N.A., Hester, S.,
Tkachev, D., Mimmack, M.L., Yolken, R.H., Webster, M.J., Torrey, E.F. and
Bahn, S. (2004) Mitochondrial dysfunction in schizophrenia: evidence for compro-
mised brain metabolism and oxidative stress. Mol Psychiatry 9(7):684–97.
49. Wang, D., Jensen, R., Gendeh, G., Williams, K. and Pallavicini, M.G. (2004)
Proteome and transcriptome analysis of retinoic acid-induced differentiation of
human acute promyelocytic leukemia cells,NB4. J Proteome Res 3(3):627–35.
50. Zhang, W. and Chait, B.T. (2000) ProFound: an expert system for protein
identification using mass spectrometric peptide mapping information. Anal Chem
72(11):2482–89.
51. Zhang, Y.Q., Matthies, H.J., Mancuso, J., Andrews, H.K., Woodruff, E. 3rd,
Friedman, D. and Broadie, K. (2004) The Drosophila fragile X-related gene
regulates axoneme differentiation during spermatogenesis. Dev Biol 270(2):
290–307.
52. Yokoo, H., Kondo, T., Fujii, K., Yamada, T., Todo, S. and Hirohashi, S. (2004)
Proteomic signature corresponding to alpha fetoprotein expression in liver cancer
cells. Hepatology 40(3):609–17.
53. Wang, S.E., Narasanna, A., Whitell, C.W., Wu, F.Y., Friedman, D.B. and
Arteaga, C.L. (2007) Convergence of P53 and TGFbeta signaling on activating
expression of the tumor suppressor gene maspin in mammary epithelial cells.J
Biol Chem 4:4.
7
MALDI/SELDI Protein Profiling of Serum
for the Identification of Cancer Biomarkers
Lisa H. Cazares, Jose I. Diaz, Rick R. Drake, and O. John Semmes
Summary
The ability to visualize the full depth of the serum proteome in a high-throughput
manner is a major goal of clinical proteomics. Methodologies, which combine higher
throughput with the ability to observe differential protein expression levels, have been
applied to this goal. An example of such a system is the coupling of robotic sample
processing to matrix-assisted laser desorption time of flight mass spectrometry (MALDI-
TOF-MS). Within this paradigm is a modification of MALDI-TOF termed surface-
enhanced laser desorption/ionization-TOF (SELDI-TOF). Both conventional MALDI and
SELDI have been used to generate protein expression profiles reflective of potential
peptide changes in serum. This information can be used to identify proteins, which may
enable new diagnostic and therapeutic strategies.
Key Words: matrix-assisted laser desorption ionization; surface-enhanced laser
desorption ionization; mass spectrometry; protein profiling; proteomics.
1. Introduction
Mining the serum proteome for the discovery of new biomarkers is
a major goal of many clinical proteomics efforts. Surface-enhanced laser
desorption/ionization (SELDI) and matrix-assisted laser desorption ionization
(MALDI) have been used extensively for protein profiling in efforts to discover
biomarkers in serum from cancer patients including prostate, lung, head and
neck, ovarian, and colon (1,2,3,4,5,6). MALDI techniques usually require some
up-front fractionation of the serum to reduce the complexity of the sample
(7,8,9) and the ease of use in sample fractionation is considered an advantage
From: Methods in Molecular Biology, vol. 428: Clinical Proteomics: Methods and Protocols
Edited by: A. Vlahou © Humana Press, Totowa, NJ
125
126 Cazares et al.
in SELDI. An advantage of MALDI-TOF instrumentation is the improved
resolution over SELDI instruments and the ability to directly identify peaks
of interest by analyzing samples in TOF/TOF mode. For routine linear mode
profiling both types of instrumentation give similar results with human serum
(see Fig. 1).
Besides the instrumentation and methodologies related to mass spectrometry
analysis, the quality and quantity of the clinical samples to be tested is an
important consideration. Serum is one of the most common sample types
used in biomarker discovery, because it is routinely obtained in the clinic, a
large proportion of blood clotting factors are removed, and it is a rich source
of molecules that may indicate systemic function. Blood plasma is an alter-
native source; however, clinical plasma collection utilizes various anticoagu-
lants, which should be standardized to allow for universal analysis. Whether
serum or plasma is used, every effort to standardize the sample collection and
processing protocols should be made. Several studies have highlighted this
and determined that multiple factors can affect the resulting spectra generated
from serum samples (10,11). These factors include the elapsed time between
venipuncture and separation of plasma and serum, type of serum collection tube,
1
4000 6000 8000 10,000
5904.6
4212.3
3266.1
2663.4 7762.3
5337.6 9282.0
Bruker IMAC Cu
2+
beads
Ciphergen IMAC Cu
2+
chip
2 3
Three primary peaks
used for instrument
standardization
A.
B.
Fig. 1. Comparison of SELDI and MALDI spectra using QC sera. (A) MALDI
spectra generated using QC processed with IMAC Cu2magnetic beads. (B) SELDI
spectra from QC sera processed on IMAC Cu2chips. The three peaks used for instrument
optimization are indicated.
MALDI/SELDI Protein Profiling of Serum 127
storage conditions, and the number of freeze thaw cycles. In our laboratory, we
routinely use serum for proteomic profiling. The following protocols outline
our method for collection and storage of serum samples for subsequent analysis
via MALDI-MS.
Reduction of sample complexity is an essential step in the generation of
high quality TOF mass spectrometry data from serum. One method of MALDI
sample preparation that reduces the complexity of serum while remaining
robust and easily amenable to automated high throughput applications is sample
fractionation using magnetic beads (MBs) combined with prestructured MALDI
sample supports (AnchorChip technology). Several MB types with different
surface chemistries can be used to fractionate serum and increase the number
of detectable peaks (12) (see Fig. 2). In addition, depletion of high abundant
WCX
=
84 peaks
IMAC
=
85 peaks
C18
=
62 peaks
WAX
=
80 peaks
203 total unique peaks mass range 1000–10000
8
.
2095
0.1
1
24
1.
8
729
5.4
5
9
2
4.446
4
3.2662
0.6335
3
.
56
2
3
6.2904
3.649
1
4
.
7
4
51
9
.46
9
4
4.9577
4
.0
1
2
2
7.7266
8.
3
3
7
1
8.7806
8.48
83
3.9053
7.1631
8
.3
2
98
0.00
0.25
0.50
0.75
1.00
1.25
1.50
×104
Intens. [a.u.]Intens. [a.u.]Intens. [a.u.]Intens. [a.u.]
6.4095
3
.
2124
1.66
2
3
9.
764
1
4.3
66
2
3.2677
6.73
3
5
5
.5
5
92
0.28
2
9
3.74
9
1
7.3904
1.
6
464
1
.5
883
2
.
802
1
9.9266
7.
21
22
1
.
7
5
5
2
0
.5
6
9
4
2.2346
5.
72
9
8
7.6
0
71
8
.
6
8
06
3
.
8
3
1
8
4
.0
5
43
0.0
0.5
1.0
1.5
×104
7.6101
5.2
1
24
8.539
2
2
.66
23
2.
8
641
0.70
6
2
4.0677
9.65
9
3
3.
8
2
6
6
4.8
0
21
4.2832
7.
0
3
4
6
4
.4
1
02
8.4219
3.
4
149
7.3095
0
.
0
9
7
1
0.
7
09
8
9.
5
318
5.446
4
1.7335
0.0
0.5
1.0
1.5
2.0
×104
7.
8
57
7
2.1124
7.2
6
2
1
1
.7266
2
.8451
1.9246
2.5504
0.
5
098
5.
9
6
4
4
6
.
1
1
4
9
1.4
3
18
1
.
1
2
19
9.6786
4.7062
4.6197
3.0716
5.701
2
3
.
2
60
5
9
.
6443
7.7
5
7
4
0.00
0.25
0.50
0.75
1.00
1.25
1.50
×104
1000 3000 4000 5000 7000 8000 9000 100002000 6000
m/z
Fig. 2. MALDI spectra of serum fractionated with magnetic beads. Example of
spectra produced on the Ultraflex-TOF/TOF when serum is fractionated with different
magnetic bead types. A total of 203 unique peaks are resolved in the m/zrange of
1000–10,000.
128 Cazares et al.
proteins such as albumin and IgG (13,14) serves to reduce ion suppression
phenomena as well as to reveal less abundant species. Unfortunately, fraction-
ation greatly increases the number of samples to be processed, which in
turn increases the complexity of the experimental procedure. Processing of
samples is, therefore, best facilitated by the use of robotics, which increases
throughput and produces reproducible results, however, manual processing of
small sample sets can be accomplished with careful attention to detail, and the
protocols and methods contained in this chapter. Another caveat to depletion
strategies is that highly abundant proteins such as albumin inadvertently bind
low abundant species (15,16). For comprehensive biomarker discovery, the
benefits of depletion and fractionation often outweigh these factors. We have
used both depleted and nondepleted serum strategies for biomarker discovery,
and this continues to be a major area of methodological development.
2. Materials
2.1. Serum Collection and Storage
1. Becton Dickinson vacutainer serum separator tube (SST) plus blood collection
tube (16 mm×100 mm, draw volume 8.5 mL) (Becton Dickenson #367988)
2. Screw cap microtubes for cryo-storage (2.0 mL) (Sarstedt Inc.# 72.609.001, with
caps # 65.716)
3. Microcentrifuge tubes for aliquots (1.7 mL) (Corning-Costar #3620)
2.2. Serum Processing for MALDI Using MB-Based Fractionation
1. The MB kit(s) (immobilized metal affinity-Cu, hydrophobic interaction, weak
cationic, or weak anionic exchange) (Bruker Daltonics, Billerica, MA)
2. Optional: ClinProt robotic workstation (Bruker Daltonics)
3. Magnetic separators for manual processing: large (1.5 mL) or small tube (0.5 mL)
format (Bruker Daltonics)
4. -Cyano-4-hydroxycinnamic acid (CHCA) (Bruker Daltonics)
5. Ethanol ultra pure 100%
6. Acetone ultra pure 100%
7. Micropipette capable of delivering 1 μL accurately
8. Peptide standard mix (Bruker)
9. Microtiter plate AnchorChip 600/384 MALDI target 600 μm diameter (Bruker
Daltonics)
2.3. Serum Processing for SELDI
1. Water high performance liquid chromatography (HPLC) grade (Fisher Scientific,
Hampton, NH)
MALDI/SELDI Protein Profiling of Serum 129
2. Copper sulfate, anhydrous (Sigma-Aldrich, St. Louis, MO)
3. Sodium acetate trihydrate salt
4. Phosphate buffered saline (PBS) buffer pH 7.4
5. Urea, at least 99% pure (Promega Madison, WI)
6. CHAPS ultra purity (Fisher Scientific)
7. Sinapinic acid (SPA) (5 μg tube)(Ciphergen Biosystems, Palo Alto, CA)
8. IMAC protein chip arrays (Ciphergen)
9. Bioprocessor holder (Ciphergen) for the processing or 12 chips in a 96-well
format
10. Bioprocessor accessory, 96-well disposable reservoir and gasket (Ciphergen)
11. Acetonitrile ultra high purity grade
12. Trifluoroacetic acid (TFA) (100%, 1 mL ampules) [Sigma/Aldrich Chemical
Company 26,977-8, (589-37-37)]
13. Plate seals
14. For calibration: (all from Ciphergen biosystems) NP20 ProteinChip arrays All-
in-one peptide standard All-in-one protein standard
15. Optional: BioMek 2000 robotic workstation, adapted to process ProteinChip
arrays (Ciphergen biosystems)
15. DPC MicroMix 5 shaker (Diagnostic Products Corporation, Los Angeles, CA)
or another type of rotary or platform shaker
16. Micropipet capable of delivering 1 μL accurately
17. Pooled serum for quality control (QC)
18. 100 mM CuSO4in water [room temperature (RT)]: 1.6 g CuSO4 (MW = 159.6)
made up to 100 mL in HPLC grade water
19. 100 mM sodium acetate, pH 4.0 (RT): 9.0 mL 0.2 M sodium acetate stock
(27.2 g/L), 50 mL HPLC water, 41.0 mL 0.2 M acetic acid (add gradually to
get to pH 4.0) (11.6 mL/L made from concentrated).
20. The PBS Buffer pH 7.4 (RT): 10 mL PBS Buffer (10) made up to 100 mL in
HPLC water. Check pH.
21. 10% TFA stock: 1 mL TFA (100%), 9 mL HPLC water (store in amber bottle)
22. 1% TFA working solution (store in amber bottle and make fresh every 2 weeks):
take 1 mL TFA (10%) and add 9 mL HPLC water
23. 8 M Urea, 1% CHAPS in PBS, pH 7.4: 48.05 g Urea, up to 90 mL PBS pH
7.4; stir until dissolved, may need warming. Add 1 g CHAPS. Bring the final
volume to 100 mL with PBS. Filter through 0.4 μm filter. Aliquot into 5 mL
volumes and freeze.
24. 1 M Urea, 0.125% CHAPS in PBS, pH 7.4: dilute the 8 M stock above in PBS
(100 mL8Min700mLPBS).
2.4. SELDI and MALDI Spectra Acquisition
1. SELDI PBS II, IIc, or PCS 4000 instrument (Ciphergen biosystems)
2. Ultraflex I or II MALDI-TOF–TOF (Bruker Daltonics)
130 Cazares et al.
3. Method
3.1. Serum Collection
Obtain proper patient consent:
1. Perform venipuncture into a 10 cc SST vacutainer tube (without anticoagulant).
2. Allow blood to clot at RT for 30 min.
3. Spin blood at 1700 rcf for 10 min, immediately decant and freeze serum at –70°C
in a screw cap freezer vial (Sarstedt). If this is not possible, the serum can be
stored at –20 for 5 days, before moving to a –70 freezer.
4. Prior to SELDI or MALDI analysis, the sample should be thawed and divided
into small volume aliquots to avoid multiple freeze thaws. When possible, no
sample should be taken through more than two freeze thaw cycles, and the number
of freeze/thaw cycles should be recorded if unused volumes are returned to the
freezer.
3.2. Preparation of Human Serum
Expression profiling of proteins/peptides utilizes both peak mass and
intensity to quantify changes in differential spectra. This necessitates the use
of a QC standard to monitor instrument performance (17). The QC sample
routinely used in our lab is pooled human serum collected using the same
serum collection protocol used to collect (see above SOP) the experimental
samples. Efforts have been made to develop a standardized QC sample for
serum mass spectrometry profiling (18). However, until that end, a large volume
of serum can be pooled and aliquoted to be run with every experimental
sample set. This QC sample should be assayed using the same processing
technique, which will be employed for the experimental samples and the data
from multiple runs analyzed. In this way, the inter- and intra-assay variability
can be determined. Additionally, the spectra obtained from the QC sample can
be used as a benchmark for the integrity of processing, instrument optimization,
and ProteinChip variability. We, therefore, recommend including several QC
samples on a MALDI target and one QC spot on each SELDI ProteinChip.
Acceptable levels of reproducibility need to be established for any new
technology, and sample preparation is the most critical step to the production
of reproducible spectra (see Notes 3,4, and 5). We have optimized the SELDI
system with high-throughput robotics, and previous studies in our laboratory
have determined that the mass accuracy of SELDI spectra is highly reproducible
with CV’s of 0.05%. Operating in linear mode, we have found the mass accuracy
of an Ultraflex-TOF–TOF to be 0.01% CV. Overall normalized intensity values
for individual peaks using QC sera are routinely below a 20% CV for samples
prepared robotically in our lab using either SELDI or MALDI-MS.
MALDI/SELDI Protein Profiling of Serum 131
3.3. Serum Protein Profiling on the MALDI-TOF–TOF
3.3.1. MB Fractionation of Human Serum
These steps are performed by the ClinProt robot. Below is an outline of
a comparable manual method. Sequential fractionation can also be performed
with multiple bead types.
1. Vortex MBs thoroughly for at least 1 min.
2. In a 0.5 mL eppendorf, pretreat 5 μL of MBs with 50 μL MB-IMAC Cu binding
solution.
3. Place the tube in the magnetic bead separator (MBS) and move it between
adjacent wells 10 times.
4. Collect the beads on the wall of the tube for 20 s and remove the supernatant
carefully with a pipette.
5. Repeat this pretreatment two more times.
6. Add 20 μL of serum and mix carefully with the beads by pipetting up and down
five times.
7. Keep at RT for 2 min.
8. Place the tube in the MBS and wait for 20 s for beads to separate.
9. Remove the supernatant with a pipette tip carefully (the unbound fraction can
be discarded or saved for analysis or a second fractionation step, if desired).
10. To wash, add 80 μL MB-IMAC Cu wash solution and place tube in the MSB
again. Move the tube back and forth to adjacent wells 10 times.
11. Collect the beads on the tube wall for 20 s and remove the supernatant carefully
with a pipette.
12. Repeat this wash two more times.
13. To elute, add 10 μL MB-IMAC Cu elution solution and mix. Let the beads sit
for 5 min at RT.
14. Place the tube on the MBS and wait 20 s for beads to separate.
15. Transfer the eluate to a fresh tube.
3.3.2. Data Collection on MALDI-TOF–TOF Instrument
To best detect proteins over the entire mass range on a MALDI instrument,
it is necessary to optimize the instrument settings for both low mass (typically
2000–20,000 Da) and high mass (20,000–100,000 Da or greater). The best
sensitivity and resolution is in the mass range below m/z20,000, and this is the
mass range we routinely use for most profiling experiments.
1. Prepare samples on an anchor plate by making dilutions of the eluates of 1:10
in CHCA matrix prepared according to the anchor chip protocol (0.3 mg/mL in
ethanol:acetone 2:1). SPA and/or 2,5-dihydroxybenzoic acid may also be used.
2. Spot 1 μL of the sample diluted in matrix onto the 600 μm diameter AnchorChip
target. Also spot 1 μL of the peptide standard diluted according to the manufac-
turer’s instructions.
132 Cazares et al.
3. Allow spots to dry.
4. Perform external calibration with the peptide standard using a linear mode method.
5. Collect at least 300 shots in linear mode, adjusting the laser energy and detection
sensitivity to maximize signal and resolution of the major peaks using a QC spot.
Typically, in linear mode the resolution of the three major peaks should be greater
than 600.
6. Instrument settings will vary based on instrument set-up, and are more numerous
that is feasible to describe in this book chapter but the most important settings to
optimize are acceleration voltage (IS1), laser power, time lag focusing (or PIE),
detector settings, and matrix suppression. Our basic instrument settings in linear
mode are as follows:
IS1, 22
Laser, 37% with laser attenuation offset at 48%, range at 40%
Time lag focus, 200 ns
Detector Gain, 24×
Matrix suppression, gated with suppression up to m/z800
All spectra should be processed using the same baseline subtraction protocol.
Perform peak detection using a uniform definition of requisite signal-to-noise
ratio and mass window. Although MALDI techniques have the potential to
produce protein profiles that contain patterns capable of distinguishing disease
and identifying biomarkers, a single analysis may produce many hundreds of
protein peaks (see Note 2). Therefore, the data analysis required to discern
the differentiating patterns poses a major challenge, and the analysis and inter-
pretation of the enormous volumes of proteomic data remains an unsolved
bioinformatics challenge. Many different classification tools are currently being
used with success for the analysis of MALDI data. These approaches include
Fisher discriminative analysis, CART (19,20), support vector machine (21),
artificial neural network (22), boosted decision tree analysis (23), and genetic
algorithm (24). General considerations for data preparation before any type
of analysis should include averaging intensity values for duplicate samples,
baseline subtraction, and peak picking.
3.4. Protein Identification Using MALDI-TOF/TOF
Biomarker candidates detected by protein profiling can be subjected to
TOF/TOF analysis for the identification of peptides directly from serum profiles
using the same sample spot and/or respotting of the sample. Initial analysis in
the reflectron mode will allow for visualization of the target or parent peak.
Metastable fragment ions of the respective precursor ion are then analyzed after
a second acceleration step, and the resulting fragment pattern is interpreted and
MALDI/SELDI Protein Profiling of Serum 133
Peptide View
MS/MS Fragmentation of DSGEGDFLAEGGGVR
Found in gi|229185, fibrinopeptide A
Start - End
2 - 16
Observed
1465.72
Mr(expt)
1464.72
Mr(calc)
1464.65
Delta
0.07
Miss
0
Sequence
DSGEGDFLAEGGGVR
Matched peptides shown in Bold Red
1 ADSGEGDFLA EGGGVR
1468.0
1868.2
1208.2
1619.0 2675.6
1352.5 1780.8 2024.5 2557.2
2297.6
1SLin, Baseline subtracted
A
0
1
2
3
×10
4
Intens. [a.u.]
1200 1400 1600 1800 2000 2200 2400 2600 m/z
C
B
Fig. 3. Identification of a serum peptide directly from the serum profile. Serum
profile (A) was generated in linear mode on the Ultraflex-TOF/TOF, from which a
peptide (m/z1469.09) was selected for MS/MS analysis resulting in a fragmentation
spectra (B). This peptide showed homology to fibrinopeptide A using the Mascot search
engine (C).
used for peptide identification via database search. The possibility to directly
sequence the peptides of interest is a powerful feature of this method (see Fig. 3).
3.5. Serum Protein Profiling on SELDI-TOF
3.5.1. Preparation of Serum
Note: All of the following steps including the ProteinChip preparation and
serum incubation on the arrays are performed robotically by the BioMek 2000
robot. The protocols below outline a manual method.
1. Thaw human serum samples on ice. Use separate aliquots to set up duplicates or
triplicates.
2. Add 20 μL human serum into a 1.7 mL microcentrifuge tube (alternatively, this
can be performed in a v-bottom 96-well plate for large sample sets).
134 Cazares et al.
3. Add 30 μL of 8 M Urea, 1% CHAPS in PBS pH 7.4.
4. Vortex tube at 4°C for 10 min or if using a plate, seal and place on MicroMix 5
shaker at 4°C for 10 min: shaker settings: form 20, amplitude 5, time 10 min.
5. Add 100 μL 1 M Urea, 0.125% CHAPS in PBS pH 7.4.
6. Vortex or pipette up and down to mix (total volume 150 μL).
7. Dilute sample 1:5 in PBS pH 7.4 by adding 600 μL PBS. If using a plate, remove
35 μL of serum–urea mixture from first plate and transfer to a second plate. Then
add 140 μL of PBS. Mix by vortexing tube or pipetting up and down.
8. Store on ice until ready to add samples to a bioprocessor containing ProteinChip
arrays.
3.5.2. Preparation of ProteinChip Arrays
This protocol describes the preparation of IMAC-Cu2+ProteinChips. Other
types of chips should be prepared according to the manufacturer’s (Ciphergen)
instructions.
1. Label or number IMAC chips on the reverse side and place them into the
bioprocessor according to the manufacturer’s instructions. (see Note 1)
2. Add 50 μL of 100 mM CuSO4onto each spot or array.
3. Shake on Micromix 5 for 10 min at RT.
4. Shaker settings: form 20, amplitude 5, time 10 min
5. Flick plate to remove CuSO4to waste and pat upside down onto a clean paper
towel to remove residual liquid (liquid can also be removed by aspiration, but
be careful no to touch array surface with pipette tip).
6. Wash with 200 μL of HPLC water 2 min × 5 min at RT on Micromix shaker at
the same settings for form and amplitude as before.
7. Flick plate and pat on paper towel.
8. Add 50 μL of 100 mM sodium acetate pH 4.0.
9. Shake on Micromix shaker for 5 min at RT.
10. Flick plate and pat as before.
11. Wash with HPLC water 2 min × 5 min at RT on Micromix.
12. Add 200 μL PBS pH 7.4.
13. Flick plate and pat as before.
14. Wash with PBS pH 7.4 2 min × 5 min at RT on Micromix.
Leave last volume of PBS on plate until ready to use.
3.5.3. Incubation of Serum on ProteinChip Arrays
1. Remove PBS from bioprocessor with multichannel pipettor, one row at a time
to avoid drying chips.
2. Add 100 μL of each sample to respective arrays. Note: samples should be
randomized as to their placement on the ProteinChip arrays. Duplicate samples
should also be randomly placed.
MALDI/SELDI Protein Profiling of Serum 135
3. Seal plate and shake bioprocessor on micromix (form 20, amplitude 5) for
30 min at RT.
4. Remove samples carefully with a pipette, changing tips to avoid cross contami-
nation.
5. Add 200 μL PBS pH 7.4 to each array and shake on micromix for 5 min at RT
using same shaker settings.
6. Remove PBS with multichannel pipettor changing tips for each row.
7. Wash with 200 μL HPLC water, shake on micromix for 5 min at RT.
8. Remove water with multichannel pipettor.
9. Repeat water wash.
10. Remove chips from bioprocessor and allow chips to dry completely.
3.5.4. Adding SPA Matrix to the Chips
1. To one tube of SPA, add 200 μL acetonitrile (100%).
2. Add 200 μL 1% TFA (final concentration of SPA:12.5 mg/mL in 50% acetonitrile,
50% 0.5% TFA).
3. Vortex for 5 min at RT.
4. Quick spin.
5. Add 1.0 μL SPA matrix to each dry spot, being careful not to touch the pipette
tip to the array surface.
6. Allow to dry.
7. Arrays are now ready to read on the SELDI instrument. Note: The arrays should
be stored in the dark in a cool dry place. It is recommended to read the chips
within a few hours of the addition of the matrix. Some signal degradation may
occur if the arrays are stored for more than 24 h).
3.5.5. Collection of Spectra on SELDI-TOF
We describe here the collection of spectra using the PBS II Ciphergen
instrument.
3.5.5.1. Calibration
Calibration of the SELDI instrument is crucial to the accurate mass analysis
of the proteins present in samples. Smaller ions fly faster than larger ions, and
their m/zratio can be calculated from their flight time using compounds of
known mass. For the most accurate mass assignments, the instrument should be
calibrated using conditions identical to the experimental conditions. Calibration
should be performed at the beginning of an experimental run, and thereafter
everyday the experimental data is collected. When obtaining calibration spectra,
use instrument settings as close to the settings used for serum profiling (i.e.,
detector voltage, lag time, etc.) as possible.
1. Reconstitute one vial each of the seven-in-one peptide and protein standards,
according to the manufacturer’s instructions. Aliquot and freeze.
136 Cazares et al.
2. Mix standards with SPA according to package insert.
3. Deposit 1 μL of each standard onto an array of an NP20 ProteinChip.
4. Air-dry the arrays completely, usually 30–60 min.
5. Read the array in the SELDI instrument using a spot protocol created to read
the experimental samples (see below). The laser intensity should be lowered
such that the peaks from the standards do not exceed 75% maximum signal
intensity.
6. Follow the calibration dialogue in the software of the PBSII SELDI instrument
to save the calibration equations.
3.5.5.2. SELDI Instrument Settings Optimization
The SELDI instrument optimization refers to the adjustment of settings
necessary for data collection, which will maximize signal intensity while
retaining the optimal resolution and the lowest noise. In our studies, there are
three consistently present protein peaks (m/z5900, 7764, 9284 ±0.2%) in the
QC sera processed on IMAC-Cu2+ProteinChips, which are used as bench-
marks for instrument optimization (see Fig. 1). Based on multiple runs, the
instrument settings are adjusted to maximize signal to noise and resolution for
these three peaks. Thereafter specific criteria were set to ensure instrument
optimization (refer to paper Semmes et al.(17)). Generally, when trying to
obtain a specific overall intensity level (e.g., to get two instruments to behave
similarly, or to obtain similar intensity levels over time), three parameters can
be adjusted. These include laser intensity, detector sensitivity, and detector
voltage. The following spot protocols for data collection on the SELDI reader
are a starting point. The settings will be different from instrument to instrument
and will change over time, based on cumulative laser utilization and detector
settings.
Data collection: standard spot protocol for QC serum on IMAC-Cu (for a
PBSII)
1. Set detector voltage to 1650.
2. Set high mass to 100,000 Da, optimized from 3000 to 50,000 Da.
3. Set starting laser intensity to 220.
4. Set starting detector sensitivity to 7.
5. Focus lag time at 900 ns.
6. Set data acquisition method to SELDI quantitation.
7. Set SELDI acquisition parameters 20 delta to 4 transients per to 12 ending position
to 80.
8. Set warming positions with two shots at intensity 230 and do not include warming
shots.
When adjusting to meet QC criteria:
MALDI/SELDI Protein Profiling of Serum 137
•Increasing detector voltage typically increases signal and noise. Change this in units
of 25 V.
•Increasing laser increases signal and generally decreases resolution. Change this in
units of 10.
•Increasing sensitivity increases signal intensity. Typical working range is six to eight.
For example, if the settings above are not meeting QC specifications, try the
following:
If S/N passes easily but resolution is low, reduce detector voltage or laser
intensity:
1. Set detector voltage to 1625.
2. Set high mass to 100,000 Da, optimized from 3000 to 50,000 Da.
3. Set starting laser intensity to 220.
4. Set starting detector sensitivity to 7.
5. Focus lag time at 900 ns.
6. Set data acquisition method to SELDI quantitation.
7. Set SELDI acquisition parameters 20 delta to 4 transients per to 12 ending position
to 80 (192 total shots).
8. Set warming positions with two shots at intensity 230 and do not include warming
shots.
If resolution passes but S/N is low increase laser intensity or detector voltage:
1. Set detector voltage to 1650.
2. Set high mass to 100,000 Da, optimized from 3000 to 50,000 Da.
3. Set starting laser intensity to 230.
4. Set starting detector sensitivity to 7.
5. Focus lag time at 900 ns.
6. Set data acquisition method to SELDI quantitation.
7. Set SELDI acquisition parameters 20 delta to 4 transients per to 12 ending position
to 80.
8. Set warming positions with two shots at intensity 230 and do not include warming
shots.
If intensity is too high (i.e., generally stay under 65), reduce laser intensity
and/or sensitivity:
1. Set detector voltage to 1650.
2. Set high mass to 100,000 Da, optimized from 3000 to 50,000 Da.
3. Set starting laser intensity to 220.
4. Set starting detector sensitivity to 6.
5. Focus lag time at 900 ns.
6. Set data acquisition method to SELDI quantitation.
138 Cazares et al.
7. Set SELDI acquisition parameters 20 delta to 4 transients per to 12 ending position
to 80.
8. Set warming positions with two shots at intensity 230 and do not include warming
shots.
After data collection, each spectrum should be calibrated for mass using the
current peptide calibration. If higher molecular weight data is included for
analysis, the protein standard calibration should be used for the peaks in this
mass range. Spectra should be normalized using total ion current (this is a
feature in the Ciphergen software) with the same normalization coefficient
and low mass cutoff (2000 Da for SPA matrix to exclude matrix peaks). All
spectra should also be processed using the same baseline subtraction protocol.
Perform peak detection using a uniform definition of requisite signal-to-noise
ratio (usually 3) and mass window (usually 0.2–0.3%).
4. Notes
1. Use powder-free nitrile (not latex) gloves when processing SELDI ProteinChips.
Repetitive peaks at 3000–4000 Da will appear in the spectra if samples are
contaminated with latex.
2. Use sample sets of sufficient size. A sample set of at least 30 should be included
in each classification group in order to do multivariate analysis and to give >90%
statistical confidence in a single marker with pvalues <0.01.
3. Make every effort to ensure that serum samples are collected, processed, and
stored in a consistent manner. Retrospective studies may be problematic, because
in most cases, samples were not processed or stored similarly.
4. Avoid hemolytic samples as hemoglobin can affect affinity binding interactions
and may confound data interpretation.
5. Initially when designing a protein profiling study, avoid using serum samples
collected at separate sites unless strict standard operating procedures for serum
collection have been followed and documented.
References
1. Adam, B. L., Qu, Y., Davis, J. W., Ward, M. D., Clements, M. A., Cazares, L. H.,
Semmes, O. J., Schellhammer, P. F., Yasui, Y., Feng, Z., and Wright, G. L. Jr.
(2002). Serum protein fingerprinting coupled with a pattern-matching algorithm
distinguishes prostate cancer from benign prostate hyperplasia and healthy men.
Cancer Res, 62: 3609–3614.
2. Wadsworth, J. T., Somers, K. D., Cazares, L. H., Malik, G., Adam, B. L.,
Stack, B. C. Jr., Wright, G. L. Jr., and Semmes, O. J. (2004). Serum protein profiles
to identify head and neck cancer. Clin Cancer Res, 10: 1625–1632.
3. Petricoin, E. F., Ardekani, A. M., Hitt, B. A., Levine, P. J., Fusaro, V. A.,
Steinberg, S. M., Mills, G. B., Simone, C., Fishman, D. A., Kohn, E. C., and
MALDI/SELDI Protein Profiling of Serum 139
Liotta, L. A. (2002). Use of proteomic patterns in serum to identify ovarian cancer.
Lancet, 359: 572–577.
4. de Noo, M. E., Mertens, B. J., Ozalp, A., Bladergroen, M. R., van der Werff, M. P.,
vandeVelde,C.J., Deelder, A. M.,andTollenaar,R.A.(2006). Detection of colorectal
cancer using MALDI-TOF serum protein profiling. Eur J Cancer, 42: 1068–1076.
5. Sidransky, D., Irizarry, R., Califano, J. A., Li, X., Ren, H., Benoit, N., and Mao, L.
(2003). Serum protein MALDI profiling to distinguish upper aerodigestive tract
cancer patients from control subjects. J Natl Cancer Inst, 95: 1711–1717.
6. Howard, B. A., Wang, M. Z., Campa, M. J., Corro, C., Fitzgerald, M. C., and
Patz, E. F. Jr. (2003). Identification and validation of a potential lung cancer serum
biomarker detected by matrix-assisted laser desorption/ionization-time of flight
spectra analysis. Proteomics, 3: 1720–1724.
7. Baumann, S., Ceglarek, U., Fiedler, G. M., Lembcke, J., Leichtle, A., and Thiery, J.
(2005). Standardized approach to proteome profiling of human serum based on
magnetic bead separation and matrix-assisted laser desorption/ionization time-of-
flight mass spectrometry. Clin Chem, 51: 973–980.
8. Orvisky, E., Drake, S. K., Martin, B. M., Abdel-Hamid, M., Ressom, H. W.,
Varghese, R. S., An, Y., Saha, D., Hortin, G. L., Loffredo, C. A., and Goldman, R.
(2006). Enrichment of low molecular weight fraction of serum for MS analysis of
peptides associated with hepatocellular carcinoma. Proteomics, 6: 2895–2902.
9. Feuerstein, I., Rainer, M., Bernardo, K., Stecher, G., Huck, C. W., Kofler, K.,
Pelzer, A., Horninger, W., Klocker, H., Bartsch, G., and Bonn, G. K. (2005).
Derivatized cellulose combined with MALDI-TOF MS: a new tool for serum
protein profiling. J Proteome Res, 4: 2320–2326.
10. Rai, A. J., Gelfand, C. A., Haywood, B. C., Warunek, D. J., Yi, J., Schuchard, M. D.,
Mehigh, R. J., Cockrill, S. L., Scott, G. B., Tammen, H., Schulz-Knappe, P., Speicher,
D. W., Vitzthum, F., Haab, B. B., Siest, G., and Chan, D. W. (2005). HUPO plasma
proteome project specimen collection and handling: towards the standardization of
parameters for plasma proteome samples. Proteomics, 5: 3262–3277.
11. Banks, R. E., Stanley, A. J., Cairns, D. A., Barrett, J. H., Clarke, P., Thompson, D.,
and Selby, P. J. (2005). Influences of blood sample processing on low-molecular
weight proteome identified by surface-enhanced laser desorption/ionization mass
spectrometry. Clin Chem, 51: 1637–1649.
12. Villanueva, J., Philip, J., Entenberg, D., Chaparro, C. A., Tanwar, M. K.,
Holland, E. C., and Tempst, P. (2004). Serum peptide profiling by magnetic
particle-assisted, automated sample processing and MALDI-TOF mass
spectrometry. Anal Chem, 76: 1560–1570.
13. Guerrier, L., Thulasiraman, V., Castagna, A., Fortis, F., Lin, S., Lomas, L.,
Righetti, P. G., and Boschetti, E. (2006). Reducing protein concentration range
of biological samples using solid-phase ligand libraries. J Chromatogr B Analyt
Technol Biomed Life Sci, 833: 33–40.
14. Fountoulakis, M., Juranville, J. F., Jiang, L., Avila, D., Roder, D., Jakob, P.,
Berndt, P., Evers, S., and Langen, H. (2004). Depletion of the high-abundance
plasma proteins. Amino Acids, 27: 249–259.
140 Cazares et al.
15. Lowenthal, M. S., Mehta, A. I., Frogale, K., Bandle, R. W., Araujo, R. P.,
Hood, B. L., Veenstra, T. D., Conrads, T. P., Goldsmith, P., Fishman, D., Petricoin,
E. F. 3rd, and Liotta, L. A. (2005). Analysis of albumin-associated peptides and
proteins from ovarian cancer patients. Clin Chem, 51: 1933–1945.
16. Mehta, A. I., Ross, S., Lowenthal, M. S., Fusaro, V., Fishman, D. A.,
Petricoin, E. F. 3rd, and Liotta, L. A. (2003). Biomarker amplification by serum
carrier protein binding. Dis Markers, 19: 1–10.
17. Semmes, O. J., Feng, Z., Adam, B. L., Banez, L. L., Bigbee, W. L., Campos, D.,
Cazares, L. H., Chan, D. W., Grizzle, W. E., Izbicka, E., Kagan, J., Malik, G.,
McLerran, D., Moul, J. W., Partin, A., Prasanna, P., Rosenzweig, J., Sokoll, L. J.,
Srivastava, S., Srivastava, S., Thompson, I., Welsh, M. J., White, N., Winget, M.,
Yasui, Y., Zhang, Z., and Zhu, L. (2005). Evaluation of serum protein profiling by
surface-enhanced laser desorption/ionization time-of-flight mass spectrometry for
the detection of prostate cancer: I. Assessment of platform reproducibility. Clin
Chem, 51: 102–112.
18. Rai, A. J., Stemmer, P. M., Zhang, Z., Adam, B. L., Morgan, W. T., Caffrey, R. E.,
Podust, V. N., Patel, M., Lim, L. Y., Shipulina, N. V., Chan, D. W., Semmes, O. J.,
and Leung, H. C. (2005). Analysis of human proteome organization plasma
proteome project (HUPO PPP) reference specimens using surface enhanced
laser desorption/ionization-time of flight (SELDI-TOF) mass spectrometry: multi-
institution correlation of spectra and identification of biomarkers. Proteomics, 5:
3467–3474.
19. Semmes, O. J., Cazares, L. H., Ward, M. D., Qi, L., Moody, M., Maloney, E.,
Morris, J., Trosset, M. W., Hisada, M., Gygi, S., and Jacobson, S. (2005). Discrete
serum protein signatures discriminate between human retrovirus-associated
hematologic and neurologic disease. Leukemia, 19: 1229–1238.
20. Qian, H. G., Shen, J., Ma, H., Ma, H. C., Su, Y. H., Hao, C. Y., Xing, B. C.,
Huang, X. F., and Shou, C. C. (2005). Preliminary study on proteomics of gastric
carcinoma and its clinical significance. World J Gastroenterol, 11: 6249–6253.
21. Ressom, H. W., Varghese, R. S., Abdel-Hamid, M., Eissa, S. A., Saha, D.,
Goldman, L., Petricoin, E. F., Conrads, T. P., Veenstra, T. D., Loffredo, C. A.,
and Goldman, R. (2005). Analysis of mass spectral serum profiles for biomarker
selection. Bioinformatics, 21: 4039–4045.
22. Liu, J., Zheng, S., Yu, J. K., Zhang, J. M., and Chen, Z. (2005). Serum protein
fingerprinting coupled with artificial neural network distinguishes glioma from
healthy population or brain benign tumor. J Zhejiang Univ Sci B, 6: 4–10.
23. Qu, Y., Adam, B. L., Yasui, Y., Ward, M. D., Cazares, L. H., Schellhammer, P. F.,
Feng, Z., Semmes, O. J., and Wright, G. L. Jr. (2002). Boosted decision tree analysis
of surface-enhanced laser desorption/ionization mass spectral serum profiles discrim-
inates prostate cancer from noncancer patients. Clin Chem, 48: 1835–1843.
24. Papadopoulos, M. C., Abel, P. M., Agranoff, D., Stich, A., Tarelli, E., Bell, B. A.,
Planche, T., Loosemore, A., Saadoun, S., Wilkins, P., and Krishna, S. (2004). A
novel and accurate diagnostic test for human African trypanosomiasis. Lancet, 363:
1358–1363.
8
Urine Sample Preparation and Protein Profiling
by Two-Dimensional Electrophoresis
and Matrix-Assisted Laser Desorption Ionization
Time of Flight Mass Spectroscopy
Panagiotis G. Zerefos and Antonia Vlahou
Summary
Urine represents the most easily attainable and consequently one of the most common
samples in clinical analysis and diagnostics. However, urine is also considered one of
the most difficult proteomic samples to work with due to its highly variable contents,
as well as the presence of various proteins in low abundance or modified forms. In this
chapter, we describe simple protocols and troubleshooting tips for urinary protein prepa-
ration and profiling by two-dimensional electrophoresis or directly via matrix-assisted laser
desorption ionization time of flight mass spectroscopy. Direct dilution, protein precip-
itation, ultrafiltration, and solid phase extraction in combination to the above profiling
technologies serve the means for reliable proteomics analysis of one of the most significant
yet very complex biological samples.
Key Words: urine; 2DE; MALDI-TOF-MS; protein profiling; sample preparation.
Abbreviations: ACT: Acetone, CE: Capillary electrophoresis, CHAPS:
[3-[(3-cholamidopropyl)dimethylammonio-1-propanesulfonate], CHCA: -Cyano-4-
hydroxycinnamic acid, d: Dalton, 2DE: Two-dimensional gel electrophoresis, DHB:
Dihydroxybenzoic acid, DTE: 1,4-Dithioerythritol, IEF: Isoelectric focusing, IPG:
Immobilized pH gradient, LC: Liquid chromatography, MALDI: Matrix-assisted laser
desorption ionization, MS: Mass spectrometry, MW: Molecular weight, MWCO:
Molecular weight cut-off, ns: Nano-second, o/n: Overnight, RCF: Relative centrifugal
forces, SA: Sinapinic acid, SDS: Sodium dodecylsulfate, SELDI: Surface-enhanced laser
desorption, SPE: Solid phase extraction, TCA: Trichloroacetic acid, TFA: Trifluoroacetic
acid, TGS: Tris-Glycine-SDS, TOF: Time of flight, UF: Ultrafiltration
From: Methods in Molecular Biology, vol. 428: Clinical Proteomics: Methods and Protocols
Edited by: A. Vlahou © Humana Press, Totowa, NJ
141
142 Zerefos and Vlahou
1. Introduction
Biological fluids play a central role in clinical chemistry. Investigation
of their cellular (cell number, morphology, etc.) biochemical (metabo-
lites, biomolecules) and physicochemical (pH, transparency, absorption, etc.)
attributes assists in formulating the clinical judgment on disease prognosis,
diagnosis, and treatment. Urine, according to International Union of Pure and
Applied Chemistry, is the human fluid, which contains water and metabolic
products and is excreted by the kidneys, stored in the bladder and normally
discharged by the way of the urethra. The protein content of urine is very low
under normal conditions (1) and derives mainly from human plasma proteins,
which are not filtered through the renal glomeruli. The presence of proteins at
high concentrations in urine is usually the result of disease or pharmaceutical
treatment. Creatinine assay in urine is one of the most common clinical exami-
nations and serves this exact purpose, to assess unexpected protein excretion.
It should be noted that besides the soluble proteins, urine also contains proteins
included in exfoliated cells as well as in membrane components known as
exosomes (2). In this chapter, we focus on the description of methods for the
analysis of the soluble urinary proteins and would recommend for the interested
reader the review by Pisitkun et al.(2), for a thorough description of the other
urinary protein components.
In comparison to other proteomics samples, urine is still less explored. The
main reason for this is the fact that urine is a difficult and diverse sample. Its
composition is age, sex, health, and drug dependent. In addition, tremendous
day variations on the protein content exist between first, void, midstream,
morning and random catch urine samples of a single donor. Despite these
facts, protein markers for disease have been detected in urine and have been
approved to be utilized as adjuncts to clinical assays for disease diagnosis
and prognosis (3,4). This justifies and triggers an in-depth analysis of the
urinary proteome, particularly with the advent of contemporary proteomics
technologies, with the objective to identify novel disease diagnostic/prognostic
biomarkers.
Specifically, urine proteome has been studied thoroughly by a series
of proteomics technologies. These include, two-dimensional electrophoresis
(5,6), liquid chromatography (LC) in combination to mass spectroscopy
(MS) (7,8), matrix-assisted laser desorption ionization-time of flight (MALDI-
TOF) or surface-enhanced laser desorption (SELDI)-TOF profiling (9,10,11,
12,13), capillary electrophoresis coupled to MS (14,15) and combinations
thereof, implementing several separation steps both chromatographic and
electrophoretic (15,16,17,18,19). The great interest in the investigation of the
urinary proteome is reflected by the recent establishment of the human urine
Urine Sample Preparation 143
and kidney proteome initiative (http://hkupp.kir.jp) within the Human Proteome
Organization that targets the integration of existing research efforts in this field.
In this chapter, we provide detailed protocols and troubleshooting tips
as experienced by the authors, in the preparation and analysis of urinary
proteins by two-dimensional gel electrophoresis (2DE) or directly by MALDI-
TOF-MS. We selected these two profiling approaches since the former
is a classical high resolution profiling approach (see also Chapters 4–
6), whereas the latter offers the advantage of high throughput [see also
Chapters 7 and 13]. In general, the process of urine analysis for the inves-
tigation of its protein content can be divided into three main steps: sample
collection, usually performed at the physician’s office, protein extraction,
protein separation, and detection. Each of these steps is very crucial and
affects significantly the output of the proteomics experiment. In this chapter,
an emphasis is given on the description of the various protein prepa-
ration/extraction methodologies including: ultrafiltration, precipitation, and
solid phase extraction (SPE) as they complement 2DE and MALDI-TOF-MS
profiling. Apparently, additional protein preparation methods exist such as
dialysis, ultracentrifugation, etc. (see Note 1); however, we have focused on
the three aforementioned methods due to their simplicity, increased repro-
ducibility, and overall compatibility with the 2DE and MALDI MS profiling
approaches.
1.1. Protein Precipitation
Protein precipitation is a very common purification procedure employed
for the isolation of macromolecules. The denaturation and precipitation of
proteins occurs in solutions of extreme ionic strength, very low pH, or high
concentrations of organic solvents. In such conditions, biopolymers do not
retain a conformation capable of sustaining their solubility. Commonly used
reagents are ammonium sulfate ([NH4]2SO4), used for protein desalting at
concentrations of 3 M, trichloroacetic acid (TCA) [used at concentrations higher
than 5% (w/v)], and several organic solvents [ethanol, acetone, acetonitrile,
chloroform, methanol, and isopropanol, at final concentrations higher than
50%, (v/v)]. The choice of the precipitation methodology depends primarily on
the analytical procedure employed. In general, protein desalting is avoided in
proteomics sample preparations since residual salts inhibit further analysis by
2DE and mass spectrometry. TCA precipitation followed by acetone washes is
very popular and efficient, especially in cases of very dilute protein solutions.
Organic solvents offer very high yields but some of them are toxic (methanol,
acetonitrile) while others like chloroform (also toxic) employ rather complicated
144 Zerefos and Vlahou
precipitation procedures. A detailed description of these approaches for urinary
protein preparation is provided in Section 3.
1.2. Ultrafiltration-SPE
Ultrafiltration is a technique based on the use of molecular filters in combi-
nation to centrifugal forces. The whole procedure is performed in a centrifuge
and in temperatures varying from 4°C to ambient conditions. It presents many
advantages; for example, proteins are kept in solution and are more easily
handled. A major disadvantage is the cost of the approach and the fact that
even traces of the filter materials, when eluted, produce significant problems
in MS based methodologies.
Solid phase extraction in combination to MS for urine clinical proteomics
is a newly added approach (22). SPE in the form of magnetic particles was
recently developed as the front end of direct profiling of biological fluids by
MS (23).
We have found that acetone or TCA precipitation and ultrafiltration are very
efficient urinary protein preparation approaches, highly compatible with 2DE
analysis (Figs. 1, 2). In the case of MALDI MS profiling, we favor the utilization
of ultrafiltration, SPE as well as direct dilution of urine in MS compatible
buffers as front end protein preparation methods (see Note 2,Fig. 3). The
detailed protocols are provided below.
12 3 4 5 67 8
Fig. 1. Comparison of urinary sample preparation approaches. Lanes correspond to:
(1) marker, (2) urine starting material, (3) TCA/acetone precipitation supernatant, (4)
TCA precipitate, (5) urine supernatant after 3 h centrifugation at 200,000 RCF, (6)
protein pellet after ultracentrifugation of 5 mL urine, (7) urine filtrate after ultrafiltration
through 5 kd MWCO, and (8) urine retentate after ultrafiltration. In lanes 2, 3, 5, and
7 equal volumes of urine sample were utilized; similarly, lanes 4 and 8 correspond
to same amount of starting urine material in order to facilitate comparison of the
approaches.
Urine Sample Preparation 145
2. Materials
2.1. Sample Collection, Handling, and Storage
1. Polypropylene aliquoting tubes (1.5, 2, 15, and 50 mL), Sarstedt Corporation
(Nümbrecht, Germany)
2.2. Urine Sample Preparation/Protein Precipitation
2.2.1. TCA/Acetone Precipitation Protocol
1. Trichloroacetic acid, ultra pure (store solutions at 2–8°C), Sigma Corporation
(St. Luis, MO, USA)
2. Acetone, analytical purity grade, Sigma Corporation
2.2.2. Organic Solvent Precipitation Protocol
1. Acetone, analytical purity grade, Sigma Corporation
2. Isopropanol, analytical purity grade, Sigma Corporation
3. Ethanol, analytical purity grade, Sigma Corporation
2.2.3. Urine Ultrafiltration
1. Amicon ultrafiltration devices, Millipore Corporation (Billerica, MA, USA)
2.2.4. Urine SPE
1. Bioselect C18 SPE cartridges were from Grace Vydac (Columbia, MS, USA)
2. Methanol, high performance liquid chromatography HPLC grade, Sigma
Corporation
3. Acetonitrile, HPLC grade, Sigma Corporation
4. Trifluoroacetic acid, HPLC grade, Sigma Corporation
2.3. Analytical/Profiling Techniques
2.3.1. Two-Dimensional Separation
1. Protean isoelectric focusing (IEF) cell, Biorad (Hercules, CA, USA)
2. Nonlinear immobilized pH gradient (IPG) strips (3,4,5,6,7,8,9,10), 17 cm long
3. 2DE sample buffer: 7 M urea, 2 M thiourea, 4% CHAPS w/v, 0.4% 1,4-
dithioerythritol (DTE) w/v, 2% IPG buffer (Biorad) w/v, all components are of
molecular biology grade
4. Mineral oil
5. Equilibration buffer I: 6 M urea, 50 mM Tris–HCl, pH 8.8, 30% glycerol, 2.0%
sodium dodecylsulfate (SDS), 30 mM DTE
6. Equilibration buffer II: 6 M urea, 50 mM Tris–HCl, pH 8.8, 30% glycerol (v/v),
2.0% SDS (w/v), 230 mM iodocatemide. All components are of molecular biology
grade
146 Zerefos and Vlahou
7. Fixation solution: 5% phosphoric acid (p.a grade, Sigma) w/v, 50% methanol v/v
(HPLC grade, Sigma)
8. Colloidal coomassie brilliant blue staining kit, Invitrogen (Carlsbad, CA, USA)
9. GS-800 calibrated densitometer and PDQuest software, Biorad
2.3.2. MALDI-TOF-MS
1. Matrix solution: 50% acetonitrile v/v, 0.1% trifluoroacetic acid (TFA) v/v, 0.75%
[-cyano-4-hydroxy-cinnamic (CHCA), Sigma Corporation]. Caution: all MALDI
matrices are light sensitive; avoid unnecessary light exposure. Fresh preparation
is advised, or else keep for 1 week (maximum) and store at 4°C
2. MALDI ground steel target plate
3. Ultraflex I MALDI-TOF-TOF-MS (Bruker Daltonics, Bremen, Germany)
4. FlexAnalysis 2.2 software, Bruker Daltonics
2.4. Miscellaneous
The HPLC grade water (Resistivity >18 Mcm−1, Total organic carbon
(TOC)<2 ppb) was utilized throughout the whole experimental process, and its
application is advised for all electrophoretic and MS approaches.
3. Methods
3.1. Sample Collection, Handling, and Storage
1. Collect urine sample (see Note 3).
2. If immediate processing and storage at –80°C is not possible, store at 4°C for a
short period of time (up to 6 h) (see Note 4).
3. Remove cellular components by centrifugation at approximately 3000 relative
centrifugal forces (RCF) for 20–30 min at 4–8°C. Aliquot the supernatant (4–5 mL
each aliquot) and store at –80°C, until further use (see Note 5).
3.2. Urine Sample Preparation/Protein Precipitation
3.2.1. TCA/Acetone Precipitation Protocol
1. Thaw a urine sample aliquot (see Note 6).
2. Add TCA to a final concentration of 15%, vortex and store overnight (o/n) at 4°C
(see Note 7).
3. Centrifuge at standard refrigerated bench-top centrifuges (for eppendorf type
tubes) for 15 min at RCF of 16,000–17,000 and 4°C. Discard the supernatant (see
Note 8).
5. Wash pellet with ice-cold acetone (–20°C), leave for 5–10 min at –20°C, and
centrifuge again for 15 min at 16,000–17,000 RCF. Discard supernatant and repeat
once more the washing step (see Notes 9,10).
Urine Sample Preparation 147
A1 A2
A3 A4
B1
B2
Fig. 2. Two-dimensional profiling of (A) 24 h collected urine concentrated by
(1) ultrafiltration through 5000 MWCO, (2) TCA precipitation, (3) acetone precipitation
without washing of the protein pellet, and (4) acetone precipitation with pellet washing.
In these cases (1,2,3,4), the starting material was preconcentrated via membrane
filtration (Pellicon 2 system, Millipore, Corporation); ultrafiltration and TCA or acetone
precipitation, as applicable, were applied for the further concentration of the sample
prior to 2DE analysis. (B) Two-dimensional profiling of random catch urine (50 mL
starting volume without any preconcentration) condensed via (1) ultrafiltration through
5000 MWCO and (2) acetone precipitation. In all cases, 1 mg of protein was analyzed
and visualized with colloidal coomassie stain in 3–10 nonlinear IPG strips.
6. Let pellet dry at ambient temperature (see Note 11).
7. Solubilize pellet in 2DE sample buffer and proceed with 2DE analysis (see
Subheading 3.3.1,Note 12, and Fig. 2).
8. The protein pellet may also be subjected to solubilization with MS compatible
buffers and analyzed by MS profiling (see Note 13,Subheading 3.3.2, and
Fig. 3).
3.2.2. Organic Solvent Precipitation Protocol
1. Add to the urine sample at least equal volume of the desired organic solvent
(ethanol, acetone, or isopropanol) and mix (see Notes 14,15).
2. Keep at –20°C o/n (see Note 16).
148 Zerefos and Vlahou
1000
A
B
C
D
E
F
G
Intensity
Mass to charge
×104
2
4
6
10,000
5000
Fig. 3. MALDI-TOF-MS profiling of urine. (A) Ultrafiltration retentate through
5000 MWCO, diluted 10×in 0.1% TFA; (B)10×dilution of urine in 0.1% TFA;
(C) supernatant of urine (diluted in 0.1% TFA) after protein precipitation via acetone;
(D) urine protein pellet from acetone precipitation reconstituted in 0.1% TFA; (E) urine
protein pellet from acetone precipitation reconstituted in 50% acetonitrile 0.1% TFA;
(F) acetone precipitation (supernatant) and further purification of the supernatant by
C18-SPE followed by dilution in 0.1% TFA; (G) C18-SPE eluate in 50% acetonitrile,
0.1% TFA. Extensive reproducibility studies indicated that urine processing by ultra-
filtration or direct dilution in 0.1% TFA provides with the most robust spectra of the
methods tested. Adapted from (13).
3. Centrifuge at standard refrigerated bench-top centrifuges (for eppendorf type
tubes) for 15 min at RCF of 16,000–17,000 and 4°C. Discard the supernatant.
4. Wash pellet with ice-cold acetone, leave for 5–10 min at –20°C, and centrifuge
again. Discard supernatant and repeat once more the washing step (see Note 17).
5. Let pellet dry at ambient temperature.
6. Solubilize pellet and proceed with 2DE analysis. The protein pellet or supernatant
may also be subjected to solubilization with MS compatible buffers and analyzed
by MS profiling (see Notes 12,13,Subheading 3.3.1,Figs. 2, 3).
3.2.3. Urine Ultrafiltration
1. Place one volume of urine upon a 5000 kd molecular weight cut-offs (MWCO)
Amicon ultrafiltration device (see Notes 18–20).
Urine Sample Preparation 149
2. Spin in a refrigerated centrifuge at 3500 RCF and 8–12°C (see Notes 21,22).
3. After condensation, collect the retentate and discard or keep the filtrate depending
on the specific application (see Notes 23–25).
4. For 2DE add the appropriate volume of sample buffer to the retentate and proceed
with IEF (see Notes 26–27,Subheading 3.3.1, and Fig. 2).
5. For MALDI profiling dilute the retentate 10 times with 0.1% TFA v/v, and
proceed as described below (see Subheading 3.3.2,Fig. 3).
3.2.4. Urine SPE (see Note 28)
1. Activate cartridge with a total of 1 mL methanol (two applications of 500 μL each).
2. Wash cartridge with 2 mL acetonitrile (four applications of 500 μL each, see
Note 29).
3. Equilibrate cartridge with a total of 1 mL 0.1% TFA v/v (two applications of
500 μL each).
4. Load cartridge with 1 mL urine acidified by TFA at 0.1% (v/v) final concentration.
5. Wash cartridge with 1 mL 0.1% TFA v/v (two applications of 500 μL each).
6. Elute compounds by adding 100 μL of 50% acetonitrile, 0.1% TFA v/v.
7. Take 1 μL eluent, place on MALDI target, and process for MALDI MS profiling
(see Subheading 3.3.2,Fig. 3).
3.2.5. Direct Dilution of Urine
This method is used only in conjunction to direct MALDI MS profiling
•Dilute urine 10 times with 0.1% TFA v/v (see Notes 30,31).
•Apply 1 μL of the urine sample on MALDI target.
•Apply 1 μL matrix solution.
•Proceed with MALDI-TOF-MS (see Subheading 3.3.2,Fig. 3).
3.3. Analytical/Profiling Techniques
3.3.1. Two-dimensional Separation
1. Measure protein concentration of the sample (pretreated by precipitation or
ultrafiltration) by the use of a commercially available protein kit.
2. Take 0.5–1 mg of urinary proteins diluted in 300 μL of 2DE sample buffer (see
Note 32).
3. Distribute the sample volume equally in a lane of the IEF focusing tray.
4. Place the strip carefully, with the gel face down and in contact with the electrodes
(see Note 33).
5. Rehydrate actively for 16 h at 50 V and 20°C. Caution: do not cover the strip
with mineral oil immediately but after1hofrehydration (see Note 34).
6. After rehydration, place moistened IEF papers between the strip and electrodes.
7. Start IEF. The typical program is: 250 V for 30 min, linear increment up to
5000 V in 12 h, 5000 V for 16 h (total 110,000 V-h) (see Note 35).
150 Zerefos and Vlahou
8. After IEF is complete, equilibrate strip with 10 mL equilibration buffer I for
20 min at ambient temperature.
9. Alkylate with 10 mL equilibration buffer II for 20 min (see Note 36).
10. Place strip on top of 12.5% polyacrylamide gel, cover with 0.5% melted agarose
in TGS buffer and start second dimension. Start with 10 mA current for1hand
continue with 40 mA for approximately another4h(see Note 37).
11. Fix gel for 2 h with fixation solution.
12. Stain o/n with colloidal coomassie blue stain (Fig. 2).
3.3.2. MALDI-TOF-MS
1. Place 1 μL sample on the MALDI target plus 1 μL matrix solution and mix on
spot (dried droplet technique, see Notes 38 and 39).
2. Leave target to dry at ambient temperature in the dark.
3. Load sample in the instrument and execute the appropriate MS method. Run the
instrument in linear mode (see Note 40).
4. Optimize ion acceleration; tempering with sensitivity of the detector is not recom-
mended prior to MS method establishment (see Note 41).
5. Set pulsed ion extraction (delayed ion acceleration) according to the profiling
region in use. Typically when -cyano-cinnamic acid is utilized 50–150 ns are
applied for large peptides (3–5 kd), 150–300 ns for small molecular weight
proteins (15 kd), and higher than 300 for proteins (>20 kd, see Notes 42 and 43).
6. Collect 1000–2000 shots per sample and sum the collected data (see Note 44).
4. Notes
1. Dialysis is one of the most classical methods for buffer exchange and purifi-
cation (separation) of high from low molecular weight constituents of a specific
sample. Although it has been utilized elsewhere (20) we consider it rather
laborious, costly and serving solely purification and not condensation purposes.
Ultracentrifugation has been applied (21) for the isolation of higher molecular
weight urinary proteins prior to 2DE (Fig. 1). In our opinion, centrifugal
isolation of proteins is a very diverse and complicated issue and reproducibility is
consequently compromized. Precipitation of biopolymers by ultracentrifugation
requires the use of solutions with very well calculated composition in order to
extract the velocity for protein isolation from the theoretical Svedberg values.
Urine samples differ significantly in density (d=m/v) and pH values to serve
such purposes in a well-defined and reproducible manner.
2. It should be emphasized that extensive complementarity of the various methods
exists; thereby the combinatorial application of different methods is recom-
mended in order to increase protein resolution.
3. Urine samples can be first void, midstream, morning, random catch, or 24 h.
Due to its high bacterial content, first morning urine is usually not recommended
in biomarker discovery studies.
Urine Sample Preparation 151
4. Upon their collection, if not stored immediately in –80°C, urine samples should
be stored at 4°C. Published data support (9,10) that for analysis by 2DE or
SELDI/MALDI MS the generated proteomic profiles are usually stable for up to
24 h urine storage at 4°C prior to deep freezing. We have observed occasional
profile changes after so prolonged storage times at 4°C, and we therefore favor
shorter times.
5. An enrichment of the soluble supernatant for cellular proteins may be achieved
if prior to the centrifugation step a mild sonication (sonicator bath) for 5–10 min
is applied.
6. The volume of urine required depends on the specific downstream application.
For 2DE analysis an aliquot of at least 15 mL of urine is required. For direct
MALDI MS profiling 1 mL urine aliquot is sufficient.
7. The TCA can be added as solid to a final concentration of 15% (w/v) (TCA
is extremely hydroscopic and is easily solubilized). Alternatively, the appro-
priate volume of 100% TCA w/v may be added to the urine sample to reach
a final concentration of 15% (w/v). TCA precipitation can also be performed
at –20°C and o/n storage with occasionally slightly better efficiency. Caution:
TCA solutions may form bilayer aqueous–organic systems depending on the
salt concentration of the urine at –20°C or lower temperatures. The precipi-
tation efficiency is dependent of the protein concentration of a given sample;
in our experience, for example, the precipitation yield for a starting material
of 0.5 mg/mL protein concentration (i.e., 1 mg total protein found in 2 mL
sample) ranges from 40 to 70%; in contrast the precipitation efficiency for a
starting material of 0.1 mg/mL protein concentration (i.e., 1 mg protein in 10 mL
sample) is 0–30%. For this reason, avoid adding TCA solution in very dilute
protein samples.
8. In case where the highest available centrifugal force is only 4000–5000 RCF,
then longer centrifugation times (45 min) are recommended.
9. The volume of acetone utilized for washing depends on the size of the protein
pellet. A general rule is to use 1 mL acetone for every 1 mL of urine starting
material.
10. Acetone washes are needed to drive of excess TCA or else the pellet is extremely
acidic and buffers utilized in further steps are neutralized. In addition, TCA
(nonvolatile acid) may inhibit IEF, PAGE, LC, or MS analysis. We have found
that acetone washes of the pellet does not induce significant protein losses.
11. The pellet should not be completely dried off, since this renders difficult
its subsequent solubilization in 2DE or other buffers. Acetone evaporation at
elevated temperatures is not recommended for the same reason.
12. If the pellet does not come in solution, try mild sonication (5 min in a sonicator
bath) or incubate at ambient temperature for 30 min with intermittent vortexing.
However, heating should be avoided (particularly if the pellet is resuspended in
2DE buffer since urea decomposes when heated and reacts with amino acids).
The buffer volume required for solubilization depends on the protein content
(pellet size) and the type of downstream application (2DE or MALDI-TOF-MS).
152 Zerefos and Vlahou
13. The protein pellet may be solubilized in 0.1% TFA v/v (roughly 100 μL of
solubilization buffer for every milliliter of urine starting material) and analyzed
by MALDI-TOF-MS. However, in our experience, plasticizers possibly extracted
during the precipitation process are frequently detected and reproducibility
problems are observed. Therefore, unless additional purification steps are intro-
duced (SPE, etc.), we do not favor the application of precipitation methods at
the front end of MALDI MS profiling.
14. The use of ethanol, acetone, or isopropanol is favored. These are hydrophobic,
water mixable – even at elevated salt concentrations – nontoxic, and volatile.
In particular, we favor the use of acetone since it is cheap, extremely volatile,
and rarely forms aqueous–organic bilayers. Organic solvent mixtures e.g.,
isopropanol–acetone, do not increase precipitation efficiencies; in our experience
their use induces reproducibility problems and therefore is not recommended.
15. The sample to solvent ratio depends on the downstream application and the
sample protein concentration. For dilute urine samples (protein concentration of
micrograms per milliliter) a solvent to sample ratio of 3 provides relatively high
precipitation efficiencies. We have observed that for more concentrated samples
(for example, preconcentrated urine or in general starting material of protein
content in the micrograms per milliliter range), the precipitation efficiency for
lower MW constituents reaches its maximum at solvent to sample ratio of
about 9.
16. Precipitation is most efficient at –20°C (lower efficiencies have been observed at
4°C, whereas at –80°C bilayer systems may form, which inhibit the procedure).
17. Acetone washes of pellet in organic solvent precipitation protocols are not
accustomed. From our experience, however, washing offers great advantages
especially when 2DE separation is the downstream application since salts and
other interfering substances are removed (Fig. 2). This washing step renders
2DE gels produced after acetone precipitation equally good to those generated
following TCA precipitation. Acetone washing induces negligible protein losses.
18. There are Amicon UF devices that can accommodate up to 4 (UF4) or 15 mL
(UF15) sample volumes. We regularly utilize the UF4 devices when MALDI MS
profiling is to be performed and UF15 when 2DE is the downstream application.
19. Amicon devices have several MWCO. We propose the use of 5000 kd MWCO
for the isolation and condensation of “total” urine protein content. The use of
different MWCO is advised for specific isolation of molecular weight groups
(see also Note 25). It should be emphasized that UF is not an absolute size-
exclusion separation method and cross-contamination between different protein
size groups is expected and regularly observed.
20. UF can be performed in the presence of chemical additives. The kind of additives
in use depends on the downstream application (2DE, MALDI profiling, LC-
MS, etc.) since in all cases the chemical compatibility to the latter should be
maintained. For example, we have observed that in case of direct MALDI MS
profiling most additives (detergents such as: octyl-glucopyranoside, triton-100,
tween-20, and organic solvents such as: trifluoroethanol <5%, acetonitrile <20%,
Urine Sample Preparation 153
and isopropanol <20%) should be avoided since significant ion suppression is
caused by membrane eluting compounds. From our experience, in the case of
simple urine condensation via 5 kd MWCO for 2DE analysis no additives are
necessary. In rare cases where proteins precipitate and plunge the membrane
filter, the use of urea (4–6 M solutions) with 0.1–1% 2DE compatible detergent
is recommended. The same applies if the removal of smaller proteins (e.g.,
application of 10 kd MWCO) is intended via UF. In this case, chaotropes disrupt
protein–protein interactions and ease the separation of different MW groups.
21. The centrifugal forces are selected based on the manufacturer’s instructions. The
temperature of the UF procedure should be kept relatively low (8–12°C). In our
experience, however, lower than 8°C temperatures may generate problems such
as protein or urea precipitation. Instead, we favor the application of 8–12°C
since at this temperature the UF procedure is not significantly decelerated and
at the same time the rate of protein degradation is decreased.
22. In case where the rate of condensation is very slow, you should pause
the centrifugation, pipette up and down the retentate, and continue with the
centrifugation.
23. Good pipetting (up and down) should be performed for the collection of the
retentate in order to reduce adsorption losses. Additionally, the filter may be
washed off with a small volume of the desired buffer (e.g., sample buffer for
2DE). The latter approach is favored since it minimizes protein losses and
increases reproducibility.
24. Amicon UF devices have a minimum cut-off volume (approximately 50 and
250 μL for 4 and 15 mL devices, respectively). Always keep record of the
final volume since that will allow the estimation of the condensation factor.
For example, if you started with 15 mL of urine and finally collected 250 μL
retentate, you have concentrated your sample’s higher MW constituents by 60×.
25. When specific isolation of molecular weight groups is desired, sequential UF
may be performed starting from 50, then 30, then 10, or 5 kd MWCO; in each
case keep the retentate and use the filtrate for concentration though the next UF
device.
26. Buffer (0.1.% TFA in case of MALDI-TOF or sample buffer in case of 2DE) may
be added to the filtrate and the process of condensation through centrifugation
may be repeated. Caution: in our opinion washing of the sample with sequential
UF for MALDI profiling is not advised; we have observed an increased chemical
noise possibly attributed to accumulation of plasticizers.
27. When 2DE is to be performed, do not use wash buffers containing high concen-
trations of IEF incompatible reagents. For example, even 50 mM of Tris–HCl,
may cause problems in IEF. Always prefer no additives (for example, addition
of ultrapure water) or urea solutions (with or without CHAPS) in concentrations
compatible to the filter device. Caution: incompatible chaotrope concentrations
may cause filter corrosion and sample loss.
28. Solid phase extraction exhibits a wide selection range for many types of
separation; even isolation of proteins–peptides with specific characteristics
154 Zerefos and Vlahou
(e.g., phosphor or glycopeptides) is feasible and that is which differentiates SPE
from other sample preparation steps. From our point of view SPE in combination
to direct MS profiling is encouraged.
29. All chromatographic and SPE media contain residuals and plasticizers, which
should be driven off prior to analyte binding. Failure to perform this step may
result in complete ionization suppression during MALDI profiling.
30. The user may have to try different dilutions of the urine sample. In MALDI MS
profiling experiments, there is a range of protein concentration within which the
spectra quality is not affected. It is advised to conduct preliminary experiments
in order to address this issue.
31. In addition to TFA, the use of several additives (urea, octyl-glucopyranoside,
triton-100, tween-20, NP-40, cholate, and organic solvents) at MALDI MS
compatible concentrations has been tested on urinary peptide–protein ionization.
However, we did not observe any clear advantage on protein resolution or
ionization in these cases.
32. The recommended protein amount of 0.5–1 mg is suitable for 17–18 cm length
and 3–10 or 4–7 pH range strips. The protein amount will vary if different strip
types are utilized, according to the manufacturer’s guidelines (for additional tips
on 2DE see Chapters 4–6).
33. Noncup loading was found to provide better resolution in urine analysis by 2DE
compared to the cup loading method.
34. Direct addition of the mineral oil might cause extraction of hydrophobic proteins
to the oil layer.
35. These running conditions are for the analysis of 1 mg protein sample on wide
range (3–10 or 4–7) 17 or 18 cm IPG strips. The program will vary depending
on the sample quantity and the type of strip in use.
36. Reduction and alkylation are necessary for higher protein resolution in SDS-
PAGE and also for protein identification through peptide mass fingerprinting.
37. The low starting current is needed for the slow migration of the proteins from
the strip to the polyacrylamide gel. Direct electrophoresis with 40 mA current
may cause protein losses. Alternatively, the gel may run at 10 mA o/n. Although
slower, the latter approach provides gels of higher resolution, in our experience
(for additional tips on 2DE see Chapters 4–6).
38. Several sample application techniques were tested (thin layer preparation, double
layer, and variations of dried droplet). Of those, we found that dried droplet
(with simultaneous sample and matrix application) was the simplest, fastest,
and most reliable method. In addition, the simultaneous drying of sample and
matrix solution (rather than sample and matrix separately) increases repro-
ducibility and minimizes losses during subsequent spot washes. In contrast,
if sample and matrix are mixed prior to their application on the target, their
consumption is much higher and the sample exposure to plastics increases,
thereby increasing the chances for sample contamination and subsequent ion
suppression by plasticizers.
Urine Sample Preparation 155
39. In case that crystal formation is obscured due to high salt content in the
sample, wash the spot by pipetting two to three times with 2 μL of cool 0.1%
TFA solution v/v (let dry again, do not wipe dry). Always prefer spot to spot
washing rather than washing the entire target, in order to avoid sample cross-
contamination.
40. Instrument calibration is performed according to the manufacturer specifications.
In any case, we propose daily calibration to ensure precision and accuracy.
41. Acceleration of biomolecules is first of all affected by voltage settings of the
ion source. Settings of the analyzer (TOF) affect mainly resolution parameters,
while detector settings should be tempered only to improve signal to noise
characteristics of a given sample.
42. The mass spectrum should be divided into subregions and data of each of the
latter should be collected separately, in order to increase protein resolution.
This is because ionization kinetics (and consequently instrument settings) are
completely different for different protein sizes.
43. Different matrices (e.g., CHCA or dihydroxybenzoic acid for peptides and SA
for proteins) require different laser focusing settings. In general, large crystals
(such as the ones formed by SA) and larger protein molecules require more
concentrated energy bursts than smaller ones where more disperse hits may be
used.
44. Always sum the same amount of laser shots and select as many regions of a
spot as possible to ensure high reproducibility.
Acknowledgments
This study was supported by the Greek Ministry of Health.
References
1. Norden, G.W.A., Sharratt, P., Cutillas, P.R., Cramer, R., Gardner, S.C. and
Unwin, R.J. (2004) Quantitative amino acid and proteomics analysis: Very low
excretion of polypeptides >750 Da in normal urine. Kidney International 66,
1994–2003.
2. Pisitkun, T., Johnstone, R. and Knepper, M.A. (2006) Discovery of urinary
biomarkers. Molecular and Cellular Proteomics 5, 1760–1771.
3. Nielsen, M.E., Schaeffer, E.M., Veltri, R.W., Schoenberg, M.P., Getzenberg, R.H.
(2006) Urinary markers in the detection of bladder cancer: What’s new? Current
Opinion in Urology 16, 350–355.
4. Thongboonkerd, V. and Malasit, P. (2005) Renal and urinary proteomics: Current
applications and challenges. Proteomics 5, 1033–1042.
5. Pieper, R., Gatlin, C.L., McGrath, A.M., Makusky, A.J., Mondal, M.,
Seonarain, M., Field E., Schatz, C.R., Estock, M.A., Ahmed, N., Anderson, N.G.
and Steiner, S. (2004) Characterization of the human urinary proteome: A method
156 Zerefos and Vlahou
for high-resolution display of urinary proteins on two-dimensional electrophoresis
gels with a yield of nearly 1400 distinct protein spots. Proteomics 4, 1159–1174.
6. Oh, J., Pyo, J., Jo, E., Hwang, S., Kang, S., Jung, J., Park, E., Kim, S., Choi, J.
and Lim, J. (2004) Establishment of a near-standard two-dimensional human urine
proteomic map. Proteomics 4, 3485–3497.
7. Spahr, C.S., Davis, M.T., McGinley, M.D., Robinson, J.H., Bures, E.J., Beierle, J.,
Mort, J., Courchesne, P.L., Chen, K., Wahl, R.C., Yu, W., Luethy, R. and
Patterson, S.D. (2001) Towards defining the urinary proteome using liquid
chromatography-tandem mass spectrometry I. Profiling an unfractionated tryptic
digest. Proteomics 1, 93–107.
8. Cutillas, P.R., Norden, A., Cramer, R., Burlingame, A. and Unwin, R.J. (2003)
Detection and analysis of urinary peptides by on-line liquid chromatography and
mass spectrometry: Application to patients with renal Fanconi syndrome. Clinical
Science 104, 483–490.
9. Schaub, S., Wilkins J., Weiler, T., Sangster, K., Rush, D., Nickerson, P.
(2004) Urine protein profiling with SELDI TOF MS. Kidney International 65,
323–332.
10. Rogers, M.A., Clarke, P., Noble, J., Munro, N.P., Paul, A., Selby, P.J. and
Banks, R.E. (2003) Proteomic profiling of urinary proteins in renal cancer by
surface enhanced laser desorption ionization and neural-network analysis: Identi-
fication of key issues affecting potential clinical utility. Cancer Research 63,
6971–6983.
11. Vlahou, A., Schellhammer, P.F., Mendrinos, S., Patel, K., Kondylis, F.I., Gong, L.,
Nasim, S. and Wright, J.G. Jr. (2001) Development of a novel proteomic approach
for the detection of transitional cell carcinoma of the bladder in urine. The American
Journal of Pathology 158, 1491–1502.
12. Vlahou, A., Giannopoulos, A., Gregory, B.W., Manousakas, T., Kondylis, F.I.,
Wilson, L.L., Schellhammer, P.F., Semmes, O.J. and Wright G.L. Jr. (2004) Protein
profiling in urine for the diagnosis of bladder cancer. Clinical Chemistry 50,
1438–1445.
13. Zerefos, P.G., Prados, J., Kalousis, A. and Vlahou, A. (2007) Sample preparation
and bioinformatics in MALDI profiling of urinary proteins. Journal of Chromatog-
raphy B.Analyt Technol Biomed Life Sci. 15, 20–30.
14. Zórbig, P., Renfrow, M.B., Schiffer, E., Novak, J., Walden, M., Wittke, S., Just, I.,
Pelzing, M., NeusóÌ, C., Theodorescu, D., Root, K.E., Ross, M.M. and Mischak, H.
(2006) Biomarker discovery by CE-MS enables sequence analysis via MS/MS with
platform-independent separation. Electrophoresis 27, 2111–2125.
15. Mischal, H., Kaiser, T., Walden, M., Hillmann, M., Wittke, S., Herrmann, A.,
Knueppel, S., Haller, H. and Fliser, D. (2004) Proteomic analysis for the assessment
of diabetic renal damage in humans. Clinical Science 107, 485–495.
16. Zerefos, P.G., Vougas, K., Dimitraki, P., Kossida, S., Petrolekas, A.,
Stravodimos, K., Giannopoulos, A., Fountoulakis, M. and Vlahou, A. (2006)
Characterization of the human urine proteome by preparative electrophoresis in
combination with 2-DE. Proteomics 6, 4346–4355.
Urine Sample Preparation 157
17. Pang, J.X., Ginanni, N., Dongre, A.R., Hefta, S.A., and Opiteck, G.J. (2002)
Biomarker discovery in urine by proteomics. Journal of Proteome Research 1,
161–169.
18. Sun, W., Li, F., Wu, S., Wang, X., Zheng, D., Wang, J. and Gao, Y. (2005) Human
urine proteome analysis by three separation approaches. Proteomics 5, 4994–5001.
19. Soldi, M., Sarto, C., Valsecchi, C., Magni, F., Proserpio, V., Ticozzi, D. and
Mocarelli, P. (2005) Proteome profile of human urine with two-dimensional liquid
phase fractionation. Proteomics 5, 2641–2647.
20. Rasmussen, H.H., Orntoft, T.F., Wolf, H. and Celis, J.E. (1996) Towards a compre-
hensive database of proteins from the urine of patients with bladder cancer. The
Journal of Urology 6, 2113–2119.
21. Thongboonkerd, V., McLeish, K.R., Arthur, J.M. and Klein, J.B. (2002) Proteomic
analysis of normal human urinary proteins isolated by acetone precipitation or
ultracentrifugation. Kidney International 62, 1461–1469.
22. Glen, L., Hortin, G.L., Meilinger, B. and Drake, S.K. (2004) Size-selective
extraction of peptides from urine for mass spectrometric analysis. Clinical
Chemistry 50, 1092–1095.
23. Zhang, X., Leung, S., Morris, C.R. and Shigenaga, M.K. (2004) Evaluation
of a novel, integrated approach using functionalized magnetic beads, bench-
top MALDI-TOF-MS with prestructured sample supports, and pattern recog-
nition software for profiling potential biomarkers in human plasma. Journal of
Biomolecular Techniques 15, 167–175.
9
Combining Laser Capture Microdissection
and Proteomics Techniques
Dana Mustafa, Johan M. Kros, and Theo Luider
Summary
Laser microdissection is an effective technique to harvest pure cell populations from
complex tissue sections. In addition to using the microdissected cells in several DNA and
RNA studies, it has been shown that the small number of cells obtained by this technique
can also be used for proteomics analysis. Combining laser capture microdissection and
different types of mass spectrometers opened ways to find and identify proteins that are
specific for various cell types, tissues, and their morbid alterations. Although the combi-
nation of microdissection followed by the currently available techniques of proteomics has
not yet reached the stage of genome wide representation of all proteins present in a tissue,
it is a feasible way to find significant differentially expressed proteins in target tissues.
Recent developments in mass spectrometric detection followed by proper statistics and
bioinformatics enable to analyze the proteome of not more than 100–200 cells. Obviously,
validation of result is essential. The present review describes and discusses the various
methods developed to target cell populations of interest by laser microdissection, followed
by analysis of their proteome.
Key Words: laser capture microdissection; matrix-assisted laser desorption/
ionization; Fourier transformer mass spectrometry; time-of-flight mass spectrometry; liquid
chromatography-electrospray ionization tandem mass spectrometry; two-dimensional
polyacrylamide gel electrophoresis; differential in-gel electrophoresis; protein chip
technology.
Abbreviations: LCM: Laser Capture Microdissection, LMM: Laser Microbeam
Microdissection, LPC: Laser Pressure Catapulting, 2D PAGE: Two-dimensional Polyacry-
lamide Gel Electrophoresis, 2D DIGE: Differential In-gel Electrophoresis, SDS: Sodium
From: Methods in Molecular Biology, vol. 428: Clinical Proteomics: Methods and Protocols
Edited by: A. Vlahou © Humana Press, Totowa, NJ
159
160 Mustafa et al.
Dodecyl Sulphate, MALDI-TOF/MS: Matrix-assisted Laser Desorption/Ionization Time-
of-flight Mass Spectrometry, MALDI-FTMS: Matrix-assisted Laser Desorption/Ionization
Fourier Transformer Mass Spectrometry, LC-ESI-MS/MS: Liquid Chromatography-
Electrospray Ionization Tandem Mass Spectrometry, HPLC: High Performance Liquid
Chromatography, SELDI-TOF: Surface-enhanced Laser Desorption/Ionization Time-of-
flight, ICAT: Isotope-coded Affinity Tag
1. Introduction
Over the last years, significant progress in the analysis of the entire genome
has triggered efforts to further analyze normal and abnormal protein expression
patterns. There is, for instance, an eagerness to discover more and better
diagnostic markers for specific diseases. High expectations of the use of better
biomarkers for the purpose of improving diagnosis and monitoring treatment
initiated technical developments. Human tissues are usually composed of rather
complex mixtures of different cell types. Many techniques have been used
for the isolation of pure cell populations and each technique has its advan-
tages and limitations. For example, immunohistochemistry is an established
and relatively easy technique applicable for localizing protein expression. A
drawback of immunohistochemistry is the impossibility of quantitative assess-
ments of proteins. Another method to obtain information about particular cell
populations is growing cell cultures in order to amplify target cells. Despite
the technical feasibility of this technique, the biological characteristics of the
original cells may not be so accurate in an in vitro environment (1). Alter-
natively, by using xenografts a better mimicking of the normal situation is
reached, but again this method only reflects the real situation of cells in vivo to
some extent (2). Another way of separating cell populations for further inves-
tigation is flow cytometry, which has successfully been applied in the study of
many disease processes. Flow cytometric analysis is applied to cell suspensions
and specific markers for selection of cell population are required. To the best
of our knowledge, the combination of flow cytometry and subsequent mass
spectrometry (MS) has not yet been described for the analysis of solid tissues.
In this review, we discuss methods of cell purification and harvesting
techniques by the use of laser microdissection, which are currently applied for
further MS analysis.
2. Laser Capture Microdissection
In order to select for specific cell populations in heterogeneous tissues,
several microdissection techniques have been described. Most techniques
involve the use of a needle to scrap off cells of interest under direct microscopic
Combining LCM and Proteomics Techniques 161
visualization (3,4). This method, however, tends to be slow, tedious, and highly
operator dependent (2). In 1992, Shibata and coworkers described a new method
of cell isolation. They used a specific pigment placed over small numbers of
cells in a tissue section, which served as an umbrella preventing the covered
cells of being destroyed. Ultraviolet light was used to destroy the DNA/RNA
of the uncovered cells (5). Shortly later, laser capture microdissection (LCM)
under direct microscopic visualization was developed by Liotta and coworkers
in the National Cancer Institute. This way of target cell isolation permits rapid,
reliable laser microdissection to collect specific cell populations from a section
of a complex, heterogeneous tissue (6). For this approach, a tissue section
is placed in a holder of an inverted microscope. A transparent, thermoplastic
polymer coating [e.g., ethylene vinyl acetate (EVA) (7)] is placed in contact
with the tissue. The EVA polymer is positioned over microscopically selected
cell clusters and subsequently the polymer is precisely activated by a near-
infrared laser pulse steered by the investigator. The laser activation of the
polymer results in specific binding to the targeted area. With the removal of
the EVA and the tissue that was bound to it from the section the selected cell
aggregates are isolated for molecular analysis (8). LCM is compatible with
a variety of cellular staining methods and tissue preservation protocols (9).
Dependent on the microlaser dissection device used, the collection caps used are
positioned in different ways. For instance, the caps in the PixCell II (Arcturus
Engineering, Mountain View, CA, USA) technique make contact with the
tissue sections, therefore, strict requirements for preparations are needed. The
PALM microlaser dissector (PALM Microlaser Technologies AG, Bernried,
Germany) provides a powerful separation in which an important application of
the cutting UV-laser is laser microbeam microdissection combined with laser
pressure catapulting (10). A specific glass slides covered with polyethylene
naphthalate membrane will aid in stabilizing the morphological integrity of
the captured area (11) (Fig. 1). In this method, collecting caps do not make
any contact with the tissue sections anymore, which increase the flexibility in
respect to section preparation (12). Both LCM techniques are specific enough
to dissect single cells. The PALM can dissect smaller sections of tissue as
compared to the PixCell system. The two methods of microdissection yield
RNA retrievals of comparable quality and quantity, but they have not been
directly compared with regard to recent developments in protein retrieval for
mass spectrometric applications (13). The collection of large quantities of cells
by LCM is a time consuming procedure requiring the microscopical visual-
ization of the cells of interest in a stained tissue sections before lasering. The
software and the hardware of the different types of laser microdissection are
still developing.
162 Mustafa et al.
objective
Stage
PEN membrane Slide
Tissue section
Cap
Laser
Buffer droplet
Microdissected tissue
Fig. 1. A scheme that represents the principle of laser capture microdissection.
3. LCM and Two-Dimensional Gel Electrophoresis
A new development is the application of LCM for protein retrieval of
tissues for further analysis by proteomic techniques. So far, several approaches
have been performed on cells obtained by laser microdissection. In 2000,
Emmert-Buck and coworkers applied two-dimensional polyacrylamide gel
electrophoresis (2D PAGE) to 50,000 microdissected epithelial cells (14). They
compared tumor cells and normal controls from two patients with oesophageal
cancer (14). Staining the gels with silver yielded the visualization of 675 distinct
proteins and isoforms. Seventeen differentially expressed spots were further
analyzed by MS. This resulted in the identification of two specific proteins,
cytokeratin 1 and annexin I. It was assumed that these proteins were present
in an abundance range of 50,000–1,000,000 copies per cell (14). Using colon
cancer as a model, also Lawrie and coworkers showed the feasibility of investi-
gating protein expression by combining the technologies of LCM and proteome
analysis like 2D PAGE and MS (15).
To overcome the limitation of LCM in producing relatively low numbers of
cells, an extra step has been added to the separation method. In addition to the
2D PAGE from the microdissected cells, an extra 2D PAGE from the whole
section of the same set of samples can be useful. The comparison of silver
stained 2D gels created from microdissected epithelial cells of ovarian cancer
and the 2D gels created from the whole section of the same ovarian samples,
facilitated the discovery of 23 differentially expressed proteins between low
malignant potential and invasive ovarian cancers (16). In-gel digestion of the
specific gel spots followed by MS/MS analysis resulted in the identification of
glyoxalase I, RhoGDI, and a 52 kDa FK506 binding protein (16). In another
study based on 2D PAGE, 315 protein spots were identified by collecting
100,000 cells by LCM of normal and cancer ductal units from breast tissue
Combining LCM and Proteomics Techniques 163
sections (17). Subsequent measurement of the spots by MS resulted in the
identification of 57 differentially expressed proteins between the two groups of
samples (17).
The relative low number of microdissected cells emphasizes the importance
of loading equivalent amounts of protein on the gels. Thus, Shekouh and
coworkers (18) followed a strategy to increase the accuracy of 2D PAGE from
LCM samples. The samples were first separated by one-dimensional sodium
dodecyl sulphate (SDS)-PAGE, stained with silver and subsequently subjected
to densitometry. Evaluation of the staining intensity was used to normalize
the samples. The 2D PAGE silver stained images from 50,000 microdissected
adenocarcinoma cells were compared with the images from whole sections of
pancreatic samples. Spots of their interest were subjected to MALDI-TOF/TOF
MS, resulting in the identification of S100A6 as an over-expressed protein in
pancreatic cancer cells (18). The same methodology has been used to under-
stand the mechanism of a specific molecule such as (HER-2/neu) in breast
cancer (19). Breast cancer tissue was used to microdissect about 50,000–70,000
cells from three HER-2/neu-positive tumors and three HER-2/neu-negative
tumors. This lead to the detection of about 500–600 protein spots in each
gel. The comparison of these two groups allowed the identification of cytok-
eratin 19 (CK19) as an overexpressed protein in HER-2/neu-positive breast
cancer patients (19). In another study, the 2D PAGE of 10,000 microdissected
cells of hepatocellular carcinoma (HCC) samples was compared with normal
surrounding tissue. The investigators visualized about 868 spots of which 20
were considered as differentially expressed proteins. The digestion of these
proteins into peptides was followed by the application of ESI-MS/MS, which
allowed the identification of 11 proteins. Four out of these 11 proteins were
considered as novel candidates of hepatitis B-related HCC markers (20). This
approach of separating the microdissected cells on 2D PAGE followed by in-gel
protein digestion and MS measurements for the identification of biomarkers has
been applied to a wide range of cancers, using various numbers of microdis-
sected cells. There is a range of 10,000–100,000 cells harvested by LCM for
the successful application of 2D electrophoresis (Table 1).
4. LCM and Differential In-Gel Electrophoresis
In 2002, Zhou and coworkers described a new technique called differ-
ential in-gel electrophoresis (DIGE) (21). Two pools of proteins are labeled
with 1-(5-carboxypentyl)-1-propylindocarbocyanine halide (Cy3) N-hydroxy-
succinimidyl ester and 1-(5-carboxypentyl)-1-methylindodi-carbocyanine
halide (Cy5) N-hydroxy-succinimidyl ester fluorescent dyes (21). The labeled
proteins are mixed and separated in the same 2D gel. This strategy improves
Table 1
Overview of Different Methods to Combine Laser Microdissection and Different Proteomics Techniques
Separation
technique Number of
microdissected
cells/sample
Number of
visualized
proteins
Identification
technique Number of
significant
differentially
identified proteins
Number of
samples/study Tissue
used Reference
2D PAGE,
silver
staining
50,000 Approximately
675 distinct
proteins
including
isoforms
Mass spectrometry
and immunoblot
analysis
n= 2; cytokeratin
1 and annexin I 2 cancer samples
and 2 normal
samples
Esophageal
cancer (14)
2D PAGE,
silver
staining
1–5 μg of total
cellular protein Not determined Mass spectrometry
data from all the
protein spots cut
from the gels
n= 3; cytokeratin
8, cytokeratin 18,
and -actin
2 cancer samples
and 2 normal
samples
Colon
cancer (15)
2D PAGE,
silver
staining
50,000 23 differentially
expressed
proteins were
discussed
ESI-MS
identification from
gels made of whole
sections
n= 3; FK506
binding protein,
glyoxalase I, and
RhoGDI
3 invasive OV
and 2 noninvasive
(LMP) OV
Ovarian
cancer (16)
2D PAGE,
silver
staining
100,000 315 protein spots MS identification
from gels made of
whole sections
n= 57 observed
proteins. n=2
after confirmation
6 samples of
DCIS and 6
samples of normal
ductal/lobular
units
Breast
cancer (17)
2D PAGE,
silver
staining
50,000 800 protein spots MALDI-TOF/TOF n=1;
calcium-binding
protein, S100A6
4 cancer samples
and 4 normal
samples
Pancreas
cancer (18)
164
2D PAGE,
silver
staining
50,000–70,000 500–600 protein
spots MALDI-TOF mass
spectrometer n=7;
cytokeratin19,
tropomyosin 3,
aldolase A,
glyoxalase I,
cathepsin D chain
3, albumin, and
MnSOD
3 HER-2/neu-
positive samples
and 3 HER-
2/neu-negative
samples
HER-
2/neu-
positive
breast
cancer
cells
(19)
2D PAGE,
silver
staining
10,000 868 protein spots Nano-flow
ESI-MS/MS n= 11 proteins,
four of them were
novel markers
10 hepatic cancer
cells samples Hepatic
cancer
cells.
hepatitis B
positive
cells
(20)
2D DIGE,
lysine
specific
dyes
250,000 1038–1088
protein spots Capillary LC
tandem mass
analysis
n= 1; tumor
rejection antigen
(gp96)
One sample
contained normal
and one sample
contains cancer
cells
Esophageal
carcinoma (21)
2D DIGE,
lysine
specific
dyes
30,000 1200 protein
spots MALDI-TOF
measurements No further
identifications One sample
contained gastric
mucosa and one
SPEM
Gastric
metaplasia
samples
(22)
2D DIGE,
lysine
specific
dyes
50,000 Not applicable MALDI-TOF
and/or
immunoblotting
for protein
identification
n= 32 Five samples
contained
malignant and
normal breast
tissue
Breast
epithelium
cell
(23)
Continued
165
Table 1
Continued
Separation
technique Number of
microdissected
cells/sample
Number of
visualized
proteins
Identification
technique Number of
significant
differentially
identified proteins
Number of
samples/study Tissue
used Reference
2D DIGE,
cysteine
specific
dyes
5000 ∼1000 protein
spots MALDI-MS
and MS/MS
measurements
n= 40 cultured oncogene-
transduced
epithelial cells and
precancerous
versus cancerous
tissue
Gastric
adenocar-
cinoma
(25)
2D DIGE,
cysteine
specific
dyes
Between 100
and 10
glomeruli,
which equals
to 0.5–3 μg
protein
Between 1400
and 900 protein
spots
Nano
LC-ESI-MS/MS n= 23 between
mice glomeruli
and mice cortex
3 different protein
extracts from
human glomeruli
and 3 independent
isolated glomeruli
and cortex from 3
mice
Kidney
glomeruli (26)
(IPG-IEF)
2D-PAGE
gel
Proteins,
3.8 μg Not applicable Mass spectrometry n= 29 2 samples
contained renal cell
carcinoma and
normal kidney
tissues
Renal
carcinoma (30)
(IPG-IEF)
2D-PAGE
gel
Approximately
<180 ng per
multiplex
protein sample
per 54-cm gel
Not applicable Mass spectrometry Quantitative
differences in 6
progesterone
receptor proteins
12 ER1/PR2 and 12
ER1/PR1 tumors
were grouped into
four pools.
Breast
cancer (31)
166
HPLC
system 10,000 Not applicable ESI mass
spectrometry
followed by
MS/MS
n= 9 3 slides from the
same cell culture Breast
cancer cell
line
(SKBR-3)
(34)
16O/18 O
isotopic
labeling
peptides
10,000 Not applicable The reverse phase
of LC-ESI-MS/MS
on the ion trap
mass spectrum
n= 76 2 samples with
invasive ductal
carcinoma of the
breast
Ductal
carcinoma
of the
breast
(29)
Gel-free
method 30,000–50,000 Not applicable SELDI-TOF/MS n= 1; prostate
carcinoma-
associated protein
(PCa-24)
17 prostate
carcinoma that
contained normal
tissue and BPH
tissue and 7 BPH
samples
Prostate
cancer (41)
Gel-free
method ∼2000 Not applicable MALDI-TOF/MS n= 2; calgranulin
A and chaperonin
10
8 endometrioid
adenocarcinomas,
4 proliferative
endometria, and
4 secretory
endometria
Endometrial
cancer (36)
Gel-free
method 150 Not applicable MALDI-TOF/MS No protein
identifications.
Unique peptide
pattern of ∼35
peptides for
trophoblast and
stroma cells
1 placenta sample
contained
trophoblasts and
surrounding
stroma cells.
Placenta
samples (37)
Continued
167
Table 1
Continued
Separation
technique Number of
microdissected
cells/sample
Number of
visualized
proteins
Identification
technique Number of
significant
differentially
identified proteins
Number of
samples/study Tissue
used Reference
Gel-free
method 2000–2400 Not applicable MALDI-TOF/TOF
mass spectrometry No protein
identifications. 9
differentially
expressed peptides
6 invasive ductal
breast carcinoma
contained cancer
and normal cells
Breast
cancer (38)
Gel-free
method 3000 Not applicable Nano LC-FTICR
mass spectrometry n= 1003 proteins
identified 2 replicate samples
of breast cancer
epithelial cells
Breast
cancer Umar
et al.,
2006
ProteinChip
technology 3000–5000 Not applicable Isolation by
two-dimensional
gel electrophoresis
and tandem mass
spectrometry
analysis
n= 1; annexin V 57 head and neck
tumor samples and
44 mucosa samples
Head and
nick
cancer
(40)
ProteinChip
technology 3000–5000 Not applicable Isolation by
reverse-phase
chromatography
and SDS-PAGE
then identified by
MS/MS analysis
n= 1; heat shock
protein 10 39 colorectal tumor
samples, 40 normal
mucosa samples,
and 29 adenoma
samples
Colorectal
cancer (39)
Abbreviations: 2DE: 2 dimensional gel electrophoresis, OV: ovarian cancer, LMP: low malignant potential, DCIS: ductal/lobular units
and ductal carcinoma in situ, HCC: hepatocellular carcinoma, BPH: benign prostatic hyperplasia, SPEM: spasmolytic polypeptide expressing
metaplasia, PR: progesterone receptor, ER: estrogen receptor
168
Combining LCM and Proteomics Techniques 169
the sensitivity of detection and enlarges the range of candidate proteins
for detection. Molecular weight- and charge-matched cyanine dyes enable
multiplex labeling with different samples run on the same gel. The same inves-
tigators described a powerful tool for the molecular characterization of cancer
progression and identification of cancer-specific protein markers by combining
2D DIGE with MS. They compare the 2D DIGE of about 250,000 microdis-
sected cells from oesophageal carcinoma with normal epithelial cells from
the oesophagus. The cancer cell lysate yielded 1038 protein spots while the
normal epithelial lysate yielded 1088 protein spots. In-gel digestion of the
differentially expressed protein spots was followed by capillary high perfor-
mance liquid chromatography (HPLC) tandem mass analysis to achieve further
identification. This way, tumor rejection antigen (gp96) was found to be
upregulated in oesophageal squamousal cell cancer (21). Applying the same
procedure to smaller numbers of microdissected cells from biopsy samples
with gastric metaplasia appeared to be successful as well (22). Approximately
1200 spots were identified from 30,000 microdissected cells. Twenty-eight of
these spots were over expressed in the metaplasia samples as compared to
the normal surface cells (22). However, subsequent MALDI-TOF measure-
ments of the spots did not result in the identification of proteins. The same
procedure was applied to 50,000 microdissected cells resulting in the identi-
fication of 32 proteins in breast epithelial cancer cells (23), of which thirteen
had not been associated previously with the tumors (23). One technical aspect
of the 2D DIGE method needs special attention: the nature of the fluorescent
dyes and their ability to bind to lysine residues only (21). Proteins with high
percentages of lysine residues can be labeled more efficiently as compared to
proteins containing little or no lysine. By developing a new generation of dyes
reacting with cysteine residues, the sensitivity of DIGE has been improved (24).
Although cysteine is less abundant than lysine in proteins in general, cysteine
labeling can be carried to saturation. Lysine labeling must be limited to 1–3%
of all the residues to prevent loss of solubility when bulky hydrophobic dyes
are coupled to the polar lysine residues (24). Greengauz-Roberts and coworkers
applied the saturated labeling for cysteine residues to study about 5000 cells
obtained by LCM of metaplasia and cancer cells. A total of 1471 distinct protein
features were observed from the relatively small number of cells. Ninety-six of
these spots were further identified. Using MALDI-MS and MS/MS measure-
ments in addition to the specific position of the protein in the gel resulted in the
identification of 42 proteins in cancer samples (25). Also Sitek and coworkers
described a novel approach to analyze glomerular proteins from mice and
human samples using DIGE saturation labeling (26). Only 10 glomeruli (0.5 μg)
picked by LCM from a slide of a human kidney biopsy appeared to be suffi-
cient to visualize 900 spots using DIGE technique (26). 2D DIGE holds several
170 Mustafa et al.
advantages over the conventional 2D gel. One of the most important advantages
is the improvement of the reproducibility of 2D DIGE method. The gel-to-gel
differences are minimalized because the separation of the pooled samples takes
place in the same gel. Therefore, the comparison of protein expression from
two cell populations or samples can be more accurately assessed and easier to
be identified. The quantitative differences of protein contents are also better
measured by the application of fluorescent dyes. In addition, 2D DIGE enables
a higher throughput analysis of 2D gels by its feasibility to automatic gel
imaging. Importantly, labeling of proteins by fluorescent dyes did not affect the
protein identification by MS, because only small percentages of the molecules
of each protein are labeled. Importantly, for 2D DIGE the number of microdis-
sected cells, which are required for protein identification is less as compared
to the other 2D electrophoresis techniques (Table 1).
5. LCM and Different Labeling Techniques
The comparison of the proteome of two different samples (for instance,
normal and tumor cells) is facilitated by labeling. In 2004, Li and coworkers
described a method for qualitative and quantitative protein analysis by
combining LCM with isotope-coded affinity tag labeling technology and two-
dimensional liquid chromatography coupled with tandem mass spectroscopy
(2D-LC-MS/MS) (27). Approximately 50,000–100,000 cells of HCC and
nonHCC hepatocytes were microdissected and a total of 644 proteins in
HCC hepatocytes were qualitatively determined, and 261 differential proteins
between the two groups were quantified (28). In 2004, 16O/18 O isotopic labeled
peptides were generated from 10,000 microdissected cells of ductal carcinoma
of the breast. The approach allowed the identification of 76 proteins (29).
By using reverse phase liquid chromatography-electrospray ionization tandem
mass spectrometry (LC-ESI-MS/MS) Zang and coworkers were able to identify
proteins that were significantly upregulated in the breast tumor cells (29).
Separating the radioactive labeled peptides on the high resolution 54 cm serial
immobilized pH gradient isoelectric focusing 2D-PAGE gel provided a precise
estimate of the abundance ratio for proteins from two samples (30). The radio-
iodination of 3.8 μg renal carcinoma proteins and 3.8 μg normal kidney proteins
with both 125I and 131 I followed by mass spectrometric identification revealed
29 differentially expressed proteins (30). Applying the same methodology of
radioactive labeling to a pool of microdissected breast cancer cells provided
a sensitive method to identify some differentially expressed proteins in corre-
lation with the presence of progesterone receptor in estrogens receptor-positive
breast cancer (31).
Combining LCM and Proteomics Techniques 171
6. Combining LCM and Different Separation Methods
It has been shown previously that the number of detected and identified
peptides and proteins increases significantly by coupling MALDI-MS (32)
and ESI-MS (33) to a peptide or protein separation system. In 2003, Wu and
coworkers described a method for discovering biomarkers from microdissected
homogeneous cells from breast cancer cell lines (34). Following capturing
the cells, the peptide digest was fractionated by reversed phase HPLC and
analyzed by ion trap MS (34). HPLC fractionation of about 10,000 endothelial
cells from a breast cancer cell line (SKBR-3) followed by ESI MS resulted
in the identification of low-expressed proteins in the cell line. Capillary
isoelectric focusing combined with the reverse phase nano-LC in an automated
and integrated platform provides systematic resolution of complex peptide
mixtures generated from limited protein quantities (7). This method separated
the mixture of peptides based on differences in isoelectric points and hydropho-
bicity, and it eliminates peptide loss and analyte dilution (7). This method
of separation coupled to ESI-tandem MS assists in the detection of 6866
peptides, leading to the identification of 1820 proteins from 20,000 microdis-
sected cells of glioblastoma (7). In order to increase the number of identified
proteins from LCM of brain samples, Gozal and coworkers added an extra
separation step (35). After collecting cells by LCM, the total protein were
extracted and resolved on an SDS gel. Gels were cut out into multiple pieces
followed by trypsin digestion. Peptides were subjected to highly sensitive liquid
chromatography-tandem mass spectrometry (LC-MS/MS). This way resulted
in identifying hundreds to thousands of proteins (35).
7. LCM and Gel-Free Mass Spectrometry
There are possibilities of measuring the peptide digest of cells harvested by
LCM directly by MS, without an initial separation step on 2D PAGE (known as
“gel-free MS”). Guo and coworkers directly analyzed endometrial epithelium
cells obtained by LCM using matrix-assisted laser desorption/ionization time-
of-flight mass spectrometry (MALDI-TOF/MS) (36). A total of 16 physio-
logic and malignant endometrial samples including four proliferative and four
secretory endometria, and eight endometrioid adenocarcinomas were used for
this study. Approximately 2000 cells appeared to be sufficient to confirm
overexpression of two proteins, calgranulin A and chaperonin 10 in the
epithelial cells of endometrial adenocarcinoma samples (36). In another study,
the direct analysis of 125 trophoblast and stroma cells of placental tissue resulted
in the detection of significant expressed protein differences between these two
cell types (37). Also, differentially expressed proteins between breast cancer
and normal samples can be detected by direct MALDI-TOF/MS measurements
172 Mustafa et al.
of 2000–2400 LCM cells (38). In a recent study, it was possible to identify
over 1000 proteins from 3000 microdissected cells by the combination of
advanced nanoLC and high resolution Fourier transformer mass spectrometry
(FTMS) (39).
8. LCM and Protein Chip Technology
There are currently two approaches to produce arrays capable of generating
protein network information. The first method is the forward phase array in
which each spot on the slide represents a specific antibody. Therefore, the array
is incubated with only one test sample (9). The second method is the reverse
phase array in which each spot represents an individual test sample, and the
array is composed of multiple, different samples, which then can be tested
under the same experimental conditions. In addition, when the arrays are probed
separately with two different classes of antibodies, it is possible to specifically
detect the total and phosphorylated forms of the protein of interest (9).By
combining LCM technique to protein chip technology, Melle and coworkers
identified annexin V as a specific protein in head and neck cancer patients,
and heat shock protein 10 as a biomarker in colorectal cancer patients (40,41).
The protein lysates from 3000 to 5000 microdissected cells were analyzed on
both strong anion exchange arrays and weak cation exchange arrays, followed
by separation steps (e.g., 2D gel or reverse phase chromatography and SDS-
PAGE), MS measurements, and MS/MS analysis (40,41). In both cases, a
validation step by immunohistochemistry confirmed their findings.
In other studies surface-enhanced laser desorption/ionization time-of-flight
analysis was applied to microdissected cells because of its sensitivity to
smaller amounts of material than other techniques such as 2D gel (42). Using
30,000–50,000 cells of prostate carcinoma specimens, the unique expression
of prostate carcinoma-associated protein, called PCa-24 in the epithelial cells,
was reached (42). Protein microarrays hold several technical challenges (43).
Their application offers the advantage of scalability, flexibility, and automatic
processing (43). Arrays may also enable the control of key parameters such as
temperature, pH, and cofactor concentration, which are not easily afforded by
cell-based systems.
9. Perspectives of LCM and Mass Spectrometry Analysis
The use of LCM of (relatively) pure populations of cells to be used for
further analysis of their proteome is an important addition to the arsenal of
techniques in bioscience. However, this technique is still time consuming and
yield relatively small numbers of cells. To overcome this problem, alternative
Combining LCM and Proteomics Techniques 173
1277.71354
1475.75278
1707.77693
1994.98513
2151.08736
2368.27262
2511.14239
2706.17286
2903.42238 3265.53235
1000 1500 2000 2500 3000 3500 m/z
0.0
0.5
1.0
1.5
×10
7
Intens.
+MS
1726.89642
1793.73840
1818.99943
1840.98089
1859.95483
1873.94999
1891.97950
1943.95115
1963.92507
1978.96298
1999.99082
2025.94879
1700 1750 1800 1850 1900 1950 2000 m/z
0.0
0.2
0.4
0.6
0.8
1.0
×10
6
Intens.
+MS
GAPDH
fibrinogen
GFAP
fibrinogen
Hb
alpha 2
CD34 antigen
Tubulin
Fig. 2. MALDI FTMS spectrum obtained from 150 microdissected cells from a
frozen glioma tissue sample. The spectrum contains approximately thousand monoiso-
topic peaks between 700 and 3000 m/zat relative high peak intensities. The small box
is a zoom in for a small part of the spectra, between 1700 and 2000 m/z. It shows the
very high numbers of peaks obtained from measuring a very small number of cells.
The peaks can be identified by different sequencing MS techniques; some examples of
identified peptides are indicated in the spectrum.
steps of processing tissues are needed. Sample collection and preparation is
crucial. During the microdissection procedure, special attention should be taken
to prevent waist and contamination of target material. For instance, material
should not drop from, or stick to, the cap of the tubes used. Another consid-
eration is to minimize the steps of transferring the collected material from one
tube into the other. Therefore, the use of low protein binding tubes is recom-
mended. A protocol for sample preparation is included in this chapter (Box 1).
The 2D PAGE is a well-established technique that had been used in combi-
nation with LCM in many studies so far. The need of relative large numbers of
cells blocks the possibility to measure large numbers of samples as indicated
in Table 1. In addition, the relative low reproducibility hampers sound statis-
tical analysis. 2D DIGE improves reproducibility and also lowers the required
amount of microdissected tissue. However, this technique is suitable for exper-
imental research only.
174 Mustafa et al.
LCM sample preparation protocol:
Cryosections of 8 μm were made from glioma braintumor tissue and
mounted on polyethylene naphthalate covered glass slides (PALM Micro-
laser Technologies AG, Bernried, Germany) as described previously (38). The
slides were fixed in 70% ethanol and stored at (–20(C for not more than 2
days. After fixation and immediately before microdissection, the slides were
washed twice with Milli-Q water, stained for 10 s in haematoxylin, washed
again twice with Milli-Q water and subsequently dehydrated in a series of 50,
70, 95, and 100% ethanol solution and air dried. The PALM laser microdis-
section and pressure catapulting device, type P-MB was used with PalmRobo
v2.2 software at 40× magnification. Estimating that a cell has a volume of
10 × 10 × 10 μm, we microdissected an area of about 190,000 μm2of blood
vessels and another area of the same size of the surrounding tumor tissue from
each sample, resulting in approximately 1500 cells per sample. The microdis-
sected cells were collected in caps of PALM tubes in 5 μl of 0.1% RapiGest
buffer (Waters, Milford, MA, USA). The caps were cut and placed onto
0.5 ml Eppendorf protein LoBind tubes (Eppendorf, Hamburg, Germany).
Subsequently, these tubes were centrifuged at 12,000 g for 5 min. To make
sure that all the cells were covered with buffer, another 5 μl of RapiGest
was added to the cells. All samples were stored at –80°C. After thawing
the microdissected tissue, the tissue was disrupted by external sonification
for 1 min at 70% amplitude at a maximum temperature of 25°C (Bransons
Ultrasonics, Danbury, USA). The samples were incubated at 37 and 100°C
for 5 and 15 min, respectively, for protein solubilization and denaturation.
To each sample, 1.5 μl of 100 ng/μl gold grade trypsin (Promega, Madison,
WI, USA) in 3 mM Tris–HCL diluted 1:10 in 50 mM NH4HCO3was added
and incubated overnight at 37°C for protein digestion. To inactivate trypsin
and to degrade the RapiGest, 2 μl of 500 mM HCL was added and incubated
for 30 min at 37°C. Samples were dried in a Speedvac (Thermo Savant,
Holbrook, NY, USA) and reconstituted in 5 μl of 50% acetonitrile/0.5% triflu-
oroacetic acid/water prior to measurement. Samples were used for immediate
measurements, or stored for a maximum of 10 days at 4°C.
Recently, the improvement of resolution and detection limits in modern mass
spectrometers, particularly in FTMS, opened a new research field to analyze
small numbers of microdissected cells (in the range of 200–5000). FTMS
has specific characteristics, unrivalled high mass resolution (in the order of
100,000–1,000,000), high mass accuracy (below 1 ppm), dynamics (three to
four orders of magnitude), and its good signal to noise ratio (44). These features
facilitate combining this technique with LCM. For instance, by MALDI-FTMS,
Combining LCM and Proteomics Techniques 175
peptide digests of no more than 150 cells taken from biological samples (e.g.,
glioma vessel tissue) resulted in informative mass spectra (Fig. 2). It is expected
that techniques like FTMS soon will be implicated in the practice of routine
laboratories for the detection of disease-related proteins in clinical specimens.
References
1. Zhang, L., Zhou, W., Velculescu, V. E., Kern, S. E., Hruban, R. H., Hamilton, S. R.,
Vogelstein, B. and Kinzler, K. W. (1997) Gene expression profiles in normal and
cancer cells. Science 276, 1268–1272.
2. Curran, S., McKay, J. A., McLeod, H. L. and Murray, G. I. (2000) Laser capture
microscopy. Mol Pathol 53, 64–68.
3. Going, J. J. and Lamb, R. F. (1996) Practical histological microdissection for PCR
analysis. J Pathol 179, 121–124.
4. Zhuang, Z., Bertheau, P., Emmert-Buck, M. R., Liotta, L. A., Gnarra, J., Linehan,
W. M. and Lubensky, I. A. (1995) A microdissection technique for archival DNA
analysis of specific cell populations in lesions <1 mm in size. Am J Pathol 146,
620–625.
5. Shibata, D., Hawes, D., Li, Z. H., Hernandez, A. M., Spruck, C. H. and
Nichols, P. W. (1992) Specific genetic analysis of microscopic tissue after selective
ultraviolet radiation fractionation and the polymerase chain reaction. Am J Pathol
141, 539–543.
6. Emmert-Buck, M. R., Bonner, R. F., Smith, P. D., Chuaqui, R. F., Zhuang, Z.,
Goldstein, S. R., Weiss, R. A. and Liotta, L. A. (1996) Laser capture microdis-
section. Science 274, 998–1001.
7. Wang, Y., Rudnick, P. A., Evans, E. L., Li, J., Zhuang, Z., Devoe, D. L., Lee, C. S.
and Balgley, B. M. (2005) Proteome analysis of microdissected tumor tissue using a
capillary isoelectric focusing-based multidimensional separation platform coupled
with ESI-tandem MS. Anal Chem 77, 6549–6556.
8. Suarez-Quian, C. A., Goldstein, S. R., Pohida, T., Smith, P. D., Peterson, J. I.,
Wellner, E., Ghany, M. and Bonner, R. F. (1999) Laser capture microdissection of
single cells from complex tissues. Biotechniques 26, 328–335.
9. Espina, V., Milia, J., Wu, G., Cowherd, S. and Liotta, L. A. (2006) Laser capture
microdissection. Methods Mol Biol 319, 213–229.
10. Schutze, K., Posl, H. and Lahr, G. (1998) Laser micromanipulation systems as
universal tools in cellular and molecular biology and in medicine. Cell Mol Biol
(Noisy-le-grand) 44, 735–746.
11. Gillespie, J. W., Gannot, G., Tangrea, M. A., Ahram, M., Best, C. J., Bichsel, V. E.,
Petricoin, E. F., Emmert-Buck, M. R. and Chuaqui, R. F. (2004) Molecular profiling
of cancer. Toxicol Pathol 32(Suppl. 1), 67–71.
12. Niyaz, Y., Stich, M., Sagmuller, B., Burgemeister, R., Friedemann, G., Sauer, U.,
Gangnus, R. and Schutze, K. (2005) Noncontact laser microdissection and pressure
catapulting: sample preparation for genomic, transcriptomic, and proteomic
analysis. Methods Mol Med 114, 1–24.
176 Mustafa et al.
13. Ball, H. J. and Hunt, N. H. (2004) Needle in a haystack: microdissecting the
proteome of a tissue. Amino Acids 27, 1–7.
14. Emmert-Buck, M. R., Gillespie, J. W., Paweletz, C. P., Ornstein, D. K., Basrur, V.,
Appella, E., Wang, Q. H., Huang, J., Hu, N., Taylor, P. and Petricoin, E. F. 3rd (2000)
An approach to proteomic analysis of human tumors. Mol Carcinog 27, 158–165.
15. Lawrie, L. C., Curran, S., McLeod, H. L., Fothergill, J. E. and Murray, G. I. (2001)
Application of laser capture microdissection and proteomics in colon cancer. Mol
Pathol 54, 253–258.
16. Jones, M. B., Krutzsch, H., Shu, H., Zhao, Y., Liotta, L. A., Kohn, E. C. and
Petricoin, E. F. 3rd (2002) Proteomic analysis and identification of new biomarkers
and therapeutic targets for invasive ovarian cancer. Proteomics 2, 76–84.
17. Wulfkuhle, J. D., Sgroi, D. C., Krutzsch, H., McLean, K., McGarvey, K.,
Knowlton, M., Chen, S., Shu, H., Sahin, A., Kurek, R., Wallwiener, D.,
Merino, M. J., Petricoin, E. F. 3rd, Zhao, Y. and Steeg, P. S. (2002) Proteomics
of human breast ductal carcinoma in situ. Cancer Res 62, 6740–6749.
18. Shekouh, A. R., Thompson, C. C., Prime, W., Campbell, F., Hamlett, J., Herrington,
C. S., Lemoine, N. R., Crnogorac-Jurcevic, T., Buechler, M. W., Friess, H.,
Neoptolemos, J. P., Pennington, S. R. and Costello, E. (2003) Application of laser
capture microdissection combined with two-dimensional electrophoresis for the
discovery of differentially regulated proteins in pancreatic ductal adenocarcinoma.
Proteomics 3, 1988–2001.
19. Zhang, D. H., Tai, L. K., Wong, L. L., Sethi, S. K. and Koay, E. S. (2005)
Proteomics of breast cancer: enhanced expression of cytokeratin19 in human
epidermal growth factor receptor type 2 positive breast tumors. Proteomics 5,
1797–1805.
20. Ai, J., Tan, Y., Ying, W., Hong, Y., Liu, S., Wu, M., Qian, X. and Wang, H. (2006)
Proteome analysis of hepatocellular carcinoma by laser capture microdissection.
Proteomics 6, 538–546.
21. Zhou, G., Li, H., DeCamp, D., Chen, S., Shu, H., Gong, Y., Flaig, M.,
Gillespie, J. W., Hu, N., Taylor, P. R., Emmert-Buck, M. R., Liotta, L. A.,
Petricoin, E. F. 3rd and Zhao, Y. (2002) 2D differential in-gel electrophoresis for
the identification of esophageal scans cell cancer-specific protein markers. Mol
Cell Proteomics 1, 117–124.
22. Lee, J. R., Baxter, T. M., Yamaguchi, H., Wang, T. C., Goldenring, J. R. and
Anderson, M. G. (2003) Differential protein analysis of spasomolytic polypeptide
expressing metaplasia using laser capture microdissection and two-dimensional
difference gel electrophoresis. Appl Immunohistochem Mol Morphol 11, 188–193.
23. Hudelist, G., Singer, C. F., Pischinger, K. I., Kaserer, K., Manavi, M., Kubista, E.
and Czerwenka, K. F. (2006) Proteomic analysis in human breast cancer: identifi-
cation of a characteristic protein expression profile of malignant breast epithelium.
Proteomics 6, 1989–2002.
24. Shaw, J., Rowlinson, R., Nickson, J., Stone, T., Sweet, A., Williams, K. and
Tonge, R. (2003) Evaluation of saturation labelling two-dimensional difference gel
electrophoresis fluorescent dyes. Proteomics 3, 1181–1195.
Combining LCM and Proteomics Techniques 177
25. Greengauz-Roberts, O., Stoppler, H., Nomura, S., Yamaguchi, H.,
Goldenring, J. R., Podolsky, R. H., Lee, J. R. and Dynan, W. S. (2005) Saturation
labeling with cysteine-reactive cyanine fluorescent dyes provides increased sensi-
tivity for protein expression profiling of laser-microdissected clinical specimens.
Proteomics 5, 1746–1757.
26. Sitek, B., Potthoff, S., Schulenborg, T., Stegbauer, J., Vinke, T., Rump, L. C.,
Meyer, H. E., Vonend, O. and Stuhler, K. (2006) Novel approaches to analyse
glomerular proteins from smallest scale murine and human samples using DIGE
saturation labelling. Proteomics 6, 4337–4345.
27. Li, C., Hong, Y., Tan, Y. X., Zhou, H., Ai, J. H., Li, S. J., Zhang, L., Xia, Q. C.,
Wu, J. R., Wang, H. Y. and Zeng, R. (2004) Accurate qualitative and quanti-
tative proteomic analysis of clinical hepatocellular carcinoma using laser capture
microdissection coupled with isotope-coded affinity tag and two-dimensional liquid
chromatography mass spectrometry. Mol Cell Proteomics 3, 399–409.
28. Gygi, S. P., Rist, B., Gerber, S. A., Turecek, F., Gelb, M. H. and Aebersold, R.
(1999) Quantitative analysis of complex protein mixtures using isotope-coded
affinity tags. Nat Biotechnol 17, 994–999.
29. Zang, L., Palmer Toy, D., Hancock, W. S., Sgroi, D. C. and Karger, B. L. (2004)
Proteomic analysis of ductal carcinoma of the breast using laser capture microdis-
section, LC-MS, and 16O/18O isotopic labeling. J Proteome Res 3, 604–612.
30. Poznanovic, S., Wozny, W., Schwall, G. P., Sastri, C., Hunzinger, C.,
Stegmann, W., Schrattenholz, A., Buchner, A., Gangnus, R., Burgemeister, R. and
Cahill, M. A. (2005) Differential radioactive proteomic analysis of microdissected
renal cell carcinoma tissue by 54 cm isoelectric focusing in serial immobilized pH
gradient gels. J Proteome Res 4, 2117–2125.
31. Neubauer, H., Clare, S. E., Kurek, R., Fehm, T., Wallwiener, D., Sotlar, K.,
Nordheim, A., Wozny, W., Schwall, G. P., Poznanovic, S., Sastri, C.,
Hunzinger, C., Stegmann, W., Schrattenholz, A. and Cahill, M. A. (2006)
Breast cancer proteomics by laser capture microdissection, sample pooling, 54-
cm IPG IEF, and differential iodine radioisotope detection. Electrophoresis 27,
1840–1852.
32. Preisler, J., Hu, P., Rejtar, T., Moskovets, E. and Karger, B. L. (2002) Capillary
array electrophoresis-MALDI mass spectrometry using a vacuum deposition
interface. Anal Chem 74, 17–25.
33. Bergstrom, S. K., Samskog, J. and Markides, K. E. (2003) Development
of a poly(dimethylsiloxane) interface for on-line capillary column liquid
chromatography-capillary electrophoresis coupled to sheathless electrospray
ionization time-of-flight mass spectrometry. Anal Chem 75, 5461–5467.
34. Wu, S. L., Hancock, W. S., Goodrich, G. G. and Kunitake, S. T. (2003) An approach
to the proteomic analysis of a breast cancer cell line (SKBR-3). Proteomics 3,
1037–1046.
35. Gozal, Y. M., Cheng, D., Duong, D. M., Lah, J. J., Levey, A. I. and Peng, J. (2006)
Merger of laser capture microdissection and mass spectrometry: a window into the
amyloid plaque proteome. Methods Enzymol 412, 77–93.
178 Mustafa et al.
36. Guo, J., Colgan, T. J., DeSouza, L. V., Rodrigues, M. J., Romaschin, A. D.
and Siu, K. W. (2005) Direct analysis of laser capture microdissected endome-
trial carcinoma and epithelium by matrix-assisted laser desorption/ionization mass
spectrometry. Rapid Commun Mass Spectrom 19, 2762–2766.
37. de Groot, C. J., Steegers-Theunissen, R. P., Guzel, C., Steegers, E. A. and
Luider, T. M. (2005) Peptide patterns of laser dissected human trophoblasts
analyzed by matrix-assisted laser desorption/ionisation-time of flight mass
spectrometry. Proteomics 5, 597–607.
38. Umar, A., Dalebout, J. C., Timmermans, A. M., Foekens, J. A. and Luider, T.
M. (2005) Method optimisation for peptide profiling of microdissected breast
carcinoma tissue by matrix-assisted laser desorption/ionisation-time of flight
and matrix-assisted laser desorption/ionisation-time of flight/time of flight-mass
spectrometry. Proteomics 5, 2680–2688.
39. Umar, A., Luider, T. M., Foekens, J. A. and Pasa-Tolic, L. (2007) NanoLC-FT-
ICR Ms improves proteome coverage attainable for approximately 3000 laser-
microdissected breast carcinoma cells. Proteomics 7, 323–329.
40. Melle, C., Bogumil, R., Ernst, G., Schimmel, B., Bleul, A. and von Eggeling, F.
(2006) Detection and identification of heat shock protein 10 as a biomarker in
colorectal cancer by protein profiling. Proteomics 6, 2600–2608.
41. Melle, C., Ernst, G., Schimmel, B., Bleul, A., Koscielny, S., Wiesner, A.,
Bogumil, R., Moller, U., Osterloh, D., Halbhuber, K. J. and von Eggeling, F.
(2003) Biomarker discovery and identification in laser microdissected head and
neck squamous cell carcinoma with ProteinChip technology, two-dimensional gel
electrophoresis, tandem mass spectrometry, and immunohistochemistry. Mol Cell
Proteomics 2, 443–452.
42. Zheng, Y., Xu, Y., Ye, B., Lei, J., Weinstein, M. H., O’Leary, M. P., Richie, J. P.,
Mok, S. C. and Liu, B. C. (2003) Prostate carcinoma tissue proteomics for
biomarker discovery. Cancer 98, 2576–2582.
43. Cutler, P. (2003) Protein arrays: the current state-of-the-art. Proteomics 3, 3–18.
44. Dekker, L. J., Burgers, P. C., Guzel, C. and Luider, T. M. (2007) Ftms and
TOF/TOF mass spectrometry in concert: identifying peptides with high reliability
using matrix prespotted MALDI target plates. J Chromatogr B Analyt Technol
Biomed Life Sci 847, 62–64.
45. Mustafa, D. A., Burgers, P. C., Dekker, L. J., Charif, H., Titulaer, M. K.,
Smitt, P. A., Luider, T. M. and Kros, J. M., (2007) Identification of glioma
neovascularization-related proteins by using MALDI-FTMS and nano-LC fraction-
ation to microdissected tumor vessels. Mol Cell Proteomics 6, 1147–1157.
III
Clinical Proteomics by LC-MS Approaches
10
Comparison of Protein Expression by Isotope-Coded
Affinity Tag Labeling
Zhen Xiao and Timothy D. Veenstra
Summary
Isotope-coded affinity tag (ICAT) labeling, in combination with mass spectrometry
(MS), has been widely adopted as an effective method for comparing protein abundance
levels. This chapter describes the ICAT labeling procedure in search for the celecoxib-
regulated proteins in a colon cancer cell line. Celecoxib, a cyclooxygenase-2 (COX-2)
specific inhibitor, is used as a colorectal cancer preventative drug in clinical trials. Here,
celecoxib is used to inhibit the expression of COX-2 in a colon cancer cell line HT-29.
To elucidate the proteomic changes induced by celecoxib, the protein lysates from the
treated and control cells are prepared. The cysteine-containing proteins are labeled with the
heavy and light ICAT reagents, respectively. The labeled proteins are then combined and
digested with trypsin. The ICAT-labeled peptides are subject to the purification through
an avidin column and eventually the cleavage of the biotin tags. This chapter focuses on
the ICAT labeling procedure itself, because sample preparation is the most critical step of
an ICAT-based protein expression comparison experiment. Other related procedures such
as the cation exchange high performance liquid chromatography separation of peptides
and MS analysis are detailed elsewhere in this book.
Key Words: isotope-coded affinity tags; quantitative proteomics; mass spectrometry.
1. Introduction
The application of mass spectrometry (MS) has rapidly expanded from
simple identification of protein components to the quantitative comparison
of proteomic changes under various biological and physiological conditions
(1,2,3). In many studies, it is desirable to identify proteins and quantify their
From: Methods in Molecular Biology, vol. 428: Clinical Proteomics: Methods and Protocols
Edited by: A. Vlahou © Humana Press, Totowa, NJ
181
182 Xiao and Veenstra
levels simultaneously using MS. While the ability to target specific molecules
for quantitation is well established, there are experimental and technical issues
that limit the accuracy of direct quantitation of hundreds (or thousands) of
species in a single MS experiment and make it extremely challenging (4,5,6,7).
To resolve this hurdle, a variety of chemical-based labeling and derivatization
techniques have been developed (5,7,8,9). One of these techniques, isotope-
coded affinity tags (ICATs), has been widely adopted and remains the model
system by which most other differential labeling methods have been developed
(10). The structure of the reagent used in ICAT studies is composed of four
parts: (1) an iodoacetamide group that covalently reacts with cysteine residues
within proteins; (2) an isotope-coded linker regions, which is prepared in two
distinct versions containing either nine 13C (heavy version) or nine 12 C (light
version); (3) a biotin tag that facilitates the purification of labeled peptides via
its specific binding to avidin; and (4) an acid-labile bond that is situated between
the biotin and isotopically differential domain of the reagent (Fig. 1). After
labeling the cysteine residues, the protein mixture is enzymatically digested
(usually with trypsin) and the labeled peptides purified via avidin chromatog-
raphy. Following the enrichment of the ICAT-labeled peptides, the cleavable
linker and the biotin tag are removed using trifluoroacetic acid (TFA). The
removal of the biotin tag reduces the mass of the remaining tag attached to the
peptide and increases the fragmentation efficiency and ultimately the success
rate of peptide identification by tandem MS.
The advantage of ICAT labeling is the identical chemistry, yet differ-
ential mass, of the heavy and light reagents, which enables the protein
abundances within two complex proteome samples to be compared simul-
taneously. Following their coelution from a nanoflow reversed-phase liquid
chromatography column, the light- and heavy-labeled peptides are easily recog-
nized within the mass spectrum, being separated by ∼9 Da. The tandem MS
spectrum enables the peptide to be identified, while the ratio of the areas of
each peak is used as a measurement of the peptide’s relative abundance in
the samples being compared. Since its inception, the ICAT reagents have been
modified, improved, and made available commercially via applied biosystems
Fig. 1. The structure of cleavable isotope-coded affinity tag reagent.
Isotope-Coded Affinity Tag Labeling 183
as a kit (11). The combination of ICAT labeling, peptide fractionation, and
the liquid chromatography tandem mass spectrometry has enabled the rapid
and simultaneous identification and quantitation of changes in complex protein
mixtures (12,13,14,15,16).
In this chapter, the ICAT labeling procedure is described as part of an exper-
iment to identify celecoxib-induced proteomic changes in colon cancer cells.
Celecoxib is a nonsteroidal anti-inflammatory drug that specifically inhibits
cyclooxygenase-2 (COX-2) (17,18). In clinical trials, it has been shown to
inhibit the development of precancerous polyposis in colon (19,20). In this
study, a COX-2 expressing colon cancer cell line (HT-29) is used (21,22).
After treating the cells with celecoxib, cell lysate would be prepared and
labeled with the ICAT reagents. A schematic diagram of the ICAT labeling
and peptide analysis procedure is shown in Fig. 2. Since the core of the ICAT-
based quantitative proteomic analysis is sample preparation, this chapter is
dedicated to the details of the ICAT labeling protocol itself. For information
on strong cation exchange (SCX) high performance liquid chromatography
(HPLC) separation of peptides, analysis by nanoflow reversed-phase liquid
chromatography tandem mass spectrometry, and bioinformatics analysis, refer
to the chapter on “Analysis of the Extracellular Matrix and Secreted Vesicle
Proteomes by Mass Spectrometry,” (Subheadings 3.6–3.8). The methods
described in this chapter can be used to (1) understand the proteomic changes
in response to drug; (2) illustrate the molecular mechanisms underlying the
drug effects; and (3) search for biomarkers or endpoints that can be used to
monitor and evaluate the therapeutic and intervention approaches.
2. Materials
2.1. Cell Culture and Harvest
1. T-75 cell culture flasks
2. McCoy’s 5a medium supplemented with 10% (v/v) fetal bovine serum, 50 U/mL
penicillin, 50 μg/mL streptomycin, and 1.5 mM l-glutamine (American Type
Culture Collection (ATCC), Manassas, VA)
3. Dimethylsulfoxide (DMSO, cell culture use)
4. HT-29 cell line (ATCC, Manassas, VA)
5. Celecoxib (Pfizer, New York, NY)
6. 75 μM celecoxib: dissolve celecoxib in DMSO to make a 100 mM stock solution.
Further dilute to 75 μM with McCoy’s 5a cell culture medium. Use the same
concentration of DMSO in medium as negative control
7. Sterile phosphate-buffered saline (PBS) solution
8. 500 mM EDTA, pH 8
9. 2 mM EDTA in sterile PBS: add 80 μL of 500 mM EDTA, pH 8, in 20 mL of PBS
10. Centrifuge (maximum force: ∼17,000×g)
184 Xiao and Veenstra
Fig. 2. Schematic diagram of the ICAT labeling procedure applied to the quantitative
proteomic analysis.
2.2. Cell Lysis, Desalting, and Protein Quantitation
1. Lysis buffer: 50 mM Tris–HCl, pH 7.2, 1% Triton X-100, 10 mM sodium fluoride
(NaF), 1 mM sodium orthovanadate (Na3VO4), and 1 mM EDTA
2. Digital sonifier (Model 250, Branson Ultrasonics Corporation, Danbury, CT)
3. Bicinchoninic acid (BCA) protein assay reagent kit (Pierce, Rockford, IL)
4. D-SaltTM excellulose plastic desalting column 5 mL (maximum binding capacity
is 1.25 mg per column) (Pierce, Rockford, IL)
5. 50 mM NH4HCO3,pH8.3
Isotope-Coded Affinity Tag Labeling 185
6. Coomassie blue reagent: coomassie plus – The Better BradfordTM assay reagent
(Pierce, Rockford, IL)
7. Centrifuge (maximum force: ∼17,000×g)
8. Vacuum centrifuge
2.3. Denaturing and Reducing the Proteins
1. Denaturing buffer: 6 M guanidine in 50 mM NH4HCO3,pH8.3
2. 100 mM Tris (2-carboxyethyl) phosphine (TCEP) (Pierce, Rockford, IL)
3. Boiling water bath
2.4. Labeling with Cleavable ICAT Reagents, Desalting,
and Tryptic Digestion
1. Cleavable ICATTM reagents (light and heavy sulfhydryl modifying biotinylating
reagents). Store at –20 °C. One unit of either light or heavy reagent labels 100 μg
of protein. The regular kit offers both reagents in 1 unit/tube. The bulk kit offers
both reagents in 10 units/tube. The method described here is based on the use of
aregular kit, i.e., 1 unit that labels 100 μg of protein/tube. (Applied Biosystems,
Foster City, CA)
2. Acetonitrile
3. 37 °C water bath
4. D-SaltTM excellulose plastic desalting column 5 mL (Pierce, Rockford, IL)
5. 50 mM NH4HCO3,pH8.3
6. Coomassie blue reagent: coomassie plus – The Better BradfordTM assay reagent
(Pierce, Rockford, IL)
7. Trypsin gold, MS grade (Promega, Madison, WI)
2.5. Purifying the Labeled Peptides
1. Phenylmethanesulfonyl fluoride (PMSF) (Sigma Chemical Co., St. Louis, MO)
2. Glass wool
3. 5–3/4˝ disposable pasteur glass pipettes
4. UltralinkTM immobilized monomeric avidin slurry [50% (v/v)] (Pierce,
Rockford, IL)
5. Teflon tubing that fits the tip of the 5–3/4˝ disposable pasteur glass pipettes
6. 2× PBS buffer, pH 7.2: dissolve 14.2 g of Na2HPO4and 8.77 g of NaCl in
450 mL of H2O. Adjust pH to 7.2 by adding about 350 μL of 85% (v/v) H3PO4.
Add H2O to make a total volume of 500 mL. The final concentration is 200 mM
Na2HPO4and 300 mM NaCl
7. 1× PBS, pH 7.2: dilute 2× PBS 1:1 in H2O
8. 2 mM biotin solution: dissolve 9.8 mg of d-biotin ImmunoPure (MW 244.31,
Pierce, Rockford, IL) in 20 mL of 2× PBS, pH 7.2
9. Acetonitrile [20% (v/v)] in 50 mM NH4HCO3,pH8.3
10. Acetonitrile [30% (v/v)] containing 0.4% (v/v) formic acid
186 Xiao and Veenstra
11. pH paper (pH 2–9)
12. Dry ice
2.6. Cleaving Biotin
1. Cleaving reagent A (10 mL) (Applied Biosystems, Foster City, CA): contains
concentrated TFA. Store in fume hood at room temperature
2. Cleaving reagent B (Applied Biosystems, Foster City, CA): store at –20 °C
3. 37 °C water bath
4. Vacuum centrifuge
3. Methods
3.1. Cell Culture and Harvest
1. On day 1, plate HT-29 cells in T-75 flasks at 5 × 106cells/flask.
2. On day 2, aspirate medium. Culture cells with fresh medium containing 75 μM
of celecoxib or DMSO (negative control).
3. On day 3, 24 h after treating cells, aspirate cell culture medium. Rinse cells once
quickly with 6 mL of PBS.
4. Add 3 mL of 2 mM EDTA-PBS per flask, put flask into the 37 °C incubator.
Monitor the detachment of cells carefully. Cells usually detach within 5 min. For
the celecoxib-treated cells, it takes less than 5 min (see Note 1).
5. Tap the side of the flask against the palm of hand to dislodge cells. When the
cells are visibly detached, add 7 mL of PBS to flask. Resuspend cells and transfer
cell suspension to a 15 mL centrifuge tube. Harvest the treated and control cells
in separate tubes.
6. Centrifuge the cell suspension at 500×gfor 5 min. Remove the supernatant.
7. Wash cell pellet with 10 mL of PBS three times. Centrifuge at 500×gfor 5 min.
Remove PBS after each centrifugation.
8. Cell pellet is ready for lysis. Leave cell pellet on ice before proceeding to the
next step, or store the pellet at –80 °C.
3.2. Cell Lysis, Desalting and Protein Quantitation
1. Add 500 μL of lysis buffer to the cell pellet harvested from each T-75 flask.
Transfer the resuspended cells to a 1.5 mL eppendorf tube. Vortex briefly.
2. Clean the sonifier probe with H2O, methanol, and let it air dry before use.
3. To break the cells, set the digital sonifier amplitude at 16%. Hold up the
eppendorf tube with suspended cells. Let the probe plunge half way into the
lysis buffer. Pulse for 10 s, pause for 50 s. Repeat this cycle five times. Rest the
tube on ice between pulses. Lift the tube up again in time before the next 10 s
pulse cycle starts (see Note 2).
4. Clean the sonifier probe as in step 2 before starting the next sample.
5. Centrifuge cell lysate at 15,000×gfor 15 min at 4 °C.
Isotope-Coded Affinity Tag Labeling 187
6. Transfer cell lysate to a fresh eppendorf tube (see Note 3).
7. Quantify the protein in cell lysate using the BCA assay (see Note 4).
8. Prepare desalting column (D-SaltTMExcellulose Plastic Desalting Column, 5 mL,
Pierce) by washing column with 5× bed volume (i.e., 25 mL) of 50 mM
NH4HCO3, pH 8.3 (see Note 5).
9. Based on the BCA assay results, load up to 1.25 mg of cell lysate into each
desalting column. Discard the flow through (see Note 6).
10. Add 0.5 mL of 50 mM NH4HCO3, pH 8.3 into the column. Collect the flow
through into one eppendorf tube. Repeat this step seven times. Collect eluant in
seven 0.5 mL fractions.
11. Take 10 μL of eluant from each fraction and mix with 300 μL (1:30) of coomassie
blue reagent (Pierce). Visually examine the color of each tube. The color of
the protein-containing fractions should change from brown to blue. Proteins
normally elute in fractions 3–5.
12. Pool the tubes containing protein. Mix well. Discard the tubes that do not contain
protein.
13. Measure the protein concentration using the BCA assay (see Note 4).
14. Based on the BCA assay results, transfer 800 μg of protein from each of the
treated and control samples into two separate eppendorf tubes (see Note 7).
15. Lyophilize these two samples in vacuum centrifuge (see Note 8).
3.3. Denaturing and Reducing the Proteins
1. Freshly prepare denaturing buffer and 100 mM TCEP.
2. Add denaturing buffer and 100 mM TCEP to the protein samples. For 800 μg of
protein, add 640 μL of denaturing buffer and 8 μL of TCEP (see Note 9).
3. Vortex until the sample is completely dissolved in the buffer.
4. Boil the sample for 10 min.
5. Vortex to mix well. Spin the samples in centrifuge briefly. Cool to room
temperature.
3.4. Labeling with Cleavable ICAT Reagents, Desalting,
and Tryptic Digestion
1. Remove the ICAT reagents from the –20 °C freezer. Bring to room temperature.
Avoid exposing them to the light. To label 800 μg of protein (control or treated),
use eight tubes of reagent (light or heavy, label 100 μg of protein/tube). Spin in
centrifuge briefly to bring down the powder from the wall to the bottom of the
tube.
2. In the chemical hood with lights off, add 20 μL of acetonitrile into each of the
eight reagent tubes (light or heavy). Add 80 μL (i.e., 100 μg) of protein sample into
each tube. Tighten the tube caps. Vortex to mix well. Spin briefly in centrifuge
(see Note 10).
188 Xiao and Veenstra
3. Pool the control or treated sample mixtures (eight tubes of light or heavy),
respectively, into two tubes. This pooling should result in one light and one heavy
label tube with 800 μL of protein mixture in each.
4. Incubate the samples in the 37 °C water bath for 2 h. Keep the samples from
being exposed to light.
5. Combine the light- and heavy-labeled samples together into one tube. Proceed
with desalting.
6. Use the same desalting column as in the previous section. Since the binding
capacity per column is 1.25 mg, prepare two columns for a total of 1.6 mg of
labeled protein. Wash each column with 5× bed volume (i.e., 25 mL) of 50 mM
NH4HCO3, pH 8.3 (see Note 11).
7. Load 800 μg of the combined and labeled proteins per column. Follow steps
8–12 in Subheading 3.2. At the end of elution, pool the protein-containing eluant
fractions (usually fractions 3–5) into one 15 mL tube. (see Note 12).
8. Prepare trypsin freshly by reconstituting 20 μg of trypsin in 20 μL of 50 mM
NH4HCO3, pH 8.3. Add trypsin to the labeled protein at a trypsin-to-protein ratio
of 1:40 (w/w). For 1.6 mg of protein, add 40 μg of trypsin (see Note 13).
9. Wrap the 15 mL tube with aluminum foil. Incubate at 37 °C overnight (see
Note 14).
3.5. Purifying the Labeled Peptides
1. Boil the peptide solution for 10 min to deactivate trypsin.
2. Freshly prepare 100 mM PMSF in methanol. Vortex to dissolve well.
3. Add PMSF at a 1:100 dilution (v/v) to the trypsin-digested samples. For 3 mL
of digests, add 30 μL of PMSF. The final PMSF concentration is 1 mM. Vortex
briefly to mix.
4. Prepare the avidin column: put a small trace of glass wool gently into a 5–3/4˝
pasteur glass pipette. Push it from the top down for about 4–1/2˝. This packing
creates a support for the resin to settle onto (see Note 15).
5. Add 0.5 mL of water into the pipette. Let the water level fall till it reaches the
glass wool. At this point, the flow should stop naturally. Block the bottom of
the pipette. Then slowly add 1.5 mL of water into the pipette. Mark the water
level as an indicator for the volume of 1.5 mL.
6. Gradually add the avidin slurry to the 1.5 mL mark. Connect Teflon tubing to
the pipette tip to increase the flow rate (see Note 16).
7. Condition the column using the following washing buffers and sequence
(see Note 17)
a. 2× PBS, pH 7.2, 8 mL (5× bed volume)
b. 2 mM biotin solution, 6 mL (4× bed volume)
c. 30% (v/v) acetonitrile, 0.4% (v/v) formic acid, and 6 mL (4× bed volume)
d. 2× PBS, pH 7.2, 8 mL (5× bed volume)
8. Sample loading and incubation: take the teflon tubing off. Load 1.5 mL of the
digest sample into the column. After the sample flows through, incubate at room
Isotope-Coded Affinity Tag Labeling 189
temperature for 15 min. Load another 1.5 mL (or the rest) of sample. Incubate
for 15 min (see Note 18).
9. Connect the teflon tubing back to the tip of the pipette. Wash the column bound
with ICAT-labeled peptides with the following buffers and sequence:
a. 2× PBS, pH 7.2, 8 mL (5× bed volume)
b. 1× PBS, pH 7.2, 8 mL (5× bed volume)
c. 20% (v/v) acetonitrile in 50 mM NH4HCO3, pH 8.3, 6 mL (4× bed volume)
10. Final wash: take off the teflon tubing. Add 1.3 mL (a volume slightly less than
the bed volume) of 30% (v/v) acetonitrile, 0.4% (v/v) formic acid as a final
wash. Discard the flow through. Measure the pH of the last drop of this wash
step with pH paper. The pH should be >8 (basic), suggesting that acetonitrile
has not eluted the peptides off and that the peptides are still retained on the
beads (see Note 19).
11. Elute the peptides with 4 mL of 30% (v/v) acetonitrile, 0.4% (v/v) formic acid
in one 15 mL tube. Mix well and divide into four 1 mL aliquots. Briefly freeze
the peptides on dry ice or at –80 °C and then lyophilize in vacuum centrifuge
(see Note 20).
3.6. Cleaving Biotin
1. Prepare the cleaving reagent mixture in a chemical hood. For 1.6 mg of labeled
peptides, mix 760 μL of cleaving reagent A with 40 μL of cleaving reagent B. Add
the cleaving reagent mixture to the dry peptides. Dispense the mixture equally to
all four peptide aliquots (see Note 21).
2. Close the tube caps. Vortex well to dissolve the peptides.
3. Incubate the samples in a 37 °C water bath for 2 h.
4. Pool all the aliquots together when the incubation is finished. Freeze briefly on
dry ice or at –80 °C. Lyophilize the peptides in vacuum centrifuge.
5. Store at –80 °C prior to the next step (i.e., fractionation by SCX HPLC).
4. Notes
1. Dislodging cells using a low concentration of EDTA preserves the integrity of
cell surface proteins, which is critical in quantitative proteomic analysis.
2. For the Branson digital sonifier, use the following program settings: pulse on for
10 s; off for 50 s; amplitude = 16%. If bubbles are generated during sonication,
decrease the amplitude setting. Depending on the sample volume, the setting
can sometimes be lowered to 14%. The clumps of cells should disappear when
sonication is complete.
3. After this step the cell lysate can be stored at –80 °C. Otherwise, proceed to the
next step, i.e., BCA assay and desalting.
4. Protein quantitation is a common laboratory procedure. The instructions are
included within the BCA assay kit (Pierce); therefore, the procedure is not
described in this chapter.
190 Xiao and Veenstra
5. It is helpful to assemble a funnel reservoir on the top of the column to hold a
larger volume (up to 25 mL) of buffer.
6. The maximum binding capacity of the desalting column is 1.25 mg of protein
per column.
7. The method described here is based on the labeling of 800 μg of protein from
each of the treated and control samples. This amount of protein is desirable if
enough cell lysate is available. However, as little as 100 μg of protein from each
of the treated and control samples can be labeled using this protocol.
8. It takes about3htolyophilize the samples. If necessary, leave the samples in
the vacuum, centrifuge overnight to dry.
9. It is important to keep the pH of the cell lysate above 7 (ideally between 8 and
9). A pH below 7 will inhibit the reaction between cysteine residues and the
iodoacetamide group of the ICAT reagents.
10. Usually the control sample is labeled with the light reagent and the treated
sample is labeled with the heavy reagent.
11. To save time, it is suggested to set the two columns up on the stand during the
2-h labeling incubation time. It is better to attach a funnel reservoir to the top
of each column to hold up to 25 mL of wash buffer.
12. Normally the volume of sample after pooling is about 3 mL. Desalted samples
may have an opaque color because of the protein present in the sample.
13. Instead of using the buffer provided by the manufacturer, resuspend trypsin
in 50 mM NH4HCO3, pH 8.3. Keep the trypsin-to-protein ratio between 1:40
and 1:50.
14. The digestion mixture is incubated overnight for approximately 16–18 h.
15. Make sure the glass wool is well packed. There should be no holes present;
however, it should still allow liquid flow through at a reasonable flow rate.
Check the flow rate by adding 0.5 mL of water into the pipette. The water
should flow through quickly. Note that the flow rate will be slower consid-
erably once the avidin slurry is packed into the column. Take these recom-
mendations into consideration and not to pack too much or too little glass
wool.
16. The protein binding capacity of avidin slurry is 1.6 mg protein per milliliter of
packed avidin. One 1.5 mL column should offer sufficient capacity to enrich the
labeled peptides from 1.6 mg of protein.
17. The binding of 2 mM biotin to the column and the elution by 30% (v/v) acetoni-
trile, 0.4% (v/v) formic acid preclear the column of any potential nonspecific
binding activities.
18. The teflon tubing is a useful tool to adjust the flow rate. Connecting the teflon
tubing on to the tip of the column will increase the flow rate. On the other hand,
the flow rate will be slower without the teflon tubing attached.
19. The final wash is aimed to remove any nonspecific binding proteins. Using a
volume slightly less than the bed volume ensures that the labeled peptides are
retained on the column. The volume of the final wash buffer can be adjusted
according to the actual bed volume. When the bed volume of avidin is smaller,
Isotope-Coded Affinity Tag Labeling 191
the volume of the final wash buffer needs to be scaled down. If the pH of the
last drop is less than 3, the labeled peptides may have started to elute, meaning
potential loss of the labeled peptides.
20. The elution should be performed in a chemical fume hood to avoid inhaling
acetonitrile. The quick freezing of samples on dry ice can prevent sample spill
during vacuum centrifugation and reduce the time needed for the samples to
dry.
21. For every 200 μg of labeled peptides (i.e., 100 μg each of heavy or light labeled
in the pair), mix 95 μL of cleaving reagent A and 5 μL of cleaving reagent B
together first and transfer to the labeled peptides.
Acknowledgments
This project has been funded in whole or in part with Federal funds from
the National Cancer Institute, National Institutes of Health, under Contract No.
N01-CO-12400. The content of this publication does not necessarily reflect
the views or policies of the Department of Health and Human Services, nor
does mention of trade names, commercial products, or organization imply
endorsement by the U.S. Government.
References
1. Aebersold, R., Rist, B. and Gygi, S. P. (2000) Quantitative proteome analysis:
methods and applications. Ann N Y Acad Sci 919, 33–47.
2. Gygi, S. P., Rist, B. and Aebersold, R. (2000) Measuring gene expression by
quantitative proteome analysis. Curr Opin Biotechnol 11, 396–401.
3. Yates, J. R. 3rd. (2004) Mass spectral analysis in proteomics. Annu Rev Biophys
Biomol Struct 33, 297–316.
4. Ong, S. E. and Mann, M. (2005) Mass spectrometry-based proteomics turns quanti-
tative. Nat Chem Biol 1, 252–262.
5. Zieske, L. R. (2006) A perspective on the use of iTRAQ reagent technology for
protein complex and profiling studies. J Exp Bot 57, 1501–1508.
6. Yan, W. and Chen, S. S. (2005) Mass spectrometry-based quantitative proteomic
profiling. Brief Funct Genomic Proteomic 4, 27–38.
7. Bronstrup, M. (2004) Absolute quantification strategies in proteomics based on
mass spectrometry. Expert Rev Proteomics 1, 503–512.
8. Conrads, T. P., Issaq, H. J. and Hoang, V. M. (2003) Current strategies for quanti-
tative proteomics. Adv Protein Chem 65, 133–159.
9. Leitner, A. and Lindner, W. (2004) Current chemical tagging strategies for
proteome analysis by mass spectrometry. J Chromatogr B Analyt Technol Biomed
Life Sci 813, 1–26.
10. Gygi, S. P., Rist, B., Gerber, S. A., Turecek, F., Gelb, M. H. and Aebersold, R.
(1999) Quantitative analysis of complex protein mixtures using isotope-coded
affinity tags. Nat Biotechnol 17, 994–999.
192 Xiao and Veenstra
11. Flory, M. R., Griffin, T. J., Martin, D. and Aebersold, R. (2002) Advances in
quantitative proteomics using stable isotope tags. Trends Biotechnol 20, S23–S29.
12. Han, D. K., Eng, J., Zhou, H. and Aebersold, R. (2001) Quantitative profiling of
differentiation-induced microsomal proteins using isotope-coded affinity tags and
mass spectrometry. Nat Biotechnol 19, 946–951.
13. Conrads, K. A., Yu, L. R., Lucas, D. A., Zhou, M., Chan, K. C., Simpson, K. A.,
Schaefer, C. F., Issaq, H. J., Veenstra, T. D., Beck, G. R. Jr. and Conrads, T. P.
(2004) Quantitative proteomic analysis of inorganic phosphate-induced murine
MC3T3-E1 osteoblast cells. Electrophoresis 25, 1342–1352.
14. Gygi, S. P., Rist, B., Griffin, T. J., Eng, J. and Aebersold, R. (2002) Proteome
analysis of low-abundance proteins using multidimensional chromatography and
isotope-coded affinity tags. J Proteome Res 1, 47–54.
15. Tao, W. A. and Aebersold, R. (2003) Advances in quantitative proteomics via
stable isotope tagging and mass spectrometry. Curr Opin Biotechnol 14, 110–118.
16. Conrads, K. A., Yi, M., Simpson, K. A., Lucas, D. A., Camalier, C. E., Yu, L. R.,
Veenstra, T. D., Stephens, R. M., Conrads, T. P. and Beck, G. R. Jr. (2005) A
combined proteome and microarray investigation of inorganic phosphate-induced
pre-osteoblast cells. Mol Cell Proteomics 4, 1284–1296.
17. Koehne, C. H. and Dubois, R. N. (2004) COX-2 inhibition and colorectal cancer.
Semin Oncol 31, 12–21.
18. Sinicrope, F. A. and Gill, S. (2004) Role of cyclooxygenase-2 in colorectal cancer.
Cancer Metastasis Rev 23, 63–75.
19. Steinbach, G., Lynch, P. M., Phillips, R. K., Wallace, M. H., Hawk, E.,
Gordon, G. B., Wakabayashi, N., Saunders, B., Shen, Y., Fujimura, T., Su, L. K.
and Levin, B. (2000) The effect of celecoxib, a cyclooxygenase-2 inhibitor, in
familial adenomatous polyposis. N Engl J Med 342, 1946–1952.
20. Thun, M. J., Henley, S. J. and Patrono, C. (2002) Nonsteroidal anti-inflammatory
drugs as anticancer agents: mechanistic, pharmacologic, and clinical issues. J Natl
Cancer Inst 94, 252–266.
21. Arico, S., Pattingre, S., Bauvy, C., Gane, P., Barbat, A., Codogno, P. and Ogier-
Denis, E. (2002) Celecoxib induces apoptosis by inhibiting 3-phosphoinositide-
dependent protein kinase-1 activity in the human colon cancer HT-29 cell line.
J Biol Chem 277, 27613–27621.
22. Lev-Ari, S., Strier, L., Kazanov, D., Madar-Shapiro, L., Dvory-Sobol, H.,
Pinchuk, I., Marian, B., Lichtenberg, D. and Arber, N. (2005) Celecoxib and
curcumin synergistically inhibit the growth of colorectal cancer cells. Clin Cancer
Res 11, 6738–6744.
11
Analysis of Microdissected Cells by Two-Dimensional
LC-MS Approaches
Chen Li, Yi-Hong, Ye-Xiong Tan, Jian-Hua Ai, Hu Zhou, Su-Jun Li,
Lei Zhang, Qi-Chang Xia, Jia-Rui Wu, Hong-Yang Wang, and Rong Zeng
Summary
Laser capture microdissection (LCM) is a powerful tool that enables the isolation of
specific cell types from tissue sections, overcoming the problem of tissue heterogeneity and
contamination. We combined the LCM with isotope-coded affinity tag (ICAT) technology
and two-dimensional liquid chromatography to investigate the qualitative and quantitative
proteomes of hepatocellular carcinoma (HCC). The effects of three different histochemical
stains on tissue sections have been compared, and toluidine blue stain was proved as the
most suitable stain for LCM followed by proteomic analysis. The solubilized proteins
from microdissected HCC and non-HCC hepatocytes were qualitatively and quantita-
tively analyzed with two-dimensional liquid chromatography tandem mass spectrometry
(2D-LC-MS/MS) alone or coupled with cleavable isotope-coded affinity tag (cICAT)
labeling technology. A total of 644 proteins were qualitatively identified and 261 proteins
were unambiguously quantified. These results showed that the clinical proteomic method
using LCM coupled with ICAT and 2D-LC-MS/MS can carry out not only large-scale but
also accurate qualitative and quantitative analysis.
Key Words: hepatocellular carcinoma; laser capture microdissection; isotope-coded
affinity tag; two-dimensional liquid chromatography; mass spectrometry.
1. Introduction
Hepatocellular carcinoma (HCC) is one of the most frequent tumors
worldwide. There are 0.25–1 million newly diagnosed cases of HCC each year
(1). The highest frequencies of HCC are observed in sub-Saharan Africa and
From: Methods in Molecular Biology, vol. 428: Clinical Proteomics: Methods and Protocols
Edited by: A. Vlahou © Humana Press, Totowa, NJ
193
194 Li et al.
in Asia. In China, it has ranked the second cancer killer since 1990s. The most
risky factors of HCC are chronic hepatitis B virus (HBV) and hepatitis C virus
(HCV) infections, chronic exposure to the mycotoxin or aflatoxin B1 (AFB1),
and alcoholic cirrhosis. Till now, the mainstay for the diagnosis for HCC
includes serological tumor markers, such as alpha-fetoprotein, the L3 fraction
of alpha-fetoprotein, and PIVKA-II, as well as imaging modalities (1,2,3).
In order to improve diagnosis and prognosis from HCC, there is an
urgent need to identify molecular markers to detect the disease. Using
tissue samples from patients with HCC may be the most direct and
persuasive way to find useful diagnostic and/or prognostic markers. Recently,
proteomic analysis was applied to HCC tissues. Nineteen cases of HCC were
analyzed by two-dimensional electrophoresis (2DE) and matrix-assisted laser
desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS) by
Paik et al. (4,5,6). Proteome alterations in normal, cirrhotic, and tumorous
tissue were observed using 2DE-MALDI-TOF-MS assay by Jung et al. (7).
Kim et al. analyzed 11 cases of HCC using 2DE and delayed extraction-
matrix assisted laser desorption/ionization time-of-flight mass spectrometry
(DE-MALDI-TOF-MS) (8).
Nowadays, non-enzymatic sample preparation (NESP) is one of the regular
techniques for tissue sample preparation, which can be modified based on tissue-
type-specific properties (9). However, problems may be associated with hetero-
geneity and contaminating proteins, e.g., blood proteins. Several approaches
have been developed to resolve those problems. The selection of cell types
of interest by dissection has received a great deal of attention. Since 1996,
a laser-assisted technique, laser capture microdissection (LCM), has emerged
as a good choice. LCM under direct microscopic visualization permits rapid
one-step procurement of select cell populations from a section of complex,
heterogeneous tissue (10,11). LCM has been used to isolate specific types
of cells for protein, DNA, and RNA analysis. In the age of proteomics,
proteins obtained by laser capture microdissected cells can be analyzed by two-
dimensional gel electrophoresis (2DE gel) (12,13), immunoassay (14,15), and
surface-enhanced laser desorption and ionization time-of-flight (SELDI-TOF)
(16,17,18,19,20,21). The only shortcoming of LCM may be that it requires long
time to pick up sufficient cells for one experiment: 2–7 h for 20,000–40,000
cells per immunoassay and 15 h for 250,000 cells per 2DE gel (22).
Our previous work had applied proteomic analysis to HCC cell lines (23,24)
and HCC metastatic cells (25). Furthermore, we extended our work to clinical
tissues using LCM. However, the present LCM assay only obtains about several
hundred micrograms of proteins with dissection for several hours, which is
hard to be analyzed by traditional 2DE-MS proteomic route, especially for
preparative 2DE gels followed by MS identification.
Proteomic Analysis of Clinical HCC Using LCM 195
Since 1999, the isotope-coded affinity tag (ICAT) strategy has been a leading
technology for relative protein quantification relying on post-harvest stable
isotope labeling (26). Post-harvest labeling with stable isotopes can be used for
protein quantification in cells and tissues from any organism, and the ICAT
method as initially described has been shown to be capable of accurate quantifi-
cation of proteins in complex mixtures (26). After the first-generation 2H-
ICAT reagents, the second- generation cleavable 13C-ICAT reagents provided
improved performance (27,28,29). The 2D chromatography MS/MS method has
been shown to be capable of identifying a large number of proteins, including
proteins of low abundance (30,31).
In this study, we used LCM to isolate HCC and non-HCC hepatocytes
and firstly combined LCM with cleavable isotope-coded affinity tag (cICAT)
labeling technology and two-dimensional liquid chromatography tandem mass
Solubilized proteins
Frozen sections of HCC tissues
HCC hepatocytes
Stained with toluidine blue
Non-HCC hepatocytes
Digestion of protein mixture
2D-LC-MS/MS
Analyze by bioinformatics
Labeled with cICAT light chain Labeled with cICAT heavy chain
Laser capture microdissection
Fig. 1. Outline of accurate qualitative and quantitative proteomic analysis of clinical
hepatocellular carcinoma using laser capture microdissection coupled with isotope-
coded affinity tag and two-dimensional liquid chromatography mass spectrometry.
Reprinted with permission from (34).
196 Li et al.
spectrometry (2D-LC-MS/MS) to carry out accurate qualitative and quantitative
analysis of HCC and non-HCC tissues. The flowchart used is outlined in Fig. 1.
Totally 644 proteins in HCC hepatocytes were qualitatively determined and 261
differential proteins between HCC and non-HCC hepatocytes were quantitated.
Till now, this is one of the largest qualitative and qualitative proteomes for
HCC and non-HCC tissues. Our strategy and method provided an accurate,
fast, and sensitive approach for proteomic analysis of clinical tissues, which
will facilitate the understanding of the mechanism of HCC or other diseases
and mining of potential markers and drug targets for diagnosis and treatment.
2. Materials
2.1. Tissue Specimen and Sample Preparation by Nonenzymatic
Method (NESP)
1. Tissues from a HCC patient are isolated from fresh partially hepatectized tissues
of HCCs in Shanghai Eastern Hepatobiliary Surgery Hospital. Access to human
tissues complies with both Chinese laws and the guidelines of the Ethics
Committee.
2. Glutamine-free RPMI 1640 medium: glutamine-free, 5% fetal calf serum, 0.2 mM
phenylmethylsulfonyl fluoride, 1 mM ethylenediaminetetraacetic acid tetrasodium
salt dehydrate (EDTA), and antibiotics: oxacillin 25 μg/ml, gentamycin 50 μg/ml,
penicillin 100 U/ml, streptomycin 100 μg/ml, amphotericin B 0.25 μg/ml, nistatin
50 U/ml. Store at 4°C.
3. Ceramic mortar and pestle (SIBAS Corp. Shanghai, China).
4. Lysis buffer: 8 M urea, 4% 3-[(3-cholamidopropyl)dimethylammonio]-1-propane
sulfonate (CHAPS), 40 mM Tris-HCl (pH 8.3), 65 mM dithiothreitol (DTT).
Store in aliquots at –8°C.
5. Proteinase inhibitor tablet mixture (Roche).
2.2. Laser Capture Microdissection
1. Tissues from a HCC patient are isolated from fresh partially hepatectized tissues
of HCCs in Shanghai Eastern Hepatobiliary Surgery Hospital. Access to human
tissues complies with both Chinese laws and the guidelines of the Ethics
Committee. The tissues are from a 50-year male patient with HCC in Edmondson
grade III (HBV infected, AFP 7.3 μg/L, size 15 × 13 × 10.5 cm).
2. Freezing microtome CM1900 (Leica).
3. O.C.T. compound (Tissue-Tek).
4. Hematoxylin, eosin, and toludine blue stain (Shanghai Genebase Corp.).
5. Leica AS LMD Laser Capture Microdissection System (Leica).
6. Lysis buffer: 8 M urea, 4% CHAPS, 40 mM Tris, 65 mM DTT. Store in aliquots
at –8°C.
7. Proteinase inhibitor tablet mixture (Roche).
Proteomic Analysis of Clinical HCC Using LCM 197
2.3. Removal of Toludine Blue and Digestion of Protein Mixture
for Qualitative Analysis
1. Precipitation solution: 50% acetone, 50% ethanol, 0.1% acetic acid (HAc). Store
at –20°C.
2. Redissolved buffer: 6 M guanidine HCl, 100 mM Tris-HCl (pH 8.3). Store at
4°C.
3. DTT and iodoacetamide (IAA) are from Bio-Rad. Sequencing grade TPCK-
trypsin is from Promega.
4. YM3 ultrafiltration membranes (molecular mass cutoff, 3 kDa) are from Millipore
Corp. All buffers are prepared with Milli-Q water (Millipore).
2.4. Cleavable Isotope-Coded Affinity Tag Labeling of Proteins
1. Tri-n-butylphosphate (TBP) is from Bio-Rad.
2. cICAT light or heavy reagents, Avidin cartridge, affinity buffer–elute, affinity
buffer–load, affinity buffer–wash 1, affinity buffer–wash 2, cleaving reagents A
and B are from Applied Biosystems.
3. Sequencing grade TPCK-trypsin (Promega).
4. YM3 ultrafiltration membranes (molecular mass cutoff, 3 kDa) are from Millipore
Corp. All buffers are prepared with Milli-Q water (Millipore).
2.5. One-Dimensional and Two-Dimensional Liquid Chromatography
Coupled with Tandem Mass Spectrometry
1. Formic acid is obtained from Aldrich, and acetonitrile (HPLC gradient grade) is
obtained from Merck.
2. The LCQ™ Deca XP system, ProteomeX™ Workstation and TurboSequest
software are purchased from Thermo Electron Corporation.
2.6. Bioinformatics Analysis
1. ExPASy proteomics tools are accessed from cn.expasy.org/tools/#proteome.
2. Program TMHMM 2.0 is accessed from the Center for Biological Sequence
Analysis (www.cbs.dtu.dk/services/TMHMM/).
3. Classification tools are accessed from www.geneontology.org.
3. Methods
In brief, two keywords should be noticed during the whole process of LCM
coupled with 2D-LC-MS/MS approaches. The first one is speediness, and
the second one is impurity. Sample preparation by LCM technology must be
done as quickly as possible, including fixation of fresh tissues, preparation of
frozen sections, histochemical staining, microdissection, and so on. Impurities,
198 Li et al.
such as histochemical stains, should be removed as completely as possible
by centrifuge, precipitation, and ultrafitration before trypsin digestion and LC-
MS/MS analysis.
Fixation and histochemical staining are the two initial steps in LCM
technology. The appropriate selection of fixation and histochemical staining
methods is an important factor for the processes. In this work, we used freshly
prepared liver tissues to make frozen sections (8 μm thick), and we fixed the
sections with ethanol to avoid the effects on proteins, such as crosslinking
caused by formalin fixation. Some histochemical stains (hematoxylin, eosin,
methyl green, and toluidine blue) were tested in 2DE gel (33), which showed
that staining with single stain (hematoxylin) was better than with two stains
simultaneously (hematoxylin and eosin); methyl green and toluidine blue
staining were both compatible with the analysis of proteins by 2D-PAGE. The
results with toluidine blue staining indicated a direct link between the intensity
of tissue section staining and problems with the generation of good-quality
protein separations. In our study, the proteins from cells after LCM were
subjected to tryptic digestion and LC-MS/MS analysis. The staining material
might affect the pH of digestion buffer or inactivate the trypsin; therefore,
we tried to remove the stains using precipitation and ultrafiltration prior to
digestion. We used three histochemical stains (hematoxylin, eosin, and toluidine
blue), respectively, to stain the frozen sections. Among these three histo-
chemical stains, we found that almost all toluidine blue stain could be removed
after precipitation in the solution (50% acetone, 50% ethanol, 0.1% acetic
acid) and desalting by ultrafiltration. In addition, protein solubilization stained
by toluidine blue stain was better because some colored protein precipitation
appeared on the filtration membrane when using hematoxylin stain or eosin
stain. Therefore, we chose toluidine blue stain to optimize the experimental
conditions, including staining, microdissection, and protein digestion.
3.1. Tissue Specimen and Sample Preparation by Nonenzymatic
Method (NESP)
1. The tissues used were from a 50-year male patient with HCC in Edmondson
grade III (HBV infected, AFP 7.3 μg/L, size 15 × 13 × 10.5 cm). Tumorous
tissues and their adjacent paired nontumorous tissues (3 cm away from the edge of
HCC lesions, about 0.1 g) were isolated from fresh partially hepatectized tissues
of HBV-associated HCC. A part of the resected tissue was used for histology
analysis.
2. The tissues were rinsed several times with cold glutamine-free RPMI 1640
medium and were homogenized in liquid nitrogen-cooled mortar and pestle (see
Note 1).
3. The tissue powders obtained were dissolved in lysis buffer (see Note 2).
Proteomic Analysis of Clinical HCC Using LCM 199
4. The samples were sonicated on ice for 30 s (intensity: below 50 W) using an
ultrasonic processor and centrifuged for1hat20,627×gto remove DNA, RNA,
and any particulate materials.
5. The protein concentrations of samples were measured by Bio-Rad Protein Assay
kit. All samples were stored at –8°C until use (see Note 3).
3.2. Laser Capture Microdissection
1. Embed fresh tissues carefully in OCT in plastic mold, taking care not to trap air
bubbles surrounding the tissue. Freeze the tissue by setting mold on top of liquid
nitrogen until 70–80% of the block turns white and then put the block on top of
dry ice.
2. For cutting step, mount the frozen block on the cryostat holder. Never, at any
point, let the tissue warm up to temperatures above –15°c. Allow frozen blocks
to equilibrate in the cryostat chamber for about 5 min. Cut 8-μm sections.
3. Wash 8-μm sections of freshly prepared liver tissues by cold phosphate buffered
saline (PBS, pH 7.4), and stain with toluidine blue using standard manufacturer’s
protocols with minor modifications (see Note 4).
4. Fix the sections in cold 95% ethanol for 10 min, air-dry and microdissect with
Leica AS LMD Laser Capture Microdissection System.
5. Using laser pulses of 7.5 μm diameter, 70 mW, and with 2–3 ms duration,
microdissect approximately 50,000 or 100,000 cells of HCC and non-HCC hepato-
cytes; store in microdissection caps at –8°C until lysed (see Note 5). An example
of the results produced using hematoxylin and eosin (H&E) stained section is
shown in Fig. 2.
6. Each cell population was determined to be 95% homogeneous by microscopic
visualization of the captured cells. Dissolve the laser capture microdissected HCC
and non-HCC hepatocytes in lysis buffer (see Note 2).
7. Sonicate the samples on ice for a while using an ultrasonic processor and
centrifuge for1hat20,627×gto remove DNA, RNA, and any particulate
materials.
8. Measure the protein concentrations of samples by Bio-Rad Protein Assay kit.
Store all the samples at –8°C until use (see Note 3).
3.3. Removal of Toludine Blue and Digestion of Protein Mixture
for Qualitative Analysis
1. Deposit the samples prepared by NESP or LCM technology in precip-
itation solution (50% acetone, 50% ethanol, 0.1% acetic acid; sample
volume:precipitation solution volume = 1:5) at least for 12 h at –20°C. Wash the
pellets with 100% acetone, 70% ethanol, and lyophilize by lyophilization (see
Note 6).
2. Redissolve the pellets in 6 M guanidine HCl, 100 mM Tris (pH 8.3); measure the
concentrations with Bio-Rad Protein Assay kit.
200 Li et al.
A.
B.
Fig. 2. HCC tissues before (A) and after (B) LCM. Reprinted with permission
from (34).
4. Reduce 200 μg solubilized proteins with DTT (final concentration 20 mM) and
subsequently alkylate with IAA (final concentration 40 mM).
5. After desalting by YM3 ultrafiltration membranes, incubate the protein mixture
with trypsin (trypsin:protein mixture = 1:30, W/W, Promega, Madison, WI) at
37°C for 16 h (see Note 7).
3.4. Cleavable Isotope-Coded Affinity Tag Labeling of Proteins
1. Reduce 100 μg HCC and 100 μg non-HCC solubilized proteins prepared by LCM
technology with TBP (final concentration 5 mM) (see Note 8).
Proteomic Analysis of Clinical HCC Using LCM 201
2. Transfer the reduced HCC and non-HCC solubilized proteins into the vial
containing cICAT light or heavy reagent, respectively, and mix. After a brief
centrifugation, incubate the proteins for2hat37°C in the dark.
3. Combine the labeled proteins into one tube. After desalting by YM3 ultrafil-
tration membranes, incubate the protein mixture with trypsin (trypsin:protein
mixture = 1:30, W/W, Promega, Madison, WI) at 37°C for 16 h (see Note 7).
4. Use Avidin cartridge (Applied Biosystems) to purify the ICAT-labeled peptides
from tryptic digests according to the manufacture’s protocol. In brief, activate
Avidin cartridge by 2 ml of the affinity buffer–elute and 2 ml of the affinity
buffer–load. Slowly inject (∼1 drop/5 s) the peptide sample onto Avidin cartridge.
Wash the Avidin cartridge by 500 μl of affinity buffer–load, 1 ml of affinity
buffer–wash 1, 1 ml of affinity buffer–wash 2, and 1 ml of Milli-Q water. To
elute the labeled peptides, slowly inject (∼1 drop/5 s) the affinity buffer–elute and
collect the elute. Dry the elute from the Avidin cartridge through lyophilization.
5. Dissolve the dried cICAT-labeled peptides in cleaving reagents and cleave for
2 h at 37°C. Condense the cICAT-labeled peptides through lyophilization.
3.5. One-Dimensional and Two-Dimensional Liquid Chromatography
Coupled with Tandem Mass Spectrometry (1D- and 2D-LC-MS/MS)
1. All the 2D HPLC separations are performed on ProteomeX™ (Thermo Finnigan
Corp., San Jose, CA) equipped with two LC pumps. The flow rates of both salt and
analytical pumps are 200 μl/min and about 2 μl/min after split. The strong cation
exchange column is the 300 μm inner diameter ones (SCX resin, 5 μm), and the
RPC column is the 150 μm inner diameter (C18 resin, 300 A, 5 μm) (see Note 9).
2. Nine different salt concentration ranges—0, 25, 50, 75, 100, 150, 200, 400, and
800 mM ammonium chloride—are used for step gradient.
3. The mobile phases used for reverse phase are A: 0.1% formic acid in water, pH
3.0, B: 0.1% formic acid in acetonitrile.
4. Load about 200 μg of peptides digested from the LCM protein to the SCX
column by the autosample. The elute condition is described in step 2. Load
the eluted peptides from each salt step to the RPC columns. The RPC columns
are washed by 95% A mobile phases in 20 column volumes. Finally, separate
the peptides using 100-min linear gradient from 5 to 80% B mobile phases.
The eluting peptide enters an LCQ ProteomeX™ mass spectrometer (Thermo
Electron, San Jose, CA) by the metal needle (see Note 10).
5. The 1D HPLC separation uses the same system/experimental steps, but without
the use of a strong cation exchange column.
6. An electrospray (ESI) ion-trap mass spectrometer (LCQ Deca XP, Thermo
Finnigan, San Jose, CA) is used for peptide detection.
7. The positive ion mode is employed and the spray voltage is set at 3.2 kV. The
spray temperature is set at 150°C for peptides.
8. The collision energy is automatically set by LCQ Deca XP. After the acquisition
of full scan mass spectra, three MS/MS scans are acquired for the next three
most intense ions using dynamic exclusion.
202 Li et al.
9. Peptides and proteins are identified using TurboSequest(Thermo Finnigan,
San Jose, CA), which uses the MS and MS/MS spectrum of peptide ions
to search against the publicly available NCBI non-redundant protein database
(www.ncbi.nlm.nih.gov).
10. The protein identification criteria that we used are based on Delta CN (≥0.1)
and Xcorr (one charge ≥1.8, two charges ≥2.2, three charges ≥3.7). An
example of the results produced is shown in Table 1 (see Note 11).
11. For quantitative analysis with cICAT technology and 2D-LC-MS/MS, manual
check is followed after database searching and quantification by Xpress
(TurboSequestsoftware). Quantitative analysis results of 261 proteins from
LCM-ICAT-2D-LC-MS/MS are shown in Fig. 3. In our experiment, a total of
149 differentially expressed proteins with at least twofold quantitative alterations
in HCC and non-HCC hepatocytes were detected, including 55 upregulated
proteins (32 with 2∼5 folds, 13 with 5∼10 folds, 10 with >10 folds) and 94
downregulated spots in HCC hepatocytes (62 with 2∼5 folds, 17 with 5∼10
folds, 15 with >10 folds).
3.6. Bioinformatics Analysis
1. The pIand Mr of the proteins are analyzed using ExPASy proteomics tools
accessed from http://cn.expasy.org/tools/#proteome. Examples of the results
produced are shown in Table 1 and Fig. 5A and 5B.
2 ≤ Ratio(HCC/non-HCC) ≤ 5
2 ≤ Ratio(non-HCC/HCC) ≤ 5
5 < Ratio(HCC/non-HCC) ≤ 10
5 < Ratio(non-HCC/HCC) ≤ 10
Ratio(HCC/non-HCC) > 10
Ratio(non-HCC/HCC) > 10
Ratio(HCC/non-HCC or non-HCC/HCC) < 2
62
17
15 32
13
10
112
Fig. 3. Quantitative analysis results of 261 proteins from LCM-ICAT-2D-LC-
MS/MS. A total of 149 differentially expressed proteins with at least twofold quanti-
tative alterations in HCC and non-HCC hepatocytes were detected, including 55 upreg-
ulated proteins (32 with 2∼5 folds, 13 with 5∼10 folds, 10 with >10 folds) and 94
downregulated spots in HCC hepatocytes (62 with 2∼5 folds, 17 with 5∼10 folds, 15
with >10 folds). Reprinted with permission from (34).
Proteomic Analysis of Clinical HCC Using LCM 203
Table 1
Summary of Total Proteins Identified in HCC-NESP-1D-LC-MS/MS,
HCC-NESP-2D-LC-MS/MS and HCC-LCM-2D-LC-MS/MS
HCC-
NESP-1D-
LC-MS/MS
HCC-
NESP-2D-
LC-MS/MS
HCC-
LCM-2D-
LC-MS/MS
Protein quantity 200μg 200μg 200μg
Total proteins identified 208 626 644
Hydrophobic proteins 25(12.0%) 64(10.2%) 80(12.4%)
Trans-membrane proteins 8(3.9%) 30(4.8%) 54(8.4%)
Proteins with Mr >100KD or < 10KD 19(9.1%) 77(12.3%) 75(11.6%)
Proteins pI >9 21(10.1%) 78(12.5%) 126(19.6%)
2. The general average hydropathicity (GRAVY) score is calculated as the arithmetic
mean of the sum of the hydropathic indices of each amino acid (32). Examples
of the results produced are shown in Table 1 and Fig. 5C.
3. The trans-membrane prediction is conducted using the computer server
program TMHMM server 2.0, which can be accessed from the CBS
(http://www.cbs.dtu.dk/services/TMHMM/). Examples of the results produced are
shown in Table 1 and Fig. 5D.
4. All identified proteins are classified by their molecular function, cellular
component, and biological process with the tools on http://www.geneontology.org.
An example of the results produced is shown in Fig. 4.
4. Notes
1. Glutamine-free RPMI 1640 medium must be cold (4°C) before use. Washing
should be done as quickly as possible, until there are no contaminations (blood,
etc.) on tissues. Glutamine-free RPMI 1640 medium could be replaced by PBS
(pH 7.4), 0.9% NaCl solution, or any other isotonic buffer.
2. Store the lysis buffer in small aliquots at –8°C to avoid multiple freeze-thaw
cycles. Protease inhibitor tablet mixture (Roche Molecular Biochemicals) should
be dissolved in lysis buffer.
3. Store the samples in small aliquots at –8°C to avoid multiple freeze-thaw cycles.
Protein concentrations of the samples should be about 10 μg/μl for subsequent
experiments.
4. The sections should be very lightly stained with toluidine blue only to distinguish
hepatocytes during microdissection. Otherwise, the redundant stains could affect
follow-up experiments.
5. In fact, in order to reduce microdissection time, manipulators could choose to
capture hepatocytes or remove other cells based on the condition of each section.
204 Li et al.
A.
B.
Fig. 4. Classification of differentially expressed proteins obtained by LCM-ICAT-
2D-LC-MS/MS. (A) shows proteins with at least twofold increased expression levels
in HCC hepatocytes. (B) shows proteins with at least twofold decreased expression
levels in HCC hepatocytes. Reprinted with permission from (34).
6. Precipitation solution, acetone, and ethanol must be cold at –20°C before use.
7. Ultrafiltration is very important to remove redundant salts, stain, and other
impurities, and ensure follow-up steps.
8. TBP is a much stronger but more toxic reducing agent for labeling ICAT reaction
than DTT.
Proteomic Analysis of Clinical HCC Using LCM 205
A. M
r distribution
7
62
79 80
33
0
20
40
60
80
100
<10 kDa 10 – 30
kDa
30 – 50
kDa
50 – 100
kDa
>100 kDa
M
r range
Protein number
D. Trans-membrane protein distribution
17
4
1
5
0
5
10
15
20
123
>3
Number of trans-membrane region
Protein number
B. pI distribution
00
21
58 53
21
61
37
10
0
10
20
30
40
50
60
70
<3
(3~4)
(4~5)
(5~6)
(6~7)
(7~8)
(8~9)
(9~10)
>10
pI range
Protein number
C. Hydrophile and hydrophobicity distribution
669
15 11
37
18
31
39
30 27
13 12
43
0
5
10
15
20
25
30
35
40
45
<–1.0
(–1.0~–0.9)
(–0.9~–0.8)
(–0.8~–0.7)
(–0.7~–0.6)
(–0.6~–0.5)
(–0.5~–0.4)
(–0.4~–0.3)
(–0.3~–0.2)
(–0.2~–0.1)
(–0.1~0)
(0 ~0.1)
(0.1–0.2)
(0.2–0.3)
>0.3
H
y
drophilic and h
y
drophobic value
Protein number
Fig. 5. Characteristics of differentially expressed proteins obtained by LCM-ICAT-
2D-LC-MS/MS. (A) shows the Mr distribution; (B) shows the pIdistribution; (C)
presents the hydrophile and hydrophobicity distribution; and (D) shows the trans-
membrane proteins. Reprinted with permission from (34).
9. The LCQ ProteomeX™ Workstation (Thermo Electron, San Jose, CA) is an
automatic 2D LC/MS system, which can be used in high-throughout proteomic
research. However, you may use another equipment to separate the proteomics
sample by offline SCX fractionation. The step involved in offline SCX fraction-
ation is almost the same as online. The difference is that you need to manually
load the step salt-eluted peptides to RPC column.
10. If you use the nanospay kit in the mass spectrometer and the 75-μm
inner diameter RPC column, the eluted peptides can directly enter the mass
spectrometer. The sensitivity in the nanospay mode is higher than in the metal
needle mode.
11. The protein identification criteria can vary based on the type of mass
spectrometer or other analytic needs. For example, we use Delta CN (≥0.1) and
Xcorr (one charge ≥1.9, two charges≥2.2, three charges ≥3.75) as criteria
when using LTQ linear ion trap mass spectrometer (Thermo Finnigan, San Jose,
CA).
Acknowledgments
This work was supported by National High-Technology Project
(2001AA233031, 2002BA711A11) and Basic Research Foundation
(2001CB210501).
206 Li et al.
References
1. Feitelson M.A., Sun B., Satiroglu Tufan N.L., Liu J., Pan J. and Lian Z. (2002)
Genetic mechanisms of hepatocarcinogenesis. Oncogene 21, 2593–2604.
2. Fujiyama S., Tanaka M., Maeda S., Ashihara H., Hirata R. and Tomita K. (2002)
Tumor markers in early diagnosis, follow-up and management of patients with
hepatocellular carcinoma. Oncology 62(Suppl 1),57–63.
3. Qin L.X. and Tang Z.Y. (2002) The prognostic molecular markers in hepatocellular
carcinoma. World J Gastroenterol 8, 385–392.
4. Park K.S., Cho S.Y., Kim H. and Paik Y.K. (2002) Proteomic alterations of the
variants of human aldehyde dehydrogenase isozymes correlate with hepatocellular
carcinoma. Int J Cancer 97, 261–265.
5. Park K.S., Kim H., Kim N.G., Cho S.Y., Choi K.H., Seong J.K. and Paik Y.K.
(2002) Proteomic analysis and molecular characterization of tissue ferritin light
chain in hepatocellular carcinoma. Hepatology 35, 1459–1466.
6. Cho S.Y., Park K.S., Shim J.E., Kwon M.S., Joo K.H., Lee W.S., Chang J.,
Kim H., Chung H.C., Kim H.O. and Paik Y.K. (2002) An integrated proteome
database for two-dimensional electrophoresis data analysis and laboratory infor-
mation management system. Proteomics 2, 1104–1113.
7. Lim S.O., Park S.J., Kim W., Park S.G., Kim H.J., Kim Y.I., Sohn T.S., Noh J.H.
and Jung G. (2002) Proteome analysis of hepatocellular carcinoma. Biochem
Biophys Res Commun 291, 1031–1037.
8. Kim J., Kim S.H., Lee S.U., Ha G.H., Kang D.G., Ha N.Y., Ahn J.S., Cho
H.Y., Kang S.J., Lee Y.J., Hong S.C., Ha W.S., Bae J.M., Lee C.W. and
Kim J.W. (2002) Proteome analysis of human liver tumor tissue by two-
dimensional gel electrophoresis and matrix assisted laser desorption/ionization-
mass spectrometry for identification of disease-related proteins. Electrophoresis 23,
4142–4156.
9. Franzen B., Hirano T., Okuzawa K., Uryu K., Alaiya A.A., Linder S. and
Auer G. (1995) Sample preparation of human tumors prior to two-dimensional
electrophoresis of proteins. Electrophoresis 16, 1087–1089.
10. Emmert-Buck M.R., Bonner R.F., Smith P.D., Chuaqui R.F., Zhuang Z.,
Goldstein S.R., Weiss R.A. and Liotta L.A. (1996) Laser capture microdissection.
Science 274, 998–1001.
11. Bonner R.F., Emmert-Buck M., Cole K., Pohida T., Chuaqui R., Goldstein S. and
Liotta L.A. (1997) Laser capture microdissection: molecular analysis of tissue.
Science 278, 1481–1483.
12. Ornstein D.K., Gillespie J.W., Paweletz C.P., Duray P.H., Herring J., Vocke
C.D., Topalian S.L., Bostwick D.G., Linehan W.M., Petricoin E.F., III and
Emmert-Buck M.R. (2000) Proteomic analysis of laser capture microdissected
human prostate cancer and in vitro prostate cell lines. Electrophoresis 21,
2235–2242.
13. Jones M.B., Krutzsch H., Shu H., Zhao Y., Liotta L.A., Kohn E.C. and
Petricoin E.F., III (2002) Proteomic analysis and identification of new biomarkers
and therapeutic targets for invasive ovarian cancer. Proteomics 2, 76–84.
Proteomic Analysis of Clinical HCC Using LCM 207
14. Simone N.L., Remaley A.T., Charboneau L., Petricoin E.F., III, Glickman J.W.,
Emmert-Buck M.R., Fleisher T.A. and Liotta L.A. (2000) Sensitive immunoassay
of tissue cell proteins procured by laser capture microdissection. Am J Pathol 156,
445–452.
15. Ornstein D.K., Englert C., Gillespie J.W., Paweletz C.P., Linehan W.M., Emmert-
Buck M.R. and Petricoin E.F., III (2000) Characterization of intracellular prostate-
specific antigen from laser capture microdissected benign and malignant prostatic
epithelium. Clin Cancer Res 6, 353–356.
16. Sauter E.R., Zhu W., Fan X.J., Wassell R.P., Chervoneva I. and Du Bois G.C.
(2002) Proteomic analysis of nipple aspirate fluid to detect biologic markers of
breast cancer. Br J Cancer 86, 1440–1443.
17. Verma M., Wright G.L., Jr., Hanash S.M., Gopal-Srivastava R. and Srivastava
S. (2001) Proteomic approaches within the NCI early detection research network
for the discovery and identification of cancer biomarkers. Ann N Y Acad Sci 945,
103–115.
18. Jain K.K. (2002) Recent advances in oncoproteomics. Curr Opin Mol Ther 4,
203–209.
19. Jr G.W., Cazares L.H., Leung S.M., Nasim S., Adam B.L., Yip T.T., Schellhammer
P.F., Gong L. and Vlahou A. (1999) ProteinChipsurface enhanced laser
desorption/ionization (SELDI) mass spectrometry: a novel protein biochip
technology for detection of prostate cancer biomarkers in complex protein mixtures.
Prostate Cancer Prostatic Dis 2, 264–276.
20. Batorfi J., Ye B., Mok S.C., Cseh I., Berkowitz R.S. and Fulop V. (2003) Protein
profiling of complete mole and normal placenta using ProteinChip analysis on
laser capture microdissected cells. Gynecol Oncol 88, 424–428.
21. Wulfkuhle J.D., Paweletz C.P., Steeg P.S., Petricoin E.F., III and Liotta L. (2003)
Proteomic approaches to the diagnosis, treatment, and monitoring of cancer. Adv
Exp Med Biol 532, 59–68.
22. Seow T.K., Liang R.C., Leow C.K. and Chung M.C. (2001) Hepatocellular
carcinoma: from bedside to proteomics. Proteomics 1, 1249–1263.
23. Yu L.R., Shao X.X., Jiang W.L., Xu D., Chang Y.C., Xu Y.H. and Xia Q.C. (2001)
Proteome alterations in human hepatoma cells transfected with antisense epidermal
growth factor receptor sequence. Electrophoresis 22, 3001–3008.
24. Yu L.R., Zeng R., Shao X.X., Wang N., Xu Y.H. and Xia Q.C. (2000) Identification
of differentially expressed proteins between human hepatoma and normal liver cell
lines by two-dimensional electrophoresis and liquid chromatography-ion trap mass
spectrometry. Electrophoresis 21, 3058–3068.
25. Ding S.J., Li Y., Tan Y.X., Jiang M.R., Tian B., Liu Y.K., Shao X.X., Ye S.L.,
Wu J.R., Zeng R., Wang H.Y., Tang Z.Y. and Xia Q.C. (2004) From proteomic
analysis to clinical significance: overexpression of cytokeratin 19 correlates with
hepatocellular carcinoma metastasis. Mol Cell Proteomics 3(1), 73–81.
26. Gygi S.P., Rist B., Gerber S.A., Turecek F., Gelb M.H. and Aebersold R. (1999)
Quantitative analysis of complex protein mixtures using isotope-coded affinity
tags. Nat Biotechnol 17, 994–999.
208 Li et al.
27. Li J., Steen H. and Gygi S.P. (2003) Protein profiling with cleavable isotope
coded affinity tag (cICAT) reagents: the yeast salinity stress response. Mol Cell
Proteomics 2(11), 1198–204.
28. Oda Y., Owa T., Sato T., Boucher B., Daniels S., Yamanaka H., Shinohara Y.,
Yokoi A., Kuromitsu J. and Nagasu T. (2003) Quantitative chemical proteomics
for identifying candidate drug targets. Anal Chem 75, 2159–2165.
29. Hansen K.C., Schmitt-Ulms G., Chalkley R.J., Hirsch J., Baldwin M.A. and
Burlingame A.L. (2003) Mass spectrometric analysis of protein mixtures at
low levels using cleavable 13C-isotope-coded affinity tag and multidimensional
chromatography. Mol Cell Proteomics 2, 299–314.
30. Washburn M.P., Wolters D. and Yates J.R., III (2001) Large-scale analysis of
the yeast proteome by multidimensional protein identification technology. Nat
Biotechnol 19, 242–247.
31. Gygi S.P., Corthals G.L., Zhang Y., Rochon Y. and Aebersold R. (2000) Evaluation
of two-dimensional gel electrophoresis-based proteome analysis technology. Proc
Natl Acad Sci USA 97, 9390–9395.
32. Kyte J. and Doolittle R.F. (1982) A simple method for displaying the hydropathic
character of a protein. J Mol Biol 157, 105–132.
33. Craven R.A., Totty N., Harnden P., Selby P.J. and Banks R.E. (2002) Laser
capture microdissection and two-dimensional polyacrylamide gel electrophoresis:
evaluation of tissue preparation and sample limitations. Am J Pathol 160, 815–822.
34. Li C., Hong Y., Tan Y.X., Zhou H., Ai J.H., Li S.J., Zhang L., Xia Q.C., Wu J.R.,
Wang Y. and Zeng R. (2004) Accurate qualitative and quantitative proteomic
analysis of clinical hepatocellular carcinoma using laser capture microdissection
coupled with isotope-coded affinity tag and two-dimensional liquid chromatog-
raphy mass spectrometry. Mol Cell Proteomics 3(4),399–409.
12
Label-Free LC-MS Method for the Identification
of Biomarkers
Richard E. Higgs, Michael D. Knierman, Valentina Gelfanova,
Jon P. Butler, and John E. Hale
Summary
Pharmaceutical companies and regulatory agencies are pursuing biomarkers as a means
to increase the productivity of drug development. Quantifying differential levels of proteins
from complex biological samples like plasma or cerebrospinal fluid is one specific
approach being used to identify markers of drug action, efficacy, toxicity, etc. Academic
investigators are also interested in markers that are diagnostic or prognostic of disease
states. We report a comprehensive, fully automated, and label-free approach to relative
protein quantification including: sample preparation, proteolytic protein digestion, LC-
MS/MS data acquisition, de-noising, mass and charge state estimation, chromatographic
alignment, and peptide quantification via integration of extracted ion chromatograms.
Additionally, we describe methods for transformation and normalization of the quantitative
peptide levels in multiplexed measurements to improve precision for statistical analysis.
Lastly, we outline how the described methods can be used to design and power biomarker
discovery studies.
Key Words: relative quantification; label-free quantification; biomarkers;
proteomics; LC-MS/MS.
1. Introduction
Recent advances in analytical technology, particularly mass spectrometry,
are finding broad applications in the search for biomarkers. Biomarkers may
be defined as indicators of biological processes and encompass a variety of
measures including imaging, polynucleotides, proteins, and small molecule
From: Methods in Molecular Biology, vol. 428: Clinical Proteomics: Methods and Protocols
Edited by: A. Vlahou © Humana Press, Totowa, NJ
209
210 Higgs et al.
metabolites, among others. These new biomarker discovery activities are
motivated by the need to improve diagnosis, guide-targeted therapies, and
monitor therapeutic efficacy and toxicity throughout a treatment regimen.
Biomarkers of drug efficacy or toxicity have the potential to shorten the drug
development timeline as they may provide early indications of a drug’s activity.
This potential for increased drug development productivity from high-quality
biomarkers has fueled increased attention from pharmaceutical, biotechnology,
and regulatory agencies alike (1,2). Within the field of protein biomarkers,
mass spectrometry is playing a central role in the discovery of biomarkers from
various biological sample matrices. Quantification of small organic molecules
using extracted ion chromatograms (XICs) from liquid chromatography mass
spectrometry (LC-MS) experiments has a long history in analytical chemistry.
Similar techniques using LC-MS experiments with proteolytic protein digests
are now routinely being applied to quantify peptide and protein levels in
biological samples. Early LC-MS peptide quantification methods relied on the
modification of peptides with reagents enriched in stable isotopes to introduce
mass shifts in the peptides from one sample in order to compare relative
peptide levels to another un-labeled sample (3,4). The number of biological
samples required for statistical power in many applications, the restriction that
study samples must be paired or pooled for these label-based methods, and the
increased cost due to specialized reagents have limited their application and
motivated the search for label-free methods of non-targeted protein profiling.
We report here a comprehensive analytical system to collect and automat-
ically process the data from non-targeted LC-MS/MS analyses of complex
protein mixtures. In contrast to pattern-based (5,6), difference based (7),or
identification-based quantification methods (8,9), the approach presented here
simply integrates the peptide parent ion current in order to obtain a relative
peptide level in each study sample. No labeling or pooling of study samples
is required. The output from this approach is an N×Ptable in which each
of Ppeptides has been quantified in each of the Nstudy samples. This table
maximizes the flexibility in downstream statistical data analysis including trans-
formation, normalization, and an analysis suited to the experimental design.
The described method is based on the collective efforts of the applied biochem-
istry and statistics groups within Lilly Research Laboratories (10,11,12).As
a broad-looking, discovery-oriented assay, it is important to note the limita-
tions imposed by the approach. An assay designed to detect and quantify many
analytes simultaneously compromises on sensitivity, selectivity, dynamic range,
and absolute quantification relative to a targeted assay designed for a particular
analyte. Ion suppression and co-elution of peptides from complex mixtures
have the potential to interfere with the ion current attributed to a peptide, thus
confounding any inference that may be made about the relative quantities of
Label-Free Biomarker Identification 211
the peptide. The limited dynamic range of these uncalibrated assays tends to
underestimate the magnitude of a change in protein levels for peptides that do
not lie near the linear portion of the instrument response curve. Nonetheless,
these non-targeted methods have shown promise in identifying relative changes
in protein levels that can be followed in subsequent studies using more targeted
assays (e.g., multiple reaction monitoring) (13) to verify the findings in a new
sample set.
The described method focuses on biomarker discovery from human plasma
and cerebrospinal fluid (CSF). Biomarker discovery from these fluids has
proven challenging as the highly abundant proteins (e.g., albumin, IgG) are
difficult to completely remove and tend to mask the detection of lower
abundance proteins that may be directly associated with the biology of interest.
However, the analytical and statistical methods described here are directly
applicable to more targeted sample matrices (e.g., tissues) in both clinical and
pre-clinical models that may increase the probability of technical success based
on samples more directly associated with the biology of interest with fewer
abundant, masking proteins to remove. Sample collection and handling proce-
dures are critical in reducing the overall variability in biomarker discovery
studies. Age, gender, diet, time of day, and medication may affect the plasma
or CSF protein profile and should be considered in study designs. Similarly,
consistent sample handling tailored to proteomics profiling (e.g., preservatives,
rapid sample freezing, controlling for blood contamination in CSF sampling,
number of sample freeze-thaw cycles, etc.) are important considerations to
ensure high-quality starting material. The proteome is arguably the most
modulated class of biomolecules in disease, treatment, and toxicity, resulting in
the promise of proteomics for biomarker discovery. Despite this promise and
rapid advancements in technology, progress has been slow (14,15). However,
with a refined strategy of: (1) applying non-targeted, hypothesis generation
methods like those described here to sample matrices proximal to the biology,
(2) using targeted MS assays to verify early discoveries in new sample sets,
and (3) clinical validation using established diagnostic assay formats (e.g.,
ELISAs), the potential to fulfill the promise is high by strategically applying
the right technology to the appropriate stage of the biomarker discovery life
cycle (16).
2. Materials
2.1. Albumin/IgG Depletion
1. Montage equilibration buffer, wash buffer, and columns are provided with the
Montage Albumin Deplete Kit™ (Millipore®).
2. ProteinG-Sepharose (Amersham Biosciences®).
212 Higgs et al.
2.2. Reduction, Alkylation, and Digestion
1. Denaturing solution and internal standard: 8 M urea in 100 mM (NH4)2CO3buffer
containing chicken lysozyme (Sigma, St Louis, MO; 10.4 μg/mL), pH 11.0.
2. Reduction/alkylation cocktail: 97.5% ACN, 2% iodoethanol, and 0.5%
triethylphosphine (v/v).
3. Trypsin solution: TPCK treated bovine pancreatic trypsin (Worthington,
Lakewood, NJ) is dissolved at 1 mg/mL in H2O and stored in single-use aliquots
at –80°C. Working solutions are prepared by diluting to 5 μg/mL in 100 mM
ammonium bicarbonate pH 8.0 prior to use.
2.3. HPLC
1. The C-18 reversed phase column was a Zorbax SB3001×50mm(Agilent).
2. Solvent A: 0.1% formic acid (Aldrich) in water (Burdick and Jackson HPLC
grade).
3. Solvent B: 50% acetonitrile, 0.1% formic acid (Aldrich) in water (Burdick and
Jackson HPLC grade).
4. Solvent C: 80% acetonitrile, 0.1% formic acid (Aldrich) in water (Burdick and
Jackson HPLC grade).
2.4. Mass Spectrometry
1. LTQ ion trap mass spectrometer (ThermoFinnigan).
3. Methods
3.1. Plasma Sample Preparation
3.1.1. Albumin/IgG Depletion
1. Dilute a 25 μL aliquot of plasma (1.25 mg protein assuming 50 mg/mL total
protein concentration) with Montage equilibration buffer to a volume of 200 μL
(see Note 1).
2. Add 100 μL of a 50% proteinG-Sepharose bead suspension and rock the mixture
for1hatRT.
3. Pellet the G-Sepharose beads at 2000 rpm for 2 min. and transfer 200 μL of the
effluent to a pre-equilibrated Montage column. Pre-equilibration was performed
with 400 μL of equilibration buffer and centrifugation for 2 min at 500×g
(see Note 2).
4. Centrifuge the Montage column at 500×gfor 2 min and re-apply the flow-thru to
the column and centrifuge again. Pass two consecutive 200 μL washes of Montage
wash buffer over the column via 500×g centrifugation for 2 min. (final volume
approximately 600 μL).
Label-Free Biomarker Identification 213
3.1.2. Reduction, Alkylation, and Digestion
1. Spike a 120 μL aliquot of the diluted and depleted plasma with 120 μL of the
denaturing and internal standard solution (see Note 3).
2. Add an equal volume (240 μ(L) of reduction/alkylation cocktail (see Note 4).
3. Cap the solutions and incubate for1hat37°C.
4. Speed vacuum the solutions to dryness (at least 3 h).
5. Re-dissolve the pellet in 600 μL of the working trypsin solution. Digest overnight
at 37°C (17).
3.2. Cerebrospinal Fluid Sample Preparation
3.2.1. Albumin/IgG Depletion
1. Dilute an aliquot of CSF (34 μg protein based on a Bradford total protein assay)
with Montage equilibration buffer to a volume of 200 μL (see Note 5).
2. Add 100 μL of a 50% proteinG-Sepharose bead suspension and rock the mixture
for1hatRT.
3. Pellet the G-Sepharose beads at 2000 rpm for 2 min and transfer 200 μL of
the effluent to a pre-equilibrated Montage column. Pre-equilibration is performed
with 400 μL of equilibration buffer and centrifugation for 2 min at 500×g(see
Note 2).
4. Centrifuge the Montage column at 500×gfor 2 min and re-apply the flow-thru to
the column and centrifuge again. Pass two consecutive 200 μL washes of Montage
wash buffer over the column via 500×gcentrifugation for 2 min (final volume
approximately 600 μL).
3.2.2. Reduction, Alkylation, and Digestion
1. Speed vacuum the CSF samples to approximately 30–50 μL and mix with 40 μL
of the denaturing and internal standard solution (see Note 3).
2. Add 100 μL of reduction/alkylation cocktail (see Note 4).
3. Cap the solutions and incubate for1hat37°C.
4. Speed vacuum the solutions to dryness (at least 3 h).
5. Re-dissolve the pellet in 600 μL of the working trypsin solution. Digest overnight
at 37°C (17).
3.3. HPLC Conditions
1. A Surveyor autosampler and MS HPLC pump (ThermoFinnigan) are used for
separation. 100 μL tryptic digests (4.2 μg plasma non-depleted equivalent protein
or 14 μg CSF non-depleted equivalent protein) onto the reversed phase column
at a flow rate of 50 μL/min (see Note 6). The gradient conditions are: 10–95% B
(90–5% A) over 120 min, followed by a 0.1 min ramp to 100% C, followed by
5 min at 100% C, followed by a 0.1 min ramp to 10% B (90% A), and hold for
214 Higgs et al.
17 min at 10% B (90% A). The effluent is diverted to waste for the first 2 min
to keep the mass spectrometer source clean.
2. Between each sample in the set, an injection of water is made and a shortened
(60 min) gradient, identical to the above, is performed to reduce carryover.
3.4. Mass Spectrometer Conditions
1. The total column effluent (50 μL/min) is connected to the electrospray interface
of the ion trap mass spectrometer.
2. The source is operated in positive ion mode with a 4.8 kV electrospray potential,
a sheath gas flow of 20 arbitrary units, and a capillary temperature of 225°C. The
source lenses should be set by maximizing the ion current for the 2+ charge state
of angiotensin.
3. Data are collected in the triple play mode with the following parameters: centroid
parent scan set to one microscan and 50 ms maximum injection time, profile
zoom scan set to three microscans and 500 ms maximum injection time, and a
centroid MS/MS scan set to two microscans and 2000 ms maximum injection
time (see Note 7).
4. Dynamic exclusion settings are set to a repeat count of one, exclusion list duration
of 2 min, and rejection widths of –0.75 m/z and +2.0 m/z.
5. Collisional activation is carried out with relative collision energy of 35% and an
exclusion width of 3 m/z.
6. Study samples should be injected in a random order to reduce any effects of
carryover or confounding with a non-random injection order (see Note 8).
7. All water blank samples should be analyzed by the mass spectrometer in the same
manner as study samples in order to monitor carryover (see Note 9).
3.5. Zoom Scan Data Processing
The data collected from a zoom scan triple-play experiment are used to
estimate the quality of the subsequent MS/MS spectrum, the charge state of
the peptide, and the monoisotopic and average mass of the peptide. The quality
estimate is used to eliminate those scan events that are triggered by noise
or small molecules from further downstream processing. Peptide mass and
charge state estimates are used in subsequent steps for peptide identification.
Eliminating low-quality scan events and more accurately estimating the charge
state and mass of peptides ultimately reduces the number of false positives that
must be dealt with at the peptide identification stage of the process.
1. Assume the charge state of the detected peptide is 1+.
2. Given the m/z of the scan event and the assumed charge state, estimate the
theoretical isotope distribution intensities for a peptide of the hypothesized mass
using the relationships given in Fig. 1 (see Note 10). Begin by determining the
relative intensity of the 12C peak (I0) using the relationship in Fig. 1A and the
MW for the assumed charge state. Next, estimate the relative peak intensity of
Label-Free Biomarker Identification 215
the 13C peak (I1) by multiplying the estimate of I0by the I1/I0ratio from Fig. 1B
using the MW for the assumed charge state. Isotope intensities I2and I3are
derived in a similar manner using the ratios from Fig. 1C–D at the MW for the
assumed charge state.
3. Convolve the estimated theoretical isotope stick spectrum with a Gaussian peak
shape that has a peak width similar to that produced in a typical zoom scan
spectrum (18). Linearly scale the result of this convolution such that the maximum
value is one.
0.0
0.0 1.0
1.0 2.0
l
0
/ max(l
0
,l
1
,l
2
,l
3
)
l
1
/ l
0
l
2
/ l
0
l
3
/ l
0
0.5 0.8
0.5 1.5
500 2500
Mono MVV
500 2500
Mono MVV
Mono MVV
500
(B)(A)
(D)(C)
2500
Mono MVV
500 2500
Fig. 1. Empirically derived relationships (from 15,493 example peptides) between
isotope peak intensities used to estimate the theoretical isotope pattern for a peptide
(A) I0/max(I0,I
1,I
2,I
3), non-linear least squares fit:
I0/maxI0I
1I
2I
3=1ifMW<1800
e−000132+MW −18000865 if MW ≥1800
(B) I1/I0, linear least squares fit:
I1/I0=−000498+0000560MW ,
(C) I2/I0, linear least squares fit:
I2/I0=−0367+0000516MW +159×10−7MW −1527342, and
(D) I3/I0, nonlinear least squares fit: I3/I0=00000605e000251MW −270×10−7MW 2.
Reprinted with permission from (10).
216 Higgs et al.
4. Convolve the result from step 3 above with the measured zoom scan to obtain the
matched filter output between the expected zoom scan spectrum from the assumed
charge state and the measured zoom scan spectrum. Record the maximum value
of the output of this convolution along with the x-axis (m/z) value where the
maximum occurred.
5. Repeat steps 2–4 above for an assumed charge state of 2+,3
+, and 4+. The
detected peptide charge state and mass are estimated from the best match between
the observed zoom scan spectrum and the theoretically derived spectrum for
the possible charge states of 1+,2
+,3
+, and 4+. The cross-correlation between
the best matching theoretical isotope pattern at the m/z shift value associated
with the convolution maximum and the measured zoom scan is used as an
intensity-independent matching score between the measured and the best matching
theoretical spectrum. Triple play events with a cross-correlation score greater
than 0.6 are retained for identification. Triple plays below this threshold represent
scans that are not peptides, a mixture of several peptides in the ion trap, or
very low signal-to-noise measurements. These lower quality scan events are not
retained for any further processing.
3.6. MS/MS Spectral Filtering
In order to reduce the effect of MS/MS noise peaks on the identification of
peptides, a dynamic MS/MS noise level is estimated for each spectrum. This
noise level estimate is then subtracted from all MS/MS peak intensities with
any resulting differences less than zero set to zero. The spectral noise level is
estimated based on the observation that ideal MS/MS spectra of peptides have
relatively few peaks (e.g., y-ions, b-ions, adducts, etc.) in a theoretical or high
signal-to-noise ratio spectrum, while noisy MS/MS spectra typically have a
high density of peaks within a local m/z neighborhood (interpreted as chemical
noise). Therefore, the filtering approach uses a percentile of the peak intensities
within a local m/z neighborhood as the noise estimate, where the percentile
used is based on the density of peaks in the neighborhood – a higher peak
density results in a higher percentile to estimate the local noise level, a lower
peak density results in a lower percentile to estimate the local noise level.
1. Bin the MS/MS spectrum into a vector of equally spaced m/z values (bin width
of 0.1 m/z).
2. At 200 equally spaced m/z value design points between the maximum and
minimum observed m/z values observed in the MS/MS spectrum, estimate the
local peak density by counting the number of non-zero intensities in a ±20 m/z
window around each of the 200 design points. Define the local peak density at
these 200 design points as the number of non-zero peaks counted divided by 40
(peaks per m/z).
3. Transform the local peak density values to a filtering percentile value using the
relationship shown in Fig. 2.
Label-Free Biomarker Identification 217
Fig. 2. Filtering percentile as a function of local MS/MS peak density. Peak density
is defined as the number of MS/MS peaks in a 40 m/z window divided by 40.
Filtering Percentile =0 if PeakDensity ≤01
075
1+e015−PeakDensity
005
if PeakDensity >01
Reprinted with permission from (10).
4. Obtain an initial noise level estimate by the percentile of MS/MS peak intensities
at each of the 200 design points, where the percentile used at each point is derived
from step 3 above (see Note 11).
5. Smooth the initial noise estimates with a Gaussian kernel smooth (150 m/z
bandwidth) and interpolate between the 200 design points to obtain the final
MS/MS noise estimate at each measured m/z value. Subtract this estimate from
the measured MS/MS peak intensities and set any negative values to zero. An
example of a high and low signal-to-noise MS/MS spectrum and the resulting
estimated noise levels is shown in Fig. 3.
3.7. Peptide Identification
A detailed description of peptide identification is beyond the scope of this
chapter, but some general discussion is warranted given the importance of the
subject and its linkage to quantification with the proposed method. The primary
problem with peptide identification is controlling for false-positive identifica-
tions while maintaining a reasonable sensitivity to detect correct identifications.
Our approach utilizes the outputs of two search engines, Sequest (19) and
X! Tandem (20), along with other descriptive features of identification (e.g.,
charge state, peptide length, etc.) as inputs to a classifier that has been trained
218 Higgs et al.
(A)
50,00020,0000
500 1000 1500
200 600 1000 1400
m/z
m/z
350,000150,0000
Intensity Intensity
(B)
Fig. 3. Example MS/MS spectra and their estimated noise levels. 443 original peaks
reduced to 118 peaks above estimated noise level in high-noise spectrum (A). 589
original peaks reduced to 173 peaks above estimated noise level in lower noise spectrum
(B). Reprinted with permission from (10).
to identify correct identifications (21). The output of the classifier provides a
unit-less score indicative of the likelihood of a correct identification. False-
positive identifications are controlled by running the searches against reversed
versions of the protein databases and estimating the p-values: the probability
of observing a model score from the reversed database search that exceeded
the observed score from the correct database. P-values alone are insufficient
due to the large number of tests (identifications) being done (i.e., with a 0.05
p-value cutoff, 5% of identifications declared correct would in fact be incorrect
in the null condition where there are truly no matches to any MS/MS spectra).
To account for multiple testing, false discovery rates (FDRs) (q-values) for
Label-Free Biomarker Identification 219
peptide identifications are estimated from p-values using the method described
by Benjamini and Hochberg (22). Peptides with identification q-values less than
a threshold, say 0.10, are retained for quantification. Proteins identified by only
one peptide are visually examined to eliminate obvious incorrect identifications
(e.g., less than four consecutive y- or b-ions). We estimate that the proportion of
false identifications using such a procedure is less than or equal to 2%. Overall,
the method is similar in strategy to PeptideProphet (23) with the following
extensions: multiple search engines are employed, a more flexible classifier
(e.g., Random Forests) is used, and statistical significance is estimated from a
null distribution of classifier scores derived from reversed database searching
instead of fitting a mixture model to the distribution of classifier output scores.
The method is described in detail in Higgs et al.(11).
In general, we typically restrict biomarker hypothesis generation to identified
peptides. The same relative quantification method can be used with unidentified
peptides (MS features), although in practice these features need to be identified
to be of practical use to clinicians and biologists. To maximize the coverage
of proteins identified in a study, identifications from all samples in the study
are pooled and used to create a list of peptides to quantify in each sample.
Thus, a confident identification needs to be made once out of a sample in order
for the associated peptide ion current to be quantified in all study samples.
Pooling the identifications across all samples in a study significantly increases
the number of identifications relative to the number of identifications from any
single sample.
3.8. Chromatographic Alignment
Variability in the abundance of individual peptides between different samples
may result in that peptide triggering an MS/MS scan in one sample and not in
another. The area of this peptide may still be extracted from the primary mass
spectrum in each sample. However, doing so requires high-quality chromato-
graphic alignment between the samples so that a consistent region in the
extracted ion chromatogram (XIC) is used for integration across all samples in
a study. Large biomarker studies can produce chromatographic retention time
shifts greater than 1 min between pairs of samples run several days and many
samples apart. Simply expanding the integration window by 1 or 2 min to
account for chromatographic variability is not an option in our experience as
we are analyzing complex samples with multiple co-eluting peaks at most XIC
masses. An expanded integration window that includes multiple peaks masks
the quantification of individual peptides, produces results that are confounded
with multiple peptides contributing to a value, and increases variability. Peak
picking is another option, but was not applied here due to the computational
220 Higgs et al.
cost as well as the inherent heuristic nature of peak picking algorithms with
an associated variability in what is being integrated. We have found a simple
pair-wise alignment between all samples and a select reference sample in the
study to work well for numerous biomarker discovery projects. This approach
to alignment is founded on the following assumptions: (a) the samples included
in the study are generally quite similar to each other with respect to their peptide
content (i.e., there are many peptides or landmarks in common between the
samples), (b) the same chromatographic conditions are used for each sample in
the study, and (c) in a local region of retention time, the retention time offset
between any two samples is approximately constant (see Note 12).
1. Identify the landmarks in the reference sample by taking all triple-play scan events
with a zoom scan cross-correlation score of 0.65 or greater. This set of reference
sample landmarks will be matched against other samples in the study.
2. Identify the matching landmarks in a study sample by declaring a landmark match
if the sample and reference triple-play events have: (a) the retention time of the
triple play event between the samples is within a user-specified amount (5 min),
(b) the charge state of the peptide matches, (c) the m/z value of the monoisotopic
peak from the zoom scans is within a user-specified amount (0.7 Da) between the
two samples, (d) the zoom scan cross-correlation coefficient of both peptides to
their respective theoretical isotope patterns exceeds a threshold (0.65), and (e) the
similarity between the corresponding MS/MS spectra exceeds a threshold (e.g.,
0.75). The MS/MS similarity metric has been implemented as a cross-correlation
coefficient between two MS/MS spectra following a convolution of each MS/MS
stick-spectrum with a Gaussian peak shape.
3. For each matching pair of landmarks identified in step 2 above, generate the
XIC for the feature in a local retention time window (e.g., ±5 min of scan event
time in each sample). Convolve the two XICs to identify the time shift value that
maximizes the convolution result between the landmark XICs in both samples.
Record the time shift and cross-correlation at the optimal shift value for each
landmark. The cross-correlation value will be used as a weighting factor in the
subsequent smoothing step below.
4. The optimal time shift values for each pair of landmarks between a sample and the
reference defines a warping function that can be used to transform the retention
time values of a sample to the reference. Estimate a smooth warping function
by fitting a weighted loess (24) to the time shift versus retention time values
for each sample. The loess should be done in a weighted manner using the XIC
cross-correlation values from step 3 above as weights. The result is a smooth
function that can be used to transform a sample’s retention time to a common
time defined by the reference sample Fig. 4.
5. The loess warping function for a sample is then applied to all the retention times
in the chromatogram (landmark or not). Thus, all samples in a study are projected
onto the same retention time scale. The warping function between two samples is
generally not monotonic over the entire retention time range, and no restriction
Label-Free Biomarker Identification 221
0 20 40 60 80 100 120
Ret. Time (min)
Shift (min) n
=
462
0.50.0
–0.5
Fig. 4. Example chromatographic alignment (“warping”) function between two rat
serum samples. Retention time shift (min) vs. retention time (min) for 462 landmark
peptides are plotted with the resulting loess fit. Reprinted with permission from (10).
on overall monotonicity is used in our estimate of the warping function. We
do, however, preserve the overall rank order of the retention times following
alignment by constraining the bandwidth (span = 0.5) used in the loess fitting
(24) (see Note 13).
3.9. Peptide Quantification
Relative quantification of peptides is carried out by integration of the XIC
peak (using normalized retention times from the chromatographic alignment)
from the primary mass spectrum within each sample. A list of peptides to
integrate within each sample is constructed by pooling together all triple-play
events across all the samples. This pooling can be done with or without the use
of peptide identification. As previously noted, we typically restrict the analyses
to identified peptides. For each identified peptide, perform the following
steps:
1. For each sample in which the peptide was identified, extract the XIC for the
peptide and compute the centroid (weighted average of retention time values
where weighting factor is the XIC ion current) of the XIC in a small retention
time neighborhood (–0.5 min to +1.0 min from triple-play trigger time) using the
aligned time values in the XIC. Compute the mean centroid time for the peptide
over all samples in which the peptide was identified. Also compute the mean
average m/z value estimated from the zoom scan spectrum for each sample in
which the peptide was identified.
222 Higgs et al.
2. For each sample in the study, create an XIC for the peptide using the mean zoom
scan average m/z value determined in step 1.
3. Estimate a local XIC baseline level and subtract the baseline from the XIC
intensity values from each sample. A local linear baseline can be estimated by
fitting a line between the lowest intensity XIC point before the peak and the lowest
intensity XIC point following the peak in a local neighborhood (e.g., 5 min).
This simple local linear baseline estimate always results in a baseline estimate
below the signal intensity in the local neighborhood, leading to a low bias in the
estimated baseline. For large peaks, this bias is negligible but for small peaks
the bias may have a more pronounced effect on quantification. Alternatively,
an asymmetric least squares smoothing approach may be used to estimate the
baseline XIC values in order to reduce the potential bias with the simple local
linear approach (25).
4. A fixed retention time window (±0.5 min for the chromatography described)
around the mean centroid time value described in step 1 is used for integration.
The width of this window is dependent on the chromatography method used.
For the chromatography method reported here, the peak width remains relatively
constant across the HPLC gradient (i.e., no band-broadening is observed).
If band-broadening is observed, then the integration window width should
be modeled as a function of the retention time (e.g., integration window
width = intercept + slope × retention time).
5. Integrate the baseline corrected XIC values within the fixed retention time window
for each sample in the study using a numerical integration algorithm such as the
trapezoid rule. Record the XIC area values for each peptide in each sample. An
example of XIC integration for a small study is shown in Fig. 5.
3.10. Data Transformation and Normalization
Following the integration of peptide-specific XIC peaks in all study samples,
we have a rectangular data table with Nrows corresponding to Nsamples in
the study, and Pcolumns corresponding to peptides detected in the study. The
cell values in this data table are the peptide peak areas. With this table in hand,
the usual operations of transformation and normalization may be applied prior
to any statistical analysis.
1. Peptide peak areas are approximately log-normal distributed. Apply a log2trans-
formation to all peak area values (see Note 14).
2. Normalize the log2transformed peak areas using a quantile normalization
procedure (26) (see Note 15).
3. Normalized log2peptide areas may be used directly as input to the statis-
tical analysis for the study (peptide level analysis). Additionally, the average
of normalized log2peptide areas for all the peptides identified from a protein
can be used as an overall estimate of the protein level (protein level analysis,
see Note 16).
Label-Free Biomarker Identification 223
Fig. 5. XICs from the 2+–1 macroglobulin peptide ATPLSLCALTAVDQSVLL-
LKPEAK for eight rat serum samples following chromatographic alignment. Note that
the peak from all samples fits within the highlighted [83.2, 84.2] integration region.
Reprinted with permission from (10).
224 Higgs et al.
3.11. Study Design, Power, Sample Size, and Analysis
Our strategy of producing an N×Ptable of relative peptide levels allows
the flexibility for the analysis to be done in a manner consistent with the
study design. Note that no part of the described method imposes any limitation
on the final study statistical analysis (e.g., pooling of samples, subtractive-
or difference-based methods, etc.). In general, the statistical analysis used for
identifying potential protein biomarkers in a study should follow the same
approach as a primary clinical endpoint analysis would take (i.e., a simple
paired design should be analyzed with a paired t-test, a crossover design with
repeated measures within period should be analyzed as a crossover study with
repeated measures within period, etc.).
An analysis of a single clinical endpoint may use the familiar type I error
threshold of 0.05 as a measure of statistical significance. This approach does not
work well when testing hundreds or thousands of proteins in a study because, by
definition, 5% of all p-values from a null experiment (an experiment in which
there is truly no treatment or group effect) will have a p-value less than 0.05.
The Bonferroni approach to control the family-wise type I error (controlling
for no errors in the set of declared changes) has been commonly employed as
a means to control false-positive findings (27). However, many investigators
doing proteomic hypothesis generation are willing to tolerate some level of false-
positive findings in a declared set as long as it is relatively low and estimated.
The use of FDR as a means to identify a set of declared findings with a
specified proportion of false-positives has been widely applied in genomics (22)
and is the current recommendation for proteomic hypothesis generating experi-
ments. There are numerous estimators of FDR (28,29) with the original method
described by Benjamini and Hochberg used in the work presented here (22).
Just as multiple comparisons should be considered in the analysis of study
data, these should also be considered at the design stage of a new study
aimed at generating hypotheses from highly multiplexed measurements like
proteomics. This is a relatively new field of research with several methods
recently reported (30,31,32,33). A simple approach originally suggested by
Benjamini and Hochberg (22), and adapted by Bemis (34), uses traditional
sample-size calculations with the following expression for average type I error
(ave) over a set of tested hypotheses: ave =fave q∗m1
m1+m01−q∗where fave is the
average power of hypothesis tests conducted in a study, q∗is the rate at which
FDR is to be controlled, m0is the number of true null hypotheses tested, and m1
is the number of true alternative hypotheses tested. Sample-size estimates are
made by first estimating ave using the desired values for fave and q∗, assumed
values for m0and m1, and existing sample size calculators using for a given
study design. An example set of sample-size curves using ave this approach
for the two-sample t-test design is given in Fig. 6.
Label-Free Biomarker Identification 225
Fig. 6. Estimated sample sized required to detect protein changes in a two-sample
t-test design. Number of subjects in each of the two groups is plotted against the
detectable effect size expressed as a fold-change. Four different levels of total variability
are shown (10% CV, 20% CV, 30% CV, and 40% CV). Sample size estimates were
made using 85% power, a 0.10 target FDR for declaring significance, and an estimated
proportion of true null hypotheses, m0
m0+m1, set to 0.98.
4. Notes
1. We find that plasma total protein concentration, as measured by a Bradford
assay, has a total coefficient of variation (CV) of approximately 11% (includes
inter-subject, intra-subject, and assay error) and ranges between approximately
48 and 68 mg/mL (12). Due to the apparent highly regulated plasma total protein
concentration, it is not generally necessary to measure total protein concentration
for each sample in a study in order to load a consistent amount of protein.
2. The depletion material used is based on a dye affinity removal method for
albumin. There are commercially available antibody-based depletion kits that
may improve albumin removal at a reasonable cost. Abundant protein depletion
is an open and active research area at the time of this writing.
3. Chicken lysozyme is added as a spiked internal standard at this stage in order
to qualitatively assess the digestion efficiency as well as to quantitatively assess
the measurement error across the samples in a study. Other internal standard(s)
could also be used.
4. The reduction/alkylation solution should be prepared just before use.
Triethylphosphine is pyrophoric and should be handled in a fume hood in accor-
dance with the material safety data sheet. The use of volatile reagents for this step
226 Higgs et al.
reduces the variability in the sample prep by minimizing sample handling steps
and removing the majority of reduction and alkylating reagents. The digestion is
performed with trypsin, which is sensitive to the presence of reducing reagents.
5. We find that CSF total protein concentration, as measured by a Bradford assay,
has a total CV of approximately 27% (includes inter-subject, intra-subject, and
assay error with the additional total variability relative to plasma total protein
attributed to a higher CSF inter-subject variance) with a range between approx-
imately 0.12 and 0.41 μg/mL (12). The higher overall variability is attributed
to a significantly higher inter-subject variability relative to plasma total protein
(12). Due to the higher variability with CSF total protein, we use the results of
Bradford total protein assay to process a consistent total CSF protein amount in
the proteomics assay.
6. The HPLC pumps must be capable of producing a smooth gradient at 50 μL/min.
The gradient formation should be verified by using water in A and 1% acetone
in water for B and running the gradient with UV monitoring at 254 nm. New
HPLC columns should be conditioned with at least four runs of digested serum
before use in the method.
7. The mass spectrometer’s source should be carefully cleaned to minimize chemical
noise. Monitor above 300 m/z and try to maximize the injection time as this is
directly proportional to achievable dynamic range in an ion trap mass spectrometer.
The spray conditions should be optimized for a peptide of about ˜1700 Da.
8. Alternatively, a design could be used to balance various study factors (e.g.,
treatment, gender, age, etc.) with injection order. This approach may be
most appropriate for small studies (e.g., <15 samples) where an unfortunate
randomization could result in a confounding of injection order with an important
study factor, like treatment.
9. Using the methods as described for plasma, we have found that the total area
under the base peak chromatogram for water blanks is <1.5% the total area for
the corresponding preceding plasma sample (<0.0225% total carryover between
plasma samples). The largest peak in plasma water blanks is an albumin peptide
(2+ALVLIAFAQYLQQCPFEDHVK) with a peak area of 9% the value of the
preceding plasma sample (0.81% carryover of this peptide estimated between
plasma samples). For CSF, we have found that the total area under the base peak
chromatogram for the water blanks is <6% the total area for the corresponding
plasma sample (<0.36% total carryover between CSF samples). The largest peak
in CSF water blanks is an acidic transthyretin peptide (2+TSESGELHGLT-
TEEEFVEGIYKVEIDTK) with a peak area of 32% the value of the preceding
CSF sample (10% carryover of this peptide estimated between CSF samples)
(12). A compressed, 60 min gradient can be used for water blank injections in
order to reduce the overall cycle time.
10. Theoretical isotope relative heights were derived from an analysis of 15,493
peptides ranging in length from 2 to 38 residues with a median length of 13. We
have denoted the intensities of the four isotope peaks I0for the 12C monoisotopic
peak, I1for the +1 13C isotopic peak, I2for the +2 13 C isotopic peak, and I3for
Label-Free Biomarker Identification 227
the +3 13C isotopic peak. The 15,493 example peptides were then used to derive
relationships for I0/max (I0,I
1,I
2,I
3), I1/I0,I2/I0, and I3/I0as functions of the
peptide monoisotopic molecular weight (Fig. 1).
11. Percentile transformation is done to define the noise level as the Xth percentile
of the peak intensities in a local m/z neighborhood where Xis dependent on
the peak density in the neighborhood (higher peak density–>higher percentile–
>higher estimated noise level).
12. One potential improvement to this alignment strategy would be to create a
composite list of landmarks across all study samples instead of relying on a single
sample to serve as the retention time reference. This could easily be accomplished
by grouping or clustering landmarks from all samples enforcing a match on m/z,
charge state, retention time, and MS/MS spectral similarity. This has not been
employed yet due to the increased computational cost and the lack of data demon-
strating any significant problems with the single reference sample approach. In
practice, several different samples are evaluated as potential alignment reference
samples, and the best sample based on a qualitative assessment of the alignment
warping functions is chosen.
13. A visual examination of the alignment warping functions for all samples included
in a study is an effective means to detect and diagnose chromatography problems
encountered in the analysis of dozens of study samples. For example, oscillatory
warping functions have been associated with pump mixing problems while large
magnitude mostly linear warping functions have been associated with column
degradation.
14. Log2is convenient because a unit change can be interpreted as a twofold change
on the original scale.
15. Normalization can be particularly important for minimizing systematic biases in
ion current introduced by sample collection and handling, sample concentration,
instrument sensitivity drift during the course of data acquisition, etc. The spiked
internal standard, chicken lysozyme can be helpful in diagnosing and monitoring
ion intensities before and after normalization. Quantile normalization assumes
that the overall distribution of log2peptide peak areas is unchanged from sample
to sample. This is generally a reasonable assumption, but there are cases where
a treatment effect may modulate the level of most of the proteins detected in
a study, and in such cases quantile normalization should not be used. In these
cases, the spiked internal standard, chicken lysozyme can be used to normalize
any systematic effects of the process on ion current occurring only after the
standard was spiked.
16. In practice, we will analyze a study at both the peptide and protein levels.
Peptide-level analyses are generally specific to the identified peptide and allow
the opportunity to discover biologically related changes in peptide level due
to processing of a specific region of a protein. Protein-level analyses provide
additional statistical power to detect smaller magnitude changes in protein levels
since we are averaging multiple peptide values, all of which have a high positive
covariance.
228 Higgs et al.
Acknowledgments
We thank John Saalwaechter and Andrew Kaczorek and the entire scien-
tific computing team for their efforts in developing and maintaining a high-
availability grid-computing environment used for this work. We also thank
Jude Onyia and the statistical and mathematical sciences management team for
supporting us in the development of these methods.
References
1. FDA Critical Path Initiative 2006 (http://www.fda.gov/oc/initiatives/criticalpath).
2. NIH Road Map for Medical Research 2006 (http://www.nihroadmap.nih.gov/
index.asp).
3. Gygi, S.P., Rist, B., Gerber, S.A., Turecek, F., Gelb, M.H., and Aebersold, R. 1999.
Quantitative analysis of complex protein mixtures using isotope-coded affinity
tags. Nat. Biotechnol.17: 994–999.
4. Aggarwal, K., Choe, L.H., and Lee, K.H. 2006. Shotgun proteomics using the
iTRAQ isobaric tags. Brief. Funct. Genomic. Proteomic.5: 112–120.
5. Petricoin, E.F., Ardekani, A.M., Hitt, B.A., Levine, P.J., Fusaro, V.A.,
Steinberg, S.M., Mills, G.B., Simone, C., Fishman, D.A., Kohn, E.C. et al 2002. Use
of proteomic patterns in serum to identify ovarian cancer. Lancet 359: 572–577.
6. Radulovic, D., Jelveh, S., Ryu, S., Hamilton, T.G., Foss, E., Mao, Y., and Emili, A.
2004. Informatics platform for global proteomic profiling and biomarker discovery
using liquid chromatography-tandem mass spectrometry. Mol Cell Proteomics 3:
984–997.
7. Wiener, M.C., Sachs, J.R., Deyanova, E.G., and Yates, N.A. 2004. Differential
mass spectrometry: a label-free LC-MS method for finding significant differences
in complex peptide and protein mixtures. Anal. Chem.76: 6085–6096.
8. Gao, J., Opiteck, G.J., Friedrichs, M.S., Dongre, A.R., and Hefta, S.A. 2003.
Changes in the protein expression of yeast as a function of carbon source.
J. Proteome. Res.2: 643–649.
9. Colinge, J., Chiappe, D., Lagache, S., Moniatte, M., and Bougueleret, L. 2005.
Differential Proteomics via probabilistic peptide identification scores. Anal. Chem.
77: 596–606.
10. Higgs, R.E., Knierman, M.D., Gelfanova, V., Butler, J.P., and Hale, J.E. 2005.
Comprehensive label-free method for the relative quantification of proteins from
biological samples. J. Proteome. Res.4: 1442–1450.
11. Higgs, R.E., Knierman, M.D., Freeman, A.B., Gelbert, L.M., Patil, S.T., and
Hale, J.E. 2007. Estimating the statistical significance of peptide identifications
from shotgun proteomics experiments. J. Proteome. Res.6: 1758–1767.
12. Patil, S.T., Higgs, R.E., Brandt, J.E., Knierman, M.D., Gelfanova, V., Butler, J.P.,
Downing, A.M., Dorocke, J., Dean, R.A., Potter, W.Z. et al. 2007. Identifying
pharmacodynamic protein markers of centrally active drugs in humans: a pilot
study in a novel clinical model. J. Proteome. Res.6: 955–966.
Label-Free Biomarker Identification 229
13. Anderson, L., and Hunter, C.L. 2006. Quantitative mass spectrometric multiple
reaction monitoring assays for major plasma proteins. Mol Cell Proteomics 5:
573–588.
14. Anderson, N.L., and Anderson, N.G. 2002. The human plasma proteome: history,
character, and diagnostic prospects. Mol Cell Proteomics 1: 845–867.
15. Gutman, S., and Kessler, L.G. 2006. The US Food and Drug Administration
perspective on cancer biomarker development. Nat. Rev. Cancer 6: 565–571.
16. Rifai, N., Gillette, M.A., and Carr, S.A. 2006. Protein biomarker discovery and
validation: the long and uncertain path to clinical utility. Nat. Biotechnol.24:
971–983.
17. Hale, J.E., Butler, J.P., Gelfanova, V., You, J.S., and Knierman, M.D. 2004.
A simplified procedure for the reduction and alkylation of cysteine residues in
proteins prior to proteolytic digestion and mass spectral analysis. Anal. Biochem.
333: 174–181.
18. Proakis, J.G., and Manolakis, D.G. 1992. Digital Signal Processing – Principles,
Algorithms and Applications. Prentice Hall, New York, NY.
19. Eng, J.K., Mccormack, A.L., and Yates, J.R. 1994. An approach to correlate tandem
mass spectral data of peptides with amino acid sequences in a protein database.
Journal of the American Society for Mass Spectrometry 5: 976–989.
20. Craig, R., and Beavis, R.C. 2003. A method for reducing the time required to match
protein sequences with tandem mass spectra. Rapid Commun. Mass Spectrom.17:
2310–2316.
21. Ulintz, P.J., Zhu, J., Qin, Z.S., and Andrews, P.C. 2006. Improved classifi-
cation of mass spectrometry database search results using newer machine learning
approaches. Mol Cell Proteomics 5: 497–509.
22. Benjamini, Y., and Hochberg, Y. 1995. Controlling the false discovery rate - a
practical and powerful approach to multiple testing. Journal of the Royal Statistical
Society Series B-Methodological 57: 289–300.
23. Keller, A., Nesvizhskii, A.I., Kolker, E., and Aebersold, R. 2002. Empirical statis-
tical model to estimate the accuracy of peptide identifications made by MS/MS
and database search. Anal. Chem.74: 5383–5392.
24. Cleveland, W.S., Grosse, E., and Shyu, W.M. 1992. Local regression models.
In Statistical Models in S. J.M. Chambers and T.J. Hastie, eds. Wadsworth &
Brooks/Cole, Pacific Grove, CA.
25. Boelens, H.F., Dijkstra, R.J., Eilers, P.H., Fitzpatrick, F., and Westerhuis, J.A. 2004.
New background correction method for liquid chromatography with diode array
detection, infrared spectroscopic detection and Raman spectroscopic detection. J.
Chromatogr. A 1057: 21–30.
26. Bolstad, B.M., Irizarry, R.A., Astrand, M., and Speed, T.P. 2003. A comparison
of normalization methods for high density oligonucleotide array data based on
variance and bias. Bioinformatics 19: 185–193.
27. Miller, R.G., Jr. 1991. Simultaneous Statistical Inference. Springer-Verlag,
New York.
230 Higgs et al.
28. Butler, K.W., Deslauriers, R., Geoffrion, Y., Storey, J.M., Storey, K.B., Smith, I.C.,
and Somorjai, R.L. 1985. 31P nuclear magnetic resonance studies of crayfish
(Orconectes virilis). The use of inversion spin transfer to monitor enzyme kinetics
in vivo. Eur. J. Biochem.149: 79–83.
29. Efron, B. 2004. Large-scale simultaneous hypothesis testing: the choice of a null
distribution. J. Am. Stat. Soc.99: 96–104.
30. Pounds, S., and Cheng, C. 2005. Sample size determination for the false discovery
rate. Bioinformatics 21: 4263–4271.
31. Hu, J., Zou, F., and Wright, F.A. 2005. Practical FDR-based sample size calcula-
tions in microarray experiments. Bioinformatics 21: 3264–3272.
32. Jung, S.H. 2005. Sample size for FDR-control in microarray data analysis. Bioin-
formatics 21: 3097–3104.
33. Li, S.S., Bigler, J., Lampe, J.W., Potter, J.D., and Feng, Z. 2005. FDR-controlling
testing procedures and sample size determination for microarrays. Stat. Med.24:
2267–2280.
34. Bemis, K.G. 2005. Statistical Issues with Mass Spectrometry Proteomics for
Biomarker Discovery. In International Workshop on Statistical Methodology in
Clinical and Nonclinical R&DDIA conference, Nice, France.
13
Analysis of the Extracellular Matrix and Secreted Vesicle
Proteomes by Mass Spectrometry
Zhen Xiao, Thomas P. Conrads, George R. Beck, Jr.,
and Timothy D. Veenstra
Summary
The extracellular matrix (ECM) and secreted vesicles are unique structures outside of
cells that carry out dynamic biological functions. ECM is created by most cell types and
is responsible for the three-dimensional structure of the tissue or organ in which they
are originated. Many cells also produce or secrete specialized vesicles into the ECM,
which are thought to influence the extracellular environment. ECM is not s a physical
structure to connect cells in a tissue or organ. The proteins in ECM and secreted vesicles
are critical to cell function, differentiation, motility, and cell-to-cell interaction. Although
a number of major structural proteins of ECM and secreted vesicles have long been
known, an appreciation of the role of less-abundant non-collagenous proteins has just
begun to emerge. This chapter outlines a series of methods used to isolate and enrich
ECM constituents and secreted vesicles from bone-forming osteoblast cells, enabling
comprehensive profiles of their proteomes to be obtained by mass spectrometry. These
methods can be easily adapted to study ECM and secreted vesicles in other cell types,
primary cell cultures derived from animal models, or tissue specimens.
Key Words: extracellular matrix; matrix vesicle; osteoblast; proteomics; mass
spectrometry.
1. Introduction
Most cells reside in a matrix environment called the extracellular matrix
(ECM), which offers the structural and nutritional support as well as a protective
barrier required for cells to survive, interact, and differentiate. In addition to
From: Methods in Molecular Biology, vol. 428: Clinical Proteomics: Methods and Protocols
Edited by: A. Vlahou © Humana Press, Totowa, NJ
231
232 Xiao et al.
the intracellular and tissue-related processes, it is becoming increasingly clear
that alterations in the ECM can affect the pathogenesis of the disease. While
much effort has been devoted to the understanding of intracellular processes,
the characteristics and functions of ECM have not been equally well studied.
The evidence gathered to date has shown that ECM is a complicated organelle
formed of various proteins that play central roles in cell differentiation,
migration, and cell-to-cell communication (1,2,3). The complexity of ECM is
exemplified in the structure of a skeleton. The formation and homeostasis of
bone is an ongoing process throughout life, and involves the recruitment, repli-
cation, and differentiation of osteoblasts and osteoclasts (4). Osteoblasts are
derived from mesenchymal stem cells and have the potential to further develop
into either osteocytes or lining cells. When induced by the appropriate stimuli,
such as ascorbic acid and -glycerophosphate, osteoblasts undergo proliferation
and maturation toward the osteocyte phenotype (Fig. 1)(5). This process is
accompanied by the accumulation of an ECM and ultimately mineralization of
the ECM in the form of hydroxyapatite (6). The deposition of hydroxyapatite
in ECM is initiated by a unique type of vesicles secreted by osteoblasts, called
matrix vesicles (MVs). With diameters ranging from 30–300 nm, these vesicles
reside in the ECM and play a critical role in mineralization (7,8). They serve
as nucleation sites for mineralization and sustain the accumulation of ECM (9).
A number of proteins, such as annexins and phosphatases, have been identified
within MVs. These proteins are responsible for the enrichment of calcium
and phosphate within the vesicles (8,10,11,12,13). Although the presence and
Fig. 1. The three-stage timeline of the osteoblast cell differentiation. The mineral
deposition is visualized by alizarin red staining of the osteoblasts cultured in the
differentiation medium.
Analysis of ECM and Secreted Vesicle Proteomes 233
function of other proteins are largely unknown, changes in ECM and MV
proteins are associated with diseases such as osteoporosis (14), arteriosclerosis
(15,16,17,18), tumor development, and metastasis (19,20,21,22). A compre-
hensive profile of the proteins present in these extracellular organelles enables
a greater understanding of pathophysiology underlying these clinical manifesta-
tions. The development of mass spectrometry (MS) technology combined with
appropriate protein enrichment and peptide separation strategies has made this
aim achievable (23,24,25,26).
This chapter describes the extraction of ECM constituents and MVs from
an osteoblast cell line MC3T3-E1 followed by the analysis of their respective
proteomic profiles by liquid chromatography (LC) fractionation combined with
MS analysis (27). The ECM and MVs are isolated and enriched using centrifu-
gation and enzymatic approaches. The enrichment of MVs is confirmed by the
measurement of elevated alkaline phosphatase (ALP) activity. Following the
creation of a complex mixture of peptides via a tryptic digestion of the extracted
proteins, this mixture is fractionated using strong cation exchange (SCX) LC.
These fractions are analyzed by nanoflow reversed-phase LC-tandem mass
spectrometry (nanoRPLC-MS/MS), and proteins are identified by searching the
data against appropriate proteomic database.
2. Materials
2.1. Cell Culture
1. MC3T3-E1 pre-osteoblast cell line (see Note 1)
2. Cell culture medium MEM (Irvine Scientific, Santa Ana, CA)
3. Fetal bovine serum (Atlanta Biologicals, Atlanta, GA)
4. Penicillin-streptomycin solution (10,000 I.U./ml penicillin, 10,000 μg/ml strepto-
mycin) (Invitrogen Corp., Carlsbad, CA)
5. 200 mM of l-glutamine (Invitrogen Corp.)
6. Growth medium: MEM supplemented with 10% fetal bovine serum, 50 U/ml
penicillin, 50 μg/ml streptomycin, and 2 mM l-glutamine
7. Differentiation medium: growth medium supplemented with 50 μg/ml ascorbic
acid (Sigma Chemical Co., St. Louis, MO) and 10 mM -glycerophosphate (Sigma
Chemical Co.)
8. Phosphate-buffered saline (PBS)
9. Trypsin/EDTA (0.25% (w/v) trypsin/0.53 mM EDTA solution in Hank’s BSS
without calcium or magnesium) (ATCC, Manassas, VA)
2.2. Extraction of the ECM Constituents
1. Liberase/blendzyme 1 (0.14 Wünsch units/ml) (Roche Applied Science, Indiana-
polis, IN)
2. Centrifuge
3. Bicinchoninic acid (BCA) protein assay reagent kit (Pierce, Rockford, IL)
234 Xiao et al.
2.3. Enrichment of MVs from the ECM
1. Liberase/blendzyme 1 (0.14 Wünsch units/ml) (Roche Applied Science, Indiana-
polis, IN)
2. Centrifuge
2.4. Isolation of MVs from Medium
1. Ultra-Clear™ centrifuge tubes: 1 × 3.5 in (38 ml) and 5/8×4in(17ml)(Beckman,
Palo Alto, CA)
2. Optima L-90K preparative ultracentrifuge (Beckman Coulter, Inc., Palo Alto, CA)
2.5. Alkaline Phosphatase Assay
1. Mild lysis buffer: 250 mM NaCl, 50 mM HEPES, pH 7.5, 0.1% NP-40
2. ALP assay kit, including alkaline buffer (1.5 mM 2-amino-2-methyl-1-propanol,
pH 10.3), p-nitrophenyl phosphate (PNPP) (4 mg/ml) and p-nitrophenol (PNP)
standard solution (10 μmol/ml) (Sigma, St. Louis, MO)
3. Flat bottom 96-well plate
4. Lumimark microplate reader (Bio-Rad, Hercules, CA)
2.6. Strong Cation Exchange Liquid Chromatography of Peptides
1. Trypsin Gold, mass spectrometry grade (Promega, Madison, WI)
2. 25% (v/v) acetonitrile containing 0.1% (v/v) formic acid
3. SCX-LC column (1 mm × 150 mm, polysulfoethyl A) (PolyLC, Columbia, MD)
Fig. 2. Transmission electron microscopic image of matrix vesicles in the ultracen-
trifuge pellets (A). The high magnification image (B) shows fine-needle deposits and
black dots, likely signs of calcification, both inside and around the vesicles. Also note
the bilayer membrane of the vesicles (arrowhead).
Analysis of ECM and Secreted Vesicle Proteomes 235
4. Mobile phase A: 25% (v/v) acetonitrile
5. Mobile phase B: 25% (v/v) acetonitrile containing 0.5 M ammonium formate, pH 3
6. 0.1% (v/v) formic acid
7. Vacuum centrifuge
8. Laser-induced fluorescence (LIF) detector
2.7. Nanoflow Reversed-phase Liquid Chromatography Tandem Mass
Spectrometry
1. Slurry packer model 1666 (Alltech, Columbia, MD)
2. Ceramic cutter
3. 75 μm i.d. × 360 μm o.d. × 12 cm long fused silica capillary column (Polymicro
Technologies, Phoenix, AZ)
4. 5 μm, 300 Å pore size C-18 silica-bonded stationary RP particles (Jupiter,
Phenomenex, Torrance, CA)
5. Agilent 1100 nanoLC system (Agilent Technologies, Palo Alto, CA) coupled with
a linear ion-trap (LIT) mass spectrometer (LTQ, ThermoElectron, San Jose, CA)
6. Glass sample injection vials 12 × 32 mm (Wheaton, Millville, NJ)
7. Mobile phase A: 0.1% (v/v) formic acid
8. Mobile phase B: 0.1% formic acid (v/v) in acetonitrile
2.8. Bioinformatic Analysis
1. 20-node Beowulf cluster computer server
2. SEQUEST Cluster version 3.1 SR1 (Thermo Electron Corp., Waltham, MA)
3. Bioworks Browser software 3.2 (Thermo Electron Corp.)
2.9. Validation by Immunofluorescence Staining
1. Primary antibodies: anti-annexin V, anti-emilin-1, anti-IQGAP1 (Santa Cruz
Biotechnology, Inc., Santa Cruz, CA)
2. Secondary antibodies: goat anti-rabbit IgG-FITC, and donkey anti-goat IgG-TR
(Santa Cruz Biotechnology)
3. PBS solution
4. 18 × 18 × 0.15 mm thick glass cover slips
5. Regular microscope glass slides
6. Blocking serum: 10% normal blocking serum in PBS. The blocking serum is
derived from the same species in which the secondary antibody is raised. For
example, if the secondary antibody is raised in goat, use the normal goat serum
diluted to 10% in PBS as the blocking serum.
7. Fixative solution: 3.7% (v/v) formaldehyde in PBS
8. DAPI diluted 1:50,000 in PBS (Invitrogen, Carlsbad, CA)
9. ProLong mounting reagent (Invitrogen)
10. Confocal fluorescence microscope LSM 510 Meta NLO (Carl Zeiss,
Oberkochen, Germany)
236 Xiao et al.
3. Methods
The ECM proteins are extracted from cultured cells by a short exposure
to an ECM-degrading enzyme. To isolate MVs that are either confined to
the ECM or reside in the cell culture medium, two approaches may be used:
(1) For MVs confined to the ECM, an ECM-degrading enzyme is first applied
followed by centrifugation and ultracentrifugation; (2) for MVs in the medium,
centrifugation and ultracentrifugation are applied. The characterization of ECM
and MV proteomes is performed using LC fractionation and MS analysis.
3.1. Cell Culture
1. Grow the murine calvaria-derived osteoblast MC3T3-E1 cells in growth medium.
The medium is changed every two or three days. Passage the cells with
trypsin/EDTA (see Note 1).
2. Once the cell culture reaches ∼50% confluency, replace the growth medium with
10 ml of differentiation medium per plate to induce osteoblast differentiation.
3. Extract the ECM or harvest culture medium on the day indicated in the methods
below.
3.2. Extraction of the ECM Constituents
1. Grow MC3T3-E1 cells in differentiation medium on 10-cm plates. Change the
medium every two or three days (see Note 2).
2. On day 21, aspirate the medium from the plates. Wash the cells with 10 ml of
PBS solution three times.
3. Add 3 ml of liberase/blendzyme 1 solution to each plate. Incubate at 37°C for
30 min.
4. Carefully collect the digested supernatant from the plates without disturbing the
cells.
5. Centrifuge the supernatant at 2000×gfor 5 min to remove any free cells. The
resulting supernatant contains ECM proteins.
6. Quantify the amount of ECM proteins using the BCA assay (see Note 3).
3.3. Enrichment of MVs from the ECM
1. Follow the same procedure described earlier to grow and prepare cells (see
Subheading 3.2,steps 1 and 2, and Note 2).
2. On day 21, aspirate the medium and wash the cells three times with PBS.
3. Add 3 ml of liberase/blendzyme 1 solution to each plate. Incubate at 37°C for
30 min (see Note 4).
4. Collect the supernatant from the plates without disturbing the cells. Centrifuge
the supernatant at 2000×gfor 5 min to remove any cells that may have been
detached from the plate. Collect the supernatant.
5. Centrifuge the supernatant at 20,000×gat 4°C for 30 min.
Analysis of ECM and Secreted Vesicle Proteomes 237
6. Transfer the supernatant to the Ultra-Clear™ centrifuge tubes. Use the centrifuge
tubes that fit the volume of the supernatant. Fill the tubes with PBS up to about
2 –3 mm from the top.
7. Subject the supernatant to ultracentrifugation at 100,000×gat 4°C for 60 min.
Carefully remove the supernatant without disturbing the pellet.
8. The pellets are enriched with MVs designed as collagenase-released MVs
(CRMVs) (see Note 5).
9. Confirm the enrichment of CRMVs by assaying the ALP activity using an
aliquot of the pellet (see Note 6 and Subheading 3.5).
10. Resuspend the rest of the pellet in 25 mM NH4HCO3, pH 8.4. Quantify the
amount of CRMV proteins in the pellet by BCA assay (see Note 3).
3.4. Isolation of MVs from Medium
1. Grow MC3T3-E1 cells in differentiation medium in four 10-cm plates.
2. On day 15, collect the media from multiple plates (see Note 2).
3. Separate cellular debris from the medium by centrifugation at 20,000×gfor 30 min
at 4°C.
4. Transfer the supernatant to Ultra-Clear™ centrifuge tubes. Use the centrifuge
tubes that fit the volume of the supernatant.
5. Further centrifuge the supernatant by ultracentrifugation at 100,000×gfor 60 min.
6. Carefully remove the supernatant. The MVs in the pellet are designated as medium
MVs (MMVs) (see Note 5 and Fig. 1).
7. Resuspend an aliquot of the MMV sample in 25 mM NH4HCO3, pH 8.4.
Determine the protein concentration in the pellet by BCA assay.
3.5. Alkaline Phosphatase Assay
1. For the standard curve: Dilute PNP standard 1:10 in dH2O. Add 0, 2, 4, 6, 8, 10,
20, 30, 40, and 50 μl of the standard (i.e., 0, 2, 2, 4, 6, 8, 10, 20, 30, 40, and
50 nmol, respectively) to the wells of a flat-bottom 96-well microtiter plate. Add
mild lysis buffer to make a total volume of 135 μl.
2. For the CRMV and MMV samples: Resuspend an aliquot of the ultracentrifuged
pellet in mild lysis buffer. Quantify the protein by BCA assay. Based on the BCA
assay results, add 25 μg of protein to the 96-well microtiter plate. Add mild lysis
buffer further to make a total volume of135 μl/well.
3. Add 25 μl of alkaline buffer and 25 μl of p-nitrophenyl phosphate (PNPP) to each
well.
4. Incubate the microtiter plate at 37°C for up to 3 h. Monitor the colorimetric
change every hour by measuring absorbance at 405 nm using the microtiter plate
reader. Stop incubation when the absorbance of the sample reaches the range of
the standards.
5. Determine the ALP activity in MV samples by comparing to the PNP standard
curve. Report the ALP activity as nmol PNP produced per minute per milligram
of protein used (see Note 6).
238 Xiao et al.
3.6. Strong Cation Exchange Liquid Chromatography of Peptides
1. Digest 100 μg of ECM, CRMV, or MMV proteins in 25 mM NH4HCO3, pH 8.4,
with trypsin using a trypsin-to-protein ratio of 1:40. For 100 μg of protein, add
2.5 μg of trypsin. Incubate the digestion at 37°C overnight (see Note 7).
2. Lyophilize the peptide digests in a vacuum centrifuge.
3. Dissolve peptide digests in 100 μl of 25% (v/v) acetonitrile containing 0.1% (v/v)
formic acid.
4. Inject the peptides onto a SCX-LC column (1 ×150 mm, polysulfoethyl A).
5. Maintain the flow rate of the column at 50 μl/min. Mobile phase A is 25% (v/v)
acetonitrile, and mobile phase B is 25% (v/v) acetonitrile with 0.5 M ammonium
formate (pH 3).
6. Elute the peptides using the following 96-min gradient method: 3% B for 3 min,
followed by a linear increase to 10% B in 43 min, a further increase to 45% B
in 40 min, and then to 100% B in 10 min. Monitor the peptide separation by
fluorescence (266 nm excitation/350 nm emission). Collect fractions every minute
for 96 min (see Note 8).
7. Based on the chromatogram, pool the adjacent fractions into a total of 20 fractions
and lyophilize (see Notes 9 and 10).
8. Resuspend each pooled fraction in 20 μl of 0.1% (v/v) formic acid prior to
nanoRPLC-MS analysis.
3.7. Nanoflow Reversed-Phase Liquid Chromatography Tandem Mass
Spectrometry
1. Cut a 12-cm piece of 75 μm i.d. ×360 μm o.d. fused silica capillary column. Use
a torch to briefly flame the section about 2 cm near one end. Once the flamed
section is soft, pull the column to make a 10-cm long section with a closed tip.
To make a fine and flat opening at the end of the tip, lightly score near the end
of the closed tip using a ceramic cutter, and then break the end away.
2. Connect the column to the slurry packer. Pack the column with 5 μm, 300 Å pore
size C-18 silica-bonded stationary reversed-phase particles.
3. Connect the column to an Agilent 1100 nanoLC system coupled with a LIT mass
spectrometer (LTQ, ThermoElectron, operated with Xcalibur 1.4 SR1 software).
4. Transfer the peptide fractions into glass vials. Inject 6 μl of the solution.
5. Mobile phase A is 0.1% (v/v) formic acid and B is 0.1% (v/v) formic acid in
acetonitrile. Elute the peptides using the following gradient method: 2% B at
500 nl/min in 30 min; a linear increase of 2–42% B at 250 nl/min in 110 min;
42–98% in 30 min including the first 15 min at 250 nl/min and then 15 min at
500 nl/min; 98% at 500 nl/min for 10 min.
6. Set the capillary temperature and electrospray voltage at 160°C and 1.5 kV,
respectively. The LIT-MS is operated in a data-dependent MS/MS mode where
the five most abundant peptide molecular ions in every MS scan are sequentially
selected for collision-induced dissociation (CID) using a normalized collision
Analysis of ECM and Secreted Vesicle Proteomes 239
energy of 35%. Apply dynamic exclusion to minimize repeated selection of
peptides previously selected for CID (see Notes 11 and 12).
3.8. Bioinformatic Analysis
1. Search the tandem mass spectra against the UniProt proteomic database from
the European Bioinformatics Institute (http://www.ebi.ac.uk/) with SEQUEST
operating on a 40-node Beowulf cluster (SEQUEST Cluster version 3.1 SR1,
Bioworks Browser 3.2). Limit the search to peptides generated with fully tryptic
cleavage constraints.
2. Set legitimate peptide identification criteria as follows: charge state and cross-
correlation (Xcorr) scores of 1.9 for [M + H]1+, 2.2 for [M + 2H]2+, 3.1 for
[M + 3H]3+, and a minimum delta correlation (Cn) of 0.08.
3. Base protein identification exclusively on unique peptide hits, i.e., peptides whose
sequence is unique to a given protein (see Notes 13 and 14).
3.9. Immunofluorescence Staining
1. Plate 50,000 cells on glass cover slips in 6-well plates. Culture in differentiation
medium.
2. On day 15, briefly wash the cells with PBS.
3. Fix the cells in 3.7% (v/v) formaldehyde in PBS for 10 min.
4. Incubate with 10% (v/v) normal blocking serum in PBS.
5. Briefly wash the cells with PBS; incubate with primary antibodies for 1.5 h.
6. Wash the cells three times with PBS for 5 min each, and then incubate with
secondary antibodies conjugated with fluorochrome (FITC or Texas Red) for 1 h.
7. Wash the cells three times with PBS for 5 min each, including once with DAPI
diluted 1:50,000 in PBS to stain nuclei.
8. Mount the cover slips on microscope glass slides with ProLong mounting reagent.
9. Observe the cells using a confocal fluorescence microscope (see Note 14).
4. Notes
1. MC3T3-E1 pre-osteoblast cells are derived from newborn murine calvaria (28).
These cells closely resemble primary cell cultures in their proliferation, differ-
entiation, and mineralization (29,30,31). The combination of ascorbic acid and
-glycerophosphate stimulates MC3T3-E1 to undergo differentiation, which is
characterized by substantial matrix mineralization (32,33). Therefore, it is a
suitable model for the enrichment of ECM and isolation of MVs.
2. It is necessary to culture multiple 10-cm plates (four or more at approximately
4×106cells /plate) in order to obtain sufficient amount of protein from ECM
or MVs.
3. Protein quantitation is a common laboratory procedure. The instructions are
included within the BCA assay kit (Pierce); therefore, the procedure is not
described in this chapter.
240 Xiao et al.
4. The liberase/blendzyme 1 is a mixture of highly purified collagenase and
dispase that offers gentle protease activity as compared to other ECM-degrading
enzymes. Note that four blendzyme mixtures with increasing levels of enzymatic
strength are available from Roche. Blendzyme 1 is the mildest version. The
digestion time varies depending on the cell or tissue type. Alternatively, colla-
genase/dispase (1 mg/ml of collagenase/dispase in PBS-containing collagenase,
0.1 U/ml and dispase, 0.8 U/ml) (Sigma Chemical Co., St. Louis, MO) can
be used. Collagenase/dispase enzyme mixture is commonly used to digest the
ECM.
5. Two approaches are designed to isolate MVs either from the ECM or directly
from the cell culture medium. In the first approach, enzymatic digestion and
ultracentrifugation are combined to release MVs embedded in the ECM (desig-
nated as CRMVs). In the second approach, ultracentrifugation is applied to the
medium to isolate MVs, designated as MMVs (34). To confirm the enrichment of
MVs, the ultracentrifugation pellets are fixed and examined using transmission
electron microscopy (Fig. 2).
6. Measurement of the enzymatic activity of ALP is a standard marker for MV
isolation (35,36).
7. Instead of using the buffer provided along with trypsin, it is desirable to
resuspend trypsin in 25 mM NH4HCO3, pH 8.4. The trypsin-to-protein ratio
should be between 1:40 and 1:50. The digestion mixture is incubated overnight
(approximately 16 h).
8. The LIF detector used in this method can be constructed in-house (37). The
LIF detector is more sensitive than a conventional lamp-based fluorescence
detector. The use of a LIF detector is particularly advantageous when a narrow
bore column (<1 mm i.d.) or a micro column (<300 μm i.d.) is used. Some
conventional fluorescence detectors can be used with the narrow bore or micro
column; however, the sensitivity is lower. When the peptide content is low and a
narrow diameter column is being used, the LIF detector offers better sensitivity.
For a peptide to be detectable using fluorescence detection, it must contain
at least one aromatic residue, particularly tryptophan. Although tryptophan-
containing proteins are comparatively rare, the complexity of the peptide mixture
compensates to provide a good estimate of the separation. An alternative to LIF
detection is UV. The advantage of an UV detector is that it detects amide bonds,
which are universally present in peptides. The main disadvantage, however, is its
limited compatibility with biological buffers. Volatile salts, such as ammonium
formate, used in this method are incompatible with UV detection since formate
absorbs strongly at 214 nm, which is the wavelength used for peptide detection.
Sodium chloride is compatible with UV detection, but it is non-volatile. In that
case, desalting of the peptide fractions is needed for the down stream. This
desalting step may lead to sample loss.
9. All the automated two-dimensional (online) LC systems use chloride as the salt
in the SCX first dimension, and a desalting step has to be implemented in the
program. However, we found that the offline multi-dimensional separation of
Analysis of ECM and Secreted Vesicle Proteomes 241
peptide is capable of identifying more proteins than the online procedure. Thus,
the offline separation is described in this chapter.
10. The pooling step is optional. The peptide fractions can be pooled based on the
complexity of the chromatogram. In general, pooling to about 20 fractions is
appropriate. It will save LC-MS running time without compromising the number
of proteins that the approach can identify.
11. In general, the MS data acquisition time is set to 150 min, starting 30 min after
the beginning of the peptide elution gradient and synchronized to end with the
elution gradient.
12. An alternative approach: the resulting ECM, CRMV, or MMV protein samples
can be resolved by SDS-PAGE and the proteins visualized by Coomassie
staining. The protein bands that are of greater intensity than those prepared
from undifferentiated cells can be excised and subjected to in-gel digestion with
trypsin and analyzed using nanoRPLC-MS/MS (27).
13. Proteins that are identified in both CRMV and MMV purifications can be
considered as authentic MV proteins with a higher degree of confidence than
those that were identified in only one of the preparations.
14. Gene ontology (GO) (www.geneontology.org) can be used to annotate the
identified proteins and categorize them according to their cellular location,
molecular function, and cellular processes they are associated with.
15. The validation of known MV proteins is conducted using Western blotting
or immunofluorescence staining. Annexin V, a known constituent of MVs, is
used as a protein landmark to locate vesicles in these experiments (38). The
osteoblast cells can be double- stained with anti-annexin V and an additional
antibody against either the extracellular protein emilin-1 or the ras GTPase,
IQGAP1 (27).
Acknowledgments
This project has been funded in whole or in part with Federal funds from
the National Cancer Institute, National Institutes of Health, under Contract No.
N01-CO-12400. The content of this publication does not necessarily reflect
the views or policies of the Department of Health and Human Services, nor
does the mention of trade names, commercial products, or organization imply
endorsement by the US Government.
References
1. Holmbeck, K. and Szabova, L. (2006) Aspects of extracellular matrix remodeling
in development and disease. Birth Defects Res C Embryo Today 78, 11–23.
2. Brooke, B. S., Karnik, S. K. and Li, D. Y. (2003) Extracellular matrix in vascular
morphogenesis and disease: structure versus signal. Trends Cell Biol 13, 51–56.
3. Tahinci, E. and Lee, E. (2004) The interface between cell and developmental
biology. Curr Opin Genet Dev 14, 361–366.
242 Xiao et al.
4. Harada, S. and Rodan, G. A. (2003) Control of osteoblast function and regulation
of bone mass. Nature 423, 349–355.
5. Beck, G. R., Jr. (2003) Inorganic phosphate as a signaling molecule in osteoblast
differentiation. J Cell Biochem 90, 234–243.
6. Aubin, J. E. (2001) Regulation of osteoblast formation and function. Rev Endocr
Metab Disord 2, 81–94.
7. Anderson, H. C. (1995) Molecular biology of matrix vesicles. Clin Orthop Relat
Res, 266–280.
8. Anderson, H. C. (2003) Matrix vesicles and calcification. Curr Rheumatol Rep 5,
222–226.
9. Anderson, H. C., Garimella, R. and Tague, S. E. (2005) The role of matrix vesicles
in growth plate development and biomineralization. Front Biosci 10, 822–837.
10. Kirsch, T. (2005) Annexins – their role in cartilage mineralization. Front Biosci
10, 576–581.
11. Hessle, L., Johnson, K. A., Anderson, H. C., Narisawa, S., Sali, A., Goding, J. W.,
Terkeltaub, R. and Millan, J. L. (2002) Tissue-nonspecific alkaline phosphatase
and plasma cell membrane glycoprotein-1 are central antagonistic regulators of
bone mineralization. Proc Natl Acad Sci USA 99, 9445–9449.
12. Johnson, K. A., Hessle, L., Vaingankar, S., Wennberg, C., Mauro, S., Narisawa, S.,
Goding, J. W., Sano, K., Millan, J. L. and Terkeltaub, R. (2000) Osteoblast tissue-
nonspecific alkaline phosphatase antagonizes and regulates PC-1. Am J Physiol
Regul Integr Comp Physiol 279, R1365–1377.
13. Morris, D. C., Masuhara, K., Takaoka, K., Ono, K. and Anderson, H. C. (1992)
Immunolocalization of alkaline phosphatase in osteoblasts and matrix vesicles of
human fetal bone. Bone Miner 19, 287–298.
14. Baldini, V., Mastropasqua, M., Francucci, C. M. and D’Erasmo, E. (2005) Cardio-
vascular disease and osteoporosis. J Endocrinol Invest 28, 69–72.
15. Dao, H. H., Essalihi, R., Bouvet, C. and Moreau, P. (2005) Evolution and
modulation of age-related medial elastocalcinosis: impact on large artery stiffness
and isolated systolic hypertension. Cardiovasc Res 66, 307–317.
16. Reynolds, J. L., Joannides, A. J., Skepper, J. N., McNair, R., Schurgers, L. J.,
Proudfoot, D., Jahnen-Dechent, W., Weissberg, P. L. and Shanahan, C. M. (2004)
Human vascular smooth muscle cells undergo vesicle-mediated calcification in
response to changes in extracellular calcium and phosphate concentrations: a
potential mechanism for accelerated vascular calcification in ESRD. J Am Soc
Nephrol 15, 2857–2867.
17. Abedin, M., Tintut, Y. and Demer, L. L. (2004) Vascular calcification: mechanisms
and clinical ramifications. Arterioscler Thromb Vasc Biol 24, 1161–1170.
18. Tintut, Y. and Demer, L. L. (2001) Recent advances in multifactorial regulation
of vascular calcification. Curr Opin Lipidol 12, 555–560.
19. Stewart, D. A., Cooper, C. R. and Sikes, R. A. (2004) Changes in extracel-
lular matrix (ECM) and ECM-associated proteins in the metastatic progression of
prostate cancer. Reprod Biol Endocrinol 2, 2.
Analysis of ECM and Secreted Vesicle Proteomes 243
20. Yin, J. J., Pollock, C. B. and Kelly, K. (2005) Mechanisms of cancer metastasis to
the bone. Cell Res 15, 57–62.
21. Mundy, G. R. (2002) Metastasis to bone: causes, consequences and therapeutic
opportunities. Nat Rev Cancer 2, 584–593.
22. Roodman, G. D. (2004) Mechanisms of bone metastasis. N Engl J Med 350,
1655–1664.
23. Yates, J. R., III. (2004) Mass spectral analysis in proteomics. Annu Rev Biophys
Biomol Struct 33, 297–316.
24. Yates, J. R., III, Gilchrist, A., Howell, K. E. and Bergeron, J. J. (2005) Proteomics
of organelles and large cellular structures. Nat Rev Mol Cell Biol 6, 702–714.
25. Domon, B. and Aebersold, R. (2006) Mass spectrometry and protein analysis.
Science 312, 212–217.
26. Aebersold, R. and Mann, M. (2003) Mass spectrometry-based proteomics. Nature
422, 198–207.
27. Xiao, Z., Camalier, C. E., Nagashima, K., Chan, K. C., Lucas, D. A., de la
Cruz, M. J., Gignac, M., Lockett, S., Issaq, H. J., Veenstra, T. D., Conrads, T. P.
and Beck Jr, G. R. (2006) Analysis of the extracellular matrix vesicle proteome in
mineralizing osteoblasts. J Cell Physiol, In press.
28. Sudo, H., Kodama, H. A., Amagai, Y., Yamamoto, S. and Kasai, S. (1983) In vitro
differentiation and calcification in a new clonal osteogenic cell line derived from
newborn mouse calvaria. J Cell Biol 96, 191–198.
29. Choi, J. Y., Lee, B. H., Song, K. B., Park, R. W., Kim, I. S., Sohn, K. Y.,
Jo, J. S. and Ryoo, H. M. (1996) Expression patterns of bone-related proteins during
osteoblastic differentiation in MC3T3-E1 cells. J Cell Biochem 61, 609–618.
30. Quarles, L. D., Yohay, D. A., Lever, L. W., Caton, R. and Wenstrup, R. J.
(1992) Distinct proliferative and differentiated stages of murine MC3T3-E1 cells
in culture: an in vitro model of osteoblast development. J Bone Miner Res 7,
683–692.
31. Franceschi, R. T., Iyer, B. S. and Cui, Y. (1994) Effects of ascorbic acid on collagen
matrix formation and osteoblast differentiation in murine MC3T3-E1 cells. J Bone
Miner Res 9, 843–854.
32. Beck, G. R., Jr, Sullivan, E. C., Moran, E. and Zerler, B. (1998) Relationship
between alkaline phosphatase levels, osteopontin expression, and mineralization in
differentiating MC3T3-E1 osteoblasts. J Cell Biochem 68, 269–280.
33. Beck, G. R., Jr, Zerler, B. and Moran, E. (2001) Gene array analysis of osteoblast
differentiation. Cell Growth Differ 12, 61–83.
34. Johnson, K., Moffa, A., Chen, Y., Pritzker, K., Goding, J. and Terkeltaub, R. (1999)
Matrix vesicle plasma cell membrane glycoprotein-1 regulates mineralization by
murine osteoblastic MC3T3 cells. J Bone Miner Res 14, 883–892.
35. Ali, S. Y., Sajdera, S. W. and Anderson, H. C. (1970) Isolation and characterization
of calcifying matrix vesicles from epiphyseal cartilage. Proc Natl Acad Sci USA
67, 1513–1520.
36. Dean, D. D., Schwartz, Z., Bonewald, L., Muniz, O. E., Morales, S., Gomez, R.,
Brooks, B. P., Qiao, M., Howell, D. S. and Boyan, B. D. (1994) Matrix vesicles
244 Xiao et al.
produced by osteoblast-like cells in culture become significantly enriched in
proteoglycan-degrading metalloproteinases after addition of beta-glycerophosphate
and ascorbic acid. Calcif Tissue Int 54, 399–408.
37. Chan, K. C., Muschik, G. M. and Issaq, H. J. (2000) Solid-state UV laser-induced
fluorescence detection in capillary electrophoresis. Electrophoresis 21, 2062–2066.
38. Wang, W., Xu, J. and Kirsch, T. (2005) Annexin V and terminal differentiation of
growth plate chondrocytes. Exp Cell Res 305, 156–165.
IV
Clinical Proteomics and Antibody Arrays
14
Miniaturized Parallelized Sandwich Immunoassays
Hsin-Yun Hsu, Silke Wittemann, and Thomas O. Joos
Summary
This chapter describes the development and use of bead-based miniaturized multi-
plexed sandwich immunoassays for focused protein profiling. Bead-based protein arrays
or suspension microarrays allow simultaneous analysis of a variety of parameters within
a single experiment. In suspension microarrays capture antibodies are coupled onto color-
coded microspheres.
The applications of suspension microarrays are described, which allow to analyze
proteins present in different types of body fluids, such as serum or plasma, cerebrospinal,
pleural and synovial fluids, as well as cell culture supernatants. The chapter is divided into
the generation of suspension microarrays, sample preparation, processing of suspension
microarrays, validation of analytical performance, and finally pattern generation using
bioinformatics tools.
Key Words: suspension microarray; microspheres; immunoassay; protein profiling;
biological fluids; serum; pleura; cell culture supernatants; cerebrospinal fluid; synovial
fluid.
1. Introduction
Protein microarray technology allows simultaneous determination of a large
variety of analytes from a minute amount of sample within a single experiment.
Assay systems based on this technology are currently applied for identification
and quantitation of proteins. Protein microarray technology is of major interest
for proteomic research in basic and applied biology as well as for diagnostic
applications. Miniaturized and parallelized assay systems have reached adequate
sensitivity, and hence have the potential to replace singleplex analysis systems.
From: Methods in Molecular Biology, vol. 428: Clinical Proteomics: Methods and Protocols
Edited by: A. Vlahou © Humana Press, Totowa, NJ
247
248 Hsu et al.
Beside the well-known planar microarray-based systems, which are perfectly
suited to screen a large number of target proteins, bead-based systems named
suspension assays are a very interesting alternative, especially when the number
of parameters of interest is comparably low. Suspension assay systems employ
different color-coded or size-coded microspheres as the solid support for capture
molecules. A flow cytometer, which is able to identify each individual type of
bead and quantify the amount of captured targets on each individual bead, is
used as a readout system. In the first step, antigen-specific capture antibodies
are immobilized on the individual bead type. Different bead types are combined
and incubated with the sample of interest. A labeled secondary antibody
detects the captured analytes and is visualized with a fluorescent reporter
system. Sensitivity, reliability, and accuracy are similar to those observed with
standard microtiter ELISA procedures (1). Color-coded microspheres can be
used to perform up to a hundred different assay types simultaneously. The flow
cytometer identifies several thousand microspheres in a second, and simul-
taneously quantitates the amount of captured analytes (2,3,4,5,6). Suspension
microarrays are currently advanced within the field of miniaturized multiplexed
ligand binding assays with respect to automation and throughput (7).
Miniaturized parallelized assay systems have to demonstrate appropriate
sensitivity, precision, and reliability before they will be applied for screening
or diagnostic purposes.
This chapter describes the development and use of suspension antibody
microarrays for protein profiling of several human body fluids. The standard
methodology guidance is described to validate immunoassays (10,11,12) and to
determine the sensitivity, precision, and accuracy of the multiplexed analysis.
In the final section, data analysis is described to show how to deal with high-
dimension data sets (13,14).
2. Materials
2.1. Equipment
1. Centrifuge: 5415D (Eppendorf)
2. Vortex Mixer (Neolab)
3. Ultrasonic bath
4. Thermomixer (Eppendorf)
5. Luminex100 instrument (Luminex Corp.)
6. Vacuum manifold (Millipore)
7. Filterplates (Millipore 96-well plate, cat. # MAB1250)
8. Microcentrifuge tubes (Starlab 1.5 ml, cat. # I1415-2500)
9. Carboxylated Beads (Qiagen, cat. # 922400 or Luminex Corp.)
10. Deionized water
Miniaturized Parallelized Sandwich Immunoassays 249
2.2. Common Reagents and Materials
1. Bovine serum albumin (BSA, Roth T844.2)
2. PBS (Fischer Scientific, cat. # 9472615)
3. EDC (Pierce)
4. Sulfo-NHS (Pierce)
5. Detection reagent: Streptavidin-phycoerythrin (Streptavidin-PE) stock solution
(1 mg/ml) in 100 mM NaCl, 100 mM sodium phosphate, pH 7.5, containing
2 mM sodium azide (Molecular Probes, cat. #S21388)
2.3. Buffers
1. Activation buffer [100 mM sodium phosphate (Na2HPO4), pH 6.2]
2. Coupling buffer (50 mM MES, pH 5.0)
3. Washing buffer [PBS, pH 7.4, and 0.05 % (v/v) Tween-20]
4. Blocking/storage (B/S) buffer: 1% BSA fraction IV (Roth, cat. # T844.2) in 1×
PBS
5. Assay buffer formulation: 1% BSA fraction IV in 1×PBS
3. Methods
3.1. Principle
The principle of suspension antibody microarrays is based on sandwich
immunoassays as represented in Fig. 1. First-capture antibodies are coupled to
carboxylated microspheres. For performing suspension antibody microarrays,
the samples are incubated with coupled microspheres. Bound analytes are
detected with biotinylated antibodies. Phycoerythrin-labeled streptavidin is used
for signal detection. Finally, microspheres are identified by a flow cytometer,
hence allowing the quantitation of the captured analytes.
3.2. Production of Suspension Microarrays—Antibody Coupling to
Carboxylated Microspheres (see Note 1)
Using proven carbodiimide coupling chemistry, the antibodies are covalently
immobilized on carboxylated beads via the amine groups in lysine side chains.
Before coupling, the beads are first activated using EDC/Sulfo-NHS.
Fig. 1. Processing of suspension microarrays. Schematic representation of the steps
required for performing a suspension microarray immunoassay. Figure reproduced from
Proteomics of Human Body Fluids: Principles,Methods and Applications, edited by
Thongboonkerd (2006). (Continued)
250 Hsu et al.
Miniaturized Parallelized Sandwich Immunoassays 251
The antibodies should not contain foreign protein, azide, glycine, Tris, or
any other reagent containing primary amine groups. Otherwise, the antibodies
must be purified by gel-filtration chromatography or dialysis before use.
3.2.1. Bead Activation
1. Sonicate the carboxylated bead stock suspension for 15–20 s to yield a homoge-
neous bead suspension. Thoroughly vortex the bead stock suspension for at least
10 s. Take 2.5 × 106beads per coupling reaction.
2. Transfer the bead stock suspension to Starlab microcentrifuge tube.
3. Briefly centrifuge the bead suspension (a quick spin up to 3000×gis sufficient)
and discard the supernatant.
4. Wash the beads with 80 μl activation buffer. Briefly vortex and centrifuge at
10,000×gfor 2 min. Discard the supernatant and repeat washing.
5. Resuspend the beads in 80 μl activation buffer. Sonicate for 15–20 s to yield a
homogeneous bead suspension.
6. Freshly prepare EDC solution (50 mg/ml) and Sulfo-NHS solution (50 mg/ml)
(see Notes 2 and 3).
7. Add 10 μl of EDC solution and 10 μl of Sulfo-NHS solution to the bead sus-
pension. Incubate for 20 min at room temperature (15–25°C) in the dark.
3.2.2. Coupling of Antibodies to Activated
Carboxylated Beads
8. Dilute the protein stock solution with coupling buffer to a concentration of
100 μg/ml in a volume of 500 μl.
9. Centrifuge the beads at 10,000×gfor 2 min and discard the supernatant.
10. Wash the beads with 500 μl of coupling buffer. Briefly vortex and centrifuge at
10,000×gfor 2 min. Discard the supernatant and repeat washing.
11. Add the diluted antibody solution (500 μl) from step 8.
12. Wrap the tube in aluminum foil to exclude light. Gently agitate the tube with
activated beads and antibody solution on a plate shaker for2hatroom temper-
ature (15–25°C).
3.2.3. Washing and Storage of Coupled
Carboxylated Beads
13. Centrifuge the beads at 10,000×gfor 2 min and carefully remove and discard
the supernatant.
14. Wash the beads with 500 μl of washing buffer. Briefly vortex and centrifuge at
10,000×gfor 2 min. Discard the supernatant and repeat washing.
15. Resuspend the bead pellet in 1 ml B/S buffer including 0.05% (w/v) azide.
16. Determine the bead concentration of the suspension using a cell-counting
chamber.
252 Hsu et al.
3.2.4. Counting Beads Using a Cell-Counting Chamber
1. Add 5 μl of beads to 45 μl of PBS and mix.
2. The hemacytometer is filled with 10 μl of the sample by placing the pipette tip
against the loading “V” of the hemacytometer at a 45° angle. The sample is
slowly released between the slide and the cover slip until the counting chamber
is loaded. It is important to fill both sides of the chamber and wait for 2–3 min
to allow the beads to settle.
3. Count the cells at two opposite corners of the scored chamber and take an average.
Each of the nine squares on the grid has an area of 1 mm2, and the coverglass
rests 0.1 mm above the floor of the chamber. Thus, the volume over the central
counting area is 0.1 mm3or 0.1 ml. Multiply the average number of beads in
each central counting area by 10,000 to obtain the number of beads per milliliter
of diluted sample. Multiply by the dilution factor of 10 to get beads/ml.
4. Store the beads at 25×, typically5×10
6beads/ml.
3.3. Processing of Bead-Based Multiplex Assays
3.3.1. Sample Preparation
Here, the preparation of proteins for use in multiplexed assay from clinical
specimens or cell culture is described. Subheading 3.3.1.1 describes the use
of serum or plasma; Subheading 3.3.1.2 describes the analysis of proteins
present in cell culture supernatants; Subheading 3.3.1.3 describes the sample
preparation of cerebrospinal, synovial, and pleural fluids.
3.3.1.1. Serum or Plasma Samples
Serum and plasma samples should be spun down (8000×g) prior to assay
to remove particulate and lipid layers. This will prevent the blocking of wash
plate as well as sample needle. The samples should be handled as biohazards
since they may carry infectious agents. Freezing-thawing cycles might result in
a measurable breakdown of some proteins (e.g., cytokines), and so the samples
should be aliquoted before any experiment. The storage of aliquoted samples at
–80°C is recommended. When we analyzed eight matched serum and plasma
samples on the Luminex platform, no differences were seen between samples
that underwent a freeze-thaw for levels of TNF, Eotaxin, IL-13, MCP-1, IFN,
IL-12p70, MIP-1, IP-10, or GM-CSF. There was, however, a significant
increase in IL-1after freeze-thaw, suggesting that this process may liberate
IL-1from insoluble receptors. IL-1and MCP-1 levels