Methods in

Molecular Biology 1871

Xing Wang
Matthew Kuruc
Editors

Functional
Proteomics
Methods and Protocols
METHODS IN MOLECULAR BIOLOGY

Series Editor
John M. Walker
School of Life and Medical Sciences
University of Hertfordshire
Hatfield, Hertfordshire, AL10 9AB, UK

For further volumes:

http://www.springer.com/series/7651
Functional Proteomics

Methods and Protocols

Edited by

Xing Wang
Array Bridge Inc., St. Louis, MO, USA

Matthew Kuruc
Biotech Support Group LLC, Monmouth Junction, NJ, USA
Editors
Xing Wang Matthew Kuruc
Array Bridge Inc. Biotech Support Group LLC
St. Louis, MO, USA Monmouth Junction, NJ, USA

ISSN 1064-3745 ISSN 1940-6029 (electronic)

Methods in Molecular Biology
ISBN 978-1-4939-8813-6 ISBN 978-1-4939-8814-3 (eBook)
https://doi.org/10.1007/978-1-4939-8814-3
Library of Congress Control Number: 2018957271

© Springer Science+Business Media, LLC, part of Springer Nature 2019

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction
on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation,
computer software, or by similar or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply,
even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations
and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to
be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty,
express or implied, with respect to the material contained herein or for any errors or omissions that may have been made.
The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

This Humana Press imprint is published by the registered company Springer Science+Business Media, LLC, part of
Springer Nature.
The registered company address is: 233 Spring Street, New York, NY 10013, U.S.A.
Preface

Over the past two decades, tremendous progress has been made in the field of proteomics,
the purpose of which is to find systemic differences in protein populations. Once established,
measurable protein markers can then help to define biological mechanism, disease, identify
therapeutic targets, and offer better precision for personalized medical interventions.
Proteomics, like other “omics” analyses, is data driven and can generate unbiased
protein profiles for a variety of end points that can contrast, for example, treated vs. untreated
cell models or healthy vs. diseased tissue; it has provided us with more in-depth understanding
of many biological systems and diseases. This progress in the field of proteomics parallels with
the advancement in many analytical technologies, especially in mass spectrometry, which has
been evolved from a less sensitive and qualitative tool to high sensitivity and quantitation system
for protein analysis and characterization. Currently, system biology and proteomics in particular
are advancing biology at two opposite but equally important polarities; one is the holistic
understanding of a biological system, be it an organism, organ, tissue, or the human circulation
system, and the other one is for single-cell analysis whereby biological heterogeneity can be
minimized and a more discrete picture of biological processes can be modeled within a more
homogeneous context. Having the tools and methods necessary to address these important
studies will promote the significant impact that is foreseen in precision medicine and other
biological fields.
In the most common view of proteomics, data is generally acquired after proteolytic
processing of the parent proteomes. The derived peptides are then analyzed on instruments
coupling Nano-Liquid Chromatography to Mass Spectrometry (LC-MS). Such instruments
generate mass spectra of peptides and the MS2 spectra through further fragmentation,
which can be compared to theoretical amino acid spectra definable through public gene
repositories. Peptide sequence matches are thus computationally derived, and from that
data, protein identifications are inferred. From such analyses, peptide markers can be used as
surrogates for the gene products from which they are derived. Through differential expres-
sion analysis of these peptide markers, proteomics can thus help identify those gene products
that define a phenotype. However the functions of the proteome, the driving force for
almost all biological actions, are not adequately annotated through the current infrastruc-
ture of methods surrounding LC-MS sequence annotation. This book is intended to fill in
this knowledge and technology gap with a specific collection of technologies that have been
developed for the study of protein function at a proteome scale. In organizing the content of
this book, the following points were taken into consideration: (1) It should bridge the
understanding of biology from protein functions to other aspects of protein analysis,
especially in post-translational modification, as most cellular proteins use this mechanism
to carry out their unique role in cellular regulation. (2) The book should also act as a bridge
to other levels of system biology research including genomics and metabolomics, so that the
readers will gain a relatively complete picture of how one might study the biological system
of their interest. (3) Technologies are categorized toward different aspects of protein
functional analysis, so that readers can understand what is available to them in functional
proteomics research. (4) Finally, the selection of technologies also takes into consideration
the impact on current and future research in a variety of disease areas.

v
vi Preface

It is hoped that by using these novel technologies, new frontiers in biological research
will be created, important drug targets can be identified, and clinically validated biomarkers
and diagnostic tests can be developed. The aim of the editors of this book is to provide the
most precise description of our technological capabilities in functional proteomics research
and give our readers the tools they will need to create the new functional domains of our
knowledge in the understanding of various biological systems.

St. Louis, MO, USA Xing Wang

Monmouth Junction, NJ, USA Matthew Kuruc
Contents

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
Contributors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
1 Making the Case for Functional Proteomics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Ray C. Perkins
2 Methods to Monitor the Functional Subproteomes of SERPIN
Protease Inhibitors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Swapan Roy and Matthew Kuruc
3 Two-Dimensional 16-BAC/SDS Polyacrylamide Gel Electrophoresis
of Mitochondrial Membrane Proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Gary Smejkal and Srikanth Kakumanu
4 Systematic Glycolytic Enzyme Activity Analysis from Human
Serum with PEP Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
David Wang
5 A Protein Decomplexation Strategy in Snake Venom Proteomics . . . . . . . . . . . . . 83
Choo Hock Tan, Kae Yi Tan, and Nget Hong Tan
6 Fractionation Techniques to Increase Plant Proteome Coverage:
Combining Separation in Parallel at the Protein and the Peptide Level . . . . . . . . 93
Martin Černý, Miroslav Berka, and Hana Habánová
7 A Systematic Analysis Workflow for High-Density Customized
Protein Microarrays in Biomarker Screening. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
Rodrigo Garcı́a-Valiente, Jonatan Fernández-Garcı́a,
Javier Carabias-Sánchez, Alicia Landeira-Viñuela, Rafael G
ongora,
Marı́a Gonzalez-Gonzalez, and Manuel Fuentes
8 Metaproteomics Study of the Gut Microbiome . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Lisa A. Lai, Zachary Tong, Ru Chen, and Sheng Pan
9 Double One-Dimensional Electrophoresis (D1-DE) Adapted
for Immunoproteomics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
Youcef Shahali, Hélène Sénéchal, and Pascal Poncet
10 BioID: A Proximity-Dependent Labeling Approach in Proteomics Study . . . . . . 143
Peipei Li, Yuan Meng, Li Wang, and Li-jun Di
11 Functional Application of Snake Venom Proteomics in In Vivo
Antivenom Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
Choo Hock Tan and Kae Yi Tan
12 Proteomic Detection of Carbohydrate-Active Enzymes (CAZymes)
in Microbial Secretomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
Tina R. Tuveng, Vincent G. H. Eijsink, and Magnus Ø. Arntzen
13 An Overview of Mass Spectrometry-Based Methods
for Functional Proteomics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
J. Robert O’Neill

vii
viii Contents

14 Functional Proteomic Analysis to Characterize Signaling Crosstalk. . . . . . . . . . . . 197

Sneha M. Pinto, Yashwanth Subbannayya, and T. S. Keshava Prasad
15 Identification of Unexpected Protein Modifications by Mass
Spectrometry-Based Proteomics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
Shiva Ahmadi and Dominic Winter
16 Label-Free LC-MS/MS Strategy for Comprehensive Proteomic
Profiling of Human Islets Collected Using Laser Capture Microdissection
from Frozen Pancreata . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
Lina Zhang, Giacomo Lanzoni, Matteo Battarra, Luca Inverardi,
and Qibin Zhang
17 Targeted Proteomics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
Yun Chen and Liang Liu
18 Metabolomic Investigation of Staphylococcus aureus Antibiotic
Susceptibility by Liquid Chromatography Coupled to High-Resolution
Mass Spectrometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
Sandrine Aros-Calt, Florence A. Castelli, Patricia Lamourette,
Gaspard Gervasi, Christophe Junot, Bruno H. Muller,
and François Fenaille
19 Nuts and Bolts of Protein Quantification by Online Trypsin
Digestion Coupled LC-MS/MS Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
Christopher A. Toth, Zsuzsanna Kuklenyik, and John R. Barr
20 Proteases: Pivot Points in Functional Proteomics . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
Ingrid M. Verhamme, Sarah E. Leonard, and Ray C. Perkins
21 The Use of Combinatorial Hexapeptide Ligand Library (CPLL)
in Allergomics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393
Youcef Shahali, Hélène Sénéchal, and Pascal Poncet
22 Efficient Extraction and Digestion of Gluten Proteins . . . . . . . . . . . . . . . . . . . . . . . 405
Haili Li, Keren Byrne, Crispin A. Howitt, and Michelle L. Colgrave
23 Glycosylation Profiling of Tumor Marker in Plasma Using
Bead-Based Immunoassay. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413
Hongye Wang, Zheng Cao, Hu Duan, and Xiaobo Yu
24 Protein-Specific Analysis of Invertebrate Glycoproteins . . . . . . . . . . . . . . . . . . . . . . 421
Alba Hykollari, Daniel Malzl, Iain B. H. Wilson,
and Katharina Paschinger
25 The Use of Proteomics Studies in Identifying Moonlighting Proteins . . . . . . . . . 437
Constance Jeffery
26 Two-Dimensional Biochemical Purification for Global Proteomic
Analysis of Macromolecular Protein Complexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 445
Reza Pourhaghighi and Andrew Emili
27 A Data Analysis Protocol for Quantitative Data-Independent
Acquisition Proteomics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 455
Sami Pietil€ a , Tomi Suomi, Juhani Aakko, and Laura L. Elo

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 467
Contributors

JUHANI AAKKO  Turku Centre for Biotechnology, University of Turku and Åbo Akademi
University, Turku, Finland
SHIVA AHMADI  Institute for Biochemistry and Molecular Biology, University of Bonn, Bonn,
Germany
MAGNUS Ø. ARNTZEN  Faculty of Chemistry, Biotechnology and Food Science, Norwegian
University of Life Sciences (NMBU), Ås, Norway
SANDRINE AROS-CALT  Service de Pharmacologie et d’Immunoanalyse, Laboratoire d’Etude
du Métabolisme des Médicaments, CEA, INRA, Université Paris Saclay, MetaboHUB,
Gif-sur-Yvette, France; bioMérieux S.A., Marcy l’Etoile, France
JOHN R. BARR  Division of Laboratory Sciences, Centers for Disease Control and Prevention,
Atlanta, GA, USA
MATTEO BATTARRA  Diabetes Research Institute, University of Miami, Miami, FL, USA
MIROSLAV BERKA  Faculty of AgriSciences, Department of Molecular Biology and
Radiobiology, CEITEC—Central European Institute of Technology, Phytophthora Research
Centre, Mendel University in Brno, Brno, Czech Republic
KEREN BYRNE  CSIRO Agriculture and Food, St Lucia, QLD, Australia
ZHENG CAO  Department of Laboratory Medicine, Beijing Obstetrics and Gynecology
Hospital, Capital Medical University, Beijing, China
JAVIER CARABIAS-SÁNCHEZ  Proteomics Unit, Cancer Research Centre (IBMCC/CSIC/
USAL/IBSAL), Salamanca, Spain
FLORENCE A. CASTELLI  Service de Pharmacologie et d’Immunoanalyse, Laboratoire d’Etude
du Métabolisme des Médicaments, CEA, INRA, Université Paris Saclay, MetaboHUB,
Gif-sur-Yvette, France
MARTIN ČERNÝ  Faculty of AgriSciences, Department of Molecular Biology and
Radiobiology, CEITEC—Central European Institute of Technology, Phytophthora Research
Centre, Mendel University in Brno, Brno, Czech Republic
RU CHEN  Department of Medicine, University of Washington, Seattle, WA, USA
YUN CHEN  School of Pharmacy, Nanjing Medical University, Nanjing, China
MICHELLE L. COLGRAVE  CSIRO Agriculture and Food, St Lucia, QLD, Australia
LI-JUN DI  Faculty of Health Sciences, Cancer Center, University of Macau, Macau, China
HU DUAN  State Key Laboratory of Proteomics, Beijing Proteome Research Center, National
Center for Protein Sciences (PHOENIX Center, Beijing), Beijing Institute of Lifeomics,
Beijing, China
VINCENT G. H. EIJSINK  Faculty of Chemistry, Biotechnology and Food Science, Norwegian
University of Life Sciences (NMBU), Ås, Norway
LAURA L. ELO  Turku Centre for Biotechnology, University of Turku and Åbo Akademi
University, Turku, Finland
ANDREW EMILI  Donnelly Centre for Cellular and Biomolecular Research, University of
Toronto, Toronto, ON, Canada; Department of Biology, Boston University, Boston, MA,
USA; Department of Biochemistry, Boston University, Boston, MA, USA; Center for
Network System Biology, Boston University, Boston, MA, USA

ix
x Contributors

FRANÇOIS FENAILLE  Service de Pharmacologie et d’Immunoanalyse, Laboratoire d’Etude du

Métabolisme des Médicaments, CEA, INRA, Université Paris Saclay, MetaboHUB,
Gif-sur-Yvette, France
JONATAN FERNÁNDEZ-GARCÍA  Proteomics Unit, Cancer Research Centre (IBMCC/CSIC/
USAL/IBSAL), Salamanca, Spain
MANUEL FUENTES  Proteomics Unit, Cancer Research Centre (IBMCC/CSIC/USAL/
IBSAL), Salamanca, Spain; Department of Medicine and Cytometry General Service-
NUCLEUS, Cancer Research Centre (IBMCC/CSIC/USAL/IBSAL), Salamanca, Spain
RODRIGO GARCÍA-VALIENTE  Proteomics Unit, Cancer Research Centre (IBMCC/CSIC/
USAL/IBSAL), Salamanca, Spain
GASPARD GERVASI  bioMérieux S.A., Marcy l’Etoile, France
RAFAEL GÓNGORA  Proteomics Unit, Cancer Research Centre (IBMCC/CSIC/USAL/
IBSAL), Salamanca, Spain; Department of Medicine and Cytometry General Service-
NUCLEUS, Cancer Research Centre (IBMCC/CSIC/USAL/IBSAL), Salamanca, Spain
MARÍA GONZALEZ-GONZALEZ  Proteomics Unit, Cancer Research Centre (IBMCC/CSIC/
USAL/IBSAL), Salamanca, Spain; Department of Medicine and Cytometry General
Service-NUCLEUS, Cancer Research Centre (IBMCC/CSIC/USAL/IBSAL),
Salamanca, Spain
HANA HABÁNOVÁ  Faculty of AgriSciences, Department of Molecular Biology and
Radiobiology, CEITEC—Central European Institute of Technology, Phytophthora Research
Centre, Mendel University in Brno, Brno, Czech Republic
CRISPIN A. HOWITT  CSIRO Agriculture and Food, Canberra, ACT, Australia
ALBA HYKOLLARI  Department für Chemie, Universit€ a t für Bodenkultur, Vienna, Austria
LUCA INVERARDI  Diabetes Research Institute, University of Miami, Miami, FL, USA
CONSTANCE JEFFERY  Department of Biological Sciences, University of Illinois at Chicago,
Chicago, IL, USA
CHRISTOPHE JUNOT  Service de Pharmacologie et d’Immunoanalyse, Laboratoire d’Etude du
Métabolisme des Médicaments, CEA, INRA, Université Paris Saclay, MetaboHUB,
Gif-sur-Yvette, France
SRIKANTH KAKUMANU  Focus Proteomics, Hudson, NH, USA
T. S. KESHAVA PRASAD  Center for Systems Biology and Molecular Medicine, Yenepoya
Research Centre, Yenepoya (Deemed to be University), Mangalore, India
ZSUZSANNA KUKLENYIK  Division of Laboratory Sciences, Centers for Disease Control and
Prevention, Atlanta, GA, USA
MATTHEW KURUC  Biotech Support Group LLC, Monmouth Junction, NJ, USA
LISA A. LAI  Department of Medicine, University of Washington, Seattle, WA, USA
PATRICIA LAMOURETTE  Service de Pharmacologie et d’Immunoanalyse, Laboratoire d’Etude
du Métabolisme des Médicaments, CEA, INRA, Université Paris Saclay, MetaboHUB,
Gif-sur-Yvette, France
ALICIA LANDEIRA-VIÑUELA  Proteomics Unit, Cancer Research Centre (IBMCC/CSIC/
USAL/IBSAL), Salamanca, Spain; Department of Medicine and Cytometry General
Service-NUCLEUS, Cancer Research Centre (IBMCC/CSIC/USAL/IBSAL),
Salamanca, Spain
GIACOMO LANZONI  Diabetes Research Institute, University of Miami, Miami, FL, USA
SARAH E. LEONARD  Chemical and Biomolecular Engineering, University of Illinois
Champaign-Urbana School of Chemical Sciences, Champaign, IL, USA
 Contributors xi

HAILI LI  CSIRO Agriculture and Food, St Lucia, QLD, Australia; Institute of Animal
Husbandry and Veterinary Science, Henan Academy of Agricultural Sciences, Zhengzhou,
Henan, China
PEIPEI LI  Faculty of Health Sciences, Cancer Center, University of Macau, Macau, China
LIANG LIU  School of Pharmacy, Nanjing Medical University, Nanjing, China
DANIEL MALZL  Department für Chemie, Universit€ a t für Bodenkultur, Vienna, Austria
YUAN MENG  Faculty of Health Sciences, Cancer Center, University of Macau, Macau,
China
BRUNO H. MULLER  bioMérieux S.A., Marcy l’Etoile, France
SHENG PAN  Institute of Molecular Medicine, University of Texas Health Science Center at
Houston, Houston, TX, USA
KATHARINA PASCHINGER  Department für Chemie, Universit€ a t für Bodenkultur, Vienna,
Austria
RAY C. PERKINS  New Liberty Proteomics Corporation, New Liberty, KY, USA
SAMI PIETIL€a  Turku Centre for Biotechnology, University of Turku and Åbo Akademi
University, Turku, Finland
SNEHA M. PINTO  Center for Systems Biology and Molecular Medicine, Yenepoya Research
Centre, Yenepoya (Deemed to be University), Mangalore, India
PASCAL PONCET  Allergy and Environment Team, Biochemistry Department, Armand
Trousseau Children Hospital (AP-HP), Paris, France; Center for Innovation and
Technological Research, Institute Pasteur, Paris, France
REZA POURHAGHIGHI  Donnelly Centre for Cellular and Biomolecular Research, University
of Toronto, Toronto, ON, Canada
J. ROBERT O’NEILL  Cancer Research UK Edinburgh Centre, MRC Institute of Genetics
and Molecular Medicine, The University of Edinburgh, Edinburgh, UK; Department
of Clinical Surgery, Royal Infirmary of Edinburgh, Edinburgh, UK
SWAPAN ROY  Biotech Support Group LLC, Monmouth Junction, NJ, USA
HÉLÈNE SÉNÉCHAL  Allergy and Environment Team, Biochemistry Department, Armand
Trousseau Children Hospital (AP-HP), Paris, France
YOUCEF SHAHALI  Razi Vaccine and Serum Research Institute, Agricultural Research,
Education and Extension Organization (AREEO), Karaj, Iran
GARY SMEJKAL  Focus Proteomics, Hudson, NH, USA
YASHWANTH SUBBANNAYYA  Center for Systems Biology and Molecular Medicine, Yenepoya
Research Centre, Yenepoya (Deemed to be University), Mangalore, India
TOMI SUOMI  Turku Centre for Biotechnology, University of Turku and Åbo Akademi
University, Turku, Finland
CHOO HOCK TAN  Venom Research and Toxicology Laboratory, Department of
Pharmacology, Faculty of Medicine, University of Malaya, Kuala Lumpur, Malaysia
KAE YI TAN  Protein and Interactomic Laboratory, Department of Molecular Medicine,
Faculty of Medicine, University of Malaya, Kuala Lumpur, Malaysia
NGET HONG TAN  Protein and Interactomic Laboratory, Department of Molecular
Medicine, Faculty of Medicine, University of Malaya, Kuala Lumpur, Malaysia
ZACHARY TONG  Department of Medicine, University of Washington, Seattle, WA, USA
CHRISTOPHER A. TOTH  Division of Laboratory Sciences, Centers for Disease Control and
Prevention, Atlanta, GA, USA
TINA R. TUVENG  Faculty of Chemistry, Biotechnology and Food Science, Norwegian
University of Life Sciences (NMBU), Ås, Norway
xii Contributors

INGRID M. VERHAMME  Department of Pathology, Microbiology and Immunology,

Vanderbilt University School of Medicine, Nashville, TN, USA
DAVID WANG  University of Iowa School of Medicine, Iowa City, IA, USA
HONGYE WANG  State Key Laboratory of Proteomics, Beijing Proteome Research Center,
National Center for Protein Sciences (PHOENIX Center, Beijing), Beijing Institute
of Lifeomics, Beijing, China
LI WANG  Faculty of Health Sciences, Cancer Center, University of Macau, Macau, China
IAIN B. H. WILSON  Department für Chemie, Universit€ a t für Bodenkultur, Vienna, Austria
DOMINIC WINTER  Institute for Biochemistry and Molecular Biology, University of Bonn,
Bonn, Germany
XIAOBO YU  State Key Laboratory of Proteomics, Beijing Proteome Research Center,
National Center for Protein Sciences (PHOENIX Center, Beijing), Beijing Institute
of Lifeomics, Beijing, China
LINA ZHANG  Center for Translational Biomedical Research, University of North Carolina
at Greensboro, Kannapolis, NC, USA
QIBIN ZHANG  Center for Translational Biomedical Research, University of North Carolina
at Greensboro, Kannapolis, NC, USA; Department of Chemistry and Biochemistry,
University of North Carolina at Greensboro, Greensboro, NC, USA
 Chapter 1

Making the Case for Functional Proteomics

Ray C. Perkins

Abstract
“Making the Case for Functional Proteomics” first differentiates the Functional Proteome from the
products of genetic protein expression. Qualitatively, the prevalence of posttranslational modifications
(PTMs) virtually insures that individual, functional proteins do not equate to their genetic expression
counterparts. Quantitatively, considering the frequency of PTMs and a conservative estimate of the number
of functional entities arising from protein interactions, the size of the Functional Proteome exceeds that of
the human genome by at least two orders of magnitude. The human genome does not, cannot, map the
Functional Proteome. Further, the collective genome of the human microbiome dwarfs the human
genome. With these facts established, “Making the Case. . .” proceeds to examine Functional Proteomics
(of which both “gene expression” and “epigenetics” are but parts of a larger whole) within the context of
Systems Biology, concluding that functionally related networks comprise the dominant motif for biological
activity. Creating just such a network focus is essential in not only expanding basic knowledge but also in
applying that knowledge in the pragmatic efforts of drug and biomarker development. Outlines for
development of drugs and biomarkers, as well as the realization of precision medicine, within a functional
proteomics-based, network motif are provided. The chapter proceeds to asses both the knowledge base and
the tools to fully embrace Functional Proteomics. Given the decades-long infatuation with the reduction-
ism of genomics, it is not surprising that both the proteomics knowledge base and tools are assessed as poor
to fair. However, even a minor shift in research funding and a renewed challenge to methods developers will
rapidly improve the current situation. Adoption of the included “Roadmap” will realistically make the
twenty-first century the century of a long-awaited revolution in biology.

Key words Protein, Gene, Genome, Proteome, Functional Proteome, Proteomics, Functional prote-
omics, Microbiome, Posttranslational modifications, Protein interactions, Epigenetics, Gene expres-
sion, Biological networks, Systems biology, Drug development, Biomarker development, Precision
medicine

1 Introduction

The title of this chapter, “Making the Case for Functional Proteo-
mics,” can elicit a sense of the absurd. For example, it is difficult to
name a single biological property or process that does not rely on
the “function” of proteins. Basic properties such as size and shape
of the organism, its interaction with the environment, and the
existence of functionally specialized structures directly reflect the

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_1, © Springer Science+Business Media, LLC, part of Springer Nature 2019

1
2 Ray C. Perkins

properties of proteins. Large-scale processes such as metabolism,

digestion, temperature maintenance, and organism reproduction
all depend on the integrated functioning of proteins. At the small
scale of transmembrane transport, pathogen detection, transcrip-
tion, and translation, protein function is essential to the health and
well-being of the organism. It is the latter of the noted small-scale
activates—processes related to genetics—that creates the need for
this chapter and this book: the reductionism and distraction of
decades of DNA sequencing. In this context, “Making the Case
for Functional Proteomics” is crucial.
This chapter interprets “Making the Case. . .” in two ways, one
broad in scope and one more narrow though no less essential. The
broad scope encompasses: (1) Specification of the “Functional
Proteome,” (2) The interrelationship of Functional Proteomics
and Systems Biology, and (3) The potential utility of Functional
Proteomics in areas of fundamental importance (e.g., Drug and
Biomarker Discovery). Specification of the Functional Proteome
includes the factual presentation and logic in determining the size
and nature of the “Functional Proteome,” especially as compared
to output associated with the human genome. Examination of
Functional Proteomics within the context of Systems Biology
insures biological relevance, especially in the evaluation of the
utility of Functional Proteomics for disease detection and treat-
ment. The narrower interpretation of “Making the Case. . .” is an
open-eyed evaluation of the existing knowledge that comprises
Functional Proteomics. Accessibility of fundamental proteomic
metrics, e.g., protein identity, localization, and activity, is assessed
and metrics are indexed to pertinent methods. Comparison of
fundamental metrics to available information sources and methods
identifies areas of weakness and strength for both the methods per
se and the status of existing coverage of the Functional Proteome.
Major headings and subheadings for “Making the Case. . .” are:
l The Functional Proteome in Relation to the Genome
l The Functional Proteome and Systems Biology
l Functional Proteomics Utility:
– Biomarker and Diagnostic/Prognostic Discovery
– Drug Discovery
– Precision Medicine
l Functional Proteome Knowledge Base
– Level of Knowledge/Ignorance
– Proteomics Methods Appraisal
l Closing Comments
Summarizing key observations in this chapter it is noted that
the Functional Proteome is orders of magnitude larger than the
 Making the Case for Functional Proteomics 3

Genome. Functional Proteomics is a natural context for Systems

Biology regardless of the level of complexity. Biomarkers developed
within a Functional Proteomics context are naturally multiplexed,
selecting not for single proteins but for functional Networks and
Pathways. Functional Proteomics naturally integrates with the fore-
front of drug discovery, phenotypic drug discovery. Unfortunately,
given the enormous potential of Functional Proteomics as a gener-
ator of new knowledge and therapeutics, the existing knowledge
base must be considered poor. Among other areas, this ignorance
inhibits proteomic contributions to Precision Medicine. Fortu-
nately the cure for ignorance is well known, comprising hard
work, ingenuity, and commitment.

2 The Functional Proteome in Relation to the Genome

The gene COL1A1 on the minus strand of chromosome 17 starting

at 50 million base pairs does not provide a primary basis for the
structural integrity of the human body—the protein, collagen, does
[1]. The gene PRSS1 on the plus strand of chromosome 7 starting
at 143 million base pairs does not digest food—the protein, trypsin,
does. The genes HBB, HBA1, and HBA2 located on chromosomes
11 and 16 do not serve as a blood circulation transport vehicle for
oxygen in blood—the protein, hemoglobin, does. These examples
can be even more instructive in establishing the non-relationship
between genome and proteome. The translation products for both
collagen and trypsin are not active. The translation product for
collagen includes an N-terminus signal peptide and propeptides at
both the N and C termini. All three peptides, totaling 406 amino
acids, are excised from the translation product to produce the
collagen alpha1 chain. Similarly, both an N-terminus signal peptide
and N-terminus propeptide (required for activation) are excised
from the translation product to produce Trypsin 1. Turning to
the third protein in this example, as already noted, the subunits of
hemoglobin are individually translated: three translation products
combine to produce one functional protein. For none of the proteins
cited can the translational products be considered functional proteins.
The generality of this three-protein exercise is detailed throughout
this chapter.
Defining the Functional Proteome has already begun and con-
tinues immediately below, maintaining comparisons to the
genome. Mining publicly accessible resources, the number of
expressed proteins, those arising from full-gene transcription and
alternative expression, will be estimated. A second estimate of the
number of distinct single-protein entities with human genetic ori-
gins is made, accounting for annotated frequencies of posttransla-
tion modification. Included is a brief but telling estimation of the
impact of human microbiota on the number of organism-wide
4 Ray C. Perkins

proteins. Enumerating distinct protein entities is only the begin-

ning. Included also is a summary of activities or events that lead to
differential functionality, extending the scope and nature of the
“Functional Proteome.”
How many proteins are in the human body? This is the starting
point and, surprisingly, finding a consensus number is tricky—
estimates range from 250,000 to millions [2]. A straightforward
exercise helps sort out relative orders of magnitude. Data for this
exercise are those accessible at Uniprot [1], a public database with
multilevel search criteria and flexible output formats. Uniprot dis-
tinguishes manually annotated (“Reviewed”) data from data mined
without review from literature searches or from other databases.
Selecting for “Homo sapiens (Human) [9606]” as “Organism” at
UniProt produces 160,566 hits, of which 20,239 are manually
annotated or curated (search date: November, 2017). The data in
the manually annotated base (aka “Swiss-Prot”) include, among a
host of other information, the number of posttranslational modifi-
cation (PTM) products for individual proteins by kind and
sequence location. Data was downloaded and collated for the four
most frequently occurring single amino acid PTM’s and the results
are summarized in Table 1 (“Modified Residue” includes single
amino acid modifications including phosphorylation, acetylation,
hydroxylation, etc. “Chain” includes key proteolytic events such as
“Signal,” “Initiator,” and “Propeptide.” Proteolysis is extensively
addressed in Chapter 20).
For each expressed protein, 6.1 distinct proteins exist (follow-
ing other treatments, the existence of the protein as expressed is
assumed, though as is later argued, may not represent an active
form of the protein). With this number in hand along with the

Table 1
Posttranslational categories and incidence

PTM processing category PTMs per protein

Crosslink 0.21
Modified residue 2.67
Glycosylation 1.01
Disulfide bond 0.88
Chain 0.33
Average PTMs per protein 5.1
This table summarizes the posttranslational modifications (PTMs) for an average protein,
derived from data maintained by Uniprot. (Although glycosylation is technically a
modified residue, it is common enough to be given its own category.) Assuming the
unmodified protein and all of its modified forms are active, each protein-coding gene
yields an average of six functionally distinct proteins
 Making the Case for Functional Proteomics 5

number of genes and the number of alternative expression products

per gene, the number of distinct proteins may be estimated. Pre-
Human Genome Project (HGP) estimates of the number of human
genes range into the millions, with most estimates between 40,000
and 140,000 [3]. As the HGP progressed the number progressively
decreased and, even today, continues to decrease with recent
reports placing the number of genes close to 19,000 [4]. The
number of proteins expressed per gene is cited at 3.4 [5] (Uniprot
data indicate an average of 2.5 proteins expressed per gene from
alternative splicing alone, in reasonable agreement with published
data). Assuming that alternatively expressed proteins are compara-
bly subject to PTM processes, an estimate of the Functional Prote-
ome size is:
ðGenome SizeÞ  ðNumber of Proteins per GeneÞ


 Number of PTM’ s per Protein
¼ 329, 460 proteins
How then does the gene-expressed proteome relate to the
Functional Proteome? The number of directly expressed proteins
tallies to some 65,000 (19,000 genes  3.4 proteins per gene).
Each of the expressed proteins is subject, on average, to five post-
translation modifications. Thus, given the high rate of posttransla-
tional modification, it is likely that few or no proteins have zero
modifications: the gene-expressed proteome comprises little or
none of the Functional Proteome. Put more prosaically, the
human genome does not map the human proteome either qualita-
tively or quantitatively. And it follows that. . .
ProteinGene 6¼ ProteinFunctional
The exercise thus far defines the number of individual proteins
that comprise the human proteome, but this is just a starting point
for defining the “functional” proteome. The exercise continues,
then, with expanding the number of functional entities based on
protein interactions with other biomolecules.
Protein interactions are at the heart of a high percentage of
biological activity—including genetics. Proteins self-associate and
hetero-associate, create complexes with multiple centers, associate
with polynucleotides, and bind to membranes. In each case the
interaction induces structural changes in all participating entities
and, as function follows structure, the composite entity produced
by any interaction must be classified as a distinct “functional”
entity. The question, then, for purposes of the exercise, is in how
many interactions is each protein a participant? As might be
expected, estimates vary but a reasonable and conservative figure
for binary protein interactions alone is five (a number also borne
out by Uniprot data). Adding the fact of five interactions per
6 Ray C. Perkins

a b
1,976,760
Functional Proteome Size Estimate Non-Reduntant
(Human) Genes: Human vs
Microbiota

19,000 Human Genes & 2.4 Alternative Human:

Expression Products 19,000
5.1 Post-Translational Modifications Microbiota:
per Protein 3,300,000
5 Binary Protein-Protein Interactions
per Protein

329,460

64,600
19,000

Human Genes Alternative Post Translation Protein-Protein

Expression Modifications Interactions
Human Microbiota

Fig. 1 (a) The challenge of proteomics lies in its fundamental, irreducible complexity. Taking into account
alternative splicing and posttranslational modifications, a single human gene can produce an average of
12 unique proteins, all needing to be identified and catalogued. The true interest in proteins lies in their
functions, however, and for that we must discern, quantify, catalogue, and compare all of their many
interactions. With an average of five binary interactions for every protein (ignoring multi-center protein
complexes and interactions with other types of molecules), we are confronted with a staggering sum of
nearly two million to identify and study! Compare this to the measly 19,000 protein-coding genes in the human
genome, and it is clear why reductionist approaches will never be up to the task. (b) The impact of the
microbiome, based on its sheer size alone, must be significant. However, there is scarce data on the impact of
non-infectious bacterial proteins on human proteins. (The effect of some infectious bacterial proteins are
discussed in “Proteases: Pivot Points in Functional Proteomics”)

protein which, when incorporated into the proteome size estimate

of 308,142, results in an estimate for the Functional Proteome of
(Fig. 1a):
ðNumber of Bound Entities þ Free ProteinÞ
 ðNumber of Distinct ProteinsÞ
¼ 1, 976, 760 Functional Protein Entities
Interestingly, the exercise converges on the widely divergent
estimates cited above that range from a quarter-million to millions.
The difference is a consideration of individual entities vs. functional
entities. Note that the relative cumulative size of the Functional
Proteome is also a reflection of relative information content—a
 Making the Case for Functional Proteomics 7

topic that will be revisited in the segment on Precision Medicine.

Even now, the exercise is not yet complete given that the human
organism comprises both genetic and proteomic contributions
from more than one species—or more than one source depending
on how human microbiota are considered. If now the microbiota
resident in each human are to be included, at least one report notes
that there are 3.3 million, nonredundant genes in the human
microbiome (Fig. 1b). At a minimum the microbiota proteins
that are not strictly intracellular functionally contribute to human
biological activity. The collective “human” proteome could readily
exceed a million distinct proteins and up to five million additional
functional entities. Ignoring the microbiome altogether
(in recognition of the paucity of proteomic data), the size of just
the Functional Proteome is two orders of magnitude larger than
the human genome. Even this is an understatement. In addition to
interactions with other large molecules, proteins interact with,
indeed are activated by, small molecules—small nucleotides, meta-
bolites, peptides, lipids, and even water. Each of these interactions,
when viewed from a protein function perspective, adds yet another
member to the Functional Proteome. Even then the exercise is not
technically complete as each conformation of a single protein is also
a potential driver of activity (e.g., protein “folding” and “misfold-
ing”) and, thus, functionality. Distilling the manifestations of func-
tionality that are biologically relevant is a significant challenge but
one that must be accepted. The payoff is potentially enormous.
When viewed as an information resource, the Functional Proteome
is not only orders of magnitude larger than the genome, but the
quality of the information reflects the instant to instant dynamic
that reflects life.
Quantitatively and qualitatively, the vital activities that com-
prise a complex organism, that reflect health and disease, and that
drive the interaction of that organism with its environment, cannot
be defined or anticipated by their genome. The activity that is
biology must focus on the actors: proteins, both free and bound.

3 The Functional Proteome and Systems Biology

This section of “Making the Case” is the context for the remainder
of the chapter. The backdrop is, once again, decades of emphasis on
the genome. The relatively recent expectation of genomics is
expressed in the following attributed to Francis Collins (currently
director of the US National Institutes of Health) in 2006 [6]:
“Essentially, we are now able to read our own instruction books. It is also a
history book explaining how humans have evolved over time. It’s a shop manual
that describes with incredible precision how to build every cell in the human
body. And it’s a medical textbook containing insights that will help doctors
predict and, eventually, cure disease.”
8 Ray C. Perkins

Systems Biology Hierarchy

Molecules Networks Cells Tissues Organism

And Pathways And Particles And Organs

Inter-Molecular Inter-Network Inter-Cellular Inter-Tissue Inter-Organism

Detail, Flexibility, Economy Relevance, Complexity, Cost

Fig. 2 The human body is so complex that we have traditionally approached it as a multilayered hierarchy of
tissues, cells, and molecules. While this is a useful organizational tool, in reality these layers are not separate.
They inform and influence each other. Therefore, our approach to studying them must not treat them as
separate but as an integrated whole, Systems Biology. When investigating the detailed mechanisms of protein
function, an awareness must be maintained of their cellular environment, and any deviations from biological
conditions should be minimized and must be acknowledged. Testing on the organism level should be done
with an eye toward how the interconnected pathways of molecular networks may cause unintended effects.
More than anything else, a common vocabulary of materials identification and handling must be developed
and maintained between the “disciplines” so that apples-to-apples comparisons can be made between
studies

Now, little more than a decade later, the reality, a “genome

bubble” [7] is much different [8].
“Having the complete gene set on the table, the knowledge of the genetic map
and sequence is now considered by experts to be only a starting point for future
research in biology and medicine,”

In this aftermath of the “genome bubble,” the context for

continued research and translation of that research must change,
and that change hinges upon the adoption of a more integrative
and pragmatic paradigm, namely Systems Biology (Fig. 2) [9]. The
nature and purpose of genes is accommodated as a part of a larger
whole, as is “epigenetics,” the functional activity that interfaces
genes to the larger System. Systems Biology also encompasses
earlier and ongoing schisms of, for example, the relative merit of
in vitro versus in vivo research. Systems Biology not only acknowl-
edges the value of multiple endeavors and test-material selection, it
insists on their integration and synthesis: a tall order, to state the
obvious.
Systems Biology then is a multilevel, zoom-in and zoom-out
process, at one moment focused on single molecules and the next
moment on the response of the organism to external stimuli.
Systems Biology as an approach is the perfect context for multiple
goals: increase in basic knowledge, understanding disease, develop-
ment of new therapeutics and diagnostics, and the implementation
of precision medicine. Systems Biology imposes a discipline and a
context—“translatability,” as will be further explored in the seg-
ment on Drug Discovery [10]. Molecular studies must be per-
formed and analyzed in relation to their presumed context in the
 Making the Case for Functional Proteomics 9

organism. Where molecular methods force a deviation from a

physiological context, that deviation must be acknowledged and
its potential qualification explicitly stated in terms understandable
for the non-expert. Network, Cellular, Tissue, and Organism work
must handle all materials, most especially proteins, in compliance
with diverse molecular observations in order to avoid introducing
artifacts. As with reports on Molecular activity, all area reports must
state the qualifications of the results in approachable terms. Work
performed at variable levels of complexity is both iterative and
recursive—knowledge gained, e.g., on cells may well inform new
work to be done on molecules. Without the knowledge gained,
e.g., on Networks, work performed on Cells is ill-informed. There
are no shortcuts. A series of related and ongoing efforts that
include the author and colleagues, and multiple, international
laboratories provides an example.
In 2015, New Liberty Proteomics (NLP) [11] was engaged to
assess the ability of a library of test molecules to modulate the
interaction of a peptide and a protein, the former associated with
well-known disease pathology and the latter identified as a genetic
risk factor for the same disease. Thus the Molecular assessment was,
from the outset, connected to the Organism in the forms of disease
manifestation and genetic predisposition to disease. Within the
capabilities of the selected methodology (electron paramagnetic
resonance spectroscopy and spin labeling) both the peptide-protein
interaction and its modulation would, predictably, assume charac-
teristic patterns. Upon completion of the screening, a portion of
the data did, indeed, correspond to the expected pattern. However,
that pattern was only one of four distinct apparent mechanisms of
action exhibited by the library of molecules. Instead the library
manifested four distinct apparent mechanisms of action: inhibition
of peptide-protein binding, promotion of peptide-protein binding,
and two of which altered the peptides conformation. Or so it
seemed. NLP’s work became a touchstone for additional biophysi-
cal work in other labs, which confirmed the two peptide-protein
interaction modulations—some molecules inhibited the interaction
and others promoted it (limitations of the second method
employed missed the apparent changes in peptide conformation).
These collective results drove support for further work in which
additional, disease-connected proteins were incorporated into both
biophysical laboratories. In the case of the NLP work, combina-
tions of three or more proteins were examined, thus moving into
the arena of Networks (NLP also pursued the peptide alone and
confirmed earlier suspicions that some test molecules did indeed
impact the peptide directly, though the precise interpretation was
still elusive.). Work in both labs demonstrated once again diverse
activity of the test library among multiple combinations of peptides
and proteins, thus laying the foundation for other labs focused on
Cells and Tissues continuing the progression toward Organism.
10 Ray C. Perkins

Work on both Cells and Tissues proceeded with a reduced set of

molecules, identified from early biophysical results, and continued
success was and is forthcoming. Thus the progression from Mole-
cules (selected for direct connection to Organism) to Cells pro-
ceeds in a rational fashion despite the work being performed in labs
with diverse locations and specialties. Incorporation of Organisms,
mice in this instance, awaits further funding. Additional reference
will, as appropriate, be made to this body of work in latter segments
of this chapter.
As in the example, implementation of Systems Biology is inti-
mately dependent on the selection of experimental observable(s).
Optimistic views of the “instruction book” or “shop manual”
potential of genomics have not proven viable in the face of ongoing
data collection. Disease cannot be readily diagnosed nor are sys-
tems’ level data forthcoming. Instead, given the multiplicity of
environments, the variety of tasks performed, and the responsive-
ness to stimuli, one logical selection of molecules-as-observables is
proteins. Categorically proteins are active drivers of digestion,
metabolism, pathogen response, and muscle contraction. They
also serve as primary structural elements, inter-tissue transport
carriers, and storage devices. At a cellular level proteins control
entry and egress of diverse molecules, recycle spent molecules and
cells, and regulate genetic processes. At a molecular level, proteins
engage in diverse interactions with small molecules, macromole-
cules, and membrane surfaces. They produce and are subject to
multiple revisions, assume multiple conformations, and exist in
multiple oligomeric states. Modifying protein activity is a primary
goal and/or outcome of therapeutic treatment and assessing enzy-
matic activity is a dominant indicator for medical diagnostics. These
latter examples—drug and diagnostic development—will be
addressed in detail in subsequent segments of this chapter.
Implementation of Systems Biology in any context is challeng-
ing, even intimidating. The number of research specialties repre-
sented by the test-material-complexity hierarchy tallies into the
hundreds. Publications within those specialties would correspond-
ingly tally into the tens of thousands per year. Few if any single
individuals exist with expertise in all areas, and cross-specialty com-
munication is difficult and no funding practices accommodate such
breadth and depth in academia or industry. Experiments on
humans are, rightly, subject to regulation. Nonetheless Systems
Biology establishes a logical, progressive, and self-correcting para-
digm. Application and entry points for Functional Proteomics are
works-in-progress as seen in the discussions below on Biomarker
and Diagnostics Development, Drug Discovery and implementa-
tion of Precision Medicine.
 Making the Case for Functional Proteomics 11

4 Functional Proteomics Utility: Biomarker and Diagnostic/Prognostic Discovery

One working definition of “biomarker” is provided by the NIH

[12], “. . . a characteristic that is objectively measured and evaluated
as an indicator of normal biological processes, pathogenic processes, or
pharmacologic responses to a therapeutic intervention.” Blood pres-
sure, pulse rate, and body temperature are longstanding biomar-
kers, as is the analysis of body fluids such as urine [13]. Indeed, the
laboratory analysis of urine in the twenty-first century echoes third
century BC Hippocrates’ noting the color—and taste—of urine.
This also serves as an example of the close relationship between the
object or property associated with a biomarker and available tech-
nology. Fever has long been noted as an indicator of disease.
However, measurement of body temperature with precision and
accuracy depends on access to a reliable thermometer—such as that
of Galileo in 1592 or Fahrenheit in 1714. Given advances and
changes in technology, available biomarkers of the twenty-first
century range from system-level measurements, e.g., body temper-
ature, to identification of pathogens to measurement of protein
levels and gene sequences. The utility of validated biomarkers con-
tinues to expand. Biomarkers aid in diagnosis of disease, suggest
therapy selection, and serve as monitors of treatment efficacy.
Diseases change the function of the organism, an observation
that leads to Fig. 3a which provides a categorization of biomarkers
by function. The function, e.g., of a diagnostic biomarker is disease
differentiation given that the presence of a patient in a physician’s
office is prior evidence of disease per se. The function of a bio-
marker for selection of therapy assumes prior disease diagnosis for
matching therapy to patient—the goal of precision medicine.
Tracking disease progress is the function of a prognostic biomarker.
Within this context there is no a priori reason to assume that a
biomarker that differentiates disease also serves as a biomarker for
therapy selection or for tracking disease progress. However, within
the prevalent model of genomics, the biomarker equivalence for
this apparent diversity is assumed: one gene to bind them. This is
especially true for chronic diseases. In sharp contrast is the pragma-
tism of Functional Proteomics as the vehicle for biomarker devel-
opment. Not only does the Functional Proteome encompass the
diverse activity that is biology, it is immediately responsive to dis-
ease—and treatment—modification of that biology. In like manner,
integration of a patient’s manifestation of disease with known
mechanisms of therapeutic action argues for a Functional Proteo-
mics context. Similar arguments favor Functional Proteomics for
tracking disease progress. Predictive biomarkers, those that puta-
tively anticipate future disease risk, are included in Fig. 3a for the
sake of completeness, and are discussed in the chapter segment on
Precision Medicine.
12 Ray C. Perkins

a b

Diagnostic Prognostic
(Disease Differentiation) (Track Disease Progress)
FDA-Approved, In
2,200
Biomarkers Vitro Diagnostics
Therapy Selection Predictive
(Match Disease & Patient) (Anticipate Disease)
"Companion"
c
“Pattern Diagnostics” Cancer Nucleotide
Diagnostic in Development Protein

9 66

Fig. 3 (a) Biomarkers are, in essence, anything that we can measure that informs us about a person’s health.
Historically, we have used them to diagnose disease and to track disease progress. Recently, there has been
an effort to use them as a means of determining which patients would be best served by which treatments,
which has met with some success. There has also been an effort to use them to predict which diseases a
patient is likely to be afflicted by, which has met with less success and more controversy. (b) In vitro
diagnostic tests are regulated by the FDA, and tests of protein quantities and/or activities outnumber tests for
gene variants 33 to 1. “Companion” diagnostics (of which there are only 9) are tests that select treatment-
suitable patients already diagnosed with a specific disease condition, the hopeful beginnings of the modern
Precision Medicine movement. (c) A recent study of cancer patients [13] detected a pattern of elevated serum
levels of multiple proteins associated with Hemostasis, Inflammation, and the Complement System. For more
information on the connections between these systems, see “Proteases: Pivot Points in Functional
Proteomics”

This chapter segment addresses biomarker discovery and devel-

opment within the paradigm of Systems Biology and the context of
Functional Proteomics. Key areas of focus include biomarker rela-
tionship to disease diagnosis, prognosis, and selection of patient-
appropriate therapy (aka precision medicine). Prediction or risk
assessment is briefly addressed as a separate topic. Examination of
existing diagnostic tests approved by the US Food and Drug
Administration (FDA) is a useful starting point.
The battery of 59,707 in vitro diagnostic tests approved by the
FDA [14] covers a wide range of “Test Systems”: hormone and
metabolite quantitation, detection and identification of pathogens,
measurement of pH, etc. Proteins constitute 22% of the approved
 Making the Case for Functional Proteomics 13

tests, dominated by enzyme activity with the balance being single

protein quantitation (Fig. 3b). Many redundancies appear in the
approved list as multiple companies provide diagnostic tools/ser-
vices for a single assay. Overall, the number of distinct approved
diagnostic assays totals some 10,000 of which some 2200 are
protein assays. Separate listings exist for approved diagnostics that
target nucleotides and the number of entries is much smaller: 99, of
which a third are redundant for purposes of this accounting. Even
shorter is the list of approved “Companion” diagnostics (tests that
select treatment-suitable patients already diagnosed with a specific
disease condition), which includes only nine entries, of which three
are proteomic and six assess specific gene variations. Summarizing
and accounting for redundancy in the listings, the FDA in vitro
diagnostics lists include 2200 protein assays, 66 nucleotide assays,
and 9 companion diagnostics. These numbers reflect both historical
development of biomarkers and the changing face of medicine.
They are also tests that are directly regulated. Other tests are
certainly performed but are not subject to direct FDA regulation,
tests that include cellular examination, e.g., for which expert exe-
cution and interpretation is required. For these tests, the FDA
regulates only the laboratory’s operation under the Clinical Labo-
ratory Improvement Amendments (CLIA) guidelines. The tests
performed, however, are not subject to rigorous validation or
regulation.
Historically, in vitro diagnostics follow the progression from
small molecule to protein to nucleotide. This reflects both the level
of understanding over time and the available technology. Further,
the object of analysis is almost exclusively singular—concentration
of a single protein, a limitation of past technologies. The small
number of assayed proteins (less than 1% of total human proteins)
and nucleotides presages enormous opportunity.
A companion diagnostic is a test administered to select patients
for clinical drug trials. If successful, the companion diagnostic is
subsequently employed in patient treatment as well. The fact that
only nine companion diagnostics have been approved severely lim-
its progress in developing new therapeutics. All nine are either
genetic tests or assays for a gene-associated protein.
The breadth of Functional Proteomics, armed with significant
improvements in key technologies, far exceeds the scope of existing
diagnostics—both in quality and quantity. Whereas the Functional
Proteome numbers in the millions, only a few thousand proteins
have been exploited to date. Furthermore, within those few thou-
sand all focus on a single protein, whether via quantitation of
concentration or assessment of enzyme activity. Within the para-
digm of Systems Biology existing diagnostics address only a small
fraction of the potential of Molecules alone. One opportunity for
new development lies, therefore, in increasing the number of pro-
teins examined while searching for more biologically relevant
14 Ray C. Perkins

patterns. Technology now allows examination of large numbers of

proteins and disease-related variation may be minimally examined
at the level of both Molecular and Network complexity. This pro-
cess is already showing signs of success as will be discussed in the
segment on Diagnostics/Prognostics.
The Functional Proteome is conservatively estimated at two
million functional entities comprising some 300,000 proteins
(Fig. 1a)—reluctantly neglecting the microbiome, if with consider-
able trepidation. The Functional proteome is dynamic in multiple
senses including posttranslational modifications and the interac-
tions of proteins to other proteins, polynucleotides, small mole-
cules, and cell membranes. It is also dynamic in that proteins are
continually being “recycled” with lifetimes of individual proteins
ranging from minutes to months. The Functional Proteome is
multifaceted, plastic, and immediately responsive to internal
changes and external stimuli. By comparison the genome is static
(“epigenetic” and “gene expression” events are covered under the
Functional Proteomics umbrella). Systems Biology is the organism
milieu in which Functional Proteomics operates and responds. The
combination is the context in which discovery and development of
new Diagnostic biomarkers must occur.
As summarized above the majority of existing, protein-based
in vitro diagnostics assess concentrations or activity of single pro-
teins. Within a Systems Biology paradigm, existing in vitro diag-
nostics barely scratch the surface of the least organism-like stage:
Molecules. No protein, indeed no biological molecule including
genes, acts alone. Any deviation in the concentration, locale, or
activity of one protein produces a ripple effect across multiple net-
works and, therefore, through the entire complex hierarchy to the
organism. Any disease state therefore comprises alterations in
numerous Molecules and Networks that in turn alter functionality
at the Cellular, Tissue, and Organism levels. It follows that the
minimum complexity for general discovery of Diagnostics is the
Network level with a sound foundation laid at the Molecule level.
This very approach—referred to by some as “pattern diagnos-
tics”—actively engages many laboratories, academic and commer-
cial, and one promising effort serves as an example [15].
The study is a classic “bottom-up” liquid chromatography-
mass spectrometry (LC-MS/MS) approach followed by progres-
sive focus on promising “hits” on human blood serum. Three
factors distinguish the study from the outset:
1. No a priori expectations or modeling.
2. Removal of serum albumin.
3. Panel restricted to high-detectability proteins.
Willful naiveté is essential as prior expectations or model-based
preconceptions consciously or unconsciously influence both design
 Making the Case for Functional Proteomics 15

and analysis. The high concentrations of serum albumin

(3.5–5.5 g/dL or 500–800 μM) obscure the detection of many
serum proteins. Finally, though low-level proteins are of interest,
the overriding goal in pursuit of a diagnostic-quality biomarker is
repeatability and robustness: a basic set of proteins numbered in the
hundreds versus thousands without the omnipresent confounding
of serum albumin. The company, Biotech Support Group (BSG),
LLC, elected to compare blood serum obtained from cancer
patients versus normal controls, and details are presented elsewhere
in this book. The outcome of their work, differentiation of cancer
patients relative to controls, is of general interest for Functional
Proteomics within a Systems Biology context.
The company has succeeded in demonstrating that the serum
proteome from cancer patients (Fig. 3c) differs from age/sex-
matched controls (biomarker deemed Stroma Liquid Biopsy).
The analysis of their findings, however, goes far beyond creating a
simple list of differentiating proteins. Proteins from three
interconnected pathways or networks emerged. Further, differenti-
ation is heavily dependent on posttranslational modifications and
control mechanisms of those modifications. This provides a bio-
marker that is self-reinforcing with internal controls. Even more
impressive is that the interconnectedness of the biomarker pro-
duces new knowledge—knowledge that potentially contributes on
many fronts. A “bottom line”: Functional Proteomics (in this
instance concentration determination of multiple proteins) points
not to single proteins but to functionally related groups of pro-
teins—consistent with the behavior expected of biological systems.
Thus even nominally Molecular studies can elucidate the behavior
of more biologically complex entities—even the organism itself.
And this only scratches the surface of Functional Proteomics’
potential.
The complex hierarchy of Systems Biology presents numerous
opportunities for application of Functional Proteomics in discovery
and development of diagnostics and prognostics. Application at the
Molecular level is an obvious example. For soluble protein sources
(e.g., serum, plasma, cerebral spinal fluid, lymph, urine, extracellu-
lar fluid, cellular or tissue extracts) protein identification and con-
centration determination should be seen as just the starting point.
As implemented by the Biotech Support Group, the first goal in
discovery of a biomarker is development of an experimental proto-
col that produces robust and reproducible results (protein survey
studies are not “cookbook” experiments). The standards for diag-
nostic discovery are much higher than for simple publication of
results. How reproducible must results be? Once in use as a diag-
nostic, a “good” assay will be 90% “Sensitive” and 90% “Specific”
implying false-negative and false-positive rates of 10%. The assay
must be reproducible well inside the false reporting rates expected
for the diagnostic itself. If, for example, the assay exhibits scatter of
16 Ray C. Perkins

Soluble Protein Levels Diagnostic

Robust Assay

Differentiation Set V0

Comparison to Functional Networks

Revise & Test Differentiation Set

Diagnostic V0

Fig. 4 Represented is a scheme to develop a reliable diagnostic based on protein

concentrations or activity in bodily fluids. It is imperative that the selected assay
be robust and reproducible from the outset, with metrics confined to fit those
criteria. The assay should initially provide breadth so that a wide range of
proteins are assessed (“shotgun”). Differences in test populations, should
differences exist in reality, will suggest connections to biological networks.
These “pattern-based” connections allow the assay test conditions to focus on
the network(s) implicated in greater detail, i.e., selection of a progressively
narrow set of test proteins. Lather, rinse, and repeat until populations
differentiate with both high sensitivity and specificity: a candidate Diagnostic

a few percent of the mean for control samples, that scatter is a

lower-bound for biomarker performance. Therefore, sound metrics
are the first order of business. Assuming that such basic robustness
is achieved and that test populations differentiate with sufficient
precision, analysis of the differentiating proteins leads to consider-
ation of the next level of complexity hierarchy: Networks (Fig. 4).
It is unlikely that an initial set of differentiating proteins—
Differentiation Set V0—is functionally unrelated. It follows that
examination of the proteins in Differentiation Set V0 will direct
attention to one or more known Networks or pathways. Having
inferred which Networks may be represented by Differentiation Set
V0, the potential exists for adding other proteins to a progressively
focused group of proteins, a subproteome (see the Chapter 2,
“Methods to Monitor the Functional Subproteomes of SERPIN
Protease Inhibitors”). Iterating between subproteome selection
and its ability to differentiate the targeted patient group converges
on optimum subproteome selection and diagnostic quality simulta-
neously. The outcome for a successful exploration is Diagnostic V0.
Other options exist within this loop such as selection of various
PTM products for a given protein, selection being based on known
PTMs associated with network processes. The result of this process
is designation of the zeroth version of a diagnostic candidate. While
such an approach shows great promise, it only scratches the poten-
tial of Functional Proteomics as a driver of Diagnostics and
Prognostics.
 Making the Case for Functional Proteomics 17

The example presented covers the most basic level of Func-

tional Proteomics: determination of soluble protein identity and
concentration coupled to known Networks or Pathways. Moving
beyond this relatively simple strategy invokes shifts in Systems
Biology complexity and/or experimental observables. Having dis-
cerned, e.g., that the concentrations of select, soluble proteins
comprise an extended, network-focused diagnostic, a next logical
refinement is assessment of protein interactions within those net-
works. From there, extending the strategy to include Cells is, at
least hypothetically, a progression to increasingly physiological test
materials. Observables at the cellular level expand to include
protein-cell interactions and cell-associated protein distribution.
However, each change in hierarchy complexity and experimental
observable demands a clear-eyed assessment of methods currently
available. These issues are the subject of the chapter segment on
Proteomics Methods Appraisal below. Foreshadowing that analysis,
methods at their current stage of development are a limiting step
for full exploitation of Functional Proteomics—not just for diag-
nostic discovery but for drug discovery and precision medicine
as well.
Closing this segment is a brief summary on the use of diagnos-
tics to predict disease well in advance of clinical manifestation of
that disease. This is certainly not a new concept and is, indeed, the
underpinnings for widespread administration of cholesterol-
lowering drugs to ameliorate or eliminate cardiovascular disease
(the effectiveness of this ongoing clinical practice has been and is
a subject for debate). Nonetheless the desire for early detection of
disease persists. A couple of points must be made. The value of early
detection only exists if disease-specific remedies or practices are
immediately available, i.e., is the prediction actionable? Given that
positive diagnostics and treatments for chronic diseases do not yet
exist, any prescribed disease-avoidance remedy will, at best, be
driven by the disease model du jour—as in the case of cholesterol-
lowering drugs. These simple facts do not, however, temper admin-
istration of non-validated “risk assessment” tests, and this intro-
duces the second point: Administration of any diagnostic to a
general population significantly raises the quality demands for any
diagnostic [16]. Consider, for example, a diagnostic with Sensitivity
and Specificity values of 90. Within a demonstrably ill population
these are acceptable numbers. Now imagine administration of the
same test on a nominally well population. While missing 10% of an
at-risk population may still be deemed acceptable, a 10% false-
positive rate does considerable harm. The logic is inescapable. For
any disease the relative incidence is a small fraction of the total
population. Therefore in almost all cases a high false-positive
value for any given test incorrectly identifies an at-risk population
that far exceeds the true incidence of the disease itself. Current
examples already exist for breast and prostate cancers [17] in which,
18 Ray C. Perkins

for breast cancer, nine out of ten identified as at-risk will not have
the disease. Unfortunately, in addition to the considerable emo-
tional trauma induced by such errors, many women are unnecessar-
ily subject to both invasive biopsies and exposure to hard radiation
(ironically a procedure that increases risk of cancer). The bottom
line is that the bar for Predictive Diagnostics must be set orders of
magnitude higher than for Clinical Diagnostics. Unfortunately, in
today’s “wild west” practices, the reverse is true [16].

5 Functional Proteomics Utility: Drug Discovery

The productivity (number of new molecular entities per billion

dollars invested) of the pharmaceutical industry has been in steep
decline [18] for over half a century (Fig. 5a), leveling to less than
10% of its productivity in 1975 [19]. While this has had little

a b
100

New Pharmaceuticals
NMEs per Dollar (normalized to 1970-75)

Productivity Decline
Drug Discovery
“Chain of Translatability”

Assay

Biology Disease

0
1975 1980 1985 1990 1995 2000
Five-Year Rolling Average

Fig. 5 (a) The efficiency of drug discovery research has plummeted to less than 10% of what it had been in
1975 (data adapted from reference 19). While many factors are at play, a significant shift in approach
coincides with this period of decline: Target-based Drug Discovery (TDD). TDD is largely gene-centric,
selecting “target” proteins based on gene mutations—a model that has failed. A resurgence of more
phenotypic approaches is occurring, a resurgence highly intertwined with expanded characterization of the
Functional Proteome. (b) A succinct summary of Phenotypic Drug Discovery approaches that emphasizes the
fundamental importance of the Assay. The “Chain of Translatability” (reference 22) implies that the Assay
must reflect and inform Biology and Disease pathology
 Making the Case for Functional Proteomics 19

influence on industry revenues the impact on healthcare has been

predictably negative. No small part of this decline relates to a
fundamental shift in the drug discovery paradigm from a pheno-
typic to a target-based approach. Historically, drug discovery fol-
lowed a phenotypic approach in a protracted but largely successful
effort to curb infectious disease. Would-be-drug molecules were
tested for their ability to kill pathogen cells. Successful cell-killers
were then tested in infected animals and molecules that were effica-
cious and safe in animals were subsequently tested in human
patients. The success of those efforts, along with development
and widespread use of vaccines, led to increased life span which,
in turn, forced a shift in focus to diseases largely manifest with
aging: cardiovascular disease, cancer, and neurodegenerative dis-
ease. However, translation of phenotypic drug discovery (PDD)
methods proved elusive as cell and animal models for chronic
disease did not exist. The stage was set for testing alternative
discovery approaches and in less than a decade PDD declined to a
small fraction of its former use. With equal doses of reductionist
(and wishful) thinking and improvements in molecular evaluation
technologies, the drug discovery paradigm shifted to a target-based
drug discovery (TDD) approach. The thinking was (and is) that
detrimental behavior of individual proteins lay at the heart of
disease and that the detrimental behavior could be modified
through “rational drug design.” Some early success with ACE
(angiotensin-converting enzyme) inhibitors hastened the transition
from PDD to TDD. The more pervasive influence, however, in the
switch from PDD to TDD was predicated on the possibility of full
genome sequencing. According to the model, knowledge of the
genome, specifically a single disease-related gene mutation, serves
to identify the target protein whose activity could be modulated
into a more acceptable state. The underpinnings of TDD rested
(and rests) on the direct association of single gene mutations as
origins for disease. Unfortunately the single gene mutation/single
protein target model was generalized from experience with rare
diseases, generally familial in origin and for which a Companion
Diagnostic is obvious and effective. The reality for the majority of
diseases, now decades following this errant generalization, is much
different and much more complicated. A recently published meta-
analysis is instructive [20].
Genomic data for 23 different cancers determined that
164,000 single-nucleotide polymorphisms (SNPs) exist—per can-
cer type. Attempts to functionally categorize the SNPs into diag-
nostically meaningful patterns were largely unsuccessful. Similarly, a
host of apparent genetic mutations are catalogued for neurodegen-
erative and cardiovascular disease, none of which are definitive. The
expectations for genomics as “a shop manual that describes with
incredible precision how to build every cell in the human body” will
not be met. By derivation, TDD has lost both its target source and
20 Ray C. Perkins

primary rationale. Not surprisingly a resurgence of interest in

returning to phenotypic drug discovery (PDD) has occurred,
though TDD continues to dominate the pharma pipeline and
much of funded research. Nevertheless, analysis of PDD outcomes
in the clinic, despite low rates of PDD usage, is promising.
A 2011 survey [21] of drugs approved by the FDA between
1999 and 2008 indicated that phenotypic discovery methods were
superior to target-based methods by 60% for “First-in-Class”
drugs. The proportion was strongly reversed for “Follower”
drugs. Noting the timeline, this is particularly impressive as the
number of PDD-origin programs was vastly exceeded by those of
TDD origin. Since the time of this report emphasis on developing
and employing PDD has accelerated rapidly. Much, of course,
remains to be done particularly for chronic diseases for which
animal and cell models have proven elusive. However, it is appro-
priate to point out the essential compatibility of PDD with Func-
tional Proteomics [22] within a Systems Biology context
(underlining added for emphasis):
“Here, we propose the term chain of translatability to describe the presence of a
shared mechanistic basis for the disease model, the assay readout and the biology
of the disease in humans, as a framework for developing phenotypic screening
assays with a greater likelihood of having strong predictive validity.”

The concept of “Chain of Translatability” (Fig. 5b) intimately

connects PDD to Systems Biology and thus to Functional Proteo-
mics. This is reinforced by realization that the analytical tools used
in Functional Proteomics overlap with those used in PDD. The
same is true for assay readouts. Thus, as was noted in 2004 [10], the
core objectives of PDD and Functional Proteomics are, within a
Systems Biology context, identical. Success in one is success in the
other and the same is true for deficiencies and dead ends. To meet
such a challenging and potentially beneficial goal, PDD efforts
must extend into multiple arenas concurrently.
In keeping with historical success, most PDD efforts focus on
activity at the Cellular level—note that Cellular is midway in the
complexity hierarchy of Systems Biology. Historical precedent also
drives a prominent use of stains/dyes in many, though not all,
assays. One such example [23] is colorfully known as “. . . Cell
Painting, which is a morphological profiling assay that multiplexes
six fluorescent dyes, imaged in five channels, to reveal eight broadly
relevant cellular components or organelles. . .” And “. . . automated
image analysis software identifies individual cells and measures
1,500 morphological features. . . .” “Cell Painting” is, therefore,
a data-rich, “high-content” foundation for evaluating cellular mod-
ification by any number of biological or pharmacological modula-
tors. Further, in keeping with theme of this book, the dyes
(presumably) bind selected proteins and, thus, the readout reflects
both the identity and locations of those proteins. The driving force
 Making the Case for Functional Proteomics 21

of the assay is protein functionality. Still, the question naturally

arises as to how such assays fit into the translatability triad of
Assay-Disease-Biology. As of this writing a dominant “solution”
for drug discovery is comparison of the “high-content” cellular
patterns created by test library compounds to the patterns created
by a reference library of molecules with “known” mechanisms of
biological action. A “match” between a test molecule and reference
molecule assumes a match in mechanism of action of the test
molecule to the reference molecule. The connection, then, between
the Assay and Biology (and Disease) is dependent on the reliability
and suitability of the reference library, selection of the cell system,
and the likely perturbations induced on cell activity by the dyes.
Cell Painting and related approaches are powerful differentiators of
members of a test library—but differentiation may or may not
imply correct mechanism selection. Still, within the context of this
book and chapter, it must be emphasized that the core responses of
Cell Painting are driven by Functional Proteomics.
The cited approach is but one of hundreds of newly developing
assay protocols in support of PDD. In concert are exciting devel-
opments in managing cells in increasingly biological structures.
Developments also continue in isolation of an individual’s stem
cells and their conversion [24] to diverse organ-compatible cells.
In this vein, enormous resources are being poured into so-called
organs-on-a-chip [25]. These all augur well for relevant and trans-
latable test beds for PDD. Even so, the predilection for starting in
the middle of the Systems Biology complexity hierarchy—ignoring
Molecules and Networks—is a mindset that must be reconsidered.
A case in point arises out of the author’s personal experience.
As cited above in the segment on Systems Biology, the author
and colleagues engaged in successful library screenings to gauge
modulation of the interaction between a pathology-related peptide
and disease-risk-related protein [11]. Results demonstrated signifi-
cant activity by members of the test library with 40% of the mole-
cules expressing activity. However, in contrast to a simple binary
interaction model, four distinct mechanisms of action were mani-
fest, with some molecules apparently acting on the peptide alone.
In the course of post-screening follow-up it was learned that the
assay response for the peptide alone included contributions from
not one but four distinct peptide entities—monomer plus three
soluble oligomers. While assignment of these oligomers is still
being made, it is known that they appear in variable concentrations
as a function of total peptide concentration, concentrations selected
to be within physiological ranges. Further, having adjusted data
analysis procedures to visualize the response from the oligomers,
the original protein partner was introduced. The data appear to
clearly indicate two phenomena of both biological and pathological
relevance: (1) Oligomeric peptides constitute the protein-
interacting form. (2) Isoforms of the protein, one disease-related
22 Ray C. Perkins

and one not, exhibit selective affinity for different peptide oligo-
mers. Work is underway to reanalyze the library screening in accord
with this new knowledge. Had the original screening work begun at
the Cellular level, the Molecular activity would have been missed.
More to the point of PDD practices, any impact of test molecules
on the peptide screening readout would be misinterpreted and
assignment of mechanisms of action compromised.
This example is telling within both PDD and Functional Pro-
teomics contexts. Translatability to both biology and disease is
established through choice of test materials, the peptide and pro-
tein. In a direct sense, those crucial connections define the assay.
The assay itself directly tested for protein functionality, namely
interaction of a peptide and protein. Once executed, the assay
further expanded knowledge of the behavior of the peptide and
connected to disease via differential and entity-selective binding to
the protein. Finally as was mentioned above, the assay was immedi-
ately extended into assessments of modulation of protein networks
and ex vivo application of the selfsame assay is in planning.
What, then, is the future of phenotypic drug discovery within
the multiple contexts of Functional Proteomics, Systems Biology
and Translatability? First, starting in the middle of the Systems
Biology complexity hierarchy—Cells—may be a high-risk decision.
Until and unless cell systems are developed that mimic chronic
disease, the classic starting point for pathogenic PDD is problem-
atic. By contrast, most diseases have identifiable relationship to one
or more proteins so that a PDD development program can launch
with the least complexity but still in a disease-relevant mode. As a
bonus, molecular level screening is both faster and less expensive.
Next, as with the example provided, extending a molecular study
into network or pathway assays is, in principle if not in practice,
straightforward. The combination of molecular and network find-
ings will predictably inform both biology/disease and assay
planning for cells. Extension, then, into original source material
(patient-derived cells, fluids, extracts, etc.) is both abbreviated and
as fully informed as can be made possible. A realistic bonus is the
acquisition of knowledge that informs likely avenues for either a
disease or companion diagnostic. Finally, the example cited points
to yet another desirable outcome attainable through phenotypic
approaches—selection of test molecules for their multiple or “poly-
pharmacological” activities [26].
Within the complexity of any organism no molecule exhibits a
single activity at all time points. The very functional attribute of any
molecule that defines one activity, e.g., inhibition of a protein
interaction, statistically matches to multiple, comparable sites on
other proteins, cell, or nucleotides. Further, other attributes of the
same molecule enable its ability to engage in additional activities.
Thus multiple activities or functionalities are manifest by a single
test-library molecule (or any molecule). Aspirin is perhaps the
 Making the Case for Functional Proteomics 23

poster child for multiple, “mechanisms of action” combined with

polypharmacological applications. Parallel multiple manifestations
of disease also exist. At an organism level, disease is differentially
manifest by combinations of symptoms. Aches and pains may or
may not be accompanied by fever, fever may or may not be accom-
panied by intestinal distress, and all the above may or may not be
accompanied by fatigue. At a molecular level, disease diagnosis
derives from comparisons of multiple tests implying multiple man-
ifestations of a single disease at a molecular level. Connecting the
two realities—multiple activities of molecules and multiple mani-
festations of disease—combine in the concept of test molecule
selection based on its polypharmacological activity. By its very
conceptual strictures, target-based assessment is a non sequitur.
Given its broader field of view, PDD more correctly and directly
addresses multiple attributes simultaneously—and those multiple
attributes are accessible through the Functional Proteome.
In closing this section on drug discovery, a watchword of sorts
is forthcoming from the composite of this segment and the seg-
ment on Diagnostics: A properly designed assay fills in gaps in the
collective understanding of biology and disease, thereby supporting
co-development of new therapeutics and diagnostics. After all, both
reflect properties of the same Organism.

6 Functional Proteomics Utility: Precision Medicine

Matching treatment to patient is the goal of Precision Medicine

(PM) [27]. As many have observed, PM has long been in practice.
Consider the cyclical relationship between disease diagnosis, ther-
apy selection, and disease prognosis (Fig. 6b). A patient presents
with apparent illness that may be attended by measurable attributes
such as elevated temperature, blood pressure, or pulse rate. Upon
questioning by the physician or nurse, other symptoms can often
provide a working diagnosis. At this point multiple therapies may
be available. Selecting the most appropriate is dependent not only
on the diagnosis but also patient or family history with certain
drugs and whether the patient is currently taking other medica-
tions. From this information a therapy, often a drug, is selected and
the patient begins the recommended regimen. Disease progress is
noted as the regimen proceeds, usually by the patient. If successful,
i.e., symptoms are reduced or eliminated, the process ends except
for any recovery from the therapy itself. If unsuccessful or if the
patient reacts poorly to the selected therapy, an alternate therapy
may be selected and the cycle continues. In the event that no
therapy is successful the possibility of a more serious disease is
considered, and expanded diagnostic procedures are brought into
play. It should be noted that a full slate of Companion Diagnostics,
a slate that includes the existing pharmacopeia, eliminates much of
24 Ray C. Perkins

a 42 Imprecise Medicine: b
Effectiveness of Common Traditional
Medications Precision Medicine Cycle

"Number Needed to Treat" (to Diagnosis Prognosis

"Work" for One Patient)
Drug Effectiveness, Per Cent
of Total Population
25 Therapy
23
20 20

4 5 5
4
2

Aspirin (1200 mg) Aspirin (650 mg) Nexium Crestor Advair Diskus

Fig. 6 (a) The drive for precision (meaning individualized) medicine is clear when looking at “Number Needed
to Treat” figures for common medications. 1200 mg of Aspirin (equivalent to 2 extra-strength tablets or
capsules) is only effective for 42% of the American population due to variations in personal biochemistry. For
other popular medications a standard dose is effective for 5% or less. Knowing whether or not the patient is
one of the lucky 5% is critically important. (b) The cycle of precision medicine has been practiced for
millennia: Diagnose the patient. Determine what would be the most effective treatment and administer
it. (Hope fervently that the patient suffers no serious ill effects.) Check the patient’s progress and adjust
treatment, or even diagnosis, according to the results. With the development of Companion Diagnostics there
is the hope that the cycle can be made more efficient and minimize ill effects from inappropriate treatments

the treatment cycle. The benefits—reduced patient exposure to

multiple drugs with multiple side effects, lower consumption of
drugs, and the need for fewer visits to the clinic—could be
considerable.
Precision Medicine as envisioned and heavily funded interna-
tionally in the twenty-first century is a more intensive and, at this
point, genome-centered process. Before expanding on that process
and “Making the Case” for Functional Proteomics, the relationship
between the existing pharmacopeia and disease efficacy must be
examined.
It is well known by medical professionals that not all drugs
work for all people. This was acknowledged in the introduction to
this section in the common practice of examination of family his-
tory and current drug usage. What is not so well known outside of
the profession is the degree to which existing drugs don’t work.
 Making the Case for Functional Proteomics 25

Consider available medications for pain relief, some available over-

the-counter (OTC) and some via prescription (Fig. 6a). For OTC
medications most people find relief from one but not another
medication, or potential side effects of a medication affect some
people but not others. The personally observed combination of
pain relief and toleration of side effects define an individual’s selec-
tion for pain relief. Studies have been made on the incidence of use
of pain relievers, one such study being the being “The Oxford
League Table of Analgesic Efficacy” [28]. The relevant category
heading is “NNT” that stands for “number needed to treat.” If, for
example, the NNT is 2.4, then for every 2.4 people who take the
medication, it only works for one person. This is the NNT for a
most common pain medication, aspirin, at a 1200 milligram
(mg) dosage (a common OTC tablet in the USA contains
325 mg). Another way to approach the NNT for a therapy is
calculation of the fraction of the population effectively served by
the medication. Aspirin, for example, at 1200 mg only works for
around 42% of the population and only for 23% at 650 mg—two
OTC tablets. These numbers seem high, but are actually consistent
with a “good” drug. Examination of heavily prescribed [29] drugs
such as Nexium®, Crestor®, and Advair Diskus® have NNTs of
25, 20, and 20, respectively, or only 4–5% of the diagnosed popu-
lation. This presentation establishes a broader context for the
twenty-first century discussion of PM, namely one in which existing
medications are prescribed more rationally—or not at all.
Precision Medicine as touted by pharmaceutical companies,
thousands of small companies and government funding agencies
is intimately tied to highly publicized “successes” particularly in
treatment of cancer. The presence of a particular gene mutation for
patients diagnosed and previously treated for a particular cancer is
selected for yet another treatment, one “targeted” at the protein
expressed by that mutation. On average about a third of the
selected population respond, with measureable declines in tumor
size for an average of 3–8 months. At that time the cancer starts up
again. Calculation of the NNT is bit tricky but, at best, is around
three assuming that a 3–8 month relief for six-figure treatment
costs is considered to have “worked.” Detailed discussions on
whether PM is proving successful within this context are available
[16, 30, 31]. The purpose of the exercise here provides background
for the true potential of PM if fully and properly implemented.
Full implementation of PM—matching therapy to patient for
the whole of the pharmacopeia—requires knowledge of both the
disease as manifest in the patient and the manifestation of therapeu-
tic activity in the patient. A multitude of questions related to the
functioning of the proteome arises, most not yet on any funding
sources’ view screen. How, e.g., does the Functional Proteome vary
within a person over time? Or in relation to the wake/sleep cycle?
Or in response to pathogens? Or as part of the development of
26 Ray C. Perkins

chronic disease? Comparable questions can be posed for response

to a medication. They can also be asked within the context of a
single patient sitting in the consulting room awaiting diagnosis.
The conclusion by all who consider the application of Functional
Proteomics to PM is simple and telling: The necessary background
knowledge does not exist. Why this is the case is abundantly appar-
ent—a lack of focus in the face of hundreds of billions of dollars
invested in sequencing the genome. For now, and in the foreseeable
future, the overwhelming potential of Functional Proteomics in the
rationalization of therapy selection is an untapped potential
[9]. This thought provides a natural transition to a clear-eyed
view of the Functional Proteomics knowledge base.

7 Functional Proteome Knowledge Base

Earlier chapter segments have established the need for refocusing

attention on Functional Proteomics. The Functional Proteome is
shown to be two orders of magnitude larger than the human
genome. This is a conservative estimate as ubiquitous events that
modulate protein activity and functionality, e.g., cofactor binding,
are ignored, as is the significant but underexplored contribution of
the microbiome to the Functional Proteome. Functional Proteo-
mics encompasses all activity associated with genetics including
transcription, translation, epigenetics, and the somewhat
ill-defined “gene expression.” Functional Proteomics is fertile
ground for development of Biomarkers that lead to Diagnostics/
Prognostics while simultaneously contributing new knowledge in
both biology and disease. For comparable reasons, Functional Pro-
teomics within a Systems Biology context is the natural paradigm
for twenty-first century Drug Discovery: Phenotypic Drug Discov-
ery. However, given the great potential of Functional Proteomics,
completeness demands that two pragmatic questions be addressed:
1. How complete is the existing Functional Proteomics’ knowl-
edge base?
2. How effectively do existing methods contribute to that knowl-
edge base?
Without the answers to these questions, efforts cannot be
prioritized nor can a realistic strategy be developed. This chapter
segment addresses both these key questions and ends with a prior-
itized roadmap to a highly desirable endpoint: a curated Functional
Proteome and a full complement of robust tools to achieve that
endpoint. Evaluating today’s knowledge base relative to the desired
endpoint is begun through comparison of efforts such as the
“Human Proteome Project” (HPP) [32] to the data presented in
Fig. 1a.
 Making the Case for Functional Proteomics 27

An ongoing goal of the HPP is creation of an organism-scale

map of the human proteome, and the necessary inclusion of alter-
native expression, PTMs and protein interactions was recognized
from the outset. However, data collected to date is gene-centric by
conscious choice. Thus the current proteome knowledge base is
largely defined by the smallest column in Fig. 1a representing some
19,000 distinct proteins. None would argue that such a knowledge
base is complete or adequate but concerted efforts to expand that
knowledge base are not being made. Therefore the databases upon
which protein identification efforts rely restrict analysis to a highly
limited, gene-centric misrepresentation of the proteome. Identifi-
cation of proteins in a mixture that relies on gene-centric databases
is almost certainly incorrect (as was argued above, gene-expressed
proteins may not exist as active, functional entities). A last, ironic
point must be made: the individual proteins tested to define the
human proteome are not gene-expressed products at all, but rather
are human-gene-sequence-compatible proteins subject to PTM
processes active in the host cell from which the proteins were
expressed. Despite the apparent negativity, this rendition of the
proteomic knowledge base is the outcome expected from a gene-
centric approach: an exchange of genome coverage at the expense
of specificity.
The existing knowledge base, by design, relates the proteome
to an admittedly incomplete reference set of data, namely the index
provided by genes alone. However, that incompleteness, or lack of
specificity, is progressively remedied through logical expansion of
the reference set, an expansion that can take many forms. One of
those forms is the ongoing, and never abandoned, protein-centric
work that examines relatively simple protein systems but with a high
degree of specificity. Thus the proteomic knowledge base expands
in a reliable though piecemeal manner. Further, combining the two
approaches—gene-centric/bottom-up/shotgun [33] and protein-
centric/top-down/rifle [34]—can and is being used with great
effectiveness. A case in point is the development of proteomic-
based biomarkers discussed above [15]. Bottom-up approaches
can provide, e.g., robust disease differentiation that lays the basis
for subsequent highly specific selection of proteins that define that
differentiation. This theme, the duality of coverage and specificity,
is consistent for the whole of the Functional Proteomics knowledge
base, encompassing not just the knowledge itself but also the
methods by which that knowledge is derived.
An ideal Uniprot profile, e.g., for any given protein would
include curated data on distribution of PTMs, structures for all
possible isoforms and conformers, protein distribution at both
tissue and cellular levels, concentrations for all possible locales,
interactions with other molecules of any type within those locales,
and . . . the list goes on. Such an idealized listing is unlikely nor is it
truly required, but the breadth of proteomic attributes does
28 Ray C. Perkins

Table 2
Proteomic properties, methods, and proteome coverage

Methods’ Coverage Proteome

Proteomic property Method examples reliability potential coverage to date
Identity Mass spectrometry Good Full Limited
Antibody Fair Limited
Structure X-ray Good Limited Poor
Nuclear magnetic resonance Good Limited
Quantitation Mass spectrometry Fair-good Full Poor
Antibody Fair Limited
Localization Immunofluorescence Fair Limited Poor
Fluorescent-protein tagging Fair-good Limited
Activity Yeast two-hybrid Poor Extremely limited Poor
Fusion Poor
Surface affinity Poor
Our knowledge of proteomics is only as accurate and as deep as our data. Summarized here are the five basic protein
properties. For each property is listed the most common methods used for measurement, graded in terms of the accuracy
and precision of individual measurements (Methods’ reliability), and their applicability to all proteins and protein
interactions (Coverage potential). The final column indicates the degree to which any given property has been evaluated
for the proteome (Proteome coverage to date). The lack of Functional Proteome Coverage To Date is the most serious
impediment for advancing the frontiers in biomarker discovery and development of patient-specific therapies

establish the context for proteomic research and application of that

research. Further, the list may be practically categorized in such a
manner as to maintain generality while establishing a basic set of
Functional Proteome attributes. It is against this basic set that the
status of the Functional Proteomics knowledge base is appraised in
parallel with the ability of existing methods to assess the key metrics
of: Identity, Structure, Quantitation, Localization, and Activity (see
Table 2).

7.1 Proteomics Table 2 cross indexes key Functional Proteomics metrics to: Meth-
Methods Appraisal ods employed to assess the metric (Method Examples), the reliabil-
ity of a given measurement by the Method (Methods’ Reliability),
the potential for the Method to address the whole of the Functional
Proteome (Coverage Potential) for the Proteomic Property, and
the extent to which all existing knowledge addresses the whole of
the Functional Proteome for a given Proteomic Property. A couple
of examples are helpful. The structural Method, X-ray, provides
exquisitely detailed spatial information and is, therefore, a Reliable
method. However, its ability to Cover the entire proteome is
Limited by the inability to crystallize every protein in the proteome.
By comparison, the Activity method for protein interactions, Yeast
Two-Hybrid, is prone to a high incidence of false-positive and false-
 Making the Case for Functional Proteomics 29

negative results (discussed in more detail below). Its Reliability is,

therefore, poor and its ability to Cover the proteome is extremely
limited.
The low assessments for existing Proteome Coverage derive
from a combination of Methods limitations and lack of concerted
effort. Summaries of Methods and their ability to contribute to
fundamental protein Properties are provided immediately below.

7.2 Identity The majority of “modern” efforts in proteomics has been and is
focused on protein identification, especially for mixtures of pro-
teins. Dominant in the effort is a variety of separation procedures
(“Sep-Sci,” gel electrophoresis, liquid chromatography, etc.) and
analysis by mass spectrometry (MS). Indeed the use of MS is often
(and wrongly) equated to proteomics and vice versa. Other identi-
fication methods include Edman sequencing, quantitative amino
acid analysis, and antibody-based analysis (e.g., enzyme-linked
immunosorbent assay or ELISA). Sep-Sci alone can be used for
rough identification though molecular weight resolution is typi-
cally insufficient to differentiate closely spaced isoforms or the
extent or kind of posttranslational modification. MS has somewhat
similar issues for some approaches though not for others. So-called,
“bottom-up” methods [33], typically applied to mixtures of pro-
teins, rely heavily on comparison of post-digestion sequence pat-
terns to databases of those sequences for individual proteins. Two
issues must be, and are, acknowledged: (1) Algorithm-driven ana-
lyses can only assess identification in accord with members of the
database (heavily weighted to proteins expressed by cDNA) and
(2) “Sequence coverage” for any given identified protein can range
from 30% to 99%, the range often reflecting relative concentrations
of proteins in the mixture. In contrast to bottom-up approaches,
several “top-down” MS approaches [34] are also available. As the
name implies, whole proteins comprise the test sample. Top-down
approaches can resolve closely spaced isoforms and can inform the
degree and kind of PTM. As with bottom-up, however, if algorith-
mic comparison to existing databases is employed, those analyses
are only trustworthy if the tested protein(s), including PTMs, is
already in the reference database. Further, Top-down methods are
typically applied to mixtures of fewer proteins. The bottom line for
Sep-Sci/MS is that identification results are reasonable within the
stated qualifiers. Confirmation, if needed, can be supplied by
“orthogonal” methods such as Edman Sequencing. Then and
only then can a non-qualified protein identification be reasonably
certain.
In sharp contrast to instrument-based methods such as MS,
antibody/antigen binding can be exploited to identify proteins
[35]. Generally, antibodies are prepared and isolated for single-
protein antigens. Subsequent binding to the antibody by a protein
from a mixture of proteins is taken as evidence that the protein is
30 Ray C. Perkins

the original protein/antigen. Inherently, the relationship between

antibody and protein/antigen is a one-to-one—a single-protein
assay. However, multiple antibodies can and have been
incorporated into array panels with the net result that mixtures of
proteins may be examined and antibody-selected proteins identi-
fied. In principle antibody-based assays and Sep-Sci/MS assays
serve as orthogonal approaches for protein identification. In prac-
tice, both are seldom used in a single report and, in the few
instances of rough comparability, the two approaches do not fully
reinforce each other. Thus, it should not be assumed that Sep-Sci/
MS and antibody-based identification are orthogonal methods. At
the heart of all antibody-based work is the inherent fact that no
antibody can be proven to be selective for a single antigen.
Attempts to prove selectivity are, in the final analysis, attempts to
prove a negative. On a practical note, antibody promiscuity should
be assumed.
What’s needed? The knowledge base for Identification is incon-
sistent, detailed for some proteins, nonexistent for others. Tissue
distribution is a particular concern—one that must be addressed for
even a rudimentary foundation for Systems Biology. A worldwide
effort would be welcome. As for methods, two issues for MS
methods stand out: (1) Expansion of the reference bases to more
closely represent the actual proteome (inclusion of PTM’s, e.g.)
and (2) Resolution of the dependence on digestion methods for
bottom-up approaches. For antibody work the overriding issue is
selectivity and, as was noted above, proving absolute selectivity is
not possible. However, demonstrating non-selectivity for key data
is perhaps tenable. Reports from either major approach should not
be considered definitive, but rather as direction indicators for more
targeted methods.

7.3 Quantitation Hand in hand with the need for protein identification is determina-
tion of protein concentration. The list of applicable methods for
quantitation is comparable to those for identification, though exe-
cution differs. In the case of antibody-based work [35], addition of
a reporter group to the antibody, e.g., as is done in preparation of
ELISA assays, identification and quantitation occur in a single assay.
For MS-based work [36], a variety of labeling approaches provide
information on the relative concentrations of proteins in a mixture.
These methods coupled to carefully prepared control mixtures can
serve to provide some indication of absolute protein concentrations
in the original source material. However, for optimum determina-
tion of absolute protein concentration coupled to dramatically
improved detection of low-concentration proteins, application of
so-called “Multiple Reaction Monitoring (MRM)” [37] is the
approach of choice. It must be noted that MRM demands intensive
method development and absolute concentrations are available
only at the expense of “coverage.”
 Making the Case for Functional Proteomics 31

What’s needed? The knowledge base for Quantitation is incon-

sistent given its dependence on Identification as a prerequisite. A
way forward is execution of MS labeling experiments which provide
relative concentrations in concert with a select subproteome ana-
lyzed by MRM. Absolute concentrations from MRM provide, in
principal, the appropriate adjustments to relative concentrations
from the broader coverage of MS labeling results.

7.4 Structure Structure and function are intimately related and only two methods
dominate determination of protein structures: X-ray crystallogra-
phy and “solutions” nuclear magnetic resonance spectroscopy
(NMR) [38]. Of these two, X-ray structures listed with the Protein
Data Bank (PDB) outnumber NMR structures listed by 10:1.
Combined, the two methods account for 98.4% of all structures
archived (cryo-Electron Microscopy, the object of a 2017 Nobel
prize, accounts for only 1.4% of PDB structures) [39]. The quality
of data forthcoming from these methods is outstanding—the issue
is coverage of the proteome.
As its name implies, X-ray crystallography demands samples
with significant, through-space uniformity. Coaxing a molecule
into such a form is a combination of science, method, and art,
and many proteins simply are not amenable to such forced unifor-
mity. This fact limits the range of proteins that can be studied by
X-ray. Coupled to this limitation is the inverse implication of cov-
erage: proteins forced into an extended, uniform state almost cer-
tainly do not represent the multiplicity of states of proteins in
solution. In the best of cases the crystallized structure may be
deemed a dominant or active conformer, in the worst the structure
may reflect no biologically active state. Solutions NMR largely
avoids this potential pitfall but has limitations of its own.
Samples for solutions NMR range in concentration from 0.1 to
5 mM [40] and, for the overwhelming majority of protein work,
contain proteins smaller than 50 kDa (slightly smaller than the
average for human proteins (53 kDa) and slightly larger than the
median (42 kDa)). For the most-practiced NMR methods as
applied to proteins, half the human proteome is not accessible
(NMR methods do exist to virtually eliminate the molecular weight
limitation but are limited by magnet technology—and cost). For
the 50% of human proteins accessible to solutions NMR, the infor-
mation content reflects not only secondary structures but also
partially inform variation in conformational states. Further, impacts
on structure by solution variation and small molecule effectors may
be assessed. Here, though, the second limitation of NMR, sensitiv-
ity/concentration, comes into play. With a preferred lower limit of
2 mM (1 mM ¼ 10 mg/mL for a 10 kDa protein), NMR sample
concentrations exceed the observed physiological concentrations of
nearly all proteins. Further, a typical sample size of slightly over
0.5 mL, samples can be costly, a consideration that is amplified by
32 Ray C. Perkins

the need to express proteins with local or large-scale isotopic sub-

stitution of one or more elements (C, H and N).
The “dark horse” for protein structure work may well be the
peptide-protein interaction methods cited by the author in this
chapter: electron paramagnetic resonance spectroscopy (EPR) cou-
pled to spin labeling [41, 42]. Not only are interactions discernible,
but intra-protein structure and individual amino acid mobility may
be determined.
What’s needed? The apparent solution to expanding the col-
lective methods’ capabilities is a wholly new method. Crystalliza-
tion of proteins is a limiting factor for X-ray just as molecular
weight is limiting factor for NMR. More widespread application
of, e.g., cryo-Electron Microscopy may pick up some of the slack.
As has been mentioned, EPR/spin labeling is an underexploited
method, one that avoids the limitations of both X-ray and NMR.
Improvement of proteome coverage can only be made with
increased support.

7.5 Localization Two widely divergent methodologies must be considered for asses-
sing the status of protein localization: fluid-based and cell-based.
Biological systems naturally afford access to extracellular fluid,
examples being blood plasma, lymph fluid, cerebral spinal fluid,
and interstitial fluid. For these cases methods such as those dis-
cussed under Identification and Quantitation apply. Fluid-based
approaches can also be applied for selective isolation and lysis of
subcellular organelles. Thus, within the qualifiers noted above,
fluid-based localization studies are hypothetically amenable to
examination. Cell-based localization of proteins demands entirely
different approaches, exemplars being Immunofluorescence (IF)
and fluorescent-protein tagging (FP) [43].
Distinct proteins are selected for either direct modification
required for detection or as the antigen for protein-specific, fluo-
rescently labeled antibodies. Images of cells with incorporated
monitors reveal the localization of the labeled entities under a
variety of conditions of interest. For either labeling approach cer-
tain cautionary flags are raised. Extracellular introduction of labeled
proteins, whether FP or IF, cannot insure bio-relevant distribution,
especially to intracellular organelles. Next, as with all labeling
approaches the label itself may significantly alter behavior. In the
specific case of IF, lack of antibody selectivity produces false-
positive results. Given that they are by design active agents, anti-
bodies may also shift equilibria in the direction of the captured
antigen, thus perturbing the very system under observation. The
impact of these concerns is quantified in a side-by-side study of
506 target proteins [43].
Comparative results for the two labeling methods were cate-
gorized as identical, similar, or dissimilar among different subcellu-
lar locales. Identical results ranged from 15 to 70% among those
 Making the Case for Functional Proteomics 33

locales, with an average of approximately 40% (number estimated

from visual inspection of graphic material). Identical plus similar
results (similar implying overlap in localization of the two methods)
yield ranges of 75–95%, and dissimilar results were observed for
10–25% of the tested proteins. While the sum of identical and
similar results is encouraging, the relative number of both identical
and dissimilar results is cautionary. Though not a conclusion of the
study it’s clear that the researchers favor labeled protein (FP) results
over antibody localization (IF). For example, 37 (8%) of the anti-
bodies gave no staining results even though RNA sequencing
argued for the presence of those proteins. On the false-positive
side, researchers suspected cross-reactivity of antibodies when tar-
get proteins existed in low concentrations.
What’s needed? Localization of extracellular proteins is
obtained via methods described above for Identification and Quan-
titation. Methods for localizing proteins in viable cells remain
problematic, though anecdotal data favor a labeled protein
approach over antibody detection of localization. More methods
development is required prior to wholesale implementation of any
given approach.

7.6 Activity This segment primarily deals with the most prevalent protein activ-
ity, protein interactions—not simply binary protein-protein inter-
actions but the whole of protein interactions. In accord with the
analysis in this chapter, the dominant Functional Proteomics enti-
ties (83%) are bound proteins. Both Functional Proteomics and
Systems Biology demand assessment of protein interactions in
bio-relevant contexts. This is abundantly clear for translation into
essential arenas such as Biomarker and Drug Discovery, and Preci-
sion Medicine. Effectiveness in these arenas comprises the ability to
assess interactions among proteins (binary and groups), between
proteins and polynucleotides and between proteins and membrane
surfaces. Further, assessment must include the impact on those
interactions of variations in solution conditions, and the introduc-
tion of effector molecules such as cofactors, substrates or new drug
candidates. However, as two references attest, measurement even of
protein-protein interactions is the weakest link in the methodology
chain.
In 2002 a composite of existing data was analyzed [44]. High-
lights from that study concluded that only 3% of reported interac-
tions are supported by more than one method. That low “hit” ratio
continues in a 2009 study [45] in which multiple methods agreed
on only 8% of tested protein pairs—all of which were “known”
interacting pairs. Further, from the 2009 study the best method
test on known interacting pairs of proteins missed a full two-thirds
of the interactions—a false-negative value of 66%. Such outcomes
lead authors to conclude that, “. . . large datasets of protein-protein
interactions vary enormously in their error rates and there is no
34 Ray C. Perkins

simple way to compare different interaction data sets.” Other

recent reviews, noting the poor collective performance, opt for
wholly comparative analyses of, e.g., homology, or in silico predic-
tions of various sorts [46]. Indeed, many university courses take
this approach, virtually eschewing the particulars of actual measure-
ment as a confirmation of protein sequence comparisons. There is a
circularity to such logic given that the basis for homology compari-
son is the poor performance of the methods from which the
homology comparisons arise. Functional Proteomics cannot pro-
ceed under these conditions. A brief review of existing methods is,
nonetheless, necessary.
The most employed method is yeast two-hybrid (Y2H) which
features downstream activation of, e.g., a transcriptional event
upon the interaction of two fusion proteins in the nucleus of a
yeast cell. One portion of each fusion protein includes the sequence
for one of two proteins to be tested. Since the downstream event is
only triggered by interaction of the fusion proteins, a trigger of that
event is presumed to represent interaction of the test proteins. In
the 2009 report the best Y2H performance was 25%. Direct trans-
lation of Y2H methods to essential applications such as drug dis-
covery is impossible. The majority of other existing protein
interaction measurements include either capture/fusion of an inter-
acting pair or analysis of interactions in which one test protein is
bound to a surface. Among these are tandem affinity-tagged, coim-
munoprecipitation and surface plasmon resonance. While all have
application in limited circumstances, none afford the robustness or
flexibility demanded by Functional Proteomics within a Systems
Biology paradigm [46].
As was mentioned under Structure above one serious candi-
date, though little known, is the use of EPR and spin labeling. As
demonstrated in the report on the author’s ongoing work and
elsewhere [41], the collective abilities of the method is amenable
to detection of protein interactions [47] with any interacting part-
ner and has been employed in test media of any complexity. Indeed
at least one example exists that demonstrates its potential use in
diagnosis of Duchenne Muscular Dystrophy. EPR’s range of appli-
cation, once translated, might be appreciable.
What’s needed? The glaringly obvious answer is new methods.
No one existing method stands forth as a standard by which other
methods may be assessed. Nor do methods agree sufficiently to
trust a variety of methods as checks and balances. It is time to focus
on new approaches among them being EPR and spin labeling.

7.7 Prioritized Table 2 assesses the overall knowledge base of the Functional
Roadmap Proteome as poor, and that Methods in most cases have limited
coverage potential. Any Roadmap must simultaneously focus on
accelerated data collection in areas of strength while improving
basic capabilities in areas of weakness. Areas of strength include
 Making the Case for Functional Proteomics 35

the ability to Identify and Quantify proteins. This argues for aggres-
sive support for both updating reference materials to include
PTMs, etc., and widespread data collection. This work is founda-
tional and receives high priority. Structural determinations may
proceed in parallel along with the need for new methods develop-
ment to address know limitations. Localization methods must, for
now, be deemed developmental with Methods showing great
promise but lacking in robustness and reproducibility. Proteomic
Activity, as related to detection of protein interactions, is essentially
moribund. Wholly new approaches, especially approaches that per-
mit assessment in complex media, must be developed—now.

7.8 A Cautionary As was briefly mentioned at the beginning of this chapter segment,
Lesson on the Use of recombinant proteins (the major source of laboratory proteins) are
Recombinant Proteins modified in accord with the posttranslational product of the host
cell from which the protein is expressed. Bacterial cells will produce
bacterial PTMs, e.g., and human cells will produce human PTMs.
The degree and kind of PTMs are almost never specified or, as is
likely, known by the supplier. Therefore, almost the whole of pro-
tein research that employs recombinant proteins cannot be fully
specified. The content of those little vials opened thousands of
times a day is simply not known. Researchers must insist on full
protein specification, including PTMs (and protein content which
is typically only 50% of the total mass delivered). Finally, in-house
protein sequencing must become commonplace.

8 Closing Comments

Biology is the adaptation of chemical and physical processes and

attributes in support of life. In this context, two classes of biological
molecules warrant primary attention: polypeptides and polynucleo-
tides—proteins and DNA/RNA. Two, simple statements define
the relationship between the two, essential classes of biomolecules:
Without genes there are no proteins.
Without proteins there are no genetic events.

Addressing the first biological truism, it becomes increasingly

clear that the word, “genome,” is not singular. Considerable
genetic variation exists not just among individuals but within a
single individual. It’s likely that a person’s genome(s) varies over
time as well. Addressing the second truism, few biological events
occur without mediation by proteins including epigenetics, “gene
expression,” transcription, translation and DNA repair—which
introduces a third truism:
Biology is change, all organisms change with each tick of the clock.
36 Ray C. Perkins

Proteins are agents of action, both driving change and respond-

ing to change. Clearly research must shift from the limited and
passive perspective of genomics to the systemic and active perspec-
tive of Functional Proteomics. This chapter closes with a review of
the demands of Functional Proteomics and how (or whether) those
demands can (or cannot) be met. As prelude, the potential payoff
for full characterization of the Functional Proteome is reviewed.
The literature is replete with reports of biomarkers potentially
suitable for identifying druggable targets, diagnosing disease or
selection of medication. Many are founded on the discovery of
some previously unreported gene variation. It’s a dark sort of
consolation to realize that any given report has a 50/50 chance of
being repeatable or that genomic variation is now known to be
quite common. As a result no positive diagnostic exists for many
diseases including Alzheimer’s Disease, and truly predictive tests
exist for no chronic disease. Both the focus and discipline must
change. Functional Proteomics, even in this relatively unexplored
stage, is already pointing the way forward in many areas. Even in the
lowest level of Systems Biology, namely Molecular, diagnostic can-
didates are coming forward that not only differentiate disease but
inform the nature of that disease [15]. In the forefront of drug
discovery a renewed focus on phenotypic approaches include as a
matter of course expansion of knowledge for biology and disease in
addition to simply “screening” test library molecules [22]. Even
further, assay demands include a coordinated search for reliable
Companion Diagnostics to support the highly desired goals of
Precision Medicine. If progress continues, the artificial boundaries
between biological research and translation of that research will
cease to exist. As they should. The question is, can that progress
continue?
As with all scientific endeavors, progress relies as heavily on
timing, resources, and will coupled to wisdom. The time for a
focus on Functional Proteomics is now. This statement would not
have been realistic even a few, short decades ago. Whereas access to
proteins has traditionally been limited, the breakthrough of recom-
binant DNA affords today’s researcher thousands of proteins avail-
able at the click of mouse. Standardization of the production of
polyacrylamide gels didn’t occur until 2004. The Nobel Prize that
would launch the general application of NMR determination of
protein structures would not be awarded until 1991. In 1980, only
a few dozen protein structures had been determined by X-ray. Now,
over 100,000 protein structures have been deposited in the PDB.
As with NMR, MS has only recently evolved from a small molecule
tool to one routinely used to identify multiple proteins simulta-
neously. Timing is apparently propitious, what about resources?
“Resource” in this context has at least two meanings. Staying
first with the capability theme, the analytical resources required for
Functional Proteomics is a mixed bag (see “Functional Proteome
 Making the Case for Functional Proteomics 37

Knowledge Base” above). The ability to identify and quantify pro-

teins is reasonably good. Areas that require improvement are noted
above with the most needed improvement being rationalization of
reference bases and analysis algorithms. Both demands, while con-
siderable, can be met with the other resource—funding. The pay-
off, a reference base of protein identification and quantification by
PTM and by tissue distribution, provides benefits that can’t yet be
imagined. Of course, that’s only scratching the surface of Func-
tional Proteomics. Moving on to determination of protein struc-
ture the picture is less rosy. Dominant methods, while more than
adequate in their niches, cannot readily be expanded outside those
niches. Localization of proteins in viable cells may well spring from
existing protein labeling methods, but more work is needed at the
foundational level (adding the dimension of time is likely a require-
ment). The last proteomic property examined above, activity, is the
most difficult to forecast. Existing protein interaction methods
simply are inadequate to the task and, therefore, a significant pro-
portion of the functional proteome—interacting entities—is cur-
rently out of reach. As has been mentioned earlier, new methods
must be forthcoming for both Structural and Activity assessments,
with one promising method in both arenas being EPR/spin label-
ing. Overall analytical resources are available to lay the foundation
of Functional Proteomics, while development of latter stages
require new methods development. Now for the second interpre-
tation of resources—funding.
R&D spending by pharma and agencies such as the US
National Institutes of Health (NIH) have grown tremendously
over the last half-century. Although total spending has plateaued
in the last decade, levels remain high: $100 billion USD per year, or
$1 trillion over the decade (this does not include venture invest-
ment and sales within the gene sequencing market for which total
spending is not readily available, although can be estimated at
between $500 and 1000 billion USD). This would seem to be
adequate funding for a diversity of support but such is sadly not
the case. Genomics, more specifically polynucleotide sequencing,
has by any reckoning sucked the air out of the room. In the US, for
example, would-be NIH applicants are advised that basic research
in biochemistry has zero chance of being funded. Within such an
environment, securing funds for a full-scale effort in Functional
Proteomics would seem unlikely. Nonetheless there are indicators
of optimism.
Two topics covered in this chapter are noteworthy. The early
success of the approach described under “Biomarker and Diagnos-
tic/Prognostic Discovery” above is but a harbinger of successes to
come. The unbiased assembly of network-connected proteins is the
future of diagnostics and prognostics. This optimism extends to the
forefront of drug discovery for precisely the same reasons.
The unbiased determination of test molecules’ ability to modulate
38 Ray C. Perkins

the phenotype is precisely in accord with the manifestation of

disease. In both cases the concept of “Chain of Translatability”
applies: discovery of both biomarkers and drug candidates must
simultaneously add to our knowledge of biology and disease, and
inform likely candidates for diagnostics of diverse application. This
growing shift to a phenotypic basis for biomarker and drug discov-
ery is accompanied by a multitude of new ways of handling cells and
organelles in increasingly bio-relevant settings. A reevaluation of
non-mainstream methods can open new vistas, as in the cited
detection and characterization of heretofore unseen, disease-
illuminating, functional entities. In step is continued improvement
in isolation of stem cells and their conversion to multiple cell
types—at the level of the individual. Even manufactured, patient-
derived, organ-like structures are being developed at a rapid pace.
All this progress, when coupled to a full arsenal of Functional
Proteomics tools, augurs for the twenty-first century as the century
of Biology.
The time has come and all the pieces are in place.

References
1. Pundir S, Martin M, O’Donovan C (2017) 9. Weston AD, Hood L (2004) Systems biology,
UniProt protein knowledgebase. Methods proteomics, and the future of health care:
Mol Biol 1558:41–55. https://doi.org/10. toward predictive, preventative, and persona-
1007/978-1-4939-6783-4_2 lized medicine. J Proteome Res 3:179–196
2. Savage N (2015) Proteomics: high-protein 10. Butcher EC, Berg EL, Kunkel EJ (2004) Sys-
research. Nature 527:S6. https://doi.org/10. tems biology in drug discovery. Nat Biotechnol
1038/527S6a 22:1253–1259. https://doi.org/10.1038/
3. Pennisi E (2012) ENCODE project writes nbt1017
eulogy for junk DNA. Science 337 11. Perkins RC. Paul Kenis, Deborah Berthhold &
(6099):1159–1161. https://doi.org/10. Sarah-Ellen Leonard, University of Illinois,
1126/science.337.6099.1159 Urbana/Champaign; Jonathan Lee, recently
4. Ezkurdia I, Juan D, Rodriguez JM, Frankish A, of Eli Lily; and Ray Perkins, New Liberty
Diekhans M, Harrow J, Vazquez J, Valencia A, Proteomics
Tress ML (2014) Multiple evidence strands 12. Strimbu K, Tavel JA (2010) What are biomar-
suggest that there may be as few as 19 000 kers? Curr Opin HIV AIDS 5(6):463–466
human protein-coding genes. Hum Mol 13. Berger D (1999) A brief history of medical
Genet 23(22):5866–5878 diagnosis and the birth of the clinical labora-
5. Ponomarenko EA, Poverennaya EV, Ilgisonis tory. MLO Med Lab Obs 31(7). 28–30,
EV, Pyatnitskiy MA, Kopylov AT, Zgoda VG, 32, 34–40
Lisitsa AV, Archakov AI (2016) The size of the 14. FDA (2018.) In vitro diagnostics. https://
human proteome: the width and depth. Int J www.fda.gov/MedicalDevices/ProductsandM
Anal Chem 2016:7436849 edicalProcedures/InVitroDiagnostics/default.
6. Collins FS (2006) The language of god. Fran- htm
cis S. Collins on unveiling the human genome. 15. Kuruc M (2017) Stroma liquid biopsy—bio-
Free Press, New York, p 1–3 markers of the dysregulation of the serum pro-
7. Ball P (2010) Bursting the genomics bubble. teome in cancer First presented at NJ cancer
Nature. https://www.nature.com/news/2010/ Retreat, May 25, 2017 New Brunswick, NJ
100331/full/news.2010.145.html. https:// USA. https://www.biotechsupportgroup.com/
doi.org/10.1038/news.2010.145 v/vspfiles/templates/257/pdf/NJ%20Cancer
8. Gisler M (2010) The rise and fall of the human %20Retreat%20Stroma%20Liquid%20Biopsy%
genome project. MIT Technology Review 20Poster.pdf
 Making the Case for Functional Proteomics 39

16. Lowe D (2016) In the pipeline: precision uk/booth/painpag/Acutrev/Analgesics/lftab.

oncology isn’t quite there yet Science Transla- html
tional Medicine weblog, Lowe D (2016). 29. Schork NJ (2015) Personalized medicine: time
http://blogs.sciencemag.org/pipeline/arc for one-person trials. Nature 520:609–611.
hives/2016/09/12/precision-oncology-isnt- https://doi.org/10.1038/520609a
quite-there-yet 30. Prasad V (2016) Perspective: the precision-
17. Gigerenzer G (2014) Risk savvy. Penguin oncology illusion. Nat Biotechnol 537(S63).
Group, New York, NY https://doi.org/10.1038/537S63a
18. Booth B, Zemmel R (2004) Opinion: pro- 31. Brock A, Huang S (2017) Precision oncology:
spects for productivity. Nat Rev Drug Discov between vaguely right and precisely wrong.
3:451–456. https://doi.org/10.1038/ Cancer Res. https://doi.org/10.1158/0008-
nrd1384 5472.CAN-17-0448
19. Scannell JW, Blanckley A, Boldon H, Warring- 32. HUPO (2016) The human proteome project.
ton B (2012) Diagnosing the decline in phar- https://hupo.org/human-proteome-project
maceutical R&D efficiency. Nat Rev Drug 33. Zhang Y, Fonslow BR, Shan B, Baek M-C,
Discov 11. https://doi.org/10.1038/ Yates JR (2013) Protein analysis by shotgun/
nrd3681 bottom-up proteomics. Chem Rev 113
20. Lawrence MS, Stoianov P, Polak P, Kryukov (4):2343–2394. https://doi.org/10.1021/
GV, Cibulskis K, Sivachenko A, Carter SL cr3003533
et al (2013) Mutational heterogeneity in cancer 34. Catherman AD, Skinner OS, Kelleher NL
and the search for new cancer-associated genes. (2014) Top down proteomics: facts and per-
Nature 499. https://doi.org/10.1038/ spectives. Biochem Biophys Res Commun 445
nature12213 (4):683–693. https://doi.org/10.1016/j.
21. Swinney D (2013) Phenotypic vs. target-based bbrc.2014.02.041
drug discovery for first-in-class medicines. Clin 35. Solier C, Langen H (2014) Antibody-based
Pharmacol Ther 93(4):299–301 proteomics and biomarker research—current
22. Moffat JG, Vincent F, Lee JA, Eder J, Prunotto status and limitations. Proteomics 14
M (2017) Opportunities and challenges in (6):774–783. https://doi.org/10.1002/
phenotypic drug discovery: an industry per- pmic.201300334
spective. Nat Rev Drug Discov 16:531–543. 36. Wasinger VC, Zeng M, Yau Y (2013) Current
https://doi.org/10.1038/nrd.2017.111 status and advances in quantitative proteomic
23. Bray M-A, Singh S, Han H, Davis CT, mass spectrometry. Int J Proteomics
Borgeson B, Hartland C, Kost-Alimova M, 2013:180605
Gustafsdottir SM, Gibson CC, Carpenter AE 37. Wolf-Yadlin A, Hautaniemi S, Lauffenbuger
(2016) Cell painting, a high-content image- DA, White FM (2007) Multiple reaction mon-
based assay for morphological profiling using itoring for robust quantitative proteomic anal-
multiplexed fluorescent dyes. Nat Protoc ysis of cellular signaling networks. Proc Natl
11:1757–1774. https://doi.org/10.1038/ Acad Sci U S A 104(14):5860–5865. https://
nprot.2016.105 doi.org/10.1073/pnas.0608638104
24. Avior Y, Sagi I, Benvenisty N (2016) Pluripo- 38. Berman HM, Westbrook J, Feng Z,
tent stem cells in disease modelling and drug Gilliland G, Bhat TN, Weissig H, Shindyalov
discovery. Nat Rev Mol Cell Biol 17. https:// IN, Bourne PE (2000) The protein data bank.
doi.org/10.1038/nrm.2015.27 Nucleic Acids Res 28(1):235–242
25. Esch EW, Bahinski A, Huh D (2015) Organs- 39. Wang H, Wang J (2017) How cryo-electron
on-chips at the frontiers of drug discovery. Nat microscopy and X-ray crystallography comple-
Rev Drug Discov 14(4). https://doi.org/10. ment each other. Protein Sci 26(1):32–39.
1038/nrd4539 https://doi.org/10.1002/pro.3022
26. Boran AD, Ivengar R (2010) Systems 40. MSU 900 MHz NMR sample requirements.
approaches to polypharmacology and drug dis- https://www2.chemistry.msu.edu/facilities/
covery. Curr Opin Drug Discov Devel 13 nmr/900mhz/MCSB_NMR_sample.html
(3):297–309
41. Claxton DP, Kazmier K, Mishra S, Mchaourab
27. Ashley EA (2016) Towards precision medicine. HS (2015) Navigating membrane protein
Nat Rev Genet 17. https://doi.org/10.1038/ structure, dynamics, and energy landscapes
nrg.2016.86 using spin labeling and EPR spectroscopy.
28. Bandolier (2007) The Oxford league table of Methods Enzymol 564:349–387. https://doi.
analgesic efficacy. http://www.bandolier.org. org/10.1016/bs.mie.2015.07.026
40 Ray C. Perkins

42. Yang Y, Ramelot TA, McCarrick RM, Ni S, 45. Braun P, Tasan M, Dreze M, Barrios-Rodiles-
Feldmann EA et al (2010) Combining NMR M, Lemmens I, Yu H, Sahalie JM, Murray RR,
and EPR methods for Homodimer protein Roncari L, A-Sd S, Venkatesan K, Rual J-F,
structure determination. J Am Chem Soc 132 Cusick ME, Pawson T, Hill DE, Tavernier J,
(34). https://doi.org/10.1021/ja105080h Wrana JL, Roth FP, Vidal M (2009) An experi-
43. Stadler C, Rexhepaj E, Singan VR, Murphy RF, mentally derived confidence score for binary
Pepperkok R, Uhlén M, Simpson JC, Lund- protein-protein interactions. Nat Methods 6
berg E (2013) Immunofluorescence and (1):91–97. https://doi.org/10.1038/nmeth.
fluorescent-protein tagging show high correla- 1281
tion for protein localization in mammalian 46. Rao VS, Srinivas K, Sujini GN, Kumar GNS
cells. Nat Methods 10:315–323. https://doi. (2014) Protein-protein interaction detection:
org/10.1038/nmeth.2377 methods and analysis. Int J Proteomics 2014:12.
44. von Mering C, Krause R, Snel B, Cornell M, https://doi.org/10.1155/2014/147648
Oliver SG, Fields S, Bork P (2002) Compara- 47. Klare J (2013) Site-directed spin labeling EPR
tive assessment of large-scale data sets of pro- spectroscopy in protein research. Biol Chem
tein–protein interactions. Nat Biotechnol 394(10):1281–1300. https://doi.org/10.
417:399–403 1515/hsz-2013-0155J
 Chapter 2

Methods to Monitor the Functional Subproteomes

of SERPIN Protease Inhibitors
Swapan Roy and Matthew Kuruc

Abstract
Conformational variants of the unique family of protease inhibitors annotated as SERPINs are most often
underrepresented in proteomic analyses. This limits understanding the complex regulation that this family
of proteins presents to the networks within the protease web of interactions. Using bead-based separation
provided by a family of proteomic enrichment products—notably AlbuVoid™ and AlbuSorb™, we
demonstrate their utility to satisfy investigations of serum SERPINs. We also suggest their use to develop
functional profiles of the SERPIN proteoforms, and how those can establish relationships to disease
phenotypes, gene mutations, and dysregulated mechanisms.

Key words SERPIN, SERPIN function, Functional proteomics, SERPIN mechanism, SERPIN
biomarkers, SERPIN proteoforms

1 Introduction

The balance and regulation of proteolytic activity within serum is

essential to blood based biomarker discovery and possibly to thera-
peutic intervention. Changes in blood components often reflect
acute responses to thwart external stresses, such as coagulation
when skin is severed, or inflammatory response during microbial
infections. These fast-acting responses are controlled by proteolytic
cascades, essentially modifying functionality by the controlled deg-
radation of protein structures. While necessary for acute response,
persistent activation of these proteolytic cascades can lead to
chronic conditions. So, there is a balance and regulation of these
proteolytic cascades which is necessary to keep aberrant proteolysis
controlled.
This is done through systemic regulatory protein factors, called
protease inhibitors or antiproteases. It is now quite apparent that
the influence of inhibition can be just as important as zymogen
activation in rapid switch cascades controlling subnetworks within
the protease web [1]. One such example of this web’s complexity, is

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_2, © Springer Science+Business Media, LLC, part of Springer Nature 2019

41
42 Swapan Roy and Matthew Kuruc

that one substrate (Neutrophil Elastase) for Alpha-1-Antitrypsin

(the inhibitor), can activate the inactive zymogen proMMP-2, a
metalloproteinase involved in tumor invasion and angiogenesis
[2]. So it becomes necessary to consider that inhibitors are them-
selves being regulated under different and often complex means of
regulation. Within this context, therein lies the special case of the
SERPIN superfamily of protease inhibitors.

1.1 The SERPIN The SERPIN family of suicidal serine protease inhibitors plays an
Superfamily of integral role in regulating a wide variety of biological activities, and
Suicidal Inhibitors represent 2–10% of circulating plasma proteins. SERPINs regulate
coagulation, hormone transport, complement and inflammation,
angiogenesis, and blood pressure along with many other pathways.
Among the key regulators in blood serum, SERPINA1 (also known
as ɑ1-antitrypsin) protects lung tissue from neutrophil elastase,
SERPINC1 (also known as antithrombin) controls coagulation
proteases, SERPING1 (also known as plasma C1 inhibitor) regu-
lates complement activation, and SERPINF2 (also known as ɑ-2-
antiplasmin) inhibits plasmin and regulates fibrinolysis [3, 4].
This unique family of protein inhibitors has been associated
with progression or remission of cancer and so they may become
valuable biomarkers for therapeutic or diagnostic use. Of clinical
utility, prostate-specific antigen (PSA), also known as kallikrein-3, is
commonly used as a biomarker for prostate cancer. However, the
kallikrein protease family of proteins is of very low abundance in
plasma, making observation and quantification difficult. Neverthe-
less, PSA is regulated by the SERPIN inhibitor family; in men with
prostate cancer the ratio of free (unbound) PSA to total PSA is
decreased, suggesting a greater role of inhibitory capacity in cancer.
By way of these examples, rather than focusing proteomic
discovery efforts on low-abundance proteins like the tissue kallik-
reins, it may be advantageous to profile much higher abundance
Tissue Kallikrein inhibitors like SERPINA5 (Protein C Inhibitor),
SERPINA3 (Antichymotrypsin), and SERPINA4 (Kallistatin), to
better understand underlying disease mechanisms and potentially
generate new biomarkers. However, the role of SERPINs in these
critical junctures is rarely straightforward as would be the case for
more simplistic binary binding inhibition. For functional interpre-
tation, reliance on strict abundance measurements, such as data that
might be derived by ELISA or quantitative LC-MS, does not
differentiate the subpopulations of the seemingly opposing out-
comes of the SERPIN interaction with its target protease.
This is because SERPINs differ from all other families of pro-
tease inhibitors in having a complex mechanism of action that
involves a drastic change in their shape, forming the basis of a
suicidal substrate inhibition mechanism [3, 4]. The reactive center
loop (RCL) extends out from the body of the protein and directs
binding to the target protease. The protease cleaves the SERPIN at
 Methods to Monitor Functional SERPINs 43

6(53,1SURWHDVHLQKLELWRUVPXVWEHDFFRXQWHGIRU
GLIIHUHQWO\WKDQELQDU\ELQGHUV
%LQDU\ELQGLQJHYHQWSURWHDVHLQKLELWLRQLVUHJXODWHGE\UHODWLYH
FRQFHQWUDWLRQVRIWKHUHDFWDQWVದSURWHDVHDQGSURWHDVHLQKLELWRU
DQWLSURWHDVH

(TXLOLEULXP

,QDFWLYDWHG3URWHDVH3URWHDVH )UHH3URWHDVH )UHH

,QKLELWRUFRPSOH[ ,QKLELWRU SURWHDVH

6(53,1LQLWLDOLQWHUDFWLRQFDQSURGXFHWZR
IXQFWLRQDOO\RSSRVLQJRXWFRPHV

,QWHUPHGLDWH
&OHDYHG5&/
3URWHDVH3URWHDVH,QKLELWRU
3HUPDQHQWO\LQDFWLYH
FRPSOH[

6XLFLGDOWUDQVIRUPDWLRQRIWKHLQKLELWRUದ
5HDFWLYH&HQWHU/RRS FDQQRWEHUHJHQHUDWHGEDFNWRDFWLYHIRUP
5&/ UHJLRQZKHUH
WKHLPSRUWDQWELRORJ\
WDNHVSODFH

,UUHYHUVLEOHPRGLILFDWLRQDQG
VWDELOL]HGSURWHDVHFRPSOH[ERWK
SURWHDVHDQGLQKLELWRUDUH
SHUPDQHQWO\LQDFWLYH

Fig. 1 SERPIN protease inhibitors must be accounted for differently than binary binders

the reactive bond site within the RCL, establishing a covalent

linkage between the carboxyl group of the SERPIN reactive site
and the serine hydroxyl of the protease [4]. The resulting inactive
serpin-protease complex is highly stable, and the structural disorder
induces its proteolytic inactivation. As a consequence, the protease
is permanently inhibited and functionally inactivated. Nevertheless,
the story does not end there for the inhibitor, as after the initial
interaction with the substrate protease, one of two possible out-
comes can occur, Fig. 1.
One possible outcome is driven by covalent modification per-
manently inactivating the inhibitory capacity as the SERPIN pep-
tide reactive bond region is irreversibly bound to the protease, and
thus cannot be reconstituted back to an active form. The second
possible outcome is a permanently inactive variant of the SERPIN
44 Swapan Roy and Matthew Kuruc

as the peptide RCL region is cleaved and can no longer bind target
substrates [4]. As a result, even minor changes in the structure due
to genetic variation and posttranslational modifications can modify
the function of SERPINs and give rise to a variety of clinical pre-
sentations. Some 200 different mutations in serpins are known to
result in disease [5]. In particular, mutations affecting antithrombin
confer a predisposition to thrombosis, those affecting C1 Inhibitor
confer a predisposition to angioedema, and those affecting anti-
plasmin confer a predisposition to hemorrhage. Interestingly, an
alternative function is made possible by a mutation in which the
methionine in the RCL region of Alpha-1-Antitrypsin is replaced
by an arginine converting its function as an inhibitor of neutrophil
elastase to a highly effective inhibitor of the coagulation proteases,
the consequence of which is life-threatening hemorrhagic
disease [6].
Mutations can affect function throughout the sequence. How-
ever, the most common loss of serpin function from mutation are
those affecting the mobile hinges of the molecule within or near the
RCL. These lead to spontaneous changes in conformation that
allow either the insertion of the intact reactive loop into the main
β sheet, resulting in the formation of an inactive “latent” form, or
the insertion of the loop of one molecule into the β sheet of the
next, resulting in the formation of polymers. Polymerization occurs
in Alpha-1-Antitrypsin with the common Z variant mutation, lead-
ing to decreased secretion from the liver into the circulation, result-
ing in emphysema and cirrhosis [7]. Amino acid substitution in the
RCL region is the likely event transforming the non-inhibitory
serpins. Posttranslational modifications at the RCL region such as
oxidation of methionine in Alpha-1-Antitrypsin have also been
proposed as a source of dysfunction [8].
So understanding the underlying mechanisms, contributions
from genetic wiring or environmental stresses, and their relation-
ships with aberrant proteolysis is necessary to characterize disease.
Functional proteomic analyses offer a new lens of observation to
examine the resulting conformational variants that can be reported
as potential biomarkers of disease phenotypes. As an example, one
such inhibitor SERPINA1, known more commonly as Alpha-1-
Antitrypsin (AAT), has several isoforms observed in plasma using
2-DE, and often serves as a model for conformational diseases
[5, 9]. Circulating levels of AAT are between 1.2 and 2 mg/mL
in healthy persons, but are known to increase during acute phases of
inflammation and infection. Its function and activity is controlled
by the many variants attributable to its conformational nexus of
features; the term “proteoform” is often used to describe such
conformational variability and we adopt that term here.
Other reports observe that the conformational properties of
AAT have multiple effects on tumor cell viability and diverse roles in
tumorigenesis, suggesting such isoforms may display a specific basis
 Methods to Monitor Functional SERPINs 45

for diagnosis of cancer and neurodegenerative disorders [8, 10,

11]. Yet, most often in proteomics, all subpopulations of SERPINs
are rolled into and counted as one homogeneous population. As a
result, the regulation, balance and dynamism within these systems
and its impact on the protease web of disease progression cannot be
properly investigated, and indeed conclusions based on such mea-
surements may be very misleading.
So methods that account for important distinctions among the
many subpopulations generated by conformational variants within
this superfamily of proteins are considered in this chapter. Specifi-
cally, a functional proteomic investigation of the two seemingly
opposing outcomes of the initial inhibitor-protease interaction
can be monitored:
1. The total amount of potentially inhibitory SERPIN activity as
reported by an intact RCL region.
2. A transformed subpopulation of the inhibitor, as reported by a
cleaved RCL region permanently inactivating its inhibitory
potential.

1.2 New Methods to By combining unique strategies of binding and voiding high-
Functionally Profile abundance proteins, we can observe different subpopulations with
SERPINs characteristic binding biases. We have previously reported for
Alpha-1-Antitrypsin that the resultant cleaved-RCL proteoform
and the uncleaved-RCL proteoform are very distinctive subpopula-
tions, separated by AlbuVoid™, and reported at the peptide fea-
ture level by LC-MS [12]. In this chapter, we consider how
Albumin Removal products—AlbuVoid™ and AlbuSorb™ (Bio-
tech Support Group LLC, Monmouth Junction, NJ, USA), can
help to functionally profile and unravel this complex biology of the
SERPIN superfamily of proteins.
Through a proprietary polymer coating, 50 μm porous silica
beads are crosslinked and passivated. This is the foundation of the
NuGel™ surface chemistry. Mixed-mode of binding interactions
form the basis of general nonspecific protein adsorbents or beads
with weak affinity or imperfect fit interactions. In this way, binding
behavior is very different from classical high affinity binding which
demands near perfect fits. Under protein saturation conditions,
progressive displacement provides a separation bias towards or
against select proteins. As a result, all derivative NuGel™ products
were empirically characterized to meet the needs of the application,
for example, AlbuVoid™ to selectively void (not bind) Albumin
with special bias toward the vast majority of the remaining
low-abundance serum proteome on the bead. Two NuGel™
based products support Albumin Removal:
1. AlbuSorb™ and AlbuSorb™ PLUS (also binds immunoglo-
bulins) for selective binding of Albumin.
46 Swapan Roy and Matthew Kuruc

2. AlbuVoid™ for negative selection or voidance of Albumin

with consequent enrichment of the remaining serum
sub-proteome on the bead.
So, while other proteomic methods might observe this:
Past observations:

Relative Abundance of Total Serpin

Population
2.5
Normal Disease
2.

1.5

1.

0.5

0.

We describe methods to observe this:

Relatve Abundance Sub-populations ACTIVE sub-

of Serpin populations
2.5
Normal Disease
2 INACTIVE sub-
populations
1.5

0.5

In this hypothetical case, the ratio of the ACTIVE

subpopulation vs. the INACTIVE subpopulation is greatly altered
in disease, whereas simple abundance measurements of the total
population would not be very informative (see Notes 1 and 2). In
the following Table 1, we report on the SERPINs observable by
LC-MS and how they bias toward AlbuVoid™ and AlbuSorb™,
as measured by spectral counting.
We suspect that conformational changes associated with the
cleavage of the reactive bond confer more or less binding affinity
to the nonspecific interactions with our beads. Such cleavage stabi-
lizes the SERPIN structures; AlbuVoid™ binding especially biases
toward unstructured proteins, and we have previously reported the
SERPINA1 (Alpha-1-Antitrypsin) RCL-intact proteoform binding
favorably over the RCL-cleaved proteoform [12]. Noteworthy is
that several non-inhibitory SERPINs A6-8, all bind poorly to
AlbuVoid™, supporting evidence for the role of conformational
stability in binding biases.
Table 1
Serum SERPINs observable by LC-MS

AlbuSorb™
AlbuVoid™ bead flowthrough
Protein ID Also known as (conc.) Function bound S.Cts. (unbound) S.Cts. Reactive (RCL) bond site Notable variants

SERPINA1 Alpha-1-Antitrypsin Inflammation, elastase 59 (strong bias 519 Met382-Ser383 Z variant

(AAT) (1–2 mg/ inhibition toward {Glu366 ! Lys366}
mL) RCL-intact deficiency syndrome,
proteoform) Pittsburgh variant
{Met382 ! Arg382}
life-threatening
bleeding
SERPING1 Plasma protease C1 Regulates complement 51 63 Ala465-Arg466
inhibitor (0.25 mg/ cascade, levels rise ~2- chymotrypsin, Arg466-
mL) fold during Thr467
inflammation
SERPINA3 Antichymotrypsin Apoptosis, Alzheimers, 86 117 Leu383-Ser384
(100–500 μg/mL) inflammation
SERPIND1 Heparin cofactor II Coagulation, thrombin 124 28 Leu463-Ser464
(40–80 μg/mL) inhibitor activated by
heparin
SERPINA8 Angiotensinogen Angiotensin I precurser, 4 62 None Disulfide bond is labile,
(AGT) (40–60 μg/ blood pressure near 40:60 ratio with
mL) regulation, the oxidized disulfide-
non-inhibitory bonded form
SERPINC1 Antithrombin, ATIII Inhibits thrombin, 58 79 Arg425-Ser426 Mutations/variants can
(0.12 mg/mL) regulates coagulation, lead to increased risk
angiogenesis, heparin of thrombosis, alter
cofactor functional heparin and
thrombin binding
domains
Methods to Monitor Functional SERPINs

(continued)
47
 48

Table 1
(continued)

AlbuSorb™
AlbuVoid™ bead flowthrough
Protein ID Also known as (conc.) Function bound S.Cts. (unbound) S.Cts. Reactive (RCL) bond site Notable variants

SERPINF1 Pigment epithelium- Neurotrophic factor, 45 0 None

derived factor, non-inhibitory
PEDF (20–175 μg/
mL)
SERPINA4 Kallistatin (20 μg/ Kidney function, 45 0 Phe388-Ser389 Cleavage at the reactive
Swapan Roy and Matthew Kuruc

mL) inflammation site by tissue kallikreins

SERPINF2 ɑ-2-antiplasmin Fibrinolysis, inhibitor of 10 39 Arg403-Met404 plasmin, Alanine insertion at the
(60–80 μg/mL) plasmin and trypsin Met404-Ser405 reactive site promotes
chymotrypsin serious bleeding
disorders
SERPIN- Z-dependent Coagulation regulation 23 0 Tyr408-Ser409 Tyr408 ! Ala408 loss of
A10 proteinase inhibitor inhibition
(1–2 μg/mL)
SERPINA5 Protein C inhibitor Coagulation, 13 0 Arg373-Ser374 Variants near or at the
(5 μg/mL) inflammation reactive bond alter
inhibition of thrombin
activity
SERPINA6 Corticosteroid- Hormone transport, 0 26 None
binding globulin non-inhibitory
(60–80 μg/mL)
SERPINA7 Thyroxine-binding Hormone transport, 0 17 None
globulin (15 μg/ non-inhibitory
mL)
 Methods to Monitor Functional SERPINs 49

2 Materials

Items required Reagent

AlbuVoid™ beads Manufacturer supplied
Binding buffer AVBB, PH 6.0 Manufacturer supplied
Wash buffer AVWB, PH 7.0 Manufacturer supplied
SpinX centrifuge tube filters Manufacturer supplied
Trypsin, DTT, iodoacetamide Not supplied

3 Methods

For this chapter, we shall consider only the workflow supporting

AlbuVoid™, but LC-MS workflows supporting AlbuSorb™
would be similar, taking into account which fractions would con-
tain the majority of Albumin and which do not, Fig. 2.
The workflow follows the AlbuVoid™ LC-MS On-Bead sam-
ple prep method following the manufacturer’s protocol. In brief,
50 μL serum is prepared by adding a binding buffer, then applied to
the AlbuVoid™ beads, and washed. All steps are performed within
a microfuge spin-filter format. Albumin is most especially voided
out, while the majority of the remaining serum proteome is
retained on the bead. After the final wash, reduction, alkylation,
and Trypsin digestion all take place on the bead.
For best results—the serum should be clear and free of colloidal
material. We recommend first filtering through a 0.45 μm syringe-
type filter before beginning the prep.
In bold are the AlbuVoid™LC  MSOn  Beadkitcomponents:
1. Weigh out 25 mg of AlbuVoid™ bead in a spin-tube (0.45 μ
SpinX centrifuge tube filter supplied).
2. Add 125 μL of Binding Buffer AVBB. Vortex for 5 min at
room temperature followed by centrifugation at 1500  g.
Discard the supernatant.
3. Repeat step 2.
4. Condition clarified serum by adding 100 μL of AVBB to
50–100 μL of the Serum. Using a syringe-type micro-filter,
clarify the serum. Add sample to the AlbuVoid™ beads in
step 3. Vortex for 10 min and then centrifuge for 5 min at
10,000  g.
5. Discard the albumin filtrate.
6. To the beads, add 250 μL of Wash Buffer AVWB. Vortex for
5 min and centrifuge for 4 min at 10,000  g. Discard
the Wash.
50 Swapan Roy and Matthew Kuruc

Serum
diluted
in
Binding
Buffer
added
Beads in to tube
Diluted serum
Spin-X
with beads
tube
Bead-based Protein Level Separation –
Analysis can be either or both bead-bound or flow-through
sub-populations
Bead-bound
proteome sub-
population
Trypsin Digest Options
- Elution from bead
In-gel
FASP
- On-Bead Digest

Tryptic peptides can be

quantified by isobaric Flow-through
labels (i.e., TMT) or proteome sub-
spectral intensities (.ie., population
MRM).

Fig. 2 Enrichment/depletion option for serum proteome separations

7. Repeat step 6 two times.

The AlbuVoid™ beads are now enriched with albumin-
depleted low-abundance proteins. For LC-MS sample prepara-
tion, the on-bead digestion protocol is as follows. Option—the
proteins can be eluted with 0.25 M Tris, 0.5 M NaCl, pH 10
(see Note 3).
8. After the final wash steps from step 7 from the enrichment, add
10 μL 100 mM DTT + 90 μL Wash Buffer AVWB, vortex
10 min, incubate ½ h at 60  C.
9. After cooling, add 20 μL 200 mM Iodoacetamide and 80 μL
Wash Buffer AVWB, incubate in dark for 45 min at
room temp.
10. Centrifuge at 10,000  g (microfuge max setting) for 5 min,
and discard supernatant.
11. Add 40 μL sequencing-grade trypsin (0.4 μg/μL, in 50 mM
acetic acid) + 60 μL Wash Buffer AVWB to the beads. Digest
 Methods to Monitor Functional SERPINs 51

overnight (maximum) at 37  C or other suitable time period

(s).
12. Centrifuge at 10,000  g (microfuge max setting) for 5 min
and retain peptide filtrate.
13. To further extract remaining peptides, add 150 μL 10% formic
acid, vortex 10 min, centrifuge at 10,000  g (microfuge max
setting) for 5 min, and add this volume to the first volume.
14. Total is about 250 μL. Prepare to desired final concentration.
Store at 80  C until LC-MS/MS.
Example of LC-MS reporting features.
After TMT labels (Proteome Sciences plc, Surrey, UK) labeling,
the peptides are pooled and analyzed with a single LC-MS/MS 3 h
gradient run using nanoRSLC system interfaced with a Thermo
Scientific™ Q Exactive™ HF (Thermo Scientific) instrument,
using data-dependent acquisition with resolution of 60,000, fol-
lowed by MSMS scans (HCD 30% of collision energy) of 20 most
intense ions, with a repeat count of two and dynamic exclusion
duration of 60 s (Table 2).
The amino acid region of the RCL is 368–392, so the adjacent
RCL tryptic peptide at Lys367, highlighted in gray, serves as a
good comparison between the observable serum subpopulations,
Fig. 3.
Bead Bound—The subpopulation of proteins that bind and are
observed by the AlbuVoidTM methods.
Flow-through (unbound)—The subpopulation of proteins
that flow-through the AlbuVoidTM beads, unbound.
Table 2
SERPINA1 (AAT) TMT ratio: pooled pancreatic cancer/pooled normal

Sp Ct ¼ peptide spectral counts

Flow-
through Serum
Bead bound (unbound) untreated

SERPINA1 (AAT) TMT Sp TMT Sp TMT Sp

Peptide region Start Amino acid sequence End ratio Ct ratio Ct ratio Ct
Adjacent RCL 360 AVLTIDEK 367 0.35 9 1.78 21 1.53 14
Tryptic
RCL cleaved 368 GTEAAGAMFLEAIPM 382 1.05 7 1.16 23
RCL intact 368 GTEAAGAMFLEAIPM 389 0.77 5 1.75 1 1.34 50
SIPPEVK
RCL cleaved 383 SIPPEVK 389 1.45 27 1.44 18
Total all peptide features 0.54 132 1.57 372 1.44 460
52 Swapan Roy and Matthew Kuruc

Cleaved RCL:
Reactive Center Loop
Permanently inactive
(RCL), region

Bead-based separations can enrich sub-populations

for better LC-MS analysis of SERPIN sub-populations

LC-MS can report peptide features distinguishing RCL

regions that are intact, vs. those that are cleaved.

Fig. 3 SERPIN LC-MS reporting features

Untreated—The total population of proteins that are observ-

able in serum without any sample enrichment, that is without the
use of AlbuVoid™:
Highlighted in dark gray is the RCL intact peptide. High-
lighted in light gray are the two RCL peptides that are cleaved at
Met382, during suicidal substrate interaction; note that these pep-
tides were not observed in the Bead-Bound fraction. These data
suggest that the overall SERPINA (AAT) population is dominated
by the subpopulation up-regulated and collected in the Flow-
through fraction of AlbuVoid™, and this same subpopulation
dominates the analysis when untreated sera is investigated. Such
would be the case in acute AAT up-regulation commonly observed
with malignancies and inflammation. However, using our methods
we distinguish a subpopulation enriched by the bead and reporting
with the bound fraction, as being severely down-regulated with
cancer! While this observation may have potential biological signif-
icance, no conclusion about the particular cancer-specific proteo-
form uncovered can be made at this time (see Note 4).
Nevertheless, from a biomarker perspective, this serves an addi-
tional multiplier benefit; that is the ratio of the two subpopulations
report Adjacent RCL Tryptic peptide region as unbound/bound
(1.78/0.35) ¼ 5. As isobaric label ratios in discovery methods can
sometimes compress the reporting difference, this ratio may
become much greater once more targeted quantitative methods
are developed, a prospect for future tests.
 Methods to Monitor Functional SERPINs 53

4 Notes
l Bead-based proteomic enrichment methods as described can
support the functional and structural proteomic analyses neces-
sary to characterize these conformational subpopulations so that
they may become useful biomarkers for disease. It should none-
theless be recognized that the RCL reporting methods
described here only work for RCL regions where the cleavage
site is non-Tryptic, and these must be entered as special peptides
into LC-MS computational workflows. “neXtProt: a knowledge
platform for human proteins,” provides a useful web-based
resource for annotating RCL cleavage sites [13]. To distinguish
RCL regions where the cleavage site is Tryptic, it becomes
necessary to differentiate those sites that are cleaved in vivo by
those cleaved ex vivo. Several methods have been developed for
this purpose and generally fall under methods called
N-terminomics [14]. This is an area for future investigations.
l Classical high-abundance proteins like the SERPIN superfamily
(i.e., Alpha-1-Antitrypsin) are often overlooked as potential
biomarkers of disease. Yet discoveries certainly can rest in the
data-rich features of the diverse variety of conformational and
proteoform variants associated with many of the classical serum
proteins. When considering these mid- to high-abundance pro-
teins, disease differentiation can be obtained through the dis-
creet quantification of the multiple subpopulations available to
measure. The methods described in this chapter can begin to
unravel and sort these variant subpopulations so that LC-MS
peptide reporting features, and potentially other functional
reporting features (i.e., substrate turnover), can distinguish
these proteoforms with more functional details. It is our inten-
tion that these methods will lead to characteristic disease pro-
files, which can then be compared and evaluated for eventual
biomarker utility.
l Many trypsin digestion protocols have been developed to
improve the reproducibility and, in some cases, reduce the
digestion time necessary for LC-MS analysis [15]. While we
have shown methods that adapt AlbuVoid™ for on-bead diges-
tion, the bead-based enrichments described here are nonetheless
compatible, after elution from the beads, with other common
digestion methods, such as filter-aided (FASP) solution methods
and post-electrophoresis, in-gel methods.
l By using the peptide reporting features of the RCL peptide
regions within SERPIN inhibitors, both “potentially active”
and “permanently inactive” proteoforms are now distinguish-
able, adding a new level of proteomic characterization to the
underlying mechanisms of disease. As one example, hereditary
54 Swapan Roy and Matthew Kuruc

dysfunction of SERPINA1 (Alpha-1-Antitrypsin) has been pre-

viously determined as a risk factor for cancer [16]. As many
proteins within the SERPIN family proteins are of moderate-
to high-abundance quantities in serum (10 + μg range/mL),
depleted functionality would impose severe dysregulation to a
normal and healthy individual. Several of the key regulators in
the Coagulation Pathway such as SERPINA10 (Z-dependent
Proteinase Inhibitor) and SERPINA5 (Protein C Inhibitor)
have notable genomic variants that alter their inhibitory func-
tion [13]. These might therefore be risk factors for disease. So
using the methods described in this chapter, hereditary genomic
factors that associate with SERPIN function can be further
investigated.

References
1. Fortelny N, Cox JH, Kappelhoff R, Starr AE from patients during an acute coronary syn-
et al (2014) Network analyses reveal pervasive drome. J Am Coll Cardiol 44(8):1578–1583
functional regulation between proteases in the 10. Wang Y, Kuramitsu Y, Yoshino S et al (2011)
human protease web. PLoS Biol 12(5): Screening for serological biomarkers of pancre-
e1001869 atic cancer by two-dimensional electrophoresis
2. Shamamian P, Schwartz JD, Pocock BJ et al and liquid chromatography-tandem mass spec-
(2001) Activation of progelatinase A trometry. Oncol Rep 26(1):287–292
(MMP-2) by neutrophil elastase, cathepsin G, 11. Zelvyte I, Sjögren HO, Janciauskiene S (2002)
and proteinase-3: a role for inflammatory cells Effects of native and cleaved forms of α1-anti-
in tumor invasion and angiogenesis. J Cell trypsin on ME 1477 tumor cell functional
Physiol 189(2):197–206 activity. Cancer Detect Prev 26(4):256–265
3. Law RH, Zhang Q, McGowan S et al (2006) 12. Zheng H, Zhao C, Roy S et al (2016) The
An overview of the serpin superfamily. Genome commonality of the cancer serum proteome
Biol 7(5):216 phenotype as analyzed by LC-MS/MS, and its
4. Khan MS, Singh P, Azhar A et al (2011) Serpin application to monitor dysregulated wellness.
inhibition mechanism: a delicate balance Poster presented at the AACR annual meeting
between native metastable state and polymeri- 2016 conference, New Orleans, LA, USA,
zation. J Amino Acids. https://doi.org/10. April 17–20 2016
4061/2011/606797 13. Lane L, Argoud-Puy G, Britan A et al (2011)
5. Carrell RW, Lomas DA (2002) Alpha1- neXtProt: a knowledge platform for human
antitrypsin deficiency—a model for conforma- proteins. Nucleic Acids Res 40(D1):D76–D83
tional diseases. N Engl J Med 346(1):45–53 14. Lai ZW, Petrera A, Schilling O (2015) Protein
6. Owen MC, Brennan SO, Lewis JH et al (1983) amino-terminal modifications and proteomic
Mutation of antitrypsin to antithrombin: α1- approaches for N-terminal profiling. Curr
antitrypsin Pittsburgh (358 Met!Arg), a fatal Opin Chem Biol 24:71–79
bleeding disorder. N Engl J Med 309 15. Zheng H, Zhao C, Qian M et al (2015) Albu-
(12):694–698 Void™ coupled to on-bead digestion-tackling
7. Sifers RN (1992) Z and the insoluble answer. the challenges of serum proteomics. J Proteom
Nature 357(6379):541 Bioinformatics 8(9):225
8. Janciauskiene S (2001) Conformational prop- 16. Sun Z, Yang P (2004) Role of imbalance
erties of serine proteinase inhibitors (serpins) between neutrophil elastase and α1-antitrypsin
confer multiple pathophysiological roles. Bio- in cancer development and progression. Lancet
chim Biophys Acta 1535(3):221–235 Oncol 5(3):182–190
9. Mateos-Cáceres PJ, Garcı́a-Méndez A, Farré
AL et al (2004) Proteomic analysis of plasma
 Chapter 3

Two-Dimensional 16-BAC/SDS Polyacrylamide Gel

Electrophoresis of Mitochondrial Membrane Proteins
Gary Smejkal and Srikanth Kakumanu

Abstract
The substitution of the reverse polarity benzyldimethyl-n-hexadecylammonium chloride (16-BAC) poly-
acrylamide gel electrophoresis (PAGE) for isoelectric focusing (IEF) in the first dimension of electrophore-
sis improves the solubility of extremely hydrophobic proteins and their recovery compared to conventional
2D IEF/SDS PAGE. The acidic environment of 16-BAC PAGE has also been shown to better preserve the
labile methylation of basic proteins such as the histones. Several improvements of the 2D 16-BAC/SDS
PAGE method are collectively described here with particular emphasis on the separation of mitochondrial
membrane proteins of low molecular mass. Lowering the 16-BAC concentration 50-fold in the gel and
buffers decreases the formation of mixed 16-BAC/SDS micelles, which otherwise interferes with the
separation of very low molecular mass proteins in second dimension SDS PAGE, and consequently
improved the resolution of mitochondrial membrane proteins in the 10–30 kDa range.

Key words Benzyldimethyl-n-hexadecylammonium chloride, Cationic detergents, Membrane pro-

teins, Mitochondria, Polyacrylamide gel electrophoresis, Proteins, Sodium dodecylsulfate, Two-
dimensional gel electrophoresis, Transmembrane domains

Abbreviations

16-BAC Benzyldimethyl-n-hexadecylammonium chloride

CMC Critical micelle concentration
DTT Dithiothreitol
HED Hydroxyethyl disulfide
IEF Isoelectric focusing
KDS Potassium dodecylsulfate
PAGE Polyacrylamide gel electrophoresis
PMSF Phenylmethylsulfonyl fluoride
SDS Sodium dodecylsulfate
TCEP Tris (2-carboxyethyl) phosphine
TMDs Transmembrane domains

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_3, © Springer Science+Business Media, LLC, part of Springer Nature 2019

55
56 Gary Smejkal and Srikanth Kakumanu

1 Introduction

Two-dimensional gel electrophoresis, as we’ve come to know it,

combines IEF with orthogonal sodium dodecylsulfate (SDS)
PAGE. The resolution of IEF is best exemplified in several land-
mark publications by Klose et al. [1, 2] which reported the separa-
tion over 10,300 proteins from mouse tissue using very large
format IEF/SDS PAGE. The resolution of IEF is further increased
using immobilized pH gradients (IPGs) which are capable of separ-
ating protein charge isoforms differing by only 0.001 pI units, a
resolution an order of magnitude higher than carrier ampholyte
generated pH gradients [3].
However, the strict requirement of nonionic or zwitterionic
detergents limits the compatibility of many detergents with IEF.
Many extremely hydrophobic proteins are insoluble in the deter-
gents typically used for IEF, or if they are initially solubilized, they
may precipitate near their isoelectric point and are excluded from
the second dimension analysis. Klein et al. [4] showed the separa-
tion of peripheral membrane proteins from Halobacterium sali-
narum by conventional IEF/SDS PAGE, but failed to separate
integral membrane proteins having multiple transmembrane
domains (TMDs) that were irreversibly precipitated and trapped
in the first dimension IPG strip. Kalinowski et al. [5] reported that
CHAPS, a zwitterionic detergent commonly used in IEF, solubi-
lized only 52% of proteins from Corynebacterium glutamicum
membranes, compared to SDS which solubilized 90% of the mem-
brane proteins. The use of SDS to initially solubilize proteins from
Chlorobium tepidum membrane fractions, when followed by ace-
tone precipitation to remove SDS, enabled IEF and more than
doubled the number of proteins identified by IEF/SDS PAGE,
compared to fractions solubilized with Triton X-100 [6].
The use of SDS in both dimensions of orthogonal PAGE
(2D SDS/SDS PAGE) exploits the anomalous migration of
extremely hydrophobic proteins at different gel concentrations
[7]. While soluble proteins typically bind 1.4 times their mass of
SDS, some membrane proteins can bind as much as 4.5 times their
mass [8], in which case, the protein itself constitutes less than 20%
of the total mass of the nascent SDS-protein complex [9]. Hence,
the migration of membrane proteins inconsistent with their true
molecular mass can be explained, at least in part, by detergent-
induced shifts in both molecular mass and charge density. Recently,
Meisrimler and Luthje [10] used SDS/SDS PAGE as second and
third dimensions in a three-dimensional electrophoresis scheme
that identified over 10% more membrane proteins with multiple
TMDs from plant microsomes than IEF/SDS PAGE. Several other
variations of SDS/SDS PAGE have been described [11, 12].
Two-Dimensional 16-BAC/SDS Polyacrylamide Gel Electrophoresis. . . 57

MacFarlane [13, 14] first described 16-BAC PAGE to preserve

base labile protein methylation of platelet and promyelocyte pro-
teins during electrophoresis, and soon afterward published the first
methods for 2D 16-BAC/SDS PAGE [15] and preparative 2D
16-BAC/SDS PAGE [16]. Hartinger et al. [17] later applied 2D
16-BAC/SDS PAGE for the analysis of proteins in purified mem-
branes from synaptic and clathrin-coated vesicles. Using 2D
16-BAC/SDS PAGE, Zahedi et al. [18] identified 42 mitochon-
drial membrane proteins not isolated by 2D IEF/SDS PAGE pro-
teins, including the cytochrome-c oxidase subunit I which contains
12 TMDs. They later introduced 16-BAC PAGE in tube gels to
eliminate the need to excise gel lanes [19].
These seminal papers spawned over three decades of publica-
tions on 16-BAC PAGE and 2D 16-BAC/SDS PAGE with very
few improvements of the initial descriptive. Hence, some early
mistakes have been repeated for decades. For example, nearly
every paper ever published on 2D 16-BAC/SDS PAGE follows
the earlier convention of staining the first dimension gels with
Coomassie to guide lane excision prior to second dimension SDS
PAGE. This necessarily precipitates the proteins in the first dimen-
sion gel and assumes that they will be completely resolubilized
during the brief SDS equilibration that precedes second dimension.
Other proteins that are soluble under such acidic conditions are not
fixed in the gel and are washed out, unfortunately risking “throw-
ing away the baby with the wash water.” Despite this, the overnight
staining of first dimension gels [18], and even their storage in
staining solution [17], before second dimension PAGE have been
reported. Using radiolabeled proteins, Hartinger et al. [17] esti-
mated a 10% loss of the total protein when stained gels were
transferred to second dimension. To the contrary, we have experi-
enced the nearly complete loss of low molecular mass mitochon-
drial membrane proteins (in the 10–40 kDa range) from
Coomassie-stained 16-BAC gels, compared to control gels that
were immediately transferred to second dimension SDS PAGE
without prior staining.
The separation of proteins by 2D 16-BAC PAGE is based on
the differential binding of the 16-BAC and SDS detergents to
proteins. Unlike SDS, the binding of 16-BAC to proteins is not
well characterized (Fig. 1). From Ferguson plot analysis of 16-BAC
protein derivatives, radically different slopes for high and low
molecular mass proteins are obtained [20]. The lack of a common
Y intercept in these plots implies that, unlike SDS, a constant net
charge density is not obtained for all proteins. This is the prerequi-
site for separations based purely on molecular size with no influence
by charge, such as in SDS systems. The differential binding of
16-BAC to some proteins is exemplified in the case of ovalbumin,
which consistently exhibits mobility similar to other proteins half its
size in 16-BAC PAGE [13, 20].
58 Gary Smejkal and Srikanth Kakumanu

Fig. 1 Chemical structure of benzyldimethyl-n-hexadecylammonium chloride.

Molecular weight 396.1

Fig. 2 2D 16-BAC/SDS PAGE of mitochondrial membrane proteins from porcine

myocardium. First dimension 16-BAC PAGE was 12% polyacrylamide
concentration. Second dimension SDS PAGE was 15% polyacrylamide
concentration. Gel concentrations were selected to maximize separation of
proteins in the 10–30 kDa range. Gels were stained overnight in KUMASI

In another important publication, Kramer [20] determined the

critical micelle concentration (CMC) of 16-BAC in the PAGE
buffers and demonstrated that the detergent concentration could
be lowered 50-fold in the gel and running buffers (A molar excess
of 16-BAC was kept in sample buffer to ensure the complete
saturation of proteins.). Working at concentrations near or below
the CMC of 16-BAC significantly decreases the formation of mixed
16-BAC/SDS micelles, which otherwise interfere with the separa-
tion of very low molecular mass proteins in second dimension SDS
PAGE (Fig. 2). Consequently, the separation of membrane proteins
in the 10–30 kDa range is improved, even at high polyacrylamide
gel concentrations where micelles are sieved. In theory, it should be
possible to exclude 16-BAC entirely from the stacking and resolv-
ing gels, relying on the constant influx of free detergent from the
anode buffer to maintain detergency.

2 Materials

2.1 Isolation of 1. 100 mM phenylmethylsulfonyl fluoride (PMSF) in 100% iso-

Mitochondrial propanol (see Note 1).
Membrane Proteins 2. Mitochondria isolation buffer (MIB): 20 mM Tris–HCl
pH 7.4, 250 mM sucrose, 10 mM potassium fluoride, 2 mM
EGTA, 1 mM sodium vanadate, 1 mM PMSF.
 Two-Dimensional 16-BAC/SDS Polyacrylamide Gel Electrophoresis. . . 59

3. Chloroform.
4. Methanol.

2.2 First Dimension Buffers and concentrated buffer stocks can be prepared in advance
16-BAC PAGE without the addition of 16-BAC. All buffers are filter sterilized
(e.g., Millipore Steriflip or similar) and can be stored up to
3 months at 4  C. The 16-BAC detergent is added to working
solutions immediately before use and used the same day.
1. Concentrated (4) resolving gel buffer: 300 mM KH2PO4
pH 2.1 (see Note 2).
2. Concentrated (4) stacking gel buffer: 500 mM KH2PO4
pH 4.1 (see Note 3).
3. 125 mM 16-BAC stock solution (see Note 4).
4. Concentrated (10) electrode buffer: 1.5 M glycine, 500 mM
ortho-phosphoric acid (see Note 5).
5. Working (1) electrode buffer: 150 mM glycine, 50 mM
ortho-phosphoric acid, 0.05 mM 16-BAC (see Note 6).
6. 9 M urea (see Note 7).
7. AG 501-X8 mixed bed ion exchange resin (see Notes 7 and 9).
8. Bond-Breaker™ 500 mM Tris (2-carboxyethyl) phosphine
(TCEP) solution, neutral pH (Thermo-Fisher Scientific,
77720).
9. 50% glycerol.
10. Concentrated (100) Pyronin Y tracking dye: 1 mg/mL Pyr-
onin Y in water.
11. 1 sample buffer: 4 M urea, 50 mM 16-BAC, 10 mM TCEP,
10.5% glycerol, 0.005% Pyronin Y (see Note 8).
12. 29.2% acrylamide, 0.8% methylene bisacrylamide solution (see
Note 9).
13. 80 mM ascorbic acid (see Note 10).
14. 5 mM ferrous sulfate (see Note 11).
15. 30% hydrogen peroxide.
16. 25% isopropanol.
17. Reflection™ Dual Vertical Electrophoresis System (Galileo
Biosciences, 85-1614).
18. Precision Plus Protein™ Standards (Biorad, 1610374).
19. Stainless steel tissue slicing blade, 22 cm length.
60 Gary Smejkal and Srikanth Kakumanu

2.3 Neutralization 1. Neutralization Buffer: 375 mM Tris HCl pH 8.8, 3 M urea, 5%

and SDS Equilibration glycerol, 0.001% bromophenol blue (see Note 12).
2. Dithiothreitol (DTT).
3. SDS Equilibration Buffer: 375 mM Tris HCl pH 8.8, 3 M urea,
2% SDS, 5% glycerol, and 0.001% phenol red. Solid 50 mM
DTT is added immediately before use (see Note 13).
4. Polypropylene reagent reservoirs, 60 mL.
5. Medium thickness filter paper, 100  20 mm.

2.4 Second 1. Criterion Dodeca Cell Vertical Electrophoresis System (Biorad,

Dimension SDS PAGE 165-4130).
2. Criterion 15% Tris–HCl precast polyacrylamide gels (BioRad,
345-0019) or Criterion empty cassettes (see Note 14).
3. SDS PAGE electrode buffer: 25 mM Tris, 192 mM glycine,
0.1% SDS pH 8.3.
4. 30% ethanol, 10% acetic acid.
5. KUMASI stabilized colloidal Coomassie staining solution
(Focus Proteomics, FPKS-001).
6. 1 M sodium azide.

3 Methods

3.1 Isolation of Mitochondria were isolated as described by Lee et al. [21]. Porcine
Mitochondrial heart was resected within minutes of euthanasia and placed imme-
Membrane Proteins diately on wet ice. The tissue was dissected within 1 h of collection.
All steps were performed at 4  C.
1. Remove connective tissue and fat and course grind the myo-
cardium in a food grinder.
2. Suspend the macerate in five volumes of MIB.
3. Blend the suspension in a food processor for three times 30 s.
4. Centrifuge the suspension at 650 RCF for 10 min and filter the
supernatant through multiple layers of cheese cloth.
5. Resuspend the remaining pellet in additional MIB and repeat
steps 3 and 4. Combine the two supernatants.
6. Centrifuge the pooled supernatants and centrifuge at 14,000
RCF for 20 min.
7. Homogenize the resulting pellet in a ground glass homoge-
nizer with Teflon plunger.
8. Centrifuge at 400 RCF for 8 min to pellet cellular debris.
Transfer the supernatant to a new tube and centrifuge for
14,000 RCF for 20 min.
 Two-Dimensional 16-BAC/SDS Polyacrylamide Gel Electrophoresis. . . 61

9. Repeat steps 7 and 8 until the supernatant is clear and the

pellet is beige in color.
10. Resuspend the mitochondrial pellet in a small volume of MIB
and determine the protein concentration using the Lowry
method. Adjust the sample volume with MIB so that protein
concentration is 20 mg/mL. The mitochondria can be stored
at 80  C for later analysis.
11. Prior to isolation, dilute the proteins to 3 mg/mL in MIB. Add
four volumes of methanol, one volume of chloroform, and
three volumes of water, vortexing vigorously after each
addition.
12. Centrifuge at 14,000 RCF for 2 min.
13. Remove the top aqueous layer and add 400 μL of methanol.
Centrifuge at 14,000 RCF to pellet the precipitated proteins.
14. Aspirate the supernatant and allow the precipitated protein
pellet to air-dry making sure to not overdry the pellet as it
will be very difficult to solubilize for electrophoresis. Pellets can
be stored at 80  C for later analysis.

3.2 16-BAC PAGE The Fenton reaction as modified by MacFarlane [13] is used to
catalyze the polymerization reaction. Generally, 16-BAC gels are
run on the same day they are cast. If necessary, the resolving gel can
be cast on the first day, then overlaid with 1 resolving buffer and
stored overnight at room temperature. The stacking gel should be
cast within a few hours of running the gel on the second day. All
steps are performed at room temperature.
1. Assemble two 16  14 cm glass plates and two 0.8 mm thick
spacers using the alignment pins of the upper buffer chamber
and clamp securely. The entire assembly is transferred to the
casting stand and cammed into place for leakproof casting
(Fig. 3).
2. Prepare the resolving gel as prescribed in Table 1. Combine all
components except for the catalysts in a 50 mL screw cap
centrifuge tube. Cap securely and mix by gentle inversion
taking care not to introduce bubbles.
3. Add sequentially the absorbic acid, FeSO4, and H2O2. Cap
securely and mix by gentle inversion following each addition.
4. Immediately pour the polymerization mixture into the gel
cassette to within 3 cm of the top of the notched glass plate.
Carefully overlay the gel with 400–600 μL of 25% isopropanol
(see Note 15).
5. Allow the gel to polymerize for 15–20 min (see Note 16).
62 Gary Smejkal and Srikanth Kakumanu

Fig. 3 Exploded view of the Galileo Biosciences large format electrophoresis system and leakproof gel casting
system. Glass plates and spacers are clamped to the upper buffer chamber during casting and are not
removed until electrophoresis is complete. This avoids the “flexing” of gels which can result in disadherence
of the gel from the glass surface and the formation of bubbles between the gel and glass

6. Decant the isopropyl overlay and wash the gel surface twice
with water. (If storing the gel overnight, overlay with 1X
resolving buffer.)
7. Prepare the stacking gel as prescribed in Table 2. Combine all
components except for the catalysts, cap securely, and mix by
gentle inversions. Add sequentially the absorbic acid, FeSO4,
and H2O2. Cap securely and mix by gentle inversion following
each addition.
8. Fill the remaining space in the gel cassette with stacking gel and
insert the 0.8 mm thick comb taking care not to trap air
bubbles. Allow to polymerize for at least 1 h.
9. Remove the comb and forcefully flush any unpolymerized
solution from the wells with water using a transfer pipette.
 Two-Dimensional 16-BAC/SDS Polyacrylamide Gel Electrophoresis. . . 63

Table 1
16-BAC PAGE resolving gel composition

Stock solution Final concentration Volume

29.2% acrylamide, 0.8% bisacrylamide 12% 20 mL
9 M urea 2.6 M 14.9 mL
4 resolving gel buffer pH 2.1 75 mM 12.5 mL
125 mM 16-BAC 0.05 mM 20 μL
Water – –
80 mM absorbic acid 4 mM 2.5 mL
5 mM ferrous sulfate 8 μM 80 μL
30% hydrogen peroxide 0.002% 3.3 μL
Total volume 50 mL

Table 2
16-BAC PAGE stacking gel composition

Stock solution Final concentration Volume

29.2% acrylamide, 0.8% bisacrylamide 4% 1.35 mL
9 M urea 3M 3.35 mL
4 stacking gel buffer pH 4.1 75 mM 2.5 mL
125 mM 16-BAC 0.05 mM 4 μL
Water – 2.3 mL
80 mM absorbic acid 4 mM 0.5 mL
5 mM ferrous sulfate 8 μM 16 μL
30% hydrogen peroxide 0.002% 1 μL
Total volume 10 mL

10. Fill the upper and lower buffer chambers with electrode buffer.
Flush the water from the wells with electrode buffer using a
transfer pipette.

3.3 Sample 16-BAC is dissolved in the sample buffer on the same day of the
Preparation analysis. Protein samples are prepared immediately before electro-
phoresis. Hartinger et al. [17] observed protein degradation and a
loss of resolution when samples were stored in the 16-BAC sample
buffer.
64 Gary Smejkal and Srikanth Kakumanu

1. Dissolve the mitochondrial pellet (100–200 μg total protein) in

25 μL of 16-BAC sample buffer. Incubate at 60  C for
10–15 min or until completely dissolved (see Note 17).
2. Dilute the protein standards at least tenfold in 16-BAC sample
buffer.
3. Centrifuge the samples and standards at 14,000 RCF for 5 min.
Apply 20 μL of supernatant to each well, leaving a blank lane
between samples when possible.
4. Connect to the power supply in reverse polarity. Commence
electrophoresis at 50 mA constant current until the Pyronin Y
dye has migrated 8–10 cm into the resolving gel.

3.4 Neutralization 1. Immediately following electrophoresis, disassemble the upper

and SDS Equilibration buffer chamber and open the gel cassette like a book using one
of the spacers to leverage. One plate should release while the
gel remains adhered to the other plate.
2. Using the stacking gel and Pyronin Y dye front as reference,
excise a slice from the center of each lane using the tissue slicing
blade. For best results, the gel slice should not exceed 4 mm in
width. Trim off the stacking gel and excess resolving gel below
the Pyronin Y.
3. Transfer each gel slice to a clean polypropylene reagent reser-
voir and incubate for 2 2 min in 5 mL of neutralization
buffer.
4. Incubate each gel slice for 2 10 min in 5 mL of SDS equili-
bration buffer.

3.5 Second 1. Flush the surface of the second dimension gel with water to
Dimension SDS PAGE remove residual storage buffer. Fill the upper buffer chamber
with SDS PAGE electrode buffer.
2. Soak a 100  20 mm filter paper in SDS PAGE electrode
buffer.
3. With a spatula, position the equilibrated gel strip on top of the
second dimensional gel. Place the saturated filter paper on top
of the equilibrated gel strip and gently press down to keep the
gel strip in intimate contact with the second dimension gel
(Fig. 4).
4. Connect to the power supply in normal polarity. Commence
electrophoresis at 40 mA constant current for 10 min. Turn the
power off and remove the filter papers. Continue electropho-
resis until the phenol red dye front has migrated to within a few
millimeters of the bottom of the gel.

3.6 KUMASI Staining 1. Fix the gel for at least 1 h in 100 mL of 30% ethanol, 10% acetic
acid (see Note 18).
 Two-Dimensional 16-BAC/SDS Polyacrylamide Gel Electrophoresis. . . 65

Fig. 4 Second dimension SDS PAGE of excised lane from first dimension 16-BAC PAGE. SDS equilibrated gel
strips were held in intimate contact with the second dimension gel with a buffer-saturated filter paper wick
(arrow). This eliminates the need for an agarose gel overlay

2. Decant the fixative and stain the gel overnight in 100 mL of

KUMASI stain (see Note 19).
3. Rinse the gel for 30 s in 30% ethanol, 10% acetic acid.
4. Incubate the gel at least 20 min in 100 mL of activated
enhancer solution from the KUMASI staining kit. Contrast is
improved with overnight incubation in the enhancer solution.
5. If gels are to be stored for more than 1 week, add 10 mM
sodium azide (see Note 20).

4 Notes

1. The half-life of PMSF is 55 and 35 min at pH 7.5 and 8.0,

respectively, in aqueous buffers [22]. Stock solutions prepared
in 100% isopropanol are stable for months.
2. To make 4 resolving gel buffer, dissolve 8.2 g KH2PO4 in
150 mL water and adjust to pH 2.1 with 1 N HCl. Adjust
volume to 200 mL. Filter sterilize and store at 4  C for up to
3 months.
3. To make 4 stacking gel buffer, dissolve 13.6 g KH2PO4 in
150 mL water and adjust to pH 4.1 with 1 N HCl. Adjust
volume to 200 mL. Filter sterilize and store at 4  C for up to
3 months.
4. Prepare fresh 16-BAC solution daily. To make 125 mM stock
solution, dissolve 50 mg 16-BAC (Millipore-Sigma, B-4136)
in 10 mL water.
66 Gary Smejkal and Srikanth Kakumanu

5. To make 10 electrode buffer, dissolve 56 g glycine in 450 mL

water. Add 28.8 mL 85% ortho-phosphoric acid and adjust
volume to 500 mL. Do not adjust pH. Filter sterilize and
store at 4  C for up to 3 months.
6. Use 100 mL of concentrated (10) buffer per liter of working
electrode buffer. Dissolve 20 mg solid 16-BAC per liter imme-
diately before use.
7. Dissolve 24.3 g urea in 45 mL water in a 50 mL screw cap
centrifuge tube. Add 0.5 g AG 501-X8 mixed bed ion
exchange resin (Biorad, 143-7424) and incubate 30–60 min
with gentle nutation. Filter sterilize, store at room tempera-
ture, and use within 3 days.
8. Prepare sample buffer fresh daily. Admix 4.5 mL 9 M urea and
3.5 mL 30% glycerol. Add 0.2 g 16-BAC and dissolve by
nutation. Add 50 μL 500 mM TCEP and 50 μL of 100
Pyronin Y and adjust volume to 10 mL.
9. Acrylamide and methylene bisacrylamide are potent neurotox-
ins. To minimize hazards, use premixed 30% acrylamide-
bisacrylamide solution (BioRad, 161-0159). Add 0.5 g AG
501-X8 mixed bed ion exchange resin to 45 mL 30%
acrylamide-bisacrylamide solution in a 50 mL screw cap centri-
fuge tube and incubate 30–60 min with gentle nutation. Filter
sterilize and store at 4  C for up to 2 months.
10. Prepare 80 mM ascorbic acid fresh daily. Dissolve 140 mg
absorbic acid in 10 mL water.
11. Prepare 5 mM FeSO4 fresh daily. Dissolve 70 mg FeSO4 in
50 mL water.
12. Bromophenol blue is a pH indicator dye that transitions from
yellow to blue indicating when the gel strip is neutralized.
Another more important role of the neutralization step is to
remove residual KH2PO4 from the gel prior to SDS equilibra-
tion since K+ ions would otherwise drive the formation of
insoluble potassium dodecylsulfate (KDS). The Krafft point
of KDS is 36  C.
13. DTT has a half-life of approximately 1.4 h at pH 8.5 at room
temperature. Desiccated DTT solids should be dissolved in the
SDS equilibration buffer and used within 1 h. The SDS equili-
bration buffer can be prepared in advance (without DTT),
filtered sterilized, and stored at 20  C for at least 2 months.
DTT is ineffective as reducing agent below pH 7.0 where only
about 1% of the thiol groups of DTT are in the reactive thiolate
form [23].
14. Gels can be hand cast at any desired gel concentration into
Criterion empty cassettes as described in detail by Smejkal and
Bauer [9].
 Two-Dimensional 16-BAC/SDS Polyacrylamide Gel Electrophoresis. . . 67

15. The gel can be “misted” with 25% isopropanol from a spray
bottle.
16. Gel polymerization rates are affected by temperature and rela-
tive humidity. If the gel polymerizes in less than 10 min, discard
and prepare a new gel using 10% less of each catalyst.
17. Avoid heating the sample higher than 60  C. Asp-Pro linkages
are susceptible to hydrolysis at acidic pH and elevated
temperature.
18. For improved staining, fix overnight to completely remove
SDS from the gel.
19. The fixative and KUMASI staining solutions can be reused at
least two times.
20. Gels stored in enhancer solution with sodium azide added are
stable for years. We have observed improved contrast with no
significant loss of sensitivity in gels stored for 9 years.

References
1. Klose J, Kobalz U (1995) Two-dimensional preparation in biological mass spectrometry.
electrophoresis of proteins: an updated proto- Springer, Heidelberg, pp 411–434
col and implications for a functional analysis of 8. Rath A, Glibowicka M, Nadeau VG, Chen G,
the genome. Electrophoresis 16:1034–1059 Deber CM (2009) Detergent binding explains
2. Klose J, Nock C, Herrmann M, Stühler K, anomalous SDS-PAGE migration of mem-
Marcus K, Blüggel M, Krause E, Schalkwyk brane proteins. PNAS 106:1760–1765
LC, Rastan S, Brown SDM, Büssow K, 9. Smejkal GB, Bauer DJ (2012) High speed iso-
Himmelbauer H, Lehrach H (2002) Genetic electric focusing of proteins enabling rapid
analysis of the mouse brain proteome. Nat two-dimensional gel electrophoresis. Gel elec-
Genet 30:385–393 trophoresis: principles and basics. Intech,
3. Hamdan M, Righetti PG (2005) Proteomics Rijeka, pp 157–170
today: Protein assessment and biomarkers 10. Meisrimler CN, Lüthje S (2012) IPG-strips
using mass spectrometry, 2D electrophoresis, versus off-gel fractionation: advantages and
and microarray technology. Wiley & Sons, limits of two-dimensional PAGE in separation
Hoboken, NJ, pp 219–265 of microsomal fractions of frequently used
4. Klein C, Garcia-Rizo C, Bisle B, Scheffer B, plant species and tissues. J Proteome
Zischka H, Pfeiffer F, Siedler F, Oesterhelt D 75:2550–2562
(2005) The membrane proteome of Halobac- 11. Rabilloud T (2010) Variations on a theme:
terium salinarum. Proteomics 5:180–197 changes to electrophoretic separations that
5. Kalinowski J, Wolters D, Poetsch A (2008) can make a difference. J Proteome
Proteomics of Corynebacterium glutamicum 73:1562–1572
and other Corynebacteria. From Corynebac- 12. Miller M, Ivano Eberini I, Gianazza E (2010)
teria: genomics and molecular biology Other than IPG-DALT: 2-DE variants. Prote-
(Burkovski A, ed). Caister Academic Press, omics 10:586–610
Norfolk, pp 56–77 13. Macfarlane DE (1983) Use of benzyldimethyl-
6. Aivaliotis M, Corvey C, Tsirogianni I, Karas M, n-hexadecylammonium chloride (16-BAC), a
Tsiotis G (2004) Membrane proteome analysis cationic detergent, in an acidic polyacrylamide
of the green-sulfur bacterium Chlorobium gel electrophoresis system to detect base labile
tepidum. Electrophoresis 25:3468–3474 protein methylation in intact cells. Anal Bio-
7. Moller AJB, Witzel K, Vertommen A, chem 132:231–235
Barkholdt V, Svensson B, Carpentier S Mock 14. Macfarlane DE (1984) Inhibitors of cyclic
HP, Finne C (2011) Plant membrane proteo- nucleotide phosphodiesterases inhibit protein
mics: challenges and possibilities. Sample carboxyl methylation in intact blood platelets.
J Biol Chem 259:1357–1362
68 Gary Smejkal and Srikanth Kakumanu

15. Macfarlane DE (1986) Phorbol diester- 19. Zahedi RP, Moebius J, Sickmann A (2007)
induced phosphorylation of nuclear matrix Two-dimensional BAC/SDS-PAGE for mem-
proteins in HL60 promyelocytes. Possible role brane proteins. In: Bertrand E, Faupel M (eds)
in differentiation studied by cationic detergent Subcellular proteomics: from cell deconstruc-
gel electrophoresis systems. J Biol Chem tion to system reconstruction. Springer, Dor-
261:6947–6953 drecht, pp 13–20
16. Macfarlane DE (1989) Two dimensional 20. Kramer ML (2006) A new multiphasic buffer
benzyldimethyl-n-hexadecylammonium chlo- system for benzyldimethyl-n-hexadecylammo-
ride sodium dodecyl sulfate preparative poly- nium chloride polyacrylamide gel electropho-
acrylamide gel electrophoresis: a high capacity resis of proteins providing efficient stacking.
high resolution technique for the purification Electrophoresis 27:347–356
of proteins from complex mixtures. Anal Bio- 21. Lee I, Salomon AR, Yu K, Samavati L, Pecina P,
chem 176:457–463 Pecinova A, Huttemann M (2009) Isolation of
17. Hartinger J, Stenius K, Högemann D, Jahn R regulatory-competent, phosphorylated cyto-
(1996) 16-BAC/SDS-PAGE: a chrome c oxidase. Methods Enzymol
two-dimensional gel electrophoresis system 457:193–210
suitable for the separation of integral mem- 22. James GT (1978) Inactivation of the protease
brane proteins. Anal Biochem 240:126–133 inhibitor phenylmethylsulfonyl fluoride in buf-
18. Zahedi RP, Meisinger C, Sickmann A (2005) fers. Anal Biochem 86:574–579
Two-dimensional benzyldimethyl-nhexadecy- 23. Singh R, Whitesides GM (1995) Reagents for
lammonium chloride/SDS-PAGE for mem- raid reduction of disulfide bonds in proteins.
brane proteomics. Proteomics 2005 Techniq Protein Chem VI:259–266
(5):3581–3588
 Chapter 4

Systematic Glycolytic Enzyme Activity Analysis from Human

Serum with PEP Technology
David Wang

Abstract
A functional proteomics technology was used to systematically monitor metabolic enzyme activities from
resolved serum proteins produced by a modified 2-D gel separation and subsequent Protein Elution Plate, a
method collectively called PEP. Both qualitative and quantitative differences in the metabolic enzyme
activity could be detected between cancer patient and control group, providing excellent biomarker
candidates for cancer diagnosis and drug development. This technology has a wide range of applications;
it can be used for rapid functional protein purification and characterization as well as drug target identifica-
tion and validation. The ability for the PEP technology to efficiently separate and recover functional
proteins makes it useful for the analysis of any proteins and its variants; this is especially advantageous for
enzyme families with large number of enzymes such as protein kinases, phosphatases, proteases, and
metabolic enzymes.

Key words Functional proteomics, 2-D gel electrophoresis, Protein purification, Biomarkers, Protein
elution plate (PEP), Cancer diagnosis, Drug target identification

1 Introduction

In the last decade, many new technologies have been utilized for
biomarker discovery with significant progress. Each of these tech-
nologies has focused on a different type of biological entity such as
circulating tumor cells (CTC), extracellular vesicles, micro-RNAs
and cancer-derived cell-free DNA or circulating tumor-derived
DNA (ctDNA) [1–9]. However, several fundamental issues such
as tumor heterogeneity, plasticity, and diversity of cancer stem cells
(CSC) make biomarker discovery and development a challenging
endeavor. The variation introduced during sample collection and
storage and the lack of robust validation approach once biomarker
leads are identified further complicate biomarker development
[10–19]. As a result of these hurdles, there are currently no United
States FDA-approved serum tests for early detection of the disease.
Given the considerable public health importance of breast cancer, it

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_4, © Springer Science+Business Media, LLC, part of Springer Nature 2019

69
70 David Wang

is crucial to quickly identify new biomarkers with the potential to

enhance early diagnosis and to predict patient prognosis, drug
resistance development, and treatment choice. Blood-based bio-
markers have great potential in cancer screening and their role
could extend further from general population risk assessment to
treatment response evaluation and recurrence monitoring [20–27].
The rich content of diverse cellular and molecular elements in
blood, which provide information about the health status of an
individual, makes it an ideal compartment to develop noninvasive
diagnostics for cancer. However, despite a large literature collection
related to biomarkers for common cancers, blood-based diagnostic
tests that inform about the presence of cancer at an early stage and
predict treatment response have been difficult to develop.
For the past decade, proteomics has been used for the discovery
of potential biomarkers from human fluids including serum. So far,
most efforts in proteomics has been focused on the identification
and sequence annotation of the proteome by mass spectrometry
analyses of peptides derived through proteolytic processing of the
parent proteome. In such a manner, thousands of proteins have
been identified from human serum (www.serumproteome.org). It
is generally recognized that sequence annotation alone cannot
capture this vital information, so new strategies are necessary.
Two-dimensional (2-D) Gel Electrophoresis is a powerful technol-
ogy to separate complex protein samples. In the first dimension
called isoelectric focusing (IEF), the proteins are separated based
on their isoelectric points (pI), proteins with as little as 0.02 unit pI
differences could be separated, making it a high-resolution
method. In the second dimension, the proteins are separated
based on their molecular size. Because 2-D Gel Electrophoresis is
using two orthogonal parameters (charge and size) for separation
and displaying the proteins in a two-dimensional manner, it is one
of the most powerful technologies in protein separation. In a large
format gel, more than 10,000 proteins could be separated and
detected with information on their relative abundance and post-
translational modification acquired simultaneously. Because of
these advantages, 2-D Gel Electrophoresis has been used widely
in proteomics studies. However, in a typical 2-D Gel Electrophore-
sis, the proteins are denatured by the addition of reagents to disrupt
disulfide bonds (DTT or β-mercaptoethanol), chemicals to prevent
disulfide bond formation (iodoacetamide) and high concentration
of SDS (typically 1%). To keep the proteins active in 2-D Gel
Electrophoresis, a few modifications were made in the current
PEP technology. First, no reducing reagent is used in the isoelectric
focusing step, keeping the disulfide bonds in the proteins intact.
Secondly, iodoacetamide is omitted from the process. Thirdly,
much reduced SDS concentration is used in the SDS-PAGE (from
1% to 0.1%) or no SDS was used at all, again trying to maintain
enzymatic activity. Recent studies indicated that many different
 Systematic Glycolytic Enzyme Activity Analysis from Human Serum with PEP. . . 71

enzyme families from a wide variety of organisms are active in the

presence of SDS such as protein kinases, protein phosphatases,
proteases, and oxido-reductases.
In addition of method modification, a high-resolution Protein
Elution Plate (PEP) was designed. The small format PEP has
384-wells matching the current 384-well microplate dimension
for ease of sample processing. For the large format PEP, the plate
is composed of 4 384-well PEP, thus having 1536 wells. In both the
large and small format PEP, a membrane with molecular cutoff of
6000 Dalton is attached that will allow the electric current and
small charged molecules to pass through but collect proteins with
molecular weight large than 6000 Dalton in the PEP wells. Fur-
thermore, a special solution was developed for the PEP to reduce
protein diffusion after the proteins are transferred from the gel to
the PEP. After transferring the solutions from PEP to a deep-well
master plate, the enzyme activity or protein function can be ana-
lyzed using part of the sample from the master plate and purified
protein can be verified using SDS-PAGE in standard condition and
identified using mass spectrometry. It is hypothesized that the levels
and distributions of certain enzyme functions in serum could pro-
duce proteomic features and collective profiles which reflect physi-
ological changes of an individual and can serve as possible
biomarkers or diagnostic parameters [28–38]. In this chapter, we
summarized the use of PEP technology for the systematic analysis
of metabolic enzyme from human serum. We believe that the
identification and validation of those functional proteins from
human serum could lead to the development of biomarkers for
cancer diagnosis. The PEP technology can also be used for the
discovery of functional biomarkers for other diseases as well as
drug target identification and drug safety evaluation (Fig. 1).

2 Materials

2.1 Chemicals All the chemicals were purchased from MilliporeSigma (St. Louis,
MO). Isoelectric focusing (IEF) unit that is capable of running IEF
at different length is from Bio-Rad (PROTEAN IEF Cell, Hercu-
les, CA). Spectrophotometer Plate Reader capable of reading
384-well plates with a wide wavelength selection and fluorescence
reading is the SPECTRAMax Plus from Molecular Devices (Sunny-
dale, CA). Semi-Blot unit for protein transfer was Bio-Rad’s Trans-
Blot SD Semi-Dry Transfer Cell. AlbuVoid™ serum protein enrich-
ment beads were from Biotech Support Group (Monmouth Junc-
tion, NJ). Protein Elution Plate (PEP) is a product of Array Bridge
(St. Louis, MO).
1. SDS-PAGE gels: Customer can choose any format SDS-PAGE
gel to run the sample. For 1-D gel, preferentially the loading
72 David Wang

Fig. 1 Diagram of the PEP Technology (adapted from Wang DL et al., PLoS One, 2015, 10(3) with permission)

capacity of each well should be 15 μL or more. Gels from Bio-

Rad (Criterion 10–20% 18 well Tris–HCl gel, catalog number:
345-0043) was used in our studies. For 2-D gel separation,
Criterion 10–20% IPG + 1 well Tris–HCl gel (Bio-Rad, catalog
number: 345-0107) or similar gels from Invitrogen, etc. can be
used for the protein separation.
2. Isoelectric focusing strips: Immobilized pH gradient (IPG) strips
to run IEF can be purchased from either Bio-Rad (Catalog
Number: 163-2014 for 11 cm IPG strips and 163-2033 for
18 cm IPG strips) or GE Health Life Sciences (Catalog Num-
ber: 18101661 for 11 cm pH 3–10 Immobiline Dry Strips;
17123501 for 18 cm, pH 3–10 Nonlinear Immobiline Dry
Strips).
3. Electrolyte: Electrolyte used for the IEF gel can be purchased
from either Bio-Rad (Bio-Lyte buffer, pH 3–10, catalog num-
ber: 163-2094) or GE Health (Pharmalyte pH 3–10, catalog
number: 17-0456-01).
 Systematic Glycolytic Enzyme Activity Analysis from Human Serum with PEP. . . 73

4. Protein staining components: If protein staining is required, the

following conditions can be used: gels after electrophoresis first
fixed in fixing solution (10% acetic acid, 10% ethanol in Mili-Q
water) for 1 h, then stained in SYPRO Orange (Invitrogen,
catalog number: S6650) or other fluorescence dye overnight in
Mili-Q water, dilute the fluorescence dye as recommended by
the manufacturer.
5. Single- and multi-channel micro-pipettes with disposable tips to
accurately dispense volumes 5–250 μL. Plastic tubes (i.e.,
1.5–15 mL) for sample dilution. Reagent reservoirs for sample
addition.

2.2 Glycolytic The glycolytic enzymes from human serum were detected by mea-
Enzyme Activity Assay suring the first enzyme in the glycolytic pathway, hexokinase. By
using a beef liver extract to provide a low and basal level of glyco-
lytic enzymes, any additional enzymes from the PEP-eluted serum
sample can be detected by the increased hexokinase activity. There-
fore the measurement of the glycolytic enzyme activity from PEP
samples was calculated by the increased hexokinase activity from the
basal level of beef liver extract instead of the total hexokinase
activity.
Hexokinase activity can be monitored by a cascade reaction as
follows:
Hexokinase

Substrates added fD‐Glucose þ ATPg

! Products fD‐Glucose 6‐Phosphate þ ADPg
G‐6‐PDH
D‐Glucose 6‐Phosphate þ β‐NADP
! 6‐Phospho‐D‐Gluconate þ β‐NADPH
In the final assay solution, glucose was at 216 mM; MgCl2 at
7.8 mM, ATP at 0.74 mM, and NADP at 1.1 mM. 25 μL of this
enzyme assay solution was mixed with 25 μL of sample from the
Master Plate (described below) and the enzyme activity was moni-
tored by the increased 340 nm absorbance from the reduction of
NADP to NADPH. The readings at different time points such as
0, 1, and 2 h were recorded for both the normal serum and breast
cancer patient serum sample. However, in lieu of purified G-6-
PDH used for the hexokinase assay, 0.25 mg/mL beef liver protein
was used as the source of glucose-6-phosphate dehydrogenase
(G-6-PDH). The assay thus reports the additive contributions of
the endogenous hexokinase activity present in the beef liver extract,
and any exogenous activity from the presence of test sera protein in
the PEP plate, which may influence the reduction of NADP (the
reporting signal). In light of the ambiguities that may arise from
74 David Wang

Fig. 2 Measurement of hexokinase activity from normal and breast cancer serum (adapted from Wang DL
et al., BMC Biomarker Research, 2017, 5(11) with permission)

such a reporting system, the primary goal of this investigation was

to generate sufficient signal intensities and activity features which
could be monitored and compared between the two samples types
within an “omics” context. Therefore, this broader spectrum assay
was chosen that could potentially detect the activities of hexokinase
and downstream glycolytic enzymes and other cross-regulating
proteins from the test sera (Fig. 2).

2.3 Supplied 1. 384-well PEP plate. A PEP plate is provided. The plate was
Components from the treated with a special solution to reduce the binding of the
PEP Universal Protein transferred protein and increase the recovery efficiency.
Purification Kit (Small 2. 384-well mater plate. A deep-well plate is provided to contain
Format PEP) from samples recovered from the PEP plate.
Array Bridge Inc. 3. 384-well enzyme assay plate. A standard 384-well polypropylene
(Catalog Number: plate is provided for enzyme assay to identify which wells
AB-000401) contain the protein of interest.
4. 10 protein transfer buffer (50 mL). Buffer used for running
the modified SDS-PAGE or the second dimension of 2-D gel,
 Systematic Glycolytic Enzyme Activity Analysis from Human Serum with PEP. . . 75

and also used to wet the filter papers for the transfer of proteins
from the gel to the PEP plate.
5. 10 PBS (10 mL). Buffer used for the pretreatment of the
master plate and fill in each well of the Master Plate with
50 μL of PBS.
6. Standard SDS-PAGE sample buffer (0.5 mL). Solution used in
sample treatment for the standard SDS-PAGE to check enzyme
fraction purity.
7. PEP plate protein recovery buffer (25 mL). Solution used in the
PEP plate to recover proteins eluted from the gel and prevent
protein diffusion.
8. Plate sealer. For sealing the Master plate and the enzyme assay
plate during the purification process. Kit AB-00402 (two).
9. Filter papers. Used to form a sandwich in the protein transfer
process.

2.4 Instruments Gel electrophoresis unit includes power supply and gel unit.
Isoelectric focusing unit that is capable of running IEF at different
length, an example of such unit is the Bio-Rad PROTEAN IEF
Cell (Catalog Number: 165-4000).
Spectrophotometer Plate Reader capable of reading 384-well plates
with a wide wavelength selection and fluorescence reading.
Semi-Blot unit for protein transfer such as Bio-Rad’s Trans-Blot SD
Semi-Dry Transfer Cell (Catalog Number: 170-3940).

3 Methods

3.1 Sample High concentration of salt will interfere with the isoelectric focus-
Treatment ing step. If the protein concentration is less than 5 mg/mL and the
salt concentration is more than 100 mM, it is recommended to
dialyze the samples in 5 mM phosphate buffer, pH 7.2 before use or
using desalting column to reduce the salt concentration.

3.2 Running the IEF 1. It is suggested to use the 11 cm IPG strip (Bio-Rad, catalog
Gel First Followed by number: 163-2033) for the IEF. To rehydrate one IPG strip,
Running the Native or 225 μL of solution is needed. It is suggested to use 200 μL of
Modified SDS-PAGE samples with up to 200 μg total protein, add urea to a final
concentration of 8 M, add 2 μL of Ampholyte such as Bio-lyte
(Bio-Rad, catalog number: 163-2094). If the protein sample
can be lyophilized, then the lyophilized sample can be dissolved
into a sample solution with 8 M urea and 0.5% Bio-lyte.
2. The solution is first added to a rehydration tray, the IPG strip is
taken out from the storage, and the plastic cover is peeled off.
The side with the dried gel surface is facing down to make
76 David Wang

contact with the sample solution in the rehydration tray. Please

make sure to let the whole IPG strip making full contact to the
sample solution. Add enough mineral oil to cover the IPG strip
to prevent evaporation and rehydrate the sample overnight at
room temperature (sometimes, a 6 h rehydration is enough for
the IEF, this is especially important if the enzyme of interest is
not stable at room temperature).
3. After rehydration, the IPG strip is taken out from the rehydra-
tion tray and the attached mineral oil is carefully removed with
a Kim wipe paper.
4. In the IEF tray, carefully wet two pieces of Electrode Wick
(Bio-Rad, catalog. No. 165-4071) and put on the metal wire
in one lane. Carefully lay down the IPG strip face down, and
gently push the IPG strip so that it can make a close contact to
the filter paper-covered metal wire. Add enough mineral oil to
cover the IPG strip to prevent the evaporation.
5. Put the IEF tray cover to the try followed by close the IEF unit
cover (Bio-Read Protean IEF Unit).
6. In the first step, set the voltage gradient from 0 to 8000 V for
4 h; in the second step, set at constant voltage at 8000 for 24 h.
The gel will actually run overnight, but the minimum voltage-
hours are 30,000 for a good 2-D separation.
7. After the IEF is completed, turn off the unit, and carefully take
out the IPG strip and use Kim wiper paper to remove the
mineral oil from the IPG strip. Put the IPG strip into a rehy-
dration tray and incubate in Tris-Glycine transfer buffer sup-
plied in this kit, incubate for 10 min to remove the urea, and
allow the SDS to bind to the proteins (if the enzyme is sensitive
to SDS, the incubation can be carried out in Tris-Glycine only,
this will also produce acceptable protein resolution).
8. Take out a Bio-Rad Criterion gel and remove the plastic comb,
use Milli-Q water to rinse the flat well. Put the gel into the
running unit and fill both the lower and upper tank with Tris-
Glycine-SDS buffer (if the enzyme is sensitive to SDS, only
Tris-Glycine buffer will be used).
9. Carefully lay down the IPG strip in the IPG well with the acidic
side always on the left side when facing the gel. Load 5 μL of
unstained protein standard in the protein standard well (the
well next to the acidic end of the IPG strip).
10. Run first at 80 voltages for 15 min followed by 120 voltages
until the dye front from the protein standard is about 0.5 cm
from the bottom of the gel (it is important to run 80 voltages
for 15 min to allow as much of the proteins in the IEF gel
getting into the second dimensional gel).
 Systematic Glycolytic Enzyme Activity Analysis from Human Serum with PEP. . . 77

3.3 Protein Transfer 1. While the SDS-PAGE is still running, put the PEP plate in a
After the 2-D Gel tray and add 50 μL of the protein recovery solution to each well
of the plate with a multiple channel pipette, there will be some
overflow of the solution during this step, it is fine. If an eight
channel pipette is used, the solution could be dispensed every
other row. For example, in the first round, add solutions to
row A, C, E, and so on; in the second round, add solution in
the row B, D, F, and so on. Cover the tray to minimize
evaporation.
2. When the dye front is about 0.5 cm from the bottom of the gel,
stop the running and carefully take out the gel from the gel
cassette and rinse with Milli-Q water followed by adding
200 mL of the transfer buffer (supplied with the kit) in the
tray. Wet four pieces of the transfer filter paper (supplied with
the kit) completely in a different tray and lay down two pieces
on the metal plate of the Semi-Dry Trans-Blot (Bio-Rad or
similar Semi-Dry Trans-Blot from other manufacturers).
3. Lay the PEP plate on top of the filter papers followed by
carefully lay the gel on top of the PEP plate and make sure
the upper left corner of the gel align with the upper left corner
of the PEP plate.
4. Lay another two pieces of transfer filter paper on top of the gel
to form a sandwich (from the bottom it should be filter papers,
PEP plate, gel and filter papers again).
5. Cover the sandwich assembly with the other metal plate of the
Semi-Dry Trans-Blot, and transfer the proteins with constant
voltage at 20 voltages for 60 min. It is shown that under this
condition, the proteins in the gel will be efficiently transferred
into the PEP plate, longer protein transfer is not
recommended.
6. While the gel is transferring, condition the 384-well deep-well
plate by adding 100 μL PBS in each well (if protein kinase or
protein phosphatase assays are performed, a phosphate-free
buffer such as Tris–HCl should be used to minimize interfer-
ence from the buffer). This treatment will improve the protein
recovery in later steps for enzyme activity analysis and mass
spectrometry protein identification. After 30 min treatment,
completely empty the solution from each well and refill the well
with 50 μL PBS (for protein kinase assay or any other assay
where phosphate is interfering, Tris–HCl buffer or other buffer
of choice could be used).
7. When the protein transfer is completed, turn off the power,
take off the Semi-Dry Trans-Blot cover and release the top
metal plate. Wait for 10 s before lifting the top metal plate
(this is important to let some air in so that the solutions in the
PEP plate will not be sucked out to cause proteins in one well
78 David Wang

over flow to adjacent wells). After removing the metal plate,

carefully lift the two pieces of filter paper followed by remove
the gel (sometimes the filter papers and gel will stick together,
in this case lift both parts together). When removing the gel, it
is important to remove it from left to right, it should be point
out that the specific composition of the PEP transfer buffer will
reduce the protein diffusion. Carefully take the PEP plate
without the two transfer paper on the bottom of the PEP
plate and put in a tray.
8. Use multiple channel pipettes to transfer the recovered protein
solution from the PEP plate to the deep-well Master Plate in
the corresponding columns. If using an eight channel pipette,
set the transfer volume at 45 μL to make sure most of the
solution in the well is transferred. The transfer will start at
column 1 from the left side of the PEP plate, and the wells
with odd numbers (row A, C, E, and so on) is first transferred
followed by transfer of the wells from the even number wells in
the first column (rows B, D, F, and so on). Repeat the process
until all the samples from the PEP plate are transferred to the
384-well Master Plate.

3.4 Glycolytic 1. After the transfer of samples from the PEP plate to the deep-
Enzyme Analysis well Master Plate, the Master Plate should be used immediately
(preferred) for glycolytic enzyme analysis. Multiple enzymes
can be analyzed from samples collected since the total volume
in each well of the Master Plate is about 90 μL (50 μL buffer
plus 40–45 μL sample transferred from the PEP plate). Refer to
Subheading 2.2 for details of the glycolytic enzyme assay pro-
tocol. Use a spectrophotometer to measure the glycolytic
enzyme activities at 340 nm. Before the assay readout, some
wells of the enzyme assay plate may contain bubbles because of
the SDS in the protein transfer buffer (one technique to avoid
bubble is to set the dispensing volume smaller than the aspirat-
ing volume so that the pipette will not cause bubbles when
dispensing). Use a transfer pipette tip to pinch the bubbles to
get rid of them before the reading; this will reduce the interfer-
ence from the bubble.
2. When reading the enzyme assay plate, use pipette to remove
the solutions from the well P24 (lower right corner well in the
384-well plate) and use this well as blank for the reading. It is
recommended to read at least 3 data points such as 0, 60, and
120 min and save the reading in separate files.

3.5 Data 1. Export the data set from the three readings (0, 60, and
Transformation and 120 min) to an Excel file (if not already in this format).
Analysis 2. In Microsoft Excel, subtract the 60 min readings from each
well from the corresponding readings at 0 min to obtain the
 Systematic Glycolytic Enzyme Activity Analysis from Human Serum with PEP. . . 79

data set for the 340 nm absorbance difference which reflect the
glycolytic enzyme activities from the serum proteome. Use
Excel Heat Map to display the enzyme activity in a 384-well
table or use Insert function, and select the 3-D display to build
the graph of this data set in 3-D display.
3. Subtract the 120 min readings from each well from the
corresponding readings at 0 min to obtain the data set for the
340 nm absorbance difference which reflect the glycolytic
enzyme activities from the serum proteome for the second
data set.

3.6 Protein Purity 1. If the enzyme testing showed that some wells have the enzyme
Confirmation activity of interest, the next step is to test the purity of the
(Optional) protein in that well. Collect all the samples from the wells with
enzyme activity in a siliconized microcentrifuge tube, dry down
the solution, and resuspend into 20 μL of Milli-Q water. Take
10 μL and mixed with 10 μL of SDS sample buffer (this sample
buffer is a 2 SDS-PAGE sample buffer with 20 mM DTT),
incubate at 37  C for 60 min.
2. Load on a SDS-PAGE gel and run the gel as in Subheading 2 of
this protocol.
3. Fix the gel in a gel-fixing solution (10% each of ethanol and
acetic acid in Milli-Q water) for at least 2 h.
4. Rinse with distilled water and stain the gel in Sypro Ruby or
other fluorescence dye overnight.
5. The next day, remove the staining solution; wash the gel twice
with distilled water followed by incubation in the distilled
water for 5 min with moderate shaking.
6. Take the gel image with a CCD camera such as the Bio-Rad
ChemiDoc.
7. Save the image in tiff file for later image processing. The gel
image will tell whether the protein is pure or not.

3.7 Mass 1. If the gel staining in Subheading 3.6 shows that the fraction
Spectrometry to with enzyme activity is pure, the 10 μL Milli-Q water resus-
Identify the Protein of pended sample in Subheading 3.6, step 1 can be submitted for
Interest mass spectrometry analysis (sometimes fraction with more than
one protein bands can be submitted for MS analysis, and the
identity of the protein can be assigned by bioinformatics effort
based on protein homology, it is unlikely that more than one
protein from the preparation share the same type of enzyme
activity, for example, GAPDH).
2. Alternatively, if there is enough protein to be seen in Subhead-
ing 3.6, step 7 with the fluorescence staining, the protein band
can be excised and sent for MS analysis.
80 David Wang

4 Notes

1. Allow diluted reagents and buffers to reach room temperature

(18–25  C) prior to starting the assay. Once the assay has been
started, all steps should be completed in sequence and without
interruption. Make sure that required reagents and buffers are
ready when needed. Prior to adding to the plate, reagents
should be mixed gently (not vortexed) by swirling.
2. Avoid contamination of reagents, pipette tips and wells. Use
new disposable tips and reservoirs, do not return unused
reagent to the stock bottles/vials, and do not mix caps of
stock solutions.
3. For some enzymes, 8 M urea might be too strong for its
enzymatic activity; in this case, 3 M urea and 2% CHAPS will
be used in the IEF gel. If the presence of SDS also inactivates the
enzyme, the second dimension separation can use
non-denaturing gels. The gel resolution needs to be tested for
the modified conditions before PEP elution and enzyme assay.
4. Sometimes the IEF does not have to run as high as 8000
Voltage as was suggested by the manufacturer, it has been
found that setting the highest voltage at 5000 can also achieve
good protein separation.

Acknowledgments

I would like to thank Array Bridge Inc. for the supply of the PEP
Universal Protein Purification kits and the opportunity to carry out
this research in its laboratory. I would also like to thank Dr. Liang Li
for providing the breast cancer and normal people sera in this
research.

References
1. Dos Anjos Pultz B et al (2014) Far beyond the 5. Henderson MC et al (2016) Integration of
usual biomarkers in breast cancer: a review. J serum protein biomarker and tumor asso-
Cancer 5(7):13 ciated autoantibody expression data
2. Li J et al (2002) Proteomics and bioinformatics increases the ability of a blood-based proteo-
approaches for identification of serum biomar- mic assay to identify breast cancer. PLoS
kers to detect breast cancer. Clin Chem 48(8):9 One:11(8)
3. Chan MK, Cooper JD, Bahn S (2015) Com- 6. Ingvarsson J et al (2007) Design of recombi-
mercialisation of biomarker tests for mental ill- nant antibody microarrays for serum protein
nesses: advances and obstacles. Trends profiling: targeting of complement proteins. J
Biotechnol 33(12):12 Proteome Res 6:10
4. Chung L et al (2014) Novel serum protein 7. Lee JS, Magbanua MJM, Park JW (2016) Cir-
biomarker panel revealed by mass spectrometry culating tumor cells in breast cancer: applica-
and its prognostic value in breast cancer. Breast tions in personalized medicine. Breast Cancer
Cancer Res 16:R63 Res Treat 160:411–424
 Systematic Glycolytic Enzyme Activity Analysis from Human Serum with PEP. . . 81

8. Mehan MR et al (2014) Validation of a blood in cancer development and progression. Lancet

protein signature for non-small cell lung can- Oncol 5:9
cer. BMC Clin Proteomics 11(32):12 24. Wang X et al (2015) Bead based proteome
9. Ross JS et al (2003) Breast cancer biomarkers enrichment enhances features of the protein
and molecular medicine. Expert Rev Mol elution plate (PEP) for functional proteomic
Diagn 3(5):13 profiling. Proteomes 3:13
10. Ross JS et al (2004) Breast cancer biomarkers 25. Amorim M et al (2016) Decoding the useful-
and molecular medicine: part II. Expert Rev ness of non-coding RNAs as breast cancer mar-
Mol Diagn 4(2):20 kers. J Transl Med 14:15
11. Surinova S et al (2015) Prediction of colorectal 26. Mabert K et al (2014) Cancer biomarker dis-
cancer diagnosis based on circulating plasma covery: current status and future perspectives.
proteins. EMBO Mol Med 7(9):13 Int J Radiat Biol 90(8):18
12. Yezhelyev MV et al (2007) In situ molecular 27. Surinova S et al (2015) Non-invasive prognos-
profiling of breast cancer biomarkers with mul- tic protein biomarker signatures associated
ticolor quantum dots. Adv Mater 19:6 with colorectal cancer. EMBO Mol Med 7:13
13. Kirmiz C et al (2007) A serum glycomics 28. Orla T et al (2011) Metabolic signatures of
approach to breast cancer biomarkers. Mol malignant progression in prostate epithelial
Cell Proteomics 6:13 cells. Int J Biochem Cell Biol 43:8
14. Harsha HC et al (2009) A compendium of 29. Teicher BA, Marston WL, Helman LJ (2013)
potential biomarkers of pancreatic cancer. Targeting cancer metabolism. Clin Cancer Res
PLoS Med 6(6):6 18(20):9
15. Kaskas NM et al (2014) Serum biomarkers in 30. Araujo EP, Carvalheira JB, Velloso LA (2006)
head and neck squamous cell cancer. JAMA Disruption of metabolic pathways—perspec-
140(1):7 tives for the treatment of cancer. Curr Cancer
16. Wang C-H et al (2015) Current trends and Drug Targets 6:77–87
recent advances in diagnosis, therapy and pre- 31. Bryksin AV, Laktionov PP (2008) Role of glyc-
vention of hepatocellular carcinoma. Asian Pac eraldehyde-3-phosphate dehydrogenase in
J Cancer Prev 16(9):10 vesicular transport from golgi apparatus to
17. Alexander H et al (2004) Proteomic analysis to endoplasmic reticulum. Biochemistry 73:7
identify breast cancer biomarkers in nipple aspi- 32. Cairns RA, Harris IS, Mak TW (2011) Regula-
rate fluid. Clin Cancer Res 10:11 tion of cancer cell metabolism. Nat Rev Cancer
18. Ma S et al (2016) Multiplexed serum biomar- 11:11
kers for the detection of lung cancer. EBio Med 33. Chaneton B, Gottlieb E (2012) Rocking cell
11:9 metabolism: revised functions of the key glyco-
19. Evens MJ, Cravatt BF (2006) Mechanism- lytic regulator PKM2 in cancer. Trends Bio-
based profiling of enzyme families. Chem Rev chem Sci 37(8):7
106:23 34. Chang C-H et al (2015) Metabolic competi-
20. Wang DL et al (2015) Identification of multi- tion in the tumor microenvironment is a driver
ple metabolic enzymes from mice cochleae tis- of cancer progression. Cell 162:13
sue using a novel functional proteomics 35. Chiaradonna FR et al (2012) From cancer
technology. PLoS One 10:e0121826 metabolism to new biomarker and drug tar-
21. Wang DL et al (2017) Identification of poten- gets. Biotechnol Adv 30:30–51
tial serum biomarkers for breast cancer using a 36. Favaro E et al (2012) Glucose utilization via
functional proteomics technology. Biomark glycogen phosphorylase sustains proliferation
Res 5:11 and prevents premature senescence in cancer
22. Sun Z et al (2016) Identification of functional cells. Cell Metab 16:14
metabolic biomarkers from lung cancer patient 37. Ledford H (2014) Metabolic quirks yield
serum using PEP technology. Biomark Res tumour hope. Nature 508:2
4:11 38. Anderson NL, Anderson NG (2002) The human
23. Sun Z, Yang P (2004) Role of imbalance plasma proteome: history, character, and diag-
between neutrophil elastase and a1-antitrypsin nostic prospects. Mol Cell Proteomics 1:23
 Chapter 5

A Protein Decomplexation Strategy in Snake Venom

Proteomics
Choo Hock Tan, Kae Yi Tan, and Nget Hong Tan

Abstract
Snake venoms are complex mixtures of proteins and peptides that play vital roles in the survival of venomous
snakes. As with their diverse pharmacological activities, snake venoms can be highly variable, hence the
importance of understanding the compositional details of different snake venoms. However, profiling
venom protein mixtures is challenging, in particular when dealing with the diversity of protein subtypes
and their abundances. Here we described an optimized strategy combining a protein decomplexation
method with in-solution trypsin digestion and mass spectrometry of snake venom proteins. The approach
involves the integrated use of C18 reverse-phase high-performance liquid chromatography (RP-HPLC),
sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and nano-electrospray ionization
tandem mass spectrometry (nano-ESI-LC-MS/MS).

Key words Snake venom, Protein decomplexation, Venom separation, Reverse-phase high-
performance liquid chromatography, Tandem mass spectrometry, Venomics

1 Introduction

The advent of proteomics has greatly facilitated the investigation of

venom protein composition in a high-throughput and comprehen-
sive manner. For a decade, the term “venomics” has been used with
increasing popularity to denote venom-related “-omics” studies
including snake venom proteomics [1, 2]. Prior to the venomic
era, bioassay-guided protein purification was the main platform
available to identify and characterize proteins in a snake venom,
but this method was akin to finding pieces of puzzle at a time and
complete protein profiling of the snake venom was hardly possible.
The application of proteomics and bioinformatics has now enabled
the study of the global profiling of venom proteins in great details,
even for components that exist in a very low amount [3, 4]. This
revolutionizing breakthrough by venomics has propelled the
growth of knowledge tremendously on the various aspects of
snake venom research including venom evolution, envenomation

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_5, © Springer Science+Business Media, LLC, part of Springer Nature 2019

83
84 Choo Hock Tan et al.

pathophysiology, antivenom production, and in toxin-based drug

discovery [5–7].
However, snake venoms are complex mixtures of proteins and
peptides which are inherently variable [8, 9]. The depth of proteo-
mic findings for a venom also tends to vary with different experi-
mental protocols, equipment or techniques used, thus posing a
challenge when one intends to collate and compare global venomic
data. To obtain as much useful proteomic information from a
venom, it is therefore important that the protocol provides good
resolution of the proteins [10]. This can be achieved through a
protein separation method prior to the mass spectrometric analysis,
as widely adapted in many venomic studies [11–13]. Protein sepa-
ration is typically achieved by either a gel-based method such as
SDS-PAGE or two-dimensional gel electrophoresis (separation by
protein differences in isoelectric point and molecular mass), or
liquid chromatography using various columns (separation by pro-
tein differences in ionic charges, hydrophobicity or molecular mass)
[14–18]. Often, the chromatographic separation method, in par-
ticular the use of C18 reverse-phase column, is preferred over the
gel-based method for better protein resolution and advantage in
the estimation of protein abundance based on peak areas (the area
under the curve) [13, 19]. Venom proteins bind to the reverse-
phase column (stationary phase) through hydrophobic interaction,
and in general the more hydrophobic proteins will bind stronger to
the C18 beads in the column. The mobile phase is composed of an
aqueous blend of water with a miscible, polar organic solvent, e.g.,
acetonitrile, delivered under high pressure. The flow of the mobile
phase elutes the venom proteins following a stepwise increase of
acetonitrile concentration over an extended time. Proteins are col-
lected into different fractions as they are eluted, and the proteins
can be visualized on SDS-PAGE. The protein fractions are then
subjected to liquid chromatography tandem mass spectrometry and
data mining for protein identification and proteome construction.
Thus far, we have reported a number of quantitative snake venom
proteomics based on this approach and have found that the results
provide good functional correlation and insights into the complex-
ity of snake venoms and toxins.

2 Materials

2.1 Snake Venom Freeze-dry snake venom samples and store at
20  C until use.
Samples

2.1.1 Reverse-Phase Stationary phase: Reverse-phase HPLC column LiChroCART®

High-Performance Liquid 250-4 LiChrospher® WP 300 (Merck, USA), or any equivalent
Chromatography column.
(RP-HPLC)
 A Protein Decomplexation Strategy in Snake Venom Proteomics 85

Mobile phase: Equilibrium buffer (Eluent A): 0.1% trifluoroacetic

acid (TFA) in HPLC grade water. Add 1 mL TFA into 999 mL of
HPLC grade water. Elution buffer (Eluent B): 0.1% trifluoroacetic
acid (TFA) in HPLC grade acetonitrile (ACN). Add 1 mL TFA into
999 mL of HPLC grade ACN.

2.1.2 Sodium Dodecyl 30% (weight/volume) Acrylamide/Bis-acrylamide (29.2%:0.8%)

Sulfate-Polyacrylamide Gel solution: Weigh 29.2 g of Acrylamide monomer and 0.8 g of
Electrophoresis Bis-acrylamide (cross-linker) and transfer to a 100 mL Scott bottle
(SDS-PAGE) containing 50 mL of ddH2O. Add a magnetic stirring bar
(20  6 mm) and allow the mixture to mix for 10 minutes. Make
up the solution to 100 mL with ddH2O. Store at 4  C.
Sodium dodecyl sulfate (SDS)-containing resolving gel buffer: 1.5 M
Tris–HCl, pH 8.8. Weigh 181.7 g Tris–HCl and 4.0 g of SDS.
Transfer to a 1 L Scott bottle and add ddH2O to a volume of
900 mL. Mix and adjust the pH with HCl. Make up the solution
to 1 L with ddH2O. Store at 4  C (see Note 1).
SDS-containing stacking gel buffer: 0.5 M Tris–HCl, pH 6.8.
Weigh 60.6 g Tris–HCl and 4.0 g of SDS. Prepare a 1 L solution
as described in the previous step. Store at 4  C (see Note 1).
10% (w/v) ammonium persulfate (APS): Weigh 30 mg of APS and
transfer to a 1.5 mL centrifuge tube. Add 300 μL ddH2O into the
tube and dissolve it completely (Freshly prepared).
N,N,N,N0 -Tetramethyl-ethylenediamine (TEMED): Store at 4  C.
Electrophoresis buffer: 0.025 M Tris–HCl, pH 8.3, 0.192 M glycine,
0.1% SDS.
Sample loading buffer (1): 62 mM Tris–HCl (pH 6.8), 2.3%
(w/v) SDS, 5% (w/v) beta-mercaptoethanol, 0.005% (w/v) bro-
mophenol blue, 10% (w/v) glycerol.
Gel staining and fixing solution: 0.2% (w/v) Commassie blue
R-250, 40% (v/v) methanol, 10% (v/v) acetic acid in ddH2O.
Gel destaining solution: 5% (v/v) methanol, 7% (v/v) acetic acid in
ddH2O.

2.1.3 Protein Digestion Trypsin stock (0.1 μg/μL): Add 200 μL of ddH2O to 20 μg lyophi-
(In-Solution Tryptic lized trypsin with 1 mM HCl.
Digestion) Digestion buffer: 50 mM ammonium bicarbonate.
Reducing buffer: 100 mM dithiothreitol (DTT).
Alkylation buffer: 100 mM iodoacetamide (IAA).

2.1.4 Peptides Extraction Materials:

and Desalting
Millipore ZipTip® C18 Pipette Tips were supplied by Merck (USA).
86 Choo Hock Tan et al.

Solution:
Wetting solution: 50% acetonitrile (ACN).
Equilibrium/wash solution: 0.1% formic acid (FA).

Elution solution: 0.1% FA in 50% ACN.

3 Methods

3.1 Protein 1. System equilibration: Attach a C18 column (LiChroCART®

Separation 250-4 LiChrospher® WP 300) to the HPLC system. Equili-
brate the C18 column with Eluent B for 30 min, followed by
3.1.1 Reverse-Phase
Eluent A for 30 min.
High-Performance Liquid
Chromatography 2. Sample preparation: Weigh 2 mg of lyophilized venom and
(RP-HPLC)—Shimadzu transfer into a 1.5 mL centrifuge tube. Add 200 μL of 0.1%
LC-20AD HPLC System TFA into the tube and centrifuge at 10,000  g for 12 min at
(Japan) 4  C. Transfer the supernatant to a new centrifuge tube.
3. Sample fractionation: Inject 200 μL supernatant into the injec-
tion loop (at loading position). Elute the venom sample with a
linear gradient of 5% Eluent B for 10 min, 5–15% Eluent B over
20 min, 15–45% Eluent B over 120 min, and 45–70% Eluent B
over 20 min (see Note 2). Monitor the venom protein elution
by UV absorbance at 215 nm. The fractionation is conducted at
room temperature (20–24  C).
4. Fractions collection: Collect all protein fractions (according to
absorbance measurement) manually. Freeze-dry all fractions
obtained and stored at
20  C until use.
*Figure 1 shows a typical C18 RP-HPLC profile of a cobra venom
under the above experimental conditions.

3.2 Protein 1. Mix 4.5 mL of SDS-containing resolving gel buffer, 3.0 mL of

Visualization acrylamide mixture, and 1.5 mL of ddH2O in a 15-mL centri-
fuge tube. Add 100 μL of 10% APS and 10-μL of TEMED
3.2.1 15% Sodium
right before gel casting. Cast the gel immediately within a
Dodecyl Sulfate-
7.25 cm  10 cm  1.5 mm gel cassette (see Note 3). Layer
Polyacrylamide Gel
the stacking gel with isopropanol (see Note 4).
Electrophoresis
(SDS-PAGE) 2. Prepare stacking gel by mixing 1.25 mL of SDS-containing
stacking gel buffer, 0.7 mL of acrylamide mixture, and
3.05 mL of ddH2O in a 15-mL centrifuge tube. Add 80 μL
of 10% APS and 8 μL of TEMED right before the gel casting.
Insert a 10-well gel comb immediately without introducing air
bubbles.
3. Reconstitute the lyophilized fractions collected from
RP-HPLC (Subheading 3.1.1, step 4) with ddH2O and deter-
mine the protein concentration of each fraction with
 A Protein Decomplexation Strategy in Snake Venom Proteomics 87

Fig. 1 Reverse-phase HPLC fractionation of snake venom using LiChrospher® WP 300 C18 column (upper
panel) following the chromatographic condition: 5% B for 10 min, 5–15% B for 20 min, followed by 15–45% B
for 120 min and 45–70% B for 20 min. The chromatographic fractions are collected manually at 215 nm
absorbance and the lyophilized fractions are further electrophoresed on SDS-PAGE (lower panel, under
reducing conditions). Protein marker is used for molecular weight calibration. The protein bands are visualized
by Coomassie blue staining

NanoDrop Spectrophotometer (Thermo Scientific, USA). Add

sample loading buffer to venom fractions (5–50 μg) in a
one-to-one volume ratio (keep the total volume less than
20 μL). Place the mixture in boiling water for 10 min and
cool the sample to ambient temperature. Centrifuge the sample
at 6000  g for 30 s to bring down the condensate.
88 Choo Hock Tan et al.

4. Remove the gel comb prior to sample loading. Load protein

marker on the left side of the gel and heated samples in the
subsequent wells. Electrophorese the samples at 90 V until the
front dye reaches the bottom of the gel.
5. Following electrophoresis, remove the gel cassette from the
electrophoresis system. Pry the gel plates with the gel remover
and rinse the gel with ddH2O. Transfer the gel carefully to a
container and stain with Coomassie Brilliant Blue R-250 stain-
ing solution for 15 min. Destain the gel with destaining solu-
tion until the gel background is clear and scan the gel with a gel
scanner.
*A schematic drawing of the SDS-PAGE of the protein fractions
was shown in Fig. 1.

3.3 Protein 1. Venom fractions from RP-HPLC (Subheading 3.1.1, step 4)

Identification are reconstituted in ddH2O. Aliquot approximately 5 μg
venom proteins (estimated using Nano-Drop Spectrophotom-
3.3.1 Protein Digestion
eter) in 10 μL from each reconstituted fraction, and add into a
(In-Solution Tryptic
1.5 mL centrifuge tube, respectively.
Digestion) and Peptides
Extraction 2. Mix 15 μL of digestion buffer and 1.5 μL of reducing buffer to
the centrifuge tube and heat the mixture at 95  C for 5 min.
Cool the sample to ambient temperature.
3. Add 3 μL of alkylation buffer into the heated venom samples
and incubate in dark at ambient temperature for 20 min.
4. Following the incubation, add 1 μL of trypsin stock (0.1 μg/μL)
into the tube and incubate at 37  C for 3 h.
5. Add another 1 μL of trypsin stock (0.1 μg/μL) into the tube
and incubate overnight at 30  C for complete digestion.
6. Extract and desalt the digested peptides using Millipore Zip-
Tip® C18 Pipette Tips (Merck, USA). Aspirate 10 μL wetting
solution using ZipTip for three times, followed by aspirating
the equilibrium solution for three times. Next, aspirate and
dispense the digested samples using equilibrated ZipTip for
ten times to allow binding of peptides onto the C18 resins of
ZipTip. Wash the peptide-bound ZipTip with washing solution
(aspirating and dispensing for three times) to remove salt
content.
7. Lastly, elute the peptides from the C18 resins of ZipTip by
aspirating and dispensing for ten times in a new centrifuge
tube containing 10 μL of elution solution. Lyophilize the
extracted and desalted peptides and store at
20  C. These
tryptic peptides will be subjected to mass spectrometry analysis.
*The workflows are illustrated in Fig. 2.
 A Protein Decomplexation Strategy in Snake Venom Proteomics 89

Fig. 2 Step-by-step workflows for protein digestion (upper panel) and peptides extraction and desalting (lower
panel) protocol

3.3.2 Nano-Electrospray 1. Perform the detection analysis using 1260 Infinity Nanoflow
Ionization-Liquid LC system (Agilent, Santa Clara, CA, USA) that is connected
Chromatography Tandem to Accurate-Mass Q-TOF 6550 series with a nano-electrospray
Mass Spectrometry ionization source.
(ESI-LC-MS/MS) and Data 2. Reconstitute the lyophilized peptide analytes in 7 μL of 0.1%
Mining formic acid in ddH2O. Subject the peptide analytes to HPLC
Large-Capacity Chip Column Zorbax 300-SB-C18 (160 nL
enrichment column, a 75 μm  150 mm analytical column with
5 μm particles) (Agilent, Santa Clara, CA, USA).
3. Adjust the injection volume to 1 μL per sample, using a flow
rate of 0.4 μL/min, with a linear gradient of 5–70% of solvent B
(0.1% formic acid in 100% acetonitrile).
4. Flow the drying gas at a rate of 11 L/min at temperature
290  C. Set the fragmentor voltage at 175 V and the capillary
voltage at 1800 V. Acquire the mass spectra using Mass Hunter
acquisition software (Agilent, Santa Clara, CA, USA) in a
MS/MS mode with an MS scan range of 200–3000 m/z and
MS/MS scan range of 50–3200 m/z.
5. Extract the data with MH+ mass range between 50 and
3200 Da and process with Agilent Spectrum Mill MS Proteo-
mics Workbench software packages version B.04.00 against
merged database incorporating both non-redundant NCBI
90 Choo Hock Tan et al.

database of Serpentes (taxid: 8570) and in-house transcript

database (see Note 5).
6. Specify the fixed modification to carbamidomethylation and
oxidized methionine as a variable modification.
7. Validate the identified proteins or peptides with the following
filters: protein score > 20, peptide score > 10, and scored peak
intensity (SPI) > 70%.
8. Filter the identified proteins to achieve false discovery rate
(FDR) < 1% for the peptide-spectrum matches.
9. Consider all results showing 2 or greater than 2 “distinct
peptide” for identification.

3.4 Protein 1. Estimate the relative abundance of protein in venom fractions

Quantitation by the peak area measurement using Shimadzu LCsolution
Software Version 1.23 (Shimadzu, Japan).
3.4.1 Relative
Abundance and Venom 2. Implement the relative abundance (%) obtained from peak area
Protein Quantitation measurement (area under curve) to all collected fractions that
show protein band(s) in SDS-PAGE.
3. Estimate the relative abundance (%) of each individual venom
protein in a fraction based on its mean spectral intensity (MSI)
relative to the total MSI of all proteins in the fraction identified
through ESI-LC-MS/MS (Subheading 3.3.2).

Relative abundance of a protein in an HPLC fraction ð%Þ

Mean spectral intensity of protein in a fraction
¼  100%
Total spectral intensity of a fraction
4. Estimate the relative abundance of individual protein in a
venom by multiplying the % area under curve with the relative
abundance obtained based on mean spectral intensity (step 3).

Relative abundance of a protein in a venom ð%Þ

¼ % AUC of a fraction  Relative abundance of a
protein in a fraction ð%Þ
5. Accumulate the relative abundance of protein (%) according to
the protein identity and family for the characterization of
venom proteome.

4 Notes

1. SDS tends to precipitate in cold (below 15  C).

SDS-containing buffers may need to be warmed prior to use.
2. The linear gradient stated in above Subheading 3.1 is an opti-
mized protocol to fractionate venoms of elapid snakes such as
 A Protein Decomplexation Strategy in Snake Venom Proteomics 91

cobras (Naja sp.). Readers are advised to adjust and optimize

the elution protocol for venom samples from snakes other than
cobra (Naja sp.).
3. The percentage of polyacrylamide gel prepared depends on the
target protein(s) to be visualized. High percentage gel
(15–18%) is suitable for separating the low molecular weight
proteins (< 20 kDa), whereas low percentage gel provides a
better separation for the higher molecular weight proteins. In
general, the proteins for most snake venoms can be separated
and visualized on a 15% gel.
4. 4.5 mL of resolving solution mixture is required in preparation
of a resolving gel. We found that isopropanol works better than
water in layering the resolving gel. Tilt the gel casting holder
slightly (20 up and down) before leaving the gel to solidify for
an even distribution of gel level.
5. The in-house transcript database is created using data obtained
from venom-gland transcriptomic study. This use of a tran-
script database is optional in venomic studies, but can be
incorporated with up-to-date non-redundant NCBI dataset
of Serpentes (taxid: 8570) to provide a more complete database
for mass spectra matching in protein identification.

References
1. Lomonte B, Fernández J, Sanz L, Angulo Y, kaouthia (monocled cobra) venoms from three
Sasa M, Gutiérrez JM, Calvete JJ (2014) Ven- different geographical regions of Southeast
omous snakes of Costa Rica: biological and Asia. J Proteome 120:105–125. https://doi.
medical implications of their venom proteomic org/10.1016/j.jprot.2015.02.012
profiles analyzed through the strategy of snake 6. Gutiérrez JM, Lomonte B, León G, Alape-
venomics. J Proteome 105(Supplement Girón A, Flores-Dı́az M, Sanz L, Angulo Y,
C):323–339. https://doi.org/10.1016/j. Calvete JJ (2009) Snake venomics and antive-
jprot.2014.02.020 nomics: proteomic tools in the design and con-
2. Calvete JJ, Sanz L, Angulo Y, Lomonte B, trol of antivenoms for the treatment of
Gutiérrez JM (2009) Venoms, venomics, anti- snakebite envenoming. J Proteome 72
venomics. FEBS Lett 583(11):1736–1743. (2):165–182. https://doi.org/10.1016/j.
https://doi.org/10.1016/j.febslet.2009.03. jprot.2009.01.008
029 7. Vetter I, Davis JL, Rash LD, Anangi R,
3. Tan CH, Tan KY, Lim SE, Tan NH (2015) Mobli M, Alewood PF, Lewis RJ, King GF
Venomics of the beaked sea snake, Hydrophis (2011) Venomics: a new paradigm for natural
schistosus: a minimalist toxin arsenal and its products-based drug discovery. Amino Acids
cross-neutralization by heterologous antive- 40(1):15–28. https://doi.org/10.1007/
noms. J Proteome 126:121–130. https://doi. s00726-010-0516-4
org/10.1016/j.jprot.2015.05.035 8. Tan KY, Tan CH, Chanhome L, Tan NH
4. Tan KY, Tan NH, Tan CH (2018) Venom (2017) Comparative venom gland transcrip-
proteomics and antivenom neutralization for tomics of Naja kaouthia (monocled cobra)
the Chinese eastern Russell’s viper, Daboia sia- from Malaysia and Thailand: elucidating geo-
mensis from Guangxi and Taiwan. Sci Rep graphical venom variation and insights into
8(1):8545. https://doi.org/10.1038/ sequence novelty. PeerJ 5:e3142. https://doi.
s41598-018-25955-y org/10.7717/peerj.3142
5. Tan KY, Tan CH, Fung SY, Tan NH (2015) 9. Augusto-de-Oliveira C, Stuginski DR, Kitano
Venomics, lethality and neutralization of Naja ES, Andrade-Silva D, Liberato T, Fukushima I,
92 Choo Hock Tan et al.

Serrano SM, Zelanis A (2016) Dynamic rear- 15. Petras D, Sanz L, Segura A, Herrera M,
rangement in snake venom gland proteome: Villalta M, Solano D, Vargas M, Leon G, War-
insights into Bothrops jararaca intraspecific rell DA, Theakston RD, Harrison RA,
venom variation. J Proteome Res 15 Durfa N, Nasidi A, Gutierrez JM, Calvete JJ
(10):3752–3762. https://doi.org/10.1021/ (2011) Snake venomics of African spitting
acs.jproteome.6b00561 cobras: toxin composition and assessment of
10. Calvete JJ (2014) Next-generation snake congeneric cross-reactivity of the pan-African
venomics: protein-locus resolution through EchiTAb-Plus-ICP antivenom by antivenomics
venom proteome decomplexation. Expert Rev and neutralization approaches. J Proteome Res
Proteomics 11(3):315–329. https://doi.org/ 10(3):1266–1280. https://doi.org/10.1021/
10.1586/14789450.2014.900447 pr101040f
11. Tan CH, Wong KY, Tan KY, Tan NH (2017) 16. Tan NH, Fung SY, Tan KY, Yap MKK, Gna-
Venom proteome of the yellow-lipped sea krait, nathasan CA, Tan CH (2015) Functional
Laticauda colubrina from Bali: insights into venomics of the Sri Lankan Russell’s viper
subvenomic diversity, venom antigenicity and (Daboia russelii) and its toxinological correla-
cross-neutralization by antivenom. J Proteome tions. J Proteome 128:403–423. https://doi.
166:48–58. https://doi.org/10.1016/j.jprot. org/10.1016/j.jprot.2015.08.017
2017.07.002 17. Tan CH, Fung SY, Yap MK, Leong PK, Liew
12. Alape-Giron A, Sanz L, Escolano J, Flores- JL, Tan NH (2016) Unveiling the elusive and
Diaz M, Madrigal M, Sasa M, Calvete JJ exotic: Venomics of the Malayan blue coral
(2008) Snake venomics of the lancehead pitvi- snake (Calliophis bivirgata flaviceps). J Prote-
per Bothrops asper: geographic, individual, and ome 132:1–12. https://doi.org/10.1016/j.
ontogenetic variations. J Proteome Res 7 jprot.2015.11.014
(8):3556–3571. https://doi.org/10.1021/ 18. Dutta S, Chanda A, Kalita B, Islam T, Patra A,
pr800332p Mukherjee AK (2017) Proteomic analysis to
13. Wong KY, Tan CH, Tan KY, Naeem QH, Tan unravel the complex venom proteome of east-
NH (2018) Elucidating the biogeographical ern India Naja naja: correlation of venom
variation of the venom of Naja naja (specta- composition with its biochemical and pharma-
cled cobra) from Pakistan through a venom- cological properties. J Proteome 156:29–39.
decomplexing proteomic study. J Proteome https://doi.org/10.1016/j.jprot.2016.12.
175:156–173. https://doi.org/10.1016/j. 018
jprot.2017.12.012 19. Tan CH, Tan KY, Yap MK, Tan NH (2017)
14. Faisal T, Tan KY, Sim SM, Quraishi N, Tan Venomics of Tropidolaemus wagleri, the sexu-
NH, Tan CH (2018) Proteomics, functional ally dimorphic temple pit viper: unveiling a
characterization and antivenom neutralization deeply conserved atypical toxin arsenal. Sci
of the venom of Pakistani Russell’s viper Rep 7:43237. https://doi.org/10.1038/
(Daboia russelii) from the wild. J Proteome srep43237
183:1–13. https://doi.org/10.1016/j.jprot.
2018.05.003
 Chapter 6

Fractionation Techniques to Increase Plant Proteome

Coverage: Combining Separation in Parallel at the Protein
and the Peptide Level
Martin Černý, Miroslav Berka, and Hana Habánová

Abstract
Peptide spectral libraries enable targeted identification and quantitation of low-abundance proteins in a
complex plant proteome. Here we describe parallel protein and peptide fractionation techniques to improve
plant proteome coverage and facilitate construction of spectral libraries.

Key words Plant proteomics, Protein fractionation, Peptide fractionation, C18, SCX, PEG

1 Introduction

Proteins may form up to 20% of total cellular weight, and rough

estimates predict that this corresponds to a range of two to four
million proteins per cubic micron [1]. However, most of these
proteins belong to only a few, highly abundant, protein families
and the difference in concentration between a low-abundance pro-
tein and a highly abundant protein within a single cell can easily be
five to six orders of magnitude [2]. The dynamic concentration
range is further expanded in multicellular organisms. For instance,
the average human body consists of ca 37 trillion cells which can be
grouped into at least 200 different cell types. Proteome complexity
is further increased by posttranslational modifications. This overall
complexity represents a significant obstacle to proteome analyses
and even the rapid development in mass spectrometry instruments
that we have seen in recent years cannot address all these issues. For
this reason, proteome fractionation is the best approach if a reason-
able level of proteome coverage is to be achieved. However, frac-
tionation requires a relatively large amount of starting material,
which is not always readily available, and the methods are time-
consuming and constitute a limitation for quantitative analyses.
This problem can be circumvented by the targeted methods

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_6, © Springer Science+Business Media, LLC, part of Springer Nature 2019

93
94 Martin Černý et al.

Fig. 1 Native extraction and PEG fractionation

selected/multiple reaction monitoring (SRM/MRM) and/or

sequential window acquisition of all theoretical spectra (SWATH).
Both these techniques improve detection limits but require the
availability of a reference peptide spectral library [3]. Here, we
present a protocol with which to obtain data for building such a
library. This protocol employs protein precipitation and parallel
fractionations at the protein (Figs. 1 and 2) and peptide levels
(Fig. 3): it comprises inexpensive nondenaturing polyethylene gly-
col (PEG) fractionation [4, 5], protein precipitation by low pH and
acetone followed by phenol re-extraction [6], mass-based separa-
tion on sodium dodecyl sulfate polyacrylamide gel electrophoresis
(SDS-PAGE) and charge-based separation by isoelectric focusing
[7], high pH C18 peptide fractionation [8], and strong cation
exchange (SCX) peptide fractionation [9, 10]. Although this work-
flow is optimized for ca 1 g of starting material, the amount can be
reduced and the protocol adapted for a smaller-scale experiment. In
addition, it can be combined with fractionations at the tissue level,
subcellular enrichment and techniques that improve the detection
of lower abundance proteins by means of immunodepletion of
abundant proteins or proteome equalization [11–13].

2 Materials

Always wear laboratory gloves for self-protection and to prevent

sample contamination. Prepare solutions using ultrapure solvents,
preferably of LC-MS grade quality.

2.1 Homogenization 1. Mixer Mill MM 400 (Retsch), stainless steel grinding jars and
milling balls (see Note 1).
 Plant Proteome Fractionation Protocol 95

Fig. 2 Denaturing protein extraction and protein digestion. (a–g) Protein extraction and purification. (h)
Determination of protein concentration, (i1–2) protein separation, and (i3, j) digestion

2. Liquid nitrogen.
3. 2.0 mL Eppendorf LoBind tubes or similarly coated low-
protein-binding microcentrifuge tubes.
96 Martin Černý et al.

Fig. 3 Peptide desalting and fractionation.

2.2 Native Extraction 1. Eppendorf Thermomixer R (see Note 2).

and PEG Fractionation 2. Ultrasonic bath.
3. Extraction buffer: 20 mM MgCl2, 1% (v/v)
β-mercaptoethanol, 1 mM EDTA, 2% (v/v) IGEPAL, 0.5 M
Tris–HCl, pH 7.8. Prepare 50 mL and store at 4  C (see Note
3). Aliquot into 1 mL portions and supplement each with
50 μL of Protease Inhibitor Cocktail (Merck) prior to use.
4. Polyethylene glycol 4000 (PEG).
5. 2.0 mL and 5.0 mL Eppendorf LoBind tubes or similarly
coated low-protein-binding microcentrifuge tubes.

2.3 Denaturing 1. 10% (w/v) Trichloroacetic acid (TCA) in acetone. Prepare 1 L

Acetone/TCA/Phenol and store at 20  C (see Note 4).
Extraction 2. 80% (v/v) acetone in water. Prepare 250 mL and store at
20  C.
3. SDS buffer: 2% (w/v) SDS, 30% (w/v) sucrose, 5% (v/v)
β-mercaptoethanol, 5 mM ethylenediaminetetraacetic acid
(EDTA), 100 mM Tris–HCl, pH 8.0. Prepare 50 mL and
keep at 4  C; shelf life is more than a month.
4. TE-saturated phenol—phenol saturated with 10 mM Tris–HCl
buffer pH 8.0 and 1 mM EDTA.
5. Thermomixer.
 Plant Proteome Fractionation Protocol 97

6. Retsch mill with an adapter for 2.0 mL tubes.

7. 100 mM ammonium acetate in methanol; prepare 500 mL and
store at 20  C.
8. 100 mM ammonium bicarbonate, 8 M urea in water; prepare
100 mL, keep at 4  C (see Note 5).
9. IEF solubilization solvent: 7 M urea, 2 M thiourea in water, 2%
(w/v) CHAPS, 90 mM dithiothreitol. Prepare 50 mL, aliquot
into 5 mL tubes, and store at 20  C (see Notes 3 and 5).
10. Bradford Reagent (Merck), bovine serum albumin standard, a
96-well microplate, and a microplate reader (see Note 6).

2.4 In-Solution 1. Ammonium bicarbonate buffer: 50 mM NH4HCO3, 2 mM

Digestion CaCl2, 8% (v/v) acetonitrile. Prepare 100 mL, keep at 4  C.
2. Vertical rotator, incubator.
3. Immobilized Trypsin (Promega) (see Notes 7 and 8).

2.5 Protein 1. IPG strips: 7 cm ReadyStrips with immobilized nonlinear pH

Separation and In-Gel gradient 3-10 (Bio-Rad).
Digestion 2. Ampholytes pH 3-10 (Bio-Rad).
2.5.1 Isoelectric 3. PROTEAN IEF Cell unit and focusing tray (Bio-Rad).
Focusing 4. Paper wicks: electrode wicks suitable for isoelectric focusing.
5. Mineral oil.
6. Scalpel.
7. 1.5 mL Eppendorf LoBind tubes or similarly coated low-
protein-binding microcentrifuge tubes.

2.5.2 SDS-PAGE 1. Mini-PROTEAN cell and power supply (Bio-Rad).

2. Precast Mini-PROTEAN TGX gel, 4–20%, 10 wells, 50 μL (see
Note 9).
3. Running buffer: 25 mM Tris–HCl, 192 mM glycine, 0.1%
SDS, pH 8.3. Prepare 1 L using 3 g of Tris (base), 14.4 g of
glycine, and 1 g of SDS, do not adjust the pH; store at 4  C.
4. 4  Loading buffer: 10% (w/v) SDS, 20% glycerol, 10 mM
dithiothreitol, 0.05% (w/v) bromophenol blue, 200 mM
Tris–HCl, pH 6.8. Prepare 20 mL, aliquot into 1.5 mL
tubes, and store at 20  C.
5. Thermomixer.
6. Scalpel.
7. 1.5 mL Eppendorf LoBind tubes or similarly coated low-
protein-binding microcentrifuge tubes.
98 Martin Černý et al.

2.5.3 In-Gel Digestion 1. Acetonitrile.

2. SpeedVac Evaporator (Thermo Scientific).
3. Retsch mill with an adapter for 2.0 mL tubes and milling balls.
4. Digestion buffer: Dissolve 20 μg of a sequencing-grade trypsin
(e.g., Promega) in 3.0 mL of ammonium bicarbonate buffer
(Subheading 2.4). Prepare on ice and use immediately for
protein digestion. This amount is sufficient for 20 samples
obtained from isoelectric focusing and SDS-PAGE separations.
5. Thin-walled 0.5 mL PCR tubes.

2.6 Peptide 1. 1% TFA: 1% (v/v) trifluoroacetic acid (TFA) in water. Prepare

Desalting 200 mL, keep at room temperature (see Note 4).
2. 50% (v/v) acetonitrile in water. Prepare 15 mL, cover with
aluminum foil or store in the dark at room temperature.
3. VersaPlate, collection plate and C18 (25 mg) tubes (Agilent),
vacuum pump (see Note 10).
4. Thin-walled 0.5 mL PCR tubes.
5. SpeedVac Evaporator (Thermo Scientific).
6. Quantitative Colorimetric Peptide Assay kit (Thermo Scien-
tific), a 96-well microplate and a microplate reader.

2.7 Peptide 1. VersaPlate, collection plate, C18 (25 mg) and SCX (50 mg)
Fractionation tubes (Agilent), vacuum pump (see Note 10).
2. 1% TFA (prepared in Subheading 2.6).
3. Concentration series of acetonitrile in 0.1% (v/v) triethyla-
mine: Prepare 2 mL of 0.1% (v/v) triethylamine in water and
2 mL of 0.1% (v/v) triethylamine in acetonitrile. Mix 50, 75,
100, 125, 150, 175, 200, and 250 μL of triethylamine in
acetonitrile with the appropriate volume of triethylamine in
water to obtain 1 mL of each stock solution (5–50%).
4. 0.5% formic acid: Dilute formic acid (FA) with water to pro-
duce 15 mL of 0.5% (v/v) FA, store at room temperature.
5. Concentration series of ammonium acetate in 0.5% FA: Prepare
ca 1.5 mL of 500 mM ammonium acetate in 0.5% FA (dissolve
38.5 mg of ammonium acetate per 1 mL of 0.5% FA), and then
dilute this stock by mixing 50, 100, 250, 300, and 400 μL with
the necessary amounts of 0.5% FA to obtain 0.5 mL volumes of
50–400 mM stock solutions.

3 Methods

3.1 Homogenization 1. Homogenize ca 1 g of plant tissue using liquid nitrogen and an

MM400 Retsch mill with a prechilled stainless steel grinding jar
and milling balls (see Note 1).
 Plant Proteome Fractionation Protocol 99

2. Mill at 30 Hz for 60 s or until a fine powder is produced; keep

the jars cold to prevent samples from melting. Wear protective
gloves and use appropriate equipment for handling liquid
nitrogen.
3. Aliquot ca 250 mg samples into 2.0 mL Eppendorf LoBind
tubes and store the aliquots at 80  C.

3.2 Native Extraction 1. Take a homogenized aliquot, place it on ice and add 1.0 mL of
and PEG Fractionation extraction buffer.
2. Sonicate at 4  C for 5 min and then incubate in the Thermo-
mixer at 4  C, 800 rpm for 10 min.
3. Centrifuge for 10 min (10,000  g, 4  C), transfer the super-
natant to a new 2.0 mL tube, and keep it on ice. Mix the pellet
with 10% (w/v) TCA in acetone and then follow steps 2–9 of
the denaturing extraction protocol (Subheading 3.3.).
4. Add 90 μL of 50% (w/v) PEG solution to give a final concen-
tration of ca 4% (w/v), incubate in the Thermomixer (4  C,
800 rpm, 20 min), then centrifuge for 10 min (10,000  g,
4  C), collect the supernatant, and process the pellet as in the
previous step.
5. Repeat the supernatant precipitation in a stepwise PEG con-
centration gradient, adding 100 and 125 μL of 50% (w/v) PEG
to give, respectively, a 8 and a 12% (w/v) mixture. Collect
pellets, mix with 10% (w/v) TCA in acetone, and transfer the
last supernatant into a 5.0 mL tube and precipitate it with
4.0 mL of 10% (w/v) TCA in acetone, then proceed to the
second step of the denaturing extraction (Subheading 3.3).

3.3 Denaturing 1. Resuspend an aliquot of frozen homogenized tissue with 10%

Acetone/TCA/Phenol (w/v) TCA in acetone (fill to 2.0 mL), adding a stainless steel
Extraction milling ball to facilitate sample solubilization.
2. Precipitate the total protein overnight at 20  C (see Note 11).
3. Centrifuge the sample for 10 min (10,000  g, 4  C) to bring
down the precipitate.
4. Wash the pellet with 80% (v/v) acetone, centrifuge again at
10,000  g for 10 min, and then resuspend in 0.8 mL of SDS
buffer. Incubate in the Thermomixer at 800 rpm and room
temperature for 10 min.
5. Remove the milling ball, add 400 μL of TE-saturated phenol,
and shake in a Retsch mill for 20 s at 30 rpm (see Note 1).
6. Centrifuge the mixture for 10 min (10,000  g, 20  C) and
aliquot the top (phenolic) layer into three 2.0 mL LoBind
tubes.
7. Precipitate overnight in ice-cold 100 mM ammonium acetate
in methanol ( 20  C).
100 Martin Černý et al.

8. Centrifuge samples for 10 min to collect protein pellets

(10,000  g, 4  C), wash pellets with 1.0 mL 80% (v/v)
acetone in water, remove all solvent, and dry on air for 5 min.
9. Dissolve protein pellets in the Thermomixer with 300 μL
(i) 100 mM ammonium bicarbonate, 8 M urea for in-solution
digestion, or (ii, iii) IEF solubilization solvent for isoelectric
focusing and SDS-PAGE. Incubate at 30  C, 800 rpm for
30 min (see Note 5) and estimate protein concentration by
means of a Sigma-Aldrich Bradford assay in microplate format.
The isolation method should yield at least 500 μg of protein per
aliquot (see Notes 12 and 13).

3.4 In-Solution 1. Dilute 300 μg of protein with an equal volume of water and
Digestion two volumes of ammonium bicarbonate buffer.
2. Add 50 μL of immobilized trypsin beads (Promega, see Note 7)
and incubate overnight on a rotator at 30 rpm in an incubator
at 30  C (see Note 8).

3.5 Protein 1. Dilute 300 μg of protein to a final volume of 260 μL with IEF
Separation and In-Gel solubilization solvent (if needed), add 1.3 μL of ampholytes
Digestion (pH 3–10), and load onto two 7 cm 3–10 NL IPG strips
(Bio-Rad) in a rehydration tray.
3.5.1 Isoelectric
Focusing
2. Overlay with mineral oil and rehydrate overnight at room
temperature.
3. Wet four paper wicks in water, transfer IPG strips into the
focusing tray, and put wet paper wicks between gel and elec-
trode to prevent their making direct contact. Overlay again
with mineral oil.
4. Isoelectrically focus proteins at 20  C in a PROTEAN IEF Cell
unit (Bio-Rad) in six steps: 150 V (20 min), 300 V (20 min),
600 V (20 min), 1500 V (20 min), 3000 V (20 min), and
4000 V up to 12,000 Vh.
5. Put the IPG strips onto a clean filter paper with the gel facing
up to dry off the mineral oil.
6. Align one IPG strip above the other, cut gels vertically into ten
equal fractions, and collect them in 1.5 mL LoBind tubes.

3.5.2 SDS-PAGE 1. Assemble the Mini-PROTEAN cell employing a Mini-

PROTEAN TGX precast gel (4–20%, 10 wells, 50 μL) and
add 700 mL of running buffer (see Note 9).
2. Mix 300 μg of protein with 4 loading buffer (3:1), incubate at
95  C for 10 min (Thermomixer), spin down (1000  g), and
load into wells (30 μg per well).
3. Connect the Mini-PROTEAN to its power supply and separate
proteins with the following settings: 100 V (10 min) followed
 Plant Proteome Fractionation Protocol 101

by 150 V (30 min). The bromophenol blue line should be ca

1 cm above the end of the gel.
4. Disconnect the electrophoresis apparatus and carefully open
the gel cassette. Remove empty parts of the gel with a clean
scalpel blade, cut gel horizontally into ten equal pieces. Cut
each fraction into smaller pieces and collect it into a 1.5
LoBind tube.

3.5.3 In-Gel Digestion 1. Wash gel pieces twice with 1.0 mL acetonitrile and dry samples
down in a SpeedVac Evaporator.
2. Use a Retsch mill and stainless steel milling balls to obtain a fine
powder (see Note 14). Place samples on ice, add 150 μL diges-
tion buffer, incubate for 15 min, then transfer tubes to 37  C
and incubate overnight.
3. Extract peptides twice with 150 μL acetonitrile, collect extracts
into 0.5 mL thin-walled PCR tubes, and dry down to ca 100 μL
in a SpeedVac Evaporator (see Note 15).

3.6 Peptide 1. Mix a peptide sample 1:1 with 1% TFA, shake, and clarify the
Desalting sample by centrifugation (10,000  g, 5 min).
2. Wash a C18 SPE column with 0.5 mL 100% acetonitrile,
2  1 mL of water, 1  1 mL of 1% TFA.
3. Load the acidified sample solution onto the SPE column (see
Note 16).
4. Wash the column twice with 0.5 mL 1% TFA.
5. Elute peptides in two steps: 100 μL 50% (v/v) acetonitrile in
water, followed by 300 μL of acetonitrile; collect the flow-
through into 0.5 mL thin-walled PCR tubes, dry samples
down in a SpeedVac Evaporator to ca 40 μL (see Note 15),
and estimate peptide concentration by Quantitative Colorimet-
ric Peptide Assay (Thermo Scientific).

3.7 Peptide 1. Mix the peptide sample from in-solution digestion 1:1 with 1%
Fractionation TFA, shake, and clarify the sample by centrifugation
(10,000  g, 5 min).
2. Wash two C18 SPE columns with 0.5 mL of 100% acetonitrile,
2  1.0 mL of water, 1 with 1.0 mL of 1% TFA.
3. Divide the acidified sample equally between two SPE columns,
wash once with 1% TFA and then proceed to high pH or SCX
fractionation.

3.7.1 High pH C18 1. Wash bound peptides with 1.0 mL of water (see Note 16).
Fractionation 2. Elute peptides with a stepwise gradient of acetonitrile in 0.1%
triethylamine. Load, successively, 200 μL each of 5%, 7.5%,
102 Martin Černý et al.

10%, 12.5%, 15%, 17.5%, 20%, 25%, and 50% acetonitrile, and
collect fractions into 0.5 mL thin-walled PCR tubes.
3. Dry down to ca 20–30 μL in a SpeedVac Evaporator (see Note
15) and estimate peptide concentration by Quantitative Col-
orimetric Peptide Assay (Thermo Scientific).

3.7.2 Peptide SCX 1. Elute bound peptides with 100 μL of 50% (v/v) acetonitrile in
Fractionation water followed by 300 μL of acetonitrile, collect fractions into
0.5 mL thin-walled PCR tubes.
2. Dry down to ca 100 μL in a SpeedVac Evaporator (see Note 15)
and dilute in 1.0 mL of 50 mM ammonium acetate in 0.5% FA.
3. Wash an SCX SPE column with, successively, 0.5 mL of
500 mM ammonium acetate in 0.5% FA and 2  1.0 mL of
50 mM ammonium acetate, then load the peptide sample and
collect the flow-through into a 0.5 mL thin-walled PCR tube
(the first fraction).
4. Elute peptides with a stepwise gradient of ammonium acetate
in 0.5% FA. Load, successively, 200 μL each of 100, 250,
300, 400, and 500 mM ammonium acetate, and collect frac-
tions into 1.5 mL tubes (see Note 17).
5. Dilute the collected eluates with 800 μL of 0.1% (v/v) FA in
water; desalt peptides on a C18 SPE column as described in
Subheading 3.6.

4 Notes

1. The Retsch mill employed in our protocol for homogenization

and phenol extraction can be substituted with a standard mor-
tar and pestle, and a vortex mixer, respectively.
2. We recommend working in a cold room or placing the Ther-
momixer in a refrigerator.
3. You may wish to consider replacing the foul-smelling
β-mercaptoethanol with 20 mM dithiothreitol (DTT), but
this solution must be prepared fresh or stored in aliquots at
20  C (the half-life of DTT at pH 8.0 and room temperature
is only hours [14]).
4. TCA and TFA are strong acids; wear appropriate protection.
5. The dissolution of urea is endothermic, so prepare urea directly
in a closed 100 mL flask and use a magnetic stirrer at room
temperature. Avoid heating; any temperature above 30  C
increases urea decomposition, produces cyanate, and results
in protein carbamylation. Prepare this solution every 4 weeks
or store aliquots at 20  C.
 Plant Proteome Fractionation Protocol 103

6. Pipette 2 μL of each of the protein samples and corresponding

blanks (ammonium bicarbonate and IEF buffers) and then
rapidly overlay with 200 μL of the Bradford Reagent. This
will provide sufficient mixing and the assay can be carried out
immediately with reasonable precision and reproducibility.
7. The immobilized trypsin can be substituted with a standard
sequencing-grade trypsin, but the immobilized form is less
prone to self-cleavage than its counterpart and does not require
the user to work on ice. However, take care to mix the slurry of
beads properly in order to obtain a homogeneous mixture for
pipetting.
8. Some samples (e.g., seed storage tissues) contain protein inhi-
bitors of proteases that may interfere with in-solution diges-
tion. If the peptide yield is lower than expected, increase the
trypsin:protein ratio and consider predigestion with Lys-C or a
modification of the digestion buffer (e.g., an increase in the
acetonitrile concentration). The digestion step can be speeded
up by ultrasound, microwave, and heating treatment; however,
the buffer contains urea and the sample is likely to suffer from
nonenzymatic modification by carbamylation.
9. Precast gels are convenient but the available gradient range may
not be sufficient for your protein extract. You may consider
using a prestained protein ladder and running a test PAGE in
order to determine the optimal gel gradient, and/or readjust-
ing the positions of the lines when slicing the gel. We do not
recommend staining of the material for digestion, as the stain
would then need to be removed and washing the gel could
result in depletion of smaller proteins.
10. The vacuum manifold that we use for the VersaPlate solid phase
extraction can be substituted with a pipette, e.g., an Eppendorf
1 mL pipette fits well on the SPE tubes.
11. Precipitation overnight is not mandatory but a shorter time
will have a negative effect on the protein yield. However, a
longer storage time will not decrease the yield, and we have
seen excellent results with samples stored for more than
6 months at 20  C. Note that some nonenzymatic posttrans-
lational modifications may still occur and affect the quality of
the protein sample.
12. Based on our experiments, the expected protein yield from
250 mg fresh weight is >1500 μg for plantlet, seedling and
leaf tissue, at least 1000 μg for root tissue and >2500 for seed
extracts (species such as Arabidopsis thaliana, Solanum lycoper-
sicum, Nicotiana tabacum, Hordeum vulgare, Pisum sativum,
and Quercus robur [6, 7, 12]).
13. Note that our routine protocol does not include cysteine alkyl-
ation and the resulting proteome library is thus missing most
104 Martin Černý et al.

cysteine-containing tryptic peptides. The thiol side chains of

cysteine residues are highly susceptible to posttranslational
modifications and we prefer to avoid these in a quantitative
analysis. Cysteine alkylation can be included prior to the diges-
tion step. For in-gel digestion, incubate homogenized gels in
0.5 mL of 100 mM DTT in IEF buffer for 30 min, 800 rpm,
room temperature, then centrifuge, wash pellets with 0.5 mL
acetonitrile, and resuspend the pellets in 0.5 mL of 100 mM
iodoacetamide in IEF buffer. Incubate at 800 rpm for 30 min
in the dark (iodoacetamide is light-sensitive). Centrifuge, wash
with acetonitrile, and dry on the SpeedVac evaporator. Alkyl-
ation for in-solution digestion: Add DTT from a 250 mM
aqueous stock solution to the protein dissolved in urea/ammo-
nium bicarbonate to a final concentration of 10 mM DTT and
incubate for 30 min at room temperature, 800 rpm. Add
iodoacetamide (250 mM in water) to a final concentration of
30 mM; incubate in the dark for 30 min at 800 rpm. Quench
iodoacetamide with a further addition of DTT, using the same
volume as for iodoacetamide.
14. Do not add milling balls to only partially dried gel pieces, as
they will stick and the milling will be ineffective. Milling of fully
dried gel pieces is rapid and does not lead to excessive heat
production. In our experience, it significantly improves peptide
recovery rates and facilitates uniform distribution of trypsin in
the digest.
15. Volatile acetonitrile will evaporate and its concentration will
not interfere with C18 binding. Monitor the evaporation and
once the liquid volume reaches ca 100 μL (in-gel digests) or ca
40 μL (desalting), remove the sample from the SpeedVac
Evaporator. Try to avoid drying of the samples, as this limits
peptide recovery due to peptide aggregation and peptide-
surface interaction [15]. If you are unsuccessful and the sam-
ples are fully dried, reconstitute in 4% (v/v) acetonitrile in
water by sonication and carefully wash the surface of the
tube. Thin-walled PCR tubes are more suitable as these will
improve sonication efficiency.
16. Smaller peptides and hydrophilic peptides (e.g., some phos-
phopeptides and glycopeptides) will be lost in this step. To
improve sample recovery, you may employ, e.g., graphite col-
umns [16]. However, these peptides will not be retained by
C18 during the LC-MS step, as they will all elute in the first
minutes and are not usually suitable for quantitation due to the
ion suppression effect.
17. The desalting step can be replaced by evaporation, which will
remove ammonium acetate. However, in our experience the
C18 method is faster and more reliable.
 Plant Proteome Fractionation Protocol 105

Acknowledgments

This work was supported by the Ministry of Education, Youth and

Sports of the Czech Republic under the project CEITEC 2020
(LQ1601) and TE02000177 (TACR), and by Brno PhD Talent
2017 (funded by Brno City Municipality) and IGA grant no. IP
15/2017 to H.H.

References

1. Milo R (2013) What is the total number of 9. Rappsilber J, Mann M, Ishihama Y (2007) Pro-
protein molecules per cell volume? A call to tocol for micro-purification, enrichment,
rethink some published values. BioEssays pre-fractionation and storage of peptides for
35:1050–1055 proteomics using StageTips. Nat Protoc
2. Picotti P, Bodenmiller B, Mueller LN, 2:1896–1906
Domon B, Aebersold R (2009) Full dynamic 10. Mostovenko E, Hassan C, Rattke J, Deelder
range proteome analysis of S. cerevisiae by tar- AM, van Veelen PA, Palmblad M (2013) Com-
geted proteomics. Cell 138:795–806 parison of peptide and protein fractionation
3. Schubert OT, Gillet LC, Collins BC, methods in proteomics. EuPA Open Proteom
Navarro P, Rosenberger G, Wolski WE et al 1:30–37
(2015) Building high-quality assay libraries 11. Černý M, Skalák J, Kurková B, Babuliaková E,
for targeted analysis of SWATH MS data. Nat Brzobohatý BB (2011) Using a commercial
Protoc 10(3):426–441 method for rubisco immunodepletion in anal-
4. Acquadro A, Flavo S, Mila S, Albo AG, ysis of plant proteome. Chemické listy
Comino C, Moglia A, Lanteri S (2009) Prote- 105:640–642
omics in globe artichoke: protein extraction 12. Černý M, Jedelský PL, Novák J, Schlosser A,
and sample complexity reduction by PEG frac- Brzobohatý B (2014) Cytokinin modulates
tionation. Electrophoresis 30(9):1594–1602 proteomic, transcriptomic and growth
5. Wang W-Q, Song B-Y, Deng Z-J, Wang Y, Liu responses to temperature shocks in Arabidop-
S-J, Møller IM, Song S-Q (2015) Proteomic sis. Plant Cell Environ 37:1641–1655
analysis of lettuce seed germination and ther- 13. Righetti PG, Boschetti E (2016) Global prote-
moinhibition by sampling of individual seeds at ome analysis in plants by means of peptide
germination and removal of storage proteins by libraries and applications. J Proteomics
polyethylene glycol fractionation. Plant Physiol 143:3–14
167(4):1332–1350 14. Stevens R, Stevens L, Price N (1983) The sta-
6. Cerna H, Černý M, Habánová H, Šafářová D, bilities of various thiol compounds used in pro-
Abushamsiya K, Navrátil M, Brzobohatý B tein purifications. Biochem Educ 11:70
(2017) Proteomics offers insight to the mech- 15. Berka M, Luklová M (2017) Limited drying
anism behind Pisum sativum L. response to Pea and its effect on peptide recovery rates. In:
seed-borne mosaic virus (PSbMV). J Proteo- Polak O et al (eds) MendelNet 2017 Proceed-
mics 153:78–88 ings of 24th International PhD Students Con-
7. Baldrianová J, Černý M, Novák J, Jedelský PL, ference. 24th International PhD Students
Divı́šková E, Brzobohatý B (2015) Arabidopsis Conference, Brno, November 2017. p 91
proteome responses to the smoke-derived 16. Nukarinen E, Tomanov K, Ziba I,
growth regulator karrikin. J Proteomics Weckwerth W, Bachmair A (2017) Protein
120:7–20 sumoylation and phosphorylation intersect in
8. Batth TS, Francavilla C, Olsen JV (2014) Arabidopsis signaling. Plant J 91:505–517
Off-line high-pH reversed-phase fractionation
for in-depth phosphoproteomics. J Proteome
Res 13:6176–6186
 Chapter 7

A Systematic Analysis Workflow for High-Density

Customized Protein Microarrays in Biomarker Screening
Rodrigo Garcı́a-Valiente, Jonatan Fernández-Garcı́a,
Javier Carabias-Sánchez, Alicia Landeira-Viñuela, Rafael Góngora,
Marı́a Gonzalez-Gonzalez, and Manuel Fuentes

Abstract
High-density protein microarrays constitute a promising high-throughput platform for the characterization
of protein expression patterns, biomarker discovery, and validation. Different types of protein microarrays
have been described according to several features (such as content, format, and detection system) present-
ing advantages and disadvantages which are relevant for the specific application and purposes. Therefore, an
experimental design is key for any screening based on protein microarrays assays; in fact, the data analysis
strategy is directly related to the experimental design, type of protein microarray and consequently the final
outcome, the data and results interpretation, is also directly linked. Here, it is proposed a systematic
workflow for biomarker discovery based on tailor-made protein microarrays platforms which obtain
comprehensively info for the functional protein characterization in high-throughput format.

Key words Protein microarray, Analysis, Proteome, Antibodies, Fluorescence, Proteomics, Normali-
zation, Biomarker, Protein microarray

1 Introduction

Despite advances in proteomics, deciphering the proteome in one

single assay remains a challenge mainly because of the complexity,
variety, and dynamics of the proteomes. Among others, the size of
the proteome is high, as an example, the human transcriptome
comprises more than 23,000 protein-coding genes, that generate
more than 100,000 protein species, mainly derived after alternative
splicing and posttranslational modifications (PTMs) . In addition,
the wide dynamic range of the proteome is caused by huge quanti-
tative variations at the protein levels; then, for systematic analysis

Rodrigo Garcı́a-Valiente and Jonatan Fernández-Garcı́a contributed equally to this work.

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_7, © Springer Science+Business Media, LLC, part of Springer Nature 2019

107
108 Rodrigo Garcı́a-Valiente et al.

extraction and enrichment methods (which are not always efficient)

are required. Moreover, mass spectrometry only captures a glimpse
of this complexity, and many biological and technical replicates are
usually required. Overall, the comprehensive and exhaustive prote-
ome characterization is still an amazing challenge. In order to avoid
these challenges, protein microarrays have become one of the
promising strategies for biomarker and drug discovery
[1–3]. High-density protein microarrays allow to analyze hundreds
to thousands of known proteins in a single experiment and in a
high-throughput format [4].
Recently, thanks to the capacity of massive comprehensive and
systematic analysis, proteomics have been considered an useful
approach to look deeper into personalized medicine and biomarker
discovery, via the five R criteria: right patient/target, right diagno-
sis, right treatment, right drug/target, and right dose/time [5]. As
it was mentioned above, high-density tailor-made protein micro-
arrays cover a wide number of applications in personalized medicine
[6, 7]. Bearing this in mind, it is very relevant to choose the best
options of the protein microarrays available strategies (i.e., content,
format, detection system, experimental conditions) for the purpose
of the protein assay [8].
Here, most of these relevant features in experimental design are
briefly described:

1.1 Microarray The majority of the protein microarrays are developed in two for-
Format mats (depending of the surface) [6, 9–11]:

1.1.1 Planar Arrays In this type of arrays, the content (protein, peptide, aptamer, tissue,
or cell lysates) is immobilized in microspots arranged in a
two-dimensions (2D) surface (around 250 μm of diameter and
separated ~300 μm) over a solid matrix. In this 2D organized
spatial distribution, spot density around 1000 spot per square
centimeter is normally reached in most of the protein microarrays
commercially available [4, 11].
In these arrays, several aspects are directly related with the
robustness and reproducibility of the assay performance [12],
such as spot (size, morphology, and reproducibility), the ligand
(binding capacity), the sample, surface, and method (background
signal), and the detection limits. In addition, other aspects are also
related to physicochemical properties of the surface [13] or bio-
molecules, which have a consequence in the assay development,
such as spotting buffer composition affecting the protein structure,
printing method (contact or noncontact) [14], hygroscopy, and
humidity. In addition, the ratio background/signal is important;
then, the unspecific binding onto the array surface has to be eval-
uated and controlled in order to detect correctly the ligand-agent
 A Systematic Analysis Workflow for High-Density Customized Protein. . . 109

union [15]. For this purpose, classical blocking buffers as bovine

serum albumin (BSA) or skimmed milk powder at different con-
centrations are typically used.

1.1.2 Bead Arrays In this format, the ligand is bound to addressable beads (color-
coded beads or quantum dots), whose diameter typically varies
from 0.02 to 0.1 μm (nanoparticles) or from 0.1 μm (micro-
spheres) [16]. Usually, to be able to distinguish them, beads with
different ligands are previously labeled with different combination
of fluorescent dyes.
Then, the color-coded beads are easily detected by flow cyto-
metry (in which one or several lasers excite the internal dye), and a
reporter fluorescence dye (which is directly linked to the identifica-
tion or quantification of the target protein, and excitation/emis-
sion are far different from the ones used for the internal color-
coded dyes). The detector captures the color profile and identifies
the ligand and therefore the target protein (by an assigned intensity
to the analyte).

1.1.3 Microarray Content Protein microarrays offer a wide diversity of subclasses according to
the biomolecules deposited or displayed onto the surface. To sim-
plify the classification, they can be categorized as assembled arrays
or self-assembled arrays [17].

Assembled Arrays Composed by, typically, antibodies, purified proteins or other enti-
ties which are immobilized onto a functionalized surface. Some
types of arrays included in this category are:
1. Capture arrays
They are generated by printing analyte-specific reagents
(ASRs), usually antibodies [18] (Fig. 1), but sometimes phages
[19] or others, over the array surface. These ASRs serve to
identify and quantify the presence of multiple entities simulta-
neously. Therefore, they are used to find biomarkers and detect
molecular signatures. The quality of the results depends on the
quality (specificity and affinity) of the ASRs, which, in the case
of antibodies, is related to them being poly- or monoclonal.
There are different detection methods for measuring the
analyte-reagent union, being direct (e.g., fluorescent, Cy3
and/or Cy5 antibodies) or indirect (biotinylated sample
revealed with streptavidine, or secondary antibodies marked
with HRP) [17].
2. Reverse-phase arrays
The concept is opposite to the capture arrays. On these arrays,
the samples are deposited onto the surface (Fig. 2). The
reverse-phase arrays are very useful to evaluate many samples
against a single ASR. It is critical to ensure the affinity of the
110 Rodrigo Garcı́a-Valiente et al.

Fig. 1 Scheme of a capture microarray. In this particular example, it is labeled

directly

ASR in order to avoid cross-reactivity. It is good to evaluate

theoretical protein pathways in a high-throughput format
[20]. However, it is highly time-consuming, and it, contrary
to the capture arrays, may have difficulties detecting low abun-
dance ligands in complex samples [10].

Self-Assembled Arrays In these arrays, the protein is in situ expressed by an in vitro

transcription-translation system from an immobilized cDNA
encoding a protein of interest with a tag in carboxy or amino
terminus (Fig. 3). It allows functional characterization of the in
situ expressed proteins and also the identification of posttransla-
tional modifications. Several types have been described so far: PISA
[21], DAPA [22], PuCA [23], and NAPPA [24], the latter being
optimized for high-throughput analysis.
For biomarker discovery, the more generalist and accessible
platform is based on planar capture single-color microarrays printed
by noncontact and incubated with chemically labeled samples (i.e.,
biotinylated), that allow to operate with a high number of samples
independently. Therefore, this is the case that will be described in
this chapter.
 A Systematic Analysis Workflow for High-Density Customized Protein. . . 111

Fig. 2 Scheme of a reverse-phase protein microarray. A fluorescent tag is bound

to each specific antibody

2 Materials

2.1 Experimental Prepare all materials at room temperature unless indicated other-
Materials wise.

2.1.1 Array Printing l General material

Sample Preparation – JetStar™ Microarray-Specific 384 Microplates.
– Micropipettes P10, P100 and correspondent tips.
– Tube racks.
– 1.5 mL tubes
– Beaker of
200 mL.
l Reactives
– PBS Na/K 1.
– 47% Glycerol.
– BS3.
l Benchtop instruments
– Thermoblock with agitator.
– Centrifuge.
– Vortex.
112 Rodrigo Garcı́a-Valiente et al.

Fig. 3 Scheme of a self-assembled protein microarray. In this example, a nucleic

acid programmable protein array—NAPPA

Array Printing l General material

– Bel-bulb pipettor.
l Reactives
– Chemically active-surface slides.
l Benchtop instruments
– Injection array printer, e.g., Arrayjet Marathon Argus.
– Sonicator.

2.1.2 Array Assays l General material

– Microarray incubation chamber.
– Microarray washing chamber.
– Wet chamber.
– 500 mL beaker.
– Micropipettes P10, P100 and correspondent tips.
– 1.5 mL tubes.
– Cover slips.
 A Systematic Analysis Workflow for High-Density Customized Protein. . . 113

l Reactives
– Blocking buffer: PBS Na/K 1 + BSA 1% (w/v), 0.2% (w/v)
Tween 20.
– Streptavidine/fluor conjugate.
l Benchtop instruments
– Agitator.
– Orbital agitator.
– Array washer platform.

2.1.3 Array Image l Benchtop instruments.

Adquisition – Microarray scanner, e.g., Sensovation’s Fluorescent Array
Imaging Reader.

2.2 Computational Recommended hardware requisites for the full computational

Resources workflow:
l IBM-compatible computer with an Intel Quad Core processor
1.8 GHz or faster.
l Microsoft Windows 7 64-bit edition operating system or
superior.
l 8 GB RAM or more.
l Dedicated Video Card 512 MB or greater.
l 256 GB SSD unit (for image storage).
l 1280  1024 display system with 16 M colors.

2.2.1 Image Analysis Recommended software:

l GenePix Pro software v. 7 or superior.
l Notepad++ v 7.5 or superior.

2.2.2 Data Analysis Recommended software:

l R v. 3.0.1 or superior and RStudio v. 1.1.313 or superior, or,
alternatively, Microsoft Excel 2010 or superior.

3 Methods

3.1 Customize Customize arrays (Fig. 4) will be designed according to the partic-
Design ular needs of the specific screening. Some aspects have to be taken
into account in an array-based assay.
1. According to the study, select appropriate Positive and Nega-
tive Controls and include them among the ligands. A standard
negative control is the cleaning buffer. Including internal con-
trols is an important tool to have an estimation of the behavior
of the array and/or the sample. “The more, the merrier.”
114 Rodrigo Garcı́a-Valiente et al.

Fig. 4 Scheme of an antibody microarray printing followed by the assay. In the first phase (1), each specific
antibody for each target protein is prepared in a plate, eluted in its corresponding master mix. The printing of
the array (2) is carried out in a chamber with controlled temperature and humidity. Between each spot, the pin
has to be cleaned with cleaning buffer to prevent cross contamination. After printing, the batch of arrays are
dried and can be conserved for posterior use. Before using the microarray, it has to be blocked to prevent
unspecific binding (3). After blocking and also after all the following steps (4 and 5) the microarray has to be
thoroughly rinsed with distilled water. The array is incubated with the sample (4) and developed (5). After the
last rinse, the array is dried and can be scanned. The microarrays can be preserved in darkness at r.t. in
controlled humidity

2. Enough number of replicates have to be included. Technical

replicates will give information about the quality of the
biological and experimental aspects of the experiment, while
biological replicates (at least three, more if possible) will pro-
vide information about the biological issue.
3. When working with planar microarrays, if there is a high num-
ber of samples and a moderate number of biomarkers to work
with, an array may be divided in subsections called subarrays,
each of one will have the same content as the others, but will be
hybridized with a different sample. Therefore, the same sample
in different subarrays will be considered technical replicates,
while different samples of the same group in different subarrays
will be considered biological replicates. On each subarray, all
the content has to be displayed in, at least, triplicates, randomly
distributed across the surface. Ideally, from 15 to 30 spots is a
very suitable option (Fig. 4).
 A Systematic Analysis Workflow for High-Density Customized Protein. . . 115

3.2 Experimental Carry out all procedures at room temperature unless otherwise is
Workflow specified on a protocol step.
In case of a high number of samples to be printed at once, then,
technical replicates and biological replicates must be randomly and
uniformly separated between the different experimental batches, to
decrease the so-called batch effect.
It is important to optimize the amount of sample and ligand.
For that, it is recommended to do previous smaller assays combin-
ing different dilutions of each, and to choose the combination that
offers a less background/signal ratio.

3.2.1 Array Printing 1. Create an Excel file with the future and random distribution of
the samples, negative controls and positive controls to print in
Sample Preparation
the microplate (see Note 1).
2. Clean the laboratory work bench with 70% ethanol. Prepare
required samples and reactives for the microplate preparation.
3. Protein samples shall be diluted in PBS Na/K 1. As cross-
linker BS3 at 50 mg/mL is used.
4. Microplates will be loaded according to the chosen distribu-
tion. Each sample will be loaded in a 1:1 dilution (v/v) with
47% glycerol (v/v).
5. After the microplates are set up, spin them in the centrifuge,
using a microtiter adaptor.

Array Printing 1. Turn on the injection array printer, the associated control
station, and linked computer.
2. Perform maintenance routine, specified by the manufacturer.
3. Introduce in the printer the microplate with the samples to
print, and the functionalized slides, the microarrays, using the
command to Load microplates and slides, ensuring that the
microplates and the slides are correctly positioned.
4. In the computer, create a folder for the experiment. In said
folder, save the execution parameters.
5. Start the printing, pressing the corresponding command. The
machine will start printing each array automatically.
6. Once the printing process is over, save the quality results in the
earlier mentioned folder.
7. Re-initialize the system (see Note 2).
8. Extract the microplate and store them in the conditions
required by the samples.
9. Extract the printed microarrays. Label them (see Note 3).
10. Dry the printed microarrays in the stove at 37  C along with an
absorbent agent, silica gel.
11. Store the microarrays at room temperature.
116 Rodrigo Garcı́a-Valiente et al.

3.2.2 Array Assays 1. Microarray blocking. Submerge the microarrays in the micro-
array washing chambers, in 6 mL of blocking solution on each
one of them, during 1 h, in orbital agitation (see Note 4).
2. Microarray washing. Wash intensely during 10 min using
milliQ water in the array washer platform. After this, wash
three times, placing each array in the array washing chambers,
filled with distilled water, in orbital agitation for 5 min each.
They will be kept in distilled water until their processing.
3. Biologic sample processing. In this case, sample must be bioti-
nilized, following the protocol described by Sierra A. et al.,
Journal Proteome Research 2016.
4. Microarray drying. Microarrays will be dried using centrifuga-
tion at 240  g during 3 min, using the adaptors for 50 mL
tubes. Once dried, they can be stored at 4  C during a maxi-
mum of 15 days.
5. Incubate biological samples at the chosen dilution (see Note 5)
overnight and at 4  C at orbital agitation.
6. Microarray washing (see Note 4). Wash for 7 min using milliQ
water in the array washer platform. After this, wash three times,
placing each array in the array washing chambers, filled with
distilled water, in orbital agitation for 5 min each. They will be
kept in distilled water until their processing.
7. Sample incubation by indirect method given marked -biotiny-
lated- samples.
(a) Preparation of the streptavidine. Dilute 1:200 (v/v) of
0.1 mg of streptavidine/fluor conjugate in milliQ water.
(b) Wet chamber preparation. Add distilled water to the wet
chamber, enough to create humidity, but not enough to
touch the arrays when they are added.
(c) Incubation. Put the cover slips over the arrays, and the
arrays in the wet chamber without touching the water.
Incubate 200 μL of the reveal sample over the full micro-
array for 20 min in the wet chamber, at no light
conditions.
8. Microarray washing (see Note 4). Wash for 7 min using milliQ
water in the array washer platform. After this, wash three times,
placing each array in the array washing chambers, filled with
distilled water, in orbital agitation for 5 min each. They will be
kept in distilled water until their processing.
9. Microarray drying. Microarrays will be dried using centrifuga-
tion at 240  g during 3 min, using the adaptors for 50 mL
tubes. Once dried, they have to be stored at no light exposure
until scanned.
 A Systematic Analysis Workflow for High-Density Customized Protein. . . 117

Fig. 5 General scheme of the analysis process after the assay. After the assay, the image is scanned in a
fluorescent array imaging reader, which generates a file with the associated data for further statistical analysis

3.2.3 Array Image This protocol (Fig. 5) is designed to be performed in a Sensovation

Adquisition Fluorescent Array Imaging Reader. When operating with a differ-
ent scanner, it should be adapted according to the characteristics of
the instrument.
0. Turn on the equipment. Wait until everything is correctly
loaded.
1. Open the Sensovation program.
2. Open the device to load as far as four arrays. Steps 2–5 shall
be repeated until all arrays have been scanned.
3. Click the Setup button, Rack Configuration to establish time
exposure and focus parameters, to establish the correct spot visual-
ization regarding the background signal (Fig. 6). Spots should be
well delimited, and background noise should be homogeneous and
well contrasted. Save parameters.
4. Click the Setup button, Assay Configuration to establish
scanning options. Save them when correct.
5. Scan and save .tiff results (see Note 6).
118 Rodrigo Garcı́a-Valiente et al.

Fig. 6 Unacceptable rack configurations (left) vs. correct rack configurations (right)

3.3 Computational For this step, we recommend using the GenePix software, which
Analysis allows us to obtain a GenePix Results file (GPR). This image
intensity file is needed for ulterior steps.
3.3.1 Image Analysis
For each generated image:
1. Open image (Ctrl+O).
2. Choose correct wavelength.
3. Adjust brightness and contrast.
4. Open the gal file/Array List (Alt+Y).
5. Adjust gal (manually and/or pressing F5 for automatic mode).
6. Analyze (Ctrl+A).
7. Configure. Background subtraction method: local.
8. Save results as .gpr.
9. Explore .gpr files with Notepad++, focusing on spot IDs, and
checking they are correct.

3.3.2 Data Analysis In order to obtain reliable and meaningful biological information
from a microarray experiment, it has to be analyzed with a consis-
tent statistical method (see Note 7). There are different available
strategies to analyze microarrays, but not all of them, which were
developed to analyze other microarray technologies such as DNA
microarrays, are easily translatable to protein microarrays. The
common steps to perform an analysis for this kind of assay are the
following:
1. Import data sets. The Genepix output must be read into the
software that will be used for the analysis. Suitable software
choices for this task could be Python, R, or Matlab. GPR files
contain, among other parameters, the mean, median, and stan-
dard deviation of the pixel intensity, as well as the total intensity
for a spot at a given wavelength.
2. Background subtraction. A simple way to subtract the back-
ground for each intensity is using the default background
measure from Genepix. Working with median values is
recommended.
 A Systematic Analysis Workflow for High-Density Customized Protein. . . 119

Fig. 7 Boxplots showing the distribution of the median intensities of the background (gray) and foreground
(yellow) of every spot after logarithmic transformation. Seven assays are shown

3. A quality control check can be performed at this point, by

detecting assays with an abnormally low overall signal (Fig. 7).
4. After subtracting the background, negative values are expected,
which must be substituted by null values before step 5.
5. Apply a logarithmic transformation to the data sets.
6. Set a cutoff point to differentiate negative and positive spots
(Fig. 8). This can be automatically achieved by determining the
minimum between the positive and negative distribution, this
can be achieved with a kernel density estimation of the data
distribution.
7. Data normalization is required for establishing comparison
across several assays. Various scaling methods are suited for
this task (Fig. 9). Standard scoring is easy to implement, but
it has to be applied to each assay individually with its respective
mean and standard deviation. Quantile normalization could be
used on demand [25] if some distributions display noisy pat-
terns caused by technical variability but it could also subtract
biological variability and trade off further statistical power.
120 Rodrigo Garcı́a-Valiente et al.

Fig. 8 Smooth histogram of the logarithmic transformed median intensity (after background is subtracted) of
all spots in a microarray. Different colors for each subarray, in case multiple assays are being carried over the
same slide. The cutoff point is plotted as the vertical blue dotted line. All points with less intensity than the
cutoff point will be considered negative and the ones with higher intensity will be further evaluated

Fig. 9 Smooth histogram (top) and Boxplots (bottom) showing the median intensities of the positive spots (after
logarithmic transformation) for every assay. Data not normalized (left), with standard scoring (center) and with
a quantile normalization (right) are shown

8. Positive spots must be compared with negative control spots

which act as a true nonarbitrary threshold for positivity on each
individual assay.
 A Systematic Analysis Workflow for High-Density Customized Protein. . . 121

9. To evaluate inter-array variability, same analyte spots that show

true positive intensity (selected after the previous step) are
counted. An index of confidence (IC), that ranges from 0 to
1, from negative to positive detection in all replicates, respec-
tively, is assigned to every target protein (number of positive
spots divided by total number of spots for that analyte).
10. Selection of differentially expressed proteins can be achieved by
comparing the mean of the IC for each protein between groups
of samples. This can be achieved with a standard t-test with a
procedure to control FDR.

4 Notes

1. One of the recommended negative controls is 47% glycerol

(v/v).
2. Between different printing essays, the device must be washed.
3. It is recommended that each label includes the batch, sample,
and array information.
4. From this step on until the step 8, the slides/microarrays must
be kept wet at all points.
5. To choose an optimal dilution for the biological samples doing
previously the Subheading 3.2.2 with a reduced set of known
control samples in different dilutions. However, typically 1:100
dilutions work well.
6. It is usually recommended to generate backup files, and it is
essential to label correctly each file according to the scanned
array.
7. These analysis can be executed in the online tool ProtArray
(www.ProtArray.com).

Acknowledgments

We gratefully acknowledge financial support from the Spanish

Health Institute Carlos III (ISCIII) for the grants: FIS PI14/
01538, FIS PI17/01930, and CB16/12/00400. We also
acknowledge Fondos FEDER (EU) and Junta Castilla-León
(grant SA198A12-2). Fundación Solórzano FS38/2017. The Pro-
teomics Unit belongs to ProteoRed, PRB3-ISCIII, supported by
grant PT17/0019/0023, of the PE I + D + I 2017-2020, funded
by ISCIII and FEDER.
122 Rodrigo Garcı́a-Valiente et al.

References
1. Sierra-Sánchez Á, Garrido-Martı́n D, 14. Glökler J, Angenendt P (2003) Protein and
Lourido L, González-González M, Dı́ez P, antibody microarray technology. J Chromatogr
Ruiz-Romero C et al (2017) Screening and B Anal Technol Biomed Life Sci 797:229–240
validation of novel biomarkers in osteoarticular 15. Kusnezow W, Jacob A, Walijew A, Diehl F,
pathologies by comprehensive combination of Hoheisel JD (2003) Antibody microarrays: an
protein array technologies. J Proteome Res 16 evaluation of production parameters. Proteo-
(5):1890–1899 mics 3(3):254–264
2. Zyuzin MV, Dı́ez P, Goldsmith M, Carregal- 16. Casado-Vela J, González-González M,
Romero S, Teodosio C, Rejman J et al (2017) Matarraz S, Martı́nez-Esteso MJ, Vilella M,
Comprehensive and systematic analysis of the Sayagués JM et al (2013) Protein arrays: recent
immunocompatibility of polyelectrolyte cap- achievements and their application to study the
sules. Bioconjug Chem 28(2):556–564 human proteome. Curr Proteomics 10
3. Dı́ez P, Ibarrola N, Dégano RM, Lécrevisse Q, (2):83–97. https://doi.org/10.2174/
Rodriguez-Caballero A, Criado I et al (2017) A 1570164611310020003
systematic approach for peptide characteriza- 17. Lourido L, Diez P, Dasilva N, Gonzalez-
tion of B-cell receptor in chronic lymphocytic Gonzalez M, Ruiz-Romero C, Blanco F, et al
leukemia cells. Oncotarget 8 (2014) Protein microarrays: overview, applica-
(26):42836–42846 tions and challenges. In: Genomics and prote-
4. Merbl Y, Kirschner MW (2011) Protein micro- omics for clinical discovery and development.
arrays for genome-wide posttranslational mod- Springer. p 147–173., https://doi.org/10.
ification analysis. Wiley Interdiscip Rev Syst 1007/978-94-017-9202-8_8
Biol Med 3(3):347–356 18. LaBaer J, Ramachandran N (2005) Protein
5. Dasgupta A (2008) Handbook of drug moni- microarrays as tools for functional proteomics.
toring methods: therapeutics and drugs of Curr Opin Chem Biol 9:14–19
abuse. Humana, Totowa, NJ, pp 1–445 19. Jara-Acevedo R, Dı́ez P, González-
6. Yu X, Schneiderhan-Marra N, Joos TO (2011) González M, Dégano RM, Ibarrola N, Gón-
Protein microarrays and personalized medi- gora R et al (2018) Screening phage-display
cine. Ann Biol Clin (Paris) 69(1):17–29 antibody libraries using protein arrays. In:
7. Yu X, Schneiderhan-Marra N, Joos TO (2010) Phage display. Methods Mol Biol
Protein microarrays for personalized medicine. 1701:365–380
Clin Chem 56:376–387 20. Spurrier B, Ramalingam S, Nishizuka S (2008)
8. Dı́ez P, Dasilva N, González-González M, Reverse-phase protein lysate microarrays for
Matarraz S, Casado-Vela J, Orfao A et al cell signaling analysis. Nat Protoc 3
(2012) Data analysis strategies for protein (11):1796–1808
microarrays. Microarrays 1(3):64–83 http:// 21. He M, Taussig MJ (2001) Single step genera-
www.mdpi.com/2076-3905/1/2/64/ tion of protein arrays from DNA by cell-free
9. Gonzalez-Gonzalez M, Jara-Acevedo R, expression and in situ immobilisation (PISA
Matarraz S, Jara-Acevedo M, Paradinas S, Saya- method). Nucleic Acids Res 29(15):E73–E73
gües JM et al (2012) Nanotechniques in prote- http://www.ncbi.nlm.nih.gov/entrez/query.
omics: protein microarrays and novel detection fcgi?cmd¼Retrieve&db¼PubMed&dopt¼Cita
platforms. Eur J Pharm Sci 45:499–506 tion&list_uids¼11470888
10. Dasilva N, Dı́ez P, Matarraz S, González- 22. He M, Stoevesandt O, Palmer EA, Khan F,
González M, Paradinas S, Orfao A et al Ericsson O, Taussig MJ (2008) Printing pro-
(2012) Biomarker discovery by novel sensors tein arrays from DNA arrays. Nat Methods 5
based on nanoproteomics approaches. Sensors (2):175–177
12:2284–2308 23. Tao SC, Zhu H (2006) Protein chip fabrication
11. Matarraz S, González-González M, Jara M, by capture of nascent polypeptides. Nat Bio-
Orfao A, Fuentes M (2011) New technologies technol 24(10):1253–1254
in cancer. Protein microarrays for biomarker 24. Ramachandran N, Raphael JV, Hainsworth E,
discovery. Clin Transl Oncol 13:156–161 Demirkan G, Fuentes MG, Rolfs A et al (2008)
12. Ellington AA, Kullo IJ, Bailey KR, Klee GG Next-generation high-density self-assembling
(2010) Antibody-based protein multiplex plat- functional protein arrays. Nat Methods 5
forms: technical and operational challenges. (6):535–538
Clin Chem 56:186–193 25. Hicks SC, Irizarry RA (2014) When to use quan-
13. Fuentes M, Dı́ez P, Casado-Vela J (2016) tile normalization? bioRxiv. doi: https://doi.
Nanotechnology in the fabrication of protein org/10.1101/012203. http://biorxiv.org/con
microarrays. Methods Mol Biol 1368:197–208 tent/early/2014/12/04/012203.abstract
 Chapter 8

Metaproteomics Study of the Gut Microbiome

Lisa A. Lai, Zachary Tong, Ru Chen, and Sheng Pan

Abstract
Proteomics is a widely used method for defining the protein composition of a complex sample. As this
approach allows for identification and quantification of proteins across a broad dynamic range as well as
detection of post-translational modifications, proteomics is an ideal platform to investigate the gut micro-
biome at a functional level. The gut microbiome is a dynamic environment which is crucial for overall health
and fitness. Imbalances in the gut microbiome can influence nutrient absorption, pathogen resistance,
inflammation, and various human diseases. Metaproteomic analysis of the gut microbiome is currently
being performed on bacteria isolated from (1) fecal samples (2) colonic lavage, or (3) colon biopsies.
Investigation of the gut microbiome has demonstrated that within the colon, there are distinct commu-
nities based on spatial location, and separable from the gut microbiomes isolated from stool. In addition to
expanding our understanding of host–bacterial interactions for human health and disease, gut microbiome
analysis is being utilized for biomarker development to discriminate normal individuals and diseased (i.e.,
inflammatory bowel disease or colon cancer) patients as well as to monitor disease activity and prognosis.

Key words Microbiome, Proteomics, Metaproteomics

1 The Gut Microbiota: An Introduction

The adult human gastrointestinal (GI) tract runs from the esopha-
gus through the stomach and colon to the rectum. These organs
are host to an enormous population of microorganisms, possibly
upward of 100 trillion representing between 15,000 and 36,000
different species of bacteria [1–3]. Through host interaction, these
bacteria, fungi, and viruses respond to stimuli within their micro-
environment and impact a broad spectrum of essential functions
including assisting with digestion of food, vitamin production/
absorption, metabolism, nutrient extraction, immune response,
and conferring resistance to pathogenic organisms [4, 5]. While
the numbers and density of bacteria within the human gut are
extremely high, the diversity is surprisingly low. Bacteria from
four phyla Bacteroides, Firmicutes, Proteobacteria, and

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_8, © Springer Science+Business Media, LLC, part of Springer Nature 2019

123
124 Lisa A. Lai et al.

Actinobacteria constitute >98% of the microbes within the human

GI tract [1], posing a unique challenge for metaproteomics analysis
of gut microbiome, particularly with regard to species with low
abundance.
The human gut microbiome begins to develop in utero as
evidenced by microbiota detected in the placenta [6], amniotic
fluid [7], and umbilical cord [8] and is fully colonized shortly
after birth [9]. The development of the infant gut microbiota is
impacted by the mode of delivery [10]. Infants delivered vaginally
showed colonization by Lactobacillus and Prevotella whereas cesar-
ean section delivered infants are first colonized by Proteobacteria
and Firmicutes. This difference can persist up to the first 12 months
of life. Rapid bacterial expansion and colonization of the gut occurs
during early childhood such that by pre-adolescence, children have
similar gut microbiomes to adults [11]. Once established, however,
the gut microbiome does not stay static but rather is highly
dynamic and can respond to environmental stimuli such as changes
in diet and antibiotic treatment, among other factors.
Recently, the Human Microbiome Project (HMP) set out to
characterize the microbiome in healthy individuals at multiple sites
including the GI tract and feces [12]. Interestingly, differences in
the microbial populations found in the lumen (stool) and in colon
tissue samples from the same individual were reported [12, 13],
suggesting that microbial communities may be spatially distinct.
While the large intestine contained greater diversity than the small
intestine, the microbial populations were similar, and dominated by
bacteria of the phyla Bacteriodetes and Firmicutes [1]. The metage-
nomics data sets generated from the HMP has laid an important
foundation for metaproteomics analysis of human gut microbiome.
As the next step, the Integrative Human Microbiome Project
(iHMP) focuses on the longitudinal investigation of microbiomes
in various patient cohorts, including neonates, inflammatory bowel
disease (IBD) patients, and diabetes patients [14]. Dysbiosis or an
imbalance of the gut microbiota has been implicated in the patho-
genesis of IBD [15], obesity [16, 17], metabolic disorders, and
cardiovascular disease [18].

2 Modulation of the Gut Microbiome by Diet

There is much interest in the gut microbiome since it is a dynamic

host-interaction which is responsive to environmental stimuli,
including changes in diet. When mice were switched from a low
fat plant-based diet to a high fat “Western” diet, microbiome gene
expression changes were noted within a single day [19]. Diet stud-
ies in humans have shown changes in bacterial levels as needed for
metabolism of plant polysaccharides in study participants eating a
plant-based diet compared to increases in carbohydrate and protein
 Metaproteomics Study of the Gut Microbiome 125

fermentation for those on an animal-based diet [20]. Similar

changes in carbohydrate metabolism were reported when compar-
ing fecal microbiomes from rural vs. metropolitan communities
showed differences in bacterial populations and profiles [21].

3 Changes in Gut Microbiota in Disease

There has been great interest in studying microbiome-dependent

inflammation in diseases of the human gut, specifically, inflamma-
tory bowel disease (i.e., ulcerative colitis (UC) or Crohn’s disease
(CD) [22]) and colorectal cancer [23, 24]. It has been shown that
colorectal cancers are colonized by Fusobacterium and maintained
in distal metastases [24]. In addition, treatment of mouse xenograft
tumors with the antibiotic metronidazole reduced bacterial load
and tumor growth [24]. In both feces and tissue samples, reduced
diversity and imbalance of gut profiles have been reported in
patients with GI disease.
In addition to studying causes of disease and disease pathogen-
esis, researchers are interested in the effects of reintroduction of
bacteria to alleviate dysbiosis. Several studies have shown that fecal
transplant is an effective treatment to reduce disease activity in IBD
patients [25] or to treat recurrent diarrhea caused by C. difficile
infection [26]. A recent study showed that fresh or frozen human
feces could be successfully transplanted into gnotobiotic mice
resulting in a gut microbiome which recapitulates the human
donor within 1 week of transplant; the utility of this animal
model for drug trials targeting gut microbiota was discussed [19].

4 Metaproteomics of Complex Samples Including Gut

While proteomics provides a comprehensive identification and

quantitation of proteins within a sample, metaproteomics is the
comprehensive characterization of expressed proteins within a
microbiome community at a given point in time [27]. Metaproteo-
mics has been used to investigate microbiomes in human and
animals as well as in environmental communities such as soil [28],
sludge [27], food [29], and the ocean [30]. This application is
currently less common than metagenomics or metatranscriptomics
studies due in part to the lack of consistent protocols for metapro-
teomic sample preparation, lack of efficient bioinformatics tools
[31], and challenges of measuring low-abundance proteins within
a complex sample [32]. Metaproteomics has been dependent upon
databases and libraries generated from genome and metagenomics
data for correct peptide identification and pathway analysis as well.
Historically, metaproteomics studies of the gut focused on bac-
terial populations isolated following short-term culture, but this
126 Lisa A. Lai et al.

approach was limited to strains that could be grown in vitro and

oftentimes these cultured strains did not exhibit the same gut
microbiota profiles. However, with advances in mass spectrometry,
researchers have been increasingly successful at interrogating com-
plex microbiomes. One study used two-dimensional differential gel
electrophoresis (2D-DIGE) along with tandem mass spectrometry
(MS/MS) and validation by SRM-based targeted proteomics to
show that CD patients had overrepresentation of Bacteroides species
and underrepresentation of Clostridiales, elevated expression of
proteins involved in oxidative stress, energy saving, and IgA immu-
noglobulins and decreased GP2 (pancreatic glycoprotein 2 of zymo-
gen granule membranes) which may promote inflammation [33].
Despite great advances in the field, this is still in the develop-
mental stages due to the staggering complexity of the gut micro-
biome—more than 63 million unique proteins expressed from
upward of 2100 different taxa [34]. Metaproteomic analysis
requires enormous computing effort and power. In addition, the
highly resistant cell walls of Gram-negative bacteria (which com-
prise but a portion of the strains within a microbiome) require
additional mechanical cell wall disruption methods, such as bead
beating or sonication, for optimal protein extraction [32, 35]. Also,
there are challenges to correctly identifying peptides from homolo-
gous proteins, resulting in redundant protein identification and
possibly skewing analysis since similar proteins from distinct species
can have quite different functions.

5 Metaproteomics Using a Shotgun Proteomics Approach

Early metaproteomic analysis used two-dimensional gel electro-

phoresis (2D-GE) to separate proteins prior to mass spectrometry
(MS). Then, shotgun metaproteomics using two-dimensional liq-
uid chromatography (LC) coupled with nanospray tandem mass
spectrometry (nano 2D LC-MS/MS) was used to interrogate
infant fecal samples and demonstrated increased protein identifica-
tion following enrichment of low-abundance microbial proteins
from fecal samples [36]. This approach was used to show a signifi-
cant decrease in proteins from the Firmicutes phylum in CD
patients, higher abundance of inflammatory response proteins,
and decreased expression of proteins involved in maintaining
mucosal integrity, all of which may contribute to chronic inflam-
mation [37]. An optimized LC-MS/MS workflow for analysis of
the mouse fecal microbiome revealed identifications of 18,000
non-redundant tryptic peptides (93% of microbial origin), repre-
senting over 600 different microbial species and 250 protein
families including members of the TonB-dependent receptor family
which are involved in energy production [38].
 Metaproteomics Study of the Gut Microbiome 127

6 Sample Collection to Investigate the Gut Microbiome

Analysis of the human gut microbiome has been approached using

basically three types of clinically obtained materials—(1) stool,
(2) colonic biopsies, and (3) colonic lavage. Of note, several studies
have noted disparities between the microbiomes from paired fecal
and rectal (colon) samples, even ones taken on the same day and
without colonoscopy prep [39]. There are benefits and caveats to
analysis of each type of clinical material, some of which are outlined
below.
Many metaproteomics studies use fecal or stool samples since
this material can be collected noninvasively and there is a large
biomass of material to work with, considering that up to 30% of
fecal biomass may be bacteria. Stool is a mixture of host cells,
bacteria, food particles, and insoluble material, requiring enrich-
ment by differential centrifugation [40], filtering to isolate micro-
bial cells from larger insoluble particles [36], and/or precipitation.
Protein extraction from feces yields a combination of bacterial and
host secreted proteins, thus allowing interpretation of the interplay
between host and bacteria. The first shotgun metaproteomics study
of the human gut using fecal samples showed higher than expected
expression of proteins related to translation, energy production,
and carbohydrate metabolism, as well as proteins involved in
novel microbial pathways and host immune response [40].
Colonic biopsies can be recovered during routine colonoscopy,
and while they are minimally invasive, they do require disruption of
the mucosal layer. While these biopsies sample a small area within
the colon, they can be targeted for regions of dysplasia or specific
regions within the colon (i.e., cecum, transverse colon, proximal
colon, rectum). While the number of bacteria in a fecal sample can
be up to 103 higher than in a biopsy sample, there is a difference in
the bacterial communities isolated [41]. PCR-based techniques
have been used to compare the profiles of bacterial subcommunities
isolated from different parts of the colon as well as in feces.
Although the predominant species detected from colon biopsies
were consistently found independent of biopsy location, there was a
distinct difference with the fecal samples, suggesting that fecal
contamination during the colonoscopy was unlikely. The differ-
ences between colon and fecal microbiomes are less than between
individuals [13].
Colonic lavage is a method of collecting bacteria following
injection of a small amount of fluid into the colon during colonos-
copy [42]. However, patients undergoing colonoscopy usually have
completed a prep protocol, so much of the colon is cleansed before
the procedure. Furthermore, analysis of mucosal lavage revealed a
significant proportion of secreted human proteins (up to 63%)
mixed in with the bacterial peptides (30%) [42]. By comparison,
128 Lisa A. Lai et al.

analysis of proteins isolated from fecal samples showed approxi-

mately 30% of identified proteins were of human origin
[40]. Some studies suggest that mucosal lavage is a preferred sam-
pling methodology since it would allow for niche-specific profiling
as well as repeated sampling. Since the surface microbiota are
recovered by lavage, they may be better adapted for adherence,
host resistance, and other mucosal trophic factors than bacteria
isolated from fecal samples. One study observed that in-gel diges-
tion of mucosal lavage samples greatly increased the efficiency of
trypsin digestion, which they postulate was likely due to inactiva-
tion of the trypsin inhibitor A1AT [42].

7 Challenges with Gut Microbiome Analysis

Some of the major problems in recovering microbiota from fecal

samples are that samples are extremely heterogeneous, high
amounts of insoluble material could skew protein measurements,
and improper storage conditions could contribute to bacterial lysis
[43]. To get around the heterogeneity issue, some researchers have
homogenized the samples prior to aliquotting. Researchers have
also successfully stored fecal samples in RNAlater rather than dry
freezing which may preserve the amount of bacteria recovered—
although storage conditions can have significant impact on micro-
biome profiles [43]. Bacteria recovered by extensive washing and
then separated from the supernatant contain both lysed or secreted
bacterial proteins and secreted host proteins. Comparison of differ-
ent protein extraction methods (including beat beating and ultra-
sonication) combined with various lysis buffers (SDS, B-Per and
urea) showed that while peptide identification and protein yield
were highest with SDS buffer and bead beating, only B-Per was
able to extract proteins from Bacteriodetes, and Actinobacteria was
detected only with lysis in urea buffer. Extraction using urea
resulted in good detection of post-translational modifications but
less so using other lysis buffers [32].

8 Specialized Software Tools for Metaproteomics Analysis

Most of the reported metaproteomics studies use shotgun proteo-

mics for identification of bacterial proteins which may be influenced
by complexity and sensitivity issues such that detection of low
abundant proteins may be challenging. However, the complexity
and dynamic range issues may be addressed using the emerging
spectral library based methods, such as data-independent acquisi-
tion (DIA), which provides a library-based, broader coverage for
peptide/protein detection [44–47]. While such an approach has
been increasingly applied in quantitative analysis of the human
 Metaproteomics Study of the Gut Microbiome 129

proteome, its applications in metaproteomics has lagged behind, in

part due to the complexity involved in bioinformatics. Reference
databases are constantly integrating mass spec analysis of cultured
bacterial and pathogen species to generate reference databases for
improved species identification [48].
Recently developed software tools designed specifically for
metaproteomic data analysis have become available as well. Meta-
Lab uses spectral clustering to improve peptide identification
speeds [49]. An example of classification of the peptides identified
from the metaproteomics analysis of a human fecal specimen is
presented in Fig. 1. Others have shown improved protein identifi-
cation in metaproteomics by employing de Bruijn graph assembly
to predict protein sequences from metagenomics sequence data and

Fig. 1 Pie chart illustrates the taxonomy distribution from metaproteomics analysis of a human fecal sample.
The center circle represents the organism with each concentric circle moving outward depicting taxonomies
(i.e., domain, kingdom, phylum, class, order, family, genus, and species)
130 Lisa A. Lai et al.

generate a reference database [50]. Taxon-specific classification of

peptide sequences can be performed using UniPept (http://
unipept.ugent.be) which uses shotgun proteomics data from Uni-
Prot KB with identification noise filtering to provide enhanced
biodiversity analysis [51, 52]. Another option for analysis of meta-
proteomics data is MetaPro-IQ, which is ideal for fecal samples
since the gut microbiome gene catalog was curated from fecal
studies which negates the need for matched metagenomics data
but makes it less applicable for other types of microbiome
samples [53].

9 Conclusions and Perspectives

With recent advances in mass spectrometry instrumentation and

data analysis, the comprehensive characterization of complex gut
microbiome samples is becoming increasingly feasible. While previ-
ous efforts have focused on detection and identification of bacterial
species, it is the hope that these efforts will blossom into analyses
which can integrate metagenomics and metatranscriptomics data to
provide better insight into the very complex and dynamic gut
microbiome. The ultimate goal of these studies is to define loca-
lized and global interactions which impact human health and dis-
ease. By understanding of the intimate relationship between
microbiota and gut, the hope is to find interventions which posi-
tively impact overall human health and fitness.

References

1. Frank DN, St Amand AL, Feldman RA et al 7. Urushiyama D, Suda W, Ohnishi E et al (2017)

(2007) Molecular-phylogenetic characteriza- Microbiome profile of the amniotic fluid as a
tion of microbial community imbalances in predictive biomarker of perinatal outcome. Sci
human inflammatory bowel diseases. PNAS Rep 7:12171
104:13780–13785 8. DiGiulio DB, Romero R, Amogan HP et al
2. Cresci GA (2015) The gut microbiome: what (2008) Microbial prevalence, diversity and
we do and don’t know. Nutr Clin Pract abundance in amniotic fluid during preterm
30:734–746 labor: a molecular and culture-based investiga-
3. Lynch SV, Pederson O (2016) The human tion. PLoS One 3:e3056
intestinal microbiome in health and disease. N 9. Milani C, Duranti S, Bottacini F et al (2017)
Engl J Med 375:2369–2379 The first microbial colonizers of the human
4. Tuddenham S, Sears CL (2015) The intestinal gut: composition, activities, and health impli-
microbiome and health. Curr Opin Infect Dis cations of the infant gut microbiota. Microbiol
28:464–470 Mol Biol Rev. https://doi.org/10.1128/
5. Rakoff-Nahoum S, Paglino J, Eslami-Varzaneh MMBR.00036-17
F et al (2014) Recognition of commensal 10. Biasucci G, Rubini M, Riboni S et al (2010)
microflora by toll-like receptors is required for Mode of delivery affects the bacterial commu-
intestinal homeostasis. Cell 118:229–241 nity in the newborn gut. Early Hum Dev
6. Aagaard K, Ma J, Anthony K et al (2014) The 86:13–15
placenta harbors a unique microbiome. Sci 11. Hollister EB, Riehle K, Luna RA et al (2015)
Transl Med 6:237ra65 Structure and function of the healthy
 Metaproteomics Study of the Gut Microbiome 131

preadolescent pediatric gut microbiome. fecal microbiota transplantation course in pedi-

Microbiome 3:36 atric patients with inflammatory bowel disease.
12. Human Microbiome Project Consortium Springer, Heidelberg, Boston, MA, pp 1–7
(2012) Structure, function and diversity of 26. Fischer M, Sipe B, Cheng YW et al (2017)
the healthy human microbiome. Nature Fecal microbiota transplant in severe and
486:207–212 severe-complicated Clostridium difficile: a
13. Eckburg PB, Bik EM, Bernstein CN et al promising treatment approach. Gut Microbes
(2005) Diversity of the human intestinal 8:289–302
microbial flora. Science 308:1635–1638 27. Wilmes P, Bond PL (2004) The application of
14. Proctor LM, Sechi S, DiGiacomo ND et al two-dimensional polyacrylamide gel electro-
(2014) The integrative human microbiome phoresis and downstream analyses to a mixed
project: dynamic analysis of microbiome-host community of prokaryotic microorganisms.
omics profiles during periods of human health Env Microbiol 6:911–920
and disease. Cell Host Microbe 16:276–289 28. Bastida F, Hernandez T, Garcia C (2014)
15. Tamboli CP, Neut C, Desreumaux P et al Metaproteomics of soils from semiarid envi-
(2015) Dysbiosis in inflammatory bowel dis- ronment: functional and phylogenetic informa-
ease. Gut 53:1–4 tion obtained with different protein extraction
16. Ley RE, Turnbaugh PJ, Klein S et al (2006) methods. J Proteomics 101:31–42
Microbial ecology: human gut microbes asso- 29. Maier TV, Lucio M, Lee H et al (2017) Impact
ciated with obesity. Nature 444:1022–1023 of dietary resistant starch on the human gut
17. Kolmeder CA, Ritari J, Verdam FJ et al (2015) microbiome, metaproteome, and metabolome.
Colonic metaproteomic signatures of active MBio 8:e01343–e01347
bacteria and the host in obesity. Proteomics 30. Williams TJ, Cavicchioli R (2014) Marine
15:3544–3522 metaproteomics: deciphering the microbial
18. Kang Y, Cai Y (2017) Gut microbiota and metabolic food web. Trends Microbiol
hypertension: from pathogenesis to new thera- 22:248–260
peutic strategies. Clin Res Hepatol Gastroen- 31. Heyer R, Schallert K, Zoun R et al (2017)
terol. https://doi.org/10.1016/j.clinre.2017. Challenges and perspectives of metaproteomic
09.006 data analysis. J Biotechnol 261:24–36
19. Turnbaugh PJ, Ridaura VK, Faith JJ et al 32. Zhang X, Li L, Mayne J et al (2017) Assessing
(2009) The effect of diet on the human gut the impact of protein extraction methods for
microbiome: a metagenomic analysis in huma- human gut metaproteomics. J Proteome.
nized gnotobiotic mice. Sci Transl Med https://doi.org/10.1016/j.jprot.2017.07.
1:6ra14 001
20. David LA, Maurice CF, Carmody RN et al 33. Juste C, Kreil DP, Beauvallet C et al (2014)
(2014) Diet rapidly and reproducibly alters Bacterial protein signals are associated with
the human gut microbiome. Nature Crohn’s disease. Gut 63:1566–1577
505:559–563 34. Wilmes P, Heintz-Buschart A, Bond PL (2015)
21. Yatsunenko T, Rey FE, Manary MJ et al (2012) A decade of metaproteomics: where we stand
Human gut microbiome viewed across age and and what the future holds. Proteomics
geography. Nature 486:222–227 15:3409–3417
22. Wright EK, Kamm MA, Teo SM et al (2015) 35. Glatter T, Ahrne E, Schmidt A (2015) Com-
Recent advances in characterizing the gastroin- parison of different sample preparation proto-
testinal microbiome in Crohn’s disease: a sys- cols reveals lysis buffer-specific extraction biases
tematic review. Inflamm Bowel Dis in gram-negative bacteria and human cells. J
21:1219–1228 Proteome Res 14:4472–4485
23. Arthur JC, Perez-Chanona E, Myhlbauer M 36. Xiong W, Giannone RJ, Morowitz MJ et al
et al (2012) Intestinal inflammation targets (2015) Development of an enhanced metapro-
cancer-inducing activity of the microbiota. Sci- teomic approach for deepening the micro-
ence 338:120–123 biome characterization of the human infant
24. Bullman S, Pedamallu CS, Sicinska E et al gut. J Proteome Res 14:133–141
(2017) Analysis of Fusobacterium persistence 37. Erickson AR, Cantarel BL, Lamendella R et al
and antibiotic response in colorectal cancer. (2012) Integrated metagenomics/metapro-
Science. https://doi.org/10.1126/science. teomics reveals human host-microbiota signa-
aal5240 tures of Crohn’s disease. PLoS One 7:e49138
25. Karolewska-Bochenek K, Grzesiowski P, 38. Tanca A, Palomba A, Pisanu S et al (2014) A
Banaszkiewicz A et al (2017) A two-week straightforward and efficieint analytical pipeline
132 Lisa A. Lai et al.

for metaproteome characterization. Micro- 46. Rosenberger G, Liu Y, Rost HL et al (2017)

biome 2:49 Inference and quantification of peptidoforms
39. Durban A, Abellan JJ, Jimenez-Hernandez N in large sample cohorts by SWATH-MS. Nat
et al (2011) Assessing gut microbial diversity Biotechnol 35:781–788
from feces and rectal mucosa. Microb Ecol 47. Rost HL, Aebersold R, Schubert OT (2017)
61:123–133 Automated SWATH data analysis using tar-
40. Verberkmoes NC, Russell AL, Shah M et al geted extraction of ion chromatograms. Meth-
(2009) Shotgun metaproteomics of the ods Mol Biol 1550:289–307
human distal gut microbiota. ISME J 48. Alispahic M, Hummel K, Jandreski-Cvetkovic
3:179–189 D et al (2010) Species-specific identification
41. Zoetendal EG, von Wright A, Vilpponen- and differentiation of Arcobacter, Helicobac-
Salmela T et al (2002) Mucosa-associated bac- ter, and Campylobacter by full-spectral
teria in the human gastrointestinal tract are matrix-associated laser desorption/ionization
uniformly distributed along the colon and dif- time of flight mass spectrometry analysis. J
fer from the community recovered from feces. Med Microbiol 59:295–301
Appl Environ Microbiol 68:3401–3407 49. Cheng K, Ning Z, Zhang X et al (2017) Meta-
42. Li X, LeBlanc J, Truong A et al (2011) A Lab: an automated pipeline for metaproteomic
metaproteomic approach to study human- data analysis. Microbiome 5:157
microbial ecosystems at the mucosal luminal 50. Tang H, Li S, Ye Y (2016) A graph-centric
interface. PLoS One 6:e26542 approach for metagenome-guided peptide and
43. Choo JM, Leong LEX, Rogers GB (2015) protein identification in metaproteomics. PLoS
Sample storage conditions significantly influ- Comput Biol 12:e1005224
ence faecal microbiome profiles. Sci Rep 51. Mesuere B, Devreese B, Debyser G et al (2012)
5:16350 Unipept: tryptic peptide-based biodiversity
44. Chapman JD, Goodlett DR, Masselon CD analysis of metaproteome samples. J Proteome
(2014) Mulitplexed and data-independent tan- Res 11:5773–5780
dem mass spectrometry for global proteome 52. Mesuere B, Van der Jeugt F, Willems T et al
profiling. Mass Spec Rev 33:452–470 (2018) High-throughput metaprotomics data
45. Nigjeh EN, Chen R, Allen-Tamura Y et al analysis with Unipept: a tutorial. J Proteome
(2017) Spectral library-based glycopeptide 171:11–22
analysis--detection of circulating galectin-3 53. Zhang X, Ning Z, Moore JI et al (2016) Meta-
binding protein in pancreatic cancer. Proteo- Pro-IQ: a universal metaproteomic approach
mics Clin Appl 11:1700064 to studying human and mouse gut microbiota.
Microbiome 4:31
 Chapter 9

Double One-Dimensional Electrophoresis (D1-DE) Adapted

for Immunoproteomics
Youcef Shahali, Hélène Sénéchal, and Pascal Poncet

Abstract
The classical proteomics approach for the identification of allergen candidates consists on the separation of
proteins by high-resolution two-dimensional electrophoresis (2-DE) with subsequent IgE immunoblotting
and further analysis of IgE-reactive protein spots with mass spectrometry. In this approach at least two gels
most be run. One gel is used for staining and the other is for immunoblotting by antibodies labeled with
specific immunostains. Additional functional characterizations require either protein purification or 2-DE
replicates and appear to be time- and reagent-consuming. Here we described a modified double
one-dimensional electrophoresis (D1-DE) allowing the conversion of a protein spot previously visualized
by 2-DE into an extended protein band. In D1-DE, the purity of the protein of interest is similar to 2-DE
spots, but its abundance is many times higher than what can be found in a 2-DE single spot allowing many
other functional analyses from a single D1-DE separation.

Key words Double one-dimensional electrophoresis, D1-DE, 2-DE, Allergens, Proteomics

1 Introduction

Two-DE combined with immunoblotting and mass spectrometry

analyses is now routinely used in immunoproteomics to isolate,
identify, and characterize IgE-binding proteins of complex extracts
from allergenic sources [1–4]. To date, this comprehensive
approach led to the characterization of more than 850 allergens,
improving the diagnosis and therapy of allergic diseases [1]. Despite
its numerous advantages, 2-DE presents some limitations for func-
tional studies on allergens. One of the main drawbacks of this
technique is that only a single supplementary experiment can be
performed on each single 2-DE protein spot (e.g., mass spectrom-
etry or immunoreactivity after Western blotting) [5]. Additional
functional or immunological characterizations will require either
protein purification or 2-DE replicates. Another limitation of this
technique is in the detection of low-abundance protein spots that
constitute an important part of any proteome [6, 7]. In 2-DE

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_9, © Springer Science+Business Media, LLC, part of Springer Nature 2019

133
134 Youcef Shahali et al.

immunoblotting, although disease-associated proteins often cover

a very small percentage of the whole 2-DE separation, the entire gel
must be transferred onto a membrane and individually incubated
with a sizeable volume of serum (at least 250–300 μL for a mem-
brane of 100 cm2) [5–8]. In addition, with regard to allergomic
studies, since the 2-DE has not yet been miniaturized, this method
appeared to be tedious, time- and reagent-consuming when several
allergenic sources have to be studied [9]. To overcome these draw-
backs, we propose the use of a modified double one-dimensional
electrophoresis (D1-DE) allowing the simultaneous screening of at
least 30 allergic patient sera on the same blotted protein separated
according to the isoelectric point (pI) and molecular mass (Mr),
just as a 2-DE spot [10]. It is an extension of the first D1-DE
concept originally reported by Atland et al. (1981) to study protein
genetic variants [11]. This method consists of the sequential
one-dimensional combination of two electrophoretic separations
while the migration axis remains unchanged. The main advantage
of this technique was obviously in the simultaneous separation and
high resolution of proteins of interest in dual sorting parameters.
For allergomics application, we designed and developed a D1-DE
for the conversion of a protein spot previously visualized by 2-DE
into an extended protein band. Therefore, the purity of the protein
of interest is similar to 2-DE spots, but its abundance is many times
higher than what can be found in a 2-DE single spot allowing many
other functional analyses from a unique D1-DE separation.

2 Materials

2.1 Equipment 1. Horizontal flatbed electrophoresis apparatus (e.g., Multiphor

II, GE Healthcare, Uppsala, Sweden).
2. Electrophoresis power supply apparatus of minimum 1000 V
(e.g., EPL 3501 XL, GE Healthcare).
3. Thermostatic circulator (e.g., Multitemp II, GE Healthcare).
4. Semidry blotting apparatus (e.g., Multiphor II electro-semidry
transfer apparatus, GE Healthcare).
5. Gel rehydration pool (e.g., GelPool, GE Healthcare).
6. Rehydration tray for IEF strips (e.g., rehydration trays, Serva,
Heidelberg, Germany).
7. Flat Forceps.
8. A pair of scissors.
9. Magnetic stirrer.
10. Water bath.
 Double One-Dimensional Electrophoresis (D1-DE) Adapted for Immunoproteomics 135

2.2 First 1. Polyacrylamide gel 4%T, 3%C (CleanGel, GE Healthcare BioS-

Dimension IEF ciences AB, Uppsala, Sweden).
2. Carrier ampholytes for IEF (Servalyt pH 2–11) were from
Serva, Heidelberg, Germany.
3. Cathode and anode buffers (Serva IEF buffers, Serva, Heidel-
berg, Germany).
4. Electrode GF/B glass fiber strips (Whatman).

2.3 Second 1. Chemicals composing the equilibration buffer: 114 mmol/L

Dimension Separation Tris pH 6.8 and 12% w/v sodium dodecyl sulfate (SDS) were
Using SDS-PAGE all supplied by Sigma-Aldrich (St Louis, MO, USA).
2. Polyacrylamide 8–18% gradient gel (ExcelGel gradient 8–18%)
were from GE Healthcare.
3. ExcelGel SDS buffer strips (anode and cathode; GE
Healthcare).
4. Molecular masses (Mr) standard protein mixture was from
Bio-Rad (Hercules, CA, USA).
5. Whatman N 1 and 3 paper (GE Healthcare).

3 Methods

3.1 Principles Like standard 2-DE, the D1-DE reported here consists of IEF as a
of D1-DE first dimension followed by SDS-PAGE. The difference between
the two techniques mainly resides in the migration axis of the
second separation. This means that an acidic, neutral, or basic
horizontal band from the first dimension (IEF) is transferred to
SDS-PAGE (Fig. 1). Therefore, proteins are isolated as a long
continuous band with purity comparable to 2-DE spots. In the
present protocol, the sequential one-dimensional combination of
IEF and SDS-PAGE followed by immunoblotting is described for
the IgE screening of patients sera under equal conditions. For a
better correlation between 2-DE and D1-DE patterns, the same
IEF gel could be used as the first dimension of both 2-DE and
D1-DE immunoblotting.

3.2 Preparation 1. A polyacrylamide gel 4%T, 3%C (CleanGel, GE Healthcare

of IEF BioSciences AB, Uppsala, Sweden) is first hydrated in a solu-
tion containing carrier ampholytes (CleanGel Dry
IEF-12.5  26.0 cm).
2. For this purpose, first prepare the rehydration buffer by adding
5% v/v Servalytes pH 2–11, (40% carrier w/v solution in
water) (Serva, Heidelberg, Germany) in distilled water.
3. To rehydrate a complete CleanGel, pipet 25 mL rehydration
solution into a Gelpool (tray for rehydration).
136 Youcef Shahali et al.

IEF
Mpl Sample +

Orthogonal
90°
Coaxial

2-DE D1-DE
+ -
Mr Mr

kDa - kDa -
94 94

67 67 2
1 2
3
45 45
3
30 30

20 20
4 4
14.4 14.4
+ +

- All proteins are analysed - Only pl-selected proteins are analyzed

- Low amount of protein for subsequent - Large amount of protein for functional
analysis (one spot) analyses (a 10-cm-long band)

Fig. 1 Schematic illustration of a D1-DE adapted from Shahali et al. (2012) [10]. Both 2-DE and D1-DE consist
of the sequential combination of IEF and SDS-PAGE. The difference between the two techniques resides in the
axis of migration of the second separation

4. Place the CleanGel IEF precast gel into the pool for rehydra-
tion by laying the edge of the CleanGel, with the gel surface
facing down into the rehydration buffer, slowly lower it in
order to avoid air bubbles.
5. Using forceps, lift the film at the edges up to the middle and
lower it again without catching air bubbles to distribute evenly
the liquid. Very even rehydration is also obtained by shaking
the GelPool at a slow rotation rate.
6. Meanwhile, switch on the cooling system (thermostatic circu-
lator or Peltier cooling plate) of the horizontal flatbed
apparatus.
7. Remove the gel from the Gelpool after 1 h making sure the
rehydration has been evenly achieved.
8. Remove the excess buffer from the gel surface using the edges
of a Whatman filter paper N 1, until the gel surface is
completely dry (see Note 1).
9. The rehydrated gel is now ready for IEF run.
10. Pipet a low amount of kerosene (about 2.5 mL) onto the
horizontal cooling plate (e.g., Multhiphor II apparatus) to
improve the cooling contact.
 Double One-Dimensional Electrophoresis (D1-DE) Adapted for Immunoproteomics 137

11. Place the gel (gel surface up/gel film or GelBond down) onto
the center of the cooling plate.
12. Cut the electrode strips at the gel length (26 cm).
13. Align strips on the cathodal and anodal edges of the gels, the
electrode strips overlapping the gel-edge with about 3 mm.
14. Apply evenly the cathode and anode buffers (Serva IEF buffers,
Heidelberg, Germany) on the cathode and anode strips,
respectively.
15. Cut two 10-cm-long and up to 0.7-cm-wide strips of a dry
Whatmans N 1 paper (GE Healthcare) for sample application
on anode side.
16. Cut a small piece of Whatman N 1 paper (0.5  0.5 cm) for
IEF protein standards (pI marker) and place it directly onto the
upper center of the gel (anode side).
17. Use forceps to align the strips on both sides of the pI marker at
the same distance from the paper piece (minimum 1 cm).
18. Pipet 2 μL of pI marker (Bio-Rad) for wide or short range IEF
as comparative references.
19. Load samples (55–60 μg of proteins from the allergen extract)
using a micropipette on each 10-cm-long sample paper piece.
20. Clean platinum electrode wires with a wet tissue paper.
21. Then, move electrodes on anode/cathode strips to ensure a
complete overlapping between buffer strips, gel, and electrode
wires.
22. For Multiphor, connect the cables of the electrodes to the
electrophoresis power supply apparatus (EPS 3501 XL, GE
Healthcare). Finally lower and close the safety lid.
23. Select the running program and start the IEF migration (see
Note 2).
24. Stop the migration and wash the cathode and anode borders of
the gel with PBS pH 7.5.
25. Lift the film at the edges up to the middle to form a U shape
and pour promptly PBS on the middle of the gel.

3.3 Preparation 1. After IEF running, cut out the gel at the pI marker along with a
of D1-DE small part of the sample separation for Coomassie blue or silver
staining.
2. Lay the gel face down on a clean plastic film or a glass plate.
Mark the area that should be cut out for transferring. The size
of this area should be 10 cm long and up to 7 mm wide.
3. Cut out horizontally the IEF strip in the selected narrow pI
range containing the allergenic fraction of interest by reference
to pI markers and sample stained.
138 Youcef Shahali et al.

4. At this step, the IEF strips could be transferred to the

SDS-PAGE separation or placed in a sealed bag until use,
conserve the gel at 20  C until use (up to 1 years).
5. Otherwise, incubate the IEF strips in the equilibration buffer
containing 114 mmol/L Tris pH 6.8 and 12% w/v SDS for
3  10 min.
6. Meanwhile, switch on the thermostatic circulator adjusted to
12  C.
7. Pipet a low amount of kerosene (about 2.5 mL) onto the
horizontal cooling plate (e.g., flatbed electrophoretic chamber,
Multiphor II, GE Healthcare) to improve the cooling contact.
8. During the equilibration procedure, place the thin 8–18% gra-
dient polyacrylamide gel (ExcelGel; GE Healthcare) surface
up/gel film down onto the center of the cooling plate.
9. Align the ExcelGel SDS buffer strips (anode and cathode; GE
Healthcare) on the cathodal and anodal edges of the gels, the
electrode strips overlapping the gel-edge with at least 5 mm.
10. Cut a small piece of Whatman N 1 paper (0.5  0.5 cm) for
molecular masses protein standards (Mr) and place it directly
onto the upper center of the gel (anode side).
11. Remove the 10-cm IEF strips from the equilibration trays.
12. Remove the excess buffer using the edges of a Whatman filter.
13. Using two forceps, place each IEF strip gel-side down into the
top of the SDS ExcelGel and push them carefully toward the
anode.
14. Pipet 2–5 μL of Mr (Bio-Rad) onto the small piece of Whatman
paper.
15. Clean platinum electrode wires with a wet tissue paper.
16. Then, move electrodes on anode/cathode strips to ensure a
complete overlapping between buffer strips, gel, and electrode
wires.
17. For Multiphor, connect the cables of the electrodes to the basic
electrophoresis power supply apparatus. Finally lower and close
the safety lid.
18. Select the running program and start the SDS-PAGE migra-
tion (see Note 3).
19. Stop the migration and wash carefully the entire SDS gel with
PBS pH 7.5. At this step, D1-DE separations could either be
silver-stained according to Blum et al. [12], Coomassie blue-
stained or blotted onto a cyanogen bromide-activated nitrocel-
lulose (Fig. 2) (Optitrans BA-S 83, Schleicher and Schuell,
Dassel, Germany) sheet [13] for further functional analyses
(see Note 4) [10, 14].
 Double One-Dimensional Electrophoresis (D1-DE) Adapted for Immunoproteomics 139

Fig. 2 D1-DE combined with IgE immunoblots (adapted from Shahali et al. 2012) [10]. (a) Represents the initial
IEF separation. The silver staining was performed after excision of the basic and neutral bands. IgE
immunoblots of the basic (b) and neutral (c) proteins of cypress pollen (CP) extracts probed with sera of
30 CP allergic patients: lanes 1–30; lane 31: healthy donor serum; lane 32: no serum (negative control). D1-DE
allowed the MS/MS characterization of an allergenic polygalacturonase (PG) of 43 kDa which overlapped
(in previous 1-DE experiments) with another CP major allergen belonging to the pectate lyase (PL) family
referenced as Cup s 1 (see Note 6)

20. After staining, the protein band of interest could be cut out,
digested, and submitted to mass spectrometry analyses (see
Note 5) [10, 14].

4 Notes

1. Remove excess buffer from the gel surface with a piece of

Whatman N 1 paper until you can hear a “squeaking.”
2. Running conditions and parameters for IEF experiment:
Running temperature: 12  C; total running time: 2 h 30 min.
Phase 1: Set the voltage as constant at 50 V during 60 min.
Remove the strips at the end of this phase.
140 Youcef Shahali et al.

Phase 2: Set the voltage as constant at 200 V during 60 min.

Phase 3: Set the voltage as constant at 150 V during 40 min.
Phase 4: Set the watt as constant at 1 W during 90 min.
Phase 5: Set the watt as constant at 2 W during 50 min.
Phase 6: Set the watt as constant at 3 W during 120 min.
3. Stop the IEF when cytochrome C (red color) marker is near
cathode side
Running conditions and parameters for SDS-PAGE. Under-
lined values are constant.

Time Voltage Current Power Temperature

(min) (V) (mA) (W) ( C)
75 100 40 40 12
110 750 20 15 12
120 900 25 15 12

4. Up to 40 blotted strips could be tested with various antibodies,

with or without inhibitors, which gave rise to fundamental
results concerning the specificity of the allergen recognition
among cypress pollen allergic patients [14].
5. Besides multiplexing for immunoblotting, the D1-DE circum-
vents the problem of low protein amounts often encountered
in a single 2-DE spot, which generally makes MS and micro-
sequencing experiments difficult. The whole protein band can
be excised, digested, and processed for analysis. A novel
low-abundant allergen has been recently identified and charac-
terized in cypress pollen using this approach [15].
6. D1-DE IgE screenings on the neutral and basic IEF fractions of
CP extracts demonstrate that among 30 tested patients,
21 (70%) showed a positive IgE response to the novel 43 kDa
basic allergen (PG), while 22 (~73%) were sensitized to Cup
s 1 (PL). This novel CP major allergen has been recently
indexed as Cup s 2 by the WHO/IUIS allergen
nomenclature [14].

References

1. Nony E, Le Mignon M, Brier S, Martelet A, 3. Hoffmann-Sommergruber K (2016) Proteo-

Moingeon P (2016) Proteomics for allergy: mics and its impact on food allergy diagnosis.
from proteins to the patients. Curr Allergy EuPA Open Proteom 12:10–12
Asthma Rep 16:64 4. Tiotiu A, Brazdova A, Longé C, Gallet P,
2. Mousavi F, Majd A, Shahali Y, Morisset M, Leduc V et al (2016) Urtica dioica
Ghahremaninejad F, Shoormasti RS, Pourpak pollen allergy: clinical, biological, and allergo-
Z (2017) Immunoproteomics of tree of heaven mics analysis. Ann Allergy Asthma Immunol
(Ailanthus atltissima) pollen allergens. J Pro- 117:527–534
teome 154:94–101
 Double One-Dimensional Electrophoresis (D1-DE) Adapted for Immunoproteomics 141

5. Poncet P, Sénéchal H, Clement G, Purohit A, of cypress pollen allergens using double and
Sutra JP, Desvaux FX et al (2010) Evaluation of triple one-dimensional electrophoresis.
ash pollen sensitization pattern using proteo- Electrophoresis 33:462–469
mic approach with individual sera from allergic 11. Altland K, Silke R, Hackler R (1981) Demon-
patients. Allergy 65:571–580 station of human prealbumin by double
6. D’Amato A, Bachi A, Fasoli E, Boschetti E, one-dimensional slab gel electrophoresis. Elec-
Peltre G, Sénéchal H et al (2010) In-depth trophoresis 2:148–155
exploration of Hevea brasiliensis latex prote- 12. Blum H, Beier H, Gross HJ (1987) Improved
ome and “hidden allergens” via combinatorial silver staining of plant proteins, RNA and DNA
peptide ligand libraries. J Proteome in polyacrylamide gels. Electrophoresis
73:1368–1380 8:93–99
7. Shahali Y, Sutra JP, Fasoli E, D’Amato A, Righ- 13. Demeulemester C, Peltre G, Laurent M,
etti PG, Futamura N et al (2012) Allergomic Panheleux D, David B (1987) Cyanogen
study of cypress pollen via combinatorial pep- bromide-activated nitrocellulose membranes:
tide ligand libraries. J Proteome 77:101–110 a new tool for immunoprint techniques. Elec-
8. Shahali Y, Sutra JP, Peltre G, Charpin D, trophoresis 8:71–73
Sénéchal H, Poncet P (2010) IgE reactivity to 14. Shahali Y, Sutra JP, Hilger C, Swiontek K,
common cypress (C. Sempervirens) pollen Haddad I, Vinh J et al (2017) Identification
extracts: evidence for novel allergens. W Allergy of a polygalacturonase (Cup s 2) as the major
Organ J 3:229–234 CCD-bearing allergen in Cupressus sempervi-
9. Shahali Y, Nicaise P, Brazdova A, Charpin D, rens pollen. Allergy 72:1806–1810
Scala E, Mari A et al (2014) Complementarity 15. Sénéchal H, Šantrůček J, Melčová M,
between microarray and immunoblot for the Svoboda P, Zı́dková J, Charpin D et al (2018)
comparative evaluation of IgE repertoire of A new allergen family involved in pollen food-
French and Italian cypress pollen allergic associated syndrome: Snakin/gibberellin-
patients. Folia Biol 60:192 regulated proteins. J Allergy Clin Immunol
10. Shahali Y, Sutra JP, Haddad I, Vinh J, 141:411–414
Guilloux L, Peltre G et al (2012) Proteomics
 Chapter 10

BioID: A Proximity-Dependent Labeling Approach

in Proteomics Study
Peipei Li, Yuan Meng, Li Wang, and Li-jun Di

Abstract
Biological activities are mainly executed by proteins and in most of the occasions these activities are
accomplished by protein complexes or through protein–protein interactions (PPI). So it is critical to reveal
how the protein complexes are organized and demonstrate the PPIs involved in the biological processes. In
addition to the traditional biochemical approaches, proximity-dependent labeling (PDL) has recently been
proposed to identify the interacting partners of a given protein. PDL requires the fusion expression of the
target protein with an enzyme which catalyzes the attachment of a reactive molecule to the interacting
partners in a distance-dependent manner. Further analysis of all the proteins that are modified by the
reactive molecule discloses the identity of these proteins which are presumed to be interacting partners of
the target protein. BioID is one of those representative PDL methods with the most widely applications.
The enzyme used in BioID is a biotin ligase BirA which catalyzes the biotinylation of target protein with the
presence of biotin. Through streptavidin-mediated pull-down and mass spectrometry analysis, the inter-
acting protein candidates of a given protein can be obtained.

Key words BioID, Protein–protein interactions, Proximity-dependent labeling

1 Introduction

Either stably or transiently, the proteins need to form complexes via

PPIs to perform important biological functions [1, 2]. So the PPIs
are the most fundamental biological activity that are critical to
accomplish the biological processes. Thus the depicting of protein
complex relies on the recognition of PPIs. Traditionally, the identi-
fication of PPI relies on the biochemical approaches, i.e., the inter-
acting proteins can be captured by immunoprecipitating the target
protein [3]. The co-immunoprecipitation requires the lysis of cells
and known as in vitro technology. Some technologies are able to
capturing PPIs in vivo such as two/three yeast hybridization system
[4]. Comparing to the technologies in vitro, the in vivo determina-
tion of PPIs in live cell is preferred because it discloses the truly
interacting proteins at the physiological condition. However, the

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_10, © Springer Science+Business Media, LLC, part of Springer Nature 2019

143
144 Peipei Li et al.

specificity of the antibody used in these in vivo methods has always a

big problem.
Fluorescence resonance emission transfer (FRET) technology
has been applied to directly demonstrate the PPI in live cells. This
technology is based on the fact that when the suspected interacting
pair of proteins are both fused to different fluorescent proteins, the
fused fluorophore will be close enough to transfer energy from the
fluorescent donor to the fluorescent receptor that will alternate the
wavelength of the emitted fluorescent signal of the energy receptor.
Through analyzing the change of the fluorescent signal emitted by
the energy receptor, the PPI can be quantitated. Another technol-
ogy is known as bimolecular complementation technology. The
proteins that are expected to interact are fused to complementary
fragments of a fluorescent protein and the PPI will bring the
complementary fragments in proximity and the fluorescent signal
recovers [5, 6]. However, both of these methods can only validate
the suspected interacting pairs of proteins but not discover
novel ones.
The PDL methods are recently proposed and provide much
more flexibility and reliability in identifying novel PPIs in vivo. The
advantage of PDL methods is that both the direct interacting
proteins, but also the proteins in close proximity can be identified.
Roughly, the frequency or interacting affinity of PPIs can be quan-
titated through analyzing the hit rate of the peptide in mass analy-
sis. The development of PDL methods relies on the enzymes that
are capable of modifying the nearby proteins in vivo by catalyzing
the attachment of reactive groups to the target amino acids of any
proteins in proximity. Several PDL approaches have been reported
such as peroxidase-based techniques including selective proteomic
proximity labeling assay using tyramide (SPPLAT), enzyme-
mediated activation of radical sources (EMARS), and Ascorbate
Peroxidase (APEX) [7–9]. Among these methods, BioID is the
earliest and most widely applied PDL method [10].
The prototype of BioID is based on the discovery that the
biotin ligase BirA, isolated from Escherichia coli, can biotinylate
the target proteins in mammalian cells. The limitation of the pro-
totype of BioID is that the target protein has to have a consensus
sequence which is recognized by BirA [11, 12]. The reason is that
two continuous steps are required in BirA catalyzed biotinylation.
Firstly, BirA catalyzes the formation of the reactive molecule 5-
0
-AMP-Biotin; this molecule cannot be released but stay at the
reaction center. At the second step, the ε-amino groups of lysine
residues of a target protein binds BirA reactive center, attacks the
anhydride of the 50 -AMP-biotin to generate an amide bond and the
biotin is transferred to the target protein [13, 14]. By mutating one
amino acid (R118G; also known as BirA*), Roux et al. demon-
strated that BirA* can be engineered to release 50 -AMP-biotin and
biotinylate almost any proteins even when the required consensus
 BioID: A Proximity-Dependent Labeling Approach in Proteomics Study 145

sequence is absent. The upgraded technology is now known as

BioID [15]. Since the instability of the reactive molecule, only the
proteins that are close to the BirA reaction center have the chance
to be modified. The biotinylated proteins can be captured by
streptavidin-coated agarose beads and further analyzed by untar-
geted mass spectrometry or other targeted technologies [16].
BioID has been successfully applied in determining the com-
position of insoluble protein complexes including lamina and cen-
trosomes [17, 18]; the components of cytoplasm membrane-
bound protein complexes such as the tight junctions [19]; the
Hippo pathway PPIs [20]; and the composition of protein com-
plexes from infectious pathogens such as Toxoplasma gondii, HIV,
and EBV [21–23]. Since BioID is an extremely sensitive labeling
method in vivo, the nonspecific modifying of proteins is inevitable.
To overcome this problem, some studies tried multi-bait in apply-
ing BioID and the real PPIs can be identified more efficiently [24].
In this manual, we introduce an application example of BioID
technology and the detailed protocol is also described.

2 Materials

NOTE: All the reagents should be analytical grade; preparing the

solutions by using 18 MΩ-cm of ultrapure water; all the reagents
and solutions should be stored at indicated temperature. Only the
specially required reagents are listed here and other regular
required reagents are omitted.

2.1 BioID Vector 1. Expression vectors for BirA* can be obtained from Addgene
(pcDNA3.1 MCS-BirA(R118G)-HA #36047 or pcDNA3.1
mycBioID #35700).
2. Cloned coding sequence of the interested protein.

2.2 Validation of 1. 293T or other appropriate cell lines and the appropriate cell
BioID Fusion Protein culture medium.
2. 1 mM Biotin: 12.2 mg biotin (Sigma) dissolved in 50 mL
H2O, sterilize by passing through a 0.22-μm syringe-driven
filter, store at 4  C.
3. HRP-streptavidin (Sigma).
4. Primary antibodies for BioID fusion protein (e.g., anti-HA/
MYC, anti-interest protein).
5. Streptavidin-Alexa Fluor.
6. DNA labeling reagent (e.g., DAPI).
146 Peipei Li et al.

2.3 BioID Pull-Down 1. Streptavidin conjugated beads (MyOne Streptavidin C1,

Invitrogen).
2. Wash buffer 1: 2% SDS in H2O, store at room temperature.
3. Wash buffer 2: 0.1% deoxycholate, 1% Triton X-100, 500 mM
NaCl, 1 mM EDTA, and 50 mM Hepes, pH 7.5, store at room
temperature.
4. Wash buffer 3: 250 mM LiCl, 0.5% NP-40, 0.5% deoxycholate,
1 mM EDTA, and 10 mM Tris, pH 8.0, store at room
temperature.
5. Wash buffer 4: 50 mM Tris, pH 7.4, store at room
temperature.
6. ACN buffer: 50 mM ammonium bicarbonate, store at room
temperature.

3 Methods

NOTE: Conducting all procedures at room temperature unless

otherwise specified.

3.1 Generation of In order to study the interacting proteins of a give protein, this
BioID Construct protein needs to be expressed in fusion with BirA*. The expression
vectors of BirA* have provided multiple cloning site and the target
protein coding sequence can be cloned into this site. The gene
clone strategy is not given in this protocol. The fusion protein
(now named as BioID vector) shouldn’t affect the function and
localization of the original protein, gene fragment inserted into
either the N- or C-terminus of biotin ligase should be carefully
considered. Additionally, the fusion protein may be observed to
ensure the proper intracellular localization and showing the
expected function.
NOTE: The BioID vector should contain HA tag or Myc tag
for easy detection of the fusion protein. Also, it is important to
select N-terminal or C-terminal for insertion of the target protein
because the fusion protein may be influenced functionally if the
insertion site hasn’t been carefully tested.

3.2 Validation of 1. Preparing BioID vector by using QIAGEN Plasmid

BioID Fusion Protein preparation kits.
2. Preparing a six-well plate cell culture (and/or an eight-well
chamber cell culture) ready for transfection by the following
vectors: empty vector with or without biotin, BioID vector
with or without biotin.
BioID: A Proximity-Dependent Labeling Approach in Proteomics Study 147

NOTE: The empty vector transfection is important negative

control for validation of BioID fusion protein and subsequent
LC-mass analysis.
3. Transfecting the cells with the above conditions using Lipofec-
tamine 3000, add 50 μM biotin into the culture medium 3 h
post-transfection.
NOTE: The selection of transfection reagent should depend on
cell lines. Excess biotin enhances protein biotinylation, but the
concentration of 50uM should be sufficient for experiment.
4. After transfection for 24 h, the six-well plate cultured cells are
ready for immunoblot assay and the eight-well chamber
cultured cells are ready for immunofluorescence assay.
5. For immunoblot analysis, wash the cells with PBS prior to lysis,
apply 100 μL cell lysis buffer into the cells on ice for 30 min
with the presence of protease inhibitors, sonicate the sample,
and centrifuge at 15,000  g for 10 min at 4  C. If the target
protein is nuclear protein, nuclear extract is recommended.
6. Adding 5 SDS-PAGE sample loading buffer and heat to
98  C for 5 min to denature proteins.
7. Performing SDS-PAGE electrophoresis and protein transfer,
incubate membrane in 1% BSA blocking buffer for 30 min at
room temperature.
8. Incubating membrane in primary antibodies for BioID fusion
protein (e.g., anti-HA/MYC antibodies or anti-target protein
antibody) for 1 h at room temperature.
9. Washing the membrane five times with PBST, 5 min each time.
10. Incubating membrane in secondary antibodies to detect pri-
mary antibody for 1 h at room temperature.
11. Washing the membrane five times with PBST, 5 min each time.
12. Taking images of the fusion protein by ECL. A representative
western blotting of CtBP2-BirA* is provided as Fig. 1a.
13. Stripping the membrane in stripping buffer for 20 min, and
rinsing membrane with PBST several times to remove the
striping buffer.
14. Blocking the cells with 1% BSA for 0.5 h, followed by incuba-
tion of streptavidin-HRP (1:20,000) for 1 h at room
temperature.
15. Washing the membrane five times with PBST, 5 min each time.
16. Observing the biotinylated proteins using ECL. The represen-
tative western blotting is provided as Fig. 1b. Be noted that the
appearance of the spear indicates the BioID pull-down is
successful.
148 Peipei Li et al.

Fig. 1 Detection of biotinylated proteins by BioID in HEK293 cells expressing CtBP2-BirA*. (a) Cells expressing
CtBP2-BirA* and control cells were incubated for 24 h with and without 50 μM biotin, the expression of CtBP2-
BirA* was detected by anti-HA, and β-actin served as loading control. (b) Biotinylated proteins were identified
by HRP-streptavidin in different conditions. (c) Fluorescence microscopy was used to observe the biotinylated
proteins in CtBP2-BirA* overexpressed cells incubating for 24 h with 50 μM biotin, biotinylated proteins were
detected by streptavidin-Alexa Fluor 594 (red), DNA was detected with DAPI (blue)

17. For immunofluorescence assay, fix the cell with 4% PFA for
10 min at room temperature, then permeate the cells with 0.2%
TritonX-100 for 10 min.
18. Blocking the cells with 1% BSA for 0.5 h, followed by incuba-
tion of streptavidin-Alexa Fluor (1:1000) and DAPI, observe
the fluorescence by microscopy (Fig. 1c).
NOTE: BSA has better effect in getting rid of free biotin than
milk.

3.3 BioID Pull-Down This protocol describes the cells transiently expressing BioID fusion
Assay protein to process large-scale (6  107 cells) BioID pull-down assay
and LC-mass (Liquid chromatograph mass spectrometry) analysis.
Cells stably expressing BioID fusion protein are also subjected to
BioID pull-down as well.
1. Plating two 15-cm dishes for each experimental condition (cells
expressing BioID construct and control cells).
BioID: A Proximity-Dependent Labeling Approach in Proteomics Study 149

2. Transfecting empty vector or BioID vector into the cells using

Lipofectamine or other appropriate transfection reagents and
supplement 50 μM biotin into the culture medium 3 h post-
transfection.
3. Incubating cells for 24 h.
NOTE: 24 h is enough for protein biotinylation, and extend-
ing incubation time will reduce the amount of biotinylated
protein.
4. Washing the cells twice with PBS to remove the free biotin.
5. Adding 1.2 mL cell lysis buffer per dish into the cells, scrape
and collect the cells.
6. Placing the tube on ice for 30 min, sonicate to shear DNA and
centrifuge the samples at 15,000  g for 10 min at 4  C.
NOTE: Mix the sample well by vortex, and pipet the lysates up
and down every 10 min during the incubation. Sonication is to
break down the DNA and solubilize the proteins.
7. The supernatant of cell lysates are gently transferred to 2-mL
tubes and are diluted to 2.5-fold with prechilled 50 mM
Tris·Cl, pH 7.4. Subsequently the lysates are aliquoted to
1.5 mL per tube.
8. The magnetic streptavidin beads need to be equilibrium in 1:1
lysis buffer and 50 mM Tris·Cl, pH 7.4. Using the magnetic
separation stand to collect the magnetic beads and remove the
buffer after equilibrium.
9. The supernatant in step 7 are transferred to the tube in step 8,
mix the samples and beads gently, and incubate the mixture on
the rotator over night at 4  C.
10. The tubes are placed on magnetic separation stand for 3 min
until the beads accumulate at one side of tube, remove the
supernatant by pipet. Avoid disturbing the beads on the
tube wall.
11. Washing the magnetic beads once sequentially by Wash Buffer
1 to 3 once and twice by Wash Buffer 4.
12. Resuspending the beads with 200 μL of 50 mM Tris·Cl,
pH 7.4.
13. 10% of the sample is reserved for SDS-PAGE, other 90% sam-
ple is washed twice by 200 μL of 50 mM ammonium bicarbon-
ate. Sample volume is adjusted to 50 μL by 50 mM ammonium
bicarbonate for LC-mass preparation or storage at 80  C.
NOTE: The protein identification by LC-mass can be per-
formed by specialist or commercial service providers and the details
are omitted here.
150 Peipei Li et al.

4 Summary

We have successfully applied BioID to identify the interacting

proteins of the nuclear transcriptional factors for several cases. Of
noting is that BioID is extremely sensitive and produces unneglect-
able background. So for each experiment, several experimental
repeats are strongly recommended and the empty vector control
should also be included for the LC-mass analysis. To our experi-
ence, only the proteins that are repeatedly identified throughout
the experiment repeats but not in the negative control groups are
potential candidates, given their score in LC-mass analysis is among
the top.
We also notice that after BioID pull-down, the western blotting
is sensitive enough to validate the suspected PPIs. So BioID can be
a reliable tool to validate the PPIs besides being an explorative tool
to identify novel PPIs.

Acknowledgments

This work is supported by the Science and Technology Develop-

ment Fund (FDCT) of Macao SAR (FDCT/0014/2018/A1), the
Multi-Year Research Grant from the University of Macau
(MYRG2018-00158-FHS), and the National Natural Science
Foundation of China (NSFC 81772980) to LD. This work is also
supported by the Multi-Year Research Grant from the University of
Macau to LW (MYRG2016-00251-FHS).

References

1. Nooren IM, Thornton JM (2003) Diversity of Rev 76(2):331–382. https://doi.org/10.

protein-protein interactions. EMBO J 22 1128/MMBR.05021-11
(14):3486–3492 5. Kenworthy AK (2001) Imaging protein-
2. Ngounou Wetie AG, Sokolowska I, Woods protein interactions using fluorescence reso-
AG, Roy U, Loo JA, Darie CC (2013) Investi- nance energy transfer microscopy. Methods
gation of stable and transient protein-protein 24(3):289–296. https://doi.org/10.1006/
interactions: past, present, and future. Proteo- meth.2001.1189
mics 13(3–4):538–557. https://doi.org/10. 6. Sjohamn J, Bath P, Neutze R, Hedfalk K
1002/pmic.201200328 (2016) Applying bimolecular fluorescence
3. Vermeulen M, Hubner NC, Mann M (2008) complementation to screen and purify aqua-
High confidence determination of specific porin protein:protein complexes. Protein Sci
protein-protein interactions using quantitative 25(12):2196–2208. https://doi.org/10.
mass spectrometry. Curr Opin Biotechnol 19 1002/pro.3046
(4):331–337. https://doi.org/10.1016/j. 7. Li XW, Funk PE, Rees JS, Farndale RW, Xue P,
copbio.2008.06.001 Lilley KS et al (2014) New insights into the
4. Stynen B, Tournu H, Tavernier J, Van Dijck P DT40 B cell receptor cluster using a proteomic
(2012) Diversity in genetic in vivo methods for proximity labeling assay. J Biol Chem 289
protein-protein interaction studies: from the (6):14434–14447. https://doi.org/10.1074/
yeast two-hybrid system to the mammalian jbc.M113.529578
split-luciferase system. Microbiol Mol Biol
 BioID: A Proximity-Dependent Labeling Approach in Proteomics Study 151

8. Miyagawa-Yamaguchi A, Kotani N, Honke K form distinct filamentous networks with differ-

(2014) Expressed glycosylphosphatidylinositol- ential nuclear pore complex associations. Curr
anchored horseradish peroxidase identifies Biol 26(19):2651–2658. https://doi.org/10.
co-clustering molecules in individual lipid raft 1016/j.cub.2016.07.049
domains. PLoS One 9(3):e93054. https://doi. 18. Firat-Karalar EN, Stearns T (2015) Probing
org/10.1371/journal.pone.0093054.g001 mammalian centrosome structure using
9. Rhee HW, Zou P, Udeshi ND, Martell JD, BioID proximity-dependent biotinylation.
Mootha VK, Carr SA et al (2013) Proteomic Methods Cell Biol 129:153–170. https://doi.
mapping of mitochondria in living cells via spa- org/10.1016/bs.mcb.2015.03.016
tially restricted enzymatic tagging. Science 19. Van Itallie CM, Aponte A, Tietgens AJ,
339:1328–1331 Gucek M, Fredriksson K, Anderson JM
10. Li P, Li J, Wang L, Di LJ (2017) Proximity (2013) The N and C termini of ZO-1 are
labeling of interacting proteins: application of surrounded by distinct proteins and functional
BioID as a discovery tool. Proteomics 17(20). protein networks. J Biol Chem 288
https://doi.org/10.1002/pmic.201700002 (19):13775–13788. https://doi.org/10.
11. Parrott MB, Barry MA (2000) Metabolic bio- 1074/jbc.M113.466193
tinylation of recombinant proteins in mamma- 20. Couzens AL, Knight JD, Kean MJ, Teo G
lian cells and in mice. Mol Ther 1(1):96–104. (2013) Protein interaction network of the
https://doi.org/10.1006/mthe.1999.0011 mammalian Hippo pathway reveals mechan-
12. Parrott MB, Barry MA (2001) Metabolic bio- isms of kinase-phosphatase interactions. Sci
tinylation of secreted and cell surface proteins Signal 6:rs15
from mammalian cells. Biochem Biophys Res 21. Nourani E, Khunjush F, Durmus S (2015)
Commun 281(4):993–1000. https://doi.org/ Computational approaches for prediction of
10.1006/bbrc.2001.4437 pathogen-host protein-protein interactions.
13. Chapman-Smith A, Morris TW, Wallace JC, Front Microbiol 6:94. https://doi.org/10.
Cronan JE (1999) Molecular recognition in a 3389/fmicb.2015.00094
post-translational modification of exceptional 22. Le Sage V, Cinti A, Valiente-Echeverria F,
specificity. J Biol Chem 274(3):1449–1457 Mouland AJ (2015) Proteomic analysis of
14. Prakash O, Eisenberg MA (1979) Biotinyl 5- HIV-1 Gag interacting partners using
0 proximity-dependent biotinylation. Virol J
-adenylate corepressor role in the regulation of
the biotin genes of Escherichia coli K-12. Proc 12:138. https://doi.org/10.1186/s12985-
Natl Acad Sci U S A 76:5592–5595 015-0365-6
15. Roux KJ, Kim DI, Raida M, Burke B (2012) A 23. Holthusen K, Talaty P, Everly DN Jr (2015)
promiscuous biotin ligase fusion protein iden- Regulation of latent membrane protein 1 sig-
tifies proximal and interacting proteins in naling through interaction with cytoskeletal
mammalian cells. J Cell Biol 196(6):801–810. proteins. J Virol 89(14):7277–7290. https://
https://doi.org/10.1083/jcb.201112098 doi.org/10.1128/JVI.00321-15
16. Kuroishi T, Rios-Avila L, Pestinger V, Wijeratne 24. Gupta GD, Coyaud É, Gonçalves J, Mojarad
SS, Zempleni J (2011) Biotinylation is a natural, BA, Liu Y, Wu Q et al (2016) A dynamic pro-
albeit rare, modification of human histones. Mol tein interaction landscape of the human
Genet Metab 104(4):537–545. https://doi. centrosome-cilium interface. Cell 163
org/10.1016/j.ymgme.2011.08.030 (6):1484–1499. https://doi.org/10.1016/j.
17. Xie W, Chojnowski A, Boudier T, Lim JS, cell.2015.10.065
Ahmed S, Ser Z et al (2016) A-type lamins
 Chapter 11

Functional Application of Snake Venom Proteomics

in In Vivo Antivenom Assessment
Choo Hock Tan and Kae Yi Tan

Abstract
Reverse-phase high-performance liquid chromatography is commonly employed as a decomplexing strat-
egy in snake venom proteomics. The chromatographic fractions often contain relatively pure toxins that can
be assessed functionally for toxicity level through the determination of their median lethal doses (LD50).
Further, antivenom efficacy can be evaluated specifically against these venom fractions to understand the
limitation of the antivenom as the treatment for snake envenomation. However, methods of toxicity
assessment and antivenom evaluation vary across laboratories; hence there is a need to standardize the
protocols and parameters, in particular those related to the neutralizing efficacy of antivenom. This chapter
outlines the important in vivo techniques and data interpretation that can be applied in the functional study
of snake venom proteomes.

Key words Immuno-neutralization, Toxin-specific neutralization, Median lethal dose (LD50),

Median effective dose (ED50), Potency (P)

1 Introduction

The complexity of snake venom can be greatly resolved through

various proteomic techniques (for instance: decomplexing proteo-
mics, Chapter 5) [1]. The identification and quantitative estimation
of proteins in a venom has become relatively more achievable and
time-saving [2, 3]. In decomplexing venom proteomics, high-
resolutional reverse-phase high-performance liquid chromatogra-
phy separates venom into pure/partially pure fractions that can be
subjected to mass spectrometric analysis [4, 5]. In addition, the
HPLC fractionation allows functional assessment of the venom
components from a bottom-up approach, where the toxicity of
the individual venom fractions and neutralization by antivenom
can be determined [4, 6]. The approach for the functional correla-
tion of venom proteome typically involves laboratory animal mod-
els to elucidate the in vivo pathophysiology of snake envenomation,

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_11, © Springer Science+Business Media, LLC, part of Springer Nature 2019

153
154 Choo Hock Tan and Kae Yi Tan

and is essential for the robust preclinical evaluation of antivenom

efficacy in neutralizing the venom toxicity [7].
To interpret the systemic toxicity of a snake venom, the venom
fractions are usually administered through intravenous route to
ensure a full systemic access of the venom components into the
animal. The use of in vivo model is favored as it involves the whole
biological system in which the snake venom can act. Using the
in vivo model, the evolution of clinical syndrome upon envenom-
ation can also be closely monitored, for instance the development
of neuromuscular paralysis induced by the venom [8]. Through the
animal model, the lethal effect of the various venom fractions can
be determined. By correlating the data with the venom proteome, it
is possible to determine the principal toxins responsible for venom-
induced lethality. Of note, the lethality parameter is essential in
antivenom studies as the neutralization of lethality has been
regarded as the gold standard for antivenom assessment by WHO
[9, 10]. Hence, functional venom proteomics can provide valuable
information on the strength and weaknesses of an antivenom in
neutralizing a venom as well as its principal toxins [11, 12]. How-
ever, the determination of the efficacy parameters of an antivenom
often varies among different laboratories, and the interpretation
(as well as comparison) of the data across different studies has been
challenging. This chapter illustrates the common protocol that can
be adopted to study the lethality of venom and toxin components,
and recommends the standardization of parameters that are essen-
tial to gauge the efficacy or potency of antivenom.

2 Materials

2.1 Snake Venom/ Lyophilize the snake venom or its toxin fractions (obtained from
Toxin Fractions protein decomplexation approach, Chapter 5). Store at
20  C
until use.

2.2 Snake Lyophilized antivenom: Reconstitute the lyophilized antivenom

Antivenoms according to manufacturer’s instruction. Aliquot and store at

20  C until use.
Liquid form antivenom: Dilute the liquid antivenom accord-
ingly for the neutralization study. Store at 2–8  C until use.

2.3 Chemical and Reconstitution buffer for antivenom: normal saline.

Solution Protein concentration estimation: Bicinchoninic acid (BCA)
assay/Lowry assay.
 Functional Application of Snake Venom Proteomics in In Vivo Antivenom Assessment 155

3 Methods

3.1 Lethality 1. Divide ICR mice into four groups (n ¼ 5 per group, 20–25 g
Determination of body weight). Each group of mice receives venom or toxin
Venom/Toxin Fraction fraction at different concentrations (treated as “doses,” see
below). Estimate the median lethal dose using at least four
doses of the venom or individual toxin fraction (see Note 1).
2. Sample preparation: Estimate the protein concentration of
venom/toxin fraction using bicinchoninic acid assay or Lowry
assay. Dilute the venom/toxin fraction with normal saline to
serial concentrations. Fix the volume of injection at 100 μL per
mouse.
3. Hold the mice using a rodent restrainer. Inject the appropri-
ately diluted venom/toxin fraction intravenously into the mice
via caudal vein (see Note 2).
4. Allow the mice access to food and water ad libitum. Monitor
and record the survival ratio of the mice in each group for 48 h
(see Note 3).
5. Determine the median lethal dose (LD50) of the venom/toxin
fraction using Probit analysis (see Note 4).
*Figure 1 shows the schematic drawing of the assessment of
median lethal dose (LD50) of venom/toxin fraction.

3.2 Lethality 1. Divide ICR mice into four groups (n ¼ 5 per group, 20–25 g
Neutralization of body weight). Each group of mice receives venom or toxin
Venom/Toxin Fraction fraction that has been preincubated with varying doses of anti-
venom (see below). Estimate the antivenom efficacy and

Fig. 1 A schematic diagram shows the assessment of median lethal dose (LD50) of venom or toxin fraction
(flow following red arrows) and neutralization efficacy as well as potency of antivenom (flow following green
arrows). The results are determined using Probit analysis based on the recorded survival ratio throughout the
experiment. The test samples (venom/toxin fraction and antivenom mixtures) are intravenously injected into
mice to ensure a full systemic access under a control titration
156 Choo Hock Tan and Kae Yi Tan

potency using at least four antivenom doses for the neutraliza-

tion of each venom or individual toxin fraction (see Note 5).
2. Antivenom protein concentration: Estimate the protein concen-
tration of antivenom using bicinchoninic acid assay or Lowry
assay.
3. Antivenom incubation: Prepare a challenge dose of venom/
toxin fraction (5 LD50) in 50 μL normal saline. Mix this
challenge dose with various dilutions of antivenom in normal
saline to give a total volume of 250 μL per injection. Incubate
the venom/toxin fraction-antivenom mixture at 37  C for
30 min.
4. Hold the mice using a rodent restrainer. Inject the 250 μL
preincubated mixture intravenously into the mice via
caudal vein.
5. Allow the mice access to food and water ad libitum. Monitor
and record the survival ratio of the mice for 48 h.
6. Determine the efficacy and potency of antivenom neutraliza-
tion through the following parameters:
(a) Median effective dose, ED50: Volume of reconstituted/
liquid antivenom in μL at which 50% of mice survived (see
Note 6).
(b) Median effective ratio, ER50: Ratio of the amount of
venom/toxin (mg) to the volume dose of antivenom
(mL) at which 50% of mice survived.
(c) Potency, P: The amount of venom/toxin
(mg) neutralized completely by 1 mL antivenom (see
Note 7).
(d) Normalized potency, n-P: The amount of venom/toxin
(mg) that is completely neutralized by 1 g of antivenom
proteins (see Note 8).
*Figure 1 shows the schematic drawing of the assessment of
antivenom neutralization against venom/toxin fractions.
7. Adopt the following formulae for calculating the challenge
dose of venom/toxin fraction and the various neutralization
parameters for antivenom:
Challenge dose ðμgÞ
¼ n  LD50 ðμg=gÞ  mouse weightðgÞ
[As indicated in Subheading 3.2, step 3]

Median effective dose, ER 50 ðstandardize the unit as mg=mLÞ

½n  LD50 ðμg=gÞ  mouse weightðgÞ
¼
ED50 ðμLÞ
[As indicated in Subheading 3.2, step 6b]
 Functional Application of Snake Venom Proteomics in In Vivo Antivenom Assessment 157

Potency,P ðstandardize the unit as mg=mLÞ

½ðn
1Þ  LD50 ðμg=gÞ  mouse weightðgÞ
¼
ED50 ðμLÞ
[As indicated in Subheading 3.2, step 6c]

Normalized potency, n
P ðstandardize the unit as mg=gÞ

Potency, P ðmg=mLÞ
¼  1000
Antivenom protein concentration ðmg=mLÞ
[As indicated in Subheading 3.2, step 6d].

4 Notes

1. The initial starting dose for median lethal dose (LD50) deter-
mination can be estimated based on reported values for venoms
or toxin derived from closely related species.
2. The route of administration is intravenous to ensure a full
systemic access of the venom or toxin into the animal. This
enables the assessment and interpretation of the systemic tox-
icity of venom/toxin that becomes fully bioavailable to the
animal.
3. The survival ratios obtained from the lethality assay should
contain an upper dose which shows 100% death of mice
(n ¼ 5), a lower dose with 100% survival (n ¼ 5), and interme-
diate doses with a mix of death and survival.
4. Median lethal dose (LD50) is determined with Probit analysis
applying the Finney method.
5. If 200 μL of reconstituted antivenom failed to provide full
protection to the mice, a lower challenge dose (2.5 or 1.5
LD50) can be used. All challenge doses should be proven to be
above the lethal dose 100% (LD100) when injected intrave-
nously into the mice. This can be assessed in an additional
group of mice constituting the control.
6. The survival ratios obtained from the neutralization assay
should contain an upper dose which shows 100% survival of
mice (n ¼ 5), a lower dose with 100% death (n ¼ 5), and
intermediate doses with a mix of death and survival. Median
effective dose (ED50) is determined using Probit analysis.
7. The neutralization potency (P) is an indicator of antivenom
neutralizing capacity and is theoretically independent of the
dosage of challenge dose. This is because it takes into consider-
ation the dose of antivenom that is able to completely neutral-
ize the lethal effect of venom/toxin by subtracting 1 LD50
158 Choo Hock Tan and Kae Yi Tan

(n
1) from the total challenge dose as shown the formula

under Subheading 3.2, step 7.
8. The normalized potency (n-P) takes into consideration the
antivenom protein amount which could be varied between
different products. By normalizing the P values of different
antivenoms by their respective protein amounts, the n-P values
can be used to compare the efficacy of neutralization across
different antivenom products.

References
1. Calvete JJ (2013) Snake venomics: from the Laticauda colubrina from Bali: insights into
inventory of toxins to biology. Toxicon 75 subvenomic diversity, venom antigenicity and
(Suppl C):44–62 https://doi.org/10.1016/j. cross-neutralization by antivenom. J Proteome
toxicon.2013.03.020 166:48–58. https://doi.org/10.1016/j.jprot.
2. Tan KY, Tan CH, Fung SY, Tan NH (2015) 2017.07.002
Venomics, lethality and neutralization of Naja 8. Tan KY, Tan CH, Sim SM, Fung SY, Tan NH
kaouthia (monocled cobra) venoms from three (2016) Geographical venom variations of the
different geographical regions of Southeast Southeast Asian monocled cobra (Naja
Asia. J Proteome 120:105–125. https://doi. kaouthia): venom-induced neuromuscular
org/10.1016/j.jprot.2015.02.012 depression and antivenom neutralization.
3. Wong KY, Tan CH, Tan KY, Naeem QH, Tan Comp Biochem Physiol C Toxicol Pharmacol
NH (2018) Elucidating the biogeographical 185–186:77–86 https://doi.org/10.1016/j.
variation of the venom of Naja naja (specta- cbpc.2016.03.005
cled cobra) from Pakistan through a venom- 9. World Health Organization (2010) WHO
decomplexing proteomic study. J Proteome Guidelines for the production control and reg-
175:156–173. https://doi.org/10.1016/j. ulation of snake antivenom immunoglobulins.
jprot.2017.12.012 WHO publication, 1–141
4. Tan CH, Tan KY, Lim SE, Tan NH (2015) 10. Faisal T, Tan KY, Sim SM, Quraishi N, Tan
Venomics of the beaked sea snake, Hydrophis NH, Tan CH (2018) Proteomics, functional
schistosus: a minimalist toxin arsenal and its characterization and antivenom neutralization
cross-neutralization by heterologous antive- of the venom of Pakistani Russell’s viper
noms. J Proteome 126:121–130. https://doi. (Daboia russelii) from the wild. J Proteome
org/10.1016/j.jprot.2015.05.035 183:1–13. https://doi.org/10.1016/j.jprot.
5. Oh AMF, Tan CH, Ariaranee GC, Quraishi N, 2018.05.003
Tan NH (2017) Venomics of Bungarus caeru- 11. Tan KY, Tan CH, Fung SY, Tan NH (2016)
leus (Indian krait): comparable venom profiles, Neutralization of the principal toxins from the
variable immunoreactivities among specimens venoms of Thai Naja kaouthia and Malaysian
from Sri Lanka, India and Pakistan. J Proteome Hydrophis schistosus: insights into toxin-specific
164:1–18. https://doi.org/10.1016/j.jprot. neutralization by two different antivenoms.
2017.04.018 Toxins 8(4):86. https://doi.org/10.3390/
6. Tan CH, Tan KY, Yap MK, Tan NH (2017) toxins8040086
Venomics of Tropidolaemus wagleri, the sexu- 12. Wong KY, Tan CH, Tan NH (2016) Venom
ally dimorphic temple pit viper: unveiling a and purified toxins of the spectacled cobra
deeply conserved atypical toxin arsenal. Sci (Naja naja) from Pakistan: insights into toxic-
Rep 7:43237. https://doi.org/10.1038/ ity and antivenom neutralization. Am J Trop
srep43237 Med Hyg 94(6):1392–1399. https://doi.org/
7. Tan CH, Wong KY, Tan KY, Tan NH (2017) 10.4269/ajtmh.15-0871
Venom proteome of the yellow-lipped sea krait,
 Chapter 12

Proteomic Detection of Carbohydrate-Active Enzymes

(CAZymes) in Microbial Secretomes
Tina R. Tuveng, Vincent G. H. Eijsink, and Magnus Ø. Arntzen

Abstract
Secretomes from microorganisms growing on biomass contain carbohydrate-active enzymes (CAZymes) of
potential biotechnological interest. By analyzing such secretomes, we may discover key enzymes involved in
degradation processes and potentially infer the mode-of-action of biomass conversion. Some of these
enzymes may have predicted functions in carbohydrate degradation, while others may not, while yet
exhibiting a similar expression pattern; these latter enzymes constitute potential novel enzymes involved
in the degradation process and provide a basis for further biochemical exploration. Hence, secretomes
represent an important source for the study of both predicted and novel CAZymes. Here we describe a
plate-based culturing technique that allows for collection of protein fractions that are highly enriched for
secreted proteins, bound or unbound to the substrate, and which minimizes contamination by intracellular
proteins trough unwanted cell lysis.

Key words Secretomics, Proteomics, Protein secretion, Carbohydrate-active enzymes, CAZymes

1 Introduction

Polysaccharides such as cellulose, hemicellulose, pectin, and chitin

are abundantly produced in Nature but do not accumulate due to
removal by the concerted action of specialized microbes and micro-
bial consortia, including fungi and bacteria. These microorganisms
exploit sophisticated enzyme systems to degrade biomass and the
enzymes involved have potential in biotechnological applications,
such as in biofuel production [1]. Microorganisms tend to degrade
polysaccharides outside of the cell and then import the generated
oligo- or monosaccharides for further intracellular metabolism. To
do so, the microorganisms secrete a variety of carbohydrate-active
enzymes (CAZymes) [2–5] depending on the growth substrate and
the degrading strategy applied. The secretomes of microorganisms
thus represent an important protein subfraction for the study of
CAZymes.

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_12, © Springer Science+Business Media, LLC, part of Springer Nature 2019

159
160 Tina R. Tuveng et al.

Proteomics applied to secreted proteins is often referred to as

secretomics or exoproteomics, and should ideally only concern
proteins exported by the organisms. However, many seemingly
cytosolic proteins (by prediction) have been reported in secretome
studies [5–7], and this is commonly ascribed to cell lysis. To avoid
that these cytosolic proteins mask the identification of truly
secreted CAZymes involved in biomass conversion, it is of interest
and importance to prepare secretome samples in a way that limits
contamination by cell lysis to a minimum.
Recently we have developed a plate-based method for growing
microorganisms on solid substrates that allows to selectively obtain
fungal [8] and bacterial [9] protein subfractions that are highly
enriched for secreted proteins. Here we present this method as a
step-by-step protocol and provide detailed notes for successful
identification of CAZymes in secretomes.

2 Materials

Prepare all solutions using ultrapure water (prepared by purifying

deionized water, to attain a conductivity of 18 MΩ cm at 25
C)
and analytical grade reagents. Prepare and store all reagents at room
temperature unless indicated otherwise. Follow all waste disposal
regulations when disposing waste materials. There is no need to
adjust pH of buffers, unless indicated otherwise.

2.1 Predicting 1. Computer with access to the internet.

CAZymes and Secreted 2. FASTA file containing the predicted proteins encoded by the
Proteins organism of interest. In most cases, this can be downloaded
from UniProt (http://www.uniprot.org/) or NCBI (https://
www.ncbi.nlm.nih.gov/), or in the case of an un-sequenced
genome, be generated by genome sequencing, assembly and
translation of the DNA sequences into protein sequences, to
create a FASTA file of the proteins encoded by the organism of
interest.

2.2 Culture Plates 1. M9 minimal medium agarose plates:

with a Membrane Filter (a) Stock solution with 5 M9 salts (500 mL): Weigh in
17.04 g Na2HPO4, 7.5 g KH2PO4, 1.25 g NaCl, and
2.5 g NH4Cl. Dissolve in approx. 400 mL ultrapure water
and add ultrapure water to a total volume of 500 mL.
Autoclave at 121
C for 20 min.
(b) 1 M MgSO4 stock solution (25 mL): Weigh in 3.009 g
MgSO4. Dissolve in approx. 20 mL ultrapure water and
add ultrapure water to total volume of 25 mL. Autoclave
at 121
C for 20 min.
 CAZymes in Microbial Secretomes 161

(c) 0.1 M CaCl2 stock solution (25 mL): Weigh in 0.227 g

CaCl2. Dissolve in approx. 20 mL ultrapure water and add
ultrapure water to total volume of 25 mL. Autoclave at
121
C for 20 min.
(d) Carbon source of interest:
l Insoluble carbon sources, such as microcrystalline cel-
lulose, chitin, or filter paper, are available as powder,
flakes, or similar and are used “as is” or after milling to
a manageable particle size. Insoluble carbon sources
will be sterilized later in the method, see Subheading
3.3.1.
l Soluble carbon sources, such as carboxymethyl cellu-
lose or glucose, are to be prepared in a suitable sterile
stock solution, e.g., 20% (w/v). Preparation of 100 mL
20% (w/v) glucose stock solution: Dissolve 20 g glu-
cose in approx. 80 mL sterilized ultrapure water. Add
sterilized ultrapure water to total volume of 100 mL.
Carry out a final sterilization by passing the solution
through a sterile 0.22 μm filter.
(e) Agarose (see Note 1).
2. Sterile grade QM-A Quartz Filters, circle, 47 mm diameter
(GE Healthcare Life Sciences, Oslo, Norway).
3. Glass Petri dish with a diameter of 80 mm.
4. Heating/cooling cabinet capable of delivering a stable temper-
ature suitable for the organism of interest.

2.3 Protein 1. Water bath that can reach 100
C.

Extraction, Sample 2. 50 mL polypropylene tubes, MS-friendly (e.g., Falcon from
Preparation, and Fisher Scientific, New Hampshire, USA).
Analysis
3. 400 mM Dithiothreitol (DTT) (1 mL) stock solution: Weigh
in 0.062 g DTT and dissolve in 1 mL ultrapure water. Aliquot
into volumes of 20 μL and store at 20
C. Thaw just before
use and do not refreeze.
4. 100 mM NH4HCO3 (100 mL) stock solution: Weigh in
0.791 g NH4HCO3. Dissolve in approx. 80 mL ultrapure
water and add ultrapure water to total volume of 100 mL.
5. Trypsin, dissolved in liquid according to the manufacturer’s
recommendation, typically in 50 mM acetic acid, to a concen-
tration of 500 ng/μL (see Note 2). Aliquot into volumes of
5 μL and store at 70
C. Thaw just before use.
6. Disposable syringes, 2 mL.
7. Pipette tips of epT.I.P.S. quality (Eppendorf, Hamburg, Ger-
many) or similar.
162 Tina R. Tuveng et al.

8. Protein LoBind tubes 2 mL (Eppendorf, Hamburg, Germany)

or similar.
9. 10% (v/v) trifluoroacetic acid (TFA), LC-MS grade stock: In a
fume hood, add 1 mL 100% TFA to 9 mL ultrapure water.
Store at 20
C.
10. C18 ZipTip pipette tips (Merck Millipore, Massachusetts,
USA).
11. NanoHPLC-MS/MS system.

2.4 Data Integration 1. Computer with a spreadsheet application (e.g., Excel from
and Heat Map Microsoft, Washington, USA).
Generation 2. An installation of Perseus. Perseus is a free software package for
the analysis of (prote)omics data and can be downloaded from
http://www.coxdocs.org/doku.php?id¼perseus:start. The ver-
sion used in this chapter is 1.6.0.7.

3 Methods

3.1 Predicting The CAZy database (http://www.cazy.org) is a database specialized

Carbohydrate-Active in the display and analysis of genomic, structural, and biochemical
Enzymes (CAZymes) information of CAZymes [10]. CAZy contains more than
300 families of catalytic and auxiliary modules, classified as glyco-
side hydrolases (GHs), glycosyltransferases (GTs), polysaccharide
lyases (PLs), carbohydrate esterases (CEs), auxiliary activities
(AAs), and carbohydrate-binding modules (CBMs). CAZy identi-
fies families of evolutionary-related proteins using a classification
based on significant amino acid sequence similarity with at least one
biochemically characterized founding member [11]. Although the
CAZy web page does not allow for automatic annotation of a given
protein sequence query, it contains CAZy annotations of several
sequenced genomes (CAZome, http://www.cazy.org/Genomes.
html). If the CAZome of the organism under study is not found
in the CAZy database, alternatives exist for automated CAZyme
annotation, including dbCAN [12] and CAT [13]. Importantly,
dbCAN has generated hidden Markov models (HMMs) represent-
ing the signature domains present in every CAZy family, and
enables automatic annotation via these.

3.1.1 Predicting 1. Access dbCAN at http://csbl.bmb.uga.edu/dbCAN/anno

CAZymes Using the dbCAN tate.php.
Web Server 2. Enter/paste one or more protein sequences in the submission
form, or select a protein FASTA file (see Subheading 2.1, step 2)
for upload of multiple protein sequences.
3. Leave your email address if many sequences are loaded; the
results will be emailed when the job is done.
 CAZymes in Microbial Secretomes 163

Fig. 1 The figure shows an example of CAZyme prediction using the dbCAN web server. The FASTA-formatted
sequence of the chitin-binding protein (UniProt: B3PDT6) from Cellvibrio japonicus Ueda107 was used as
input. A table shows the HMM hits to different parts of the protein sequence and a graphical representation
shows the domain architecture within the protein

4. Click submit, and after some computation time, the results will
be presented as a table and as a graphical representation of the
domain architecture (Fig. 1). Note that several CAZy modules
may be found within one protein sequence, leading to multiple
rows in the table per protein.
5. Click “Download parsed output” to download a tab-separated
file containing the results. This can be used for integration with
proteomic expression values in a spreadsheet application, see
Subheading 3.4.

3.2 Predicting Several algorithms have been developed for in silico prediction of
Protein Secretion different N-terminal signal peptides and, consequently, the subcel-
lular locations of their cognate proteins. These prediction tools
have since the 1980s developed from signal peptide prediction
based on weight matrices and the presence of specific amino acid
motifs to more sophisticated machine learning approaches [14],
including also so-called leaderless proteins, i.e., secreted proteins
without a typical signal peptide. Today, a combination of several
prediction servers is often used to obtain the most reliable predic-
tions possible [15]. It is important to remember that in silico
prediction of the subcellular location of proteins is not trivial, and
that the presence of a signal peptide is not necessarily a guarantee
for secretion. There is a plethora of prediction servers available
[16–18], some applicable to all types of organisms, while others
are more “specialized” for, e.g., bacteria. In any case, the SignalP
server (http://www.cbs.dtu.dk/services/SignalP/ [19]) offers a
good starting point for prediction of signal peptides and is
164 Tina R. Tuveng et al.

explained in detail in this chapter (see Subheading 3.2.1). Then,

depending on the type of organism, additional prediction tools can
be used to supplement the analysis to get an as good, and complete,
prediction as possible.
A prerequisite for the prediction analysis is a sequenced
genome in the form of a FASTA file (see Subheading 2.1, step 2).
Use the FASTA file for the following prediction of secreted pro-
teins. It is advisable to organize all predictions (secretion and
CAZy) in a spreadsheet (see Subheading 3.4 and Table 1) where
agreement between different prediction algorithms also can be
assessed.

3.2.1 Step-by-Step 1. Access SignalP at http://www.cbs.dtu.dk/services/SignalP/

Prediction of Secreted 2. Enter/paste one or more protein sequences in the submission
Proteins Using SignalP form, or select a protein FASTA file for upload of multiple
protein sequences. See Note 3.
3. Select organism group: Eukaryotes, Gram-negative bacteria, or
Gram-positive bacteria.
4. One can choose between four output formats where “Stan-
dard” is the most applicable for a single/low number of protein
sequences. This option provides a graphical representation of
the N-terminal protein sequence(s) with highlighting of the
predicted cleavage site, and several scores to aid the interpreta-
tion. The D-score (discrimination score) is the score that is
used to discriminate signal peptides from non-signal peptides
and this is accompanied by “YES” if a signal peptide is pre-
dicted. The output format “Short” is applicable to multiple
protein entries such as FASTA files. This provides a single line
per protein sequence including the different scores, and the
letter “Y” is displayed when a signal peptide is predicted. This
latter format is useful for copying into a spreadsheet application
and to subsequentially integrate with proteomic abundance
values, see Subheading 3.4. The two other output formats
“Long” and “All” add extra information regarding each posi-
tion in the sequence.
5. Choose “Short” output format and click “Submit.”
6. After some computation time, the results will be presented as a
table that can be selected and copied into a spreadsheet appli-
cation for integration with proteomics expression values, see
Subheading 3.4.

3.2.2 Prediction in l Prediction of lipoproteins. The LipoP web server (http://

Bacteria: Additional www.cbs.dtu.dk/services/LipoP/ [20]) discriminates between
Considerations classical signal peptides (presence of signal peptidase I (SpI)
cleavage site), signal peptides for lipoproteins (presence of signal
peptidase II (SpII) cleavage site), transmembrane helices
 CAZymes in Microbial Secretomes 165

(TMH), and cytosolic proteins (CYT). LipoP thus provides

additional information compared to SignalP, which only predicts
presence of SpI cleavage site, without further discrimination.
LipoP was trained using Gram-negative bacteria, but also
shows good performance for sequences from Gram-positive
bacteria [20, 21]. Notably, dedicated prediction tools for lipo-
proteins in Gram-positive bacteria exist, such as PRED-LIPO
(http://www.compgen.org/tools/PRED-LIPO [22]).
l Prediction of twin-arginine signal peptides. Translocation of
completely folded proteins from the cytoplasm may occur via the
twin-arginine translocation (TAT) pathway. The signal peptides
involved in this secretion pathway are longer and less hydropho-
bic than normal signal peptides, and contain a distinctive pattern
of two consecutive arginines in the N-terminal region. Several
servers are available for prediction, such as PRED-TAT (http://
www.compgen.org/tools/PRED-TAT) [23] and TatP (http://
www.cbs.dtu.dk/services/TatP/) [24].
l Prediction of transmembrane helices (TMHs). A large pro-
portion of proteins contain membrane-spanning helices and
knowledge of the presence and location of the TMHs is impor-
tant for structural, as well as functional, annotation of the pro-
teins. In general, proteins containing TMHs are not expected to
be found in true secretomes unless they are part of vesicles
produced by the microorganism. Albeit, LipoP occasionally
reports proteins containing a TMH prediction, even in secre-
tomes prepared according to the method described in this chap-
ter, and these could be confused with a signal peptide, especially
if the TMH is close to the N-terminus and only one TMH is
predicted. These TMH predictions can be further validated
using the TMHMM server, a very accurate predictor of TMHs
(http://www.cbs.dtu.dk/services/TMHMM/) [25].

3.2.3 Prediction in Fungi: To get a reliable prediction of secreted proteins in fungi, it is

Additional Considerations possible to complement the SignalP prediction with other predic-
tion servers where eukaryotic proteins can be analyzed. Two alter-
natives include Phobius (http://phobius.sbc.su.se/) [26], which
predicts TMHs and signal peptides, and WoLF PSORT (https://
wolfpsort.hgc.jp/) [27], which uses a combination of signal motifs
and sequence-derived features, such as amino acid content, to
predict secreted proteins. For a protein to be considered as
secreted, it is advisable that two of three prediction servers agree
on the prediction, such as applied in [8].

3.2.4 Nonclassically Although most secreted proteins have a signal peptide that guides
Secreted Proteins them to a secretory pathway, a limited number of proteins without
such a signal peptide (also called leaderless proteins) are secreted
166 Tina R. Tuveng et al.

[15, 28]. For example, two chitinases encoded by Serratia marces-

cens lack a classical signal peptide, while they are known to be
secreted via a process that remains unclear [29]. Such proteins are
said to be subject to so-called nonclassical secretion. SecretomeP
(http://www.cbs.dtu.dk/services/SecretomeP/) is a server that
predicts nonclassically secreted proteins in either mammalian cells
or bacteria by utilizing several sequence-derived features, including
various posttranslational modifications and localization predictions,
found in proteins known to be secreted [30, 31]. Proteins pre-
dicted as cytosolic proteins by other prediction servers could be
candidates for further analysis using SecretomeP.

3.3 Culture Plates for Carry out all procedures at room temperature unless otherwise
Secretome Enrichment specified.

3.3.1 Casting of Plates 1. The weights and volumes used below allow for preparation of
with Membrane (Work in 250 mL M9 agarose medium, which will give approx. 15 plates.
Sterile Hood) 2. (a) Using an insoluble carbon source: To achieve a 1% (w/v)
concentration of carbon source in the final plates (see Note
4): Weigh in 2.5 g insoluble carbon source and mix with
2.5 g agarose in 199.5 mL ultrapure water. Include a
magnet for later homogenization (see step 9 and Note 5).
(b) Using a soluble carbon source: Weigh in 2.5 g agarose and
mix with 187 mL ultrapure water. The soluble carbon
source will be added to the medium in step 7.
3. Autoclave at 121
C for 20 min and cool down to approx.
70
C.
4. Add 250 μL 0.1 M CaCl2 (to achieve a final concentration of
0.1 mM).
5. Add 250 μL 1 M MgSO4 (to achieve a final concentration of
1 mM).
6. Add 50 mL 5 M9 salt (to achieve a final concentration of 1).
7. If using a soluble carbon source: To achieve 1% (w/v) concen-
tration of carbon source in the final plates (see Note 4): Add
12.5 mL of the sterile carbon source stock solution (given that
the concentration of the stock is 20% (w/v), see Subheading
2.2).
8. Mix gently to ensure homogenization.
9. Poor 8 mL M9 minimal agarose medium (see Note 5) in an
80 mm glass Petri dish and let it solidify for 5–10 min. While
waiting for the agarose to solidify, store the bottle with melted
agarose at 60
C.
 CAZymes in Microbial Secretomes 167

10. Gently place a sterile QM-A Quartz Filter in the middle of the
dish and poor another 8 mL (see Note 5) of the M9 minimal
agarose medium over the filter. Let it solidify for 5–10 min.
11. Store plates at 4
C until use. Always let the plates equilibrate to
the temperature of interest before use.

3.3.2 Inoculation and 1. Grow a liquid pre-culture in the desired medium.

Growth 2. Measure the OD600 of the pre-culture and plate an appropriate
Inoculation of Bacteria amount of cells (e.g., 100 μL of a culture with OD600  0.5).
See Note 6 for considerations regarding this step.
3. Incubate at the desired temperature for an appropriate amount
of time (see Note 7).

Inoculation of Fungi 1. Collect an agar plug with a diameter of ~7 mm containing

fungus grown on a M9 agarose plate with the carbon source
of interest using a pipette tip (Fig. 2, step 1A).
2. Transfer the plug to a new plate with an embedded membrane.
Position the plug in the center over the membrane (Fig. 2, step
1B).
3. Incubate at the desired temperature for an appropriate amount
of time (see Note 7 and Fig. 2, step 2).

3.3.3 Protein Extraction As with all proteomics work, it is important to work clean to avoid
and Sample Preparation contamination from fingers, hairs, and similar.
1. Weigh an empty 50 mL Falcon tube and note the weight.
2. Flip the agar disc and punch out the agar with the 50 mL
Falcon tube directly under the filter (Fig. 2, step 3). Transfer
the plug to the Falcon tube, weigh again, and calculate the
sample net weight by subtracting the weight of the empty
Falcon tube in step 1.
3. Add 10 μL of 400 mM DTT stock solution per g sample to
achieve a final concentration of 4 mM DTT (by assuming 1 g
sample ¼ 1 mL sample).
4. Heat the sample until agar melts (i.e., by placing the tube in an
80–100
C water bath) and vortex, then boil the sample for
30 min. In this step, proteins bound to the solid substrate are
likely to be released.
5. Cool the sample to room temperature (the agar resolidifies).
6. Use a 2 mL disposable syringe (without needle), remove the
plunger, and transfer the solid sample into the syringe from the
top. Insert the plunger and crush the agar by pressing it
through the syringe, back into the Falcon tube.
7. Add 1 mL of 100 mM NH4HCO3 stock solution per g sample
to achieve a final concentration of 50 mM NH4HCO3. Mix by
gentle vortexing.
168 Tina R. Tuveng et al.

Fig. 2 Growth of microbes on membrane-containing plates and secretome sampling. Step 1: An agar plug
containing fungus grown on a normal plate is collected using the back of a sterile pipette tip (1A) and
transferred to a membrane-containing plate with the same carbon source and positioned in the center of the
plate, i.e., over the membrane (1B). In the case of bacteria, one applies a cell suspension instead of an agar
plug. Step 2: Incubate plates for the desired time. The pictures show growth of the fungus Hypocrea jecorina.
Step 3: After incubation, the agar is flipped out of the Petri dish, thus exposing the cell free agar between the
bottom of the dish and the agar-embedded membrane (3A). Use a sterile Falcon tube (or the back of a sterile
pipette tip if a smaller sample is desirable) to punch out an agar disc containing secreted proteins that have
passed through the filter (3B–C). This figure is reproduced from [8] with permission from Elsevier

8. Add 2 μg trypsin (e.g., 4 μL of a 500 ng/μL solution) and

incubate overnight at 37
C.
9. Freeze (20
C) and thaw the sample before centrifuging the
Falcon tube briefly (4500  g for 1 min); collect all liquid.
10. Transfer the liquid fraction to a 2 mL Eppendorf LoBind tube
(see Note 8).
11. Centrifuge at 16,000  g for 10 min.
12. Transfer the supernatant to a new 2 mL Eppendorf LoBind
tube. Repeat the centrifugation if necessary (no agar pieces
should be transferred).
13. Reduce the sample volume to 10–15 μL using a speed-vacuum
centrifuge and add TFA to a final concentration of 0.1% (v/v)
(see Note 9).
14. ZipTip the sample to concentrate the sample and remove
buffers, using the supplier’s recommendations (see Note 10).
 CAZymes in Microbial Secretomes 169

15. Dry the sample using a speed-vacuum centrifuge and then

dissolve in 10 μL of a solution that is appropriate for
subsequent loading on the mass spectrometer (e.g. 0.1% TFA
in ultrapure water).
16. Analyze the samples using a nanoHPLC-MS/MS system, e.g.,
as described previously [8, 9].

3.4 Integrating Quantitative proteomics has become an indispensable analytical

Quantitative tool for microbial research. Quantitative data can be acquired
Proteomics Data with using a vast number of analytical techniques including classical
Functional Annotation gel-based procedures, or modern MS-based quantitative techni-
ques with metabolic or chemical labeling, or by using label-free
approaches. A detailed description of these methods and their
strengths and limitations is beyond the scope of this chapter and a
recent review of current methods can be found in [32]. Indepen-
dent of the quantification technique used, it is most practical to
arrange the data in a tabular form with each row representing a
protein (or protein group) with the quantitative data for the differ-
ent conditions in separate columns. In addition, functional predic-
tion and prediction of secretion should also be in separate columns.
This can be achieved in a spreadsheet application. For an example,
see Table 1. For final publication of proteomics data, it is highly
advisable to deposit the data to the ProteomeXchange Consortium
accessible at http://www.proteomexchange.org/. This ensures
transparency and enables potential reuse of the data in the future.

3.4.1 Building a Heat Perseus is a free software package for analyzing quantitative (prote)
Map Using Perseus omics data and can be used with many different quantification
techniques [33], see Note 11.
1. In Perseus, click on Generic matrix upload, a small green
arrow in the upper-left corner.
2. Select the tabular file containing the quantitative proteomics
data and the predicted functional data, similar to Table 1.
Perseus supports tab- or comma-delimited text files.
3. Columns containing quantitative values should be selected as
“Main” columns.
4. Select dbCAN and secretion predictions as “Categorical”
columns.
5. Select accession number and protein name as “Text” columns.
Click OK.
6. Click Annotation of rows, and select Categorical annotation
of rows. Give biological replicates the same name, and keep
default settings. Click OK.
7. If you have quantification values ¼ 0, meaning that the protein
was not detected in quantifiable amounts, click Quality and
 170

Table 1
An example of how to structure quantitative proteomics data in a spreadsheet application

α-Chitin β-Chitin Glucose

Accn. Protein name SignalP LipoP SecretomeP Secreted? dbCAN R1 R2 R3 R1 R2 R3 R1 R2 R3

Tina R. Tuveng et al.

B3PDT6 Chitin-binding protein, Y SpI 0.90 Y AA10 32.9 32.8 33.3 33.7 33.2 33.6 28.2 27.5 28.3
putative, cbp33/10B CBM10
P14768 Endo-1,4-beta-xylanase A Y SpI 0.96 Y CBM2 26.8 26.8 27.5 33.4 32.6 34.0 25.8 ND 25.5
CBM10
GH10
B3PDV8 Pullulanase, putative, pul13B Y SpII 0.94 Y CBM48 29.4 30.3 31.5 29.6 29.3 28.7 25.3 26.0 25.7
GH13
B3PK74 Alpha-glucosidase, putative, Y SpI 0.58 Y GH97 25.6 24.0 24.5 23.9 23.9 24.0 ND ND ND
adg97B
B3PBG2 Pilin N TMH 0.93 N 31.0 30.5 30.9 30.2 30.2 30.7 30.6 30.4 30.8
B3PF53 Putative lipoprotein N SpII 0.90 Y 28.8 26.7 27.3 28.3 28.6 27.9 26.2 ND 25.8
B3PI93 SrpA-related protein N CYT 0.18 N 30.6 30.8 31.2 27.8 27.4 28.8 ND ND 24.0
The data describes selected proteins detected in the secretome of Cellvibrio japonicus Ueda107 growing on α- or β-chitin or on glucose. The quantitative values are log2-
transformed LFQ-values from the MaxQuant [37] software. The complete proteome FASTA file was used for predicting secreted proteins using different prediction servers (see
Subheading 3.2) and for predicting CAZymes (see Subheading 3.1). The table shows results for four expressed CAZymes with varying expression levels on the different substrates
and three non-CAZymes for comparison; the data is adapted from [9]. The column labeled “Secreted?” is based on the use of three prediction servers SignalP, LipoP, and
SecretomeP and was marked with Y when at least two algorithms predicted secretion. ND not detected, R1–3 replicate 1–3
 CAZymes in Microbial Secretomes 171

select Convert to NaN. For “Invalid values should be,” select

Less or equal. The “Threshold” should be set to 0. Click OK.
8. Click Visualization, Histogram and OK to see histograms for all
samples. The graphs should look like bell-shaped distributions
nicely spread across the intensity range (x-axis), but not necessarily
normally distributed for all samples. If this is not what you observe,
the quantitative data may need log-transformation. Click Basic,
Transform and OK and then redo the histograms to reevaluate.
9. Assuming three biological replicates: Click Filter rows, select
Filter rows based on valid values. “Min. valids” should be
Number and set to 2. “Mode” should be In at least one
group. Other options are kept default. See Note 12. Click OK.
10. The generated matrix can now be used for further analysis of
the data, e.g., calculation of the fraction of secreted proteins
(see Subheading 3.4.2). Perseus offers a wide variety of statisti-
cal and visual tools (see Note 11); however, it is out of the
scope of this chapter to go into the details of these options.
11. To filter the data to only contain CAZymes, click Filter rows
and then Filter rows based on categorical column. Select the
dbCAN column, and add all items in “Values” to the right
container box. Change “Mode” to Keep matching rows.
Then a new matrix will be generated where only CAZymes
are visible, see Note 13. Click OK.
12. Click on Clustering/PCA and select Hierarchical clustering.
Default cluster parameters are generally sufficient, but in some
cases, it is more practical to deselect Columns tree and manually
select the column order using the box labeled Use for clustering.
13. When clicking OK, the heat map will be generated as a new tab
called Clustering. Therein it is possible to adjust the heat map
color scale, the size of text headers, as well as the size and
thickness of the dendrograms.
14. To visualize category columns in the heat map, click the button
labeled Configure row names and choose the desired category
as Row color bar (creates a color bar next to the heat map)
and/or Addtl. row names (adds the category value as text).
15. Clusters can be defined manually by clicking on nodes in the
dendrogram, or automatically by clicking on the button labeled
Define row clusters and then either enter the number of desired
clusters, or select clustering based on a distance threshold.
16. Remember to save your Perseus file, so you can continue to
work with it at a later stage.
17. The heat map can be exported by clicking on the PDF button
and select PDF as file type, see Note 14. An example of a heat map
prepared for publication using Perseus can be found in Fig. 3.
172 Tina R. Tuveng et al.

Fig. 3 A heat map representation of quantitative proteomics data where every protein (row) has a CAZyme
annotation (see Subheading 3.1) and predicted cellular location (from LipoP, see Subheading 3.2). The figure is
a filtered subset of the complete data set, showing only proteins with a CAZy annotation. The CAZy annotations
are colored as indicated below the figure; GH glycoside hydrolase, CE carbohydrate esterase, PL polysaccha-
ride lyase, AA auxiliary activity, CBM carbohydrate-binding module. The heat map generated by Perseus was
further manually sectioned into six clusters based on similar expression patterns. The colors in the heat map
indicate protein abundance, ranging from high (red color, MaxQuant LFQ 5  1010) to low (green color,
MaxQuant LFQ 7  106). Table 1 shows an example of the data used for generating this figure. The figure is
reproduced from [9] with permission from John Wiley & Sons
 CAZymes in Microbial Secretomes 173

Table 2
Table for calculating the secretome enrichment

Protein Predicted NOT predicted Secreted fraction

count secreted secreted (%)
In experimental 351 267 84 76
secretome
In complete proteome 3711 1076 2635 29
The secreted fraction in the experimental secretome is compared to the secreted fraction in the complete proteome. The
data used in this example apply to Cellvibrio japonicus Ueda107 growing on α-chitin. The secretome was collected after
144 h growth. Data is adapted from [9]

3.4.2 Assessment of 1. Calculate the percentage of predicted secreted proteins in the

Secretome Enrichment in complete proteome according to Table 2 (see Note 15).
Samples 2. Calculate the percentage of predicted secreted proteins in the
secretome sample using the same criteria as for the complete
proteome. Comparing this percentage with the percentage
calculated for the complete proteome is a good indication of
the enrichment of secreted proteins in the secretome sample
(see Note 16).

4 Notes

1. Agarose is used instead of agar due to its purity, hence reducing

the risk of transferring contaminants to the mass spectrometer
later on. In addition, compared to agar, agarose has less inter-
actions with biomolecules such as proteins and DNA.
2. Trypsin must be of high quality and suitable for protein
sequencing, e.g., Sequencing Grade Modified Trypsin from
Promega (Wisconsin, USA).
3. Some prediction servers have limits when it comes to the
number of sequences per submission and the number of
amino acids per protein entry. In the case of complete pro-
teomes, a FASTA file may often exceed these limits and
splitting the FASTA file is necessary. This can readily be done
using a text editor or online tools such as FaBox [34], available
at http://users-birc.au.dk/biopv/php/fabox/ (click on “Fasta
dataset splitter/divider,” select your FASTA file and the desired
number of sequences in each file).
4. The ideal concentration of carbon source might vary for differ-
ent microorganisms and depends on the type of carbon source.
5. Homogenization is important before pouring out the 8 mL.
We advise to use a magnet stirrer to mix the M9 agarose
medium (include a magnet when sterilizing the agar and
174 Tina R. Tuveng et al.

carbon source). It can be useful to use a measuring cylinder to

measure 8 mL before transferring the medium to the plate.
6. When plating bacteria from a liquid pre-culture you can con-
sider harvesting cells, i.e., centrifuge and remove the medium,
and resolubilize cells in a suitable medium before plating.
Alternatively, streaking out directly from a fresh plate or from
a 80
C stock may be considered.
7. It is strongly recommended to do pre-experiments to establish
how the microorganism behaves during growth on plates with
the desired carbon source. Too much growth is not advisable,
and since measuring growth is difficult on plates, sampling at
different time-points is needed to find optimal conditions for
secretome analysis. When comparison between different
growth conditions (e.g., different carbon sources) is to be
performed, and when, hence, comparison of “equal growth
phases” is desirable, one can use the number of proteins
detected in the secretome as a very rough estimate of growth.
Note that prolonged growth inevitably leads to cell lysis, mean-
ing that artificial high numbers of proteins will be detected and
that the fraction of proteins predicted to be secreted will
decrease. In our experience, it is important to invest time in
finding optimal incubation times; optimal meaning that one
obtains a high number of proteins and little contamination
with cytosolic proteins.
8. When transferring the sample from the Falcon tube to the
LoBind Eppendorf tube, it can be useful to cut the pipet tip
to get a wider tip opening. Some agar pieces may be trans-
ferred, hence the centrifugation in the next step.
9. If the sample is accidentally dried completely during the speed-
vacuum centrifugation process, dissolve the sample in
10–15 μL 0.1% TFA before the ZipTip procedure.
10. If you have many samples, it can be useful to elute the peptides
bound to the ZipTip directly in the desired HPLC-vial (given
that the speed-vacuum centrifugation unit is able to handle
these tubes). This will reduce the number of LoBind Eppen-
dorf tubes used and also potentially reduce sample loss.
11. In this chapter, we only explain how to generate a heat map
representation of the quantitative data (see Subheading 3.4.1),
but we recommend the reader to explore the other functional-
ity of the software, such as data transformation, profile plots,
statistical tests, and volcano plots that can aid the data analysis.
Tutorials, user examples, etc. can be found at http://www.
coxdocs.org/doku.php?id¼perseus:start.
12. This filtering step removes proteins that are only identified in
one out of three biological replicates. We recommend setting a
threshold that a protein should be identified in at least two out
 CAZymes in Microbial Secretomes 175

of three biological replicates, in at least one group (e.g., one

carbon source), to be included in the analyses.
13. This filtering will remove all proteins without predicted
CAZyme annotation and hence also remove (novel) CAZymes
that have not yet been recognized as such, i.e., potentially
novel enzymes involved in biomass conversion. Potential
novel CAZymes or other enzyme possibly involved in conver-
sion of the biomass in question are likely to be abundant
proteins showing similar expression patterns as known
CAZymes.
14. All plots in Perseus can be exported as PDF files, but typically,
the figures need further improvement prior to publication. The
PDF files can be imported into a vector graphics software, e.g.,
Inkscape (https://inkscape.org/en/) or Adobe Illustrator
(http://www.adobe.com/products/illustrator.html), for gen-
erating figures suitable for publication.
15. When calculating the percentage of secreted proteins, include
proteins predicted with a SpI and Tat signal peptide, and
proteins predicted to be nonclassically secreted by SecretomeP.
Lipoproteins (SpII) must be used with caution as many
membrane-anchored proteins face the periplasm rather than
the extracellular milieu.
16. In the majority of cases where we have applied the plate
method for collecting secreted proteins, we have obtained
protein fractions that are clearly enriched for secreted proteins
([8, 9] and in unpublished results). However, for Serratia
marcescens growing on chitin, we observed larger fractions of
cytosolic proteins than expected [35]. This could possibly be
explained by nonclassical secretion systems used by this bacte-
rium, as experimental data indicated that cell lysis was not a big
problem (see discussion in [35]). Some bacteria transfer
CAZymes to the external environment by expelling enzyme-
loaded vesicles, which may lead to less good enrichment statis-
tics, despite the secretome data being correct and relevant
[36]. This emphasizes that the success of the method may
vary between microorganisms, but also that increased knowl-
edge of the organism under study enables better evaluation of
the results.

References

1. Himmel ME, Xu Q, Luo Y, Ding S-Y, 2. Payne CM, Knott BC, Mayes HB, Hansson H,
Lamed R, Bayer EA (2010) Microbial enzyme Himmel ME, Sandgren M, Ståhlberg J, Beck-
systems for biomass conversion: emerging ham GT (2015) Fungal cellulases. Chem Rev
paradigms. Biofuels 1(2):323–341. https:// 115(3):1308–1448. https://doi.org/10.
doi.org/10.4155/bfs.09.25 1021/cr500351c
176 Tina R. Tuveng et al.

3. Benz JP, Chau BH, Zheng D, Bauer S, Glass 12. Yin Y, Mao X, Yang J, Chen X, Mao F, Xu Y
NL, Somerville CR (2014) A comparative sys- (2012) dbCAN: a web resource for automated
tems analysis of polysaccharide-elicited carbohydrate-active enzyme annotation.
responses in Neurospora crassa reveals carbon Nucleic Acids Res 40(Web Server issue):
source-specific cellular adaptations. Mol W445–W451. https://doi.org/10.1093/
Microbiol 91(2):275–299. https://doi.org/ nar/gks479
10.1111/mmi.12459 13. Park BH, Karpinets TV, Syed MH, Leuze MR,
4. Suzuki K, Suzuki M, Taiyoji M, Nikaidou N, Uberbacher EC (2010) CAZymes Analysis
Watanabe T (1998) Chitin binding protein Toolkit (CAT): web service for searching and
(CBP21) in the culture supernatant of Serratia analyzing carbohydrate-active enzymes in a
marcescens 2170. Biosci Biotechnol Biochem newly sequenced organism using CAZy data-
62(1):128–135. https://doi.org/10.1271/ base. Glycobiology 20(12):1574–1584.
bbb.62.128 https://doi.org/10.1093/glycob/cwq106
5. Takasuka TE, Book AJ, Lewin GR, Currie CR, 14. Caccia D, Dugo M, Callari M, Bongarzone I
Fox BG (2013) Aerobic deconstruction of cel- (2013) Bioinformatics tools for secretome
lulosic biomass by an insect-associated Strepto- analysis. Biochim Biophys Acta Proteins Pro-
myces. Sci Rep 3:1030. https://doi.org/10. teom 1834(11):2442–2453. https://doi.org/
1038/srep01030 10.1016/j.bbapap.2013.01.039
6. Siljam€aki P, Varmanen P, Kankainen M, 15. Desvaux M, Hebraud M, Talon R, Henderson
Sukura A, Savijoki K, Nyman TA (2014) Com- IR (2009) Secretion and subcellular localiza-
parative exoprotein profiling of different Staph- tions of bacterial proteins: a semantic awareness
ylococcus epidermidis strains reveals potential issue. Trends Microbiol 17(4):139–145.
link between nonclassical protein export and https://doi.org/10.1016/j.tim.2009.01.004
virulence. J Proteome Res 13(7):3249–3261. 16. Nielsen H (2017) Predicting secretory proteins
https://doi.org/10.1021/pr500075j with SignalP. In: Kihara D (ed) Protein func-
7. Adav SS, Cheow ESH, Ravindran A, Dutta B, tion prediction: methods and protocols.
Sze SK (2012) Label free quantitative proteo- Springer, New York, pp 59–73. https://doi.
mic analysis of secretome by Thermobifida fusca org/10.1007/978-1-4939-7015-5_6
on different lignocellulosic biomass. J Prote- 17. Nielsen H (2017) Protein sorting prediction.
ome 75(12):3694–3706. https://doi.org/10. In: Journet L, Cascales E (eds) Bacterial pro-
1016/j.jprot.2012.04.031 tein secretion systems: methods and protocols.
8. Bengtsson O, Arntzen MØ, Mathiesen G, Springer, New York, pp 23–57. https://doi.
Skaugen M, Eijsink VGH (2016) A novel pro- org/10.1007/978-1-4939-7033-9_2
teomics sample preparation method for secre- 18. Nielsen H (2016) Predicting subcellular locali-
tome analysis of Hypocrea jecorina growing on zation of proteins by Bioinformatic algorithms.
insoluble substrates. J Proteome In: Bagnoli F, Rappuoli R (eds) Protein and
131:104–112. https://doi.org/10.1016/j. sugar export and assembly in gram-positive
jprot.2015.10.017 bacteria. Springer International Publishing,
9. Tuveng TR, Arntzen MØ, Bengtsson O, Gard- Cham, pp 129–158. https://doi.org/10.
ner JG, Vaaje-Kolstad G, Eijsink VGH (2016) 1007/82_2015_5006
Proteomic investigation of the secretome of 19. Petersen TN, Brunak S, Heijne G, Nielsen H
Cellvibrio japonicus during growth on chitin. (2011) SignalP 4.0: discriminating signal pep-
Proteomics 16(13):1904–1914. https://doi. tides from transmembrane regions. Nat Meth-
org/10.1002/pmic.201500419 ods 8:785. https://doi.org/10.1038/nmeth.
10. Lombard V, Golaconda Ramulu H, Drula E, 1701
Coutinho PM, Henrissat B (2014) The 20. Juncker AS, Willenbrock H, Von Heijne G,
carbohydrate-active enzymes database (CAZy) Brunak S, Nielsen H, Krogh A (2003) Predic-
in 2013. Nucleic Acids Res 42(D1): tion of lipoprotein signal peptides in Gram-
D490–D495. https://doi.org/10.1093/nar/ negative bacteria. Protein Sci 12
gkt1178 (8):1652–1662. https://doi.org/10.1110/
11. Cantarel BL, Coutinho PM, Rancurel C, ps.0303703
Bernard T, Lombard V, Henrissat B (2009) 21. Rahman O, Cummings SP, Harrington DJ,
The Carbohydrate-Active EnZymes database Sutcliffe IC (2008) Methods for the bioinfor-
(CAZy): an expert resource for Glycoge- matic identification of bacterial lipoproteins
nomics. Nucleic Acids Res 37(Database): encoded in the genomes of Gram-positive bac-
D233–D238. https://doi.org/10.1093/nar/ teria. World J Microbiol Biotechnol 24
gkn663 (11):2377. https://doi.org/10.1007/
s11274-008-9795-2
 CAZymes in Microbial Secretomes 177

22. Bagos PG, Tsirigos KD, Liakopoulos TD, 207(5):615–626. https://doi.org/10.1083/

Hamodrakas SJ (2008) Prediction of lipopro- jcb.201404127
tein signal peptides in Gram-positive bacteria 30. Bendtsen J, Kiemer L, Fausboll A, Brunak S
with a Hidden Markov Model. J Proteome Res (2005) Non-classical protein secretion in bac-
7(12):5082–5093. https://doi.org/10.1021/ teria. BMC Microbiol 5(1):58. https://doi.
pr800162c org/10.1186/1471-2180-5-58
23. Bagos PG, Nikolaou EP, Liakopoulos TD, 31. Bendtsen J, Jensen L, Blom N, von Heijne G,
Tsirigos KD (2010) Combined prediction of Brunak S (2004) Feature based prediction of
Tat and Sec signal peptides with hidden Mar- non-classical protein secretion. Protein Eng
kov models. Bioinformatics 26 Des Sel 17:349–356. https://doi.org/10.
(22):2811–2817. https://doi.org/10.1093/ 1093/protein/gzh037
bioinformatics/btq530 32. Otto A, Becher D, Schmidt F (2014) Quanti-
24. Bendtsen J, Nielsen H, Widdick D, Palmer T, tative proteomics in the field of microbiology.
Brunak S (2005) Prediction of twin-arginine Proteomics 14(4–5):547–565. https://doi.
signal peptides. BMC Bioinformatics 6:167. org/10.1002/pmic.201300403
https://doi.org/10.1186/1471-2105-6-167 33. Tyanova S, Temu T, Sinitcyn P, Carlson A,
25. Krogh A, Larsson B, von Heijne G, Sonnham- Hein MY, Geiger T, Mann M, Cox J (2016)
mer E (2001) Predicting transmembrane pro- The Perseus computational platform for com-
tein topology with a hidden Markov model: prehensive analysis of (prote)omics data. Nat
application to complete genomes. J Mol Biol Methods 13(9):731–740. https://doi.org/
305:567–580. https://doi.org/10.1006/ 10.1038/nmeth.3901
jmbi.2000.4315 34. Villesen P (2007) FaBox: an online toolbox for
26. K€all L, Krogh A, Sonnhammer EL (2007) fasta sequences. Mol Ecol Resour 7
Advantages of combined transmembrane (6):965–968. https://doi.org/10.1111/j.
topology and signal peptide prediction—the 1471-8286.2007.01821.x
Phobius web server. Nucleic Acids Res 35 35. Tuveng TR, Hagen LH, Mekasha S, Frank J,
(Suppl 2):W429–W432. https://doi.org/10. Arntzen MØ, Vaaje-Kolstad G, Eijsink VGH
1093/nar/gkm256 (2017) Genomic, proteomic and biochemical
27. Horton P, Park K-J, Obayashi T, Fujita N, analysis of the chitinolytic machinery of Serra-
Harada H, Adams-Collier C, Nakai K (2007) tia marcescens BJL200. Biochim Biophys Acta
WoLF PSORT: protein localization predictor. Proteins Proteom 1865(4):414–421. https://
Nucleic Acids Res 35(suppl_2):W585–W587. doi.org/10.1016/j.bbapap.2017.01.007
https://doi.org/10.1093/nar/gkm259 36. Arntzen MO, Varnai A, Mackie RI, Eijsink
28. Costa TR, Felisberto-Rodrigues C, Meir A, VGH, Pope PB (2017) Outer membrane vesi-
Prevost MS, Redzej A, Trokter M, Waksman cles from Fibrobacter succinogenes S85 contain
G (2015) Secretion systems in Gram-negative an array of carbohydrate-active enzymes with
bacteria: structural and mechanistic insights. versatile polysaccharide-degrading capacity.
Nat Rev Microbiol 13(6):343–359. https:// Environ Microbiol 19(7):2701–2714.
doi.org/10.1038/nrmicro3456 https://doi.org/10.1111/1462-2920.13770
29. Hamilton JJ, Marlow VL, Owen RA, Costa 37. Cox J, Mann M (2008) MaxQuant enables
Mde A, Guo M, Buchanan G, Chandra G, high peptide identification rates, individualized
Trost M, Coulthurst SJ, Palmer T, Stanley- p.p.b.-range mass accuracies and proteome-
Wall NR, Sargent F (2014) A holin and an wide protein quantification. Nat Biotechnol
endopeptidase are essential for chitinolytic pro- 26(12):1367–1372. https://doi.org/10.
tein secretion in Serratia marcescens. J Cell Biol 1038/nbt.1511
 Chapter 13

An Overview of Mass Spectrometry-Based Methods

for Functional Proteomics
J. Robert O’Neill

Abstract
The mechanism underlying many biological phenotypes remains unknown despite the increasing availabil-
ity of whole genome and transcriptome sequencing. Direct measurement of changes in protein expression is
an attractive alternative and has the potential to reveal novel processes. Mass spectrometry has become the
standard method for proteomics, allowing both the confident identification and quantification of thousands
of proteins from biological samples. In this review, mass spectrometry-based proteomic methods and their
applications are described.

Key words Mass spectrometry, Proteomics, Quantitation, Label-free, Selective reaction monitoring,
MALDI

1 The Challenge of Measuring the Proteome

The study of the entire protein content of an organism, tissue, or

cell was first described as proteomics nearly 20 years ago [1]. Mass
spectrometry has become the de facto standard method for prote-
omics, allowing the confident identification of proteins from com-
plex mixtures [2].
Although the goal of measuring an entire eukaryotic proteome
has been achieved [3], the human proteome has yet to be described
in toto despite the publication of the complete human genome at
the turn of the century [4]. The human proteome project has
delivered progressive increments toward this goal [5, 6] yet as of
the August 2017 data release of neXtProt, the most comprehensive
human protein database available, no direct experimental evidence
has been provided for 3031 (15%) of the predicted 20,199 proteins
comprising the human proteome [7]. Several reasons can be pro-
posed for this disparity.
The polymerase chain reaction (PCR) allows template nucleo-
tide sequences to be copied with an increase in number of many
orders of magnitude and very low error rates [8]. Complementary-

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_13, © Springer Science+Business Media, LLC, part of Springer Nature 2019

179
180 J. Robert O’Neill

base pairing also allows cryptic nucleotide sequences to be rapidly

deciphered [9]. The combination of these methods and advances in
computational assembly of short sequence reads allows nucleotide
sequencing to proceed in massively parallel configurations to
sequence entire genomes within hours [10].
In contrast, the de novo identification of protein sequences
contains greater intrinsic challenges. No method exists to amplify
protein or peptide sequences and therefore proteomic methods are
always restricted by the input mass. Similarly amino acids do not
exhibit complementation and identification relies on mass measure-
ment or, historically, chromatography or electrophoresis [11]. The
proteome is also significantly larger than the genome with alterna-
tive splicing and alternative transcription start sites contributing to
transcriptome and ultimately proteome diversity [12].
A further challenge is posed by greater combinatorial possibi-
lities with up to 21 amino acids used interchangeably to generate
peptides. This complexity is further increased by posttranslational
modifications (PTMs) including the addition of biochemical
groups such as a phosphate (phosphorylation), a carbohydrate
(glycosylation), and at least 25 other distinct moieties or
modifications [11].
A final compounding difficulty is the dynamic nature of the
proteome. The genome sequence of an organism is constant across
all cells in that organism and is relatively stable in the face of DNA
extraction methods even allowing DNA sequences to be obtained
from ancient specimens [13]. In contrast, the proteome varies from
cell to cell [14] and is highly context-dependent with the post-
translational state of a single protein varying across subcellular
locations [15]. Extracting the proteome for quantification is also
confounded by the rapid alterations in the PTM state induced by
hypoxia and changes in intracellular pH with some phosphoryla-
tions reported to be lost within 60 min of tissue biopsy [16]. Many
of these challenges have been addressed with technological
advances, the most significant of which is the use of high scanning
speed, high accuracy mass spectrometry [17].

2 Mass Spectrometry for Proteomics

The fundamental components of a mass spectrometer consist of an

ion detector coupled to a mass analyzer that measures both the
number and the mass-to-charge ratio (m/z) of ions generated into
the gas phase from an ionization source. Variations on the instru-
mentation abound, however, each with their own strengths and
weaknesses [18–22]. Despite this, all combine high sensitivity and
high mass accuracy to finally bring the measurement of whole
proteomes within reach.
 MS-Based Proteomics 181

Electrospray ionization (ESI) sources ionize analytes directly

into the gas phase from liquid, commonly a polar volatile solvent
eluted from a chromatography column [23]. These sources are
most commonly used for the analysis of complex mixtures includ-
ing cell lysates. Alternatives include matrix-assisted laser desorption
ionization (MALDI) sources which use a laser to ionize analytes
directly into the gas phase out of a solid matrix [24]. These sources
are limited in the number of ions that can be generated and have
previously been reserved for relatively homogenous analytes.

3 Protein Identification Using MS: Bottom-Up Approach

The de novo identification of proteins from a complex mixture can

be achieved by several means. The most common method, termed
“shotgun” or “bottom-up” proteomics relies on the identification
of peptides generated by proteolytic digestion of the protein mix-
ture. The presence of a protein in the original mixture is then
inferred by interrogation of a protein sequence database with the
identified peptide sequences. Matching a peptide sequence unique
to a particular protein provides evidence of the protein in the
original mixture [2]. An example overview of the workflow is
illustrated in Fig. 1.
Shotgun proteomics relies on tandem mass spectrometry
(MS/MS) where peptides are ionized to generate precursor ions,
analyzed and separated according to their mass-to-charge ratio (m/
z) in the primary mass spectrometry run (MS1) . Precursor ions are
then fragmented, usually by collision ion dissociation, and the
fragment ions are separated and analyzed in the second MS run
(MS2) [25]. Multiple fragment species are generated from the same
peptide and, with high-quality spectra and sufficient fragment ions,
species differing by each individual amino acid in the peptide will be
discernible as discrete ion peaks separated by a measured mass
difference. As amino acids all have a fixed, defined mass, the
measured difference can be used to identify the amino acid
[26]. Thus the sequence of the peptide can be determined directly,
defined as de novo peptide sequencing [27]. In practice, with
complex peptide mixtures it is rarely possible to sequence all pep-
tides directly and this labor-intensive approach is reserved for
organisms with limited genome sequence information and there-
fore limited or absent potential protein databases.
More commonly, database searching is performed to generate
peptide-spectrum matches. Several algorithms have been described
but they generally follow the same principle; the measured precur-
sor mass is used to filter a database of peptides generated by in silico
digestion of a list of potentially identifiable proteins. Theoretical
fragment-ion mass differences are generated for all the candidate
peptides with a matching precursor mass [28]. These are compared
182 J. Robert O’Neill

Tissue Proteins
Proteins A BC D A BC Identifications
enzymatic
digestion
protein

Computational
Experimental

inference
Peptides
fractionation
Peptide IDs
*
tandem mass database
spectrometry search

Fragment ions MS/MS spectra

Fig. 1 An overview of protein identification by shotgun proteomics. A complex protein mixture, in this example
a tissue sample containing proteins A–D, is proteolytically digested to yield peptides. Each peptide is
illustrated as a colored box. To reduce the mixture complexity, peptides are fractionated by a common
property such as isoelectric point. Peptide fractions are subjected to tandem mass spectrometry to yield
fragment ion spectra. Peptide-spectrum matches (IDs) are made using a protein database, the peptide
(precursor) ion masses, and a database search tool [91]. Not all fragment ion spectra result in a peptide
match and some peptide matches are of low confidence (e.g., light green peptide; *). Using further statistical
tools [28], proteins are identified with unique peptide matches confirming the presence of a protein in the
original mixture. Each shotgun experiment only identifies a subset of the proteome from complex mixtures
such as tissue lysate, so in this example protein “D” has not been identified

with the identified fragment-ion spectra and candidates ranked

using a scoring algorithm, specific to the database search
method [29].
These methods can identify peptides without the requirement
for prior mass spectrometry. Organism-specific spectral libraries
generated using stringent identification thresholds and evaluation
of millions of published experimentally derived peptide spectra are
now available [30, 31]. An alternative, or complementary, approach
is to search identified spectra against these libraries, incorporating
other spectral features such as relative ion intensity. This has been
reported to enhance the number of peptide identifications com-
pared to standard database search strategies [32, 33].
The sequences of identified peptides are then used to identify
proteins using the original search database. A variety of statistical
approaches are included in commonly used software packages to
deal with protein inference problems such as repeated peptide
sequencing events, peptides shared between multiple proteins and
estimating the false-discovery rate [28].

3.1 Fractionation A significant limitation of mass spectrometry is the throughput of

ions that can be analyzed. Although this has improved with the
current generation of instruments, the number of analytes that can
be studied simultaneously is still often the limiting factor. Tissue
 MS-Based Proteomics 183

lysates contain highly diverse mixtures of proteins. This diversity is

further compounded by proteolytic digestion, presenting signifi-
cant challenges for peptide spectrum matching. Samples are often
fractionated to reduce this complexity. Approaches include strong
cation exchange [34], subcellular fractionation [35], isoelectric
focusing electrophoresis [36], high pH (basic) reversed phase
[37], and other chromatography methods [38]. By delivering frac-
tions with reduced numbers of unique peptides into the mass
spectrometer, homogenous m/z fractions can be produced during
the MS1 phase which can be accurately sequenced during the MS2
phase [39, 40].

4 Data-Dependent and -Independent Shotgun Proteomics

A key feature of the shotgun proteomic method as described previ-

ously is the selection of precursor ions for fragmentation in the
MS2 phase. This is usually performed on the basis of precursor ion
intensity and is referred to as the data-dependent approach
[41]. This method has the limitation that a precursor ion must be
detected to allow peptide sequencing and places an intrinsic bias
toward abundant precursor species. An alternative strategy is to
systematically fragment all precursor ions within windows of a
defined m/z range regardless of whether a precursor ion was
detected or not [21, 42]. In one iteration of this method, the
precursor mass used for peptide spectrum matching is assigned as
the center of the MS1 m/z window. When this method is applied,
fragment ions yielding high-confidence peptide spectrum matches
can be detected in up to 10% of cases in the absence of a precursor
ion [42, 43] and this approach can enhance the dynamic range of
detection by identifying low-abundance peptides. A disadvantage is
the long data acquisition times required to obtain spectra across all
m/z windows although faster instruments and optimized chroma-
tography have reduced this [44].

5 Protein Identification Using MS: Top-Down Approach

An exciting recent development has been the ability to identify

intact proteins by mass spectrometry, a so-called “top-down”
approach [45]. Proof of concept studies have demonstrated the
capacity to identify several thousand distinct protein isoforms (pro-
teoforms) using cultured mammalian cells and extensive orthogo-
nal fractionation in the liquid phase [46–48]. An advantage of this
method is the direct identification of proteins, rather than inference
from peptide identifications using the shotgun approach. This pro-
vides the potential to characterize the entire population of proteo-
forms generated from a single gene and identify dynamic changes in
184 J. Robert O’Neill

protein-processing, alternative-splicing or posttranslational modifi-

cation often not possible from peptide-level data alone. Although
not currently capable of proteome-scale analysis, with further
developments in automated fractionation, instrumentation and
data analysis methods, this may become feasible in the future
[49]. Measuring dynamic changes in cellular states, the goal of
most biological proteomic experiments, however, requires quanti-
tation in addition to protein identification. Methods to undertake
this using a hypothesis-free top-down approach are in early devel-
opment and still lack the robustness of shotgun approaches [50].

6 Selective Reaction Monitoring

Mass spectrometry provides an ideal method to allow the

hypothesis-free discovery of expressed proteins in biological sam-
ples using either the “bottom-up” or “top-down” approaches
described. The development of high-quality comprehensive spec-
tral libraries and the availability of synthetic peptides have allowed
the development of robust mass spectrometry “assays” covering the
entire human proteome and that of several model organisms
[30, 51, 52]. These databases can be used for hypothesis-driven
studies to quantify protein expression across samples. The com-
monest application of this method is selective reaction monitoring.
In this method, a peptide unique to the protein of interest and
consistently identifiable by mass spectrometry is selected (proteo-
typic peptide) [53]. The spectral features of this peptide are then
used to isolate precursor ions using defined mass windows. This
significantly reduces the complexity of the ion mixture for
subsequent fragmentation and peptide identification. This also
significantly reduces the analysis time so that higher numbers of
samples can be analyzed.
Many dozens of proteins can be assayed simultaneously by
multiplexing this strategy (multiple reaction monitoring; MRM).
By spiking-in isotopically labeled synthetic proteotypic peptides at a
defined concentration, the absolute concentration of peptide, and
by inference the protein, of interest can be determined with high
accuracy. The higher throughput of MRM approaches means that
they are commonly employed in the validation phase of biomarker
development studies when shortlisted biomarker candidates deter-
mined in a discovery proteomic experiment in a small number of
samples are assessed in several hundred further samples [54].
Improvements in instrument scanning speeds and the application
of data analysis approaches from selective reaction monitoring have
been employed in a further hybrid proteomic method. This tech-
nique, termed sequential window acquisition of all theoretical
fragment-ion spectra or “SWATH MS” [21], has high technical
reproducibility and quantitative accuracy [19, 55, 56]. In this
 MS-Based Proteomics 185

method, data are acquired using a data-independent shotgun proteo-

mic method, and peptide identifications are made on a candidate basis
using the SRM approach. Proponents of this technique claim that a
“digital” representation of the protein state of a biological sample is
created and this can be assessed retrospectively as hypotheses are
subsequently developed without the need for further mass spectrom-
etry. Although compelling as a concept, complete proteome coverage
is still not routinely achieved using current instruments and identify-
ing biologically significant changes in protein modifications such as
phosphorylation still requires careful experimental control and
modification-specific sample preparation methods.

7 Quantitative Proteomics

A striking common finding of the increasing number of large-scale

proteomic studies is that few proteins exhibit tissue-specific expres-
sion [57, 58]. In almost all cases, therefore, diverse phenotypes are
manifest through changes in protein expression level, subcellular
localization, or posttranslational modifications rather than the pres-
ence or absence of protein expression. If the experimental objective
is to understand the mechanism governing an observed phenotype,
then quantifying protein expression is of central importance.

7.1 Gel-Based A typical proteomic experimental design is to compare a biological

Methods for sample under two or more conditions and attempt to identify differ-
Quantitative entially expressed proteins. Historically, 2D gel electrophoresis would
Proteomics be used to separate the lysates from each condition according to
protein mass and isoelectric point [59]. Gels could then be stained
using a silver-based or other similar methods and differentially
expressed proteins could be identified as spots of differing intensities
[60]. A variation of this method minimized the gel-to-gel variability
by labeling all the proteins in each sample with a different fluorophore
and running all the samples together on the same gel [61]. By quan-
tifying the relative emission from each fluorophore across the spots,
the relative expression could be determined.
In both examples, differentially expressed spots containing
proteins of unknown identify are excised, digested to peptides
using proteolytic enzymes, and subjected to mass spectrometry
for peptide and subsequent protein identification using similar
strategies to shotgun proteomics. This method has the advantage
of limiting the protein identifications to a small number of differ-
entially expressed proteins, and providing a relatively homogenous
sample for mass spectrometry. Unfortunately, despite advances in
the automation of spot detection and quantification, these methods
were only semiquantitative, labor-intensive, the data quality was
highly user dependent, and protein identifications were limited to
a few dozen per experiment.
186 J. Robert O’Neill

7.2 Quantitative Advances in sample processing and instrumentation have enabled

Shotgun Proteomics the development of quantitative shotgun proteomic methods.
These rely on lysis, digestion, and usually fractionation of samples
prior to liquid chromatography (LC) and MS/MS. A labeling
phase can be incorporated into the sample preparation stages
prior to MS/MS or peptides can be quantified directly using
label-free strategies [62].

7.3 Quantitative Chemical labeling can take place at the protein or peptide level. The
Shotgun Proteomics use of stable carbon, hydrogen, and nitrogen isotopes allows differ-
Using Labeling ential labeling of amino acids such as Leucine, Lysine, and Arginine
that will remain biochemically identical but through their mass
7.3.1 Stable Isotope differences are resolvable as discrete spectral peaks. This approach,
Labeling of Amino Acids in termed Stable Isotopic Labeling of Amino Acids in Culture
Culture (SILAC) (SILAC), allows the proteins in mammalian cells in culture to be
isotopically labeled by the use of medium containing only “heavy”
amino acids [63]. A typical experiment would comprise one trea-
ted, “heavy”-labeled cell line and a control unlabeled, “light”, cell
line. Cell lysates are mixed in a 1:1 ratio and then subjected to
standard LC-MS/MS workflow. Peptides are identified in the usual
fashion and the relative expression between cell line conditions
identified at the MS1 level by the ratio of heavy to light peptide
ion intensities. This approach has been shown to be reproducible
across the proteome with a coefficient of variation of ~30% [62]. By
using both heavy lysine and heavy arginine combinations, three
conditions can be compared simultaneously.
A disadvantage of SILAC approaches is the requirement for
complete label uptake by cultured cells, which limits the application
to cells which express stable phenotypes of interest across several
passages. The requirement for prior labeling in the conventional
SILAC method also precludes the study of human tissues samples
although fully isotopically labeled organisms have been described
which may have application in disease models [64–66].

7.3.2 Super-SILAC A variation of the SILAC method, termed super-SILAC, has been
applied to quantify the proteome of human cancer samples [67]. In
this procedure a mixture of cell lines derived from the cancer tissue
of interest and approximately covering the expression profile of the
tissue of interest are heavy-labeled using the SILAC method. A
mixture of lysates from these cell lines with a defined protein mass
is spiked-in to each tissue lysate in a 1:1 ratio before digestion,
fractionation, and LC-MS/MS using standard procedures. Peptide
identification and quantitation then proceeds as for a standard
SILAC experiment. The ratio of expression between heavy and
light peptides is calculated for each tissue sample. The constant
SILAC spike-in mass provides a method of normalizing between
experimental runs and also, by calculating the ratio of ratios, allows
the relative expression between tissue types to be calculated [68].
 MS-Based Proteomics 187

An advantage is the spike-in standard can be used in multiple

experiments on multiple platforms and still allow normalization
between experiments and, once the spike-in standard is generated,
there is no further labeling steps or reagent costs. A disadvantage is
that the accuracy of SILAC is highest at ratios <2 and therefore a
relatively close match to the tissue expression profile is required
[62]. SILAC media is not yet available for many primary cells and
therefore primary human tissues or cancers with few available cell
lines may be difficult to analyze with this technique. Similarly,
proteins unique to a tissue sample will not be quantified.

7.3.3 Isotope-Coded In this method, the cysteine residues of reduced proteins are labeled
Affinity Tags (ICAT) with tags comprising a composite of a sulfhydryl reacting group, a
deuterated linker, and a biotin affinity tag [69]. Proteins from
discrete samples can be differentially labeled as both “light” and
“heavy” isotopes of the linker are available. Labeled samples are
then pooled and digested together. Cysteine-containing peptides
are then enriched by avidin-affinity chromatography. Peptides can
then be further fractionated or directly subjected to LC-MS/MS.
The different isotopes of the deuterated linker provide discrete
mass peaks during MS1 analysis to allow differential expression
analysis.
Unfortunately only cysteine-containing proteins can be stud-
ied, limiting proteome-wide efforts and the bulky affinity group,
biotin, introduces significant background into the MS/MS spectra
[70]. Furthermore, deuterated labels are more hydrophobic and
therefore are differentially eluted during reverse phase LC, compli-
cating the MS analysis [71]. This technique still has a role, however,
as the affinity enrichment step allows the study of low-abundance
proteins, not easily accessible by other methods.
18
7.3.4 O Labeling An advantage of this strategy is that it can be applied to almost any
sample. In an approach that predates the SILAC method, samples
for comparison are either proteolytically digested in 18O-contain-
ing water for the “heavy” sample or standard “light” water [72]. As
the protease, in most cases trypsin, cleaves the peptide bonds, the
heavy isotope is incorporated, so all tryptic peptides will be labeled.
The subsequent data analysis is identical to SILAC methods. A
disadvantage of this approach is the relative expense of H218O.

7.3.5 Dimethyl Isotopic A further method uses standard and deuterium isotopes of formal-
Labeling dehyde to label the amino-terminus of peptides or the amino group
of Lysine residues [73]. The isotopes are subsequently resolved by
their mass differences allowing peptide-level quantitation from the
MS1 scan. A further limitation common to SILAC, 18O, and
Dimethyl labeling is that maximum of three samples can be com-
pared per mass spectrometry analysis.
188 J. Robert O’Neill

Reporter Group Balance Amine Reactive

O

O
N N

N O

O
Mass 114-117 Da Mass 28-31 Da
m/z )
m/z
m/z
m/z N (+0)

Mass 145 Da

Fig. 2 Schematic of the four-plex iTRAQ peptide label. A reporter group with a
defined mass between 114 and 117 Da is connected to a balancing linker.
Together the reporter and linkers have a fixed mass of 145 Da and they are
connected to an amine-reactive group which binds peptide amino-termini and
lysine residues. The label is cleaved at the balancing linker during MS2 fragment
ion generation to allow reporter ion detection. Figure adapted from [74]

7.3.6 Isobaric Peptide Isobaric peptide labels offer greater multiplexing capabilities with
Labeling 4-plex or 8-plex (Isobaric Tag for Relative and Absolute Quantifi-
cation; iTRAQ) [74] or 6-plex, 10-plex or 11-plex (Tandem Mass
Tags; TMT) commercial kits available [75]. These kits all rely on
the same underlying principle.
Each label consists of an amine-reactive ester, a balancing car-
bonyl linker, and a reporter ion (Fig. 2). Tryptic peptides form
amide linkages with the labels via N-termini or lysine residues. A
label with a different reporter is used for each different sample and
all the samples are mixed prior to fractionation and LC-MS/MS.
Each label has the same total mass and chromatographic properties
and therefore the LC retention time and mass/charge (m/z) sepa-
ration of each sample are not differentially affected during the MS1
scan [74].
Precursor ions are then sampled for MS/MS analysis and the
ionized-labeled peptides are fragmented with dissociation of the
reporter ions from the balancing carbonyl linker. The peptide frag-
ments are detected generating mass spectra in the usual manner.
The reporter ions are also detected as peaks at a predefined m/z.
For a four-plex iTRAQ experiment the reporters are detected at
114.1, 115.1, 116.1, and 117.1 m/z [74]. For a six-plex TMT
 MS-Based Proteomics 189

experiment, the reporters are detected between 126.1 and

131.1 m/z [75].
Assuming complete peptide labeling of each sample, the more
abundant peptides within each sample will have accumulated more
label. When equal amounts of each sample are mixed and subjected
to LC-MS/MS together, those samples with a greater original
concentration of a peptide of interest will produce higher reporter
ion peak intensities in the MS/MS scan. By comparing the relative
reporter ion intensities, the relative peptide and therefore protein
abundances in the original samples can be determined [76].
The multiplexing capabilities of isobaric labels are directly off-
set by the consequent dilution of each sample leading to challenges
in identifying low-abundance peptides [44]. Samples also require to
be lysed and digested separately which has the potential to intro-
duce error. In contrast, cell populations can be mixed prior to lysis
in SILAC experiments. The quantitative accuracy and dynamic
range offered by isobaric labels are excellent, however, surpassing
SILAC in a direct comparison [62].

7.4 Label-Free All sample manipulation steps during a proteomic workflow reduce
Quantitative Shotgun the data yield due to loss of proteins [77] and are additional sources
Proteomics of variation [78]. Eliminating the sample processing steps to incor-
porate labels for quantitation is clearly an advantage and underlies
the rationale to develop label-free methods of quantitation.
The total number of spectra matched to each peptide contri-
buting to a protein identification, termed the spectral count, has
been reported to correlate with absolute protein abundance
[79]. Various methods have been proposed to refine the spectral
count such as normalizing for protein length [80], or combination
scores including peptide count and fragment-ion intensity
[81]. For complex protein mixtures, spectral counts are still subject
to significant between-run variability and are highly dependent on
LC conditions and precursor ion selection. As a result, the quanti-
tative reproducibility of spectral counting is inferior to isobaric
labeling methods [82].
An alternative relies on the capture of precursor ion intensity as
a function of time to produce an ion chromatogram. The area
under the ion chromatogram curve is linearly proportional to the
peptide concentration [83]. Challenges exist in applying this
method across LC-MS/MS runs to allow differential analysis as
the same peptide ion must be identified and quantified despite
background noise, co-eluting peptides causing signal overlap, tech-
nical variations in retention time and total protein loading among
other factors [84].
A simultaneous advantage and disadvantage of both label-free
approaches is the requirement to analyze one sample per LC-MS/
MS run. This prevents sample dilution, provides maximum poten-
tial coverage, and prevents the potential failure to identify
190 J. Robert O’Neill

dysregulated low-abundance proteins that occurs with multiplexed

approaches. By comparing conditions across separate LC-MS/MS
runs, however, the inherent changes in LC performance and the
stochastic nature of protein identification by shotgun proteomics
both contribute to data heterogeneity. Until these concerns are
addressed, labeling strategies will still be widely employed.

8 MALDI-Imaging MS (MALDI-IMS)

A major disadvantage of lysing tissue biopsies for downstream mass

spectrometry analysis is the loss of microscopic spatial information
relating to protein expression. The local microenvironmental con-
text of a cell is critical in determining behavior with cancer tissues
being a well-recognized example [85]. Understanding the changes
in protein expression that occur within a defined cellular niche may
unveil novel insights not apparent from analysis of tissue biopsies in
toto [86].
To preserve this heterogeneity, matrix-assisted laser desorption
and ionization (MALDI) techniques have been adapted to allow
direct ionization and mass spectrometry from tissue sections
[87]. By co-registering spectra and histological images, patterns
of protein expression can be interpreted within their biological
and cellular context. The significant advantages of this imaging
mass spectrometry (MALDI-IMS) method are offset by some of
the limitations common to all MALDI approaches [88].
MALDI-IMS generates spectral features (m/z) which can be
used to differentiate samples but does not provide protein identi-
ties. Hybrid approaches with downstream tandem mass spectrome-
try allow low mass proteins to be identified directly although
proteome coverage has not yet reached parity with LC-MS/MS
analysis of tissue lysates [89]. A further significant limitation is in
the resolution of ionization sources. Current technologies allow a
minimum resolvable area of 10 μm but most analyses are practically
limited to areas of 100 μm [90]. This allows a granular expression
map to be generated but the goal of identifying subcellular expres-
sion patterns, for example, at the tumor-stromal interface, remains
elusive.

9 Conclusion

Biological phenotypes are governed by protein interactions and the

correlation between RNA and protein expression is limited in many
circumstances. The large-scale, direct measurement of protein
expression is therefore an attractive prospect for the biologist.
Mass spectrometry offers the potential to identify expressed pro-
teins from biological samples in a hypothesis-free manner.
 MS-Based Proteomics 191

Technological advances have allowed the goal of proteome-wide

measurement to become a reality in some model systems.
Monitoring dynamic changes in protein abundance has become
feasible using both biochemical labeling strategies to provide highly
accurate protein quantitation and label-free techniques. The evolu-
tion of selective reactive monitoring methods allows robust identi-
fication and quantitation across conditions. MALDI-IMS allows
the spatial diversity of protein expression in biological tissues to
be preserved and novel insights into disease processes such as cancer
can be gleaned directly from tissue sections. By applying this array
of proteomic techniques, scientists can address fundamental ques-
tions and begin to understand biological processes and disease
pathophysiology.

References
1. Wasinger VC, Cordwell SJ, Cerpa-Poljak A, 7. Gaudet P, Michel PA, Zahn-Zabal M, Britan A,
Yan JX, Gooley AA, Wilkins MR, Duncan Cusin I, Domagalski M, Duek PD, Gateau A,
MW, Harris R, Williams KL, Humphery- Gleizes A, Hinard V, Rech de Laval V, Lin J,
Smith I (1995) Progress with gene-product Nikitin F, Schaeffer M, Teixeira D, Lane L,
mapping of the Mollicutes: mycoplasma geni- Bairoch A (2017) The neXtProt knowledge-
talium. Electrophoresis 16(7):1090–1094 base on human proteins: 2017 update. Nucleic
2. Aebersold R, Mann M (2003) Mass Acids Res 45(D1):D177–D182. https://doi.
spectrometry-based proteomics. Nature 422 org/10.1093/nar/gkw1062
(6928):198–207. https://doi.org/10.1038/ 8. Saiki RK, Gelfand DH, Stoffel S, Scharf SJ,
nature01511 Higuchi R, Horn GT, Mullis KB, Erlich HA
3. de Godoy LM, Olsen JV, Cox J, Nielsen ML, (1988) Primer-directed enzymatic amplifica-
Hubner NC, Fröhlich F, Walther TC, Mann M tion of DNA with a thermostable DNA poly-
(2008) Comprehensive mass-spectrometry- merase. Science 239(4839):487–491
based proteome quantification of haploid ver- 9. Liu L, Li Y, Li S, Hu N, He Y, Pong R, Lin D,
sus diploid yeast. Nature 455 Lu L, Law M (2012) Comparison of next-
(7217):1251–1254. https://doi.org/10. generation sequencing systems. J Biomed Bio-
1038/nature07341 technol 2012:251364. https://doi.org/10.
4. Consortium IHGS (2004) Finishing the 1155/2012/251364
euchromatic sequence of the human genome. 10. Pareek CS, Smoczynski R, Tretyn A (2011)
Nature 431(7011):931–945. https://doi.org/ Sequencing technologies and genome
10.1038/nature03001 sequencing. J Appl Genet 52(4):413–435.
5. Legrain P, Aebersold R, Archakov A, https://doi.org/10.1007/s13353-011-0057-
Bairoch A, Bala K, Beretta L, Bergeron J, x
Borchers CH, Corthals GL, Costello CE, 11. Niall HD (1973) Automated Edman degrada-
Deutsch EW, Domon B, Hancock W, He F, tion: the protein sequenator. Methods Enzy-
Hochstrasser D, Marko-Varga G, Salekdeh mol 27:942–1010
GH, Sechi S, Snyder M, Srivastava S, 12. Sperling J, Azubel M, Sperling R (2008) Struc-
Uhlén M, Wu CH, Yamamoto T, Paik YK, ture and function of the Pre-mRNA splicing
Omenn GS (2011) The human proteome proj- machine. Structure 16(11):1605–1615.
ect: current state and future direction. Mol Cell https://doi.org/10.1016/j.str.2008.08.011
Proteomics 10(7):M111.009993. https://doi. 13. Ermini L, Olivieri C, Rizzi E, Corti G,
org/10.1074/mcp.M111.009993 Bonnal R, Soares P, Luciani S, Marota I, De
6. Omenn GS, Lane L, Lundberg EK, Overall Bellis G, Richards MB, Rollo F (2008) Com-
CM, Deutsch EW (2017) Progress on the plete mitochondrial genome sequence of the
HUPO Draft Human Proteome: 2017 Metrics Tyrolean iceman. Curr Biol 18
of the Human Proteome Project. J Proteome (21):1687–1693. https://doi.org/10.1016/j.
Res 16:4281. https://doi.org/10.1021/acs. cub.2008.09.028
jproteome.7b00375
192 J. Robert O’Neill

14. Elguoshy A, Hirao Y, Xu B, Saito S, Quadery top down LC MS/MS and versatile peptide
AF, Yamamoto K, Mitsui T, Yamamoto T, fragmentation modes. Mol Cell Proteomics
JProS CXPTo (2017) Identification and valida- 11(3):O111.013698. https://doi.org/10.
tion of human missing proteins and peptides in 1074/mcp.O111.013698
public proteome databases: data mining strat- 21. Gillet LC, Navarro P, Tate S, Röst H,
egy. J Proteome Res 16:4403. https://doi. Selevsek N, Reiter L, Bonner R, Aebersold R
org/10.1021/acs.jproteome.7b00423 (2012) Targeted data extraction of the
15. Thul PJ, Åkesson L, Wiking M, Mahdessian D, MS/MS spectra generated by data-
Geladaki A, Ait Blal H, Alm T, Asplund A, independent acquisition: a new concept for
Björk L, Breckels LM, B€ackström A, consistent and accurate proteome analysis.
Danielsson F, Fagerberg L, Fall J, Gatto L, Mol Cell Proteomics 11(6):O111.016717.
Gnann C, Hober S, Hjelmare M, https://doi.org/10.1074/mcp.O111.016717
Johansson F, Lee S, Lindskog C, Mulder J, 22. Souza GH, Guest PC, Martins-de-Souza D
Mulvey CM, Nilsson P, Oksvold P, (2017) LC-MS(E), multiplex MS/MS, ion
Rockberg J, Schutten R, Schwenk JM, Siverts- mobility, and label-free quantitation in clinical
son Å, Sjöstedt E, Skogs M, Stadler C, Sullivan proteomics. Methods Mol Biol 1546:57–73.
DP, Tegel H, Winsnes C, Zhang C, https://doi.org/10.1007/978-1-4939-6730-
Zwahlen M, Mardinoglu A, Pontén F, von 8_4
Feilitzen K, Lilley KS, Uhlén M, Lundberg E 23. Hardman M, Makarov AA (2003) Interfacing
(2017) A subcellular map of the human prote- the orbitrap mass analyzer to an electrospray
ome. Science 356(6340):eaal3321. https:// ion source. Anal Chem 75(7):1699–1705
doi.org/10.1126/science.aal3321
24. Krutchinsky AN, Kalkum M, Chait BT (2001)
16. Espina V, Edmiston KH, Heiby M, Automatic identification of proteins with a
Pierobon M, Sciro M, Merritt B, Banks S, MALDI-quadrupole ion trap mass spectrome-
Deng J, VanMeter AJ, Geho DH, Pastore L, ter. Anal Chem 73(21):5066–5077
Sennesh J, Petricoin EF, Liotta LA (2008) A
portrait of tissue phosphoprotein stability in 25. Johnson RS, Martin SA, Biemann K, Stults JT,
the clinical tissue procurement process. Mol Watson JT (1987) Novel fragmentation pro-
Cell Proteomics 7(10):1998–2018. https:// cess of peptides by collision-induced decompo-
doi.org/10.1074/mcp.M700596-MCP200 sition in a tandem mass spectrometer:
differentiation of leucine and isoleucine. Anal
17. Aebersold R, Mann M (2016) Mass- Chem 59(21):2621–2625
spectrometric exploration of proteome struc-
ture and function. Nature 537 26. Hughes C, Ma B, Lajoie GA (2010) De novo
(7620):347–355. https://doi.org/10.1038/ sequencing methods in proteomics. Methods
nature19949 Mol Biol 604:105–121. https://doi.org/10.
1007/978-1-60761-444-9_8
18. Hu Q, Noll RJ, Li H, Makarov A, Hardman M,
Graham Cooks R (2005) The Orbitrap: a new 27. Johnson RS, Taylor JA (2002) Searching
mass spectrometer. J Mass Spectrom 40 sequence databases via de novo peptide
(4):430–443. https://doi.org/10.1002/jms. sequencing by tandem mass spectrometry.
856 Mol Biotechnol 22(3):301–315. https://doi.
org/10.1385/MB:22:3:301
19. Collins BC, Hunter CL, Liu Y, Schilling B,
Rosenberger G, Bader SL, Chan DW, Gibson 28. Nesvizhskii AI, Vitek O, Aebersold R (2007)
BW, Gingras AC, Held JM, Hirayama-Kurogi- Analysis and validation of proteomic data gen-
M, Hou G, Krisp C, Larsen B, Lin L, Liu S, erated by tandem mass spectrometry. Nat
Molloy MP, Moritz RL, Ohtsuki S, Methods 4(10):787–797. https://doi.org/
Schlapbach R, Selevsek N, Thomas SN, Tzeng 10.1038/nmeth1088
SC, Zhang H, Aebersold R (2017) Multi- 29. Sadygov RG, Cociorva D, Yates JR (2004)
laboratory assessment of reproducibility, quali- Large-scale database searching using tandem
tative and quantitative performance of mass spectra: looking up the answer in the
SWATH-mass spectrometry. Nat Commun 8 back of the book. Nat Methods 1
(1):291. https://doi.org/10.1038/s41467- (3):195–202. https://doi.org/10.1038/
017-00249-5 nmeth725
20. Michalski A, Damoc E, Lange O, Denisov E, 30. Kusebauch U, Campbell DS, Deutsch EW,
Nolting D, Müller M, Viner R, Schwartz J, Chu CS, Spicer DA, Brusniak MY, Slagel J,
Remes P, Belford M, Dunyach JJ, Cox J, Sun Z, Stevens J, Grimes B, Shteynberg D,
Horning S, Mann M, Makarov A (2012) Hoopmann MR, Blattmann P, Ratushny AV,
Ultra high resolution linear ion trap Orbitrap Rinner O, Picotti P, Carapito C, Huang CY,
mass spectrometer (Orbitrap Elite) facilitates Kapousouz M, Lam H, Tran T, Demir E,
 MS-Based Proteomics 193

Aitchison JD, Sander C, Hood L, Aebersold R, an LTQ Orbitrap mass spectrometer. Mol Cell
Moritz RL (2016) Human SRMAtlas: a Proteomics 7(9):1702–1713. https://doi.
resource of targeted assays to quantify the com- org/10.1074/mcp.M800029-MCP200
plete human proteome. Cell 166(3):766–778. 40. Garbis SD, Roumeliotis TI, Tyritzis SI, Zorpas
https://doi.org/10.1016/j.cell.2016.06.041 KM, Pavlakis K, Constantinides CA (2011) A
31. Picotti P, Clément-Ziza M, Lam H, Campbell novel multidimensional protein identification
DS, Schmidt A, Deutsch EW, Röst H, Sun Z, technology approach combining protein size
Rinner O, Reiter L, Shen Q, Michaelson JJ, exclusion prefractionation, peptide zwitterion-
Frei A, Alberti S, Kusebauch U, Wollscheid B, ion hydrophilic interaction chromatography,
Moritz RL, Beyer A, Aebersold R (2013) A and nano-ultraperformance RP
complete mass-spectrometric map of the yeast chromatography/nESI-MS2 for the in-depth
proteome applied to quantitative trait analysis. analysis of the serum proteome and phospho-
Nature 494(7436):266–270. https://doi.org/ proteome: application to clinical sera derived
10.1038/nature11835 from humans with benign prostate hyperplasia.
32. Dasari S, Chambers MC, Martinez MA, Car- Anal Chem 83(3):708–718. https://doi.org/
penter KL, Ham AJ, Vega-Montoto LJ, Tabb 10.1021/ac102075d
DL (2012) Pepitome: evaluating improved 41. Michalski A, Cox J, Mann M (2011) More than
spectral library search for identification com- 100,000 detectable peptide species elute in sin-
plementarity and quality assessment. J Prote- gle shotgun proteomics runs but the majority is
ome Res 11(3):1686–1695. https://doi.org/ inaccessible to data-dependent LC-MS/MS. J
10.1021/pr200874e Proteome Res 10(4):1785–1793. https://doi.
33. Lam H (2011) Building and searching tandem org/10.1021/pr101060v
mass spectral libraries for peptide identifica- 42. Panchaud A, Scherl A, Shaffer S, Haller P,
tion. Mol Cell Proteomics 10(12): Kulasekara H, Miller SI, Goodlett DR (2009)
R111.008565. https://doi.org/10.1074/ Precursor acquisition independent from ion
mcp.R111.008565 count: how to dive deeper into the proteomics
34. Jmeian Y, El Rassi Z (2009) Liquid-phase- ocean. Anal Chem 81:6481–6488
based separation systems for depletion, prefrac- 43. Scherl A, Shaffer SA, Taylor GK, Kulasekara
tionation and enrichment of proteins in HD, Miller SI, Goodlett DR (2008)
biological fluids for in-depth proteomics analy- Genome-specific gas-phase fractionation strat-
sis. Electrophoresis 30(1):249–261. https:// egy for improved shotgun proteomic profiling
doi.org/10.1002/elps.200800639 of proteotypic peptides. Anal Chem 80
35. Boisvert FM, Lam YW, Lamont D, Lamond AI (4):1182–1191. https://doi.org/10.1021/
(2010) A quantitative proteomics analysis of ac701680f
subcellular proteome localization and changes 44. Panchaud A, Jung S, Shaffer SA, Aitchison JD,
induced by DNA damage. Mol Cell Proteomics Goodlett DR (2011) Faster, quantitative, and
9(3):457–470. https://doi.org/10.1074/ accurate precursor acquisition independent
mcp.M900429-MCP200 from ion count. Anal Chem 83
36. Chenau J, Michelland S, Sidibe J, Seve M (6):2250–2257. https://doi.org/10.1021/
(2008) Peptides OFFGEL electrophoresis: a ac103079q
suitable pre-analytical step for complex eukary- 45. Savaryn JP, Catherman AD, Thomas PM, Abe-
otic samples fractionation compatible with cassis MM, Kelleher NL (2013) The emer-
quantitative iTRAQ labeling. Proteome Sci gence of top-down proteomics in clinical
6:9. https://doi.org/10.1186/1477-5956-6- research. Genome Med 5(6):53. https://doi.
9 org/10.1186/gm457
37. Batth TS, Olsen JV (2016) Offline high pH 46. Tran JC, Zamdborg L, Ahlf DR, Lee JE,
reversed-phase peptide fractionation for deep Catherman AD, Durbin KR, Tipton JD,
phosphoproteome coverage. Methods Mol Vellaichamy A, Kellie JF, Li M, Wu C, Sweet
Biol 1355:179–192. https://doi.org/10. SM, Early BP, Siuti N, LeDuc RD, Compton
1007/978-1-4939-3049-4_12 PD, Thomas PM, Kelleher NL (2011)
38. Boersema PJ, Mohammed S, Heck AJ (2008) Mapping intact protein isoforms in discovery
Hydrophilic interaction liquid chromatogra- mode using top-down proteomics. Nature 480
phy (HILIC) in proteomics. Anal Bioanal (7376):254–258. https://doi.org/10.1038/
Chem 391(1):151–159. https://doi.org/10. nature10575
1007/s00216-008-1865-7 47. Catherman AD, Durbin KR, Ahlf DR, Early
39. Bantscheff M, Boesche M, Eberhard D, BP, Fellers RT, Tran JC, Thomas PM, Kelleher
Matthieson T, Sweetman G, Kuster B (2008) NL (2013) Large-scale top-down proteomics
Robust and sensitive iTRAQ quantification on of the human proteome: membrane proteins,
194 J. Robert O’Neill

mitochondria, and senescence. Mol Cell Prote- (11):1130–1136. https://doi.org/10.1038/

omics 12(12):3465–3473. https://doi.org/ nbt.3685
10.1074/mcp.M113.030114 56. Röst HL, Aebersold R, Schubert OT (2017)
48. Fornelli L, Durbin KR, Fellers RT, Early BP, Automated SWATH data analysis using tar-
Greer JB, LeDuc RD, Compton PD, Kelleher geted extraction of ion chromatograms. Meth-
NL (2017) Advancing top-down analysis of the ods Mol Biol 1550:289–307. https://doi.org/
human proteome using a Benchtop 10.1007/978-1-4939-6747-6_20
Quadrupole-Orbitrap mass spectrometer. J 57. Geiger T, Velic A, Macek B, Lundberg E,
Proteome Res 16(2):609–618. https://doi. Kampf C, Nagaraj N, Uhlen M, Cox J, Mann
org/10.1021/acs.jproteome.6b00698 M (2013) Initial quantitative proteomic map of
49. Fornelli L, Toby TK, Schachner LF, Doubleday 28 mouse tissues using the SILAC mouse. Mol
PF, Srzentić K, DeHart CJ, Kelleher NL Cell Proteomics 12(6):1709–1722. https://
(2017) Top-down proteomics: where we are, doi.org/10.1074/mcp.M112.024919
where we are going? J Proteome 175:3. 58. Uhlén M, Fagerberg L, Hallström BM,
https://doi.org/10.1016/j.jprot.2017.02. Lindskog C, Oksvold P, Mardinoglu A, Siverts-
002 son Å, Kampf C, Sjöstedt E, Asplund A,
50. Cai W, Guner H, Gregorich ZR, Chen AJ, Olsson I, Edlund K, Lundberg E, Navani S,
Ayaz-Guner S, Peng Y, Valeja SG, Liu X, Ge Szigyarto CA, Odeberg J, Djureinovic D,
Y (2016) MASH suite pro: a comprehensive Takanen JO, Hober S, Alm T, Edqvist PH,
software tool for top-down proteomics. Mol Berling H, Tegel H, Mulder J, Rockberg J,
Cell Proteomics 15(2):703–714. https://doi. Nilsson P, Schwenk JM, Hamsten M, von
org/10.1074/mcp.O115.054387 Feilitzen K, Forsberg M, Persson L,
51. Desiere F, Deutsch EW, King NL, Nesvizhskii Johansson F, Zwahlen M, von Heijne G,
AI, Mallick P, Eng J, Chen S, Eddes J, Loeve- Nielsen J, Pontén F (2015) Proteomics.
nich SN, Aebersold R (2006) The PeptideAtlas Tissue-based map of the human proteome. Sci-
project. Nucleic Acids Res 34(Database issue): ence 347(6220):1260419. https://doi.org/
D655–D658. https://doi.org/10.1093/nar/ 10.1126/science.1260419
gkj040 59. Görg A, Weiss W, Dunn MJ (2004) Current
52. Zolg DP, Wilhelm M, Schnatbaum K, two-dimensional electrophoresis technology
Zerweck J, Knaute T, Delanghe B, Bailey DJ, for proteomics. Proteomics 4
Gessulat S, Ehrlich HC, Weininger M, Yu P, (12):3665–3685. https://doi.org/10.1002/
Schlegl J, Kramer K, Schmidt T, Kusebauch U, pmic.200401031
Deutsch EW, Aebersold R, Moritz RL, 60. Granier F, de Vienne D (1986) Silver staining
Wenschuh H, Moehring T, Aiche S, of proteins: standardized procedure for
Huhmer A, Reimer U, Kuster B (2017) Build- two-dimensional gels bound to polyester
ing ProteomeTools based on a complete syn- sheets. Anal Biochem 155(1):45–50
thetic human proteome. Nat Methods 14 61. Dowsey AW, Dunn MJ, Yang GZ (2003) The
(3):259–262. https://doi.org/10.1038/ role of bioinformatics in two-dimensional gel
nmeth.4153 electrophoresis. Proteomics 3(8):1567–1596.
53. Mallick P, Schirle M, Chen SS, Flory MR, https://doi.org/10.1002/pmic.200300459
Lee H, Martin D, Ranish J, Raught B, 62. Altelaar AF, Frese CK, Preisinger C, Hennrich
Schmitt R, Werner T, Kuster B, Aebersold R ML, Schram AW, Timmers HT, Heck AJ,
(2007) Computational prediction of proteoty- Mohammed S (2013) Benchmarking stable
pic peptides for quantitative proteomics. Nat isotope labeling based quantitative proteomics.
Biotechnol 25(1):125–131. https://doi.org/ J Proteome 88:14–26. https://doi.org/10.
10.1038/nbt1275 1016/j.jprot.2012.10.009
54. Ebhardt HA, Root A, Sander C, Aebersold R 63. Ong SE, Blagoev B, Kratchmarova I, Kristen-
(2015) Applications of targeted proteomics in sen DB, Steen H, Pandey A, Mann M (2002)
systems biology and translational medicine. Stable isotope labeling by amino acids in cell
Proteomics 15(18):3193–3208. https://doi. culture, SILAC, as a simple and accurate
org/10.1002/pmic.201500004 approach to expression proteomics. Mol Cell
55. Navarro P, Kuharev J, Gillet LC, Bernhardt Proteomics 1(5):376–386
OM, MacLean B, Röst HL, Tate SA, Tsou 64. Krüger M, Moser M, Ussar S, Thievessen I,
CC, Reiter L, Distler U, Rosenberger G, Luber CA, Forner F, Schmidt S, Zanivan S,
Perez-Riverol Y, Nesvizhskii AI, Aebersold R, F€assler R, Mann M (2008) SILAC mouse for
Tenzer S (2016) A multicenter study bench- quantitative proteomics uncovers kindlin-3 as
marks software tools for label-free proteome an essential factor for red blood cell function.
quantification. Nat Biotechnol 34
 MS-Based Proteomics 195

Cell 134(2):353–364. https://doi.org/10. Saccharomyces cerevisiae using amine-reactive

1016/j.cell.2008.05.033 isobaric tagging reagents. Mol Cell Proteomics
65. Sury MD, Chen JX, Selbach M (2010) The 3(12):1154–1169. https://doi.org/10.1074/
SILAC fly allows for accurate protein quantifi- mcp.M400129-MCP200
cation in vivo. Mol Cell Proteomics 9 75. Thompson A, Sch€afer J, Kuhn K, Kienle S,
(10):2173–2183. https://doi.org/10.1074/ Schwarz J, Schmidt G, Neumann T,
mcp.M110.000323 Johnstone R, Mohammed AK, Hamon C
66. Larance M, Bailly AP, Pourkarimi E, Hay RT, (2003) Tandem mass tags: a novel quantifica-
Buchanan G, Coulthurst S, Xirodimas DP, tion strategy for comparative analysis of com-
Gartner A, Lamond AI (2011) Stable-isotope plex protein mixtures by MS/MS. Anal Chem
labeling with amino acids in nematodes. Nat 75(8):1895–1904
Methods 8(10):849–851. https://doi.org/10. 76. Bouchal P, Roumeliotis T, Hrstka R,
1038/nmeth.1679 Nenutil R, Vojtesek B, Garbis SD (2009) Bio-
67. Geiger T, Cox J, Ostasiewicz P, Wisniewski JR, marker discovery in low-grade breast cancer
Mann M (2010) Super-SILAC mix for quanti- using isobaric stable isotope tags and
tative proteomics of human tumor tissue. Nat two-dimensional liquid chromatography-
Methods 7(5):383–385. https://doi.org/10. tandem mass spectrometry (iTRAQ-2DLC-
1038/nmeth.1446 MS/MS) based quantitative proteomic analy-
68. Geiger T, Wehner A, Schaab C, Cox J, Mann M sis. J Proteome Res 8(1):362–373. https://doi.
(2012) Comparative proteomic analysis of org/10.1021/pr800622b
eleven common cell lines reveals ubiquitous 77. Luk VN, Wheeler AR (2009) A digital micro-
but varying expression of most proteins. Mol fluidic approach to proteomic sample proces-
Cell Proteomics 11(3):M111.014050. sing. Anal Chem 81(11):4524–4530. https://
https://doi.org/10.1074/mcp.M111. doi.org/10.1021/ac900522a
014050 78. Rai AJ, Gelfand CA, Haywood BC, Warunek
69. Gygi SP, Rist B, Gerber SA, Turecek F, Gelb DJ, Yi J, Schuchard MD, Mehigh RJ, Cockrill
MH, Aebersold R (1999) Quantitative analysis SL, Scott GB, Tammen H, Schulz-Knappe P,
of complex protein mixtures using isotope- Speicher DW, Vitzthum F, Haab BB, Siest G,
coded affinity tags. Nat Biotechnol 17 Chan DW (2005) HUPO Plasma Proteome
(10):994–999. https://doi.org/10.1038/ Project specimen collection and handling:
13690 towards the standardization of parameters for
70. Zhou H, Ranish JA, Watts JD, Aebersold R plasma proteome samples. Proteomics 5
(2002) Quantitative proteome analysis by (13):3262–3277. https://doi.org/10.1002/
solid-phase isotope tagging and mass spec- pmic.200401245
trometry. Nat Biotechnol 20(5):512–515. 79. Lundgren DH, Hwang SI, Wu L, Han DK
https://doi.org/10.1038/nbt0502-512 (2010) Role of spectral counting in quantita-
71. Zhang R, Sioma CS, Wang S, Regnier FE tive proteomics. Expert Rev Proteomics 7
(2001) Fractionation of isotopically labeled (1):39–53. https://doi.org/10.1586/epr.09.
peptides in quantitative proteomics. Anal 69
Chem 73(21):5142–5149 80. Carvalho PC, Hewel J, Barbosa VC, Yates JR
72. Antonov VK, Ginodman LM, Rumsh LD, (2008) Identifying differences in protein
Kapitannikov YV, Barshevskaya TN, Yavashev expression levels by spectral counting and fea-
LP, Gurova AG, Volkova LI (1981) Studies on ture selection. Genet Mol Res 7(2):342–356
the mechanisms of action of proteolytic 81. Griffin NM, Yu J, Long F, Oh P, Shore S, Li Y,
enzymes using heavy oxygen exchange. Eur J Koziol JA, Schnitzer JE (2010) Label-free,
Biochem 117(1):195–200 normalized quantification of complex mass
73. Hsu JL, Huang SY, Chow NH, Chen SH spectrometry data for proteomic analysis. Nat
(2003) Stable-isotope dimethyl labeling for Biotechnol 28(1):83–89. https://doi.org/10.
quantitative proteomics. Anal Chem 75 1038/nbt.1592
(24):6843–6852. https://doi.org/10.1021/ 82. Li Z, Adams RM, Chourey K, Hurst GB, Het-
ac0348625 tich RL, Pan C (2012) Systematic comparison
74. Ross PL, Huang YN, Marchese JN, of label-free, metabolic labeling, and isobaric
Williamson B, Parker K, Hattan S, chemical labeling for quantitative proteomics
Khainovski N, Pillai S, Dey S, Daniels S, on LTQ Orbitrap Velos. J Proteome Res 11
Purkayastha S, Juhasz P, Martin S, Bartlet- (3):1582–1590. https://doi.org/10.1021/
Jones M, He F, Jacobson A, Pappin DJ pr200748h
(2004) Multiplexed protein quantitation in 83. Chelius D, Bondarenko PV (2002) Quantita-
tive profiling of proteins in complex mixtures
196 J. Robert O’Neill

using liquid chromatography and mass spec- of peptides and proteins using MALDI-TOF
trometry. J Proteome Res 1(4):317–323 MS. Anal Chem 69(23):4751–4760
84. Listgarten J, Emili A (2005) Statistical and 88. Aichler M, Walch A (2015) MALDI Imaging
computational methods for comparative pro- mass spectrometry: current frontiers and per-
teomic profiling using liquid chromatography- spectives in pathology research and practice.
tandem mass spectrometry. Mol Cell Proteo- Lab Investig 95(4):422–431. https://doi.
mics 4(4):419–434. https://doi.org/10. org/10.1038/labinvest.2014.156
1074/mcp.R500005-MCP200 89. Minerva L, Clerens S, Baggerman G, Arckens L
85. Hanahan D, Weinberg RA (2011) Hallmarks (2008) Direct profiling and identification of
of cancer: the next generation. Cell 144 peptide expression differences in the pancreas
(5):646–674. https://doi.org/10.1016/j.cell. of control and ob/ob mice by imaging mass
2011.02.013 spectrometry. Proteomics 8(18):3763–3774.
86. Gerlinger M, Rowan AJ, Horswell S, Larkin J, https://doi.org/10.1002/pmic.200800237
Endesfelder D, Gronroos E, Martinez P, 90. Balluff B, Schöne C, Höfler H, Walch A (2011)
Matthews N, Stewart A, Tarpey P, Varela I, MALDI imaging mass spectrometry for direct
Phillimore B, Begum S, McDonald NQ, tissue analysis: technological advancements and
Butler A, Jones D, Raine K, Latimer C, Santos recent applications. Histochem Cell Biol 136
CR, Nohadani M, Eklund AC, Spencer- (3):227–244. https://doi.org/10.1007/
Dene B, Clark G, Pickering L, Stamp G, s00418-011-0843-x
Gore M, Szallasi Z, Downward J, Futreal PA, 91. Perkins DN, Pappin DJ, Creasy DM, Cottrell
Swanton C (2012) Intratumor heterogeneity JS (1999) Probability-based protein identifica-
and branched evolution revealed by multire- tion by searching sequence databases using
gion sequencing. N Engl J Med 366 mass spectrometry data. Electrophoresis 20
(10):883–892. https://doi.org/10.1056/ (18):3551–3567. https://doi.org/10.1002/(
NEJMoa1113205 SICI)1522-2683(19991201)20:18<3551::
87. Caprioli RM, Farmer TB, Gile J (1997) Molec- AID-ELPS3551>3.0.CO;2-2
ular imaging of biological samples: localization
 Chapter 14

Functional Proteomic Analysis to Characterize Signaling

Crosstalk
Sneha M. Pinto, Yashwanth Subbannayya, and T. S. Keshava Prasad

Abstract
The biological activities of a cell are determined by its response to external stimuli. The signals are
transduced from either intracellular or extracellular milieu through networks of multi-protein complexes
and post-translational modifications of proteins (PTMs). Most PTMs including phosphorylation, acetyla-
tion, ubiquitination, and SUMOylation, among others, modulate activities of proteins and regulate
biological processes such as proliferation, differentiation, as well as host pathogen interaction. Convention-
ally, reverse genetics analysis and single molecule-based studies were employed to identify and characterize
the function of PTMs and enzyme-substrate networks regulated by them. With the advent of high-
throughput technologies, it is now possible to identify and quantify thousands of PTM sites in a single
experiment. Here, we discuss recent advances in enrichment strategies of various PTMs. We also describe a
method for the identification and relative quantitation of proteins using a tandem mass tag labeling
approach combined with serial enrichment of phosphorylation, acetylation and succinylation using anti-
body enrichment strategy.

Key words Signaling pathways, Mass spectrometry, Phosphorylation, Crosstalk, Quantitation

1 Introduction

The biological activities of a cell are governed by various cellular

and physiological processes. These processes are mediated by pro-
teins, the functional units that either act in toto or in concert with
other proteins. Distinct post-translational modifications (PTMs)
dynamically regulate protein function and thereby are involved in
driving a majority of cellular processes. These include changes
mediated by phosphorylation and activation of downstream sub-
strates, ubiquitination or SUMOylation of target proteins resulting
in their degradation, subcellular trafficking or regulation of tran-
scriptional mechanisms. In most biological processes, the relay of
information is mediated by multiple protein PTMs. The crosstalk
provides cues to decide the fate of the proteins and consequently
govern cellular activity. In order to understand cellular

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_14, © Springer Science+Business Media, LLC, part of Springer Nature 2019

197
198 Sneha M. Pinto et al.

mechanisms, it is imperative to identify and characterize PTMs,

thereby understanding its role in health and disease.
It is estimated that there are over 200 types of post-translation
modifications and 300 types of chemical and biological modifica-
tions [1, 2]. These PTMs are mediated by enzymes including
kinases, phosphatases, transferases, and ligases under the influence
of various stimuli. Conventionally, PTMs, especially phosphoryla-
tion, have been identified using in vitro kinase assays or site-
directed mutagenesis-based assays. In case of ubiquitination or
acetylation modifications, the site of modification and its effect
have been studied depending on the availability of the antibodies
against the protein of interest using immunoblotting approaches.
In recent years, the utility of protein/peptide array chip for the
detection of post-translational modifications and PTM-dependent
interactions has been on the rise [3–5]. However, a vast majority of
these studies have focused on methods for interrogating protein
phosphorylation-mediated signaling systems in various diseases
([6, 7], reviewed in [8]). The advent of mass spectrometry tech-
nique has not only enabled identification and quantification of
proteins in a high-throughput fashion but has rapidly become
popular for the discovery of PTMs, thereby revolutionizing the
current understanding of signaling pathways and its role in regulat-
ing cellular processes. With the unveiling of the two drafts of the
human proteome [9, 10], the PTMome represents the next frontier
of grand challenges. In the forthcoming sections, we will briefly
describe advances made in the strategies for PTM enrichment, mass
spectrometry methods, data analysis, and data-mining approaches.

1.1 Strategies Identification and characterization of PTMs and enzyme-substrate

to Enrich Post- networks regulated by them were conventionally studied using
translational reverse genetics or by single molecule-based studies. With the
Modifications advent of high-throughput technologies, it is now possible to
identify and quantify thousands of PTM sites in a single experi-
ment. However, the low stoichiometry of PTM-modified peptides
necessitates the use of enrichment strategies to enable their detec-
tion in biological samples. Toward this end, substantial progress has
been made toward the development and optimization of enrich-
ment and fractionation strategies for global studies on PTM
(reviewed in [11]). Specific enrichment strategies have primarily
been developed based on the properties of affinity chromatography
either by utilizing the ionic charge properties of the modifying
group or based on antibody recognition. The enrichment strategies
are applied either at the protein level or at the peptide level.
Peptide-based enrichment strategies are by far the more common
and amenable method. A brief description of the various enrich-
ment strategies has been provided.
 Functional Proteomic Analysis to Characterize Signaling Crosstalk 199

1.1.1 Antibody-Based One of the widely employed methodology to assay PTMs of inter-
Enrichment methods est is the use of antibody-based enrichment strategies coupled with
LC-MS/MS analysis. The most common as well as the most popu-
lar PTM assayed includes the use of antibodies that specifically
enrich for tyrosine [12–14] or serine/threonine phosphorylated
proteins [15]. A detailed review of this topic has been provided
by Harsha et al. [16]. Antibody-based enrichment strategies have
now been extended to profile and quantify other PTMs including
ubiquitination, SUMOylation, lysine modifications such as acetyla-
tion, succinylation, and malonlylation, among others. Further-
more, Cell Signaling Technologies and PTM Biolabs have also
developed antibodies targeting methylation, crotonylation [17],
and modifications on lysine and arginine residues [18]. In addition
to pan-PTM-specific antibodies, antibodies specific to PTM motifs
have also been employed to identify and quantify substrates [19].

1.1.2 Metal-Ion-Based The alternative approach for enrichment of PTMs especially phos-
Enrichment Strategies phorylation moieties are chromatographic methods based on metal
affinity enrichment referred as immobilized metal affinity chroma-
tography (IMAC) and titanium dioxide (TiO2) based enrichment
[20]. Metal ions such as Fe(3+), Zr(4+), and Ga(3+) are employed
in IMAC, which is immobilized on beads and covalently bound to
negatively charged phosphopeptides [21–23]. TiO2 also works on a
similar principle and both approaches are employed in combination
with chromatography-based separation techniques such as strong
cation exchange chromatography and high pH reverse phase chro-
matography [14, 16, 24]. Sequential enrichment using IMAC
followed by TiO2 has also been reported to predominantly enrich
for monophosphorylated peptides [25].

1.1.3 Chemical With advances in chemical biology, PTM enrichment using chemi-
Proteomics cal tagging or derivatization has enabled detection and enrichment
of several PTMs including N-linked and O-linked glycosylation
[26, 27], acylation [28], palmitoylation [29, 30], and S-nitrosyla-
tion. The tagging methods include both in vitro and in vivo label-
ing of small chemical tag at the site of the modification and
subsequent enrichment using a capture reagent such as biotin. In
case of enrichment and separation of N-linked glycopeptides,
boronic acid chemistry, hydrazide chemistry, and alkyne click
chemistry, among others, have been employed [31]. A detailed
description of advances in chemical proteomics to study the
PTMome has been described by Tate [32].

1.1.4 Serial Enrichment It is known that proteins can be simultaneously modified by multi-
of PTMs ple PTMs and these PTMs may form the basis of regulatory cross
talk events [33]. Therefore, it is ideal to simultaneously study
numerous PTMs in a single experiment. This can be achieved by
carrying out PTM enrichment in a serial manner. Serial enrichment
200 Sneha M. Pinto et al.

technique has been explored for the enrichment of phosphoryla-

tion and ubiquitylation in Saccharomyces cerevisiae to understand
cross talk mechanisms [34]. Mertins and colleagues have also
employed serial enrichments of different posttranslational modifi-
cations (SEPTM) and extensively studied the changes in the protein
expression, phosphorylation, acetylation, and ubiquitination events
in the same sample [35].

1.2 Mass Over the years, mass spectrometry-based proteomic analysis has
Spectrometry to considerably increased our understanding of the occurrence and
Characterize Multiple dynamics of protein PTMs. Following PTM enrichment, sequence
Post-translational identification and confident site localization is vital. In case of
Modifications bottom-up approach which involves proteolytic digestion of pro-
teins followed by enrichment of PTMs, information on MS2 frag-
mentation provides data that is being used to deduce the sequence
as well as identify the site of modification. The fragmentation
approaches include collision-induced dissociation (CID) and
higher energy collisional dissociation (HCD), both of which gen-
erate predominantly b- and y-type ions and are universally
employed for global protein profiling as well as for the analyses of
stable PTMs. In the CID mode, fragmentation and data collection
occurs in ion trap at a faster acquisition speed resulting in higher
number of data points. On the contrary, HCD mode of fragmenta-
tion results in Fourier transform detection of the MS2 fragments in
the Orbitrap mass analyzer with data being acquired at a higher
resolution albeit lower scan rate. In case of labile PTMs such as
acetylation and glycosylation, electron transfer dissociation (ETD)
has been described as a preferred method for data acquisition
[36]. The mass accuracy of mass spectrometers can affect the iden-
tification and accurate site localization. Mass accuracy can also
significantly impact distinction of isobaric/indistinguishable mass
PTMs when mass analyzers such as ion trap are employed rather
than high-resolution mass analyzers such as Orbitrap. It is therefore
essential to choose the right mass spectrometry method for data
acquisition. To maximize the detection of PTM-modified peptides,
a combination of fragmentation methods can also be employed.

1.3 Quantitative In addition to detection of PTMs on thousands of proteins in a

Proteomic Strategies single experiment, recent advances in mass spectrometry-based
to Assess Dynamic techniques also allow quantitation of the relative changes of modi-
States of Proteins fication in multiple biological conditions. Several quantitative pro-
teomic technologies are currently available to determine differences
in protein abundances across biological states including metabolic
labeling techniques such as SILAC [37] and chemical labeling
techniques such as iTRAQ [38] and TMT (Tandem Mass Tags)
[39]. These tools have enabled multiplexing and achieve quantita-
tion of PTMs across multiple conditions. In addition, the develop-
ment of label-free approaches such as iBAQ [40], NSAF [41],
 Functional Proteomic Analysis to Characterize Signaling Crosstalk 201

SWATH [42], and APEX [43] have also enabled quantitation of

PTMs without the need for chemical or metabolic labeling. Soft-
ware algorithms such as ptmRS [44] and A-Score [45] now allow
localization of modification sites, thereby enabling quantitation of
the extent of PTMs.

1.4 Studies on Mass A partial list of studies on mass spectrometry-based PTM analysis is
Spectrometry-Based provided in Table 1. For the purpose of this chapter, we have briefly
PTM Analysis described studies on phosphorylation, succinylation, acetylation,
ubiquitination, glycosylation, and palmitoylation.
Reversible protein phosphorylation at serine, threonine, and
tyrosine residues is one of the most common and widely studied
PTM of proteins. Other residues such as histidine, asparagine,
arginine, cysteine, and lysine are also known to be phosphorylated.
However, the extent of phosphorylation is relatively less known
owing to its occurrence in a low frequency or challenges in efficient
enrichment and identification. A previous study on human renal
cancer used a Ni-NTA-based enrichment method and identified
44,728 phosphosites on 6415 proteins [46]. Another study
expanded the understanding of the IL-33 signaling pathway by
identifying 7191 phosphorylation sites on 2746 proteins
[14]. Phosphorylation events have also been shown to be wide-
spread in prokaryotes using metal affinity-based enrichment
techniques [24].
Succinylation of lysine was identified as a new posttranslational
modification a few years ago [47]. Succinylation has been observed
to occur widely across both prokaryotes and eukaryotes
[48]. Enrichment of succinylated peptides has been typically carried
out using antibody-based affinity enrichment [49]. These enriched
peptides are then subjected to strong cation exchange chromatog-
raphy and subjected to LC-MS/MS.
Acetylation is known to regulate diverse cellular processes and
has several implications in the disease pathophysiology [50]. One of
the early studies surveyed acetylation sites in HeLa cells and mouse
liver mitochondria using an antibody-based enrichment approach
identified 388 sites [51]. Lysine acetylation has also been found to
be conserved in prokaryotes such as Escherichia coli [52] and Sac-
charomyces cerevisiae [53] and rat models [54] using antibody-
based enrichment methods. Quantitative proteomic techniques
such as TMT labeling are amenable to antibody-based acetyl lysine
peptide enrichment and have been applied to study lysine acetyla-
tion sites on perturbation [55]. In addition, technologies such as
SWATH are being increasingly used for quantitation of acetylated
proteins [56].
The ubiquitination PTM drives the localization, stability, and
activity of the protein by covalent attachment of the ubiquitin
moiety to cellular proteins [57]. Protein ubiquitination usually is
Table 1
202

Details of studies on mass spectrometry-based PTM analysis

Mass
Quantification Number of protein and spectrometer
PTM type Enrichment method method sites Model used type Reference
Phosphorylation Ni-NTA 44,728 phosphosites on Human renal cancer LTQ Orbitrap Peng et al. [46]
6415 proteins Velos
TiO2-based phosphopeptide TMT 512 phosphorylation sites Mycobacterium Orbitrap Verma et al.
enrichment derived from tuberculosis Fusion [24]
Sneha M. Pinto et al.

257 proteins Tribrid

TiO2-based phosphopeptide SILAC 7191 phosphorylation sites RAW264.7 cells LTQ- Pinto et al.
enrichment on 2746 proteins Orbitrap [14]
Elite
Succinylation Anti-succinyl lysine antibody SILAC 2004 lysine succinylation HeLa cells Q-Exactive Weinert et al.
sites on 738 proteins [48]
Anti-succinyl lysine antibody 1931 lysine succinylated Vibrio parahaemolyticus Q Exactive Pan et al. [77]
peptides from Plus
642 proteins
Anti-succinyl lysine antibody TMT 815 succinylation sites on Rat Q-Exactive Cheng et al.
407 proteins [78]
Acetylation Anti-acetyl lysine antibody SILAC 2803 lysine acetylation sites Escherichia coli LTQ Orbitrap Colak et al.
in 782 proteins Velos [79]
Anti-acetyl lysine antibody 1206 lysine acetylation sites Human LTQ- Sun et al. [80]
in 576 proteins Orbitrap
Anti-acetyl lysine antibody 1128 lysine acetylation sites Mycobacterium Q-Exactive Xie et al. [81]
in 658 proteins tuberculosis
Ubiquitination Anti-lysine ubiquitination SILAC 1067 ubiquitination sites A549 cells Q Exactive Wu et al. [58]
(Kub) antibody on 613 proteins Plus
Anti-lysine ubiquitination TMT 1248 ubiquitinated Mouse J1 and E14 Q Exactive Karg et al. [82]
antibody peptides embryonic stem cells HF
Anti-lysine ubiquitination SILAC 2299 ubiquitination sites Saccharomyces cerevisiae Q-Exactive Iesmantavicius
antibody et al. [59]
Glycosylation Lectin affinity 720N-glycosylated sites on Human seminal plasma LTQ Orbitrap Yang et al. [60]
chromatography 372 proteins Velos
Boronic acid-based chemical 816N-glycosylation sites on Saccharomyces cerevisiae LTQ Orbitrap Chen et al.
enrichment 332 proteins elite [62]
Hydrazide chemistry method Label-free 608 N- glycosylation sites Human Bronchial LTQ- Sudhir et al.
on 317 proteins epithelial cells Orbitrap [61]
XL
Palmitoylation 17-octadecynoic acid SILAC 338 proteins BW5147-derived mouse Orbitrap Velos Martin et al.
(17-ODYA)-based click T-cell hybridoma cells [65]
chemistry
Acyl-biotinyl exchange SILAC 280 proteins Primary CD4 + T cells, Orbitrap elite
Morrison et al.
chemistry Jurkat cells [29]
Acyl-biotinyl exchange 401 proteins Toxoplasma gondii Linear Caballero et al.
chemistry quadrupole [83]
ion trap
Functional Proteomic Analysis to Characterize Signaling Crosstalk
203
204 Sneha M. Pinto et al.

enriched using an antibody affinity-based approach. An antibody-

based ubiquitination enrichment approach followed by mass spec-
trometry identified 1067 ubiquitination sites on 613 proteins in
A549 lung cancer cells [58]. A similar approach identified 2299
ubiquitination sites in Saccharomyces cerevisiae, of which 581 sites
were found to be differentially regulated in response to
rapamycin [59].
Glycosylation, a PTM where a carbohydrate moiety is cova-
lently attached to the proteins, is one of the most abundant post-
translational modifications. Protein glycosylation events are
enriched either using lectin affinity chromatography [60] or hydra-
zide chemistry [61]. A study on human seminal plasma identified
720 N-glycosylated sites on 372 proteins using lectin affinity-based
enrichment. Hydrazide-based enrichment method identified
608 N- glycosylation sites on 317 proteins in human bronchial
epithelial cells. An enrichment method using boronic acid-based
chemical enrichment has also been described previously
[62]. Using a yeast model, this method was used to identify
816 N-glycosylation sites on 332 proteins.
Palmitoylation is one of the most common mechanisms of lipid
modifications besides prenylation and myristoylation. Palmitoyla-
tion is known to drive mechanisms of protein targeting toward the
membrane and subcellular protein trafficking [63]. Several studies
have used acyl-biotinyl exchange chemistry to enrich palmitoylated
proteins. A previous study used this technique to identify 280 pal-
mitoylated proteins in primary and Jurkat T cells [29]. Protein
S-palmitoylation has also been enriched by using 17-octadecynoic
acid (17-ODYA)-based click chemistry. 17-ODYA is metabolically
incorporated into palmitoylation machinery. This can be combined
with metabolic labeling techniques such as SILAC. The extent of
palmitoylation modification can be assessed using click chemistry
ligation to biotin-azide [64]. Using this method, global quantita-
tive profiling of dynamic protein palmitoylation was carried out in
mouse T-cell hybridoma cells [65].

1.5 Data-Mining Advances in PTM enrichment techniques and mass spectrometry

Approaches for PTM technologies have led to increasingly large amounts of studies on
Analysis various PTMs. Researchers are cataloging these data into compre-
hensive public databases accessible to the scientific community.
Several PTM-specific databases are currently available to researchers
for comparison and analysis, some which are provided in Table 2. A
few databases currently provide information pertaining to multiple
PTMs. These include PhosphositePlus (https://www.phosphosite.
org) [66], Human Protein Reference Database (HPRD) (http://
hprd.org) [67], dbPTM (http://dbptm.mbc.nctu.edu.tw/) [68],
and CPLM (Compendium of Protein Lysine Modifications)
(http://cplm.biocuckoo.org/) [69]. However, several databases
 Functional Proteomic Analysis to Characterize Signaling Crosstalk 205

Table 2
Databases pertaining to posttranslational modifications

Database PTM type Link Reference

CPLM Nε-lysine acetylation, ubiquitination, http://cplm. Liu et al. [69]
(Compendium of methylation, sumoylation, glycation, biocuckoo.org/
Protein Lysine butyrylation, crotonylation,
Modifications) malonylation, propionylation,
succinylation, phosphoglycerylation,
pupylation
dbGSH S-Glutathionylation http://csb.cse.yzu. Chen et al. [84]
edu.tw/dbGSH/
dbPTM 20 types of PTM http://dbptm.mbc. Huang et al.
nctu.eDU.tw/ [68]
dbSNO S-nitrosylation http://140.138. Chen et al. [85]
144.145/
~dbSNO/
DEPOD Dephosphorylation http://depod.bioss. Duan et al.
uni-freiburg.de/ [86]
Human Protein 27 types of PTM http://hprd.org/ Prasad et al.
Reference index_html [67]
Database (HPRD)
MYRbase Myristoylation http://mendel.imp. Maurer-Stroh
ac.at/myristate/ et al. [87]
myrbase/
O-GlycBase O- and C-glycosylation http://www.cbs. Gupta et al.
dtu.dk/ [88]
databases/
OGLYCBASE/
PHOSIDA phosphorylation, acetylation, and http://141.61.102. Gnad et al.
N-glycosylation 18/phosida/ [89]
index.aspx
Phospho.ELM Phosphorylation http://phospho. Dinkel et al.
elm.eu.org/ [90]
PhosphositePlus Acetylation, Caspase cleavage, https://www. Hornbeck et al.
Di-methylation, Methylation, Mono- phosphosite.org/ [66]
methylation, O-Galnac, O-Glcnac, homeAction.
Phospho-Ser, Phospho-Thr, Phospho- action
Tyr, Succinylation, Sumoylation,
Tri-methylation, Ubiquitylation
PRENbase Farnesylation, Geranylgeranylation http://mendel.imp. Maurer-Stroh
ac.at/PrePS/ et al. [91]
PRENbase/
PupDB Pupylation http://cwtung. Tung C.
kmu.edu.tw/ W [92].
pupdb/
(continued)
206 Sneha M. Pinto et al.

Table 2
(continued)

Database PTM type Link Reference

SNObase S-nitrosylation http://www. Zhang et al.
nitrosation.org/ [93]
index.html
Succinsite Succinylation http://systbio.cau. Hasan et al.
edu.cn/ [94]
SuccinSite/
UbiProt Ubiquitination http://ubiprot.org. Chernorudskiy
ru/ et al. [95]

specialize in one type of PTM and exclusively provide information

pertaining to these. PTM-specific databases will enable the growth
and development of specific assays for multiple reaction monitoring
of modified peptides. Technological innovations such as IMAC-
MRM (Immobilized Metal Affinity Chromatography coupled to
Multiple Reaction Monitoring) now enable measurement of mod-
ified peptide levels [70]. Given that mass spectrometry and enrich-
ment technologies are becoming increasingly accessible to
researchers, the number of specialized PTM databases is expected
to grow phenomenally.
Several data-mining approaches are also being developed to
carry out analysis of mass spectrometry-derived PTM data. These
methods have enabled the measurement of PTM abundances at the
basal level. Several unassigned spectra in proteomics datasets have
been attributed to the presence of unknown or unexpected post-
translational modifications. Data from high-resolution mass spec-
trometers may be searched for potential post-translational
modifications using wide-tolerance searches. One such approach
using a large precursor ion tolerant search identified several post-
translational modifications including phosphorylation, N-terminal
acetylation, glycosylation, and monomethylation demethylation
and rarer PTM forms including glycerol phosphorylethanolamine
(GPE) and glutamylation in HEK293 cells [71]. This approach can
be potentially used to mine for PTM data from existing proteomic
datasets.

1.6 PTM Cross Talk PTM profiling studies have shown that numerous proteins can be
in Biological Systems modified by multiple PTMs, suggesting the possibility of PTM
crosstalk. Several examples of complex PTM crosstalk events reg-
ulating biological processes exist in the literature. An example of a
complex crosstalk involving PKCδ, Caspase-3, and p53 through
multiple PTMs is known to regulate various biological functions
 Functional Proteomic Analysis to Characterize Signaling Crosstalk 207

[72]. Co-occurrence of PTM sites has been used to measure PTM

co-evolution across eukaryotes [73]. Minguez and colleagues col-
lected data for 13 PTMs across 8 eukaryotes and derived a global
network of functionally associated PTMs.
Complex PTM cross talk mechanisms are not fully understood
due to nonavailability of enrichment methods for most types of
PTMs. To date, ~200 PTM crosstalk pairs in 77 human proteins
have been reported in the literature [13]. Recent advances in
affinity enrichment of PTMs have enabled the analysis of a few of
the multiple PTMs in a single sample. To a certain extent, PTM
crosstalk events can be identified using serial enrichment techni-
ques [33]. A previous study in murine synaptosomes identified
crosstalk between O-GlcNAcylation and phosphorylation events
using a mass spectrometry-based method [74]. The serial use of
multiple enrichment techniques coupled with high-resolution mass
spectrometry will help explore the dynamics of cross talk between
various PTMs will help us in a better understanding of the cellular
processes and underlying mechanisms of disease pathophysiology.

1.7 Challenges in As is evidenced by several studies, studying protein PTMs is vital as

Multi-PTM Analysis it provides a strong foundation for a better understanding of cellu-
lar networks and mechanisms of cell regulation. However, there are
several challenges when it comes to identification and characteriza-
tion of PTMs including sub-stoichiometric levels of expression,
high dynamic range as protein modification is a reversible process.
Furthermore, the existence of multiple PTM on the same site or
same modifications on different sites can bring about further com-
plexity. Lastly, suppression of modified peptide signal by
non-modified peptides at MS level can severely deter their identifi-
cation. This necessitates strategies that will enable efficient identifi-
cation of the PTM proteome. Advance in enrichment techniques
such as chemical methods, affinity enrichment, and chro-
matographic approaches together with improvements in
MS-based detection has dramaticalyly influenced comprehensive
characterization of PTM repertoire. PTMs can be erroneously
assigned in the mass spectrometry-based analysis due to several
factors discussed in detail by Kim et al. [75]. Some of these factors
may be attributed to isobaric PTMs and chemical modifications,
amino acid substitutions, SNPs, shared peptides, and poor MS/MS
fragmentation. Most of these errors can be resolved with counter-
measures including use of high-resolution analyzers such as Orbi-
trap, use of multiple proteolytic enzymes, and employment of
statistical measures for site localization.
208 Sneha M. Pinto et al.

2 Materials

2.1 Protein 1. SDS-Lysis buffer: 2% (w/v) SDS, 50 mM TEABC, pH 8.0,

Extraction and 1 mM sodium fluoride, 1 mM sodium orthovanadate, 2.5 mM
Estimation sodium pyrophosphate, 1 mM beta glycerophosphate).
2. Sonicator—Branson Sonifier 250.
3. Benchtop centrifuge.
4. Pierce BCA protein assay kit (Thermo Scientific, Catalog #
23225).

2.2 In-Solution 1. Reducing reagent (100 mM Dithiothreitol (DTT) in 50 MM

Digestion of Proteins TEABC buffer.
2. Alkylating reagent (200 mM Iodoacetamide (IAA) in 50 MM
TEABC buffer (see Note 1).
3. Ice-cold acetone.
4. Triethylammonium bicarbonate (TEABC) (Sigma-Aldrich,
Catalog # T7408-500ML).
5. Sequencing grade modified trypsin (Promega, Catalog
#V5111).

2.3 Tandem Mass 1. TMT 6plex or 10plex isobaric label reagent set.
Tag Labeling of 2. Anhydrous acetonitrile.
Peptides
3. 5% hydroxylamine (Quenching Reagent).
4. Ultrasonic water bath.
5. Benchtop centrifuge and vortex mixer.

2.4 Basic pH Reverse 1. TEABC 1.0 M, pH8.4–8.6.

Phase 2. Acetonitrile.
Chromatography
3. Waters XBridge column (Waters Corporation, Milford, MA;
(bRPLC)
130 Å, 5 μm, 250  9.4 mm).
4. HPLC system.
5. Solvents:
(a) Solvent A: 10 mM TEABC buffer, pH 9.5.
(b) Solvent B: 10 mM TEABC buffer, 90% acetonitrile, pH
9.5.
6. 96-well plates for fraction collection.
7. Speedvac concentrator with cold trap.

2.5 Enrichment of 1. Titansphere 10 μm or 5 μm, GL Sciences Inc., Japan (P/N

Phosphorylated 5020-75010).
Peptides 2. 2,5-Dihydrobenzoic acid (DHB), (Sigma Catalog #. 149357-
100G).
 Functional Proteomic Analysis to Characterize Signaling Crosstalk 209

3. Ammonium hydroxide (J.T. Baker Catalog #9721-01).

4. Trifluoroacetic acid (TFA) (Fisher Scientific Catalog #
A116-50).
5. Empore C8 disks.
6. Solvents:
(a) Wash solution 1: 80% acetonitrile, 3% TFA.
(b) Wash solution 2: 80% acetonitrile, 1% TFA.
(c) Wash solution 3: 80% acetonitrile, 0.1% TFA.
(d) DHB solution (5% 2,5-dihydroxybenzoic acid (DHB),
i.e. 50 mg of DHB in 1 mL of Wash solution 1).
7. Ammonia solution (4% NH4OH, pH ~10.5, i.e. 40 μL of
NH4OH in 1 mL of 40% acetonitrile).

2.6 SepPak-Based 1. Sep-Pak C18 columns (Waters Corporation, Milford, MA,

Sample Cleanup USA).
2. 5-mL syringes.
3. Solvents:
(a) 100% acetonitrile.
(b) Solvent A: 0.1% formic acid.
(c) Solvent B: 40% acetonitrile, 0.1% formic acid.

2.7 Immunoaffinity 1. Anti-Acetyl-Lysine Motif [Ac-K] Immunoaffinity Beads (Cell

Enrichment of Signaling Technology, Catalog # 13416) or Pan anti-
Acetylated and acetyllysine antibody agarose conjugated beads (PTM Bio
Succinylated Lysine PTM-104).
Residues 2. Pan anti-succinyllysine antibody agarose conjugated beads
(PTM Bio PTM-402).
3. Universal pH Indicator strips.
4. Gel loading tips.
5. Benchtop centrifuge.
6. Tube rotator.
7. Solvents:
(a) Immunoaffinity purification buffer (1): 50 mM MOPS
pH 7.2, 10 mM sodium phosphate, 50 mM NaCl
(pH 7.0–7.5) (see Note 2).
(b) 1 M Tris base pH ~10 (1.21 mg in 1 mL water).
(c) MilliQ water.
(d) 0.15% TFA.

2.8 Sample Cleanup 1. Empore C18 disks.

of Enriched Peptides 2. 200 μL pipette tips.
210 Sneha M. Pinto et al.

Experiment

Protein extraction and estimation

In-solution digestion

Tandem Mass Tag labeling of peptides

Pooling TMT tagged peptides and fractionation using high pH HPLC

Enrichment of post-translational modifications using metal affinity/immunoaffinity methods

Phosphorylation Ubiquitylation Acetylation Glycosylation

TiO2 Anti-pTyr Anti-K-ε-GG Anti-Acetyl K Multi-lectin affinity

LC-MS/MS analysis

Protein quantitation and data analysis

100
Relative Abundance
Relative Abundance

50

Quantion using reporter

ion intensities
0
126 127 128 129 130 131
m/z

m/z
Chemotactic factor
LPS INS Chemotactic factor
F2 GF FN1

F2RCD14 RTK ITG GPCR

Pathway/Network analysis
Gα12,13 Sos FAK Gβγ
Cas
Raf Ras CrkII JNK, p38 MAPK signaling
Gene expression
c-Src FGD 1/3 pathway
MEK PI3K Rac1GEF
Dock180

ERK PIP3
Actin polymerization
Vav/Tiam1 Drf3 IRSp53 Mena F-actin
Adherens junction IQGAP Cdc42 Arp2/3
+p Filopodia
NWASP
RhoGEF GRLF1

Asef APC
Rho Rac
TMSB4
PIR121
PXN GIT1 PIX HSPC300
Nap125
+p +p WAVE1
Focal adhesion +p PAK Abi2 PFN Actin
ERM NHE1 ROCK +p polymerization
assembly +p +p
VCL +p +p IRSp53 WAVE2
PI4P5K MLCP MLCK PI4P5K LIMK F-actin
ACTN PIP2 +p Arp2/3
-p +p
Focal adhesion CFN Stabilization
MLC
mDia -p of actin Lamellipodia
PIP2
SSH
PFN
?
Actin VCL GSN
polymerization
stress fiber
Actomyosin
Focal adhesion Focal adhesion
assembly contraction
assembly
F-actin MyosinII
Stress fibers

Fig. 1 Workflow for the identification and quantification of PTM-modified peptides using LC-MS/MS analysis.
The cells are cultured in appropriate growth media and stimulated with growth factors/cytokines or treated
with inhibitors to study their effect. Proteins from each condition are harvested by cell lysis, enzymatically
 Functional Proteomic Analysis to Characterize Signaling Crosstalk 211

3. Syringes.
4. Blunt end needle.
5. Benchtop centrifuge for microfuge tubes.
6. Vacuum concentrator.
7. Solvents.
(a) Acetonitrile (LC-MS grade).
(b) 0.1% formic acid.
(c) 80% acetonitrile, 0.5% acetic acid.
(d) 50% acetonitrile, 0.1% formic acid.

2.9 LC-MS Analysis 1. High-resolution mass spectrometer (Orbitrap Fusion Tribrid

and Data Analysis mass spectrometer, Thermo Electron, Bremen, Germany).
2. Nanoflow liquid chromatography system such as Easy-
nLC1200 or Ultimate 3000 Quaternary RSLC (Thermo Elec-
tron, Bremen, Germany).
3. Data analysis software such as Proteome Discoverer software
suite, MaxQuant, ProteinPilot.

3 Methods

A brief overview of the methodology employed for the identifica-

tion and quantitation of modified peptides using high-resolution
LC-MS/MS analysis is provided in Fig. 1.

3.1 Protein 1. Culture cells in appropriate growth media. Prior to stimulation

Extraction and with growth factors/cytokines or treatment with drugs to
Estimation study their effect, it is recommended to suspend the cells in
media free of serum/growth factor supplements.
2. Once the cells have reached the desired confluency and density,
aspirate media from the culture plates/flasks. In case of adher-
ent cells, aspirate media, wash the cells with chilled 1 PBS
thrice (see Note 3).

Fig. 1 (continued) digested and subsequently labeled with Tandem mass tags (TMT). The TMT-labeled
samples are pooled, fractionated using high pH reverse phase chromatography. The fractionated samples
are dried and subjected to the serial enrichment of posttranslational modifications using metal affinity/
immunoaffinity methods. It is preferable to first enrich for phosphorylated residues using metal affinity
chromatography as the solvents used are volatile and the flowthrough can be then subjected to immunoaffinity
purification. The enriched peptides as well as a fraction of the total peptides are analyzed using high-
resolution nano-LC-MS/MS. The raw data derived from the data acquisition are processed using software
suites to identify and quantify PTM-modified peptides and proteins. The reporter ions are used for the relative
quantitation of the abundance of peptides
212 Sneha M. Pinto et al.

3. Add 0.5–1 mL SDS lysis buffer (2% SDS, 50 mM TEABC, pH

8.0, 1 mM sodium fluoride, 1 mM sodium orthovanadate,
2.5 mM sodium pyrophosphate, 1 mM beta glycerophosphate)
to dish/cell pellet (see Note 4).
4. Gently scrape cells into the lysis buffer using cell scraper and
transfer to 1 mL microfuge tubes, keep on ice. In case of cell
pellet obtained from suspension, add lysis buffer to the cell
pellet and vortex to disrupt cell pellet.
5. Sonicate the lysed samples using a probe sonicator-12 pulses of
10 s each at an amplitude of 30–40%. Heat the lysate at 95  C
for 5 min to ensure complete denaturation. Cool the tubes to
room temperature.
6. Centrifuge the samples at 14,000  g for 10 min at 4  C.
Transfer the supernatant to a new microfuge tube (see Note 5).
7. Estimate the protein concentration of the cleared lysates using
BCA assay.

3.2 Protein Digestion 1. Transfer 600–800 μg per condition into a new tube. Adjust the
final volume such that it is equal in all tubes.
2. Add reducing agent to a final concentration of 10 mM and
incubate at 60  C for 20 min. Cool the tubes to room
temperature.
3. Add alkylating agent to the final concentration of 20 mM and
incubate for 10 min in dark at room temperature.
4. Add six volumes of ice-cold acetone to each tube and incubate
at 20  C overnight (see Note 6).
5. Centrifuge the samples at 14,000  g for 10 min at 4  C.
Discard the supernatant and air-dry the pellet. Resuspend the
protein pellet in 200 μL of 50 mM TEABC buffer (see Note 7).
6. Prepare trypsin solution at a concentration of 1 mg/mL in
50 mM TEABC buffer. Add Trypsin solution at 1:20 (Enzyme
to substrate ratio) to the peptide digest and incubate the
microfuge tubes overnight at 37  C (see Note 8).
7. After the confirmation of digestion efficiency evaporate the
sample to complete dryness.

3.3 Tandem Mass 1. Bring down the TMT reagents to room temperature and add
Tag (TMT) Labeling of 41 μL of anhydrous acetonitrile to each of the TMT reagent
Peptides (0.8 mg) tube.
2. Briefly vortex the tubes to ensure complete dissolution.
3. Centrifuge the tubes and keep for 5 min at room temperature.
Check the pH (7.5–8).
4. Dissolve the vacuum-dried sample in 100 μL of 50 mM
TEABC (see Note 9). Centrifuge the tubes at 12,000  g for
10 min.
 Functional Proteomic Analysis to Characterize Signaling Crosstalk 213

5. Transfer the supernatant directly to TMT reagent vial. Briefly

vortex the samples and centrifuge the tubes to bring down the
solution. Incubate at room temperature for 1 h.
6. Add 8 μL of 5% hydroxylamine to quench the reaction and
incubate at room temperature for about 15 min.
7. Labeling efficiency check: Pipette out volume equivalent to
2 μg from each TMT channel tube and pool the sample. Vortex
briefly and centrifuge at 14,000  g for 10 min. Transfer the
supernatant and desalt using C18 StageTip.
8. Carry out LC-MS analysis of the desalted sample to check the
labeling efficiency. If labeling is efficient, pool the samples from
all TMT channels and evaporate the sample to dryness.

3.4 Fractionation 1. Fractionate the peptide digest using basic reverse phase chro-
matography. Reconstitute the TMT-labeled pooled peptide
digest in 1 mL Solvent A (10 mM TEABC buffer, pH 9.5).
2. Load the sample onto XBridge C18 column. Resolve the pep-
tides using a gradient of solvent A and B for 120 min with a
flow rate of 1 mL/min.
3. Collect the fractions in 96-well plate using a fraction collector
programmed to collect fractions from the start of the gradient
(see Note 10).
4. Reduce the volume of fractions at freezing temperature. Con-
catenate the fractions to obtain a total of 6 fractions. Dry the
samples using vacuum concentrator equipped with refrigerated
vapor trap.

3.5 Enrichment of 1. Reconstitute the fractions with 500 μL of 5% DHB solution.

Post-translational Ensure the peptides are entirely dissolved (see Note 11).
Modifications 2. For enrichment of phosphorylated peptides, the beads-to-pep-
3.5.1 Enrichment of tide ratio is 1:2. Weigh required amount of TiO2 beads.
Phosphorylated Peptides
Pre-warm the TiO2 beads by keeping the tube in the dry bath
Using Metal Affinity for 15–20 min.
Chromatography 3. Suspend TiO2 beads in 5% DHB solution and incubate for
15 min on the rotator at room temperature.
4. Add beads to each tube at a ratio of 1:2 (TiO2: peptide) and
incubate the peptide-DHB-TiO2 mix on a rotator for 30 min at
room temperature.
5. Centrifuge the tubes at 1500  g for 1 min. Transfer superna-
tant to another microfuge tube.
6. Wash the peptide-bound beads with 500 μL wash solution
1 (80% ACN, 3% TFA). Centrifuge the tubes at 1500  g for
1 min. Discard supernatant.
214 Sneha M. Pinto et al.

7. Wash the beads with 500 μL wash solution 2 (80% ACN, 1%

TFA). Centrifuge the tubes at 1500  g for 1 min. Discard
supernatant.
8. Prepare StageTips with C8 plug. Resuspend TiO2 beads in
200 μL wash solution 3 and transfer the entire content to C8
StageTip.
9. Connect the StageTips to 5-mL syringes with the plunger
pulled out. Slowly push the plunger down to remove the
wash solution. The peptide-bound-TiO2 beads remain in
the tip.
10. In separate collection tubes, add 40 μL of 3% TFA and place it
on ice. Elute phosphopeptides from TiO2 beads by adding 4%
NH4OH solution. Prepare the elution buffer shortly
before use.
11. Repeat elution twice. Dry the pooled samples in vacuum con-
centrator with cold trap attached.
12. Resuspend the dried peptides in 0.1% formic acid and desalt
them by C18 StageTip. Store the enriched phosphopeptides in
80  C until LC-MS/MS analysis.
13. The supernatants obtained following incubation with TiO2
beads from each fraction can be further used to enrich other
PTMs. For this, the samples must first be desalted to
remove DHB.
SepPak C18 cleaning:
All the steps for SepPak purification of peptides should be carried
out at room temperature.
1. Connect Sep-Pak C 18 columns to the shorter end of 5-mL
syringes. Remove the plunger as the body of the syringe serves
as a reservoir to hold the solvents as well as the sample.
2. Activate the column using 5 mL of 100% acetonitrile. Connect
the plunger and apply slight pressure such that the solvent
starts to elute. Remove the plunger and allow the solvent to
flow due to gravity.
3. Equilibrate the columns with 7 mL of Solvent A (0.1% Formic
acid) twice. Ensure the column does not dry up at any time
during the course of the sample cleanup.
4. Load the supernatant/peptide sample and follow the proce-
dure described in step 2. Reload the eluate.
5. Wash the column with 12 mL of Solvent A added in two
batches of 6 mL each.
6. Elute the bound peptides using 2 mL of Solvent B (40%
acetonitrile, 0.1% formic acid).
7. Evaporate the purified peptides to dryness using vacuum con-
centrator. Proceed for downstream processing.
 Functional Proteomic Analysis to Characterize Signaling Crosstalk 215

3.5.2 Enrichment of All steps involving enrichment using anti-phosphotyrosine, anti-

PTM-Modified Peptides acetyllysine, or anti-succinyllysine antibody should be carried out
Using Immunoaffinity on ice or at 4  C.
Beads
1. Dissolve the lyophilized peptide digest with 1.4 mL of 1
immunoaffinity purification (IAP) buffer. Sonicate the digest
until a clear solution is obtained with no turbidity observed.
Centrifuge the sample at 3000  g for 30 s.
2. Pipette a small volume (1–2 μL) onto a pH indicator strip. If
the pH is less than 7.0 adjust to the desired pH range (7.0–7.5)
using 1 M Tris base. The volume of 1 M Tris solution should
not exceed 30 μL.
3. Centrifuge the sample at 10,000  g for 5 min. Transfer the
supernatant to a fresh tube.
4. Wash the antibody-bead slurry with 1 mL of 1 PBS buffer.
Care should be taken not to introduce air bubbles. Gently
invert the tube 3–4 times to obtain a homogenous suspension
(see Note 12).
5. Centrifuge at 2000  g for 1 min. Aspirate the buffer leaving
behind ~50 μL volume as there are chances of losing beads
when complete aspiration is performed. Repeat step 4 twice.
Finally wash the beads with 1 IAP buffer.
6. Transfer the peptide solution to the antibody beads containing
vial. Pipette the sample directly on top of the beads. Gently
invert the tube 3–4 times to ensure homogenous mixing. Care
should be taken not to introduce any air bubbles.
7. Incubate the peptide-antibody beads on a rotator for 2 h at
4  C.
8. Centrifuge at 1500  g for 1 min. Keep tubes on ice for
1–2 min for the beads to settle down completely. Carefully
transfer supernatant using a fresh microfuge tube without dis-
turbing the beads (see Note 13).
9. Wash the peptide-bound antibody beads twice with 1 mL of 1
IAP buffer added to the beads each time. Gently mix the
contents by inverting the tubes 4–5 times. Ensure that no air
bubbles are introduced. Centrifuge at 1500  g for 30 s and
remove supernatant.
10. Wash the beads to remove unbound/nonspecific bound pep-
tides using 1 mL chilled HPLC grade water. Mix well to create
a homogenous suspension. Centrifuge at 1500  g for 1 min
and remove supernatant. Repeat the steps for a total of three
washes.
11. Following the last washing step, carefully remove the entire
solvent using a gel loading tip.
216 Sneha M. Pinto et al.

12. Add 55 μL of 0.15% TFA to elute the enriched PTM peptides.

Gently tap the bottom of the tube several times such that a
homogenous slurry is formed. Incubate for 10 min at room
temperature (see Note 14).
13. Centrifuge at 1500  g for 30 s. Transfer motif-enriched
peptides to fresh microfuge tubes.
14. Repeat step 13 twice and pool the eluates. Use gel loader tips
to recover the enriched peptides without disturbing the beads.
C18 StageTip cleanup:
1. Prepare C18StageTip using Empore C18 disk.
2. Equilibrate the C18 disk with 120 μl of 80% ACN, 0.5% acetic
acid solution followed by 120 μL of 50% ACN, 0.1% Formic
acid, and 120 μL of 0.15% TFA.
3. Load the pooled eluate onto the equilibrated C18 StageTip
connected to the end of a 2-mL syringe. Elute the sample
into a microfuge tube and reload the eluate.
4. Add 40 μL of 0.15% TFA to wash the bound peptides. Ensure
that no beads are present.
5. Elute the bound peptides with 10 μL of 50% ACN, 0.1%
Formic acid. Repeat the process thrice. Combine the eluate
and evaporate to dryness using vacuum concentrator.
6. Store the enriched modified peptides in 80  C until LC-MS/
MS analysis.

3.6 Mass To characterize dynamic states of protein complexes in signaling

Spectrometry Data pathways, the TMT-labeled fractions enriched for PTM-modified
Acquisition of PTM peptides can be analyzed using high-resolution nanoflow HPLC-
Experiments MS/MS setup that is being mostly employed for shotgun proteo-
mics analysis (see Note 15). The PTM enriched sample from each
fraction can be reconstituted in 0.1% formic acid and loaded onto
trap column that is typically of 2 cm length and is packed with 3 μm
C18 material. The enriched peptides are resolved on 15–50 cm C18
analytical columns with 75 μm diameter. The typical flow rate
employed for resolving the peptides is 250–300 nL/min with a
column oven temperature set at 40  C. A linear gradient of 5–35%
solvent B (80% acetonitrile in 0.1% formic acid) over 100 min is
generally used in our setting to resolve the peptide mixture with a
total run time of 120 min. Data dependent acquisition with full
scans in 400–1600 m/z range is typically carried out using an
Orbitrap mass analyzer at a mass resolution of 120,000 at
200 m/z in the positive ion mode. From the precursor survey
scan, we generally select the most intense precursor ions for an
MS2 scan using TopSpeed mode. The selected precursor ions
were fragmented using HCD mode with 40–42% normalized colli-
sion energy. The scan range is typically set as 100–1600 m/z and
 Functional Proteomic Analysis to Characterize Signaling Crosstalk 217

detected at a mass resolution of 60,000 at 200 m/z. Studies have

reported that for most isobaric-based quantification methods,
co-isolating precursor species result in distorted reporter ion inten-
sities, thereby providing an inaccurate estimate of its abundance in
any given sample. One approach would be to narrow down the
precursor isolation width; however, this may hamper the peak
picking of the isotopic cluster. This issue has been partly circum-
vented with the advent of Orbitrap Fusion mass spectrometer
which enables synchronous precursor selection (SPS) [76]. Avail-
ability of three mass analyzers enables increased scan rates as the
MS2 events occur in the IonTrap and the MS3 events occur in
parallel in the Orbitrap mass analyzer. Enabling SPS to include
10 MS2 fragment ions with an increased collision energy of up to
55% and a narrow m/z range provides TMT reporter ion abun-
dance. SPS increase the sensitivity and quantitation accuracy by
isolating multiple fragment ions in the MS2 spectrum. The data
obtained can be searched against reference databases for the organ-
ism under the study. The search parameters should include dynamic
modification of the PTMs under study. If quantitative methodol-
ogy is employed, then the appropriate quantitation node with
TMT/iTRAQ should be enabled. Statistical assessment of site
localization can be carried out using algorithms such as ptmRS
[44] and A-Score [45].

4 Notes

1. Prepare IAA shortly before use and since it is light sensitive, it is

recommended to be stored in the dark until use.
2. Prepare a stock of 10 IAP buffer. This solution can be stored
at 4  C for 2 months. Prior to use, dilute the requisite volume
with MilliQ water to 1 concentration before use. Store for up
to 1 month at 4  C.
3. In case of suspension cells, ensure the cells are suspended
homogenously, centrifuge and pellet down the cells. Aspirate
the media and resuspend the cell pellet in 15 mL chilled 1
PBS buffer. Make a homogenous suspension. Repeat centrifu-
gation and completely aspirate PBS. Repeat the process thrice.
4. The volume of lysis buffer can be adjusted such that the final
protein concentration does not exceed 5 mg/mL. For acetyl or
succinyl enrichment, appropriate inhibitors such as sodium
butyrate (final concentration 5 mM) should be added.
5. In case of microorganisms or infectious agents employed to
study host-pathogen interactions, it is recommended to filter
the lysate using 0.22 μm filter to ensure no viable microbes are
retained.
218 Sneha M. Pinto et al.

6. Ensure the centrifuge tubes are made of acetone compatible

material else the plasticizers may leach and interfere with
LC-MS analysis. Acetone precipitates the protein in solution
and based on our experience, the protein loss is minimal when
the samples are reduced and alkylated prior to precipitation. In
addition to effective removal of SDS, acetone also enables
removal of excess DTT and IAA added to the lysates.
7. Acetone-precipitated pellet may not dissolve completely. It is
recommended to use sonicator water bath to disrupt the pellet
and dissolve the samples. In case the pellet does not dissolve, an
additional 100 μL of 50 mM TEABC buffer can be added.
8. It is important to check the digestion efficiency before proceed-
ing to TMT labeling of the samples. Pre-digest samples can be
taken prior to addition of chilled acetone and the post-digest
sample (10 μg/condition) can be taken post 14–16 h incuba-
tion with trypsin.
9. Use a water bath sonicator to ensure complete dissolution of
peptide digest.
10. Add 50 μL of 1% Formic acid to 96 wells prior to fraction
collection to neutralize peptides.
11. The tubes can be kept in the thermomixer set at 1150 rpm for
20 min to ensure complete dissolution. Alternatively, the tubes
can be placed on vortex mixer set at low speed for 20 min.
12. Pipette the desired volume of IAP buffer directly onto the
beads such that the beads are dislodged and enter into the
solution. Forceful ejection of the solvent must be avoided as
it can introduce air bubbles. Care should also be taken to
ensure that the beads are not agitated which can result in the
breaking of the beads further leading to nonspecific binding.
13. The supernatant obtained at this step can be subjected to
enrichment using other PTM affinity antibody beads. For
instance, after phosphopeptide enrichment using TiO2 beads
which predominantly enriches phosphorylation at serine and
threonine residues, the supernatant can be subjected to enrich-
ment using anti-phosphotyrosine antibody beads. The super-
natant thereby obtained can be further subjected to
enrichment using anti-acetyllysine or anti-succinyllysine
antibody.
14. Tap the tubes gently to ensure that the beads do not settle at
the bottom and remain in suspension every 2–3 min.
15. In case of a quantitative PTMomic experiment, it is ideal to
carry out a corresponding quantitative proteomic profiling
experiment using the same sample. This will help in ascertain-
ing that changes of PTM abundance are not due to changes in
the protein abundance.
 Functional Proteomic Analysis to Characterize Signaling Crosstalk 219

Acknowledgements

The authors acknowledge Yenepoya (Deemed to be University) for

access to mass spectrometry instrumentation facility. We also thank
Karnataka Biotechnology and Information Technology Services
(KBITS), Government of Karnataka, for the support to the Center
for Systems Biology and Molecular Medicine at Yenepoya Univer-
sity under the Biotechnology Skill Enhancement Programme in
Multiomics Technology (BiSEP GO ITD 02 MDA 2017). SMP is
a recipient of INSPIRE Faculty Award from Department of Science
and Technology (DST), Government of India.

References
1. UniProt C (2015) UniProt: a hub for protein kinome profiling by rational selection of rele-
information. Nucleic Acids Res 43(Database vant spots. Sci Rep 6:26695. https://doi.org/
issue):D204–D212. https://doi.org/10. 10.1038/srep26695
1093/nar/gku989 8. Baharani A, Trost B, Kusalik A, Napper S
2. Creasy DM, Cottrell JS (2004) Unimod: pro- (2017) Technological advances for interrogat-
tein modifications for mass spectrometry. Pro- ing the human kinome. Biochem Soc Trans 45
teomics 4(6):1534–1536. https://doi.org/10. (1):65–77. https://doi.org/10.1042/
1002/pmic.200300744 BST20160163
3. Zhang H, Shi X, Pelech S (2016) Monitoring 9. Kim MS, Pinto SM, Getnet D, Nirujogi RS,
protein kinase expression and phosphorylation Manda SS, Chaerkady R, Madugundu AK,
in cell lysates with antibody microarrays. Meth- Kelkar DS, Isserlin R, Jain S, Thomas JK,
ods Mol Biol 1360:107–122. https://doi.org/ Muthusamy B, Leal-Rojas P, Kumar P, Sahas-
10.1007/978-1-4939-3073-9_9 rabuddhe NA, Balakrishnan L, Advani J,
4. Shi J, Sharif S, Ruijtenbeek R, Pieters RJ George B, Renuse S, Selvan LD, Patil AH,
(2016) Activity based high-throughput screen- Nanjappa V, Radhakrishnan A, Prasad S,
ing for novel O-GlcNAc transferase substrates Subbannayya T, Raju R, Kumar M, Sreenivasa-
using a dynamic peptide microarray. PLoS One murthy SK, Marimuthu A, Sathe GJ, Chavan S,
11(3):e0151085. https://doi.org/10.1371/ Datta KK, Subbannayya Y, Sahu A, Yelamanchi
journal.pone.0151085 SD, Jayaram S, Rajagopalan P, Sharma J,
5. Zhu B, Farris TR, Milligan SL, Chen H, Murthy KR, Syed N, Goel R, Khan AA,
Zhu R, Hong A, Zhou X, Gao X, McBride Ahmad S, Dey G, Mudgal K, Chatterjee A,
JW (2016) Rapid identification of ubiquitina- Huang TC, Zhong J, Wu X, Shaw PG,
tion and SUMOylation target sites by micro- Freed D, Zahari MS, Mukherjee KK,
fluidic peptide array. Biochem Biophys Rep Shankar S, Mahadevan A, Lam H, Mitchell
5:430–438. https://doi.org/10.1016/j. CJ, Shankar SK, Satishchandra P, Schroeder
bbrep.2016.02.003 JT, Sirdeshmukh R, Maitra A, Leach SD,
Drake CG, Halushka MK, Prasad TS, Hruban
6. Al-Ejeh F, Miranda M, Shi W, Simpson PT, RH, Kerr CL, Bader GD, Iacobuzio-Donahue
Song S, Vargas AC, Saunus JM, Smart CE, CA, Gowda H, Pandey A (2014) A draft map
Mariasegaram M, Wiegmans AP, Chenevix- of the human proteome. Nature 509
Trench G, Lakhani SR, Khanna KK (2014) (7502):575–581. https://doi.org/10.1038/
Kinome profiling reveals breast cancer hetero- nature13302
geneity and identifies targeted therapeutic
opportunities for triple negative breast cancer. 10. Wilhelm M, Schlegl J, Hahne H, Gholami AM,
Oncotarget 5(10):3145–3158. https://doi. Lieberenz M, Savitski MM, Ziegler E,
org/10.18632/oncotarget.1865 Butzmann L, Gessulat S, Marx H,
Mathieson T, Lemeer S, Schnatbaum K,
7. Scholma J, Fuhler GM, Joore J, Hulsman M, Reimer U, Wenschuh H, Mollenhauer M,
Schivo S, List AF, Reinders MJ, Peppelenbosch Slotta-Huspenina J, Boese JH, Bantscheff M,
MP, Post JN (2016) Improved intra-array and Gerstmair A, Faerber F, Kuster B (2014) Mass-
interarray normalization of peptide microarray spectrometry-based draft of the human
phosphorylation for phosphorylome and
220 Sneha M. Pinto et al.

proteome. Nature 509(7502):582–587. 19. Matsuoka S, Ballif BA, Smogorzewska A,

https://doi.org/10.1038/nature13319 McDonald ER 3rd, Hurov KE, Luo J, Baka-
11. Zhao Y, Jensen ON (2009) Modification- larski CE, Zhao Z, Solimini N, Lerenthal Y,
specific proteomics: strategies for characteriza- Shiloh Y, Gygi SP, Elledge SJ (2007) ATM
tion of post-translational modifications using and ATR substrate analysis reveals extensive
enrichment techniques. Proteomics 9 protein networks responsive to DNA damage.
(20):4632–4641. https://doi.org/10.1002/ Science 316(5828):1160–1166. https://doi.
pmic.200900398 org/10.1126/science.1140321
12. Sathe G, Pinto SM, Syed N, Nanjappa V, 20. Pinkse MW, Uitto PM, Hilhorst MJ, Ooms B,
Solanki HS, Renuse S, Chavan S, Khan AA, Heck AJ (2004) Selective isolation at the fem-
Patil AH, Nirujogi RS, Nair B, Mathur PP, tomole level of phosphopeptides from proteo-
Prasad TSK, Gowda H, Chatterjee A (2016) lytic digests using 2D-NanoLC-ESI-MS/MS
Phosphotyrosine profiling of curcumin- and titanium oxide precolumns. Anal Chem
induced signaling. Clin Proteomics 13:13. 76(14):3935–3943. https://doi.org/10.
https://doi.org/10.1186/s12014-016-9114- 1021/ac0498617
0 21. Li Y, Xu X, Qi D, Deng C, Yang P, Zhang X
13. Yu Y, Gaillard S, Phillip JM, Huang TC, Pinto (2008) Novel Fe3O4@TiO2 core-shell micro-
SM, Tessarollo NG, Zhang Z, Pandey A, spheres for selective enrichment of phospho-
Wirtz D, Ayhan A, Davidson B, Wang TL, peptides in phosphoproteome analysis. J
Shih Ie M (2015) Inhibition of spleen tyrosine Proteome Res 7(6):2526–2538. https://doi.
kinase potentiates paclitaxel-induced cytotoxic- org/10.1021/pr700582z
ity in ovarian cancer cells by stabilizing micro- 22. Feng S, Ye M, Zhou H, Jiang X, Jiang X,
tubules. Cancer Cell 28(1):82–96. https://doi. Zou H, Gong B (2007) Immobilized zirco-
org/10.1016/j.ccell.2015.05.009 nium ion affinity chromatography for specific
14. Pinto SM, Nirujogi RS, Rojas PL, Patil AH, enrichment of phosphopeptides in phospho-
Manda SS, Subbannayya Y, Roa JC, proteome analysis. Mol Cell Proteomics 6
Chatterjee A, Prasad TS, Pandey A (2015) (9):1656–1665. https://doi.org/10.1074/
Quantitative phosphoproteomic analysis of mcp.T600071-MCP200
IL-33-mediated signaling. Proteomics 15 23. Thingholm TE, Jensen ON (2009) Enrich-
(2–3):532–544. https://doi.org/10.1002/ ment and characterization of phosphopeptides
pmic.201400303 by immobilized metal affinity chromatography
15. Zahari MS, Wu X, Pinto SM, Nirujogi RS, Kim (IMAC) and mass spectrometry. Methods Mol
MS, Fetics B, Philip M, Barnes SR, Godfrey B, Biol 527:47–56, xi. https://doi.org/10.
Gabrielson E, Nevo E, Pandey A (2015) Phos- 1007/978-1-60327-834-8_4
phoproteomic profiling of tumor tissues iden- 24. Verma R, Pinto SM, Patil AH, Advani J,
tifies HSP27 Ser82 phosphorylation as a robust Subba P, Kumar M, Sharma J, Dey G,
marker of early ischemia. Sci Rep 5:13660. Ravikumar R, Buggi S, Satishchandra P,
https://doi.org/10.1038/srep13660 Sharma K, Suar M, Tripathy SP, Chauhan DS,
16. Harsha HC, Pinto SM, Pandey A (2013) Pro- Gowda H, Pandey A, Gandotra S, Prasad TS
teomic strategies to characterize signaling (2017) Quantitative proteomic and phospho-
pathways. Methods Mol Biol 1007:359–377. proteomic analysis of H37Ra and H37Rv
https://doi.org/10.1007/978-1-62703-392- strains of mycobacterium tuberculosis. J Prote-
3_16 ome Res 16(4):1632–1645. https://doi.org/
17. Bao X, Wang Y, Li X, Li XM, Liu Z, Yang T, 10.1021/acs.jproteome.6b00983
Wong CF, Zhang J, Hao Q, Li XD (2014) 25. Thingholm TE, Jensen ON, Robinson PJ, Lar-
Identification of ‘erasers’ for lysine crotony- sen MR (2008) SIMAC (sequential elution
lated histone marks using a chemical proteo- from IMAC), a phosphoproteomics strategy
mics approach. elife 3. https://doi.org/10. for the rapid separation of monophosphory-
7554/eLife.02999 lated from multiply phosphorylated peptides.
18. Gu H, Ren JM, Jia X, Levy T, Rikova K, Mol Cell Proteomics 7(4):661–671. https://
Yang V, Lee KA, Stokes MP, Silva JC (2016) doi.org/10.1074/mcp.M700362-MCP200
Quantitative profiling of post-translational 26. Bertozzi CR, Kiessling LL (2001) Chemical
modifications by immunoaffinity enrichment glycobiology. Science 291(5512):2357–2364
and LC-MS/MS in cancer serum without 27. Vocadlo DJ, Hang HC, Kim EJ, Hanover JA,
immunodepletion. Mol Cell Proteomics 15 Bertozzi CR (2003) A chemical approach for
(2):692–702. https://doi.org/10.1074/mcp. identifying O-GlcNAc-modified proteins in
O115.052266 cells. Proc Natl Acad Sci U S A 100
 Functional Proteomic Analysis to Characterize Signaling Crosstalk 221

(16):9116–9121. https://doi.org/10.1073/ Khainovski N, Pillai S, Dey S, Daniels S,

pnas.1632821100 Purkayastha S, Juhasz P, Martin S, Bartlet-
28. Lanyon-Hogg T, Faronato M, Serwa RA, Tate Jones M, He F, Jacobson A, Pappin DJ
EW (2017) Dynamic protein acylation: new (2004) Multiplexed protein quantitation in
substrates, mechanisms, and drug targets. Saccharomyces cerevisiae using amine-reactive
Trends Biochem Sci 42(7):566–581. https:// isobaric tagging reagents. Mol Cell Proteomics
doi.org/10.1016/j.tibs.2017.04.004 3(12):1154–1169. https://doi.org/10.1074/
29. Morrison E, Kuropka B, Kliche S, Brugger B, mcp.M400129-MCP200
Krause E, Freund C (2015) Quantitative anal- 39. Thompson A, Schafer J, Kuhn K, Kienle S,
ysis of the human T cell palmitome. Sci Rep Schwarz J, Schmidt G, Neumann T,
5:11598. https://doi.org/10.1038/ Johnstone R, Mohammed AK, Hamon C
srep11598 (2003) Tandem mass tags: a novel quantifica-
30. Roth AF, Wan J, Bailey AO, Sun B, Kuchar JA, tion strategy for comparative analysis of com-
Green WN, Phinney BS, Yates JR 3rd, Davis plex protein mixtures by MS/MS. Anal Chem
NG (2006) Global analysis of protein palmi- 75(8):1895–1904
toylation in yeast. Cell 125(5):1003–1013. 40. Schwanhausser B, Busse D, Li N, Dittmar G,
https://doi.org/10.1016/j.cell.2006.03.042 Schuchhardt J, Wolf J, Chen W, Selbach M
31. Zhang Y, Zhang C, Jiang H, Yang P, Lu H (2011) Global quantification of mammalian
(2015) Fishing the PTM proteome with chem- gene expression control. Nature 473
ical approaches using functional solid phases. (7347):337–342. https://doi.org/10.1038/
Chem Soc Rev 44(22):8260–8287. https:// nature10098
doi.org/10.1039/c4cs00529e 41. Zybailov B, Mosley AL, Sardiu ME, Coleman
32. Tate EW (2008) Recent advances in chemical MK, Florens L, Washburn MP (2006) Statisti-
proteomics: exploring the post-translational cal analysis of membrane proteome expression
proteome. J Chem Biol 1(1–4):17–26. changes in Saccharomyces cerevisiae. J Prote-
https://doi.org/10.1007/s12154-008-0002- ome Res 5(9):2339–2347. https://doi.org/
6 10.1021/pr060161n
33. Webb K, Bennett EJ (2013) Eavesdropping on 42. Keller A, Bader SL, Kusebauch U,
PTM cross-talk through serial enrichment. Nat Shteynberg D, Hood L, Moritz RL (2016)
Methods 10(7):620–621. https://doi.org/10. Opening a SWATH window on posttransla-
1038/nmeth.2526 tional modifications: automated pursuit of
modified peptides. Mol Cell Proteomics 15
34. Swaney DL, Beltrao P, Starita L, Guo A, (3):1151–1163. https://doi.org/10.1074/
Rush J, Fields S, Krogan NJ, Villen J (2013) mcp.M115.054478
Global analysis of phosphorylation and ubiqui-
tylation cross-talk in protein degradation. Nat 43. Lu P, Vogel C, Wang R, Yao X, Marcotte EM
Methods 10(7):676–682. https://doi.org/10. (2007) Absolute protein expression profiling
1038/nmeth.2519 estimates the relative contributions of tran-
scriptional and translational regulation. Nat
35. Mertins P, Qiao JW, Patel J, Udeshi ND, Clau- Biotechnol 25(1):117–124. https://doi.org/
ser KR, Mani DR, Burgess MW, Gillette MA, 10.1038/nbt1270
Jaffe JD, Carr SA (2013) Integrated proteomic
analysis of post-translational modifications by 44. Taus T, Kocher T, Pichler P, Paschke C,
serial enrichment. Nat Methods 10 Schmidt A, Henrich C, Mechtler K (2011)
(7):634–637. https://doi.org/10.1038/ Universal and confident phosphorylation site
nmeth.2518 localization using phosphoRS. J Proteome
Res 10(12):5354–5362. https://doi.org/10.
36. Kim MS, Zhong J, Kandasamy K, Delanghe B, 1021/pr200611n
Pandey A (2011) Systematic evaluation of
alternating CID and ETD fragmentation for 45. Beausoleil SA, Villen J, Gerber SA, Rush J,
phosphorylated peptides. Proteomics 11 Gygi SP (2006) A probability-based approach
(12):2568–2572. https://doi.org/10.1002/ for high-throughput protein phosphorylation
pmic.201000547 analysis and site localization. Nat Biotechnol
24(10):1285–1292. https://doi.org/10.
37. Ong SE, Blagoev B, Kratchmarova I, Kristen- 1038/nbt1240
sen DB, Steen H, Pandey A, Mann M (2002)
Stable isotope labeling by amino acids in cell 46. Peng X, Xu F, Liu S, Li S, Huang Q, Chang L,
culture, SILAC, as a simple and accurate Wang L, Ma X, He F, Xu P (2017) Identifica-
approach to expression proteomics. Mol Cell tion of missing proteins in the phosphopro-
Proteomics 1(5):376–386 teome of kidney cancer. J Proteome Res 16
(12):4364–4373. https://doi.org/10.1021/
38. Ross PL, Huang YN, Marchese JN, acs.jproteome.7b00332
Williamson B, Parker K, Hattan S,
222 Sneha M. Pinto et al.

47. Zhang Z, Tan M, Xie Z, Dai L, Chen Y, Zhao Y Biochim Biophys Acta 1866:464. https://doi.
(2011) Identification of lysine succinylation as org/10.1016/j.bbapap.2017.11.011
a new post-translational modification. Nat 56. Meyer JG, D’Souza AK, Sorensen DJ, Rardin
Chem Biol 7(1):58–63. https://doi.org/10. MJ, Wolfe AJ, Gibson BW, Schilling B (2016)
1038/nchembio.495 Quantification of lysine acetylation and Succi-
48. Weinert BT, Scholz C, Wagner SA, nylation stoichiometry in proteins using mass
Iesmantavicius V, Su D, Daniel JA, Choudhary spectrometric data-independent acquisitions
C (2013) Lysine succinylation is a frequently (SWATH). J Am Soc Mass Spectrom 27
occurring modification in prokaryotes and (11):1758–1771. https://doi.org/10.1007/
eukaryotes and extensively overlaps with acety- s13361-016-1476-z
lation. Cell Rep 4(4):842–851. https://doi. 57. Pickart CM, Eddins MJ (2004) Ubiquitin:
org/10.1016/j.celrep.2013.07.024 structures, functions, mechanisms. Biochim
49. Xie Z, Dai J, Dai L, Tan M, Cheng Z, Wu Y, Biophys Acta 1695(1–3):55–72. https://doi.
Boeke JD, Zhao Y (2012) Lysine succinylation org/10.1016/j.bbamcr.2004.09.019
and lysine malonylation in histones. Mol Cell 58. Wu Q, Cheng Z, Zhu J, Xu W, Peng X,
Proteomics 11(5):100–107. https://doi.org/ Chen C, Li W, Wang F, Cao L, Yi X, Wu Z,
10.1074/mcp.M111.015875 Li J, Fan P (2015) Suberoylanilide hydroxamic
50. Drazic A, Myklebust LM, Ree R, Arnesen T acid treatment reveals crosstalks among prote-
(2016) The world of protein acetylation. Bio- ome, ubiquitylome and acetylome in non-small
chim Biophys Acta 1864(10):1372–1401. cell lung cancer A549 cell line. Sci Rep 5:9520.
https://doi.org/10.1016/j.bbapap.2016.06. https://doi.org/10.1038/srep09520
007 59. Iesmantavicius V, Weinert BT, Choudhary C
51. Kim SC, Sprung R, Chen Y, Xu Y, Ball H, Pei J, (2014) Convergence of ubiquitylation and
Cheng T, Kho Y, Xiao H, Xiao L, Grishin NV, phosphorylation signaling in rapamycin-
White M, Yang XJ, Zhao Y (2006) Substrate treated yeast cells. Mol Cell Proteomics 13
and functional diversity of lysine acetylation (8):1979–1992. https://doi.org/10.1074/
revealed by a proteomics survey. Mol Cell 23 mcp.O113.035683
(4):607–618. https://doi.org/10.1016/j. 60. Yang X, Liu F, Yan Y, Zhou T, Guo Y, Sun G,
molcel.2006.06.026 Zhou Z, Zhang W, Guo X, Sha J (2015) Pro-
52. Zhang J, Sprung R, Pei J, Tan X, Kim S, teomic analysis of N-glycosylation of human
Zhu H, Liu CF, Grishin NV, Zhao Y (2009) seminal plasma. Proteomics 15
Lysine acetylation is a highly abundant and (7):1255–1258. https://doi.org/10.1002/
evolutionarily conserved modification in pmic.201400203
Escherichia coli. Mol Cell Proteomics 8 61. Sudhir PR, Chen CH, Pavana Kumari M, Wang
(2):215–225. https://doi.org/10.1074/mcp. MJ, Tsou CC, Sung TY, Chen JY, Chen CH
M800187-MCP200 (2012) Label-free quantitative proteomics and
53. Henriksen P, Wagner SA, Weinert BT, N-glycoproteomics analysis of KRAS-activated
Sharma S, Bacinskaja G, Rehman M, Juffer human bronchial epithelial cells. Mol Cell Pro-
AH, Walther TC, Lisby M, Choudhary C teomics 11(10):901–915. https://doi.org/10.
(2012) Proteome-wide analysis of lysine acety- 1074/mcp.M112.020875
lation suggests its broad regulatory scope in 62. Chen W, Smeekens JM, Wu R (2014) A uni-
Saccharomyces cerevisiae. Mol Cell Proteomics versal chemical enrichment method for
11(11):1510–1522. https://doi.org/10. mapping the yeast N-glycoproteome by mass
1074/mcp.M112.017251 spectrometry (MS). Mol Cell Proteomics 13
54. Lundby A, Lage K, Weinert BT, Bekker-Jensen (6):1563–1572. https://doi.org/10.1074/
DB, Secher A, Skovgaard T, Kelstrup CD, mcp.M113.036251
Dmytriyev A, Choudhary C, Lundby C, 63. Guan X, Fierke CA (2011) Understanding pro-
Olsen JV (2012) Proteomic analysis of lysine tein palmitoylation: biological significance and
acetylation sites in rat tissues reveals organ enzymology. Sci China Chem 54
specificity and subcellular patterns. Cell Rep 2 (12):1888–1897. https://doi.org/10.1007/
(2):419–431. https://doi.org/10.1016/j.cel s11426-011-4428-2
rep.2012.07.006 64. Martin BR (2013) Nonradioactive analysis of
55. Xie C, Shen H, Zhang H, Yan J, Liu Y, Yao F, dynamic protein palmitoylation. Curr Protoc
Wang X, Cheng Z, Tang TS, Guo C (2017) Protein Sci 73(Unit 14):15. https://doi.org/
Quantitative proteomics analysis reveals altera- 10.1002/0471140864.ps1415s73
tions of lysine acetylation in mouse testis in 65. Martin BR, Wang C, Adibekian A, Tully SE,
response to heat shock and X-ray exposure. Cravatt BF (2011) Global profiling of dynamic
 Functional Proteomic Analysis to Characterize Signaling Crosstalk 223

protein palmitoylation. Nat Methods 9 73. Minguez P, Parca L, Diella F, Mende DR,
(1):84–89. https://doi.org/10.1038/nmeth. Kumar R, Helmer-Citterich M, Gavin AC, van
1769 Noort V, Bork P (2012) Deciphering a global
66. Hornbeck PV, Zhang B, Murray B, Kornhauser network of functionally associated post-
JM, Latham V, Skrzypek E (2015) PhosphoSi- translational modifications. Mol Syst Biol
tePlus, 2014: mutations, PTMs and recalibra- 8:599. https://doi.org/10.1038/msb.2012.31
tions. Nucleic Acids Res 43(Database issue): 74. Trinidad JC, Barkan DT, Gulledge BF,
D512–D520. https://doi.org/10.1093/nar/ Thalhammer A, Sali A, Schoepfer R, Burlin-
gku1267 game AL (2012) Global identification and
67. Keshava Prasad TS, Goel R, Kandasamy K, characterization of both O-GlcNAcylation
Keerthikumar S, Kumar S, Mathivanan S, and phosphorylation at the murine synapse.
Telikicherla D, Raju R, Shafreen B, Mol Cell Proteomics 11(8):215–229. https://
Venugopal A, Balakrishnan L, Marimuthu A, doi.org/10.1074/mcp.O112.018366
Banerjee S, Somanathan DS, Sebastian A, 75. Kim MS, Zhong J, Pandey A (2016) Common
Rani S, Ray S, Harrys Kishore CJ, Kanth S, errors in mass spectrometry-based analysis of
Ahmed M, Kashyap MK, Mohmood R, Rama- post-translational modifications. Proteomics
chandra YL, Krishna V, Rahiman BA, Mohan S, 16(5):700–714. https://doi.org/10.1002/
Ranganathan P, Ramabadran S, Chaerkady R, pmic.201500355
Pandey A (2009) Human protein reference 76. Hughes CS, Spicer V, Krokhin OV, Morin GB
database—2009 update. Nucleic Acids Res 37 (2017) Investigating acquisition performance
(Database):D767–D772. https://doi.org/10. on the Orbitrap fusion when using tandem
1093/nar/gkn892 MS/MS/MS scanning with isobaric tags. J
68. Huang KY, Su MG, Kao HJ, Hsieh YC, Jhong Proteome Res 16(5):1839–1846. https://doi.
JH, Cheng KH, Huang HD, Lee TY (2016) org/10.1021/acs.jproteome.7b00091
dbPTM 2016: 10-year anniversary of a 77. Pan J, Chen R, Li C, Li W, Ye Z (2015) Global
resource for post-translational modification of analysis of protein lysine succinylation profiles
proteins. Nucleic Acids Res 44(D1): and their overlap with lysine acetylation in the
D435–D446. https://doi.org/10.1093/nar/ marine bacterium vibrio parahemolyticus. J
gkv1240 Proteome Res 14(10):4309–4318. https://
69. Liu Z, Wang Y, Gao T, Pan Z, Cheng H, doi.org/10.1021/acs.jproteome.5b00485
Yang Q, Cheng Z, Guo A, Ren J, Xue Y 78. Cheng Y, Hou T, Ping J, Chen G, Chen J
(2014) CPLM: a database of protein lysine (2016) Quantitative succinylome analysis in
modifications. Nucleic Acids Res 42(Database the liver of non-alcoholic fatty liver disease rat
issue):D531–D536. https://doi.org/10. model. Proteome Sci 14:3. https://doi.org/
1093/nar/gkt1093 10.1186/s12953-016-0092-y
70. Kennedy JJ, Yan P, Zhao L, Ivey RG, Voytovich 79. Colak G, Xie Z, Zhu AY, Dai L, Lu Z, Zhang Y,
UJ, Moore HD, Lin C, Pogosova-Agadjanyan Wan X, Chen Y, Cha YH, Lin H, Zhao Y, Tan
EL, Stirewalt DL, Reding KW, Whiteaker JR, M (2013) Identification of lysine succinylation
Paulovich AG (2016) Immobilized metal affin- substrates and the succinylation regulatory
ity chromatography coupled to multiple reac- enzyme CobB in Escherichia coli. Mol Cell
tion monitoring enables reproducible Proteomics 12(12):3509–3520. https://doi.
quantification of phospho-signaling. Mol Cell org/10.1074/mcp.M113.031567
Proteomics 15(2):726–739. https://doi.org/ 80. Sun G, Jiang M, Zhou T, Guo Y, Cui Y, Guo X,
10.1074/mcp.O115.054940 Sha J (2014) Insights into the lysine acetylpro-
71. Chick JM, Kolippakkam D, Nusinow DP, teome of human sperm. J Proteome
Zhai B, Rad R, Huttlin EL, Gygi SP (2015) A 109:199–211. https://doi.org/10.1016/j.
mass-tolerant database search identifies a large jprot.2014.07.002
proportion of unassigned spectra in shotgun 81. Xie L, Wang X, Zeng J, Zhou M, Duan X, Li Q,
proteomics as modified peptides. Nat Biotech- Zhang Z, Luo H, Pang L, Li W, Liao G, Yu X,
nol 33(7):743–749. https://doi.org/10. Li Y, Huang H, Xie J (2015) Proteome-wide
1038/nbt.3267 lysine acetylation profiling of the human path-
72. Pan Z, Liu Z, Cheng H, Wang Y, Gao T, ogen Mycobacterium tuberculosis. Int J Bio-
Ullah S, Ren J, Xue Y (2014) Systematic analy- chem Cell Biol 59:193–202. https://doi.org/
sis of the in situ crosstalk of tyrosine modifica- 10.1016/j.biocel.2014.11.010
tions reveals no additional natural selection on 82. Karg E, Smets M, Ryan J, Forne I, Qin W,
multiply modified residues. Sci Rep 4:7331. Mulholland CB, Kalideris G, Imhof A,
https://doi.org/10.1038/srep07331 Bultmann S, Leonhardt H (2017) Ubiquitome
224 Sneha M. Pinto et al.

analysis reveals PCNA-associated factor revised database of O-glycosylated proteins.

15 (PAF15) as a specific ubiquitination target Nucleic Acids Res 27(1):370–372
of UHRF1 in embryonic stem cells. J Mol Biol 89. Gnad F, Gunawardena J, Mann M (2011)
429(24):3814–3824. https://doi.org/10. PHOSIDA 2011: the posttranslational modifi-
1016/j.jmb.2017.10.014 cation database. Nucleic Acids Res 39(Data-
83. Caballero MC, Alonso AM, Deng B, Attias M, base issue):D253–D260. https://doi.org/10.
de Souza W, Corvi MM (2016) Identification 1093/nar/gkq1159
of new palmitoylated proteins in toxoplasma 90. Dinkel H, Chica C, Via A, Gould CM, Jensen
gondii. Biochim Biophys Acta 1864 LJ, Gibson TJ, Diella F (2011) Phospho.ELM:
(4):400–408. https://doi.org/10.1016/j. a database of phosphorylation sites—update
bbapap.2016.01.010 2011. Nucleic Acids Res 39(Database issue):
84. Chen YJ, Lu CT, Lee TY, Chen YJ (2014) D261–D267. https://doi.org/10.1093/nar/
dbGSH: a database of S-glutathionylation. gkq1104
Bioinformatics 30(16):2386–2388. https:// 91. Maurer-Stroh S, Koranda M, Benetka W,
doi.org/10.1093/bioinformatics/btu301 Schneider G, Sirota FL, Eisenhaber F (2007)
85. Chen YJ, Lu CT, Su MG, Huang KY, Ching Towards complete sets of farnesylated and ger-
WC, Yang HH, Liao YC, Chen YJ, Lee TY anylgeranylated proteins. PLoS Comput Biol 3
(2015) dbSNO 2.0: a resource for exploring (4):e66. https://doi.org/10.1371/journal.
structural environment, functional and disease pcbi.0030066
association and regulatory network of protein 92. Tung CW (2012) PupDB: a database of pupy-
S-nitrosylation. Nucleic Acids Res 43(Database lated proteins. BMC Bioinformatics 13:40.
issue):D503–D511. https://doi.org/10. https://doi.org/10.1186/1471-2105-13-40
1093/nar/gku1176 93. Zhang X, Huang B, Zhang L, Zhang Y,
86. Duan G, Li X, Kohn M (2015) The human Zhao Y, Guo X, Qiao X, Chen C (2012) SNO-
DEPhOsphorylation database DEPOD: a base, a database for S-nitrosation modification.
2015 update. Nucleic Acids Res 43(Database Protein Cell 3(12):929–933. https://doi.org/
issue):D531–D535. https://doi.org/10. 10.1007/s13238-012-2094-6
1093/nar/gku1009 94. Hasan MM, Yang S, Zhou Y, Mollah MN
87. Maurer-Stroh S, Gouda M, Novatchkova M, (2016) SuccinSite: a computational tool for
Schleiffer A, Schneider G, Sirota FL, the prediction of protein succinylation sites by
Wildpaner M, Hayashi N, Eisenhaber F exploiting the amino acid patterns and proper-
(2004) MYRbase: analysis of genome-wide gly- ties. Mol BioSyst 12(3):786–795. https://doi.
cine myristoylation enlarges the functional org/10.1039/c5mb00853k
spectrum of eukaryotic myristoylated proteins. 95. Chernorudskiy AL, Garcia A, Eremin EV,
Genome Biol 5(3):R21. https://doi.org/10. Shorina AS, Kondratieva EV, Gainullin MR
1186/gb-2004-5-3-r21 (2007) UbiProt: a database of ubiquitylated
88. Gupta R, Birch H, Rapacki K, Brunak S, Han- proteins. BMC Bioinformatics 8:126. https://
sen JE (1999) O-GLYCBASE version 4.0: a doi.org/10.1186/1471-2105-8-126
 Chapter 15

Identification of Unexpected Protein Modifications by Mass

Spectrometry-Based Proteomics
Shiva Ahmadi and Dominic Winter

Abstract
Peptide identification relies in the majority of mass spectrometry-based proteomics experiments on match-
ing of experimental data against peptide and fragment ion masses derived from in silico digests of protein
databases. One of the main drawbacks of this approach is that modifications have to be defined for database
searching and therefore no unexpected modifications can be identified in a standard setup. Consequently, in
many bottom-up proteomics experiments, unexpected modifications are not identified, even if high-quality
fragment ion spectra of the modified peptides were acquired. It is therefore often not straightforward to
identify unexpected modifications. In this protocol, we describe a stepwise procedure to identify unex-
pected modifications at peptides using the database search algorithm Mascot. The workflow includes
parallel searches for the identification of known modifications at unexpected amino acids, error tolerant
searches for modifications unexpected in the sample but known to the community, and mass tolerant
searches for entirely unknown modifications. Furthermore, we suggest a follow-up strategy consisting of
(1) verification of identified modifications in the initial dataset and (2) targeted experiments using synthetic
peptides.

Key words Mass spectrometry, Unexpected modifications, Posttranslational modifications, Bottom-

up proteomics, Data analysis, Mascot, Error tolerant search, Mass tolerant search

1 Introduction

Mass spectrometry has become the most versatile and powerful

technique for the identification, quantification, and characteriza-
tion of proteins. In the large majority of studies, the so-called
bottom-up approach is used which employs proteolytic digestion
of proteins followed by analysis of the resulting peptides using
MALDI-MS or MS/MS, as well as LC-ESI-MS/MS [1]. In this
method, MS/MS data are most often collected in the data-
dependent acquisition (DDA) mode. In DDA, the instrument is
set up to select the most abundant (multiply charged) precursor
ions for fragmentation [2]. The resulting MS/MS spectra are then
usually identified by matching against theoretical spectra originat-
ing from in silico digests of protein databases [3]. Despite the

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_15, © Springer Science+Business Media, LLC, part of Springer Nature 2019

225
226 Shiva Ahmadi and Dominic Winter

widespread use of this approach, a large number of spectra

(on average 75%) remain unidentified in a standard proteomics
experiment [4]. One major factor preventing the assignment of
spectra is unexpected modifications of peptides, either due to post-
translational modifications (PTMs [5]) or chemical modifications,
which are introduced during sample preparation [6]. Up to date,
several hundreds of these modifications are known and can be
found in specific databases like Unimod (www.unimod.org [7]).
Unexpected modifications prohibit straightforward identification
of the peptides as they alter their precursor mass and therefore
prevent them from being matched to the theoretical masses calcu-
lated for the in silico digest [8]. Dependent on the experiment, it
may be of interest to identify these unknown modifications. Unex-
pected PTMs, for example, could result in the discovery of cellular
regulatory mechanisms, which may be of high importance with
respect to the investigated biological question. Unintended chemi-
cal modifications, on the other hand, can affect the performance of
the experiment, and are therefore usually undesired. Their identifi-
cation could allow adapting the parameters for sample preparation
and/or database searching and accordingly increasing the number
of identified peptides. Furthermore, once a modification is identi-
fied, characterization of the modified peptides can be improved
using a range of well-established approaches for the fractionation
of peptides due to their physicochemical properties. These include
for example affinity capture using antibodies, columns, or beads
[9], and fractionation by chromatography techniques [10].
While expected modifications are often easily identified by
defining them as variable modification for database searching, the
identification of unknown modifications is usually not straightfor-
ward and requires further data analysis and validation. In this
chapter, we provide a strategy for the discovery of unexpected
modifications in complex samples by bottom-up LC-ESI-MS/
MS. We suggest points to be considered for the preparation and
analysis of such samples and exemplify the identification of unex-
pected modifications by different strategies using the commonly
used peptide search engine Mascot (www.matrixscience.com). We
demonstrate the workflow based on a recent study from our group
addressing unspecific modifications resulting from reduction and
alkylation of proteins in proteomics experiments [11].

2 Materials

2.1 Cell Culture Cell culture medium: Dulbecco’s modified Eagle medium
(DMEM) supplemented with 10% FCS, 100 U/mL penicillin,
and 100 μg/mL streptomycin.
 Identification of Unexpected Protein Modifications by Mass Spectrometry. . . 227

2.2 Cell Harvest, 1. Phosphate Buffered Saline (PBS, 1
).

Lysis, and Preparation 2. Sucrose buffer: 250 mM sucrose, 15 mM KCl, 1.5 mM MgAc,
of Cytosolic Fraction 10 mM HEPES, 1
Protease Inhibitor Cocktail (see Note 1).
3. Acetone.
4. Lysis buffer: 0.1 M Tris HCl/ 4% SDS.
5. A kit for the determination of the protein concentration of a
sample.
6. Dounce homogenizer.
7. Cell scraper.
8. Tabletop centrifuge.
9. Ultracentrifuge.

2.3 In-Gel Reduction, 1. Thermomixer.

Alkylation, and 2. Sample loading buffer (modified from [12]): 0.25 M Tris–HCl
Digestion pH 6.8, 8% (w/v) SDS, 40% (v/v) glycerol, and 0.004% (w/v)
bromophenol blue.
3. Reduction solutions: 20 mM dithiothreitol (DTT), 20 mM
tris-(2-chlorethyl)-Phosphate (TCEP), 40 mM
β-mercaptoethanol (BME) in 0.1 M NH4HCO3.
4. Alkylation solutions: 55 mM iodoacetamide (IAA), iodoacetic
acid (IAC), acrylamide (AA), chloroacetic acid CAA in 0.1 M
NH4HCO3.
5. 10% SDS-PAGE gel.
6. Coomassie brilliant blue.
7. Gel destaining solution: 30% acetonitrile (ACN), 70 mM
NH4HCO3.
8. Wash solution: 0.1 M NH4HCO3.
9. Trypsin solution: 0.1 μg/μL sequencing grade trypsin in 0.1 M
NH4HCO3.
10. Peptide extraction solutions.
(a) 0.1% trifluoroacetic acid (TFA), 50% ACN.
(b) 0.1 M NH4HCO3.
(c) 100% ACN.
11. Resuspension solution: 0.01% acetic acid 3% ACN.
12. MS sample buffer: 5% ACN, 5% formic acid (FA).
13. Vortex.

2.4 Dimethyl 1. Dimethylation solution 1: Light label: 4% (v/v) CH2O,

Labeling medium label: 4% (v/v) CD2O, heavy label: 4% (v/v)
13
CD2O (see Note 2, [13]).
228 Shiva Ahmadi and Dominic Winter

2. Dimethylation solution 2: Light label: 0.6 M NaBH3CN,

intermediate label: 0.6 M NaBH3CN, heavy label: 0.6 M
NaBD3CN (see Note 2, [13]).
3. Quenching solution: 1% (v/v) NH4OH.
4. Formic acid (FA).
5. Vacuum Centrifuge.
6. Fume hood.
7. Thermomixer.

2.5 LC-MS/MS 1. C18 analytical column: ESI spray tip produced in house with a
Measurements Sutter Instruments P2000 laser puller from a 360 μm outer
diameter, 100 μm inner diameter fused silica capillary and
packed with 5 μm particles [Dr. Maisch, Reprosil C-18 AQ].
Alternatively, any type of commercially available nanoflow C18
column in combination with an appropriate emitter can
be used.
2. Thermo Scientific EASY-nLC 1000 or similar nanoflow high/
or ultrahigh-performance liquid chromatography system.
3. Thermo Scientific Orbitrap Velos Mass Spectrometer or any
other high-resolution / high accuracy mass spectrometer.
4. Solvent A: Water with 0.1% FA.
5. Solvent B: ACN with 0.1% FA.

2.6 Software 1. Mascot (www.matrixscience.com).

2. Proteome Discoverer (Thermo Scientific).

3 Methods

To be able to identify unexpected modifications, it is essential that

the modified peptides are fragmented by the mass spectrometer.
Peptide signals are usually selected for fragmentation in the DDA
mode based on their intensity in the survey spectrum. Since it is
unlikely that the modified peptides are among the most abundant
peptides in the sample, it is advisable to use samples of medium to
low complexity. This results in over- rather than under-sampling
and is therefore more likely to yield fragment ion spectra from
modified peptides. When analyzing single protein bands from
SDS-gels or purified proteins, the complexity is usually sufficiently
low. If, however, cells or tissues are to be analyzed at a proteome-
wide scale, the sample should be divided into several fractions either
at the protein or peptide level. Methods commonly employed for
fractionation before proteolytic digestion include SDS-PAGE
(in combination with in-gel digestion) [14], as well as size exclu-
sion [15] or ion exchange chromatography [16], followed by
 Identification of Unexpected Protein Modifications by Mass Spectrometry. . . 229

proteolytic digestion of the resulting fractions. On the peptide

level, basic reversed-phase high-performance liquid chromatogra-
phy (RP-HPLC) [17], strong anion exchange chromatography
(SAX) [18], strong cation exchange chromatography (SCX) [19],
or isoelectric focusing (IEF) [20] are the available techniques for
fractionation of digested peptides. Dependent on the desired depth
of analysis, protein and peptide fractionation can be combined.
Furthermore, subcellular fractionation (e.g., by differential centri-
fugation or enrichment of organelles) is another approach to gen-
erate samples of reduced complexity [21].
In the current study, we used a cytosolic fraction of HeLa cells,
separated the proteins by SDS-PAGE according to their molecular
weight, and performed in-gel digestion of a subset of proteins (rang-
ing from 30 to 50 kDa). Additionally, we performed in-solution
digestion of the whole sample followed by fractionation using SAX
stop and go extraction (STAGE) tips [22]. As the strategy for
LC-MS/MS measurement and data analysis is similar for both
types of sample processing, we will restrict our descriptions in this
chapter to the analysis of the in-gel digested samples. The details
concerning the in-solution digested samples can be found in [11].

3.1 Sample If the choice of the specific subcellular fraction is not influenced by
Preparation: In-Gel a biological question and it serves only as a source of proteins, the
Digestion of a cytosolic fraction is the easiest to generate, as the other parts of the
Cytosolic HeLa cell can be removed by centrifugation. Furthermore, the cytosolic
Fraction fraction is of adequate complexity in order to allow a sufficient
coverage of the proteins present in the sample and to provide a
reasonable number of proteins in order to resemble the complexity
observed in medium to large scale proteomics studies. When
planning the experiment, care should be taken that the differential
treatment of the samples is not influencing the subsequent steps,
possibly resulting in secondary effects. For this reason, we decided
to perform reduction and alkylation after performing SDS-PAGE as
differential modification of the proteins may influence their run-
ning behavior in the gel electrophoresis.
1. Place the cell culture dishes on ice, aspirate the medium, and
wash the cells 3
with 10 mL ice-cold PBS (see Note 3).
2. Add 1 mL sucrose buffer, detach the cells using a cell scraper,
and transfer the cell suspension to a dounce homogenizer (see
Note 4).
3. Homogenize cells on ice using the dounce homogenizer
(30 strokes) and transfer the suspension to a 2 mL microtube
(for higher volumes 15 or 50 mL conical tubes can be used).
4. Centrifuge at 1000
g, 4  C for 10 min in order to pellet
intact cells and nuclei. Transfer the supernatant to a new 2 mL
microtube.
230 Shiva Ahmadi and Dominic Winter

5. Resuspend the pellet from the first tube in 1 mL sucrose buffer

and repeat steps 3 and 4.
6. Combine the supernatants from both homogenization steps,
transfer them to an appropriate ultracentrifugation tube, and
centrifuge at 100,000
g, 4  C for 1 h to pellet cell organelles
and membranes. The clear supernatant represents the cytosolic
fraction.
7. In order to precipitate proteins, mix the supernatant (cytosolic
fraction) with acetone (20  C) in a ratio of 1–4 (v/v), vortex
for 30 s, and incubate at 20  C overnight (see Notes 5 and 6).
8. Centrifuge samples at 20,000
g, 4  C for 30 min, discard the
supernatant, and air-dry the pellet at 23  C.
9. Resuspend the pellet in the lysis buffer by heating at 95  C for
5 min, centrifuge at 20,000
g, RT for 30 min, and transfer
the clear supernatant to a new tube (see Note 7).
10. Determine the protein concentration using a protein assay.
11. Perform SDS-PAGE, stain the gel using Coomassie brilliant
blue (several hours to overnight, dependent on sample
amount) and destain the gel using MilliQ water.
12. Excise the region to be analyzed from the gel and cut it into
small pieces of ~1 mm2. If PTMs in general are to be analyzed
and no specific protein is targeted, cut the whole gel lane to
pieces of similar complexity (as determined by the number of
Coomassie stained bands). If unspecific modifications due to
chemical sample treatments are to be analyzed, a single section
of the gel is usually sufficient. In our example study we selected
a region between 30 and 50 kDa.
13. Destain the gel using 500 μL gel destaining solution at 25  C,
800 rpm for 30 min in a thermomixer and discard the liquid.
Repeat this step until the gel pieces are colorless.
14. Reduce disulfide bonds using 100 μL of reduction solution for
45 min at 56  C, 800 rpm.
15. Remove the reducing reagent and alkylate thiol groups by
adding 100 μL of alkylation solution. Incubate for 30 min at
RT in the dark.
16. Wash the gel pieces using 500 μL wash solution for 15 min at
800 rpm. Discard the liquid.
17. Dehydrate the gel pieces with 100% ACN for 15 min at
800 rpm. Discard the liquid and dry the gel pieces using a
vacuum centrifuge.
18. Add 10 μL of trypsin solution, incubate for 15 min, and then
add a sufficient volume of wash solution to cover the gel pieces.
Control after 30 min if all gel pieces are covered and add more
wash solution if needed. Incubate overnight at 37  C.
 Identification of Unexpected Protein Modifications by Mass Spectrometry. . . 231

19. Extract peptides consecutively using 50 μL of peptide extrac-

tion solutions A, B, and C for 15 min each at 25  C, 800 rpm
and pool the peptide extracts (see Note 8). Dry the sample
using a vacuum centrifuge.
20. Reconstitute the samples in 100 μL resuspension solution and
desalt, e.g., using STAGE tips [22].
21. Dry the eluate fractions using a vacuum centrifuge and recon-
stitute in 20 μL MS sample buffer.

3.2 LC-MS/MS Data For the identification of unexpected modifications, it is beneficial to

Acquisition measure the samples with high mass accuracy. This allows in the
subsequent steps to analyze the data with a small mass tolerance
window reducing the analysis time and probability of false positive
assignments. We therefore advice, if possible, to use Orbitrap or
QTOF instruments which provide high resolution and mass accu-
racy at a good sensitivity and scan speed. If only ion trap instru-
ments are available, it should be considered to use (at least for the
survey scan of the intact peptide ions), the zoom or ultra-zoom
scan mode and to apply longer LC gradients to compensate for the
lower survey scan speed. In our example study, we used a LTQ
Orbitrap Velos mass spectrometer acquiring the survey scans for the
intact precursor ions in the Orbitrap part of the instrument and the
fragment ion spectra by collision induced dissociation (CID) in the
ion trap. We chose this option over fragmentation by higher energy
collisional dissociation (HCD), followed by measurement of the
fragment ions in the Orbitrap, since in our experience the increased
mass accuracy of the HCD spectra cannot compensate for the
reduction in sensitivity and scan speed compared to the LTQ in
this instrument.
1. Set up the gradient for the reversed-phase (RP) chromatography
system in a way that the samples are rather over-sampled. This
means, the instrument should be able to acquire more MS/MS
spectra than necessary for the complexity of the sample which in
turn increases the chance of identifying low-abundant modified
peptides (see Note 9).
2. If the sample is too complex, an increase in the number of
MS/MS spectra performed for each survey scan may allow for
a more efficient identification of low-abundant peaks. This will
come at a cost of deterioration of chromatographic peak
shapes. Additionally, the identification rate of low-abundant
peptides can be increased by performing technical replicates
and/or definition of exclusion lists, while the latter was in our
hands not very efficient.
3. Set the capillary voltage to the lowest possible value to prevent
extensive in-source fragmentation as this may result in an
increase of spectrum complexity. We use, if the spray tip allows,
1.6 kV in the positive ion mode.
232 Shiva Ahmadi and Dominic Winter

4. Set the repeat count to one and the dynamic exclusion window
to a value larger than the average chromatographic peak width
at the basis in order to prevent repetitive fragmentation of
highly abundant ions.
5. If certain neutral losses are expected (e.g., 98 for phosphory-
lation) define them in the MSA scan setup (in case an Orbitrap
mass spectrometer is used).
6. If available, activate the lock mass option in order to achieve
maximal mass accuracy.
7. If possible, use either small RP particles or avoid the use of a
trap column to achieve maximal chromatographic resolution.
This results in increased signal intensities due to narrower peak
shapes aiding the identification of low-abundant species.
8. Perform nano(U)HPLC-MS/MS analysis of the sample.

3.3 Data Analysis Dependent on the availability of sample material and controls, differ-
ent strategies can be taken. Figure 1 shows the workflow for identifi-
cation of unexpected modifications, including possible steps and the
order of their execution. There are several algorithms available to
perform such analyses; we used Mascot (www.matrixscience.org)
either in combination with Proteome Discoverer (Thermo Scientific)
or individually by manual submission of mascot generic files (mgf).
Alternatively, dependent on their features other programs like Max-
Quant, SEQUEST, PEAKS DB, ProteinPilot, pFind, Byonic, or X!
Tandem can be used [23].

3.3.1 Initial Analyses In case of modifications due to a chemical treatment (in our case
Defining Known unspecific alkylation of amino acid side chains), whose composition
Modifications at All Amino is known but not the amino acid residue(s) modified, the possible
Acids modification(s) should initially be searched at all possible amino
acids. As the number of variable modifications which can be defined
in a Mascot search is limited to 9 at a time, three individual searches
have to be performed. This results in redundant assignments of
MS/MS spectra for the unmodified peptides in the single searches.
Furthermore, if the fragment ions are not conclusive for assigning
the modification to a distinct amino acid, the same spectrum may be
annotated with different modification positions in the separate
searches. Proteome discoverer allows dealing with these issues by
combining the results of the individual searches in one FDR analysis.
1. Extend the specificity definition of the modification in the
Mascot configuration editor to all possible amino acids and
define them as active so they can be selected for searching. In
our study, we specified carbamidomethyl (for IAA and CAA),
carboxymethyl (for IAC), and propionamide (for AA) at all
amino acids.
Identification of Unexpected Protein Modifications by Mass Spectrometry. . . 233

Fig. 1 Data analysis strategy for the identification of unexpected modifications.

Dependent on the availability of a control sample, different possible routes can
be taken. Identified novel modifications are finally validated by manual
inspection of the datasets and synthetic peptide experiments. PSM: peptide
spectral match; T  C: Number of PSMs in the treated sample is slightly lower or
similar compared to the control sample; T C: Number of PSMs in the treated
sample is significantly lower compared to the control sample

2. Create a workflow with three parallel Mascot searches, each

containing one subset of modifications.
3. Merge the results in a single percolator processing step in order
to allow FDR calculation for the combined results and to
remove redundantly assigned MS/MS spectra.
234 Shiva Ahmadi and Dominic Winter

If Proteome Discoverer is not available, alternatively, the

redundant assignments can be removed using MS Excel. Due to
the high number of variable modifications, this search strategy will
almost always result in the identification of modifications at every
amino acid. In order to estimate the credibility of the modifications,
manual analysis of the type of modified amino acid has to be
performed. In our example study, for instance, we found alkylation
at several amino acids with side chains which are highly unlikely to
react with the alkylation reagents like alanine, leucine, or valine
(Fig. 2a, b). Manual analysis revealed that all of these amino acids
were located at the peptide N-terminus and that the observed
modification was located at the N-terminal amino group and not
at the peptide side chain. It became apparent that only seven amino
acids as well as the peptide N-terminus were modified. We then

Fig. 2 Amino acids found to be modified by the reduction and alkylation procedure. Shown are the combined
results from six replicates. (a, b) Investigation for off-site alkylation at all possible amino acids in samples
reduced with DTT and alkylated with (a) AA (DA) and (b) IAA (DIA). Manual analysis of (b) revealed frequent
alkylation of seven amino acids and the peptide N-terminus. (c, d) PSMs annotated to be off-site (c) mono-
alkylated and (d) di-alkylated at Y, S, D, T, H, E, K, and the peptide N terminus (reproduced from [11]). DI
DTT/IAA, DIA DTT/IAC, DA DTT/AA, DC DTT/CAA, TI TCEP/IAA, TIA TCEP/IAC, TA TCEP/AA, TC TCEP/CAA, MI
BME/IAA, MIA BME/IAC, MA BME/AA, MC BME/CAA, N-term peptide N terminus
 Identification of Unexpected Protein Modifications by Mass Spectrometry. . . 235

performed a second round of searches, this time only including

such amino acids, which allowed us to identify a high number of
unspecific offside alkylation by monomers and dimers of the alkyl-
ation reagents (Fig. 2c, d).

3.3.2 Estimating the If the database searches with known modifications do not yield any
Abundance of Unexpected conclusive results, a reasonable initial step is to investigate which
Modifications unmodified peptides are depleted from the sample as a consequence
of the investigated treatment. Therefore, if possible, the sample
exposed to the biological stimulus or chemical treatment should
be compared to an untreated control sample (see Note 10). This
allows estimating the extent of the modification(s) either due to the
lack of identification of unmodified peptides or their reduction in
signal intensity. In our example study, we compared the samples
treated with different alkylation reagents to an untreated control.
Each sample was generated in two independent replicates and then
each replicate measured three times by LC-MS/MS in order to
exclude false conclusions due to DDA-dependent variation in pep-
tide identification. This high degree of redundancy allowed us to
assess if the observed differences are random or follow a systematic
pattern.
1. If needed, convert MS raw files to an appropriate format (see
Note 11).
2. Define parameters for the initial database search.
(a) The precursor and fragment ion tolerance should be as
narrow as possible, dependent on the performance of the
mass spectrometer used. For most high-resolution mass
spectrometers, a precursor mass tolerance of 10 ppm is
reasonable (also applied for our dataset). For fragment
ions, 50 mmU should be selected for high-resolution
MS/MS spectra (Orbitrap and QTOF instruments) and
0.6–0.8 Da for low resolution fragment ion spectra
acquired in the LTQ part of the instrument (see Note 12).
(b) Choose a database and define the taxonomy based on the
organism used. When searching for unexpected modifica-
tions, it is advisable to use rather small high confidence
databases like, e.g., SwissProt (see Note 13).
(c) Define a way of false discovery rate (FDR) estimation; we
usually use Percolator [24] (which is included in Prote-
ome Discoverer) and set the FDR to 1% on the peptide
level.
(d) Based on the alkylation reagent, define the respective fixed
modification at cysteine, as well as the expected variable
modifications (e.g., oxidation of methionine) (see Note 14).
236 Shiva Ahmadi and Dominic Winter

3. Export the search results as *.csv file and import them in MS

Excel (see Note 15). Calculate the percentage of MS/MS
spectra assigned to peptide sequences passing the threshold
for 1% FDR relative to the total number of acquired MS/MS
spectra and compare the different conditions. There are three
scenarios:
(a) The number of acquired MS/MS spectra decreases but
the identification rate does not change significantly. This
implies that less peptides were generated and interference
of the treatment with proteolytic digestion should be
considered (see Note 16).
(b) The number of acquired MS/MS spectra is constant and
the identification rate is reduced. This indicates the pres-
ence of unassigned modifications.
(c) The number of MS/MS spectra is constant and the iden-
tification rate does not vary considerably. In such a case
the modification may only affect a small fraction of the
peptides and high-abundance peptides are reduced in
intensity, but their identification efficiency is not affected.
In this case either sample complexity has to be reduced or
relative quantification of peptide signal intensities has to
be performed (see Subheading 3.3.3).

In our example study, we analyzed the identification rates of

MS/MS spectra for the different alkylation reagents using standard
database searching conditions. Figure 3 shows the resulting
assigned peptide spectral matches (PSMs) and the relative identifi-
cation rates. Dependent on the alkylation reagent, the percentage
of identified spectra varied strongly. This implied that, dependent

a) b)
10000

8000
40
% Identification
# of PSMs

6000

4000 20

2000

0 0
I TI I I TI I
D IA DA DC A A
TI T TC M IA A C N D IA DA D
C
TI
A TA TC M IA MA MC ON
D M M M CO D M C
conditions conditions

Fig. 3 Results of the analyses of six replicates of in-gel digested cytosolic fractions of HeLa cells reduced and
alkylated with different combinations of reagents. (a) Number of peptide spectral matches (PSMs); (b)
percentage of identified spectra (reproduced from [11]). DI DTT/IAA, DIA DTT/IAC, DA DTT/AA, DC DTT/CAA,
TI TCEP/IAA, TIA TCEP/IAC, TA TCEP/AA, TC TCEP/CAA, MI BME/IAA, MIA BME/IAC, MA BME/AA, MC BME/CAA,
Con control sample
 Identification of Unexpected Protein Modifications by Mass Spectrometry. . . 237

on the treatment, different portions of unexpectedly modified

peptides were present in the individual samples as these peptides
were fragmented but not identified.

3.3.3 Relative On the PSM level, the influence of a modification can only be
Quantification of Signal identified if the investigated treatment results in a strong reduction
Reduction of the peptide. If it, however, only results in the modification of a
small percentage of the peptides, it may be difficult to realize the
effect based on the PSM number; as also ions reduced in abundance
are usually sufficient to result in MS/MS spectra allowing for
peptide assignment. In such cases, relative quantification can be
used in order to identify changes and to estimate the extent of
modification. While the easiest approach is a label-free quantifica-
tion by calculation of the area under the curve, a variety of stable
isotope labeling-based methods is available which allows for more
accurate quantification. They can be grouped in methods inducing
stable isotope labels through metabolic incorporation (15N/14N
metabolic labeling and SILAC), chemical derivatization (ICAT,
iTRAQ, TMT), or during proteolytic processing (18O labeling)
[25]. As the labels are introduced at different steps of the protocol
by the varying methods, the method of choice is dependent on the
experiment. In our example study, we applied dimethyl labeling as it
allowed us to use the same starting material for all conditions and to
label the peptides after reduction and alkylation with the different
reagents.
1. After sample treatment and proteolytic digestion, desalt the
peptides, dry them using a vacuum centrifuge, and resuspend
them at 1 μg/μL in 100 mM HEPES (pH 5–8.5) (see Note
17).
2. For 100 μL of sample volume, add 16 μL of light, medium, or
heavy dimethylation solution 1 to the respective samples.
3. Add 16 μL of light, medium, or heavy dimethylation solution
2 to the samples followed by incubation in the fume hood for
1 h at RT while mixing at 800 rpm.
4. Quench the labeling with 64 μL of dimethyl quenching solu-
tion/100 μg sample.
5. Add 32 μL of FA for acidification of the sample.
6. Combine the different labeled samples, desalt them using
STAGE tips [22], and proceed to MS analysis (see Note 18).
7. Process the raw files using the relative quantification by
dimethyl labeling option in Proteome Discoverer. Exclude
peptides containing the presumably modified amino acids
(e.g., cysteine in this study) from the list of the results as they
exhibit different masses and chromatographic retention times
due to the different modified groups. Also exclude samples
with missing values in one of the channels (see Note 19).
238 Shiva Ahmadi and Dominic Winter

a) b)
8 IAA / Control 8 AA / Control
6 6
4 4
Log2 fold change

2 2
0 0
-2 -2
-4 -4
-6 -6
-8 -8
500 1000 1500 500 1000 1500
Pepde # Pepde #

Fig. 4 Quantification of changes in peptide abundance due to differential alkylation. Shown are the log2 values
of the average fold-change from two independent experiments. Proteins were reduced and alkylated in the
gel, digested by trypsin, the individual peptide samples labeled by the dimethylation reagents (IAA: light, AA:
medium, control: heavy) and combined. PSMs with a negative fold change between 1 and 2 are indicated in
light red and those with >twofold in dark red. PSMs with a positive fold change between 1 and 2 are colored
light green and >twofold dark green (reproduced from [11])

8. Investigate peptide abundance ratios between the different

treatments and the control sample in order to identify peptides
which decrease significantly in abundance.
In order to determine the significance cutoff, it is often suffi-
cient to plot the normalized treated/control ratios in log2-scale if
not enough replicates are available to calculate p-values. This allows
to visually determine a reasonable cutoff as the majority of peptides
should fall in one unregulated group. In our example study, where
samples alkylated with IAA were labeled with the light, samples
alkylated with AA with the medium, and the untreated control
samples with the heavy label, respectively, we applied a >twofold
change in abundance cutoff. The results are displayed in Fig. 4 (see
Note 20). In these data, downregulation of the unmodified version
of the peptide implies the upregulation of its modified form.

3.3.4 Identification of The MS/MS identification rate data (Subheading 3.3.2) or the
Affected Amino Acids or relative peptide abundances (Subheading 3.3.3) can now be used
Functional Groups to determine if the affected amino acids follow a certain pattern.
(at Peptide Termini or For this purpose, all peptides which were either absent or signifi-
Specific Side Chains) cantly downregulated as a consequence of the treatment are ana-
lyzed for their amino acid composition. If the unknown
modification is specific for a certain amino acid or functional
group of the peptide/protein, as it usually is the case, the affected
amino acid is expected to be overrepresented in the regulated
population of peptides. This analysis can easily be performed in
MS Excel:
 Identification of Unexpected Protein Modifications by Mass Spectrometry. . . 239

1. Copy the peptide sequences (excluding any annotations for

variable modifications) of all regulated peptides in one Excel
column.
2. In the adjacent 20 columns, define the following function:
¼LEN (PS)-LEN (SUBSTITUTE (PS;"AmAc ";"")), where
PS is the cell containing the peptide sequence and AmAc is the
amino acid of interest.
3. Summarize the abundance of each individual amino acid and
normalize it to the total number of amino acids found in the
whole data set.
In our example study, we performed this analysis for both the
initial dataset and the dimethyl-based relative quantification data-
set. The results are shown in Fig. 5. The analysis revealed that the
only amino acid strongly affected by the alkylation procedure was
methionine allowing us to narrow down our search window of the
residues affected by the unexpected modification.

3.3.5 Database Search A number of computational strategies have been developed to

Strategies for Unexpected search in complex samples for unexpected modifications. In Mas-
Modifications cot, there are two approaches available: error tolerant [26] and
mass tolerant [7] searches. While error tolerant searches have
been established more than a decade ago and can be selected in
the Mascot search interface (not through Proteome Discoverer),
mass tolerant searches have been introduced more recently. In error
tolerant searches, Mascot generates a reduced database containing
proteins which have been identified in a normal first pass search.
This database is then used for the error tolerant search in which all
modifications contained in the Unimod database (www.unimod.
org) are searched successively. Furthermore, point mutations,
changes in reading frames, and relaxed enzyme specificity are
accounted for by the algorithm. In mass tolerant searches, no
variable modifications are defined and therefore also modifications
which are not listed in Unimod can be identified. Instead, a large
mass tolerance window (up to 250 Da) for the precursor ion is
defined. This will then result in peptides which are identified as
being unmodified along with a mass error which represents the
mass of the unexpected modification. For both search strategies,
the raw files have to be converted in a file format which is accepted
by the Mascot MS/MS Ions Search interface. We usually use the
mascot generic format (*.mgf) and generate for each condition one
merged *.mgf file using Proteome Discoverer with a top N setting
of 6 in 100 Da. If Proteome Discoverer is not available, several
other algorithms can be used for this purpose (see Note 21).

Error Tolerant Search Error tolerant searching can either be executed in the initial search
by selecting the respective option in the MS/MS Ions Search
interface or on the results page of any MS/MS ion search. The
240 Shiva Ahmadi and Dominic Winter

a)
2

normalized log2 abundance

1

-1

-2
A C D E F G H I K L M N P Q R S T V W Y

b)
2
normalized log2 abundance

-1

-2
A C D E F G H I K L M N P Q R S T V W Y

DI DA TI TA MI MA C ON
D IA DC T IA TC M IA MC

Fig. 5 Normalized amino acid abundance in (a) peptides regulated in the

dimethylation experiment and (b) PSMs regulated in the whole initial dataset.
Abundances are calculated based on the over-/underrepresentation of the
indicated amino acid in the group of regulated peptides/PSMs with respect to
the distribution of all amino acids in the corresponding whole dataset (repro-
duced from [11]). DI DTT/IAA, DIA DTT/IAC, DA DTT/AA, DC DTT/CAA, TI TCEP/
IAA, TIA TCEP/IAC, TA TCEP/AA, TC TCEP/CAA, MI BME/IAA, MIA BME/IAC, MA
BME/AA, MC BME/CAA, Con control sample
 Identification of Unexpected Protein Modifications by Mass Spectrometry. . . 241

first option results in Mascot performing error tolerant searches for

all proteins which have been identified in the first pass search. In our
experience, this is only advisable if the samples are not of high
complexity, as the search is otherwise likely to occupy the server
for an extremely long time. If such searches are to be performed
nevertheless, the server time-out (which is 1 day by default) should
be adjusted accordingly to allow the search to finish. Therefore,
often the second alternative is more practicable. In order to be able
to select specific proteins from the results, they have to be displayed
in the select summary format (see Note 22). In our example study,
we selected the top 20 proteins for error tolerant searching.

Mass Tolerant Search Mass tolerant searches have to be performed on the whole dataset.
If they result in too much server occupancy, the *.mgf files can be
generated from fewer *.raw files, or if the *.raw files are too large,
they can be divided in smaller packages. This has to be determined
based on the capacity of the server used. The wide-tolerance win-
dow for the precursor mass defined in the mass tolerant search
(250 Da in our example study) cannot be defined in the Mascot
MS/MS Ions Search interface. Therefore, the *.mgf file has to be
modified manually.
1. Create the *.mgf file and open it using an appropriate software
(e.g., Wordpad or Notepad++).
2. Modify the header of the *.mgf file to include the wide mass
tolerance of your choice. In our example study, we changed the
header to: peptide mass tolerance: 250 Da; fragment ion toler-
ance: 0.8 Da; search type: SQ; fixed modifications: carbamido-
methyl, carboxymethyl, or propionamide, respectively, at
cysteine residues depending on the alkylation reagent.
3. Perform Mascot MS/MS ion searches using the modified *.
mgf file. The mass tolerance settings in the *.mgf file overwrite
the settings selected in the MS/MS Ions Search interface;
therefore the default settings can be kept.

Interpretation of Error Both the error tolerant and the mass tolerant searches result in high
Tolerant and Mass Tolerant numbers of false positive identifications and are not compatible
Search Results with decoy database searching. Therefore, great care has to be
taken which peptide matches are accepted. In such case, an arbitrary
cutoff can be selected; in our experience an ion score of 30 is a
reasonable compromise between selectivity and sensitivity. Alterna-
tively, a cutoff to achieve 1% FDR at the peptide level can be
determined in a first target/decoy database search. One should
be, however, careful as the value determined by this approach will
definitely be too low. Therefore, it should only be used as a ten-
dency rather than a strict cutoff.
242 Shiva Ahmadi and Dominic Winter

1. Export the search results with a reasonable ion score cutoff.

Open the results in MS Excel or any other software accepting
tab-delimited files.
2. Error tolerant searches:
(a) Count the frequency with which the modifications have
been annotated. A straightforward way to do this is using
Pivot Tables in Excel. Specific modifications should be
more abundant in comparison to the control sample.
(b) Correlate the presumably specifically modified peptides
with the list of affected peptides/amino acids found in
Subheading 3.3.4 (if applicable). This can either be
achieved by matching the peptide sequences themselves
or by filtering for peptides containing the amino acid
(s) found to be affected.
3. Mass tolerant searches:
(a) Transfer the peptide sequences and mass errors in Dalton
to an Excel sheet. Use the “frequency” function of Excel
to count the number of peptides falling within specific
mass ranges (5 Da is a reasonable value); count the num-
ber of peptides in these bins.
(b) Plot the results of the treated and the control sample and
compare the distribution of mass errors.
(c) If a bin with a certain mass error is identified to be over-
represented in the treated sample, extract the peptides in
this bin.
(d) Correlate these peptides with the data obtained in Sub-
heading 3.3.4.

For both the error and mass tolerant searches, the amino acids
found to be affected in Subheading 3.3.4 are expected to be present
in the identified modified peptides. If this is not the case, alterna-
tively the identification/quantification information for these pep-
tides can be extracted from the previous datasets to assess if they
were regulated. While the error tolerant search provides a sugges-
tion for the observed modification, the mass tolerant search only
gives a mass value which does not provide any explanation. In this
case, it would be ideal if it is possible to suggest modifications based
on the treatment investigated. If this is not applicable, a reasonable
first step is the determination of the modifications’ elemental com-
position. This can be achieved by calculation of its accurate mass
through the mean value of the mass error for all peptides found with
the respective modification. This then allows to suggest possible
sum formulas using the involved atoms (usually C, H, O, N, and S).
In our example study, error tolerant searches did not result in
the identification of any unusual modifications. In mass tolerant
 Identification of Unexpected Protein Modifications by Mass Spectrometry. . . 243

a)
2000
1500
1000
IAA/IAC
# PSMs

500
0
-500
AA/CAA
-1000
-1500
-2000

b)
2000 oxM
1500 -64
1000
IAA/IAC
# PSMs

500
0
-500
AA/CAA
-1000
-1500
-2000

Fig. 6 Combined results of the mass tolerant searches for two independent replicates of all reduction and
alkylation reagents. Shown are the summed abundances for peptide mass errors binned in groups of 5 Da for
all reduction reagents in combination with the iodine-containing alkylation reagents (IAA/IAC, upper panel), as
well as the non-iodine-containing reagents (AA/CAA, lower panel). (a) Searches performed without variable
modifications did not identify any differences between the treatments except of the already identified offside
alkylation (+57 Da); (b) searches including oxidation at Methionine (oxM) as variable modification resulted in
the assignment of ~1000 PSMs with a mass error of 64 Da. All of these PSMs contained methionine
(reproduced from [11])

searches, we initially also did not observe any differences except an

increased number of peptides with a mass error of 57 Da for the
iodine-containing alkylation reagents, which was due to the off-site
alkylation already detected in the previous searches (Fig. 6a). We
then repeated the mass tolerant searches with oxidation at methio-
nine as variable modification. This resulted in the identification of
~1000 PSMs annotated with a mass error of 64 Da in the samples
treated with iodine-containing alkylation reagents (Fig. 6b).
Intriguingly, almost all of these peptides contained methionine,
which was identified earlier (see Subheading 3.3.4) to be the main
affected amino acid. This implied that alkylation of peptides with
iodine-containing reagents results in a loss of the side chain of
244 Shiva Ahmadi and Dominic Winter

methionine (64 Da equals the molecular weight of the side chain of

oxidized methionine).
Theoretically, one would have expected to identify this modifi-
cation already in the first mass tolerant search, which was performed
without any variable modifications. Manual analysis revealed that
this was not the case, as the fragment ions including the modified
methionine residue were not identified by Mascot—their difference
in mass compared to the unmodified fragments interfered with
their identification. This resulted in a lower ion score and prevented
the respective peptides to pass our threshold for acceptance. It may
be therefore beneficial to perform several rounds of mass tolerant
searches including known modifications. This exemplifies a general
drawback of mass tolerant searches: The mass error observed is
solely determined from the precursor ion mass. As only unmodified
fragment ions are matched, the number of assigned b/y-ions is
reduced, which in turn decreases the peptide score. Additionally,
the MS/MS spectrum does not provide any confirmation of the
identified mass error. Therefore, results obtained from mass toler-
ant searches are more vulnerable toward false positive identifica-
tions resulting from, e.g., erroneously determined charge states or
co-isolation of another precursor ion in MS/MS data acquisition.
We therefore advise to only consider modifications identified in
mass tolerant searches which have been detected at significant
numbers and to also take lower scoring peptides into consideration
as the reduced number of matched fragment ions may be due to the
modification rather than the quality of the spectrum.

3.3.6 Manual Data The identified modifications have to be validated in order to pre-
Validation for Identified vent false conclusions. Due to the lack of an FDR analysis, this is
Modifications especially of importance in the case of error and mass tolerant
searches. Therefore, the initial database search should be repeated
with the newly discovered modification(s) as variable modification
at the respective amino acid(s). The modification can be defined in
the Mascot configuration editor. If the modifications’ elemental
composition could not be determined, one can alternatively use a
sum formula matching the detected mass error and increase the
search tolerance window to compensate for possible minor differ-
ences in the molecular weight of the modification. The amino acids
to be specified as modified are indicated in case of the error tolerant
searches; for the mass tolerant searches, however, as the mass errors
are only determined from the precursor ion, no suggestions can be
provided by Mascot. If no amino acid abundance analysis (see
Subheading 3.3.4) was performed or if it was inconclusive, another
set of parallel searches (see Subheading 3.3.1) can be performed in
order to narrow down the list of possibly affected amino acids.
Subsequently, several spectra with high ion scores should be eval-
uated manually by comparison of the fragment ion masses
 Identification of Unexpected Protein Modifications by Mass Spectrometry. . . 245

calculated by Mascot and the annotated spectrum. Alternatively,

manual de novo sequencing [27] can be performed by calculating
the theoretical fragment ion masses. If the assignment is not con-
clusive, a MS/MS spectrum of the matching unmodified peptide
fragmented in the control sample can be used for comparison—the
fragment ion series not carrying the identified modification should
match between both spectra. This allows to determine if the sug-
gested modification is indeed likely to be present at the identified
amino acid(s). In our example study, we suspected methionine
residues to be alkylated resulting in a labile side chain and a neutral
loss from the modified peptides. After defining the loss of the
methionine side chain as variable modification in Mascot, we iden-
tified several hundred peptides with this modification (data not
shown, for more information see ref. 11). Manual analysis of a
number of peptide hits revealed, that the neutral loss of the alky-
lated side chain is occurring either in the ion source of the mass
spectrometer (Fig. 7), or during MS/MS fragmentation (data not
shown, for more information see ref. 11).

Fig. 7 Carbamylated methionine results in a neutral loss of the amino acid side chain due to in-source
fragmentation. (a) Survey spectrum showing the neutral loss of the side chain from the peptide with alkylated
methionine; (b) extracted ion chromatogram (XICs) of the peptide with alkylated methionine, and its version
which lost the side chain due to in-source fragmentation, showing perfect co-elution (reproduced from [11])
246 Shiva Ahmadi and Dominic Winter

Fig. 8 (a) MALDI-MS spectra of the synthetic peptide APEIMLNSK, reduced with DTT and alkylated with IAA,
IAC, AA, and CAA, respectively, showing a loss of the methionine side chain (48 Da) in peptides alkylated
using iodine-containing reagents. (b) MALDI-MS/MS spectra of the unmodified peptide (upper panel) and the
signal at M-48 Da (lower panel). The fragment ion series confirms that the mass difference of 48 Da results
from a loss of the methionine side chain (reproduced from [11])

3.3.7 Validation of To provide final proof that the identified spectra indeed originate
Identified Modification from the proposed modification(s), experiments with synthetic
(s) Using Synthetic peptides should be performed. In the case of chemical artifacts,
Peptides this procedure is straightforward as any synthetic peptide contain-
ing the amino acid(s) susceptible to the modification can be used
(in our example study methionine). In case of biological modifica-
tions, ideally one or several of the peptides found to be modified
should be chemically synthesized and should be measured using the
same mass spectrometer in order to generate a reference spectrum,
to which the initial samples can be compared to. If the identified
modification is indeed correct, the fragment ion patterns are
expected to match perfectly. In our example study, we used a
peptide containing methionine, incubated it with the different
alkylation reagents and measured the sample by MALDI-MS/MS
(Fig. 8). This showed clearly that only iodine-containing reagents
are resulting in an unspecific alkylation of methionine followed by
the loss of its side chain (see Note 23).

3.4 Follow-Up Once a novel modification is unambiguously identified, it should be

Studies documented in online databases like Unimod to make it available to
the community. In the case of biological modifications, protocols
 Identification of Unexpected Protein Modifications by Mass Spectrometry. . . 247

for its enrichment can be established in order to determine the

extent of the modification in biological systems. Ultimately, muta-
tion experiments are needed to prove the biological significance.
For unintended chemical modifications, the protocols should be
modified in a way to prevent these modifications. Based on the type
of modification identified, it is often possible to recognize the
reagents responsible for it. If it is not possible to prevent the
modification, the strategy for database searching can be adapted
including it as a variable modification. This will at least allow
identification of a higher percentage of peptides in the samples. In
our example study, it was revealed that iodine-containing alkylation
reagents resulted in massive offside alkylation significantly reducing
numbers of identified peptides. We therefore changed our proto-
cols toward the use of the non-iodine-containing alkylation reagent
acrylamide. The identified loss of the alkylated methionine side
chain was documented in the Unimod database.

4 Notes

1. If protease inhibitor tablets are used, they can be dissolved in

water to generate a 100
stock solution. The stock can be
stored for several months at 20  C. This solution should be
added to the lysis buffer directly before performing the
experiment.
2. Dimethyl labeling reagents (all cyano- and formaldehyde solu-
tions) are highly toxic upon skin contact. Therefore, they
should be handled with extreme care.
3. It is necessary to put plates on ice and to use ice-cold PBS to
suppress any biological reactions.
4. To obtain a sufficiently pure cytosolic fraction, the cells have to
be lyzed in a gentle way keeping the organelles intact. This
allows to remove the organelles by centrifugation and prevents
excessive contamination with proteins from them. We used a
sucrose-containing buffer in combination with a dounce
homogenizer. Alternative gentle lysis methods can be used
based on the type of biological sample.
5. Protein precipitation is an optional step. If the lysis buffer is
compatible with proteolytic digestion it can be omitted.
6. If the protein pellet is not suspended in SDS, Chloroform
methanol precipitation usually leads to better results as the
precipitated proteins are more efficiently digested [28].
7. As we performed in-gel digestion, the pellet was reconstituted
in SDS buffer. In case of direct in-solution digestion, other
buffers such as Rapigest [29] can be used. Urea, which is also
248 Shiva Ahmadi and Dominic Winter

commonly used for in-solution digestion, should be avoided as

it may introduce unspecific carbamylation [30].
8. The volume of peptide extraction solutions is dependent on the
amount of the gel pieces. The volume should be sufficient to
cover the gel pieces.
9. In a top N experiment, if the majority of survey scans are
followed by the maximum number of MS/MS scans, the sam-
ple contains more peaks than the mass spectrometer can frag-
ment. This can be investigated using software like Raw Meat
(Vast Scientific): the part of the gradient in which the majority
of peptides elute should be investigated by plotting how many
MS/MS are triggered from each survey scan. If the number of
MS/MS events is constantly at the maximum value, the sample
complexity is too high for the mass spectrometer resulting in
under-sampling. In this case, low-abundant peptides are not
identified. On the other hand, if the MS/MS scans are below
the maximum number per survey scan, the sample is over-
sampled which is favorable for the identification of
low-abundant modified peptides.
10. In case no comparison to a control sample is considered in your
experimental setting, proceed to Subheading 3.3.5.
11. For a MS/MS ion search in Mascot in addition to *.mgf files
the following formats are supported: Finnigan (*.ASC),
Waters/Micromass (*.PKL), Sequest (*.DTA), PerSeptive (*.
PKS), Sciex API III, Bruker (*.XML), mzData (*.XML),
mzML (*.mzML).
12. While the right choice of precursor mass tolerance is of high
importance, in our experience the MS/MS ion tolerance is less
critical.
13. When datasets are searched against large databases, the search
will take longer and the ion score cutoff at 1% FDR will be
higher. Therefore, if only “normal” proteins from well-studied
organisms like human and mouse are identified, a smaller
database often results in better performance.
14. It is advisable to select the smallest number of variable mod-
ifications possible at this point. Higher numbers of modifica-
tions will result in higher score cutoffs due to an increased
number of decoy hits. This in turn results in lower numbers
of peptide identifications.
15. Proteome Discoverer includes a wide variety of analysis tools
which allow the visualization of data in different ways. These
tools may be used for some of these analyses as well.
16. In case of reduced enzymatic activity, it is advisable to perform
a step to remove the interfering chemicals before proteolytic
Identification of Unexpected Protein Modifications by Mass Spectrometry. . . 249

digestion. This could be achieved, e.g., through protein pre-

cipitation, SDS-PAGE, or molecular weight cutoff spin filters.
17. For resuspension of peptides, avoid using buffers containing
primary amines such as Tris or ammonium bicarbonate, since
the substances used for dimethyl labeling are amine reactive
and such buffers would prevent the efficient labeling of pep-
tides. In case of in-solution digestion setups, the tryptic diges-
tion can be carried out directly in a compatible buffer such as
HEPES and the desalting step omitted. Before dimethyl label-
ing, a protein assay can be performed to make sure that the
correct amount of peptide is used for the labeling reaction.
18. If samples with protein amounts exceeding 15–20 μg are gen-
erated, high capacity cartridges such as Oasis or Sep-Pak (both
from Waters) can be used. In case of increased sample com-
plexity, samples can be further fractionated (using, e.g., SAX
STAGE tips) and the resulting fractions desalted by individual
STAGE tips.
19. Mascot itself does not allow for quantification based on pre-
cursor ion intensities as this information is not contained in the
*.mgf files used for database searching. Therefore, quantifica-
tion has to be performed using Proteome Discoverer or other
algorithms (for instance Mascot Distiller). Also other protein
identification algorithms frequently include such quantification
options.
20. The >twofold change cutoff applied in our example study is a
frequently used value in proteomic studies. It is, however,
rather arbitrary. Another possibility, which is widely accepted,
is to compute the significance cutoff ratio based on a 95%
confidence interval in the reference group (control study).
21. The *.mgf files can be created using several tools like the Trans-
Proteomic Pipeline (TPP, [31]) or the Proteowizard MSCon-
vert GUI [32].
22. For MS/MS searches with several hundreds to a few thousands
of spectra, the “peptide summary” report can easily be accessed
on a HTML page. However, in case of large and complex
MS/MS searches, it is not practical to simply open the results
as the file may become too big. There are a number of switches
which can be modified for the individual report to allow open-
ing large MS/MS searches as a “peptide summary” report.
More information can be found in the Mascot help page
(http://www.matrixscience.com/help/results_help.html).
23. In this study we used a MALDI instrument for the synthetic
peptide experiments due to its ease of use. In principle, how-
ever, any mass spectrometer which is capable of performing
MS/MS experiments can be used.
250 Shiva Ahmadi and Dominic Winter

References
1. Aebersold R, Mann M (2016) Mass- 9:4632–4641. https://doi.org/10.1002/
spectrometric exploration of proteome struc- pmic.200900398
ture and function. Nature 537:347–355. 11. Müller T, Winter D (2017) Systematic evalua-
https://doi.org/10.1038/nature19949 tion of protein reduction and alkylation reveals
2. Kalli A, Smith GT, Sweredoski MJ et al (2013) massive unspecific side effects by iodine-
Evaluation and optimization of mass spectro- containing reagents. Mol Cell Proteomics
metric settings during data-dependent acquisi- 16:1173–1187. https://doi.org/10.1074/
tion mode: focus on LTQ-orbitrap mass mcp.M116.064048
analyzers. J Proteome Res 12:3071–3086. 12. Laemmli UK (1970) Cleavage of structural
https://doi.org/10.1021/pr3011588 proteins during the assembly of the head of
3. Eng JK, McCormack AL, Yates JR (1994) An bacteriophage T4. Nature 227:680–685.
approach to correlate tandem mass spectral https://doi.org/10.1038/227680a0
data of peptides with amino acid sequences in 13. Boersema PJ, Raijmakers R, Lemeer S et al
a protein database. J Am Soc Mass Spectrom (2009) Multiplex peptide stable isotope
5:976–989. https://doi.org/10.1016/1044- dimethyl labeling for quantitative proteomics.
0305(94)80016-2 Nat Protoc 4:484–494. https://doi.org/10.
4. Griss J, Perez-Riverol Y, Lewis S et al (2016) 1038/nprot.2009.21
Recognizing millions of consistently unidenti- 14. Shevchenko A, Wilm M, Vorm O et al (1996)
fied spectra across hundreds of shotgun prote- Mass spectrometric sequencing of proteins
omics datasets. Nat Methods 13:651–656. from silver-stained polyacrylamide gels. Anal
https://doi.org/10.1038/nmeth.3902 Chem 68:850–858. https://doi.org/10.
5. Nielsen ML, Savitski MM, Ra Z (2006) Extent 1021/ac950914h
of modifications in human proteome samples 15. Chin Y, Aiken GR, O’Loughlin E (1994)
and their effect on dynamic range of analysis in Molecular weight, polydispersity, and spectro-
shotgun proteomics. Mol Cell Proteomics scopic properties of aquatic humic substances.
5:2384–2391. https://doi.org/10.1074/ Environ Sci 28:1853–1858. https://doi.org/
mcp.M600248-MCP200 10.1021/es00060a015
6. Nesvizhskii AI, Roos FF, Grossmann J et al 16. Williams A, Frasca V (2001) Ion-exchange
(2006) Dynamic spectrum quality assessment chromatography. Curr Protoc Protein Sci
and iterative computational analysis of shotgun 15:8.2.1–8.2.30
proteomic data: toward more efficient identifi- 17. Chen J, Lee CS, Shen Y et al (2002) Integra-
cation of post-translational modifications, tion of capillary isoelectric focusing with capil-
sequence polymorphisms, and novel peptides. lary reversed-phase liquid chromatography for
Mol Cell Proteomics 5:652–670. https://doi. two-dimensional proteomics separation. Elec-
org/10.1074/mcp.M500319-MCP200 trophoresis 23:3143–3148. https://doi.org/
7. Chick JM, Kolippakkam D, Nusinow DP et al 10.1002/1522-2683(200209)23:18<3143::
(2015) A mass-tolerant database search identi- AID-ELPS3143>3.0.CO;2-7
fies a large proportion of unassigned spectra in 18. Nühse TS, Stensballe A, Jensen ON et al
shotgun proteomics as modified peptides. Nat (2003) Large-scale analysis of in vivo phos-
Biotechnol 33:743–749. https://doi.org/10. phorylated membrane proteins by immobilized
1038/nbt.3267 metal ion affinity chromatography and mass
8. Tanner S, Shu H, Frank A et al (2005) spectrometry. Mol Cell Proteomics
InsPecT: identification of posttranslationally 2:1234–1243. https://doi.org/10.1074/
modified peptides from tandem mass spectra. mcp.T300006-MCP200
Anal Chem 77:4626–4639. https://doi.org/ 19. Beausoleil SA, Jedrychowski M, Schwartz D
10.1021/ac050102d et al (2004) Large-scale characterization of
9. Jensen ON (2004) Modification-specific pro- HeLa cell nuclear phosphoproteins. Proc Natl
teomics: characterization of post-translational Acad Sci 101:12130–12135. https://doi.org/
modifications by mass spectrometry. Curr Opin 10.1073/pnas.0404720101
Chem Biol 8:33–41. https://doi.org/10. 20. Michel PE, Reymond F, Arnaud IL et al (2003)
1016/j.cbpa.2003.12.009 Protein fractionation in a multicompartment
10. Zhao Y, Jensen ON (2009) Modification- device using Off-GelTM isoelectric focusing.
specific proteomics: strategies for characteriza- Electrophoresis 24:3–11. https://doi.org/10.
tion of post-translational modifications using 1002/elps.200390030
enrichment techniques. Proteomics
 Identification of Unexpected Protein Modifications by Mass Spectrometry. . . 251

21. Huber LA, Pfaller K, Vietor I (2003) Organelle 200210)2:10<1426::AID-PROT1426>3.0.

proteomics: implications for subcellular frac- CO;2-5
tionation in proteomics. Circ Res 27. Seidler J, Zinn N, Boehm ME et al (2010) De
92:962–968. https://doi.org/10.1161/01. novo sequencing of peptides by
RES.0000071748.48338.25 MS/MS. Proteomics 10:634–649. https://
22. Rappsilber J, Ishihama Y, Mann M (2003) Stop doi.org/10.1002/pmic.200900459
and go extraction tips for matrix-assisted laser 28. Winter D, Steen H (2011) Optimization of cell
desorption/ionization, nanoelectrospray, and lysis and protein digestion protocols for the
LC/MS sample pretreatment in proteomics. analysis of HeLa S3 cells by LC-MS/MS. Pro-
Anal Chem 75:663–670. https://doi.org/10. teomics 11:4726–4730. https://doi.org/10.
1021/ac026117i 1002/pmic.201100162
23. Verheggen K, Raeder H, Berven FS et al 29. Yu YQ, Gilar M, Lee PJ et al (2003) Enzyme-
(2017) Anatomy and evolution of database friendly, mass spectrometry compatible surfac-
search engines-a central component of mass tant for in-solution enzymatic digestion of pro-
spectrometry based proteomic workflows. teins. Anal Chem 75:6023–6028. https://doi.
Mass Spectrom Rev. https://doi.org/10. org/10.1021/ac0346196
1002/mas.21543 30. Kollipara L, Zahedi RP (2013) Protein carba-
24. Brosch M, Yu L, Hubbard T et al (2009) Accu- mylation: in vivo modification or in vitro arte-
rate and sensitive peptide identification with fact? Proteomics 13:941–944. https://doi.
mascot percolator. J Proteome Res org/10.1002/pmic.201200452
8:3176–3181. https://doi.org/10.1021/ 31. Deutsch EW, Mendoza L, Shteynberg D et al
pr800982s (2015) Trans-Proteomic Pipeline, a standar-
25. Bantscheff M, Schirle M, Sweetman G et al dized data processing pipeline for large-scale
(2007) Quantitative mass spectrometry in pro- reproducible proteomics informatics. Proteo-
teomics: a critical review. Anal Bioanal Chem mics Clin Appl 9:745–754. https://doi.org/
389:1017–1031. https://doi.org/10.1007/ 10.1002/prca.201400164
s00216-007-1486-6 32. Holman JD, Tabb DL, Mallick P (2014)
26. Creasy DM, Cottrell JS (2002) Error tolerant Employing ProteoWizard to convert raw mass
searching of uninterpreted tandem mass spec- spectrometry data. Curr Protoc Bioinformatics
trometry data. Proteomics 2:1426–1434. 46:13.24.1–13.24.9. https://doi.org/10.
https://doi.org/10.1002/1615-9861( 1002/0471250953.bi1324s46
 Chapter 16

Label-Free LC-MS/MS Strategy for Comprehensive

Proteomic Profiling of Human Islets Collected Using Laser
Capture Microdissection from Frozen Pancreata
Lina Zhang, Giacomo Lanzoni, Matteo Battarra, Luca Inverardi,
and Qibin Zhang

Abstract
Diabetes mellitus is caused by either loss of pancreatic islets β-cells (Type 1 Diabetes, T1D), insufficient
insulin release in the islet β-cells coupled with insulin resistance in target tissues (Type 2 Diabetes, T2D), or
impaired insulin release (genetic forms of diabetes and, possibly, T1D subtypes). The investigation of the
islet proteome could elucidate facets of the pathogenesis of diabetes. Enzymatically isolated and cultured
(EIC) islets are frequently used to investigate biochemical signaling pathways that could trigger β-cell
changes and death in diabetes. However, they cannot fully reflect the natural protein composition and
disease process of in vivo islets due to the stress from isolation procedures and in vitro culture. The laser
capture microdissection method employs a high-energy laser source to separate the desired cells from the
remaining tissue section in an environment which is well conserved and close to the natural condition.
Here, we describe a label-free proteomic workflow of laser capture microdissected (LCM) human islets
from fresh-frozen pancreas sections of cadaveric donors to obtain an accurate and unbiased profile of the
pancreatic islet proteome. The workflow includes preparation of frozen tissue section, staining and dehy-
dration, LCM islets collection, islet protein digestion, label-free Liquid Chromatography-Tandem Mass
Spectrometry (LC-MS/MS), database search, and statistical analysis.

Key words LCM human pancreatic islets, Label-free proteomics, LC-MS/MS, MaxQuant, Perseus

1 Introduction

Diabetes mellitus is caused by either loss of pancreatic islets β-cells

(Type 1 Diabetes, T1D), insufficient insulin release in the islet
β-cells coupled with insulin resistance in target tissues (Type 2 Dia-
betes, T2D), or impaired insulin release (genetic forms of diabetes
and, possibly, T1D subtypes). The investigation of the islet prote-
ome could elucidate facets of the pathogenesis of diabetes. Enzy-
matically isolated and cultured (EIC) islets [1–3] have been
frequently used to investigate biochemical signaling pathways that
could trigger β-cell changes and death. However, such in vitro

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_16, © Springer Science+Business Media, LLC, part of Springer Nature 2019

253
254 Lina Zhang et al.

models have some limitations: they do not fully reflect what hap-
pens in vivo due to a lack of the natural environment where islets
exist and due to the changes in cell physiology induced by isolation
and culture. The procedure of enzymatic isolation of pancreatic
islets causes major structural changes and induces upregulation of
stress-related genes in islets [4]. Furthermore, EIC islets frequently
contain a significant percentage of contaminating acinar cells and
duct cells [5]. Alternatively, human pancreatic tissue can be col-
lected from cadaveric individuals and preserved frozen for further
laser-capture microdissected (LCM) isolation. LCM employs a
high-energy laser source to separate the desired cells from the
remaining tissue section [6], a strategy that can minimize the
contamination of surrounding tissue. LCM isolation also enables
the extraction of samples from an environment which is well con-
served and close to the natural condition, to better investigate cell
physiology [7], cell biology [8], cell transcriptome [4], and prote-
ome [9]. The exploration of the proteome signature of LCM islets
with an unbiased method may provide information on the changes
of protein composition occurring in dysfunctional islets, even with
limited sample amounts, which may facilitate understanding of the
pathogenesis of diabetes.
Here we describe a workflow for label-free proteomic analysis
of LCM islets obtained from sections of fresh-frozen human pan-
creas. This method enables accurate and unbiased profiling of the
pancreatic islet proteome. The strategy avoids enzymatic treatment
for cell dissociation and in vitro culture, and is designed to maintain
protein composition close to that of the original tissue. The
method can be easily adapted to other tissues, organs, and species.
The workflow covers preparation of frozen tissue sections, immu-
nohistochemical staining of reference sections, staining and dehy-
dration for LCM, LCM of pancreatic islets and acinar tissue,
preparation of samples for proteomic analysis, label-free Liquid
Chromatography-Tandem Mass Spectrometry (LC-MS/MS) data
acquisition, database search for protein identification, quantifica-
tion, and statistical analysis for the determination of the proteins
differentially expressed between LCM islets and LCM acinar tissue.

2 Materials

Common solvents and reagents [acetic acid, dithiothreitol (DTT),

iodoacetamide (IAA), Ammonium Bicarbonate (NH4HCO3),
Hydrochloric acid (HCl), Formic Acid (FA), Acetonitrile
(CH3CN)] were purchased from Sigma-Aldrich (St. Louis, MO).
Tissue-Tek O.C.T. compound and 100% ethanol were purchased
from VWR. Toluidine blue O and Anti-insulin antibody clone
K36aC10 were purchased from Sigma-Aldrich. Leica polyethylene
naphthalate (PEN) Membrane slides were purchased from Leica.
 Label Free Proteomics of LCM Collected Human Pancreatic Islets 255

PBS was purchased from Gibco Life Technologies. Kimwipes and

Drierite were purchased from VWR. Peroxo-Block™ was purchased
from Thermofisher Scientific. Histostain Plus Broadspectrum AEC
kit was purchased from Invitrogen. Elite Mini PAP Pen was pur-
chased from Diagnostic BioSystems (Pleasanton, CA). PPS Silent
Surfactant was purchased from Expedeon (San Diego, CA). The
BCA protein assay was obtained from ThermoFisher Scientific
(Rockford, IL), the sequencing-grade Trypsin was purchased
from Promega (Madison WI). All solvents used are HPLC-grade.
Instrumentation: cryotome (Leica Cryotome CM3050 S,
Leica), histology slide scanner (PathScan Enabler IV, with worksta-
tion and PathScan Enabler software, Meyer Instruments), laser
microdissection system (Leica Microscope LS LMD, with worksta-
tion and Leica LMD software, Leica), liquid chromatography and
mass spectrometry system (UltiMate 3000 RSLCnano system and a
Q Exactive HF mass spectrometer coupled with an EASY-Spray ion
source, ThermoFisher Scientific). Details of reagents and materials
used in each step are listed below.

2.1 For Frozen 1. Tissue-Tek Cryomold Standard.

Tissue Sections 2. Tissue-Tek O.C.T. compound.
3. 100% ethanol.
4. Leica PEN Membrane slides.

2.2 Immuno- 1. Elite Mini PAP Pen.

histochemical Staining 2. 10 mM phosphate-buffered saline (PBS) pH 7.4.
of Reference Sections
3. Anti-insulin antibody clone K36aC10 1:1000 dilution (see
Note 1).
4. Ready-to-use biotinylated secondary antibody (from Histos-
tain Plus Broadspectrum AEC kit).
5. HRP Substrate/Chromogen reagents: AEC Single Solution.

2.3 For LCM Staining 1. 100% ethanol.

and Dehydration 2. 70% ethanol: combine 70 mL ethanol and 30 mL H2O.
3. 90% ethanol: combine 90 mL ethanol and 10 mL H2O.
4. 0.5% w/v Toluidine blue O staining solution was prepared in
70% ethanol (see Note 2).
5. Drierite desiccant.

2.4 For LCM of 1. 50 mM NH4HCO3 pH 8: add 0.40 g of NH4HCO3 to

Pancreatic Islets and 100 mL of H2O.
Acinar Tissue
256 Lina Zhang et al.

2.5 For Protein 1. Prepare 1% PPS by adding 100 μL 50 mM NH4HCO3 to 1mg

Digestion PPS bottle (see Note 3).
2. 50 mM dithiothreitol (DTT): weigh 0.77 mg of DTT in a
microcentrifuge tube, and add 100 μL DI water (see Note 4).
3. 50 mM iodoacetamide (IAA): weigh 0.925 mg of IAA in a
microcentrifuge tube, and add 100 μL DI water (see Note 5).
4. Trypsin stock solution: prepare 1 μg/μL in 50 mM acetic acid,
and store at
20  C before use.
5. 2 M HCl: 16.52 mL 37% HCl and add H2O to 100 mL.

2.6 For LC-MS/MS Buffer A: 0.1% FA; Buffer B: 0.1% FA in CH3CN.

3 Methods

The major procedures involved in analysis of LCM human islets

proteome is shown in Fig. 1.
The human pancreas tissue from three donors was used in this
study. Three technical replicates of LCM islets were collected from
each donor and six islet equivalent human pancreatic islets were
collected from each replication. Meanwhile, the same equivalent
acinar tissue was collected from surrounding of the same islets. Two
technique replicates were collected for each donor. LCM acinar
tissue was used to confirm no contamination in LCM islets.

Fig. 1 Schematic representation of the experimental work flow

 Label Free Proteomics of LCM Collected Human Pancreatic Islets 257

3.1 Preparation of 1. Exercise care while processing pancreatic tissue: avoid squeez-
Frozen Tissue Sections ing or stretching the tissue, use a scalpel to obtain blunt cuts.
2. Resect pancreatic tissue blocks from the neck region of cadav-
eric pancreata from organ donors.
3. Collect tissue fragments of approximately 1 cm  0.5 cm and
position fragments in the center of cryomolds.
4. Embed tissue in Tissue-Tek O.C.T. compound and immedi-
ately freeze at
80  C by placing the mold holding the tissue
on top of dry ice (see Note 6).
5. Cut the blocks into sections of 10 μm thickness with a cryo-
tome, with temperature set at
20  C (see Note 7).
5. Transfer three pancreas sections onto each of 10 Leica PEN
Membrane slides.
6. Prepare extra reference sections on a regular glass slide for
insulin immunohistochemistry and mapping (see Note 8).

3.2 Immuno- 1. Fix the reference sections in 10% formalin for 15 min (see Note 9).
histochemical Staining 2. Wash four times in PBS.
of Reference Sections
3. Leave a drop of PBS on each tissue section and draw a circle
with a PAP pen to surround each section.
4. Remove the PBS.
5. Add Peroxo-Block™ for 45 s. Wash immediately.
6. Add 100 μL of anti-insulin antibody clone K36aC10 solution
(dilution 1:300) to each section to completely cover tissue.
7. Incubate in a humidified chamber at room temperature for
60 min.
8. Rinse with PBS for 5 min, three times.
9. Add 100 μL of secondary antibody to each section to
completely cover tissue and incubate for 10 min.
10. Rinse with PBS for 5 min, three times.
11. Add enough enzyme conjugate solution (from the Histostain
Plus Broadspectrum AEC kit) to each section to completely
cover tissue and incubate for 10 min.
12. Rinse with PBS for 2 min, three times.
13. Add chromogen AEC Single Solution and incubate 5–10 min.
14. Scan reference sections with a PathScan Enabler IV instrument
to obtain maps of the entire sections and identify stained
insulin-containing islets.

3.3 Staining and 1. The staining and dehydration protocol is performed with
Dehydration for LCM 8 clean Coplin jars, prefilled with 50 mL of 70% (jar # 1–5),
90% (jar #6), and 100% (jar #7 and 8) ethanol.
258 Lina Zhang et al.

2. Jars # 1–5 are maintained chilled on ice during the staining, jars
# 6–8 are maintained at room temperature to avoid condensa-
tion after dehydration.
3. The PEN membrane slides with tissue sections are dipped for
30 s in each jar from #1 to #3 of the ethanol series.
4. After jar #3, each slide is drained by gently placing the side edge
of the glass on a Kimwipe, then placed horizontally, with the
tissue sections on top.
5. The sections are stained for 90 s by adding 200 μL of Toluidine
blue O staining solution, then drained and transferred to jar #
4, to continue dehydration (see Note 10).
6. The slides are dipped for 30 s in each jar following the numeri-
cal order (jar # 4–8), to obtain dehydration.
7. The stained and dehydrated slide is drained with a Kimwipe and
placed under a laminar flow hood for 4 min to enable ethanol
evaporation.
8. The slide is placed in a slide box containing the desiccant
Drierite wrapped in Kimwipes and closed with tape (see
Note 11).

3.4 LCM of 1. The stage of the Leica Microscope LS LMD system is posi-
Pancreatic Islets and tioned in a clear acrylic box (microdissection chamber) where
Acinar Tissue (see the atmosphere can be controlled.
Note 12) 2. 1 h before microdissection, the workplace is cleaned and the
microdissection chamber is dehydrated using 2 kg of fresh
Drierite to minimize the humidity and enable membrane
microdissection.
3. The Leica LMD software is used to set up the laser (see
Table 1), to initialize the instrument and to control the move-
ments of the laser on the tissue section.
4. The PEN membrane slide with the stained tissue is positioned
on the stage, with the tissue facing down.
5. Empty sterile collection tubes are placed under the cutting area
(RNAse-free Eppendorf tubes, 500 μL volume, flat cap).
6. The scans of the reference sections are used to map the insulin-
containing islets.
7. Pancreatic islets are identified in the toluidine-stained tissue by
visualizing in bright field and in phase-contrast with 10 mag-
nification. In bright field, pancreatic islets appear as clusters of
cells with lightly colored cytoplasm, whereas the surrounding
acinar tissue is composed by cells with darker cytoplasm (see
Fig. 1). In phase-contrast, islet cells appear finely granulated.
Visualize islet borders at 20 magnification (see Note 13).
 Label Free Proteomics of LCM Collected Human Pancreatic Islets 259

Table 1
System configuration for laser capture microdissection with the Leica LMD
instrument

Parameters 10 magnification 20 magnification

Aperture 10 13
Intensity 40 35
Speed 4 7
Offset 26 34
Ap Diff 8 8
Option Med Med

8. Pancreatic islets and pancreatic exocrine tissue are collected in

separate tubes. The area of each microdissection is annotated.
9. The volume of microdissected tissue is calculated by multiply-
ing the total area collected by 10 μm (thickness of the section).
10. Microdissected islets are collected into the cap of 500 μL sterile
tubes (see Note 14, Fig. 1).
11. Acinar tissue is microdissected from neighboring areas and
collected into separate tubes.
12. Additional tissue is microdissected until the total volume for
each sample corresponds to 1.06  107 μm3, six Islet Equiva-
lents (see next paragraph).
13. Microdissection session should not last more than 60 min
(stain additional tissue sections every 60 min).
14. Carefully remove the collection tubes from the microdissection
chamber.
15. Resuspend the microdissected tissue with 50 μL of 50 mM
NH4HCO3.
16. Close the tube, centrifuge the resuspended tissue for 2 min at
13,000 rpm (see Note 15).
17. Freeze by placing on dry ice, maintain frozen at
80  C.

3.5 Conversion of The total volume of isolated islets can be expressed as number of
LCM Areas to Volumes islet equivalents (IEQ) [10]. An IEQ corresponds to the volume of
and to IEQ a “standard” islet, a sphere with a diameter d ¼ 150 μm and a
volume of VIEQ ¼ 1.77  106 μm3. 1 IEQ contains approximately
1560 islet cells [11]. The area of laser-captured tissue is recorded,
and the volume is calculated by multiplying the total area collected
for the thickness of the tissue section (10 μm). The target total
volume of each microdissected sample is 1.06  107 μm3,
260 Lina Zhang et al.

corresponding to 6 IEQ (6  1.77  106 μm3). At any time during

the collection, the total volume of laser-captured tissue can be
divided by the standard volume of 1 IEQ (VIEQ ¼ 1.77  106 μm3)
to obtain the corresponding number of laser-captured IEQ.

3.6 Protein Digestion 1. Add 6 μL of 1% pps silent surfactant (PPS) to extract and
solubilize hydrophobic proteins.
2. Add 1.5 μL of 50 mM DTT and incubate at 95  C for 6 min.
3. Sonicate sample for 3 min.
4. Alkylate with 7.5 μL 50 mM iodoacetamide for 25 min at 45  C
in the dark.
5. Add 1 μg stock trypsin at 37  C overnight.
6. Hydrolyze PPS by adding 12 μL 2 M HCl at room temperature
for 2 h.
7. Centrifuge samples at 16,000g for 12 min and separate the
supernatant for LC-MS/MS analysis (see Note 16).

3.7 LC-MS/MS Protocols for LC-MS/MS analysis can vary because of diversity of
Analysis LC systems (manufacturer, column, solvent composition, gradient,
flow rate, etc.) and MS instruments (manufacturer, electrospray
condition, fragmentation, MS parameters, analyzer, etc.). The fol-
lowing is the practice routinely used in our laboratory.
1. The LC-MS/MS platform consists of an UltiMate 3000
RSLCnano system and a Q Exactive HF mass spectrometer
coupled with an EASY-Spray ion source (ThermoFisher
Scientific).
2. Peptide separation is performed on a PepMap C18 analytical
column (2 μm particle, 50 cm  75 μm, ThermoFisher Scien-
tific). Injection volume is 2.5 μL (0.5 μg peptide amount
loaded into column) per sample (see Note 17).
3. A binary solvent system consisting of 0.1% FA in water (solvent
A) and 0.1% FA in CH3CN (solvent B) is used at a flow rate of
250 nL min
1 (see Note 18).
4. LC separation is performed using the following gradient
setting: hold at 4% B for 3 min (for desalting), from 4 to 8%
B in 0.1 min, 8 to 40% B in 90 min (effective gradient), 40 to
90% B in 0.1% min, hold at 90% B for 10 min (for washing
column), 90% to 4% B in 0.1 min, and hold at 4% B for 17 min
for re-equilibrating column (see Note 19).
5. MS data are acquired in profile mode and resolution for full
scan (400–2000 m/z) is set to 120,000 (at m/z 200) with
maximum ion injection time of 50 ms, and automatic gain
control (AGC) target of 1e6.
 Label Free Proteomics of LCM Collected Human Pancreatic Islets 261

6. MS/MS data are acquired with data-dependent method of top

15. An isolation window of 1.4 m/z is used to isolate precursor
ions for fragmentation by higher-energy collisional dissociation
(HCD) at normalized collision energy of 28. Resolution for
MS/MS spectrum is set to 15,000 (at m/z 200) with maxi-
mum ion injection time of 100 ms. AGC target for MS/MS
scans is 1e5.
7. Precursor ions with single, seven, and higher charge states are
excluded from fragmentation, and dynamic exclusion time is
set to 20 s.

3.8 Database Search Many database search software packages are available for this pur-
for Protein pose. MaxQuant is demonstrated here [12].
Identification and
1. The acquired datasets (.raw files) are analyzed using MaxQuant
Quantification and the built-in Andromeda search engine against a UniProt
human database (see Note 20).
2. Variable modifications include protein N-terminal acetylation
and methionine oxidation.
3. Fixed modifications contain cysteine carbamidomethylation.
4. A maximum of two missed cleavages are allowed for the search.
5. Trypsin/P is selected as the specific proteolytic enzyme (see
Note 21).
6. For label-free quantification, “match between runs” is selected
(see Note 22).
7. The false discovery rate (FDR) cutoff used for both peptides
and proteins is 0.01 (1%) using decoy database.
8. Only the razor/unique peptides are used for quantitative
calculations.
9. The other parameters are the default settings in MaxQuant
software for processing orbitrap-type data.

3.9 Statistical The search results in ProteinGroups.txt generated by MaxQuant

Analysis are directly processed by Perseus software [13]. The differentially
expressed proteins are identified by statistical analysis tools built in
Perseus.
1. Import the quantitative data from ProteinGroups.txt into
Perseus.
2. The potential contaminants, reverse hits and proteins only
identified by modification site are excluded.
3. Filter out the protein with unique peptides less than 1.
4. The protein intensities are log2-transformed.
5. Categorize the samples into two groups: LCM islets and LCM
acinar tissue.
262 Lina Zhang et al.

6. Filter out the proteins not quantified in all the samples

7. Two-samples tests coupled with Benjamini-Hochberg (FDR
cut off of 0.05) correction are performed to identify the differ-
entially expressed proteins [14].

3.10 Additional l http://string-db.org: a database of known and predicted protein

Resources for Data interactions. The interactions include direct (physical) and indi-
Analysis and rect (functional) associations; they are derived from four sources:
Biological Genomic context, High-throughput experiments, Coexpres-
Interpretation sion, Previous knowledge. This tool can be used to interpolate
proteins in functional and interaction networks. The participa-
tion of proteins in networks was established by references in
literature.
l http://www.proteinatlas.org: a database of Antibody-based Pro-
teomics. This tool enables the analysis of gene and protein
expression in various human tissues. The database presents
data related to the binding specificity of commercially available
antibodies.
l http://compartments.jensenlab.org: a subcellular localization
database. The database integrates evidence on protein subcellu-
lar localization from manually curated literature, high-
throughput analyses, automatic text mining, and sequence-
based prediction methods.

4 Notes

1. The dilution of the anti-insulin antibody should be prepared

freshly.
2. The 0.5% w/v Toluidine blue O staining solution in 70%
ethanol should be prepared freshly.
3. PPS solution should be prepared freshly. Once the package is
opened to air, the contents should be immediately reconsti-
tuted in aqueous buffer (pH 7–8), protected from elevated
temperatures, and used within 12 h.
4. Stock DTT solution should be freshly prepared before use.
5. Keep the IAA solution in the dark.
6. Wipe the inner chamber and the stage of the cryotome with
100% ethanol.
7. Change the blade, wipe the inner chamber and the stage of the
cryotome with 100% ethanol after each sample in order to
avoid contamination.
8. Sections are maintained frozen and stored at
80  C.
Label Free Proteomics of LCM Collected Human Pancreatic Islets 263

9. Immunohistochemical staining of reference slides with insulin

antibodies enables the identification and mapping of islets with
β-cells.
10. Toluidine blue O staining of frozen pancreas sections enables
good discrimination of islets and acinar tissue: islets appear
lightly colored compared to the surrounding acinar tissue;
moreover, endocrine cells have a characteristic granulated or
“rugged” aspect in phase-contrast illumination. If the humid-
ity in the microdissection chamber is too high (as indicated by a
pink drierite), the tissue section may rehydrate in one hour or
less: this determines visible tissue degradation and hampers
further laser-capture microdissection.
11. Insulin-containing islets were mapped via conventional immu-
nohistochemical staining of reference sections, and Toluidine
Blue O staining was used to guide the laser-capture microdis-
section in ethanol-dehydrated sections.
12. Each LCM session lasted a total of 60 min, to avoid tissue
rehydration and degradation.
13. The setup for Laser capture microdissection is optimized by the
operator and adjusted to the nature of the samples.
14. The collection tubes should be sterile and RNase/DNase/
Protease free.
15. Centrifugation of the tissue at this stage enabled us to avoid
loss of tissue.
16. A rough estimation of solvent volume used for reconstitution
of peptides in each fraction can be determined by the amount
of peptides loaded onto column for fractionation and the
number of final fractions. For instance, 100 μg divided by
24 fractions yields 4.2 μg per fraction, and the preparation of
samples at 0.2 μg/μL requires the addition of 21.0 μL of
solvent for reconstitution.
17. The injection volume depends on the sample loop of autosam-
pler, column loading capacity, and MS detector, therefore
injection volume need to be adjusted based on the actual setup.
18. A flow rate of 250 nL/min for C18 50 cm  75 μm i.d. column
results in around 550–600 bar column pressure when heating
column at 35  C.
19. The gradient used for peptide separation can be modified
depending on separation performance. However, all samples
must be run under the same condition to limit variations
between samples.
20. The database information needs to include the type, sequence
entry number, and releasing date of database.
264 Lina Zhang et al.

21. The selection of enzyme used for search is based on the enzyme
that is chosen for protein digestion in Subheading 3.6.
22. “Match between runs” should be selected because it can
improve the search results for less missing values.

Acknowledgements

This work was supported by the National Institutes of Health (R01

DK114345) and by the Diabetes Research Institute Foundation.

References

1. Schrimpe-Rutledge AC, Fontès G, Gritsenko AM, Marchetti P, Solimena M (2013)

MA, Norbeck AD, Anderson DJ, Waters M, Improved protocol for laser microdissection
Adkins JN, Smith RD, Poitout V, Metz TO of human pancreatic islets from surgical speci-
(2012) Discovery of novel glucose-regulated mens. J Vis Exp 71:50231
proteins in isolated human pancreatic islets 8. Marciniak A, Cohrs CM, Tsata V, Chouinard
using LC–MS/MS-based proteomics. J Prote- JA, Selck C, Stertmann J, Reichelt S, Rose T,
ome Res 11(7):3520–3532 Ehehalt F, Weitz J, Solimena M, Slak
2. Waanders LF, Chwalek K, Monetti M, Rupnik M, Speier S (2014) Using pancreas
Kumar C, Lammert E, Mann M (2009) Quan- tissue slices for in situ studies of islet of Lan-
titative proteomic analysis of single pancreatic gerhans and acinar cell biology. Nat Protoc 9
islets. Proc Natl Acad Sci U S A 106 (12):2809–2822
(45):18902–18907 9. Nishida Y, Aida K, Kihara M, Kobayashi T
3. Eizirik DL, Sammeth M, Bouckenooghe T, (2014) Antibody-validated proteins in
Bottu G, Sisino G, Igoillo-Esteve M, Ortis F, inflamed islets of fulminant type 1 diabetes pro-
Santin I, Colli ML, Barthson J, Bouwens L, filed by laser-capture microdissection followed
Hughes L, Gregory L, Lunter G, Marselli L, by mass spectrometry. PLoS One 9(10):
Marchetti P, McCarthy MI, Cnop M (2012) e107664
The human pancreatic islet transcriptome: 10. Ricordi C, Gray DW, Hering BJ, Kaufman DB,
expression of candidate genes for type 1 diabe- Warnock GL, Kneteman NM, Lake SP,
tes and the impact of pro-inflammatory cyto- London NJ, Socci C, Alejandro R et al (1990)
kines. PLoS Genet 8(3):e1002552 Islet isolation assessment in man and large ani-
4. Marselli L, Thorne J, Ahn YB, Omer A, Sgroi mals. Acta Diabetol Lat 27(3):185–195
DC, Libermann T, Otu HH, Sharma A, 11. Pisania A, Weir GC, O’Neil JJ, Omer A,
Bonner-Weir S, Weir GC (2008) Gene expres- Tchipashvili V, Lei J, Colton CK, Bonner-
sion of purified beta-cell tissue obtained from Weir S (2010) Quantitative analysis of cell
human pancreas with laser capture microdissec- composition and purity of human pancreatic
tion. J Clin Endocrinol Metab 93 islet preparations. Lab Invest 90
(3):1046–1053 (11):1661–1675
5. Marselli L, Thorne J, Dahiya S, Sgroi DC, 12. Tyanova S, Temu T, Cox J (2016) The Max-
Sharma A, Bonner-Weir S, Marchetti P, Weir Quant computational platform for mass
GC (2010) Gene expression profiles of Beta- spectrometry-based shotgun proteomics. Nat
cell enriched tissue obtained by laser capture Protoc 11(12):2301–2319
microdissection from subjects with type 2 dia- 13. Tyanova S, Temu T, Sinitcyn P, Carlson A,
betes. PLoS One 5(7):e11499 Hein MY, Geiger T, Mann M, Cox J (2016)
6. Bonner RF, Emmert-Buck M, Cole K, The Perseus computational platform for com-
Pohida T, Chuaqui R, Goldstein S, Liotta LA prehensive analysis of (prote)omics data. Nat
(1997) Laser capture microdissection: molecu- Methods 13(9):731–740
lar analysis of tissue. Science 278(5342):1481, 14. Tusher VG, Tibshirani R, Chu G (2001) Sig-
1483. nificance analysis of microarrays applied to the
7. Sturm D, Marselli L, Ehehalt F, Richter D, ionizing radiation response. Proc Natl Acad Sci
Distler M, Kersting S, Grutzmann R, U S A 98(9):5116–5121
Bokvist K, Froguel P, Liechti R, Jorns A,
Meda P, Baretton GB, Saeger HD, Schulte
 Chapter 17

Targeted Proteomics
Yun Chen and Liang Liu

Abstract
Targeted proteomics detects proteins of interest with high sensitivity, quantitative accuracy, and reproduc-
ibility. In a targeted proteomics assay, surrogate peptides are generated by proteolytic digestion of target
proteins and selected reaction monitoring (SRM) assays are developed to quantify these peptides using
liquid chromatography-tandem mass spectrometry (LC-MS/MS). In this report, we describe the details of
quantitative analysis of target protein in cells and tissue samples.

Key words Targeted proteomics, Liquid chromatography-tandem mass spectrometry, Protein quan-
tification, Cells and tissue samples

1 Introduction

With a growing demand for protein quantification across multiple

samples, liquid chromatography-tandem mass spectrometry
(LC-MS/MS)-based targeted proteomics has emerged as a power-
ful tool in systems biology, biomedical research, and clinical prote-
omics because of its high sensitivity, quantitative accuracy, and
reproducibility [1–4]. Targeted proteomics was selected as a
method to watch in both 2009 and 2010 and also as method of
the year in 2012 by Nature Methods [5–7]. In a targeted analysis,
target proteins are first digested into peptides using a proteolytic
enzyme, commonly trypsin. Then, surrogate peptides that can
uniquely represent the target proteins are selectively analyzed by
selected/multiple reaction monitoring (SRM/MRM), which is
typically performed on a triple quadruple mass spectrometer.
Therein, the ion mass of the precursor peptide of interest is set in
the first mass analyzer (Q1), while peptide product ions, which is
generated by collision-induced dissociation in Q2, is predefined in
the third mass analyzer (Q3). Precursor ion/product ion m/z pairs,
referred to as SRM/MRM transitions, are used to yield LC-MS/
MS chromatogram. The area under the curve of the chromatogram

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_17, © Springer Science+Business Media, LLC, part of Springer Nature 2019

265
266 Yun Chen and Liang Liu

Intensity
ESI Retention time
Source Q1 Q2 Q3 Detector

Fig. 1 A schematic of targeted proteomics strategy using a triple quadrupole mass spectrometer operating in
SRM mode

provides a quantitative measurement for each desired peptide and

target protein (Fig. 1).
To date, targeted proteomics has become increasingly popular
for different research area and applications. In our lab, we have
regularly applied this technique for protein quantification, such as
p-glycoprotein (P-gp) [8, 9], transferrin (TRF) [10], transferrin
receptor (TfR) [11], folate receptor (FR) [12], extracellular regulated
protein kinase (ERK) [13], murine double minute 2 (MDM2) [14],
and heat shock proteins 27 (HSP27) [15] in cells and tissue samples.
The typical pipeline consists of (1) sample pretreatment and protein
extraction, (2) protein digestion, (3) selection of appropriate surro-
gate peptides of target proteins, (4) chemical synthesis of internal
standard peptides by incorporation of heavy stable isotopes, (5) assay
development and validation, and (6) sample analysis (Fig. 2).

2 Materials

2.1 Cell Viability Test Hemocytometer, cover slip, and counter.

10 mM phosphate-buffered saline (PBS).
Trypan blue (0.4%).

2.2 Tissue Scissor and homogenizer.

Homogenization Homogenization buffer: 50 mM Tris–HCl, pH 7.4, 2 mM Ethy-
lenediaminetetraacetic acid (EDTA), 1 mM DL-dithiothreitol
(DTT), 150 mM NaCl, and 1% protease inhibitor (see Note 1).

2.3 Protein RIPA lysis buffer: 50 mM Tris–HCl, pH 7.4, 150 mM NaCl, 1%

Extraction Buffers Triton X-100, 1% sodium deoxycholate, 0.1% sodium dodecyl
sulfate (SDS) (see Note 2).
2.3.1 Cytosolic Proteins

2.3.2 Membrane Proteins Extraction buffer: 50 mM Tris–HCl, pH 7.4, 1 mM DTT, 2 mM

EDTA, 1% protease inhibitor, 1% Triton X-114.
Washing buffer: 50 mM Tris–HCl, pH 7.4, 1 mM DTT, 2 mM
EDTA, 0.06% Triton X-114.
 Targeted Proteomics 267

Fig. 2 A typical pipeline of targeted proteomics

2.4 Tryptic Digestion 50 mM DTT, 400 mM iodoacetamide (IAA), and 50 mM ammo-

Buffer nium bicarbonate (NH4HCO3).

2.5 LC-MS/MS An Agilent Series 1200 HPLC system and a 6410 Triple Quad
Instrument LC/MS mass spectrometer (see Note 3).

3 Methods

3.1 Sample 1. Carefully remove culture medium from cells.

Preparation 2. Wash cells twice with cold PBS.
3.1.1 Cell Pretreatment 3. Assess cell viability using trypan blue (0.4%) exclusion. Mix cell
suspension, PBS, and trypan blue in a 2:3:5 ratio, and incubate
for 5 min at 37  C.
4. Count viable cells using a hemocytometer.
268 Yun Chen and Liang Liu

3.1.2 Tissue 1. Thaw tissue samples to room temperature and rinse thoroughly
Homogenization with deionized water.
2. Remove fat tissue and cut the remaining tissue into small pieces
and transfer them to tubes.
3. Weigh approximately 50 mg of tissue and suspend it in tissue
homogenization buffer (see Note 4).
4. Homogenize the tissue suspension on ice using a Bio-Gen
PRO200 homogenizer.

3.2 Protein 1. Add cold RIPA buffer to sample. Keep the sample on ice for
Extraction 45 min and vortex every 15 min.
3.2.1 Cytosolic Proteins 2. Centrifuge the sample at 14,000 g for 15 min.
3. Transfer the supernatant to a new tube and determine protein
concentration using a BCA protein assay kit.

3.2.2 Membrane Proteins 1. Centrifuge the sample at 10,000 g for 10 min, and resuspend
the pellet in 500 μL of membrane protein extraction buffer.
2. Incubate the sample on ice for 30 min, and vortex every 10 min,
followed by incubation at 37  C for 10 min (see Note 5).
3. Centrifuge the mixture at 10,000 g for 3 min to separate
detergent and aqueous phases.
4. Add 500 μL of 1% extraction buffer and 500 μL of wash buffer
to the aqueous and detergent phases, respectively. Then repeat
the above incubation and centrifugation steps.
5. Combine the detergent phases and precipitate proteins using
cold acetone (pre-frozen at 20  C for 1 h before use).
6. Allow acetone to evaporate at room temperature.
7. Dissolve the protein pellet in 1% SDS solution.
8. Determine protein concentration of the obtained sample using
a BCA protein assay kit.

3.3 In-Solution 1. Mix 100 μL of the sample with 50 μL of 50 mM NH4HCO3.

Tryptic Digestion 2. Denature proteins at 95  C for 8 min (see Note 6).
3. Add 50 μL of 50 mM DTT to the sample, followed by incuba-
tion at 60  C for 30 min.
4. Add 30 μL of 400 mM IAA and incubate at room temperature
for 30 min in the dark.
5. Add 50 μL of sequencing grade trypsin solution (1:20 enzyme:
protein) and incubate at 37  C for 24 h.
6. Add 10 μL of 0.1% TFA to stop the reaction.
7. Dry the sample in a vacuum centrifuge.
8. Resuspend the sample in 100 μL of ACN:water (50:50, v/v)
containing 0.1% FA.
 Targeted Proteomics 269

3.4 Desalting 1. Add 100 μL of internal standard solution to the tryptic

mixture.
2. Precondition the microspin C18 column (The Nest Group,
Inc., MA, USA) with 100 μL of ACN and 100 μL of water in
advance.
3. Transfer 50 μL of the sample into the column and centrifuge at
1000 g for 1 min.
4. Wash the column with 50 μL of ACN:water (5:95, v/v) con-
taining 0.1% TFA and elute it with 50 μL of ACN:water
(80:20, v/v) containing 0.1% FA.
5. Repeat the above procedure 3–4 times, and finally combine the
collections.

3.5 Surrogate 1. The most critical step in the establishment of a targeted prote-
Peptide Selection omics assay is the selection of proteolytic peptides that (1) are
unique to a candidate protein, (2) can provide an adequate
response, (3) are completely digested, and (4) can generate
high-quality SRM [16, 17] (see Note 7).
2. The uniqueness of selected surrogate peptides is normally
checked using a BLAST search. For example, the peptides of
434STTVQLMQR442, 674GSQAQDR680, and 368IIDNKP-
SIDSYSK380 were found to be unique to P-gp (accession
no. P08183 (MDR1 HUMAN), gi: 2506118) [8].
3. A LC-MS/MS analysis with a list of SRM transition pairs based
on either in silico prediction or spectral evidence from public
repositories is usually performed to identify the peptide with
the greatest abundance [18] (see Note 8). Synthetic references
peptides are usually employed for confirmation (Fig. 3).
4. The digestion efficiency is evaluated using the substrate peptide
containing the same peptide sequence (to mimic a piece of the
target protein). The digestion efficiency was calculated by com-
paring the response ratios of the tryptic peptide after digestion
and the equimolar synthetic peptide standard in the digestion
(Fig. 4) [19].
5. Optimize SRM transitions (see Notes 9–11].

3.6 Internal Standard Synthetic stable isotope-labeled peptide is prepared according to

the selected surrogate peptide. C13 and D stable isotope-labeled
amino acids are usually employed. For instance, a stable isotope-
labeled valine with an added mass of 8 Da from deuterium was
coupled to the peptide sequence STTV*QLMQR at position 4 to
yield a molecular mass shift of 8 Da from the non-labeled peptide
STTVQLMQR [8].
270 Yun Chen and Liang Liu

a 100
y8 15eV

Relative Abundance (%)

895

y1
146 b2 y2 b8
226233 681
b4 y4 y5
b3 598 y7
455483
341
798

200 400 600 800 1000

m/z
b 400

300
Intensity, cps

200

100

0
1 2 3 4 5 6 7
Time, min

Fig. 3 The product ion spectrum and LC-MS/MS chromatogram of 368IIDNKPSIDSYSK380, a surrogate peptide
of P-gp. The characteristic sequence-specific b ions and y ions, and retention time are indicative of this
peptide (reproduced from ref. 9 with permission from Elsevier)

3.7 Immuno- 1. Add BioMagPlus IgG beads that are pre-incubated with anti-
Depleted Matrix target-protein antibody (see Note 12).
Preparation 2. Incubate the mixture at 25  C for 2 h with shaking.
3. Magnetically separate the beads and collect the supernatant.
4. Rinse the beads with 1% SDS solution and combine the eluate
with the above supernatant.
5. Examine this synthetic matrix using Western blotting and
LC-MS/MS-based targeted proteomics assay (Fig. 5).

3.8 Assay 1. Prepare 1 mg/mL stock solution by weighing the peptide

Development and (including internal standard peptide) and dissolving it deio-
Validation nized water. The solution was stored at 20  C in a brown
glass tube to protect it from light.
2. Prepare calibration standards and QC standards by serial dilu-
tion of the stock solution using immuno-depleted matrix. For
P-gp, the concentrations of the calibration standards are
 Targeted Proteomics 271

a b
6x103 6x103 Undigested peptide GKSTTVQLMQRLY
Undigested peptide GKSTTVQLMQRLY
5 5
Intensity, cps

Intensity, cps
4 4

3 3

2 2

1 1

0 0
1 2 3 4 5 6 7 1 2 3 4 5 6 7
Time,min Time,min
3 3
6x10 6x10
Digested peptide STTVQLMQR Digested peptide STTVQLMQR
5 5
Intensity, cps

Intensity, cps
4 4

3 3

2 2

1 1

0 0
1 2 3 4 5 6 7 1 2 3 4 5 6 7
Time,min Time,min

Fig. 4 LC-MS/MS chromatograms of STTVQLMQ and its substrate peptide GKSTTVQLMQRLY (a) before and (b)
after tryptic digestion (reproduced from ref. 8 with permission from Elsevier)

Fig. 5 The Western blotting image and LC-MS/MS chromatogram for P-gp depleted tissue extract (reproduced
from ref. 9 with permission from Elsevier)

10, 25, 50, 100, 250, 400, 700, and 1000 ng/mL. The QC
standards for the lower limit of quantification (LLOQ), low
QC, mid QC, and high QC were prepared at 10, 30, 200, and
800 ng/mL, respectively, and frozen prior to use.
3. Construct the calibration curve using a weighted linear regres-
sion model with a weighting factor of 1/x2. The relative peak
area ratio of the analyte and the stable isotope-labeled internal
standard was plotted as a function of concentration (Fig. 6).
4. The precision and accuracy of the assay were assessed by
observing the response of the QC samples with four different
272 Yun Chen and Liang Liu

3.5

3 r2 = 0.9974

2.5

Area Ratio
2

1.5

0.5

0
0 200 400 600 800 1000

Concentration (ng/mL)

Fig. 6 Representative calibration curves (10–1000 ng/mL) for the standards

using P-gp depleted matrices (reproduced from ref. 9 with permission from
Elsevier)

Table 1
Accuracy and precision for the QC samples using P-gp depleted matrix (reproduced from ref. 8 with
permission from Elsevier)

Nominal concentration 10 ng/mL 30 ng/mL 200 ng/mL 800 ng/mL

Mean 9.33 29.0 200 770
%Bias 6.6 3.3 0.0 3.8
Intra-day Precision (%CV) 4.4 4.6 2.8 2.7
Inter-day precision (%CV) 9.4 4.1 3.1 1.0
n 18 18 18 18
Number of runs 3 3 3 3

concentrations of P-gp in three validation runs. The intra- and

inter-day precisions were expressed as the percent coefficient of
variation (%CV). The accuracy was obtained by comparing
the average calculated concentrations to their nominal values
(%bias) (Table 1).

3.9 Sample Analysis 1. Apply the LC-MS/MS-based targeted proteomics assay to

analyze the samples.
2. Calculate the amount of target proteins in cell and tissue sam-
ples using the calibration curves built above (Figs. 7 and 8) (see
Note 13). The protein amounts can also be monitored in a
time manner (Fig. 9).
 Targeted Proteomics 273

5x104
MCF-7/WT
4

Intensity, cps
3

0
1 2 3 4 5 6 7
Time, min

5x104
MCF-7/ADR
4
Intensity, cps

0
1 2 3 4 5 6 7
Time, min

Fig. 7 LC-MS/MS chromatograms of P-gp in cells. The expression levels of P-gp

were accurately quantified to be 3.53 fg/cell in MCF-7/WT and 34.5 fg/cell in
MCF-7/ADR cells (reproduced from ref. 8 with permission from Elsevier)

4 Notes

1. DTT may not be stable and should not be stored for long
periods of time, so make up fresh before use [20].
2. If necessary, add protease and phosphatase inhibitors to RIPA
buffer immediately before use.
3. It should be noted that other instrument platforms (e.g., ion
trap, Q-TOF) are also capable of performing SRM like experi-
ments; however, quadrupole mass spectrometer is the preferred
technology for quantification and the most accessible instru-
ment in routine research and clinical lab [21].
4. Tissue collection must be approved by the Medical Ethics
Review.
274 Yun Chen and Liang Liu

2.0

1.5

ng/mg*
1.0

0.5

0.0
Normal Tumor

Fig. 8 P-gp amounts in 36 matched pairs of breast tissue samples (reproduced

from ref. 9 with permission from Elsevier)

5. Increase incubation time if necessary.

6. Currently, many denaturing conditions are available such as a
strong acid or base (e.g., acetic acid), a concentrated inorganic
salt (e.g., urea), an organic solvent (e.g., alcohol), or heat [22].
7. There are primarily two opposite approaches, prospective and
retrospective, either using in silico prediction by various algo-
rithms or based on spectral evidence using data from either
public repositories or in-house experiments (e.g., spectra
recorded during global discovery experiments). As a prospec-
tive SRM design, it would be possible to predict which peptides
and product ions are most appropriate for SRM in protein
quantification by several computational tools, for example,
ESP predictor, STEPP, Peptide sieve (PAGE-ESI), Peptide
detectability. However, it should be noted that the mechanisms
of proteolysis, ionization, and fragmentation are not yet suffi-
ciently well understood to produce accurate models from
which to make such predictions. The current models can only
assist to select high-responding peptides, particularly in the
absence of experimental data. The retrospective approach uses
experimentally obtained peptide spectra as evidence, and sev-
eral software tools have been developed. Publicly available
spectral repositories include PRIDE, GPMDB, PeptideAtlas,
NIST, and MacCoss. The software tools are Targeted Identifi-
cation for Quantitative Analysis by Multiple reaction monitor-
ing (TIQAM), MRMer, SRMCollider, MaRiMba, MRMaid,
Skyline and ATAQS, or commercial with the software platforms
provided by mass spectrometer vendors (e.g., SRM Workflow
software (based on SIEVE), Pinpoint and P3 predictor
(Thermo Scientific), mTRAQ-reagent-based MRMPilot
 Targeted Proteomics 275

3.0x104
a
2.5

2.0
Intensity, cps

1.5

1.0 P3
P2
0.5 P1

P0
0.0
2 4 6 8 10
Time, min
1.5x104
b

1.2
Intensity, cps

0.9

P3
0.6
P2

0.3
P1

P0
0.0
2 4 6 8 10
Time, min

Passages after the

0 1 2 3
treatment of DOX
HSP27 (pg/cell) 11.4 ± 0.3 9.73 ± 0.55 8.03 ± 0.25 6.21 ± 0.36
P-gp (fg/cell) 3.62 ± 0.20 4.41 ± 0.20 4.74 ± 0.14 5.83 ± 0.28

Fig. 9 The LC/MS-MS chromatograms of (a) HSP27 and (b) P-gp in freshly prepared (P0) and passage 1, 2,
3 (P1, P2, P3) MCF-7/WT cells after the treatment of DOX. The amounts are also listed in the table (reproduced
from ref. 15 with permission from Elsevier)

software and multiple reaction monitoring initiated detection

and sequencing (MIDAS) Workflow Designer (Applied Biosys-
tems), VerifyE and TargetLynx™ Application Manager
(Waters), MassHunter Optimizer (Agilent Technologies)).
8. There are several empirical criteria for selection of peptides with
high abundance, (1) length between 6 and 16 amino acids, (2) no
276 Yun Chen and Liang Liu

methionine or cysteine residues, (3) no post-translational

modifications (e.g., proteolysis, phosphorylation, or glycosyla-
tion) or single nucleotide polymorphism, (4) no transmembrane
region for membrane proteins, (5) no proline residue at the
C-terminal side of an arginine or lysine residue, and (6) no con-
tinuous sequence of arginine or lysine residues (RR, KK, RK,
KR).
9. To achieve a better peak separation and signal sensitivity, it is
important to choose appropriate column type and optimize
organic composition, flow rate and column temperature.
10. Usually, optimal collision condition for doubly and triply
charged peptides ranges between 20 and 40 eV, or even lower
(10 eV). Longer peptides and peptides with less charges may
need higher collision energy (CE) and y ions could require
higher CE than b ions [19].
11. Singly charged y ions are normally the predominant fragments
generated by collision-induced dissociation, as compared to
b ions that are low abundant or even absent in product ion
spectrum due to their lower stability and easier decomposition.
Product ion with m/z value close to the precursor should be
avoided because such transitions are usually noisy [19].
12. Antibodies from multiple species and the corresponding IgG
beads are eligible for protein depletion [23].
13. SRM transitions can be summed to quantify the peptide, or it is
possible to use one or several transitions for quantification and
the others for a confirmation of the peptide identity [19].

References
1. Yocum AK, Chinnaiyan AM (2009) Current 7. Doerr A (2013) Mass spectrometry-based tar-
affairs in quantitative targeted proteomics: geted proteomics. Nat Methods 10(1):23–23
multiple reaction monitoring–mass spectrome- 8. Yang T, Xu F, Xu J, Fang D, Yu Y, Chen Y
try. Brief Funct Genomics Proteomics 8 (2013) Comparison of liquid chromatogra-
(2):145–157 phy–tandem mass spectrometry-based targeted
2. Pan S, Aebersold R, Chen R, Rush J, Goodlett proteomics and conventional analytical meth-
DR, McIntosh MW, Zhang J, Brentnall TA ods for the determination of P-glycoprotein in
(2008) Mass spectrometry based targeted pro- human breast cancer cells. J Chromatogr B
tein quantification: methods and applications. J 936:18–24
Proteome Res 8(2):787–797 9. Yang T, Chen F, Xu F, Wang F, Xu Q, Chen Y
3. Parker CE, Pearson TW, Anderson NL, Borch- (2014) A liquid chromatography–tandem mass
ers CH (2010) Mass-spectrometry-based clini- spectrometry-based targeted proteomics assay
cal proteomics–a review and prospective. for monitoring P-glycoprotein levels in human
Analyst 135(8):1830–1838 breast tissue. Clin Chim Acta 436:283–289
4. Domon B, Aebersold R (2006) Mass spec- 10. Yu Y, Xu J, Liu Y, Chen Y (2012) Quantifica-
trometry and protein analysis. Science 312 tion of human serum transferrin using liquid
(5771):212–217 chromatography–tandem mass spectrometry
5. Marx V (2013) Targeted proteomics. Nat based targeted proteomics. J Chromatogr B
Methods 10(1):19–22 902:10–15
6. Doerr A (2010) Targeted proteomics. Nat 11. Yang T, Xu F, Zhao Y, Wang S, Yang M, Chen
Methods 7(1):34–34 Y (2014) A liquid chromatography-tandem
 Targeted Proteomics 277

mass spectrometry-based targeted proteomics MacCoss MJ (2009) Expediting the develop-

approach for the assessment of transferrin ment of targeted SRM assays: using data from
receptor levels in breast cancer. Proteom Clin shotgun proteomics to automate method
Appl 8(9–10):773–782 development. J Proteome Res 8
12. Yang T, Xu F, Fang D, Chen Y (2015) Tar- (6):2733–2739
geted proteomics enables simultaneous quanti- 18. Picotti P, Aebersold R (2012) Selected reaction
fication of folate receptor isoforms and monitoring-based proteomics: workflows,
potential isoform-based diagnosis in breast potential, pitfalls and future directions. Nat
cancer. Sci Rep 5:16733 Methods 9(6):555–566
13. Yang T, Xu F, Sheng Y, Zhang W, Chen Y 19. Gianazza E, Tremoli E, Banfi C (2014) The
(2016) A targeted proteomics approach to the selected reaction monitoring/multiple reac-
quantitative analysis of ERK/Bcl-2-mediated tion monitoring-based mass spectrometry
anti-apoptosis and multi-drug resistance in approach for the accurate quantitation of pro-
breast cancer. Anal Bioanal Chem 408 teins: clinical applications in the cardiovascular
(26):7491–7503 diseases. Expert Rev Proteomics 11
14. Zhang W, Zhong T, Chen Y (2017) LC-MS/ (6):771–788
MS-based targeted proteomics quantitatively 20. Schmidt C, Urlaub H (2012) Absolute quanti-
detects the interaction between p53 and fication of proteins using standard peptides and
MDM2 in breast cancer. J Proteome multiple reaction monitoring. Methods Mol
152:172–180 Biol 893:249–265
15. Xu F, Yang T, Fang D, Xu Q, Chen Y (2014) 21. Dillen L, Cools W, Vereyken L, Lorreyne W,
An investigation of heat shock protein 27 and Huybrechts T, de Vries R, Ghobarah H,
P-glycoprotein mediated multi-drug resistance Cuyckens F (2012) Comparison of triple quad-
in breast cancer using liquid chromatography- rupole and high-resolution TOF-MS for quan-
tandem mass spectrometry-based targeted pro- tification of peptides. Bioanalysis 4
teomics. J Proteome 108:188–197 (5):565–579
16. Anderson L, Hunter CL (2006) Quantitative 22. Mosby I (2006) Mosby’s medical dictionary.
mass spectrometric multiple reaction monitor- Mosby
ing assays for major plasma proteins. Mol Cell 23. Zolotarjova N, Martosella J, Nicol G, Bailey J,
Proteomics 5(4):573–588 Boyes BE, Barrett WC (2005) Differences
17. Prakash A, Tomazela DM, Frewen B, among techniques for high-abundant protein
MacLean B, Merrihew G, Peterman S, depletion. Proteomics 5(13):3304–3313
 Chapter 18

Metabolomic Investigation of Staphylococcus aureus

Antibiotic Susceptibility by Liquid Chromatography Coupled
to High-Resolution Mass Spectrometry
Sandrine Aros-Calt, Florence A. Castelli, Patricia Lamourette,
Gaspard Gervasi, Christophe Junot, Bruno H. Muller,
and François Fenaille

Abstract
Staphylococcus aureus is a major human pathogen that can readily acquire antibiotic resistance. For instance,
methicillin-resistant S. aureus represents a major cause of hospital- and community-acquired bacterial
infections. In this chapter, we first provide a detailed protocol for obtaining unbiased and reproducible
S. aureus metabolic profiles. The resulting intracellular metabolome is then analyzed in an untargeted
manner by using both hydrophilic interaction liquid chromatography and pentafluorophenyl-propyl col-
umns coupled to high-resolution mass spectrometry. Such analyses are done in conjunction with our
in-house spectral database to identify with high confidence as many meaningful S. aureus metabolites as
possible. Under these conditions, we can routinely monitor more than 200 annotated S. aureus metabo-
lites. We also indicate how this protocol can be used to investigate the metabolic differences between
methicillin-resistant and susceptible strains.

Key words Staphylococcus aureus, Methicillin resistance, Metabolomics, Liquid chromatography,

High-resolution mass spectrometry

1 Introduction

Staphylococcus aureus (S. aureus) is a notorious and opportunistic

pathogen causing a wide range of diseases and syndromes, includ-
ing bacteremia, pneumonia, cellulitis, osteomyelitis, and infections
affecting skin and soft tissues [1]. In addition, S. aureus is well
known for its ability to acquire resistance to various kind of anti-
biotics. Among antibiotic-resistant S. aureus strains, Methicillin-
Resistant S. aureus (MRSA) is one of the most serious threat-level
pathogens that had also developed multidrug resistance [2], with
increasing resistance to vancomycin [3]. MRSA is involved in most
of the global S. aureus bacteremia cases, and is often associated with

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_18, © Springer Science+Business Media, LLC, part of Springer Nature 2019

279
280 Sandrine Aros-Calt et al.

poor clinical outcomes (~30% mortality rate) [4, 5]. Although the
level of MRSA isolates has decreased over time in Europe, some
European countries still report 25% or more of invasive S. aureus
isolates as MRSA [5]. Acquisition of antibiotic resistance can par-
tially be explained by the misuse and mishandling of drugs to treat
infections. Unfortunately, first antibiotic resistances often appear
very soon after the introduction of an antibiotic. When combined
with drug companies’ reluctance to develop novel antibiotics, this
has led to serious health concerns [6]. Therefore, numerous efforts
have been made to better understand antibiotic-resistance mechan-
isms in order to potentially find new therapeutic strategies [7–10].
In that objective, metabolomics can greatly help digging deeper
into the pathogenicity and biochemical mechanisms behind the anti-
biotic resistance of MRSA strains. Indeed, metabolomics provides the
most direct assessment of cellular phenotype and is well suited to the
quantitative and dynamic monitoring of bacterial metabolism varia-
tions in response to particular environmental conditions. Recently
published papers clearly support this statement regarding the study
MRSA strains [11–15]. Just to cite a few, the pioneering work of
Liebeke et al. regarding S. aureus metabolomics demonstrated the
possibility to simultaneously analyze about 80 S. aureus metabolites
by using liquid chromatography coupled to mass spectrometry
(LC-MS). They reported that the S. aureus central carbon metabo-
lism, which notably includes energy transfer molecules like nucleo-
tides, sugar mono- and bi-phosphates, and cofactors, can be used to
monitor metabolic disturbances in response to genetic deletions of
serine/threonine kinase and phosphatase [11].
Keaton et al. investigated the changes in metabolic pathways
when treating MRSA strains with subinhibitory concentrations of
β-lactam antibiotics. They used a combination of LC-MS and gas
chromatography coupled to mass spectrometry (GC-MS) techni-
ques to highlight significant increases in tricarboxylic acid (TCA)
cycle intermediates under the studied conditions, which tended to
demonstrate that the energy production of MRSA strains was
redirected to supply the cell wall synthesis/metabolism and so,
contributed to their survival in the presence of β-lactam antibiotics
[16]. In a recent study, Dorries et al. investigated the impact of five
antibiotics with different cellular targets on S. aureus metabolism
by studying its global intra- and extracellular metabolic profiles
thanks to LC-MS, GC-MS, and NMR techniques [13]. Their ana-
lytical platforms largely covering primary metabolism allowed them
to highlight accumulation as well as depletion of metabolites from
various biosynthetic pathways, such as central carbon and amino
acid metabolism; peptidoglycan, purine, and pyrimidine synthesis
[13]. Also, Ammons et al. successfully implemented an
NMR-based strategy to distinguish MRSA from MSSA strains
based on the quantitative monitoring of 40 intracellular metabo-
lites (essentially amino acids and TCA-cycle intermediates) [12].
 S. aureus Metabolomics 281

Although successful in distinguishing MRSA from MSSA

strains or getting insights into S. aureus antibiotic resistance, most
of these studies investigated a limited number of metabolites or
described metabolites that were only putatively annotated accord-
ing to the Metabolomic Standard Initiative criteria [17]. Indeed, to
be considered as formally identified, metabolites need to have a
minimum of two orthogonal physicochemical parameters matching
those of an authentic standard analyzed under identical experimen-
tal conditions (i.e., retention time and mass spectrum, or accurate
mass and tandem mass spectra, for example). In the absence of an
available corresponding authentic chemical standard, metabolites
of interest can only be regarded as putatively annotated, for exam-
ple, based on their accurate measured mass and interpretation of
the MS/MS spectra when available [17]. Considerable efforts are
still required to annotate the metabolome of S. aureus, which is
predicted to be a very complex microorganism with ~500 to ~1400
expected metabolites, as deduced from genome scale models [18].
About 100–150 S. aureus metabolites have been reported as confi-
dently identified in the most exhaustive studies published, this
suggests that the availability a fully annotated and comprehensive
metabolite library is still many years away.
The prerequisites for obtaining unbiased and most comprehen-
sive intracellular metabolite profiles of S. aureus strains are appro-
priate sample preparation as well as an efficient detection method to
cope with the natural chemical diversity of the metabolome. In that
context, we have designed and carefully optimized a specific and
robust sample preparation protocol to obtain a reliable snapshot of
S. aureus bacterial metabolism under various experimental condi-
tions. We have also reported the development of two complemen-
tary liquid chromatography coupled to high-resolution mass
spectrometry (LC-HRMS) platforms to identify as many S. aureus
metabolites as possible, at the highest confidence level (Fig. 1)
[14]. Our metabolite identification workflow followed the criteria
reported by the Metabolomics Standard Initiative [17], and has led
to the successful characterization of 210 S. aureus metabolites with
up to 173 formal identifications. We thus demonstrated the ability
of the implemented protocol to reproducibly detect differences in
metabolic profiles between MRSA and MSSA strains [14].
In this chapter, the step-by-step protocols for the extraction,
preparation, and analysis of S. aureus intracellular metabolites are
described so as to enable the reader to replicate them.

2 Materials

2.1 Bacterial Strains Methicillin-resistant and susceptible Staphylococcus aureus strains

were obtained from the bioMérieux collection of microorganisms.
Presence of the mecA/mecC gene in the studied strains was
282 Sandrine Aros-Calt et al.

Fig. 1 Experimental workflow for S. aureus metabolome analysis by LC-HRMS

evaluated by multiplex PCR assay, while minimal inhibitory con-

centrations (MIC) were determined using oxacillin Etest® strips
(bioMérieux, Marcy l’Etoile, France).

2.2 Bacterial Culture 1. Mueller Hinton II broth (MHII, cation-adjusted) from Becton
Dickinson (product reference 212322, Franklin Lakes, NJ).
Dissolve 22 g of MHII medium in 1 L of ultrapure water,
autoclave the mixture at 120
C for 20 min, and store at 4
C
until use.
2. Colombia (COS) agar containing 5% sheep blood plates
(bioMérieux).
3. Cefoxitin (Sigma-Aldrich, Saint Quentin Fallavier, France).
4. Petri dishes (Sigma-Aldrich).
5. Erlenmeyer baffled cell culture flasks (Sigma-Aldrich).
6. Sterile inoculating loops and needles (VWR, Fontenay-sous-
Bois, France).
7. Serological pipettes (5 mL, Sigma-Aldrich).
8. Minitron II rotary shaker (Infors HT, Bottmingen-Basel,
Switzerland).
9. Eppendorf BioPhotometer (Eppendorf, Montesson, France)
or equivalent spectrophotometer.

2.3 Metabolite 1. Polyethersulfone (PES) sterile membrane disc filters (47 mm

Extraction diameter, 0.45 μm pore size, PALL, Washington, NY).
2.3.1 Bacteria Collection 2. Filtration system (Millipore, Darmstadt, Germany), clamps
and Metabolism Quenching (Dutscher, Brumath, France), and vacuum pump (KNF,
Village-Neuf, France).
3. Washing buffer: 0.6% NaCl. Dissolve 3 g of NaCl in 500 mL of
deionized water, filter the resulting solution at 0.22 μm, and
store at 4
C until use.
 S. aureus Metabolomics 283

4. Ice-cold 60% ethanol (Sigma-Aldrich) in deionized water,

stored at 20
C.
5. Liquid nitrogen.

2.3.2 Extraction of 1. Tissue homogenizer Precellys 24 (Bertin Technologies, Mon-

Intracellular Metabolites tigny-le-Bretonneux, France).
2. 0.1 mm glass beads and tubes (Bertin Technologies).
3. Turbovap evaporator (Caliper Life Science Inc., Roissy,
France).

2.4 LC–MS Analysis 1. Sequant ZIC-pHILIC column, 2.1150 mm, 5 μm, HPLC
of Intracellular PEEK (Merck, Darmstadt, Germany).
Metabolites 2. Discovery HSF5 Pentafluorophenylpropyl (PFPP) column,
2.4.1 Chromatographic 2.1  150 mm, 5 μm (Sigma-Aldrich).
Columns

2.4.2 Mobile Phases 1. Mobile phase A (ZIC-pHILIC): 10 mM ammonium carbonate

pH 10.5. Dissolve 960 mg of ammonium carbonate (product
reference 68392, Sigma-Aldrich) in 1 L of ultrapure water.
Adjust pH to 10.5 using a 28% NH4OH solution.
2. Mobile phase B (ZIC-pHILIC): Pure acetonitrile (ACN).
3. Mobile phase A (PFPP): Water containing 0.1% formic acid.
4. Mobile phase B (PFPP): ACN containing 0.1% formic acid.

2.4.3 LC-MS(/MS) 1. LC-MS experiments were performed using an Ultimate 3000

Systems chromatographic system (Thermo Fisher Scientific, Courta-
boeuf, France) coupled to an Exactive Orbitrap mass spectrom-
eter (Thermo Fisher Scientific) fitted with an electrospray (ESI)
source.
2. LC-MS/MS experiments were realized using an Ultimate
3000 chromatographic system coupled to a Q-Orbitrap mass
spectrometer (Q-Exactive Plus, Thermo Fisher Scientific) fitted
with an electrospray source.
3. The mass spectrometer is calibrated externally once per week in
both ESI polarities using the manufacturer’s predefined meth-
ods and the recommended calibration mixture provided by the
manufacturer. Under such routine conditions, absolute mass
accuracies are on average below 3 ppm for both negative and
positive ionization modes.

2.4.4 Internal Solutions of internal standards are prepared in pure water (Table 1).
Standards (IS)
284 Sandrine Aros-Calt et al.

Table 1
List of internal standards used for LC-MS analysis

Concentration of Compound
individual stock concentration in the
solutions (mg/mL 20 standard mixture
Compound name water) (μg/mL)
13
C1-alanine 4 200
Ethylmalonic acid 4 30
15
N1-aspartic acid 4 200
13
C1-glucose 4 200
Ampicillin 2 50
Prednisone 2 10
Dihydrostreptomycin 4 200
Roxithromycin 2 200
15
N5-AMP 1 100
15
N5-ADP 1 50
15
N5-ATP 1 50
All compounds except 15N-labeled AMP and ATP (Euriso-Top, Saint-Aubin, France)
were from Sigma-Aldrich

3 Methods

3.1 Bacterial 1. Preheat liquid and solid cultivation media a few hours at 37
C
Pre-culture and to exclude any bacterial contamination.
Culture 2. S. aureus strains are first isolated from an overnight culture at
37
C on COS plates.
3. Bacterial pre-culture is performed under aerobic conditions by
inoculating few bacterial colonies in 12.5 mL of preheated
MHII medium within a 125 mL Erlenmeyer flask (10% of
total volume to ensure sufficient aeration). Incubate the result-
ing culture medium at 37
C for 12–18 h with vigorous shak-
ing (200 rpm).
4. Withdraw an aliquot and dilute to an optical density at 600 nm
(OD600) of ~0.1 in a fresh MHII medium (in a 2 L Erlenmeyer
flask). Allow bacteria growth at 37
C with vigorous shaking
(200 rpm). For all the strains studied, the early-exponential
phase corresponded to an OD600 of 1, which was equivalent to
5  108 CFU/mL (see Notes 1 and 2).
 S. aureus Metabolomics 285

3.2 Bacteria 1. Prepare the filtration device, run the vacuum pump, and con-
Sampling and dition the filter by passing through 5 mL of MHII medium.
Metabolism Quenching 2. Withdraw a 5 mL aliquot from the main culture broth (at the
targeted OD600) and fill it into the filter system to rapidly
separate bacteria from culture medium.
3. Wash the bacteria retained on the filter with 5 mL of 0.6% NaCl
to remove culture medium (see Note 3).
4. The filter is then rapidly transferred to a 50 mL Falcon tube
containing 5 mL of ice-cold 60% ethanol (see Note 4). The
tube is subsequently quickly immersed in liquid nitrogen to
quench bacterial metabolism (see Note 5) before mechanical
cell disruption. At this step, samples can be conserved several
months at 80
C.

3.3 Mechanical Cell 1. Following quenching, tubes containing bacteria on filter in the
Disruption and extraction solution are vortexed 10 times (10 s at 4
C) to
Metabolite Extraction remove cells from the filter.
2. Transfer 1 mL of the bacterial suspension into the Precellys
tubes, while the remaining 4 mL are kept at 80
C.
3. Bacterial lysis is accomplished by performing three cycles in a
Precellys 24 homogenizer for 30 s at 3800 rpm and at ~4
C.
4. Centrifuge the tubes during 5 min at 4
C and 10,000 g to
remove glass beads and cell debris.
5. Withdraw 400 μL of the supernatant and transfer into a 1.5 mL
Eppendorf tube.
6. Evaporate under a stream of nitrogen.
7. 200 μL of each sample can also be withdrawn and pooled to
obtain a Quality Control (QC) sample. 400-μL aliquots of the
resulting mixture can then be evaporated under a stream of
nitrogen.
8. Store samples and QCs at 80
C until LC-MS analyses.

3.4 LC-MS Analysis A complete detailed protocol for running LC-MS experiments will
of Intracellular not be provided hereafter. Below, we only list the main highlights of
Metabolites our approach involving a combination of two complementary
ZIC-pHILIC and PFPP columns coupled to an Orbitrap instru-
ment to obtain an optimal coverage of the S. aureus metabolome.
For additional details, the reader can refer to Boudah et al. and
Aros-Calt et al. [14, 19]. In principle, most of routine methods
used in metabolomics laboratories are expected to be used for
S. aureus metabolite analysis with limited or no modifications,
even if they make use of another type of high-resolution mass
spectrometer (e.g., Q-TOF).
286 Sandrine Aros-Calt et al.

3.4.1 LC-MS Analysis 1. Dilute eight times the 20 internal standard mixture (Table 1)
Using a ZIC-pHILIC Column in 10 mM ammonium carbonate pH 10.5.
2. Estimate the number of CFU in each dried sample tube and
adjust the resuspension volume to obtain a concentration of
1.25  107 CFU per 10 μL. The protocol provided hereafter is
provided for a final volume of 80 μL.
3. Solubilize the dried bacterial extract and QC samples in 32 μL
of the diluted internal standard mixture (step 1). After vigor-
ous mixing, incubate the resulting mixture in an ultrasonic bath
for 5 min.
4. Centrifuge at 10,000  g for 5 min at 4
C.
5. Transfer the resulting supernatant in an injection vial and add
48 μL of acetonitrile.
6. Inject 10 μL into the LC-MS system.
7. Metabolites are eluted from the column (maintained at 15
C)
at a flow rate of 200 μL/min using the gradient reported in
Table 2.
8. The Exactive Orbitrap mass spectrometer is operated in the
negative ion mode at a resolution of 50,000 at m/z 200 (full
width at half-maximum), using the following source para-
meters: Capillary voltage, 3 kV; capillary temperature,
280
C; sheath gas pressure, 60 arbitrary units; auxiliary gas
pressure, 10 arbitrary units. The detection is performed from
m/z 75 to 1000, using an injection time set at 100 ms and an
AGC target value of 3  106.

3.4.2 LC–MS Analysis 1. Dilute 20 times the 20 internal standard mixture (Table 1) in
Using a PFPP Column water containing 0.1% formic acid.
2. Estimate the number of CFU in each dried sample tube and
adjust the resuspension volume to obtain a concentration of

Table 2
Gradient conditions used for metabolite LC-MS analysis using the
ZIC-pHILIC column

Time (min) % Mobile phase B (ACN)

0 80
2 80
12 40
12.01 0
17 0
17.01 80
42 80
 S. aureus Metabolomics 287

Table 3
Gradient conditions used for metabolite LC-MS analysis using the PFPP
column

Time (min) % Mobile phase B (ACN containing 0.1% formic acid)

0 5
2 5
20 100
24 100
24.01 5
30 5

1.25  107 CFU per 10 μL. The protocol provided hereafter is

provided for a final volume of 80 μL.
3. Solubilize the dried bacterial extract and QC samples in 80 μL
of the diluted internal standard mixture (step 1). After vigor-
ous mixing, incubate the resulting mixture in an ultrasonic bath
for 5 min.
4. Centrifuge at 10,000 g for 5 min at 4
C.
5. Transfer the resulting supernatant into the injection vial.
6. Inject 10 μL into the LC-MS system.
7. Metabolites are eluted from the column (maintained at 30
C)
at a flow rate of 250 μL/min using the gradient reported in
Table 3.
8. The Exactive Orbitrap mass spectrometer is operated in the
positive ion mode using the parameters described above, except
the source voltage that is set at 5 kV.

3.5 Analysis of 1. AMP, ADP, and ATP were quantified by the isotope dilution
LC–MS Data method using their 15N-labeled homologues, with a procedure
similar to that described by Martano et al. [20]. Under our
3.5.1 Determination of
conditions, a mass accuracy better than 3 ppm for the endoge-
the Adenylate Energy
nous nucleotides and a perfect coelution with their labeled
Charge (AEC)
homologues ensure compound identification and accurate
quantification (see Note 6).
2. Calculate the AEC by using molar concentrations with the
following formula (see Note 7):

½ATP þ 0:5  ½ADP

AEC ¼
½ATP þ ½ADP þ ½AMP
288 Sandrine Aros-Calt et al.

3.5.2 Well-Designed Obtaining reliable metabolic profiling data is a complex task and
Sample Dataset and Quality particular attention should be paid to the experimental design in
Control Samples (QCs) for order to avoid any instrumental or analytical bias. Well-designed,
Reliable Untargeted robust, and reproducible metabolomics workflows commonly
Metabolomics involve column conditioning, sample randomization, and use of
QC samples. QCs are made from a pooled sample obtained by
mixing an equal volume (10–100 μL) of each sample to be studied,
and therefore constitute a representative bulk control sample so as
the signal variations of any metabolite could be reflected in the QCs
[21]. QC samples proved really useful to correct for drifts in MS
response, mass measurement accuracy, as well as chromatographic
retention time between analytical runs or batches [22, 23]. Another
particularly relevant feature of QCs implies dilution of QCs to
evaluate the corresponding MS response linearity. Under these
conditions, the metabolites showing a linear trend can be consid-
ered as analytically relevant. Figure 2 depicts a typical sample run
order used for LC-HRMS-based metabolomics.

Injection Order Sample

1 Blank
2 Blank
3 QC
4 QC
5 QC
6 QC
7 QC
8 Blank
9 8x dil. QC
10 4x dil. QC
11 2x dil. QC
12 QC
13 Blank
14 QC
15 Sample 1
16 Sample 2
17 Sample 3
… …
24 Sample 10
25 Blank
26 QC
27 Sample 11
… …
36 Sample 20
37 Blank
38 QC
… …

Fig. 2 Typical sample run order for LC-HRMS. If needed, additional inter-batches
QCs can be added
 S. aureus Metabolomics 289

3.5.3 Data Processing Data processing as well as statistical analysis can be performed
automatically and reproducibly online using the Workflow4meta-
bolomics (W4M) platform for computational metabolomics
[24, 25]. W4M is built on the Galaxy environment providing
intuitive and powerful features that allow the analyst to set up and
run complex workflows (Fig. 3). Four main processing steps can be
considered.
1. Preprocessing: Automatic peak detection, alignment, and
extraction using the XCMS software.
2. Normalization thanks to QCs: Correction of signal variation,
selection of analytically pertinent metabolites.
3. Statistical analyses (univariate and multivariate statistical tools).
Other statistical software such as SIMCA-P (Umetrics, Umea,
Sweden) can be used to perform multivariate data analyses such
as principal component analysis (PCA) and partial least squares
discriminant analysis (PLS-DA) or Prism (GraphPad, La Jolla,
USA) to perform univariate data analyzes (e.g., t-test).
4. Compound annotation is commonly performed considering a
10 ppm mass tolerance using either our in-house database
(including ~1000 metabolites) [14, 19, 26] or publically avail-
able databases such as KEGG [27], HMDB [28], or
METLIN [29].
5. Confirmation of metabolite annotation and identification of
other statistically relevant unknown metabolites is accomplished
by MS/MS experiments. Resulting MS/MS spectra are com-
pared to those included in our in-house database or in public
databases such as METLIN. Figure 4 summarizes our workflow
for formal metabolite identification (see Notes 8 and 9).

Fig. 3 Data treatment tools available in the W4M infrastructure. The left and right panels give the tools and
parameters that can be used to build the workflow in the central panel
290 Sandrine Aros-Calt et al.

Fig. 4 Analytical workflow for HRMS-based metabolite identification

4 Notes

1. Growth curves need to be carefully determined for each strain

studied. They are used to determine the OD600 at which the
bacteria should be harvested to correspond to early-, mid-, and
late-exponential or stationary phases.
2. Bacterial metabolism is a very fast changing and versatile pro-
cess (especially during the exponential phase). Therefore, the
intracellular metabolome of each studied strain needs to be
withdrawn exactly at the same growth stage to ensure accurate
comparison. For improved data consistency, it is also recom-
mended to prepare at least three independent biological repli-
cates of each condition/strain investigated.
3. A single washing step proved to be optimal with no significant
metabolite leakages/cell lysis by comparison with three wash-
ing rounds [14].
4. Although 60% ethanol solution represents the preferred solvent
for efficiently and reproducibly extracting intracellular metabo-
lites from S. aureus, an acetonitrile/methanol/water
(40:40:20, v/v/v) mixture might also represent a viable alter-
native if some particular metabolites are targeted [14].
5. The filtration step has to be performed as fast as possible, not to
bias metabolome quality [14, 30].
 S. aureus Metabolomics 291

6. By spiking 15N-labeled nucleotides in bacterial extracts, the

limits of detection (LODs) were estimated at 55, 45, and
230 nM for AMP, ADP, and ATP, respectively.
7. Exponentially growing bacterial cells should have a stable AEC
above 0.8, while stressed cells would have lower values
[31]. Using our optimized fast-filtration protocol, we can rou-
tinely and reproducibly obtain an AEC of 0.76  0.02. This
value is slightly lower than the theoretically expected value of
0.8 and might seem insufficient at first sight. However, the
high reproducibility of the measured AEC values indicated
that these values were probably not related with an incorrect
handling of the cells during sampling but rather to a particular
physiological characteristic of these cells under our growth
conditions. Several authors have already reported AEC values
significantly and negatively impacted (down to 0.1) by para-
meters such as the growth phase and the culture medium itself
for cells with maintained metabolic capability [32, 33]. There-
fore, the AEC value might not be considered as a strict indica-
tor of metabolic integrity, but rather of particular physiological
characteristics observed under specific growth conditions.
8. Under optimal conditions and according to the Metabolomic
Standards Initiative criteria [17], we were able to characterize
up to 210 metabolites in S. aureus, of which 173 and 9 were
identified and putatively annotated, respectively, while the
remaining 28 were only characterized by their accurate
masses [14].
9. As a representative example, Fig. 5 represents the PLS-DA score
plot of features from ZIC-pHILIC/MS fingerprints of MRSA
and MSSA strains harvested at the mid-exponential growth
phase. Each bacterial strain was then treated with cefoxitin

Fig. 5 PLS-DA score plot of features from ZIC-pHILIC/MS fingerprints of MRSA and MSSA harvested at the
mid-exponential growth phase and then exposed to cefoxitin. The PLS-DA model was validated by a
permutation test (100 times)
292 Sandrine Aros-Calt et al.

(β-lactam) at its minimal inhibitory concentration (MIC  1)

during both pre-culture and culture steps. The discriminant
metabolites were selected by combining multivariate variable
importance in the projection (VIP) obtained from the PLS-DA
model and univariate p-values (nonparametric MannWhitney
statistical test). The metabolites were considered as discrimi-
nant when VIP > 1.5 and p-value <0.05. Under these condi-
tions, up to 26 metabolites discriminate MSSA and MRSA
strains. Thus, intracellular levels of some nucleotides, amino
acids, or metabolites belonging to glycolysis of Krebs cycle
were also modified. Other metabolites such as N-acetyl-mur-
amoate-6-phosphate, UDP-N-MurNAc-Ala-Glu-Lys-Ala-Ala,
N-acetyl-glucosamine/galactosamine, and glycerol-3-phos-
phate involved in cell-wall biosynthesis were also discriminant.
These data tended to reinforce what has been observed without
any antibiotic exposure [14].

Acknowledgements

This work was supported by bioMérieux S.A. and the Association

Nationale de la Recherche et de la Technologie (ANRT). S.A.-C. is
the recipient of a CIFRE fellowship (grant number 2011/1474).
This work was also supported by the Commissariat à l’Energie
Atomique et aux Energies Alternatives and the MetaboHUB infra-
structure (ANR-11-INBS-0010 grant).

References

1. Lowy FD (1998) Staphylococcus aureus infec- 6. Boucher HW, Corey GR (2008) Epidemiology
tions. N Engl J Med 339:520–532 of methicillin-resistant Staphylococcus aureus.
2. Chambers HF, Deleo FR (2009) Waves of Clin Infect Dis Off Publ Infect Dis Soc Am
resistance: Staphylococcus aureus in the antibi- 46(Suppl 5):S344–S349
otic era. Nat Rev Microbiol 7:629–641 7. Sommer MOA, Dantas G (2011) Antibiotics
3. Gardete S, Tomasz A (2014) Mechanisms of and the resistant microbiome. Curr Opin
vancomycin resistance in Staphylococcus aureus. Microbiol 14:556–563
J Clin Invest 124:2836–2840 8. Munck C, Gumpert HK, Wallin AIN et al
4. Chastre J, Blasi F, Masterton RG et al (2014) (2014) Prediction of resistance development
European perspective and update on the man- against drug combinations by collateral
agement of nosocomial pneumonia due to responses to component drugs. Sci Transl
methicillin-resistant Staphylococcus aureus after Med 6:262ra156
more than 10 years of experience with linezo- 9. Belenky P, Ye JD, Porter CBM et al (2015)
lid. Clin Microbiol Infect Off Publ Eur Soc Bactericidal antibiotics induce toxic metabolic
Clin Microbiol Infect Dis 20(Suppl 4):19–36 perturbations that lead to cellular damage. Cell
5. Hassoun A, Linden PK, Friedman B (2017) Rep 13:968–980
Incidence, prevalence, and management of 10. Ling LL, Schneider T, Peoples AJ et al (2015)
MRSA bacteremia across patient populations- A new antibiotic kills pathogens without
a review of recent developments in MRSA detectable resistance. Nature 517:455–459
management and treatment. Crit Care Lond 11. Liebeke M, Meyer H, Donat S et al (2010) A
Engl 21:211 metabolomic view of Staphylococcus aureus and
its ser/thr kinase and phosphatase deletion
 S. aureus Metabolomics 293

mutants: involvement in cell wall biosynthesis. of serum and plasma using gas chromatography
Chem Biol 17:820–830 and liquid chromatography coupled to mass
12. Ammons MCB, Tripet BP, Carlson RP et al spectrometry. Nat Protoc 6:1060–1083
(2014) Quantitative NMR metabolite profiling 23. Dunn WB, Wilson ID, Nicholls AW et al
of methicillin-resistant and methicillin- (2012) The importance of experimental design
susceptible Staphylococcus aureus discriminates and QC samples in large-scale and MS-driven
between biofilm and planktonic phenotypes. J untargeted metabolomic studies of humans.
Proteome Res 13:2973–2985 Bioanalysis 4:2249–2264
13. Dörries K, Schlueter R, Lalk M (2014) Impact 24. Giacomoni F, Le Corguillé G, Monsoor M et al
of antibiotics with various target sites on the (2015) Workflow4Metabolomics: a collabora-
metabolome of Staphylococcus aureus. Antimi- tive research infrastructure for computational
crob Agents Chemother 58:7151–7163 metabolomics. Bioinformatics 31:1493–1495
14. Aros-Calt S, Muller BH, Boudah S et al (2015) 25. Guitton Y, Tremblay-Franco M, Le Corguillé
Annotation of the Staphylococcus aureus Meta- G et al (2017) Create, run, share, publish, and
bolome using liquid chromatography coupled reference your LC-MS, FIA-MS, GC-MS, and
to high-resolution mass spectrometry and NMR data analysis workflows with the Work-
application to the study of methicillin resis- flow4Metabolomics 3.0 Galaxy online infra-
tance. J Proteome Res 14:4863–4875 structure for metabolomics. Int J Biochem
15. Schelli K, Zhong F, Zhu J (2017) Comparative Cell Biol 93:89–101
metabolomics revealing Staphylococcus aureus 26. Roux A, Xu Y, Heilier J-F et al (2012) Annota-
metabolic response to different antibiotics. tion of the human adult urinary metabolome
Microb Biotechnol 10:1764–1774 and metabolite identification using ultra high
16. Keaton MA, Rosato RR, Plata KB et al (2013) performance liquid chromatography coupled
Exposure of clinical MRSA heterogeneous to a linear quadrupole ion trap-Orbitrap mass
strains to β-lactams redirects metabolism to spectrometer. Anal Chem 84:6429–6437
optimize energy production through the TCA 27. Kanehisa M, Goto S (2000) KEGG: Kyoto
cycle. PLoS One 8:e71025 encyclopedia of genes and genomes. Nucleic
17. Sumner LW, Amberg A, Barrett D et al (2007) Acids Res 28:27–30
Proposed minimum reporting standards for 28. Wishart DS, Tzur D, Knox C et al (2007)
chemical analysis chemical analysis working HMDB: the human Metabolome database.
group (CAWG) metabolomics standards initia- Nucleic Acids Res 35:D521–D526
tive (MSI). Metabolomics 3:211–221 29. Smith CA, O’Maille G, Want EJ et al (2005)
18. Liebeke M, Lalk M (2014) Staphylococcus METLIN: a metabolite mass spectral database.
aureus metabolic response to changing envi- Ther Drug Monit 27:747–751
ronmental conditions—a metabolomics per- 30. Meyer H, Liebeke M, Lalk M (2010) A proto-
spective. Int J Med Microbiol 304:222–229 col for the investigation of the intracellular
19. Boudah S, Olivier M-F, Aros-Calt S et al Staphylococcus aureus metabolome. Anal Bio-
(2014) Annotation of the human serum meta- chem 401:250–259
bolome by coupling three liquid chromatogra- 31. Chapman AG, Fall L, DE A (1971) Adenylate
phy methods to high-resolution mass energy charge in Escherichia coli during growth
spectrometry. J Chromatogr B Analyt Technol and starvation. J Bacteriol 108:1072–1086
Biomed Life Sci 966:34–47 32. van der Werf MJ, Overkamp KM, Muilwijk B
20. Martano G, Delmotte N, Kiefer P et al (2015) et al (2008) Comprehensive analysis of the
Fast sampling method for mammalian cell met- metabolome of Pseudomonas putida S12
abolic analyses using liquid chromatography- grown on different carbon sources. Mol Bio-
mass spectrometry. Nat Protoc 10:1–11 Syst 4:315–327
21. Naz S, Vallejo M, Garcı́a A et al (2014) Method 33. Stuani L, Lechaplais C, Salminen AV et al
validation strategies involved in non-targeted (2014) Novel metabolic features in Acinetobac-
metabolomics. J Chromatogr A 1353:99–105 ter baylyi ADP1 revealed by a multiomics
22. Dunn WB, Broadhurst D, Begley P et al (2011) approach. Metabolomics 10:1223–1238
Procedures for large-scale metabolic profiling
 Chapter 19

Nuts and Bolts of Protein Quantification by Online Trypsin

Digestion Coupled LC-MS/MS Analysis
Christopher A. Toth, Zsuzsanna Kuklenyik, and John R. Barr

Abstract
Protein digestion coupled to liquid chromatography and tandem mass spectrometry (LC-MS/MS) detec-
tion enables multiplexed quantification of proteins in complex biological matrices. However, the reproduc-
ibility of enzymatic digestion of proteins to produce proteotypic target peptides is a major limiting factor of
assay precision. Online digestion using immobilized trypsin addresses this problem through precise control
of digestion conditions and time. Because online digestion is typically for a short time, the potential for
peptide degradation, a major source of measurement bias, is significantly reduced. Online proteolysis
requires minimal sample preparation and is easily coupled to LC-MS/MS systems, further reducing
potential method variability. We describe herein a method optimized for the multiplexed quantification
of several apolipoproteins in human serum using on-column digestion. We highlight key features of the
method that enhance assay accuracy and precision. These include the use of value-assigned serum as
calibrators and stable isotope-labeled (SIL) peptide analogs as internal standards. We also comment on
practical aspects of column switching valve design, instrument maintenance, tandem mass spectrometry
data acquisition, and data processing.

Key words Online digestion, Immobilized enzyme reactor, IMER, Apolipoproteins, Quantitative,
IDMS, IMER-LC-MS/MS

1 Introduction

Mass spectrometry (MS) approaches have become effective means

for selective quantitative measurement of proteins in complex
biological matrices [1]. Targeted multiplexed assays for quantifica-
tion of up to 30–50 proteins involve proteolytic digestion, most
often with trypsin, and analysis of resulting unique peptides by
liquid chromatography with tandem mass spectrometry (LC-MS/
MS) detection. The use of stable isotope-labeled (SIL) peptide or
protein analogs as internal standards further enhances precision by
counteracting (normalizing) for matrix effects. This approach is
commonly referred to as isotope dilution MS (IDMS). The IDMS
approach is based on the assumption that there is a proportionate
relationship between the measured MS/MS response ratio and

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_19, © Springer Science+Business Media, LLC, part of Springer Nature 2019

295
296 Christopher A. Toth et al.

analyte molar ratio both before and after digestion. Furthermore,

this relationship needs to hold true both in the calibrators and in the
unknown samples regardless of heterogeneous proteoforms [2].
Protein digestion is commonly performed in-solution, on a
batch of samples, off-line from IDMS analysis. A typical batch
digestion protocol involves denaturation, reduction and alkylation,
enzymatic digestion, and some form of sample cleanup, i.e., solid
phase extraction. A main challenge of batch digestion is in the
selection of optimal digestion conditions for multiple proteins,
each with a potentially wide range of concentrations in vivo. Long
digestion times and high temperatures typically result in more
complete proteolytic cleavage of the target peptides. In practice
however, incubation times
3 h and temperatures >37  C give
rise to several non-idealities, particularly the aggregation, adsorp-
tion, and chemical degradation of cleavage products and their SIL
analogs. Peptide decay during in-solution digestion is a main source
of calibration bias and method variability [3].
An emerging alternative to in-solution digestion is online or
on-column digestion [4], where trypsin is covalently bound to
porous particles of packing material in an LC column. This immo-
bilized enzyme reactor (IMER) can be directly connected to the
LC-MS/MS system. Continuous plug-flow IMERs digest proteins
on-column as they pass through the enzyme laden stationary phase.
Samples are easily combined with SIL internal standards immedi-
ately prior to digestion.
Theoretically, the recovery of the peptide cleavage products
collected at the outlet of the IMER is determined by the flow
rate, temperature, void volume, injection volume, and the concen-
tration of the proteins in the injected sample [5]. If all conditions,
barring the protein concentration, are held constant and in an
optimal range, the measured [peptide cleavage product peak
area]/[SIL peptide peak area] ratio has a linear relationship with
the [native protein]/[SIL peptide] molar ratios in the injected
sample. With constant moles of SIL peptide, the peak area ratio
also has a linear correlation with the native protein concentration in
the injected sample.
IMER-LC-MS/MS is emerging as a valuable tool to facilitate
precise relative quantitation of proteins. The main advantage of
online digestion is the accelerated cleavage of the target peptides
due to high enzyme–protein ratio inside the IMER. This allows for
short reaction times (2–8 min), which reduce the opportunity for
chemical degradation of the proteins, the peptide cleavage products
and the SIL peptides, enhancing reproducibility. Immobilization
greatly reduces the propensity of trypsin for autolysis and denatur-
ation, allowing digestion at temperatures > 37  C (further enhanc-
ing cleavage rates), and the repeated use of the IMER for the
consecutive digestion of numerous samples. In addition, minimal
sample preparation is required, lending to simple dilute-and-shoot
 Nuts and Bolts of Protein Quantification by Online Trypsin Digestion. . . 297

methods. Sample cleanup is also performed online, as target pep-

tides elute from the IMER they are trapped on a short reverse phase
LC column where the peptides can be desalted and washed prior to
LC-MS/MS analysis. All valve switching intervals, reagent addi-
tions, digestion times, and flow rates are controlled in a precise and
automated fashion. Overall, relative to off-line batch digestion, the
IMER digestion is more simple, reproducible, and economical in
terms of labor and reagent cost.
The main limitation of online IMER digestion is that in many
cases complete proteolysis cannot be achieved. Therefore, to obtain
quantitative accuracy requires the use of matrix and concentration
range representative protein calibrators. Additionally, the IMER-
LC-MS/MS technique typically works best if the target proteins are
present at >100 nmol/L concentration in serum [5].
While the use of online trypsin digestion with LC-MS/MS for
qualitative analysis of proteins is relatively well demonstrated in the
literature, its application for quantitative protein analysis is rarely
discussed [6, 7]. In this chapter, we use a panel of apolipoproteins
in serum as an example to address the practical aspects of online
trypsin digestion for multiplexed protein quantification.

2 Materials

2.1 Reagents and 1. LC-MS grade acetonitrile and 2-propanol.

Consumables 2. Ultrapure deionized water.
3. Diluting buffer: 10 mM NaHCO3, 150 mM NaCl, pH 7.4 (see
Note 1).
4. Detergent solution: 0.45% w/v Zwittergent 3–12 in sample
buffer (see Note 2).
5. 0.5 mL 96-well round bottom microplates (Agilent).
6. Slit Seal 96-well microplate covers (BioChromato).
7. 1.5 mL standard polypropylene autosampler vials and caps
(Thermo Fisher).
8. SIL (13C, 15N) analogs of the proteotypic peptides of interest
(see Note 3).

2.2 HPLC Columns 1. IMER: Trypsin column, 33  2.1 mm id (Perfinity Bios-

ciences) (see Note 4).
2. Trapping/desalting column: Halo® ES-C18 peptide,
4.6  5 mm id, 2.7 μm particle diameter (see Note 5).
3. Analytical column: Halo® C18 core shell, 2.1 mm  100 mm,
2.7 μm particle size or similar.
298 Christopher A. Toth et al.

2.3 HPLC Solvents 1. Digestion buffers (quaternary pump).

(a) Digestion buffer: 50 mM Tris–HCl, 2 mM CaCl2, pH
8.4, in deionized water.
(b) Wash buffer: 50 mM Tris–HCl, 2 mM CaCl2, 25%
2-propanol, pH 8.4, in deionized water.
2. Gradient Eluents (binary pump).
(a) Eluent A: 0.1% formic acid in water.
(b) Eluent B: 0.1% formic acid in acetonitrile.

2.4 LC-MS/MS 1. Shimadzu 20/30 series HPLC system (Perfinity Biosciences)

Instrumentation (see Note 6).
2. Mass spectrometer: QTrap® 6500+ (SCIEX) or any suitable
triple quadrupole mass analyzer.
3. Ionization source: Electrospray ionization.
4. LC Control Software: Labsolutions (Shimadzu).
5. MS Control Software: Analyst (SCIEX).

2.5 Data Processing 1. Skyline (MacCoss Lab) (see Note 7).

Software 2. MultiQuant (SCIEX) (see Note 8).

2.6 Human Serum Frozen human serum from individual donors was purchased from a
specimen depository (Bioreclamation IVT, Westbury, NY).

3 Methods

3.1 Preparation of 1. Dissolve 2–5 mg of lyophilized powder in 100 mL of 5%

SIL Peptide Stock acetonitrile/0.1% formic acid in water (see Note 9).
Solutions 2. Distribute each peptide into 200–500 μL aliquots and store at
80  C (see Note 10).

3.2 Preparation of 1. Thaw one aliquot of each SIL stock solution on the bench,
Working SIL vortex-mix, and spin down.
Peptide Mix 2. To a 15 mL Falcon™ conical centrifuge tube, add 20–450 μL of
each individual SIL peptide stock solution (see Notes 11 and 22).
3. Add 100 μL of a 0.1% Zwittergent 3–12/0.1% formic acid in
water solution, cap, and vortex-mix (see Note 12).
4. Bring to 10 mL volume with 0.1% formic acid in water.
5. Store the SIL peptide mix for maximum 4 weeks at 4  C (see
Note 13).
6. Before analysis dispense a 1.5 mL aliquot into a standard
polypropylene autosampler vial and place into the reagent
tray of the autosampler.
 Nuts and Bolts of Protein Quantification by Online Trypsin Digestion. . . 299

3.3 Preparation of 1. Allow frozen serum from individual donors to thaw on ice.
Calibrator Pool and 2. Merge equal volumes from each donor (see Note 14).
Storage
3. Vacuum filter the pool using a Nalgene™ Rapid-Flow™ filter
(0.45 μm pore size) or equivalent.
4. Continuously stir the pool at 250 rpm in an ice bath on a
magnetic stir plate while distributing a series of 100, 50, 20,
10, and 5 μL aliquots using a positive displacement repeater
pipette into 2 mL storage vials.
5. Verify the volume of each aliquot by mass using an analytical
balance and record.
6. Store the aliquots at 80  C until use.
7. For the value assignment of the calibrator pool by standard
addition methodology, see Note 15.

3.4 Preparation of 1. Pull a set of stored Calibrator Pool Aliquots (100, 50, 20,
Standard Series 10 and 5 μL) from the freezer and allow to thaw on ice.
2. Bring to volume with diluting buffer (Subheading 2.1, step 3)
following an appropriate dilution level (see Note 16).
3. Gently invert each tube 10 times to mix and spin down.
4. Use immediately or store at 4  C for no more than 4 days (see
Note 17).

3.5 Sample 1. Pipette 10 μL of serum sample to a microcentrifuge tube.

Preparation for IMER- 2. Add 990 μL of sample diluent (Subheading 2.1, step 3).
LC-MS/MS Analysis
3. Invert 10 times to mix and briefly spin down.
4. Transfer 100 μL of diluted sample to a standard PP autosam-
pler vial or microplate.
5. Add 50 μL 0.45% w/v Zwittergent 3–12 solution (Subheading
2.1, step 4) to achieve a final concentration of 0.15% w/v.
6. Place on shaker plate for 5 min at 500 rpm.
7. Store at 4  C prior to IMER-LC-MS/MS analysis or place
immediately into chilled autosampler compartment.

3.6 IMER-LC-MS/MS 1. Set the temperature of the column compartment to 50  C and

Operation autosampler compartment to 4  C.
2. Program the autosampler via the Labsolutions software to draw
5 μL from the reagent vial containing the SIL mix, followed by
of 50 μL sample (see Note 18).
3. Digest the sample for 7 min using a flow rate of 20 μL/min (see
Note 19).
4. Peptide separation: The binary gradient flow rate should be set
to 0.5 mL/min. The starting composition should be set to
2% Solvent B. The low initial organic composition improves
300 Christopher A. Toth et al.

focusing and can be quickly ramped to achieve fast and efficient

peptide separation. The start of gradient elution must be syn-
chronized with the trap column switching valve. To enhance
sample throughput, the plumbing of the column switching
system can be modified for dual operation mode, with simulta-
neous digestion/trapping and gradient elution of consecutive
samples (see Notes 20 and 21).
5. The optimal operation of the mass spectrometer; 6500+
QTRAP® (Sciex, Foster City, CA) with a heated electrospray
ionization probe in positive ion mode requires the setting of
the following parameters: Ion spray voltage 5500 V, ion source
heater temperature 450  C, source gas 50 psi, curtain gas
35 psi. The native and the isotopically labeled IS peptide chro-
matograms are acquired by multiple reaction monitoring
(MRM) with unit mass resolution, in scheduled 60 s acquisi-
tion windows with a 0.65 s target scan time (see Note 22).
6. For the highest accuracy and precision, each sample unknown
should be analyzed in triplicate, and calibrators in duplicate. After
the highest calibrator, duplicate blank injections are recom-
mended to avoid carry-over from trypsin column (see Note 23).

3.7 Data Processing 1. Pull the raw mass spectrometry data into MultiQuant (SCIEX)
for the calculation of [native cleavage peptide peak area]/[SIL
peptide peak area] ratios.
2. Examine the automated peak integrations.
3. Before reporting concentrations review the data based on
acceptable range of SIL peak intensities, product ion ratios,
and deviations between technical repeats.

4 Notes

1. Phosphate containing buffer should not be used to dilute

samples as it quickly forms insoluble precipitates with CaCl2
in the digestion buffer inside the flow lines.
2. The addition of a mild detergent at near critical micelle con-
centration was found to be critical to enhance method sensitiv-
ity. As shown in Fig. 1 using bovine serum albumin as an
example, addition of detergent gave a marked increase
(20-fold in some cases) in peptide peak areas.
3. Recommendations for the selection, preparation, and storage
of labeled synthetic peptides have been previously described
[8]. In addition, a list of target peptides specific for apolipo-
protein quantification and corresponding SIL peptides are
described in a previous publication [7].
 Nuts and Bolts of Protein Quantification by Online Trypsin Digestion. . . 301

Fig. 1 The effect of Zwittergent 3-12 concentration on the digestion efficacy is demonstrated with bovine
serum albumin as a model protein. As the final detergent concentration in the sample well increases, the
peptide peak area/protein concentration increases for all tryptic peptides. This effect levels off just beyond the
critical micelle concentration (CMC) of Zwittergent 3-12 (4 mM, 0.15% w/v)

4. Trypsin columns available from Perfinity Biosciences could be

reliably used for 2000 injections of dilute human serum and
can be kept in a 50  C column compartment for 60 days
without significant loss of efficiency. A decrease in digestion
efficacy, less than 30%, can be corrected for by the measurement
of the calibrator pool with every batch.
5. A 4.6 mm id trapping column had twice the life span compared
to 2.1 mm id trap columns. Each 4.6 mm id trapping column
reliably functioned for 200 injections of 1:100 diluted human
serum before chromatographic artifacts developed and war-
ranted replacement.
6. The fully integrated digestion and LC platform (Perfinity Bios-
ciences/Shimadzu Scientific Instruments) consisted of an
autosampler (SIL-20 ACHT with pretreatment option), col-
umn oven (CT0-20AC), a low pressure ( 3000 psi) quater-
nary pump with solvent selection valve to alternate between
digestion and wash buffers (LC-20 AD), two high-pressure
HPLC pumps (LC-20ADXR) and control module
(CBM-20A). The system must be operated in “XL mode” in
order to program the required pretreatment operations with
the SIL-20 ACHT autosampler. The LC and MS systems were
controlled from separate computers with synchronization
302 Christopher A. Toth et al.

through a trigger cable between the LC control module and

the MS instrument.
7. Skyline software was used for the generation of the scheduled
MRM acquisition method including declustering potentials
and collision energies, and were used without further
optimization.
8. Multiquant software was used for MS/MS peak integration
and assessment of accuracy and precision of concentration
measurements.
9. Peptide stock solutions should be relatively concentrated
(0.5–2 nmol/μL) for accurate determination of purity by
amino acid analysis and to reduce the significance of peptide
adsorption to the container wall during storage. Some hydro-
phobic peptides may require between 5 and 30% acetonitrile
and 0.1–1% formic acid to fully dissolve [8].
10. For long term storage (>1 year at 20 to 80  C) lyophiliza-
tion of precise aliquot volumes is recommended [8]. The con-
centration of SIL peptide is not considered when native protein
concentration is calculated. However, initial and periodical
amino acid analysis is good practice to confirm SIL peptide
purity and storage stability.
11. As a general rule, the concentration of the SIL peptides should
be optimized to match the SIL peak areas with the median
native peptide peak areas from the digestion of 1:100 diluted
samples and the calibrator pool.
12. The addition of 0.001% w/v Zwittergent 3–12 reduces peptide
adsorption to container walls during storage.
13. Because the absolute concentration of the SIL peptides are not
factored into the calculation of the protein concentration
minor peptide degradation of SIL peptides in the working
mix over the course of 30 days storage at 4  C did not nega-
tively affect assay performance. For longer storage (>4 weeks),
the working SIL peptide mix should be aliquotted and stored
frozen (20 to 80  C) until use. Avoid multiple freeze/thaw
cycles if possible.
14. Recommendations for the preparation of calibrator pools have
been proposed [9].
15. The absolute concentration of several apolipoproteins in the
calibrator pool were determined using standard addition meth-
odology. Commercially available purified proteins were pur-
chased from Academy Biomedical, Novoprotein Scientific,
and Sigma Aldrich. Lyophilized powder was preferred where
possible and purities ranged from 95 to 98% as stated by the
manufacturer. The protein stock solutions were prepared in
 Nuts and Bolts of Protein Quantification by Online Trypsin Digestion. . . 303

sodium bicarbonate buffered saline (10 mM NaHCO3,

150 mM NaCl, pH 7.4) to approximate concentrations of
600 nmol/L apoA-II, 55 nmol/L apoA-IV, 150 nmol/L
apoC-I, 80 nmol/L apoC-II, 450 nmol/L apoC-III, and
90 nmol/L apoE3. The concentrations were chosen to approx-
imate 2–5 times the expected concentration in 1:100 diluted
matrix. Concentrations of the individual protein stock solu-
tions were measured by amino acid analysis by Midwest
Bio-Tech (Fishers, IN, USA). For each protein of interest, a
value-assigned protein stock solution was diluted in triplicate
to a 7-point series with 1, 1.25, 1.67, 2.5, 5, 10, and 20-fold
dilution factors, in addition to a buffer blank. A 50 μL aliquot
at each dilution level was added to the 50 μL aliquots of 1:100
diluted serum and 50 μL of 0.45% Zwittergent 3–12. The plate
was covered with a slit-seal cover, mixed on a shaker plate for
5 min at 500 rpm, then stored in an autosampler at 4  C prior
to analysis. Samples were analyzed by IMER-LC-MS/MS.
Light/heavy area ratio vs. protein dilution ratio linear regres-
sion curves were calculated for each peptide transition, of
which one peptide transition for each protein was selected for
quantification (Fig. 2). The concentration of each protein of

1
5 ApoA-2 EQLTPLIK +2y5 ApoA-4 LEPYADQLR +2y6 ApoC-1 TPDVSSALDK +2y8
2
0.8
4
1.5
0.6
3

1
0.4
2

0.2 0.5
1 y = 3.052x + 1.830 R² = 0.998 y = 0.687x + 0.166 R² = 0.999 y = 1.280x + 0.562 R² = 0.995
y = 3.954x R² = 0.997 y = 0.634x R² = 0.997 y = 1.248x R² = 0.997
0 0 0
0 0.5 1 0 0.5 1 0 0.5 1
4
ApoC-2 TYLPAVDEK +2y7 ApoC-3 GWVTDGFSSLK +2y7 1.25 ApoE SELEEQLTPVAEETR +2y7
0.4
3 1

0.3
0.75
2
0.2
0.5

1
0.1 0.25
y = 0.315x + 0.089 R² = 0.998 y = 3.113x + 0.477 R² = 0.999 y = 3.113x + 0.477 R² = 0.998
y = 3.132x R² = 0.998 y = 3.132x R² = 0.994
y = 0.302x R² = 0.999
0 0 0
0 0.5 1 0 0.5 1 0 0.5 1

Fig. 2 Area ratio (Y-axis) versus concentration ratio (X-axis) for standard addition of purified recombinant
proteins to a spiked matrix (square) and solvent blank (circle) for slope comparison. Error bars indicate
standard deviation (N ¼ 3). The comparable slope indicates that spiked recombinant proteins digest in an
analogous way to endogenous protein in diluted matrix, a critical assumption for this method
304 Christopher A. Toth et al.

ApoA1.THLAPYSDELR+2y8 ApoB.TGISPLALIK+2y6
R2 = 0.9996 R2 = 0.9897

Fig. 3 Demonstration of the linear dynamic range for two selected peptide transitions, apoA1: THLAPYSDELR
2y8 and apoB-100: TGISPLALIK 2y6 in diluted serum calibrator. The nine point calibration curve represents
1:15 and 1:1500 fold dilutions at the ULOQ and LLOQ, respectively

interest in the calibrator serum pool was calculated by dividing

the area ratio y-intercept with the regression slope, multiplied
by the concentration of the protein stock solution (from AAA)
and the dilution factor. To compare the digestion efficiency
between endogenous proteins and their spiked recombinant
analogs, standard addition was also performed in triplicate on a
buffer blank. In the case of apoA-I and apoB-100, standard
addition was performed using certified reference materials
(CRM) from the World Health Organization (WHO), SP1-
01 and SP3-08, respectively, to assign apoA-I and apoB-100
concentrations to the calibrator serum pool.
16. See Fig. 3.
17. When the calibrator pool was diluted to prepare the standard
series, these materials were stable for only 4 days at 4  C. The
frequent re-preparation of the standard series was one of the
main sources of method variability. Storing the calibrator pool
in a range of volumes verified by mass, and dilution only after
thawing provided better calibration reproducibility.
18. For addition of internal standard, the injection sequence
should be programmed to draw the SIL peptide mix and the
protein sample sequentially into the sample loop, then transfer
both simultaneously into the IMER by the 20 μL/min diges-
tion buffer flow. The SIL peptides and the peptide cleavage
product are retained together on the trapping column and
eluted through the analytical column into the LC-MS inter-
face. The authors preferred this method over addition of SIL
peptide mix directly into the sample well, as some SIL peptides
show signs of degradation in dilute endogenous matrix. How-
ever, SIL peptides kept in a concentrated mixture within the
 Nuts and Bolts of Protein Quantification by Online Trypsin Digestion. . . 305

control rack remained stable for several weeks at 4  C. In

addition, the automated addition of internal standard
improved method precision and reduced sample preparation
time and pipetting steps. The Labsolutions pretreatment pro-
gram is shown in Table 1.
19. The combined sample & SIL mix injection-plug is carried
through the IMER by digestion buffer (Subheading 2.3, step
2a) at a rate of 20 μL/min for 7 min (Fig. 4). After exiting the

Table 1
Labsolutions pretreatment program

Line # Value
1 a1¼1001
2 vial a1
3 n.strk ns
4 aspir 5,5
5 air.a 1,5
6 rinse 200,50
7 vial sn
8 n.strk ns
9 aspir iv,ss
10 air.a 1,5
11 rinse 200,50
12 inj.p
13 s.inj
14 end

Fig. 4 Simplified diagram of the IMER-LC-MS/MS system showing sample flow path and valve positions
306 Christopher A. Toth et al.

IMER, the native cleavage products, the SIL peptides, and

remaining matrix components are retained on the trapping/
desalting column, while salts and unretained matrix compo-
nents pass into waste. From 7 to 7.5 min, the flow rate is
increased to 0.5 mL/min to purge out the remaining cleavage
products from the IMER and flow lines into the trap. At
7.5 min, the loop encompassing the trapping column is idled
(with cleavage products and SIL peptides), while the IMER is
washed at 2 mL/min with wash buffer (Subheading 2.3, step
2b) for 2 min, then re-equilibrated with digestion buffer for
2 min.
20. The start of gradient elution is synchronized with the turn of
the trap column switching valve. This allows the elution flow of
the binary gradient pumps to carry the trapped cleavage pro-
ducts and SIL peptides to the analytical column and to the
LC-MS interface. To initiate the peptide transfer from the
trapping column to the analytical column, Solvent B was
ramped in 0.25 min to 8% after injection. Solvent B is then
increased 8–16% during 0.25–7 min, 16–25% during 7–8 min,
and 25–95% during 8–9 min, held at 95% during 9–10.5 min
before re-equilibrating to initial conditions during 10.5–11.5.
For more polar peptides than those described in this method,
initial conditions may require longer loading times at lower
initial solvent B content to avoid significant peak broadening
and/or inconsistent retention times. Example LC parameters
and a corresponding solvent composition curve are shown in
Fig. 5 (Table 2).

Fig. 5 LC solvent composition curve

 Nuts and Bolts of Protein Quantification by Online Trypsin Digestion. . . 307

Table 2
LC instrument parameters

Line Time (min) Hardware Parameter Value

1 0.01 Controller Event 13
2 0.01 Pumps Pump C Flow 0.05
3 0.01 Controller Start
4 0.25 Pumps Pump B Conc. 8
5 1 Pumps Pump C Flow 0.05
6 1.01 Pumps Pump C Flow 0.02
7 7 Pumps Pump B Conc. 16
8 7.01 Pumps Pump C Flow 0.02
9 7.11 Pumps Pump C Flow 0.5
10 7.61 Controller Event 123
11 7.61 Pumps Pump C Flow 0.5
12 7.61 Pumps SV(Pump C) D
13 7.71 Pumps Pump C Flow 2
14 8 Pumps Pump B Conc. 25
15 9.01 Pumps Pump B Conc. 95
16 9.61 Pumps SV(Pump C) B
17 10.51 Pumps Pump B Conc. 95
18 10.61 Pumps Pump B Conc. 2
19 11.55 Pumps Pump C Flow 2
20 11.61 Pumps Pump C Flow 0.1
21 11.7 Controller Stop

21. The modified valve design shown in Fig. 6a, b allows simulta-
neous digestion and LC separation. This is accomplished by
employing a 10-port switching valve and two identical trapping
columns. While the cleavage products and SIL peptides are
collected on one column, the cleavage products and SIL pep-
tide from the previous sample eluted to the analytical column
and to the LC-MS interface. This alternating setup effectively
reduces total analysis time to the length of the LC gradient.
Additionally, the valve design reverses the direction of the flow
on the trapping column during desalting and elution, refocus-
ing analytes onto the analytical column and reducing peak
broadening.
22. See Fig. 7 and Table 3.
Fig. 6 Dual trapping valve design and valve positions corresponding to LC methods A and B

1
3.0e6
3 15

2 4
2.0e6
Intensity, cps

11 13 6
8 10 9
1.0e6
18
14 12
17
16 7

0.0
1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0
Time, min

Fig. 7 Representative chromatogram: 50 μL of 1:100 diluted serum calibrator (gray) and 5 μL SIL mix (black)
injected into system. Corresponding peak IDs are listed in Table 3
Table 3
MRM instrument parameters for native and labeled peptides

Native Label

Fragmention Fragmention

ID Protein Peptide sequence Precursor ion m/z 1 2 DP CE Peptide sequence Precursor ion m/z 1 2
*
1 apoA-I AKPALEDLR 506.8 716.4 813.4 68.1 27.1 AKPALED(L+7)R 510.3 723.4 820.5
*
2 ATEHLSTLSEK 608.3 664.4 777.4 75.5 30.8 ATEHLST(L+7)SEK 611.8 671.4 784.5
*
3 THLAPYSDELR 651.3 950.5 1063.5 78.6 32.3 THLAPYSDE(L+7)R 654.8 957.5 1070.6
*
4 apoA-II EQLTPLIK 471.3 470.3 571.4 65.5 25.8 EQLTPL(I+7)K 474.8 477.4 578.4
*
5 apoA-IV LEPYADQLR 552.8 765.4 862.4 71.4 28.8 LEPYADQ(L+7)R 556.3 772.4 869.5
*
6 LTPYADEFK 542.3 772.4 869.4 70.6 28.4 LTPYADEF(K+8) 546.3 780.4 877.4
*
7 apoB ATGVLYDYVNK 621.8 801.4 914.5 76.4 31.2 ATGVLYDY(V+6)NK 624.8 807.4 920.5
*
8 AAIQALR 371.7 487.3 600.4 58.2 22.2 AAIQA(L+7)R 375.2 494.3 607.4
*
9 TGISPLALIK 506.8 654.5 741.5 68.1 27.1 TGISPLAL(I+7)K 510.3 661.5 748.5
*
10 apoC-I EWFSETFQK 601.3 739.4 886.4 75 30.5 EWFSETFQ(K+8) 605.3 747.4 894.4
*
11 TPDVSSALDK 516.8 620.3 834.4 68.8 27.5 TPDVSSA(L+7)DK 520.3 627.3 841.4
*
12 apoC-II ESLSSYWESAK 643.8 870.4 957.4 78 32 ESLSSYWESA(K+8) 647.8 878.4 965.4
*
13 TYLPAVDEK 518.3 658.3 771.4 68.9 27.5 TYLPA(V+6)DEK 521.3 664.4 777.4
*
14 apoC-III DALSSVQESQVAQQAR 858.9 887.5 1016.5 93.7 39.8 DALSSVQESQVAQ(Q+7)AR 862.4 894.5 1023.5
*
15 GWVTDGFSSLK 598.8 753.4 854.4 74.8 30.4 GWVTDGFSS(L+7)K 602.3 760.4 861.4
*
16 apoE AATVGSLAGQPLQER 749.4 770.4 827.4 85.7 35.8 AATVGSLAGQPL(Q+7)ER 752.9 834.4 905.5
*
17 LGPLVEQGR 484.8 588.3 489.2 66.5 26.3 LGPLVE(Q+7)GR 488.3 496.3 595.3
*
Nuts and Bolts of Protein Quantification by Online Trypsin Digestion. . .

18 LQAEAFQAR 517.3 721.4 792.4 68.8 27.5 LQAEAF(Q+7)AR 520.8 728.4 799.4
Product ions marked with (*) were chosen for quantitation
309
310 Christopher A. Toth et al.

1.25e5
2.8e6

1.00e5

2.0e6
Intensity, cps

1st Blank
1.0e6
2nd Blank

THLAPYSDELR
TGISPLALIK
1.00e4

0.0 0.00
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5
Time, min

Fig. 8 Example of carry-over contribution to the native signal in one and two blank injections (50 μL of digest
buffer) immediately following a highly concentrated sample (50 μL of 1:15 diluted human serum on-column).
The intensities are plotted for two peptide transitions with contrasting native protein concentrations,
THLAPYSDELR (apoA-I; left axis) and TGISPLALIK (apoB; right axis)

23. Carry-over contribution for the two peptides was calculated at

0.2 and 1.3% after one blank injection, respectively, and 0.05
and 1.0% after two blank injections, respectively (Fig. 8).

Disclaimer

References in this article to any specific commercial products, pro-

cess, service, manufacturer, or company do not constitute an
endorsement or a recommendation by the U.S. Government or
the Centers for Disease Control and Prevention. The findings and
conclusions in this report are those of the authors and do not
necessarily represent the views of CDC.
 Nuts and Bolts of Protein Quantification by Online Trypsin Digestion. . . 311

References
1. Picotti P, Aebersold R (2012) Selected reaction 1461:84–91. https://doi.org/10.1016/j.
monitoring-based proteomics: workflows, chroma.2016.07.058
potential, pitfalls and future directions. Nat 7. Toth CA, Kuklenyik Z, Jones JI, Parks BA,
Methods 9(6):555–566. https://doi.org/10. Gardner MS, Schieltz DM, Rees JC, Andrews
1038/nmeth.2015 ML, McWilliams LG, Pirkle JL, Barr JR (2017)
2. Shuford CM, Walters JJ, Holland PM, On-column trypsin digestion coupled with
Sreenivasan U, Askari N, Ray K, Grant RP LC-MS/MS for quantification of apolipopro-
(2017) Absolute protein quantification by mass teins. J Proteome 150:258–267. https://doi.
spectrometry: not as simple as advertised. Anal org/10.1016/j.jprot.2016.09.011
Chem 89(14):7406–7415. https://doi.org/10. 8. Hoofnagle AN, Whiteaker JR, Carr SA, Kuhn E,
1021/acs.analchem.7b00858 Liu T, Massoni SA, Thomas SN, Reid
3. Shuford CM, Sederoff RR, Chiang VL, Muddi- Townsend R, Zimmerman LJ, Boja E, Chen J,
man DC (2012) Peptide production and decay Crimmins DL, Davies SR, Gao Y, Hiltke TR,
rates affect the quantitative accuracy of protein Ketchum KA, Kinsinger CR, Mesri M, Meyer
cleavage isotope dilution mass spectrometry MR, Qian WJ, Schoenherr RM, Scott MG,
(PC-IDMS). Mol Cell Proteomics 11 Shi T, Whiteley GR, Wrobel JA, Wu C, Acker-
(9):814–823. https://doi.org/10.1074/mcp. mann BL, Aebersold R, Barnidge DR, Bunk
O112.017145 DM, Clarke N, Fishman JB, Grant RP,
4. Regnier FE, Kim JH (2014) Accelerating trypsin Kusebauch U, Kushnir MM, Lowenthal MS,
digestion: the immobilized enzyme reactor. Moritz RL, Neubert H, Patterson SD, Rock-
Bioanalysis 6(19):2685–2698. https://doi. wood AL, Rogers J, Singh RJ, Van Eyk JE,
org/10.4155/bio.14.216 Wong SH, Zhang S, Chan DW, Chen X, Ellis
5. Kuklenyik Z, Jones JI, Toth CA, Gardner MS, MJ, Liebler DC, Rodland KD, Rodriguez H,
Pirkle JL, Barr JR (2017) Optimization of the Smith RD, Zhang Z, Zhang H, Paulovich AG
linear quantification range of an online trypsin (2016) Recommendations for the generation,
digestion coupled liquid chromatography–tan- quantification, storage, and handling of peptides
dem mass spectrometry (LC–MS/MS) plat- used for mass spectrometry-based assays. Clin
form. Instrum Sci Technol 46:1–13. https:// Chem 62(1):48–69. https://doi.org/10.
doi.org/10.1080/10739149.2017.1311912 1373/clinchem.2015.250563
6. Bonichon M, Combès A, Desoubries C, 9. Grant RP, Hoofnagle AN (2014) From lost in
Bossée A, Pichon V (2016) Development of translation to paradise found: enabling protein
immobilized-pepsin microreactors coupled to biomarker method transfer by mass spectrome-
nano liquid chromatography and tandem mass try. Clin Chem 60(7):941–944. https://doi.
spectrometry for the quantitative analysis of org/10.1373/clinchem.2014.224840
human butyrylcholinesterase. J Chromatogr A
 Chapter 20

Proteases: Pivot Points in Functional Proteomics

Ingrid M. Verhamme, Sarah E. Leonard, and Ray C. Perkins

Abstract
Proteases drive the life cycle of all proteins, ensuring the transportation and activation of newly minted,
would-be proteins into their functional form while recycling spent or unneeded proteins. Far from their
image as engines of protein digestion, proteases play fundamental roles in basic physiology and regulation at
multiple levels of systems biology. Proteases are intimately associated with disease and modulation of
proteolytic activity is the presumed target for successful therapeutics. “Proteases: Pivot Points in Functional
Proteomics” examines the crucial roles of proteolysis across a wide range of physiological processes and
diseases. The existing and potential impacts of proteolysis-related activity on drug and biomarker develop-
ment are presented in detail. All told the decisive roles of proteases in four major categories comprising
23 separate subcategories are addressed. Within this construct, 15 sets of subject-specific, tabulated data are
presented that include identification of proteases, protease inhibitors, substrates, and their actions. Said data
are derived from and confirmed by over 300 references. Cross comparison of datasets indicates that
proteases, their inhibitors/promoters and substrates intersect over a range of physiological processes and
diseases, both chronic and pathogenic. Indeed, “Proteases: Pivot Points . . .” closes by dramatizing this very
point through association of (pro)Thrombin and Fibrin(ogen) with: hemostasis, innate immunity, cardio-
vascular and metabolic disease, cancer, neurodegeneration, and bacterial self-defense.

Key words Protease, Peptidase, Proteolysis, Protease inhibitor, Protease promoter, Digestion,
Hemostasis, Complement system, Immune regulation, Signaling, Cell migration, Cell proliferation,
Programmed cell death, Protein secretion, DNA replication, DNA repair, DNA processing, Intra-
nuclear proteolysis, Transmembrane proteolysis, Intramembrane proteolysis, Cytosolic proteolysis,
Epigenetics, Inflammation, Cardiovascular disease, Metabolic disease, Stroke, Cancer, Neurodegener-
ative disease, Autoimmune disease, Infectious organisms, Drug target, Drug development, Biomarker
development, Precision medicine

1 Introduction

1.1 Proteases: More One or more proteolytic events initiate the active life for many
Than Just Protein proteins and proteolytic events terminate the relatively short life
Digestion of all proteins. Roughly a third of proteins, as translated, include
peptides that influence proteins’ life span (initiator methionine),
govern their transport (signal peptide), direct them to the appro-
priate organelle (transit peptide), and produce their active form
(propeptide). These processes are merely those proteolytic events

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_20, © Springer Science+Business Media, LLC, part of Springer Nature 2019

313
314 Ingrid M. Verhamme et al.

Translated Protein

Initiator Methionine Removal

Recycling

Shippable Protein Amino Acids

Protease-Driven
Protein Life Cycle
Signal Peptide Removal Lysosome,
Proteasome Degradation

Destination Protein Active Protein

Transit Peptide/ProPeptide Removal

Fig. 1 Stages of the Protease-Driven Protein Life Cycle

that begin the active life of proteins. Over the course of lifetimes
that may be as brief as 11 min or as long as 4 months, additional
proteolytic events remake individual proteins—one example being
the much studied A4-Human amyloid precursor transmembrane
protein whose 13 proteolytic products include a family of
pathology-related, β amyloid peptides. Having served their tempo-
rary purposes, all proteins are once again substrates for proteases
with digestion proceeding to amino acids recycled as raw materials
for ongoing expression of new proteins. In the light of these varied
and continuous processes, proteolytic activity literally reshapes the
proteome. Of course, this brief discourse is but the tip of the
iceberg. Proteolytic enzymes—and their promoters and inhibi-
tors—are essential actors in nearly all biological processes. This
chapter addresses many of those processes especially in relation to
disease and development of new therapeutics. Truly, proteases give
definition to the Functional Proteome—one protein at a time
(Fig. 1).
In the 1830s, parallel studies in Europe and the USA demon-
strated that digestion of proteins in food required a compound in
the gastric mucosa, in addition to gastric juice hydrochloric acid
[1]. The discovery of pepsin triggered the formulation of popular
therapeutic preparations for dyspepsia, and intensified pepsin puri-
fication efforts. However, it would take nearly a century before
crystallization of pure pepsin was achieved. Northrop, Kunitz,
and Herriott published the earliest characterization of classical
digestive proteases such as pepsin, chymotrypsin, trypsin, pancreatic
carboxypeptidases, and their zymogens as pure compounds
 Proteases: Pivot Points in Functional Proteomics 315

[2]. The protease field has come a long way since the discovery of
pepsin. Rapidly developing methods for protein purification,
sequencing, structure-function analysis, X-ray crystallography, and
synthetic substrate development have since then accelerated the
identification of countless mammalian, plant, fungal, and bacterial
proteolytic enzymes [3], and the number of identified sequences is
growing exponentially. As of September 2017, the MEROPS data-
base of proteolytic enzymes (https://www.ebi.ac.uk/merops/)
listed more than one million sequences [4].
The complete ensemble of human proteases, known as the
protease degradome, currently consists of 588 proteases, organized
into five classes: aspartic, cysteine, metallo-, serine, and threonine
proteases [5]. They represent >2% of the human genome, and a
database is maintained by the lab of López-Otı́n (http://
degradome.uniovi.es) [6]. Protease systems have gained consider-
able recognition since the realization that they are important reg-
ulators of countless biological mechanisms, and not simply part of a
machinery for nonspecific protein digestion. Blood coagulation,
fibrinolysis, complement activation, peptide hormone processing,
protein secretion and degradation, DNA replication and repair, cell
signaling and proliferation, and programmed cell death are just a
few of these processes in which physiological proteases target spe-
cific substrates. The peptidase field is rapidly evolving, and extensive
volumes have already been dedicated to cataloguing and docu-
menting its recently discovered and previously known members
[7]. Here we focus on novel aspects of the intimate connection
between disease and proteolysis, and the potential of mechanism-
based drug targeting.

1.2 Physiological Physiological protease activity is strictly controlled to avoid indis-

and Regulatory Roles criminate and unwanted protein degradation [8]. Proteases may be
present as inactive zymogens requiring a proteolytic activation step
followed by a conformational rearrangement to form the active site
(hemostatic protease zymogens); complexed with an inhibitory
propeptide or N-terminal domain that blocks the active site (matrix
metalloproteinases or MMPs); or in a low-reactive state requiring
allosteric activation (factor VII and tissue factor), proteolysis (single
chain tPA), or di/multimerization (caspase-8 and -9, proteasome
proteases). Active proteases may be pH-controlled, and examples of
this are lysosomal cathepsins and gastric pepsin. Many proteases are
transcriptionally regulated and only expressed in specific cells and
tissues, sometimes temporally restricted, whereas housekeeping
proteases are expressed constitutively. Physiological triggers such
as inflammation may temporarily upregulate zymogen expression,
as seen for the precursors of cathepsins and MMPs.
Because many proteases are associated with other proteins and
act within networks of effectors, ligands, and receptors, studying
their biological functions has become increasingly complex. In vitro
316 Ingrid M. Verhamme et al.

delineation of structural and kinetic properties of a protease is

required for defining its mechanisms of substrate and inhibitor
specificity, but its catalytic efficiency is often profoundly affected
in vivo by complex macromolecular interactions. Protease activity
may be up- or downregulated by physiological feedback mechan-
isms, processes at the genomic level, and stressors in the molecular
environment. Depending on the molecular environment and the
binding partners, proteases may catalyze reactions that result in
opposite physiological processes. Thrombin, the central protease
in hemostasis, is a good example of this. Uncomplexed thrombin is
procoagulant, and cleaves fibrinogen to fibrin to form a thrombus.
However, when bound to thrombomodulin, thrombin activates
the anticoagulant protein C. The proteases of the contact activation
system in the intrinsic coagulation pathway aid in sustaining clot-
ting once the extrinsic pathway is activated, but they also activate
plasminogen in the fibrinolytic pathway.
Protease networks are regulated by intrinsic inhibitors of a
protein or polypeptide nature. Serine proteases feature predomi-
nantly in coagulation, fibrinolysis, and the complement system, and
also play roles in digestion, late stage apoptosis, development,
fertilization, and membrane-associated signaling. They are typically
inhibited by serpins (serine proteinase inhibitors) in irreversible
covalent complexes, or by polypeptide inhibitors of the Kunitz,
Kazal, or elafin types, with a protease-binding loop in a conserved
canonical backbone conformation, complementary to the protease
active site. Inhibitory serpins belong to a superfamily of ubiquitous
proteins with a typical fold that includes a metastable reactive center
loop (RCL) for baiting their target proteases, and a core consisting
of 3 β-sheets and 8 or more α-helices [9]. Upon cleavage of the
RCL, the protease stays attached, and the resulting acyl-enzyme
undergoes a dramatic conformational rearrangement during which
the cleaved RCL inserts as an additional strand in β-sheet A, and the
attached protease translocates to the distal end of the serpin. Defor-
mation of the active site prohibits completion of hydrolysis, with a
stable covalent complex as the end product. Kunitz, Kazal, and
elafin-like inhibitors act by a different mechanism of tight binding
in a substrate-like mode, followed by slow and reversible cleavage of
the reactive bond [10]. The cleaved inhibitor stays bound to the
protease but typically loses inhibitory potency by several orders of
magnitude.
Analysis of known cleavage events has shown that regulation of
the proteome in vivo occurs through an interconnected protease-
inhibitor web [11], with protein protease inhibitors being protease
substrates themselves. Many inhibitors target groups of related
proteases rather than single enzymes, and proteolytic inactivation
of one inhibitor may represent a key on/off switch for an entire
protease subnetwork. Multiple regulatory mechanisms of proteoly-
sis have been identified, such as the presence of PEST (proline,
glutamic acid, serine, threonine) sequences in intracellular proteins
 Proteases: Pivot Points in Functional Proteomics 317

targeted for rapid degradation; KFERQ motifs (lysine, phenylala-

nine, glutamic acid, arginine, glutamine) guiding cytosolic proteins
to the endosome or lysosome for degradation; and RxxLxxIxN
destruction box motifs that tag proteins for degradation by the
ubiquitin-proteasome system [12]. Whereas the protease/sub-
strate/inhibitor interconnectedness adds many levels of complex-
ity, it also provides more opportunities for design of new
therapeutic approaches.

1.3 Proteases Protease functionality often depends on the concerted action of a

and Disease catalytic domain and specific nonenzymatic domains and modules,
either incorporated in the protein or as separate entities associated
with the catalytic domain. Anion-binding sites, kringle and apple
domains, epidermal growth factor (EGF) and fibronectin domains,
thrombospondin repeats and transmembrane domains have diverse
functions such as localization; recognition of substrates, inhibitors,
and effectors; and interactions with various ligands, cofactors, and
other proteins. Many of these domains are evolutionary conserved,
and are present in diverse proteases either as single units (e.g., EGF
domains) or as repeats (kringle and apple domains, thrombospon-
din domains). Not surprisingly, defects in the catalytic domain or
any of the regulatory domains result in a physiologically dysfunc-
tional protease. Intrinsic protease dysregulation is a hallmark of
many pathologies such as inflammation, cancer, hemostatic and
autoimmune diseases, and neurodegeneration [13, 14]. A selection
of recent findings with regard to abnormal protease activity in these
pathologies is discussed below. The López-Otı́n degradome data-
base currently lists 124 hereditary diseases due to protease muta-
tions, and many other pathologies are associated with
posttranslational and epigenetic changes in protease activity.
Although much progress has recently been made in inflammation-
related protease research, many processes are still unclear with
respect to upregulation of protease activity as a cause or conse-
quence. However there is ample consensus about the importance
of proteases as attractive drug targets in many disease states. Effec-
tor ligands and substrates may provide additional targets. Protease
up- or downregulation may be used as a diagnostic, and proteolytic
generation of signal peptides may provide a previously underappre-
ciated arsenal of biomarkers in various disease states.

2 Regulation of Physiological Processes

2.1 Hemostasis If you are familiar with only one proteolytically regulated process in
the human body, it is likely hemostasis. Featured prominently in
documentaries about European royalty and in nightly medication
commercials, the formation and degradation of blood clots has
been intensely studied. It is a model system for illustrating the
complexity of proteolysis: multiple pathways acting both in concert
318 Ingrid M. Verhamme et al.

and in opposition to each other, dozens of actors, extensive regula-

tion at each step, and even individual actors performing opposing
roles under different conditions. Each step is an opportunity for
both dysfunction and intervention, and both affects and is affected
by processes in cells throughout the body (Table 1).
In primary hemostasis, vascular injury exposes the highly
thrombogenic subendothelium, and platelets are recruited to the
site of injury, where they are activated and form a platelet plug. In
the “secondary hemostasis” cascade model of coagulation, pro-
posed more than 50 years ago, vascular injury triggers a stepwise
amplification of clotting factor activity. This culminates in the for-
mation of thrombin, the central coagulation protease, that cleaves
fibrinogen to fibrin [15]. In the extrinsic pathway, plasma factor
VIIa forms a highly reactive complex with tissue factor exposed

Table 1
Hemostasis clot formation: activities and Uniprot ID codes (where applicable) of proteases, protease
inhibitors, and cofactors associated with hemostasis, specific to clot formation

Protease/inhibitor Uniprot Action

Thrombin (Prothrombin) P00734 Converts fibrinogen to fibrin
Activates factors V, VII, VIII, XI, XIII
Complexes with thrombomodulin
Thrombin/thrombomodulin activates protein C
Plasma factor VIIa (Coagulation factor VII) P08709 Complexes with Tissue Factor
VIIa/TF converts/activates X to Xa
VIIa/TF converts/activates IX to IXa
Tissue factor P13726 Propagates coagulation protease cascade
Complexes with phospholipids
Complexes with circulating factor VII or VIIa
VIIa/TF converts/activates X to Xa
VIIa/TF converts/activates IX to IXa
Coagulation factor X P00742 Converts/activates prothrombin to thrombin
Complexes with phospholipids & calcium
Activates factor VII to form factor VIIa
Prekallikrein (Plasma kallikrein) P03952 Reciprocal activation of factor XII
Coagulation factor XII P00748 Reciprocal activation of Prekallikrein
Coagulation factor IX P00740 Converts/activates factor X
Activates factor VII to form factor VIIa
Activates factor X to form factor Xa
Antithrombin (Antithrombin-III) P01008 Inhibits thrombin and factors IXa, Xa and XIa
Activity enhanced by heparin
Heparin cofactor II P05546 Inhibits thrombin and factors IXa, Xa and XIa
Inhibits chymotrypsin
TFPI (Tissue factor pathway inhibitor) P10646 Inhibits factor X (Xa)
TFPI + Xa inhibits VIIa/tissue factor
 Proteases: Pivot Points in Functional Proteomics 319

during injury, and sequential activation of factor X and prothrom-

bin leads to clot formation. The intrinsic or contact activation
pathway consists of prekallikrein, and factors XII and IX. Stepwise
activation of factors XII and XI generates factor IXa that sustains
formation of factor Xa. Except for factor XIIIa, a transaminase
that crosslinks fibrin, all the hemostatic enzymes are serine pro-
teases. Intrinsic serine protease inhibitors (antithrombin, heparin
cofactor II) and the Kunitz inhibitor TFPI provide regulatory
control. Phospholipid surfaces and the nonenzymatic cofactors
VIII and V are required for activation of factor X and prothrombin,
respectively. This “waterfall” mechanism did not explain why factor
XII-deficient patients do not have a bleeding tendency, and a new
“hemostatic network” mechanism was discovered in which throm-
bin activates factor XI to sustain hemostasis [16, 17]. Whereas the
contact activation proteases are not critical in normal hemostasis,
animal studies have shown that they are important contributors to
formation of pathologic intravascular thrombi [18], and may be
suitable targets for therapeutic inhibitors.
When the blood clot has served its purpose, endothelial cells
release tPA that converts plasminogen to plasmin on the fibrin
surface. Both tPA and plasminogen bind to fibrin through their
kringle structures (binding domains specific to blood clotting fac-
tors), resulting in significant enhancement of the rate of plasmin
formation. Plasmin degrades fibrin which exposes additional
carboxy-terminal lysines for interaction with the kringles on tPA,
plasminogen and plasmin, ultimately resulting in accelerated fibrin
degradation. In a regulatory process, carboxypeptidase U, also
known as thrombin-activatable fibrinolysis inhibitor (TAFIa),
removes carboxy-terminal lysine residues and stabilizes fibrin
thrombi. TAFI proved to be a poor substrate for thrombin in the
absence of thrombomodulin, and a riveting chronicle of the
lengthy process of its discovery, purification and characterization
again demonstrates the gratifying aspect of understanding the
molecular basis of a biochemical process [19]. The serpins plasmin-
ogen activator inhibitor-1 (PAI-1) and α2-antiplasmin (α2-AP),
respectively targeting tPA and plasmin, provide additional regula-
tion of fibrin degradation. Plasmin bound to fibrin is protected
from α2-AP, whereas TAFI decreases this protection by eliminating
plasmin-binding lysine residues on fibrin (Table 2).
Platelets can both promote and counteract fibrinolysis. Acti-
vated platelets localize plasminogen and its physiological activators
tPA and urokinase via the GPIIb/IIIa (integrin αIIbβ3) complex.
Thrombospondin, released from platelet granules and exposed on
the platelet surface, also binds plasminogen and enhances its activa-
tion. Hence activated platelets provide an alternative surface for
promoting fibrinolysis. As part of a regulatory mechanism, platelets
secrete two antifibrinolytic serpins, PAI-1 and α2-AP, and as a result
platelet-rich thrombi resist fibrinolysis.
320 Ingrid M. Verhamme et al.

Table 2
Hemostasis clot degradation: activities and Uniprot ID codes (where applicable) of proteases,
protease inhibitors, and cofactors associated with hemostasis, specific to clot degradation

Protease/inhibitor Uniprot Action

tPA (Tissue-type plasminogen activator) P00750 Converts plasminogen to plasmin on the fibrin
surface
Displaces plasmin from fibrin, promoting
inhibition by alpha-2-antiplasmin
Plasmin (Plasminogen) P00747 Dissolves the fibrin of blood clots
carboxypeptidase U (carboxypeptidase B2, Q96IY4 Removes C-terminal lysine residues from fibrin
thrombin-activatable fibrinolysis Down-regulates fibrinolysis
inhibitor3) Cleaves complement proteins C3a and C5a
Activated by thrombin/thrombomodulin
complex
Plasminogen activator inhibitor 1 P05121 ‘Bait’ for tissue plasminogen activator,
urokinase, protein C and matriptase-
3/TMPRSS7
α2-antiplasmin (Alpha-2-antiplasmin) P08697 Inhibits plasmin and trypsin
Inactivates matriptase-3/TMPRSS7 and
chymotrypsin
Protein C (Vitamin K-dependent protein C) P04070 Inactivates factors Va and VIIIa in the presence
of calcium ions and phospholipids
Activated by Thrombin/thrombomodulin
complex

Thrombin may act both as a procoagulant and as an anticoagu-

lant enzyme, and has been called a “Janus-headed” protease
[20]. Crystal structures of α-thrombin bound to numerous effec-
tors have aided in identifying extended recognition sites on its
surface. By combining these discrete functional surface regions
thrombin interacts with various substrates and ligands in a selective
and specific manner. Binding of Na+ causes thrombin to adopt a
“fast” conformation which rapidly cleaves procoagulant substrates.
In the Na+-free state, thrombin bound to thrombomodulin prefer-
entially initiates the protein C anticoagulant pathway in which
factors V and VIII are proteolytically inactivated [21, 22].

2.2 The Complement The plasma complement system regulates the innate immune
System and Immune defense by opsonization and elimination of pathogens, cell debris,
Regulation and host cells that have undergone alterations [23–25]. Activation
of the complement system occurs via three pathways: in the classical
pathway, the recognition protein C1q binds to antigen-antibody
complexes, the C1 complex is activated, and a series of serine
protease activation reactions leads to formation of C3- and ulti-
mately C5-convertase; in the lectin pathway, mannose-binding
lectin binds to mannose on the surface of pathogens as a trigger
 Proteases: Pivot Points in Functional Proteomics 321

to formation of C3/C5-convertase; in the alternative pathway,

continuous low level activation of C3 and binding to the pathogen
leads to C3/C5-convertase. C3-convertases cleave C3 into the
anaphylatoxin C3a and the opsonin C3b which is deposited on
the pathogen surface and facilitates targeting by macrophages.
C5-convertases generate the pro-inflammatory anaphylatoxins
C5a and C5b, and a membrane attack complex (MAC) is formed
by C5b–C9 assembly. This complex forms a pore in the membrane
that kills the pathogen or the targeted cell. Anaphylatoxins C3a and
C5a promote chemotaxis of immune cells. The complement serine
proteases involved in these complicated interrelated processes
include C1r, C1s, MASPs 1-3 [26], C2, and Factors B, D, and I,
all with restricted specificity. The plasma serpin, C1-inhibitor, cova-
lently inactivates C1r, C1s, and MASPs 1 and 2. C2 and Factor B
activity is controlled by the Regulation of Complement Activation
(or RCA) proteins, and no endogenous inhibitors are known for
Factors I and D. Complement deficiency causes increased suscepti-
bility to infection, and clearance impairment of immune complexes
and apoptotic cells results in the development of systemic lupus
erythematosus (SLE). However, excessive complement activation is
also associated with autoimmune diseases such as SLE, rheumatoid
arthritis, and certain cancers. The monoclonal antibody Eculizu-
mab inhibits C5, and was recently approved for the treatment of
complement hyperactivation in paroxysmal nocturnal hemoglobin-
uria and atypical hemolytic uremic syndrome. It may ultimately
prove useful in the treatment of SLE as well [27] (Table 3).
Human cytotoxic T lymphocytes and natural killer cells secrete
five types of granzymes (A, B, H, K, M), serine proteases that aid in
the neutralization of virus-infected and tumor cells. Only granzyme
B and M have known intracellular inhibitors, serpinB9 (PI-9) and
serpinB4 (SCCA2), respectively [28]. Increased PI-9 expression
may be an immune evasion mechanism used by lung cancer cells
for protection from granzyme B-mediated cytotoxicity [29].
Immune regulation is tightly associated with proteolytic pro-
cesses in the gut. In immune diseases of the gut, cytokines upregu-
late protease activity, resulting in inflammation and exacerbated
immune response [30]. In inflammatory bowel disease, MMPs,
neutrophil elastase, and cathepsins are typically overexpressed in
the gut epithelium and basement membrane. The nature of the gut
microbiome is equally important for maintaining immune homeo-
stasis, and commensal and pathogenic bacteria produce a wide
range of proteases that differentially affect the integrity of the
intestinal mucosa.
The serpin α1-antitrypsin (now renamed α1-proteinase inhibi-
tor or α1-PI), produced in the liver, protects the lungs from inflam-
matory neutrophil elastase damage, and it is also an acute-phase
protein that reduces pro-inflammatory cytokine production, inhi-
bits apoptosis, blocks leukocyte degranulation and migration, and
322 Ingrid M. Verhamme et al.

Table 3
Complement system and immune regulation: activities and Uniprot ID codes (where applicable) of
proteases, protease inhibitors, and cofactors associated with the complement system and immune
regulation

Protease/inhibitor Uniprot Action

C1R (Complement C1r P00736 Cleaves/Activates C1s
subcomponent, Classical)
C1S (Complement C1s P09871 Cleaves/Activates C2 and C4
subcomponent, Classical)
C2a (Complement C2, C1s P06681 Combines with C4b to form C3 convertase (classical,
cleavage product) lectin)
Factor D (Complement P00746 Cleaves/Activates Complement Factor B
factor D, Bb fragment)
Factor B (Complement factor B, P00751 Cleavage Product Bb combines with C3b to form C3
Bb fragment) Convertase (Alternative)
C3-convertase (Classical, Lectin: Cleaves C3 into the anaphylatoxin C3a and the opsonin
C4bC2a) C3b
C3-convertase (Alternative: Cleaves C3 into the anaphylatoxin C3a and the opsonin
C3bBb) C3b
C3-convertase (Aqueous: C3: Cleaves C3 into the anaphylatoxin C3a and the opsonin
H2O) C3b
C5-convertase (Classical: Cell Cleaves C5 into the anaphylatoxin C5a and the MAC
Membrane, C4b2b3b) component C5b
C5-convertase (Alternative: Cell Cleaves C5 into the anaphylatoxin C5a and the MAC
Membrane, C3bBbC3b) component C5b
C5-convertase (Classical: Fluid, Cleaves C5 into the anaphylatoxin C5a and the MAC
C4b2boxy3b ) component C5b
MASP 1 (Mannan-binding P48740 Activates MASP2 or C2 or C3
lectin serine protease 1)
MASP 2 (Mannan-binding O00187 Cleaves/Activates C2 and C4
lectin serine protease 2)
MASP 3 (Mannan-binding P48740a Cleaves/Activates complement pro-factor Da Alternative
lectin serine protease 3) Splicing Product
C1-inhibitor (Plasma protease P05155 Complexes with/Inactivates C1r, C1s, MASP 1, MASP
C1 inhibitor)a 2, chymotrypsin, kallikrein, fXIa, FXIIa
Factor I (Complement factor I) P05156 Cleaves/Inactivates C3b, iC3b, and C4b
a
Complement activation, blood coagulation, fibrinolysis, and the generation of kinins

modulates local and systemic inflammatory responses [31]. In

monocytes, α1-antitrypsin increases intracellular cAMP, regulates
CD14 expression, and suppresses NF-κB nuclear translocation.
These functions may be related to the inhibitory activity of
 Proteases: Pivot Points in Functional Proteomics 323

antitrypsin, protein-protein interactions, or both. Preclinical use of

antitrypsin in autoimmunity and transplantation models showed
that it is capable of preventing or reversing autoimmune disease and
graft loss.

2.3 Proteolytic The lysosome and the ubiquitin-proteasome are the two major
Processing intracellular proteolytic systems keeping the protein pool in bal-
ance. Originally considered strictly degradative, these systems have
revealed regulatory functions beyond catabolism, and their molec-
ular defects are associated with various disease states. Lysosomes
contain cathepsins B, D, and L in addition to lipases, nucleases,
glycosidases, phospholipases, phosphatases, and sulfatases that are
active in an acidic milieu [32]. Lysosomes regulate autophagy
during nutrient starvation, and participate in development and
differentiation, induction of cathepsin-dependent cell death, and
degradation of apoptotic cells. Cancer cell lysosomes have a higher
membrane permeability (“leaky”) and express more cathepsin than
those of normal cells, and this property may be exploited in cancer
treatment. Agents such as tetrahydrocannabinol and chloroquine
may disrupt the lysosome and trigger killing of the cancer cells
(Table 4).
The ubiquitin-proteasome, an intracellular high molecular
weight protease complex predominantly located in the cytosol,
selectively degrades proteins tagged with ubiquitin at lysine resi-
dues [33]. Its “central pore” contains several inward facing protease
active sites, with caspase-, trypsin-, and chymotrypsin-like specific-
ity. This multi-protein construct is capped by one or two activator
complexes that conformationally regulate access of protein sub-
strates to the pore. Whereas the physiological function of the
proteasome was originally thought to be restricted to intracellular
protein catabolism, new functions have been discovered with
respect to regulation of the cell cycle progression, gene expression,
and responses to cellular stress [34]. Protein ubiquitination is
reversible, and more than 100 potentially regulatory deubiquitinase
(DUB) genes have been identified, mainly cysteine and metallopro-
teinases (MMPs). DUBs rescue proteins from degradation and
reverse ubiquitination-induced signaling. The immunoprotea-
some, containing specific subunits with increased chymotrypsin-
and trypsin-like activities, and decreased caspase-like activity, parti-
cipates in production of peptide epitopes for cytotoxic T lympho-
cytes. In the thymoproteasome, chymotryptic activity is attenuated
but the caspase- and trypsin-like activities are conserved. Its peptide
products are MHC class I ligands with moderate avidity, which
supports positive selection of CD8+ T cells [35]. Foreign peptides,
generated during the breakdown of virus and cancer cell proteins,
bind MHC class I molecules on the cell surface, and the cells are
recognized by cytotoxic T cells as potentially dangerous and are
destroyed. The proteasome discards misfolded proteins, and
324 Ingrid M. Verhamme et al.

Table 4
Proteolytic processing: activities and Uniprot ID codes (where applicable) of proteases, protease
inhibitors, and cofactors associated with proteolytic processing

Protease/inhibitor Uniprot Action

Cathepsin B P07858 Intracellular degradation and protein turnover
Upregulation of Cathepsin D, matrix metalloproteinase,
and urokinase
Implicated in metastasis and immune resistance
Cathepsin D P07339 Intracellular degradation and protein turnover
Used by macrophages to degrade bacterial proteins
Activates ADAM30, implicated in Alzheimer’s progression
Implicated in metastasis in breast cancer
Cathepsin L1 P07711 Intracellular degradation and protein turnover
Degrades collagen and elastin
Degrades alpha-1 protease inhibitor
Deubiquitinase Cleaves ubiquitin from proteins and other molecules
Group of approx. 102 cysteine proteases and
metalloproteases
Immunoproteasome Degrades proteins into peptide ligands for Major
Histocompatibility Complex (MHC)
Proteasome with β1i, β2i, and β5i subunits
Thymoproteasome Degrades proteins into peptide ligands for MHC
1, selective for CD8+ T cells
Unique to thymic cortex
Trypsin-1 P07477 Degradation of food proteins in small intestine
Trypsin-2 P07478 Degradation of food proteins in small intestine
Mesotrypsin P35030 Degradation of antitrypsin inhibitors
Chymotrypsinogen B1 P17538 Degradation of food proteins in small intestine
Chymotrypsinogen B2 Q6GPI1 Degradation of food proteins in small intestine
Serine Protease Inhibitor Kazal- P00995 Trypsin inhibitor, in pancreas protects against self-activated
type 1 (SPINK1) trypsin
Inhibits calcium binding and NO production in sperm

proteasome defects may contribute to the pathogenesis of neuro-

degenerative diseases such as Parkinson’s, Huntington’s, Alzhei-
mer’s, and ALS. A decline in proteasome activity is also a hallmark
of aging cells.
The gastrointestinal tract contains the highest concentrations
of endogenous and exogenous proteases. The intestinal mucosa is
constantly exposed to low level protease activity, from bacteria in
the lumen, immune and mesenchymal cells in the basement mem-
brane, and epithelial cells at the brush border membrane. Protease
activity is tightly controlled, as the mucosal barrier is thin and
 Proteases: Pivot Points in Functional Proteomics 325

susceptible to proteolysis. The intestinal epithelium is at the inter-

face of digestive, absorptive, and secretory functions, and signaling
processes to the mucosal immune, vascular, and nervous systems.
Endogenous growth factors, cytokines and extracellular matrix
(ECM) proteins that modulate these functions, are digestive prote-
ase substrates [36]. The biochemistry of digestive protein proces-
sing has been studied for more than 80 years since the first
crystallographic studies were published, and the functions of pan-
creatic trypsin, chymotrypsin and elastase and their inhibitors are
well known. Pancreatic trypsinogen is activated to trypsin by
membrane-bound enterokinase in the small intestine. Trypsin acti-
vates pancreatic chymotrypsinogen, procarboxypeptidases, proelas-
tases, and prolipases. Turnover of intestinal epithelium is rapid,
requiring tight control of gut protease activity under normal physi-
ological conditions. Pancreatic PRSS3/mesotrypsin, discovered in
the late 1970s, is an atypical trypsin with an evolutionary mutation
that renders the protease resistant to inactivation by the physiologi-
cal Kazal inhibitor, pancreatic secretory trypsin inhibitor
(SPINK1), and endows it with specific digestive trypsin inhibitor-
degrading properties [37]. In pancreatitis, trypsin is activated in the
pancreas, causing tissue destruction and inflammation. Mesotryp-
sin is upregulated in certain cancers, and SPINK1 deficiency is
associated with hereditary pancreatitis.

2.4 Tissue Zinc proteases feature prominently in these biological processes.

Remodeling, Signaling, They can be subdivided according to the structure of their catalytic
Cell Migration, and sites and their domain organization [38]. The human ADAM
Proliferation family (a disintegrin and metalloprotease) currently counts 13 pro-
teolytically active transmembrane and secreted members. ADAMs
are largely tissue-specific and play roles in fertilization, prolifera-
tion, migration, and cell adhesion. Transmembrane ADAMs act as
sheddases, i.e., proteases that cause extracellular shedding of adja-
cent transmembrane proteins by proteolytic cleavage at the mem-
brane. Examples of such activated proteins are TNF-α and the ErbB
family of receptor tyrosine kinases; and EGF receptor ligands such
as TGF-α, heparin-binding EGF-like growth factor, betacellulin,
epiregulin, and amphiregulin. ADAM-mediated shedding is often
followed by RIPping, or Regulated Intramembrane Proteolysis, in
which the intracellular portion of these transmembrane proteins is
cleaved off by aspartyl proteases, S2P-metalloproteases, and rhom-
boid serine proteases. The released intracellular domain participates
in signaling to the nucleus to modify gene expression. Processing of
amyloid precursor protein (APP) and Notch signaling are typical
examples of RIPping. The 19 known human ADAM-TS proteases
have a similar architecture to the ADAMs, except for the presence
of thrombospondin repeats instead of a transmembrane domain,
which makes them extracellular. They process procollagen and von
Willebrand factor, and cleave extracellular matrix aggrecan,
326 Ingrid M. Verhamme et al.

versican, brevican, and neurocan. Matrix metalloproteinases

(MMPs) typically have three common domains: the N-terminal
propeptide that keeps the protease in an inactive form, the catalytic
domain containing the Zn2+ ion, and a C-terminal hemopexin-like
β-propeller domain for protein-protein interactions. MMPs are not
only instrumental in matrix remodeling and tissue maintenance,
but also as regulators of signaling pathways [39] (Table 5).
MMPs, originally thought to degrade extracellular matrix pro-
teins in a rather indiscriminate fashion, were later shown to have
specific physiological roles in shedding, activation, and inactivation
of proteins such as growth factors and cytokines. They cleave their
substrates using a HExxHxxGxxH motif which contains three zinc-
binding histidines and a glutamate that acts as a general base/acid
during catalysis. To date, there are 23 known human MMPs,
organized in four classes according to their substrate specificity:
collagenases (MMP-1, -8, and -13), gelatinases (MMP-2 and -9),
stromelysins (MMP-3, -10, and -11), and a group containing
matrilysin (MMP-7), metallo-elastase (MMP-12), enamelysin
(MMP-20), matrilysin-2 (MMP-26), and epilysin (MMP-28).
Pro-MMPs are activated by proteolytic removal of the N-terminal
prodomain that keeps the zymogen inactive by using a cysteine
switch to bind the catalytic zinc ion. MMPs participate in multiple
processes that involve tissue remodeling, e.g., embryo implantation,
wound healing, cell proliferation, bone ossification, and blood
vessel remodeling; signaling by all 54 human chemokines; and
innate immune defense [39]. MMP activity is regulated by endog-
enous, tight-binding tissue inhibitors of metalloproteinases
(TIMPs). Abnormal MMP expression and activity have been
observed in cancer, cardiac remodeling and aneurysm formation,
impaired wound healing, neurodegeneration, and after UV radia-
tion exposure of the aging skin and the cornea [40]. The family of
endogenous tissue-inhibitors of metalloproteinases (TIMPs) con-
sists of 4 proteins that target protease activity of MMPs, ADAMs,
and ADAM-TSs. They also affect cell growth and differentiation,
cell migration, anti-angiogenesis, anti- and pro-apoptosis, and syn-
aptic plasticity in biological processes different from protease
inhibition [41].
Protease signaling is a relatively new concept, and in contrast
with other types like receptor or kinase signaling the process is
irreversible [8]. The major immediate results of protease signaling
are target protein activation or inactivation, exposure of cryptic
sites, shedding of transmembrane proteins, and receptor agonist/
antagonist interconversion. These processes may initiate down-
stream signaling, resulting in a wide variety of physiological or
pathological responses. Selection of a physiological substrate is
facilitated by protease–substrate colocalization; substrate specificity,
as determined by the complementarity of the protease active site to
the reaction transition state; interactions immediately distal from
 Proteases: Pivot Points in Functional Proteomics 327

Table 5
Tissue remodeling, signaling, cell migration, and proliferation: activities and Uniprot ID codes (where
applicable) of proteases, protease inhibitors, and cofactors associated with tissue remodeling, signaling,
cell migration, and proliferation

Protease/inhibitor Uniprot Action

ADAM family (a disintegrin Play roles in fertilization, proliferation, migration and cell
and metalloprotease) adhesion
Not all are proteases, those which are act as sheddases
ADAM-TS (a disintegrin and Process procollagen and von Willebrand factor, and cleave
metalloproteinase with extracellular matrix aggrecan, versican, brevican, and
thrombospondin motifs neurocan
MMPs (Matrix All implicated in metastasis except 12, 20, and 28
metalloproteinases)
TIMPs (tissue-inhibitors of Endogenous inhibitors of MMPs
metalloproteinases)
MMP-1 P03956 Cleaves collagens I, II, III, VII, and X
Mediates neurotoxicity of HIV viral Tat protein
MMP-8 P22894 Degrades fibrillar collagens I, II, and III
MMP-13 P45452 Cleaves collagens I, II, III, IV, XIV, and X
Degrades fibrillar collagen, fibronectin, tenascin C, and
aggrecan
MMP-2 P08253 Degrades extracellular matrix proteins, including collagen I
and IV
MMP-9 P14780 Cleaves collagen IV and V and fibronectin
Implicated in neovascularization in malignant gliomas
MMP-3 P08254 Degrades fibronectin, laminin, gelatins of type I, III, IV, and
V; collagens III, IV, X, and IX, and cartilage proteoglycans
Activates MMPs 1, 7, and 9
MMP-10 P09238 Degrades fibronectin, gelatins of type I, III, IV, and V,
collagens III, IV, and V
Activates procollagenase
MMP-11 P24347 Cleaves alpha 1-proteinase inhibitor, activated intracellularly
by furin
MMP-7 P09237 Degrades casein, gelatins of types I, III, IV, and V, and
fibronectin
Activates procollagenase
Activates MMP-2 and MMP-9
MMP-12 P39900 Cleaves elastin, implicated in aneurysm formation
MMP-20 O60882 Degrades amelogenin, aggrecan, and cartilage oligomeric
matrix protein (COMP)
MMP-26 Q9NRE1 Degrades collagen type IV, fibronectin, fibrinogen, beta-
casein, type I gelatin and alpha-1 proteinase inhibitor
Activates progelatinase B
MMP-28 Q9H239 Degrades casein
328 Ingrid M. Verhamme et al.

the substrate-binding pocket; and interactions with protease exo-

sites remote from the active site.
Protease-activated receptors (PARs) are prototypical examples
of protease signaling. These four G protein-coupled receptors are
activated irreversibly by extracellular proteases, by cleavage of the
N-terminal ectodomain and exposure of a tethered peptide ligand.
Transmembrane signaling is initiated by binding of this tethered
peptide to the body of the receptor [42]. PAR1, PAR3, and PAR4
are activated by thrombin, and signaling occurs during tissue injury,
hemostasis, and inflammation. Signaling is regulated by rapid inter-
nalization of spent receptors. PAR1 and PAR4 cleavage on platelets
causes robust platelet activation. Thrombin has a higher affinity for
PAR1, and the PAR1 antagonist vorapaxar was approved in 2014 as
antiplatelet drug. However, major bleeding side effects prompted
the development of PAR4 antagonists which are currently in clinical
testing. PAR4 signaling promotes vascular disease and cardiac post-
infarction remodeling, and these antagonists are promising candi-
dates for safer antithrombotic and anti-inflammatory therapy
[43]. PAR1 on endothelial cells is productively cleaved by activated
protein C (APC) in the presence of the endothelial protein C
receptor. This triggers expression of monocyte chemoattractant
protein-1, acting as a protective component during sepsis. PAR2
is activated by trypsin, tryptase, the coagulation factors VIIa and
Xa, and matriptase. PAR2 signaling is thought to regulate epithelial
growth and function. Thrombin-mediated PAR activation has been
implicated in vascular smooth muscle cell migration and prolifera-
tion as causative processes in restenosis after stent placement [44],
and in tumor metastasis, where a simultaneous requirement for
PARs and fibrinogen was found [45]. Thrombosis and cancer
have since long been recognized as interconnected pathologies,
and in this light argatroban, the tight-binding competitive inhibitor
of active thrombin, has been re-evaluated as a clinically useful
antiproliferative and antimetastatic agent [46].

2.5 Programmed For obvious reasons, the process of apoptosis must be both highly
Cell Death regulated and a model of organized efficiency once initiated. It is
governed by a cascade of caspases which, though they sometimes
have other functions, thoroughly dismantle the innards of a cell so
that it may be phagocytosed by immune cells without releasing
cytosolic components into the extracellular space. Caspases are
cysteine aspartate proteases involved in cell death, cellular remodel-
ing, stem cell fate determination, spermatogenesis, and red blood
cell differentiation. Their sets of substrates with regard to apoptosis
are well defined, and cooperative cleavage of these substrate sets
triggers apoptosis. With regard to apoptosis, their functions fall
into one of two categories: Initiator (caspases 2, 8, 9, and 10) or
Executioner (caspases 3, 6, and 7). Caspases 1, 4, 5, and 12L are
considered inflammatory [47]. Activation of the Initiator caspases
 Proteases: Pivot Points in Functional Proteomics 329

can be induced intrinsically, by release of cytochrome c into the

cytosol by mitochondria, or extrinsically by ligand binding to
Death Receptors. When cytochrome c is released, it binds to adap-
tor protein APAF-1, inducing it to form the apoptosome oligomer,
which then binds to the caspase activation domain (CARD) of
procaspase 9, inducing oligomerization of the procaspase. This
induces autoproteolysis of caspase 9 to activate it [48, 49]. This
process can be suppressed by the presence of the β transcription
variant of caspase 9 which lacks a catalytic domain [50, 51]. Caspase
9 then activates procaspases 3 and 7 by cleaving them at a L-G-H-
D-(cut)-X sequence [52, 53]. Caspase 9 can be downregulated by
phosphorylation, or inhibited by proteins in the Inhibitor of Apo-
ptosis (IAP) family [54]. Activated Caspase 3 is known to inhibit
the function of IAPs [55], ensuring that once the cascade is
initiated it progresses rapidly, and to activate caspases 6, 7, and
9, which further accelerates the process. Extrinsic activation is
initiated by binding of Death Factors, such as FasL, to Death
Receptors, such as FasR. The conformation change in the recep-
tor’s cytoplasmic Death Domain (DD) induces a change in the
bound adaptor protein FADD which recruits procaspase 8 to
bind at its Death Effector Domain (DED, proving gallows humor
is irresistible even at the cellular level), activating caspase
8 [56]. Caspase 8 activates caspases 3, 4, 6, 7, 9, and 10. Caspase
10 then activates Caspases 3, 4, 6, 7, 8, and 9. Caspase 6 has a
limited capacity to autoproteolyze and activate itself [57], and is
known to target both Huntingtin and Amyloid Precursor Protein
(APP), linking it to neurodegenerative diseases [58]. In apoptosis it
is responsible for disinhibition of the immune system by cleaving
interleukin-10 and interleukin-1 receptor-associated kinase
3 (IRAK3) [59]. The Executioner Caspases, as a group, are respon-
sible for cleavage of over 600 other proteins [60]. Caspase 3 can
also be activated by Granzyme B, allowing T lymphocytes and
natural killer cells to initiate apoptosis in target cells. The substrate
landscape in non-apoptotic events may be much broader [61], as
suggested by recent global proteomics studies. The “forward”
approach involves triggering endogenous caspases to identify native
substrates in intact cells, whereas in the “reverse” approach exoge-
nous caspases are added to cell lysates. Isolated cleavage products
are digested and identified by tandem mass spectrometry. The
forward method allows identification of substrates in intact cells,
rather than which caspase performs the cleavage. In the reverse
method, specific caspases are tested for their ability of cleaving
substrates, but in a cell lysate with destroyed organelles, endoge-
nous proteases may contribute to substrate cleavage, requiring the
need for strict controls. Moreover, proteolysis in organelle mem-
branes may be missed due to removal of insoluble material before
analysis. A combination of current methods has yielded several
hundred potential native substrates for caspases, and measuring
330 Ingrid M. Verhamme et al.

rates and extent of substrate cleavage allows distinguishing func-

tional from bystander targets [62]. Eight human endogenous inhi-
bitors of apoptosis (IAPs) have been identified, and their inhibitory
activity is neutralized by the mitochondrial protein Smac/DIA-
BLO. Development of Smac/DIABLO-like peptidomimetics has
been proposed as a potential therapeutic approach in cancer treat-
ment. Because of the roles of caspases in inflammation, caspase
inhibitors may also prove beneficial in treating sepsis [63]. In addi-
tion to caspases, other proteases are also associated with apoptosis,
e.g., calpains, cathepsins, granzymes, and the proteasome. These are
regulated by their respective endogenous inhibitors: calpastatins,
cystatins, the serpin PI-9, and various macromolecular proteins
(Table 6).

2.6 Protein Secretion Many current cardiovascular biomarkers are secreted proteins, gen-
erated by cleavage of their pro-proteins at the endoplasmic reticu-
lum (ER). Upon release of the mature protein, the signal peptide is
proteolytically separated from the ER by signal peptidase, an intra-
membrane aspartic protease. The long-held belief that signal pep-
tides are invariably recycled or degraded by ubiquitin-proteasome
related factors has been challenged during recent years, and several
were shown to remain intact after cleavage in the ER [64]. These
signal peptides have biological functions of their own, and play
roles in regulation of immunity, trafficking, and other processes.
In type I diabetes, a signal peptide fragment from pre-proinsulin,
presented at the surface of the pancreatic β-cells, acts as antigen and
flags the cell for destruction by cytotoxic T cells. Identifying agents
to control β-cell destruction may be a new therapeutic strategy.
Similarly, a signal peptide fragment from pre-procalcitonin, highly
abundant in several lung cancers and medullary thyroid cancers, is
an epitope for T-killer cells. This knowledge may aid in the devel-
opment of treatments of these cancers. Rapidly increasing plasma
levels of N-terminal adducted signal peptide fragments from A-, B-,
and C-type natriuretic peptides are characteristic of ST-elevated
myocardial infarction. The nature of the N-terminal adducts may
be useful in assay design and disease assessment, and development
of fast biomarker assays for these signal peptide fragments may
ultimately be beneficial in clinical decision-making (Table 7).

2.7 DNA Replication, DNA damage hinders replication, and may lead to strand breaks,
Repair, and Processing genomic instability, aging, and cancer [65]. DNA-topoisomerase
1 crosslinks (DPC) are bulky lesions that trap otherwise transient
covalent DNA-protein intermediates, and inhibit movement of
polymerases and helicase, causing stalling of the replication fork.
In yeast models, the protease Wss1 was identified as effector of
DPC repair. BLAST searches revealed a conserved family of DPC
proteases, with Spartan being the human member of this class
[66]. Spartan was recently characterized as a DNA replication-
 Proteases: Pivot Points in Functional Proteomics 331

Table 6
Programmed cell death: activities and Uniprot ID codes (where applicable) of proteases, protease
inhibitors, and cofactors associated with programmed cell death

Protease/inhibitor Uniprot Action

Caspase-2 P42575 Function uncertain, sequence homology with initiator
caspases
Caspase-8 Q14790 Activates caspases 3, 4, 6, 7, 9, and 10
Activated by death receptors via FADD
Apoptotic protease activating O14727 Forms apoptosome complex
factor 1 (APAF-1) Activated by binding of cytochrome c and ATP
Caspase-9 P55211 Activates caspase 3
Activated by the apoptosome complex
Cleaves poly(ADP-ribose) polymerase (PARP)
Implicated in activation of Abelson murine leukemia viral
oncogene homolog 1 (ABL1)
Caspase-10 Q92851 Activates caspases 3, 4, 6, 7, 8, and 9
Activated by caspase 8
Caspase-3 P42574 Activates caspases 6, 7, and 9
Activated by caspases 8 and 9
Cleaves poly(ADP-ribose) polymerase (PARP)
Cleaves and activates sterol regulatory element-binding
proteins (SREBPs)
Implicated in Huntington’s disease
Caspase-6 P55212 Dis-inhibits immune system, cleaves interleukin-10 and
interleukin-1 receptor-associated kinase 3
Cleaves poly(ADP-ribose) polymerase (PARP) and lamins
Implicated in Huntington’s and Alzheimer’s
Caspase-7 P55210 Degradation of cellular proteins in apoptosis
Cleaves poly(ADP-ribose) polymerase (PARP)
Cleaves and activates sterol regulatory element-binding
proteins (SREBPs)
Apoptosome Heptameric complex of APAF-1, activates caspase 9
FasL P48023 Tumor necrosis factor ligand, activates death receptors to
initiate apoptosis
FasR P25445 Death receptor, tumor necrosis factor receptor, activates
caspase 8

coupled metalloprotease for DPC repair and restoration of geno-

mic stability [67, 68]. Mutations have been associated with prema-
ture aging and early onset hepatocellular carcinoma, suggesting
Spartan as a tumor suppressor, and DPC repair as a protective
antitumor mechanism. Double-strand DNA breaks are repaired
by the DNA damage response, in pathways that are tightly con-
trolled by ubiquitinylation and deubiquitinylation events. The
332 Ingrid M. Verhamme et al.

Table 7
Protein secretion: activities and Uniprot ID codes (where applicable) of proteases, protease inhibitors,
and cofactors associated with protein secretion

Protease/inhibitor Uniprot Action

Signal Peptidase Removes amino terminal signal sequences from secretory
pro-proteins
Group of aspartic proteases
Insulin P01308 increases cell permeability to monosaccharides, amino acids, and
fatty acids
accelerates glycolysis, the pentose phosphate cycle, and glycogen
synthesis in liver
Calcitonin P01258 Promotes rapid incorporation of calcium and phosphate into
bone
Atrial natriuretic peptide P01160 Peptide hormone
(ANP) regulates of natriuresis, diuresis, and vasodilation
promotes trophoblast invasion and spiral artery remodeling in
uterus during pregnancy
Binds and stimulates the cGMP production of the NPR1 receptor
Binds the clearance receptor NPR3
Brain natriuretic peptide P16860 Peptide hormone
(BNP) Regulates natriuresis, diuresis, vasorelaxation, and inhibition of
renin and aldosterone secretion
Binds and stimulates the cGMP production of the NPR1 receptor
Binds the clearance receptor NPR3
Natriuretic peptide P23582 Peptide hormone
precursor C (NPPC) Regulates of cartilaginous growth plate chondrocytes
proliferation and differentiation
Binds and stimulates the cGMP production of the NPR2 receptor

latter are catalyzed by deubiquitinases (DUBs) that can positively or

negatively affect the damage response. The proteasomal deubiqui-
tinating enzyme POH1 promotes double-strand DNA break repair
[69]. Various DUBs are also associated with transcriptional and
epigenetic control of gene expression, DNA damage repair path-
ways, and cell cycle checkpoint control, often deregulated in tumor
cells [70]. These DUBs may be potential targets for therapeutic
inhibition, and they are currently the subject of small molecule
screening. DNA processing is also under indirect control of pro-
teases. DNA fragmentation and chromatin condensation are final
processes in apoptosis. The DNase CAD, catalyzing these reactions,
is normally under tight control of its ligand ICAD that acts as a
chaperone and inhibitor when bound to CAD. This prevents spon-
taneous activation of CAD in non-apoptotic cells. ICAD cleavage
by caspase-3 during the apoptotic execution phase liberates active
CAD that enters the nucleus to fragment DNA and catalyze chro-
matin condensation [71]. Human Lon protease binds to
 Proteases: Pivot Points in Functional Proteomics 333

mitochondrial single-stranded DNA sequences with a propensity

for forming G-quadruplexes [72]. The precise role of human Lon
protease is not yet clear, but it has been suggested that DNA-bound
Lon may process proteins involved in mitochondrial DNA and
RNA metabolism.

2.8 Intranuclear, In addition to Spartan and the proteasome involved in DNA pro-
Cytosolic, cessing, other proteases are known to have intranuclear functions.
Transmembrane, and Interleukin-1β-converting enzyme (ICE), also known as caspase-1,
Intramembrane plays a role in the inflammatory immune response. The N-terminal
Proteolysis prodomain of its precursor, procaspase-1, possesses a nuclear local-
ization signal. Tumor necrosis factor induces translocation of
procaspase-1 from the cytosol to the nucleus where it is activated
by proteolytic removal of the intact prodomain [73]. Cell transfec-
tion studies showed that the prodomain alone is capable of trigger-
ing apoptosis, which suggests separate nuclear roles for the
prodomain and active caspase-1. Caspase-3 also effects nuclear
changes in apoptotic cells. Its inactive precursor is cleaved in the
cytoplasm by initiator caspases in response to death signals, and also
by cytosolic MMP-2 and -9. The active caspase-3 dimerizes and
translocates to the nucleus by an active transport system. Caspase-7
is only found in the cytoplasm, suggesting that translocation is
caspase-3 specific and not the result of simple diffusion after the
nuclear-cytoplasmic barrier is disrupted [74]. The extracellular
properties of MMPs as tissue remodeling proteases are well docu-
mented but less is known about their intracellular functions. Vari-
ous MMPs have been detected in the intranuclear space, and they
are mostly associated with pathological processes [75, 76].
Oxygen-glucose deprivation in ischemic stroke induces an intrinsic,
caspase-independent apoptotic pathway in neurons, characterized
by elevated intranuclear MMP-2 and -9 activity that targets nuclear
DNA repair proteins. Similarly, MMP-2 in the nuclei of stressed
cardiac myocytes can induce apoptosis. Intranuclear MMP-3 in
osteoarthritis and certain cancers upregulates connective tissue
growth factor, a mediator of cell migration, proliferation, and
pathological fibrosis. During viral infections, macrophage-secreted
MMP-12 translocates to the nucleus of infected cells, and enhances
transcription of IκBα, a protein that promotes antiviral interferon-α
(IFN-α) secretion. Extracellular MMP-12 degrades excess secreted
IFN-α, thus limiting its systemic toxicity. It is becoming increas-
ingly clear that the same protease may exert significantly divergent
functions, depending on its microenvironment, substrates, and
effectors. Understanding these different mechanisms of interaction
on a molecular level is ultimately the key to successful design of
therapeutics (Table 8).
Promotion of angiogenesis is regulated by binding of single
chain urokinase-type plasminogen activator (scuPA) to its receptor
on the endothelial cell surface, and subsequent transport of the
334 Ingrid M. Verhamme et al.

Table 8
Intranuclear, transmembrane, intramembrane, and cytosolic proteolysis: activities and Uniprot ID
codes (where applicable) of proteases, protease inhibitors, and cofactors associated with intranuclear,
transmembrane, intramembrane, and cytosolic proteolysis

Protease/inhibitor Uniprot Action

Matriptase (Suppressor of Q9Y5Y6 Degrades extracellular matrix, trypsin-like activity
tumorigenicity 14 protein) Promotes epithelial differentiation and possibly growth
Implicated in metastasis
Matriptase 2 Q8IU80 Cleaves collagen I, fibronectin, and fibrinogen
Involved in matrix remodeling processes in liver
Regulates the expression of the iron absorption-
regulating hormone hepcidin/HAMP
Prostasin Q16651 Stimulates epithelial sodium channel (ENaC) activity
through activating cleavage of the gamma subunits
(SCNN1G)
Also found in seminal fluid
hepatocyte growth factor (HGF) O43291 Inhibits HGF, possibly inhibits serine proteases
activator inhibitor type2 generally
(SPINT2) Implicated in suppression of liver cancer
hepatocyte growth factor activator O43278 Inhibits HGF and matriptase
inhibitor type1 (SPINT1)

protease to the nucleus [77]. scuPA de-represses transcription of

the VEGF receptor 1 (VEGFR1) and 2 (VEGFR2) genes by inter-
fering with the proline-rich homeodomain protein that represses
the activity of vegfr1 and vegfr2 gene promoters. The VEGF
growth factors are known targets for control of pathologic angio-
genesis in macular degeneration, and the discovery of the scuPA-
mediated pathway may offer additional avenues for therapeutic
intervention.
Transmembrane proteases may be anchored to the membrane
by a C-terminal domain (Type I), an N-terminal domain with
cytoplasmic extension (Type II), or by glycosylphosphatidylinositol
(GPI) [78]. Their catalytic domains are extracellular. Among the
zinc-dependent proteases are MMP-14, 15, -16, -24, ADAM-10
and -17, meprins α and β (Type I), MMP-23 (Type II) and
MMP-17 and -25 (GPI). The Type I zinc proteases act as sheddases
upon proteolytic removal of their N-terminal propeptide. The
majority of transmembrane serine proteases are classified as Type
II subfamilies: hepsin/TMPRSS (transmembrane protease/serine),
matriptase, corin, and HAT/DESC (human airway trypsin-like
protease/differentially expressed in squamous cell carcinoma).
Corin in cardiomyocytes activates the atrial natriuretic factor
(ANF), a cardiac hormone that regulates blood pressure and cardiac
function by promoting natriuresis, diuresis, and vasodilation.
 Proteases: Pivot Points in Functional Proteomics 335

Tryptase γ1 is the only known Type I serine protease, and prostasin

and testisin are GPI-anchored. Prostasin plays a role in epithelial
sodium channel regulation, and testisin regulates germ cell matura-
tion. All these proteases are involved in physiological development,
but also in pathological processes of inflammation and cancer. They
activate peptide hormones, growth and differentiation factors,
receptors, enzymes, adhesion molecules, and viral coat proteins.
Matriptases 1 and 2, and prostasin are expressed in human epithe-
lium, and inhibited by their cognate Kunitz-type inhibitors,
membrane-anchored hepatocyte growth factor activator inhibitors
(HAI) 1 and 2 [79]. Matriptase overexpression elicits signaling via
the PAR-2, and promotes fibroblast activation, proliferation, and
migration in idiopathic pulmonary fibrosis [80]. Matriptase pro-
teolytically activates hepatocyte growth factor (HGF) that binds to
its receptor c-Met, a receptor tyrosine kinase. This activates critical
signaling pathways in organ development. Abnormal c-Met signal-
ing is associated with cell proliferation, migration and invasion, and
progression of lung, breast, ovary, kidney, colon, thyroid, liver, and
gastric carcinomas. There is also some matriptase crosstalk with the
hemostatic system with regard to epithelial defense and repair after
injury and infection. Exposure of membrane-anchored tissue factor
(TF) in damaged vascular endothelium to factor VIIa triggers the
extrinsic coagulation pathway and formation of factor Xa. The TF:
fVIIa complex and factor Xa activate epithelial pro-matriptase to
matriptase which cleaves the PAR2 receptor. This induces enhance-
ment of the epithelial barrier function [81].
Intramembrane proteases (IMPs), a fairly recently discovered
class, are embedded in lipid bilayers and their catalytic site is formed
by residues in different transmembrane helices [82]. The four IMP
families are metallo-, serine, aspartate, and glutamate proteases, and
they are found in the Golgi apparatus, endosomes and lysosomes,
the plasma membrane, endoplasmic reticulum, and the inner mito-
chondrial membrane. IMPs cleave their substrates with a fairly high
specificity, given the fact that of the ~2500 identified single-pass
transmembrane proteins only a limited number are identified as
IMP substrates. The function of IMPs is diverse, ranging from
transcription factor signaling, mitochondrial remodeling, and pro-
tein maturation to regulation of immunity, and quorum sensing
and parasite-host interactions in pathogens. Many IMP defects are
associated with pathogenesis.
Site-2 protease (S2P) is the only member of the metalloprotei-
nase IMP group, and mutations cause ichthyosis follicularis and
osteogenesis imperfecta. The HIV inhibitor Nelfinavir was found
to inhibit S2P in castration-resistant prostate cancer cell lines;
however, due to the multitude of other Nelfinavir targets and the
less than impressive effects on PC-3 cancer cells this drug may not
be specific enough for use in prostate cancer. Five human rhomboid
serine IMPs are known, but no function or substrates have been
336 Ingrid M. Verhamme et al.

identified for rhomboid 1 and 3. Epidermal growth factor and

thrombomodulin are rhomboid 2 substrates, and the protease is
thought to control cell migration and proliferation [83]. Lowered
expression impairs wound healing, and overexpression may be
linked to tumor metastasis. Rhomboid 4 upregulation is associated
with poor outcome in colorectal cancer, but there is no clear
consensus on the molecular mechanism of this process. Rhomboid
4 cleaves amyloid precursor protein (APP) within its ectodomain
and reduces formation of Aβ38, 40, and 42 peptides. This pathway
may be an alternative to the pathological processing of APP by
γ-secretase in Alzheimer’s disease. The role of the mitochondrial
rhomboid protease PARL in Parkinson’s disease is controversial:
some studies attribute a protective function to PARL in inducing
removal of defective mitochondria through autophagy, whereas
others suggest that PARL knockdown is responsible for this pro-
cess. A potential link of low level PARL activity with type 2 diabetes
was first seen in obese sand rats with diet-induced diabetes. Normal
PARL levels and insulin sensitivity were restored when the rats were
put on an exercise regimen. Similarly, in humans with type 2 diabe-
tes, PARL mRNA and mitochondrial DNA are reduced in skeletal
muscle. PARL was recently identified as a pro-apoptotic protease
because it cleaves mitochondrial Smac/DIABLO [84]. The pro-
cessed protein is released into the cytosol and binds an apoptosis
inhibitor, thereby triggering the caspase cascade. The most studied
member of the aspartate IMPs is the γ-secretase complex, with
presenilin as the intramembrane catalytic subunit. Presenilin and
APP mutations are linked to familial Alzheimer’s disease (AD), and
γ-secretase was considered as an anti-AD drug target. However,
severe side effects limit the use of γ-secretase inhibitors. Processing
of Notch by γ-secretase releases the intracellular Notch domain for
intranuclear modulation of gene expression. Alterations of this
pathway are linked to several types of cancer [85], and γ-secretase
inhibitors are now in clinical trials as potential anticancer drugs.
The multifunctionality of these proteases illustrates the caveats in
developing drugs that indiscriminately target protease activity with-
out taking specific molecular mechanisms and the protease micro-
environment into account.

3 Proteases and Disease

3.1 Epigenetics and Many inheritable diseases are directly related to DNA modification;
Disease however, epigenetic processes are equally prominent in the disease
state, and they are essential contributors to normal physiological
development. Environmental factors, diet, aging, and diseases such
as cancer may contribute to positive or negative changes in gene
expression that are passed onto daughter cells: DNA (hydroxyl)
methylation, covalent histone modification and chromatin
 Proteases: Pivot Points in Functional Proteomics 337

remodeling, arrangement of the chromatin-histone nucleosomes

along the DNA sequence, gene-activating transcription factor
activity of new gene products, and downregulation of messenger
RNA by noncoding microRNA. Formation of prion structures in
transmissible spongiform encephalopathies is also considered an
epigenetic phenomenon.
Proteases play regulatory roles in epigenetic mechanisms of
altered gene expression. Various nonspecific and N-tail-specific
histone proteases are thought to assist in fertilization, histone
turnover, gene de-repression, histone removal during spermato-
genesis, and reversal of N-tail methylation [86]. In N-tail clipped
histone H3, lysines for acetylation have been removed, which may
result in transcriptionally inactive chromatin. This process may be
in part responsible for age-related declining gene expression.
Cathepsin L, susceptible to stefin B inhibition, was identified as
the histone H3-cleaving protease during stem cell differentiation in
mice; however, the corresponding H3-clipping protease in human
embryonic stem cells is refractory to specific cathepsin L inhibitors
and remains to be identified [87].

3.2 Inflammation as A comprehensive, recent overview of protease activity in inflamma-

an Over-Arching tion is given by Deraison et al. [88]. The host inflammatory
Symptom of Disease response is accompanied by release of proteases from neutrophil
granulocytes, macrophages, and mast cells. These proteases form a
first line of defense in bacterial infections but, if left uncontrolled,
are also damaging to host tissues. Human neutrophil elastase,
present in six isoforms, and the related cathepsin G and
proteinase-3 are serine proteases localizing to neutrophil extracel-
lular traps (NETs) in a defense mechanism against bacterial inva-
sion. They cleave collagen-IV and elastin, and excessive secretion
may cause idiopathic pulmonary fibrosis, rheumatoid arthritis, and
adult respiratory distress syndrome. Alpha1-proteinase inhibitor
(α1-PI) and leukocyte elastase inhibitor are their major endoge-
nous irreversible serine proteinase inhibitors (serpins). Historically
α1-PI was identified as a trypsin inhibitor and later found to be
more specific for elastase inactivation. Smoking-induced oxidative
inactivation or mutation-induced misfolding of α1-PI causes
emphysema and cystic fibrosis, and may be treated with α1-PI
augmentation therapy [89], as gene therapy for α1-PI deficiency
is still in the Clinical Phase II stage. Increased risk for lung cancer
has also been observed in α1-PI-deficient patients who never
smoked [90]. Proteinase-3 generates antimicrobial peptides by
cleaving cathelicidin in neutrophils, but it is also predominant in
anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculi-
tis, a severe multisystem autoimmune disease with poor prognosis
[91] (Table 9).
Tryptase (6 isoforms), chymase, granzyme B, and
carboxypeptidase A, released by mast cells, degrade extracellular
338 Ingrid M. Verhamme et al.

Table 9
Inflammation as an over-arching symptom of disease: activities and Uniprot ID codes (where
applicable) of proteases, protease inhibitors, and cofactors associated with inflammation, specifically
as related to symptoms of disease

Protease/inhibitor Uniprot Action

Neutrophil Elastase P08246 Broad substrate specificity, chymotrypsin family
Inflammatory response triggers bacterial and host tissue
destruction
Inhibits C5a-dependent neutrophil enzyme release and
chemotaxis
Proteolysis of collagen-IV and elastin of the extracellular
matrix
Degrades pathogenic outer membrane proteins and
virulence factors
Cathepsin G P08311 Degrades ingested host pathogens
Breaks down ECM components at inflammatory sites
Cleaves complement component C3
Inhibited by Rv3364c (M. tuberculosis protein)
Indirect suppression of macrophage apoptosis
Converts angiotensin I to angiotensin II
Proteinase-3 (Myeloblastin) P24158 Degrades elastin, fibronectin, and collagen (in vitro)
Target antigen for anti-neutrophil cytoplasmic antibodies
(ANCA)
Alpha1-proteinase inhibitor P01009 Inhibits elastase, plasmin and thrombin
(a1-PI, Alpha-1-antitrypsin) Irreversibly inhibits trypsin, chymotrypsin, and
plasminogen activator
Possible non-protease inhibitor activities:
Anti- and pro-inflammation
Anti-apoptosis
Leukocyte Elastase Inhibitor P30740 Inhibits neutrophil elastase, cathepsin G, proteinase-3,
chymase, chymotrypsin, and kallikrein-3
Potent intracellular inhibitor of granzyme H
Tryptase: Major neutral protease present in mast cells
Resistant to endogenous proteinase inhibitors
Active only as heparin-stabilized tetramers
Six Isoforms: Tryptase alpha/beta-1(Q15661)
Tryptase beta-2(P20231), Tryptase delta(Q9BZJ3)
Tryptase gamma(Q9NRR2), Brain-specific serine protease
4(Q9GZN4)
Chymase (CMA1) P23946 Major secreted protease of mast cells
Release may promote inflammatory response
Converts angiotensin I to angiotensin II
Granzyme B (GZMB) P10144 Unique to cytolytic T-lymphocytes and natural killer cells
Activates caspases 3, 7, 9, and 10
Cleaves/activates BH3 interacting-domain death agonist
(BID)
Cleaves/activates Inhibitor of caspase-activated DNase
(ICAD)
Generate a cytotoxic level of mitochondrial reactive
oxygen species
(continued)
 Proteases: Pivot Points in Functional Proteomics 339

Table 9
(continued)

Protease/inhibitor Uniprot Action

Carboxypeptidase A3 (CPA3) P15088 Cleaves C-terminal aromatic or aliphatic residue
Unique to mast cells
Upregulated in sepsis and anaphylaxis
Implicated in autoimmune diseases
Alpha 1-antichymotrypsin P01011 Inhibits neutrophil cathepsin G and mast cell chymase
(SERPINA3)
Cathepsin L1 P07711 Intracellular degradation and protein turnover
Degrades collagen and elastin
Degrades alpha-1 proteinase inhibitor
Cathepsin B P07858 Intracellular degradation and protein turnover
Upregulation of cathepsin D, matrix metalloproteinase,
and urokinase
Implicated in metastasis and immune resistance
Cathepsin D P07339 Intracellular degradation and protein turnover
Used by macrophages to degrade bacterial proteins
Activates ADAM30, implicated in Alzheimer’s progression
Implicated in metastasis in breast cancer
Trypsin-3 (PRSS3) (meso- P35030 Degrades trypsin inhibitors
trypsinogen)
Serine Protease Inhibitor Kazal- P00995 Trypsin inhibitor, specifically inhibits autoactivated trypsin
type 1 (SPINK1) in the pancreas
Inhibits calcium binding and NO production in sperm
Caspase-3 P42574 Activates caspases 6, 7, and 9
Activated by caspases 8 and 9
Cleaves poly(ADP-ribose) polymerase (PARP)
Cleaves and activates sterol regulatory element-binding
proteins (SREBPs)
Implicated in Huntington’s disease

matrix components. An imbalance between tryptase and its endog-

enous inhibitors is characteristic in rheumatoid arthritis
[92]. Experimental tryptase inhibition in an in vivo model alle-
viated some, but not all, symptoms, suggesting the necessity for a
multidrug approach. Cell surface proteoglycan-bound chymase is
partially protected from endogenous inactivation by α1-PI and α1-
antichymotrypsin, and inhibitory caging by α2-macroglobulin.
Mast cell infiltration of atherosclerotic plaque aggravates the local
inflammatory status by causing smooth muscle cell (SMC) apopto-
sis in a chymase-dependent process: fibronectin cleavage by chy-
mase unmasks pro-apoptotic epitopes and disrupts the p-FAK-
dependent cell-survival signaling cascade, leading to SMC cell
death [93]. Mast cell tryptase and chymase activity is also associated
340 Ingrid M. Verhamme et al.

with several other pathological processes in atherosclerosis, abdom-

inal aortic aneurysm (AAA) formation and metabolic disease, as
discussed below [94].
Macrophages release matrix metalloproteinases (MMPs), cys-
teine proteinases (caspases and cathepsin L), and cathepsin D, an
aspartyl protease also present in lysosomes. Their proteolytic power
is a part of the diverse arsenal of mechanisms macrophages utilize in
vessel wall-localized pro-inflammatory processes associated with
the development of vasculitis, a fast-developing pathology, and
atherosclerosis, a disease developing over decades [95].
Cathepsin B, a lysosomal cysteine proteinase, preferentially
activates mesotrypsinogen in pancreatic acinar cells. Increased
mesotrypsin (trypsin-3) activity as a result from overexpressed
cathepsin B lowers protective SPINK1 levels and initiates apoptosis
by activating caspase-3 [96]. Both processes contribute to the
development of human pancreatitis. Crystallization studies identi-
fied diminazene analogues as small molecule inhibitors of meso-
trypsin, and these structures may form the basis for developing
selective, tight-binding drugs [97]. In irritable bowel syndrome
(IBS) the intestinal epithelium overproduces mesotrypsin, which
increases intestinal epithelium permeability, signals to human sub-
mucosal enteric neurons and induces visceral hypersensitivity by a
protease-activated receptor-2-dependent mechanism [98]. Meso-
trypsin may be a suitable biomarker for IBS as well as a target for
novel, specific drugs.
Inflammation and complement activation are interconnected,
and the inflammatory environment in degenerative diseases, cancer,
transplant rejection, and exposure to chronic external stimuli is
characterized by excessive activation or insufficient control of com-
plement activation [99]. The physiological response of comple-
ment involves self-recognition of normal cells; immune
recognition and clearance of diseased cells, apoptotic cell debris,
and immune complexes; elimination and danger signaling of patho-
gens; and tolerance to transplants and biomaterials. Excessive com-
plement activation triggers inflammatory reactions seen in danger
signaling and attack of “self” cells resulting in autoimmune disease,
pathogen infection, and tissue/biomaterial rejection. The proteases
and other components of the complement system may be attractive
candidates for novel anti-inflammatory drug intervention; how-
ever, the complexity and extensive cross talk of this network poses
significant challenges in the development of specific inhibitors
devoid of off-target side effects.

3.3 Cardiovascular Due to the interconnectedness of cardiovascular and metabolic

and Metabolic disease, a new term, cardiometabolic disease has been coined, and
Diseases, and Stroke recent studies underscore the critical roles of MMPs, calpains,
cathepsins, and caspases in disease progression [100]. Physiological
hemostasis is a delicate balance between cascading serine protease
 Proteases: Pivot Points in Functional Proteomics 341

zymogen activation, positive and negative feedback mechanisms,

protease inhibition, and thrombus dissolution. Many clotting and
bleeding disorders are the result of gene defects affecting the
expression or function of protease zymogens, or the substrates,
cofactors, and inhibitors of the active proteases. In the extrinsic
pathway, deficiency of factors VII, IX (hemophilia B), and X causes
bleeding. The G20210A mutation in the 3’ untranslated region of
the prothrombin gene stabilizes the precursor mRNA which results
in increased prothrombin levels and venous thrombosis risk. Up to
8% of the Caucasian population is heterozygous for this mutation.
In the contact activation pathway, deficiency of factor XI causes
mild to moderate bleeding, whereas factor XII and prekallikrein
deficiencies are generally asymptomatic. This is consistent with a
fairly recently discovered role for thrombin, generated through the
extrinsic pathway, as a factor XI activator in the absence of factor
XII [16]. Pathogen infections may also trigger hypercoagulability,
as seen in coagulase-positive Staphylococcus aureus infections. Sta-
phylocoagulase is not a protease but activates host prothrombin
conformationally, resulting in deposition of fibrin/bacterial vegeta-
tions that can embolize to the lungs, brain, and other parts of the
body [101] (Table 10).
The proteases of the contact pathway play a role in inflamma-
tory and immune processes as well as in maintaining hemostasis
[102]. Kallikrein cleaves high molecular weight kininogen to the
pro-inflammatory bradykinin. This nonapeptide is a vasodilator,
increases vascular permeability and contributes to inflammatory
pain by binding to bradykinin receptors. Elevated bradykinin levels
are seen in rheumatoid arthritis and IBS. The serpin, C1-inhibitor is
the main physiological inhibitor of factor XIIa and kallikrein. An
overactive contact pathway can be the result of decreased or dys-
functional C1-inhibitor, or a factor XII mutation leading to a more
active fXIIa form. The pathological manifestation is hereditary
angioedema, with sometimes life-threatening swelling in the
upper respiratory tract or the intestinal mucosa. Abnormal activa-
tion of factor XII by β-amyloid triggers inflammation in Alzhei-
mer’s patients [103]. These findings suggest that drug targeting of
the contact pathway proteases and regulators may prove beneficial
in a variety of pathologies.
In anticoagulant protein C deficiency, insufficient proteolytic
inactivation of factors V and VIII weakens this negative regulatory
feedback, causing thrombophilia. Protein S is an obligatory cofac-
tor in this reaction, and its deficiency also results in thrombophilia
even at normal levels of functional protein C [104]. Patients with
the Arg506Gln factor V Leiden mutation at one of the cleavage
sites for activated protein C have a higher risk for developing
venous thromboembolism due to decreased proteolytic factor Va
processing [105]. About 40–50% of inherited thrombophilia cases
are due to factor V Leiden, and 4–10% of the Caucasian population
342 Ingrid M. Verhamme et al.

Table 10
Cardiovascular and metabolic diseases, and stroke: activities and Uniprot ID codes (where
applicable) of proteases, protease inhibitors, and cofactors associated with cardiovascular disease,
metabolic diseases, and stroke

Protease/inhibitor Uniprot Action

Factor VII (F7) P08709 Complexes with Tissue Factor
VIIa/TF converts/activates X to Xa
VIIa/TF converts/activates IX to IXa
Factor IX (or Christmas factor) P00740 Converts/activates factor X
(F9) Activates factor VII to form factor VIIa
Activates factor X to form factor Xa
Deficiency results in hemophilia B
Factor X (Stuart–Prower factor) P00742 Converts/activates prothrombin to thrombin
(F10) Complexes with phospholipids and calcium
Activates factor VII to form factor VIIa
Thrombin (Prothrombin) P00734 Converts fibrinogen to fibrin
Activates factors V, VII, VIII, XI, XIII
Complexes with thrombomodulin
Thrombin/thrombomodulin activates protein C
Factor XI (Plasma P03951 Activates factor IX
thromboplastin antecedent) Inhibited by protein Z-dependent protease inhibitor
(ZPI)
Coagulation factor XII P00748 Reciprocal activation of Prekallikrein
(Hageman factor)
Prekallikrein (Plasma kallikrein) P03952 Reciprocal activation of factor XII
C1-inhibitor (Plasma protease P05155 Complexes with/Inactivates C1r, C1s, MASP 1, MASP
C1 inhibitor)a 2, chymotrypsin, kallikrein, fXIa, FXIIa
Protein C (Vitamin K-dependent P04070 Inactivates factors Va and VIIIa in the presence of calcium
protein C) ions and phospholipids
Activated by Thrombin/thrombomodulin complex
Factor V P12259 Cofactor required by factor Xa
Activated by Thrombin
Degraded by protein C
Factor VIII (FVIII) P00451 Cofactor required by factor IXa
Deficiency results in hemophilia A
High levels implicated in deep vein thrombosis and
pulmonary embolism
Antithrombin (Antithrombin- P01008 Inhibits thrombin and factors IXa, Xa, and XIa
III) Activity enhanced by heparin
Heparin cofactor II P05546 Inhibits thrombin and factors IXa, Xa, and XIa
Inhibits chymotrypsin
Alpha 2-antiplasmin (or α2- P08697 Inhibits plasmin and trypsin
antiplasmin or plasmin Inactivates matriptase-3/TMPRSS7 and chymotrypsin
inhibitor)
(continued)
 Proteases: Pivot Points in Functional Proteomics 343

Table 10
(continued)

Protease/inhibitor Uniprot Action

Plasmin P00747 Dissolves the fibrin of blood clots
tPA (Tissue plasminogen P00750 Converts plasminogen to plasmin on the fibrin surface
activator) Displaces plasmin from fibrin, promoting inhibition by
alpha-2-antiplasmin
Mast cell chymase P23946 Major secreted protease of mast cells
Release may promote inflammatory response
Converts angiotensin I to angiotensin II
Tryptase {6 isoforms, see Major neutral protease present in mast cells
previous list table} Resistant to endogenous proteinase inhibitors
Active only as heparin-stabilized tetramers
Six Isoforms: Tryptase alpha/beta-1(Q15661)
Tryptase beta-2(P20231), Tryptase delta(Q9BZJ3)
Tryptase gamma(Q9NRR2), Brain-specific serine
protease 4(Q9GZN4)
pro-MMP-9 (MMP-9) P14780 Cleaves collagen IV and V and fibronectin
Implicated in neovascularization in malignant gliomas
pro-MMP-1 (MMP-1) P03956 Cleaves collagens I, II, III, VII, and X
Mediates neurotoxicity of HIV viral Tat protein
pro-MMP-2 (MMP-2) P08253 Degrades extracellular matrix proteins, including collagen
I and IV
pro-MMP-3 (MMP-3) P08254 Degrades fibronectin, laminin, gelatins of type I, III, IV,
and V; collagens III, IV, X, and IX, and cartilage
proteoglycans
Activates MMPs 1, 7, and 9
Proteinase-3 (PRTN3) P24158 Degrades elastin, fibronectin, and collagen (in vitro)
Target antigen for anti-neutrophil cytoplasmic antibodies
(ANCA)
Kallikrein 13 Q9UKR3 Cleaves kininogen to the pro-inflammatory bradykinin
Cathepsin A P10619 Protects beta-galactosidase and neuraminidase
Cathepsin C (CTSC) P53634 Activates elastase, cathepsin G, granzymes A and B,
neuraminidase, factor XIII, chymase, and tryptase
Cathepsin D P07339 Intracellular degradation and protein turnover
Used by macrophages to degrade bacterial proteins
Activates ADAM30, implicated in Alzheimer’s
progression
Implicated in metastasis in breast cancer
Cathepsin L1 P07711 Intracellular degradation and protein turnover
Degrades collagen and elastin
Degrades alpha-1 protease inhibitor
Cathepsin X/Z/P Q9UBR2 Lysosomal protease, cleaves C-terminal residue
(continued)
344 Ingrid M. Verhamme et al.

Table 10
(continued)

Protease/inhibitor Uniprot Action

Calpain-10 Q9HC96 Limited proteolysis of substrates involved in cytoskeletal
remodeling and signal transduction
Cathepsin K P43235 Cleaves elastin, collagen, and gelatin
Involved in breakdown of bone for remodeling
Implicated in emphysema
Activated by inflammatory cytokines
Degraded by cathepsin S
Caspase-3 (CASP3) P42574 Activates caspases 6, 7, and 9
Activated by caspases 8 and 9
Cleaves poly(ADP-ribose) polymerase (PARP)
Cleaves and activates sterol regulatory element-binding
proteins (SREBPs)
Implicated in Huntington’s disease
Caspase-6 P55212 Dis-inhibits immune system, cleaves interleukin-10 and
interleukin-1 receptor-associated kinase 3
Cleaves poly(ADP-ribose) polymerase (PARP) and lamins
Implicated in Huntington’s and Alzheimer’s
Caspase-8 Q14790 Activates caspases 3, 4, 6, 7, 9, and 10
Activated by death receptors via FADD
Factor VII activating protease Q14520 Activates factor VII and pro-urokinase
(FSAP) May act as tumor suppressor
Tissue inhibitor of P01033 Irreversibly inhibits MMP 1, 2, 3, 7, 8, 9, 10, 11, 12, 13,
metalloproteinase-1 (TIMP-1) and 16
Activates integrin signaling via CD63 and ITGB1
a
Complement activation, blood coagulation, fibrinolysis and the generation of kinins

is heterozygous. The heterozygosity has been suggested as a poten-

tially protective, evolutionarily conserved factor against excessive
blood loss during child birth, and was shown to protect from post-
cardiac surgery hemorrhage [106].
Venous thrombosis can also originate from functional or
expression defects of antithrombin, the endogenous serpin that
irreversibly inactivates thrombin, and factors Xa, IXa, and XIa in
reactions that are dramatically accelerated by heparin. The anti-
thrombin mutation database currently lists 127 different mutations
[107], and major functional impairments are due to missense
mutations in the reactive site, the heparin-binding site and serpin-
protease contact regions important for serpin folding and stable
covalent complex formation. Heparin cofactor II (HCII) is an
equally potent but highly specific thrombin inhibitor, in the pres-
ence of cell surface dermatan and heparan sulfate, and small over-
sulfated molecules that do not affect the antithrombin–thrombin
 Proteases: Pivot Points in Functional Proteomics 345

interaction [108–111]. HCII deficiency is associated with arterial

thrombosis, development of atherosclerosis and in-stent restenosis
[44, 112]. As 60% of HCII is extravascular, it may control throm-
bin’s signaling properties, and other extravascular serine proteases
may be yet unidentified HCII targets. A few serpin-related bleeding
disorders are known: α1-PI Pittsburgh has a Met358Arg mutation
in its reactive site, which shifts its specificity from elastase to throm-
bin, thereby impairing normal blood clotting; and congenital α2-
antiplasmin deficiency results in premature lysis of hemostatic plugs
by excess plasmin. Selective inhibition of the anticoagulant acti-
vated protein C by mutating the reactive site-flanking residues of
α1-PI Pittsburgh to lysines has been shown successful in normal-
izing bleeding in a hemophilia B mouse model, and may show
promise as a novel hemophilia drug [113].
Hyperfibrinolysis by uncontrolled plasminogen activation is
characterized by excessive bleeding mimicking hemophilia. In the
congenital disease α2-AP or PAI-1 are deficient, and acquired
hyperfibrinolysis may occur in liver disease, trauma, or during
surgery. Treatment with tranexamic acid, ε-aminocaproic acid, or
other lysine analogs inhibits plasminogen activation by tPA on the
surface of the fibrin clot by occupying the lysine-binding sites on
plasminogen. This displaces plasminogen from the fibrin surface
and inhibits plasmin formation. Tranexamic acid may be anti-
inflammatory by inhibiting plasmin-dependent activation of com-
plement, monocytes and neutrophils. The lysine analogs also block
conformational plasminogen activation by the bacterial nonenzy-
matic cofactor streptokinase (SK), a fibrinolytic that has been dis-
placed by tPA in the USA but is still used in many European and
non-Western countries. SK has a C-terminal lysine residue that
binds to plasmin(ogen) kringles, thereby increasing the affinity of
the plasmin(ogen) complexes with SK and the rate of plasminogen
activation [114–116].
Many cardiovascular and metabolic pathologies have an inflam-
matory component throughout the development of the disease,
and extravascular infiltration of circulatory hemostatic proteases
upon tissue damage contributes to inflammation. Cellular proteases
produced by white blood cells also feature prominently in inflam-
mation. Macrophages transform into cholesterol- and lipid-laden
foam cells in the atherosclerotic vascular wall. Monocytes, neutro-
phils, lymphocytes, and mast cells in particular play a role in foam
cell formation in the arterial intima [94]. Mast cell chymase con-
verts angiotensin-I in vascular cells to the potent pro-inflammatory
angiotensin II that upregulates expression of redox-sensitive cyto-
kines, chemokines, and growth factors implicated in the formation
of atherosclerotic lesions [117]. Elevated angiotensin II causes
arterial hypertension, and has been implicated in vascular prolifera-
tion, aortic valve disease, myocardial infarction, heart failure, and
abdominal aortic aneurysm (AAA). Mice with angiotensin
346 Ingrid M. Verhamme et al.

II-induced hypertension develop arterial vascular inflammation,

dependent on thrombin-triggered activation of factor XI bound
to platelets via its receptor glycoprotein Iba [118]. Patients with
uncontrolled arterial hypertension also exhibit factor
XI-dependent, amplified platelet-localized thrombin generation
which may serve as an inflammatory marker of high blood pressure.
Blocking factor XIa activity in combination with inhibition of the
renin-angiotensin system may show promise in treating hyperten-
sion and associated vascular inflammation. Inhibition of the renin-
angiotensin system in animal models and humans also diminishes
plaque formation, and may provide an avenue for treatment and
prevention of atherosclerosis.
Chymase and tryptase degrade ApoE and HDL3, thereby
decreasing cholesterol efflux from foam cells and impairing choles-
terol reverse transport. Chymase induces SMC apoptosis, inhibits
SMC growth and collagen synthesis, and degrades endothelin-1,
leading to impaired vasodilation. Chymase activates pro-MMP-9
whereas tryptase activates pro-MMP-1, -2, and -3, all involved in
the development of atherosclerosis and abdominal aortic aneurysm.
Extracellular matrix degradation by elevated MMPs facilitates che-
mokine- and angiogenic factor-triggered migration of leukocytes
and endothelial cells, which accompanies neovascularization and
growth of the atherosclerotic lesion, and eventually facilitates
plaque rupture [119]. Chymase-activated TGF-β1 disrupts endo-
thelial function and also contributes to intima thickening. Elevated
plasma chymase and tryptase levels were detected in patients with
acute myocardial infarction (MI) or unstable angina pectoris but
not in stable angina, indicating a correlation with plaque instability.
The mast cell inflammatory cytokines IL-6, TNF-α, and IFN-γ
induce smooth muscle cell and endothelial cell expression of chy-
mase and tryptase, and their plasma levels correlate directly with the
AAA expansion rate. Elevated levels of matrix MMPs are associated
with the development of AAA, and high plasma MMP-1 and -9
concentrations are indicative of poor outcomes after aneurysm
rupture [120]. Anti-angiogenic drugs showed adverse effects in
clinical trials with cancer patients suffering from atherosclerosis,
and drugs targeting proteases may be an alternative to help combat
atherosclerosis. Caspase-mediated apoptosis occurs in atherosclero-
sis, and both beneficial and harmful caspase effects have been
reported. In a population study, apoptotic markers of 4284 subjects
were measured, and at the mean 19-year follow-up, 381 patients
presented with adverse cardiovascular events. Elevated caspase-8 at
baseline was strongly correlated with their incidence [121]. Macro-
phage apoptosis in atherosclerosis may have both pro- and anti-
atherogenic effects, and more studies are needed to elucidate these
complex mechanisms.
Plasma chymase is elevated in type 2 diabetes and prediabetes,
and clinical trials currently evaluate chymase and tryptase as drug
 Proteases: Pivot Points in Functional Proteomics 347

targets for small molecule inhibitors. Chymase-generated angioten-

sin II contributes to islet disorganization and high risk for cardio-
vascular events in diabetic patients. Urinary extracellular vesicles of
patients with diabetic nephropathy contain elevated levels of
MMP-9, proteinase-3, kallikrein 13, and cathepsins A, C, D, L
and X/Z/P compared to controls [122], and these proteases derive
from neutrophils and monocytes, recruited to the glomerular
endothelial cells. The profiles may have prognostic and diagnostic
value in the assessment of kidney damage in type 1 and 2 diabetes.
Proteinase-3 cleaves insulin-like growth factor 1 and promotes
glomerular inflammation. These combined findings illustrate how
endothelial dysfunction and inflammation may be predictors of
diabetic nephropathy.
A strong association exists between calpain-10 and type 2 dia-
betes, and blocking calpain activation prevents diabetes-associated
cardiac injury. Mast cells and macrophages produce cathepsins,
among which L and K are associated with obesity. In mouse models
of obesity, L and K knockout mice or wild-type mice given L- and
K-selective small molecule inhibitors were significantly leaner than
control mice and had improved glucose sensitivity [94]. Cathepsin
K is a marker of adiposity, and recent findings report cathepsin S
and D association with human obesity. Maternal diabetes may cause
embryonic neural tube defects, characterized by elevated levels of
caspase-3, -6, and -8. The mechanism involves proteolytic activa-
tion of the effector caspases-3 and -6 by initiator caspase-8 [123].
Caspase-3 is also a major effector of insulin-producing pancreatic
β-cell apoptosis in type 1 diabetes [124].
In ischemic stroke, arterial blockage can be caused by a throm-
bus formed within the brain, or an embolus formed elsewhere in
the body. Hemorrhagic stroke, less frequent but often more severe,
is the result of a ruptured blood vessel. In patients on anticoagu-
lants or fibrinolytics, ischemic stroke may develop a hemorrhagic
component, known as hemorrhagic transformation. Factor VII
activating protease (FSAP) is a plasma serine protease that activates
pro-urokinase (pro-uPA) rather than factor VII. The FSAP-
Marburg I polymorphism (1704G > A), which reduces FSAP
activity, increases stroke risk and mortality but seems to lower the
risk of developing carotid restenosis in atherosclerotic patients
[125]. Ischemic stroke triggers uncontrolled MMP-2 and
MMP-9 activity, associated with disruption of the blood-brain
barrier and onset of edema, and MMP-9 is also elevated in hemor-
rhagic transformation [126]. Expression of the endogenous tissue
inhibitor of metalloproteinase-1 (TIMP-1) is observed in conjunc-
tion with elevated MMP-9, as a protective response to tissue injury.
Neutrophils, rather than resident brain cells are the main source of
pro-MMP-9 following stroke, and upon degranulation, the proen-
zyme is proteolytically activated in the extracellular space. MMP-2
and TIMPs are ubiquitously expressed in tissues of the central
348 Ingrid M. Verhamme et al.

nervous system [127]. These findings suggest that focusing on

endothelial cells, pericytes, astrocytes, and infiltrating leukocytes,
rather than neurons, may prove to be more successful in identifying
new therapeutic targets. Understanding the relationship between
MMP-9 and neutrophils may help elucidate mechanisms involved
in disruption of the blood-brain barrier, and lead to more successful
therapeutic approaches.

3.4 Cancer The long-held concept that somatic mutations are the causal event
in the majority of cancers has recently come under scrutiny
[128, 129]. Priming the cellular microenvironment for develop-
ment of cancer is characterized by a sequence of events that precede
the transformation of a normal cell into a cancer cell, and somatic
mutations are actually later events in the development of many
cancers. Chronic inflammation and fibrosis have been identified as
two of these events. Hemostatic proteases have recently been
recognized to contribute to inflammatory processes in cancer. Dis-
ruption of the endothelial barrier during tissue damage allows
hemostatic zymogens to be activated. Not only do these proteases
contribute to extravascular coagulation and fibrinolysis, they also
trigger signaling through cell surface activation of PAR receptors,
binding to uPAR and LRP-1, and activation of MMPs
[130]. Inflammation also triggers the release of TGF-β that
potently induces MMP-2 and MMP-9 expression. In turn,
MMP-2, -9, and -14 proteolytically activate latent TGF-β in the
ECM. Transmembrane MMP-14 and several members of the
ADAM family are localized on invadopodia of migrating cells.
The involvement of MMPs in extracellular matrix remodeling facil-
itates tumor invasion, and MMPs also figure prominently in cancer-
related signaling. Whereas many MMPs are recognized as
pro-tumorigenic, some may negatively affect cancer progression,
depending on the microenvironment of the cell [131]. The MMP
and ADAM inhibitor Marimastat showed no broad therapeutic
anticancer potential due to lack of specificity; however, it inhibited
ADAM-17, highly expressed in renal cell carcinoma [132]. This
inhibition downregulated Notch pathway-mediated cell prolifera-
tion and invasion more effectively than γ-secretase inhibition.
Hence, Marimastat may have therapeutic potential in renal cell
cancer. Tissue inhibitors of metalloproteinases (TIMPs) are differ-
entially expressed in cancer: high TIMP1 expression is associated
with fibrotic processes and poor outcome, and TIMP3 silencing
indicates advanced disease [133]. TIMPs figure prominently in
other pathologies such as cardiovascular disease and sepsis, and
fibrosis as measured by TIMP1 levels was recently shown to predict
all-cause mortality in the AGES-Reykjavik Study [134] (Table 11).
Epigenetic processes are increasingly recognized as essential in
carcinogenesis. During epithelial-mesenchymal transition (EMT)
in cancer initiation, progression, and metastasis, epigenetic
 Proteases: Pivot Points in Functional Proteomics 349

Table 11
Cancer: activities and Uniprot ID codes (where applicable) of proteases, protease inhibitors, and
cofactors associated with cancer

Protease/inhibitor Uniprot Action

MMP-2 P08253 Degrades extracellular matrix proteins,
including collagen I and IV
MMP-9 P14780 Cleaves collagen IV and V and fibronectin
Implicated in neovascularization in
malignant gliomas
MMP-14 P50281 Degrades extracellular matrix proteins
Activates progelatinase A and MMP-15
Inhibits angiogenesis via cleavage of
ADGRB1
ADAM-17 (ADAM metallopeptidase P78536 Activates tumor necrosis factor alpha
domain 17) Activates Notch Pathway
Sheddase, activates multiple growth
factors
Implicated in tumor resistance to
radiotherapy
Tissue inhibitor of metalloproteinase-1 P01033 Irreversibly inhibits MMP 1, 2, 3, 7, 8, 9,
(TIMP-1) 10, 11, 12, 13, and 16
Activates integrin signaling via CD63 and
ITGB1
TIMP-3 (Metalloproteinase inhibitor 3) P35625 Irreversibly inhibits MMP 1, 2, 3, 7,
9, 13, 14, and 15
Thrombin (Prothrombin) P00734 Converts fibrinogen to fibrin
Activates factors V, VII, VIII, XI, XIII
Complexes with thrombomodulin
Thrombin/thrombomodulin activates
protein C
PAR-1 (Proteinase-activated receptor 1 or P25116 Stimulates phosphoinositide hydrolysis
coagulation factor II (thrombin) receptor) Activated by thrombin
May play a role in vascular development
SUMO (Sentrin-specific protease 7) Q9BQF6 Removes SUMO (Small Ubiquitin-like
Modifier protein) 2 and 3
Serine Protease HTRA1 Q92743 Degrades extracellular matrix
Degrades insulin-like growth factor
receptors and tuberin
Cathepsin L1 P07711 Intracellular degradation and protein
turnover
Degrades collagen and elastin
Degrades alpha-1 protease inhibitor
Cathepsin B P07858 Intracellular degradation and protein
turnover
Upregulation of Cathepsin D, matrix
metalloproteinase, and urokinase
Implicated in metastasis and immune
resistance
(continued)
350 Ingrid M. Verhamme et al.

Table 11
(continued)

Protease/inhibitor Uniprot Action

Cathepsin D P07339 Intracellular degradation and protein
turnover
Used by macrophages to degrade bacterial
proteins
Activates ADAM30, implicated in
Alzheimer’s progression
Implicated in metastasis in breast cancer
Mesotrypsin (Trypsin-3) P35030 Degradation of natural trypsin inhibitors
Matriptase (Suppressor of tumorigenicity Q9Y5Y6 Degrades extracellular matrix, trypsin-like
14 protein) activity
Promotes epithelial differentiation and
possibly growth
Implicated in metastasis
HAI-1 (hepatocyte growth factor activator O43278 Inhibits HGF and matriptase
inhibitor type1 (SPINT1))
HAI-2 (hepatocyte growth factor activator O43291 Inhibits HGF, possibly inhibits serine
inhibitor type2 (SPINT2)) proteases generally
Implicated in suppression of liver cancer
Proteasome Group of massive protease complexes,
stacked ring structure
Degrade proteins “tagged” with multiple
ubiquitins
Critical in protein turnover, apoptosis,
and adaptive immune response
Kallikrein-3 (hK3) P07288 Liquifies seminal fluid, degrades cervical
mucus
Elevated levels associated with prostate
cancer
Kallikrein-5 (hK5) Q9Y337 Degrades extracellular matrix proteins in
epithelium, leads to cell shedding
Gingipains (RgpA, RgpB, and Kgp)
rgpA (Gingipain R1) <human> P28784 Bacterial thiol proteases
Degrade host tissue proteins and
cytokines
SPINK6 (Serine protease inhibitor Q6UWN8 Inhibits KLK4, KLK5, KLK6, KLK7,
Kazal-type 6) KLK12, KLK13, and KLK14

mechanisms such as DNA methylation and histone modifications

regulate EMT-related genes [135]. The transformation of epithelial
cells into migratory fibroblasts and mesenchymal cells is a hallmark
of metastasis, and various protease activities are associated with this
process. In human gastric cancer cells, thrombin-catalyzed
 Proteases: Pivot Points in Functional Proteomics 351

activation of PAR-1 is thought to trigger EMT [136], and in breast

cancer, induction of the SUMO-specific protease 7 long variant
promotes gene expression favoring cell proliferation and EMT
[137]. Mammalian intracellular high-temperature requirement A
(HtrA) serine proteases contain a chymotrypsin-like domain and
play a role in protein quality control. Epigenetic silencing of the
HTRA1 gene in cancer cells may be caused by histone deacetylase
targeting of the promoter, or transcription repression of the methy-
lated promoter by binding of methyl-CpG-binding domain protein
2 (MBD2) [138]. Silencing of tumor suppressor microRNAs via
protease-activated PAR and Nf-κB signaling, and of caspase-
8 expression by DNA methylation are yet a few other epigenetic
mechanisms associated with the development of certain cancers
[139, 131].
Altered expression of secreted lysosomal cysteine cathepsins has
been associated with a variety of cancers, and several studies corre-
late either overexpression or gene knockout with progression of
malignancy, depending on the type of cathepsin, and the nature and
localization of the cancer. Overexpression of tumor tissue cathe-
psins B and L is detected in ovarian cancer but not in benign tumors
and control tissue, and plasma cathepsin L is elevated in patients
with malignant tumors. These proteases may be useful biomarkers
[140]. Native and mutated forms of the aspartic protease cathepsin
D precursor feature prominently in metastatic breast cancer. This
tumor marker is present at higher levels in invasive ductal carcino-
mas, lymph node metastases, and hormone-receptor negative can-
cers than in lobular cancers and nodal positive carcinomas [141].
About 90% of the cancers originate in the epithelium. Epithelial
cells express mesotrypsin and matriptase, and upregulation of these
proteases is observed in epithelial cancers. Increased mesotrypsin
activity is an indicator of poor prognosis in breast, prostate, pancre-
atic, and many other cancers [142]. Mesotrypsin is unusual in that
it is not inhibited by Kunitz and Kazal-like trypsin inhibitors, but
rather recognizes these proteins as substrates. It exhibits specificity
for Arg/Lys-Ser/Met bonds, targets thrombin substrates such as
the PAR1, 3, and 4 receptors, and is not inhibited by α1-antitrypsin
(Met-Ser reactive bond) but by α1-antitrypsin Pittsburgh
(Arg-Ser). Engineering a triple mutant M17G/I18F/F34V of the
human amyloid precursor protein Kunitz protease inhibitor
domain (APPI) created a selective, tight-binding inhibitor with an
inhibition constant (Ki) of 89 pM, active in cell-based models of
mesotrypsin-dependent prostate cancer cell invasiveness
[143]. The crystal structure of the APPI M17G/I18F/F34V/
mesotrypsin complex shows unique active site features that may
be critical in driving metastasis, given the observation that other
trypsins do not contribute to the invasive prostate cancer pheno-
type. This structural information may facilitate development of
therapeutic peptide inhibitors, complementary to the mesotrypsin
active site. Matriptase is the best-known member of the serine
352 Ingrid M. Verhamme et al.

proteases with a type II N-terminal transmembrane domain, and its

endogenous inhibitors are the Kunitz-type hepatocyte growth fac-
tor activator inhibitor types 1 and 2 (HAI-1 and 2). In normal
tissue, matriptase proteolytic activity is tightly regulated by excess
HAI-1 and 2, whereas in cancer tissue this balance is tilted heavily
toward excess matriptase. Transgenic expression of epidermis
matriptase in a squamous cell carcinoma mouse model causes
tumor formation that is inhibited by HAI-1 or HAI-1 co-expres-
sion [144], strongly suggesting matriptase proteolytic activity as an
essential trigger for malignancy. In highly aggressive inflammatory
breast cancer (IBC), matriptase proteolytically activates
pro-hepatocyte growth factor (pro-HGF). Binding of HGF to the
receptor tyrosine kinase, c-Met, activates signaling pathways lead-
ing to cell proliferation, migration, morphogenesis, and invasion
[145]. Both matriptase and c-Met are membrane-bound in IBC
cells, and upregulated in cancer cells of IBC patients. Proliferation
and invasion of IBC cells is halted by silencing with RNAi or
treatment with synthetic matriptase inhibitors, illustrating their
potential merit in IBC therapy.
Excessive proteasome activity occurs in certain blood cancers.
Degradation of pro-apoptotic factors such as p53 impairs pro-
grammed cell death in cancer cells, and proteasome inhibition was
proposed as a potential cancer treatment. The proteasome inhibi-
tors bortezomib, carfilzomib, and ixazomib were approved by the
FDA for the treatment of multiple myeloma, and are currently in
clinical trials of blood, lung, and breast cancers. These potent
inhibitors of the β5 peptidase activity of the 26S proteasome have
only modest activity against β1 and β2 peptidases, which appears to
limit their usefulness to multiple myeloma [146]. The β5 peptidase
inhibitors have not been successful in treatment of solid tumors,
and in recent studies triple negative breast cancer cell lines only
responded to bortezomib or carfilzomib after CRISPR gene edit-
ing to inactivate β2 [147]. Development of dual β2/β5 inhibitors,
while conceptually attractive, may prove daunting, and combina-
tion therapy with the β5 and known β2 inhibitors is a more realistic
approach.
Balanced complement-associated inflammation can be of
advantage in potentiation of immunotherapy, whereas an imbalance
may sustain tumor cell proliferation, migration, invasiveness, and
metastasis [148]. Genetic and epigenetic changes mark tumor cells
as nonself, and the innate immune cells assist in clearing opsonized
tumor cells through concerted actions of antitumor monoclonal
antibodies (mAb) and complement cytotoxicity. Rituximab and
ofatumumab, FDA-approved chimeric anti-CD20 mAbs were
developed for the treatment of B cell lymphomas and chronic
lymphocytic leukemia. Their targeting of tumor antigens elicits
complement-dependent phagocytosis. Binding of C1q to the Fc
portion of the mAbs results in formation of the proteolytically
active C1 complex that initiates the cascade. Other tumor-specific
 Proteases: Pivot Points in Functional Proteomics 353

mAbs recognize CD38 and CD52 as epitopes highly expressed on

B cell- or T cell-derived tumors, and they are well studied in terms
of their ability to elicit complement-dependent cytotoxicity. How-
ever, some solid tumors downregulate complement cytotoxicity
and opsonization by overexpressing or sequestering surface pro-
teins, thus limiting the efficacy of therapeutic antibodies.
The connection between complement activation, chronic
inflammation, and cancer is becoming increasingly evident
[149]. Complement factors and their active cleavage products
themselves contribute to mitogenic signaling cascades and growth
factor production (C3a, C5a, and the membrane attack complex
MAC), angiogenesis (C3, C3a, C5a, MAC), protection from anti-
growth signals and apoptosis (C3a, C4, C5a, MAC), cellular inva-
sion and migration through the extracellular matrix (C1q, C1s,
factor B, C3, C3a, C3d, C5, C5a, C9), proliferation (C3, C3a,
C4, C5a, MAC), and suppression of antitumor immunity (C5a).
From these observations it is clear that complement itself can
promote cancer, under the “right” circumstances. The physiologi-
cal role of the MAC complex is to disrupt the cell membrane and
cause lysis; however, subthreshold MAC activity does not kill the
cell but activates the cell cycle and triggers proliferation. Hence
inhibition of complement may become an emerging strategy for the
fight against cancer.
Excessive complement activation in colorectal, breast, pancre-
atic, lung, prostate, esophageal cancer, lymphoma, and leukemia
has suggested the use of C3 activation fragments as diagnostic or
prognostic biomarkers. However, C3 is abundant in plasma, and
mass spectrometry quantitation may not be straightforward. Intra-
tumoral C3 expression in ovarian cancer has been linked with
disease prognosis, and a C3 fragment is found in prostatic fluid
from cancer patients. Prostate-specific antigen (PSA) cleaves C3
and C5, and may act pro-tumorigenic by proteolysis of comple-
ment proteins.
Human tissue kallikreins (hK) are secreted serine proteases that
are differentially expressed in many endocrine cancers. The KLK3,
8, 10, 13, and 14 genes are thought to encode tumor suppressor
proteins, illustrating the recently recognized concept that protease
upregulation does not always reflect tumor progression
[131]. With this in mind, defining target protease specificity is
crucial when developing protease inhibitors as potential anticancer
drugs. hK3 or prostate-specific antigen (PSA) is the best-known
biomarker for screening, diagnosis, and monitoring of prostate
cancer. PSA is also elevated in benign prostatic hyperplasia, and its
extent of complex formation with the serpin, α1-antichymotrypsin,
differentiates between both pathologies. Tissues and plasma of
prostate cancer patients contain higher levels of the complex than
those of patients without cancer, and a level of 25% or more of free
PSA activity is generally a good indicator of benign hyperplasia
[150]. Many other kallikreins may also be suitable cancer
354 Ingrid M. Verhamme et al.

biomarkers. hK5 proteolytically activates PAR-2, leading to Nf-κB

activation and downregulation of tumor suppressor microRNAs in
oral squamous cell carcinoma [139]. Plasma kallikrein is capable of
activating the complement system, linking inflammatory responses
to many cancer-related processes [151].
A very strong correlation has also been observed between
cancer and hemostasis, and cancer patients invariably exhibit hyper-
coagulability, contributing to mortality and morbidity [152]. This
prothrombotic state is attributed to the ability of tumor cells to
activate coagulation, by producing procoagulant factors and
inflammatory cytokines; interacting with monocytes, platelets, neu-
trophils, and vascular cells; and triggering acute-phase reactants and
necrosis. Various anticoagulant therapies with heparins, vitamin K
antagonists, or direct oral anticoagulants have proven to be benefi-
cial in the treatment of cancer patients [153].

3.5 Neuro- Patients with Alzheimer’s disease (AD), representing up to 70% of

degenerative Diseases the dementia cases, have brain tissue containing amyloid plaque
composed of toxic Aβ peptides, and neurofibrillary tangles com-
posed of tau protein. Tau is localized in neuronal axons, promotes
tubulin polymerization, and stabilizes microtubules. Amyloid β A4
precursor protein (APP) is a highly conserved synaptic integral
membrane protein, thought to regulate synapse formation, neural
plasticity, and maintenance of homeostasis in the central nervous
system. Normal proteolytic processing of APP occurs via cleavage
by α-secretase to release extracellular APPsα, followed by cleavage
by intramembrane γ-secretase, with release of an extracellular p3
fragment, and an intracellular AICD fragment [154]. APP proces-
sing is different in AD: sequential cleavage by β- and γ-secretase
releases the extracellular APPsβ, Aβ1-40 and Aβ1-42 peptides, and
intracellular AICD. Heterogeneous proteolytic degradation yields
several other extracellular species, ranging from 37 to 49 residues.
Aβ1-40 and Aβ1-42 are considered neurotoxic, and form plaques in
the brain. The APPsβ fragment oligomerizes and mediates death
receptor signaling. The Aβ1-42 fragment in cerebrospinal fluid is
routinely used as a biomarker, combined with the measurement of
total and hyperphosphorylated tau protein. This combination assay
can diagnose AD in an early stage, and provide a prognosis of
disease progression [155]. N-truncated Aβ peptides with cyclized
terminal glutamate residues figure prominently in amyloid deposits,
and are particularly useful additional biomarkers. However, no clear
correlation could be made between plasma Aβ fragment concentra-
tions and AD, dementia and various stages of cognitive decline,
perhaps in part due to the limited sensitivity of the current analyti-
cal methods [156]. Due to observed discrepancies between AD
dementia and amyloid deposition, some groups have suggested
that either soluble oligomeric Aβ peptides may be more toxic, or
that tau neurofibrillary tangles may be the pathogenic species [157]
(Table 12).
Table 12
Neurodegenerative diseases: activities and Uniprot ID codes (where applicable) of proteases,
protease inhibitors, a critical disease-related protease target, and cofactors associated with
neurodegenerative diseases

Protease/inhibitor Uniprot Action

Amyloid β A4 precursor P05067 Cell surface receptor involved in neuronal growth, adhesion,
protein (APP) and motility
Upregulated in neuronal repair
Proteolysis generates amyloid β peptides (Abeta) from 37 to
49 residues in length, Abeta 40 and Abeta 42 implicated in
Alzheimer’s disease
α-Secretase Group of ADAM family sheddases which cleave the amyloid
precursor protein (APP)
γ-Secretase Cleaves single-pass transmembrane proteins, including APP
Transmembrane complex of presenilin-1 (PSEN1)
(P49768), nicastrin (Q92542), anterior pharynx-
defective 1 (APH-1) (Q5TB21), and presenilin enhancer
2 (PNE-2) (Q9NZ42)
β-Secretase 1 (BACE1) P56817 Transmembrane aspartic-acid protease important for myelin
sheath formation
Cleaves APP to form Abeta 40 and Abeta 42
Several BACE1 inhibitors currently being tested as
Alzheimer’s treatments
ADAM10 O14672 Primary alpha secretase in platelets
MMP-9 P14780 Cleaves collagen IV and V and fibronectin
Implicated in neovascularization in malignant gliomas
TIMPs (tissue-inhibitors of Endogenous inhibitors of MMPs
metalloproteinases)
Caspase-6 P55212 Dis-inhibits immune system, cleaves interleukin-10 and
interleukin-1 receptor-associated kinase 3
Cleaves poly(ADP-ribose) polymerase (PARP) and lamins
Implicated in Huntington’s and Alzheimer’s
Calpain (family) Family of calcium-dependent, non-lysosomal cysteine
proteases
Excessive activity implicated in cytoskeletal degradation and
altered calcium homeostasis in Alzheimer’s disease
Calpastatin P20810 Endogenous calpain inhibitor
m-AAA (Mitochondrial AAA Group of ATP-dependent mitochondrial proteases
proteases)
HtrA2 (high temperature O43464 Mitochondrial serine protease, initiates cell death by binding
requirement) IAPs (Inhibitors of Apoptosis Proteins)
Implicated in Parkinson’s disease
Presenilins-associated Q9H300 Antiapoptotic, activates optic atrophy 1 (OPA1) which
rhomboid-like protein prevents release of cytochrome C into cytosol
(PARL) Mutation controversially implicated in Parkinson’s disease
SUMO-specific protease Q9HC62 Processes SUMO1, SUMO2, and SUMO3 into mature
2 (SENP2) proteins
Deconjugates SUMO1, SUMO2, and SUMO3 from target
proteins
356 Ingrid M. Verhamme et al.

APP is also produced in platelets, and cleaved by platelet

α-secretase to release soluble APPsα in the circulation upon platelet
activation [158]. APP isoforms of 130 kDa (intact) and
106–110 kDa (cleaved) can be detected by immunoblotting of
platelet lysates. Patients with AD and mild cognitive impairment
(MCI) have a significantly lower ratio of intact vs. cleaved APP than
healthy controls. The decrease in ratio parallels cognitive decline
and may predict conversion from MCI to AD. The metalloprotease
ADAM10 in the brain processes APP through an alternative,
non-amyloidogenic pathway [159], and platelets have the same
proteolytic machinery as neurons for processing APP. The increase
in the amyloidogenic pathway in AD patients is reflected by a
decrease of platelet α-secretase and ADAM10 activity, and an
increase of platelet β-secretase activity. These observations suggest
that platelet biomarker assays for AD may be feasible, and raised the
question if Aβ peptides generated by platelets can enter the brain
and contribute to neuronal deficit. A recent elegant in vivo study
showed indeed that Aβ peptides, originating from transgenic AD
mice in prolonged parabiosis with healthy wild-type mice, accumu-
lated in the brain of the healthy mice [160]. This finding strongly
suggests a biological connection between altered platelet and neu-
ronal protease expression in AD patients.
Clinical trials have focused on the development of β- and
γ-secretase inhibitors, and antibodies targeting Aβ peptides with
the goal of peptide removal as a means of reducing plaque forma-
tion. To date the results have been disappointing, except for clinical
trials of the monoclonal Aducanumab, directed against aggregated
and soluble Aβ peptides [161]. In contrast, the monoclonal Sola-
nezumab, targeting soluble monomeric Aβ, failed in three consec-
utive trials, the latest one Expedition3, which was halted in January
2018 [162]. Small molecule inhibitors of β-secretase were also put
to the test: negative results halted the Verubecestat Epoch trial in
February 2017, but two trials studying the compound
JNJ-54861911 are set to run until 2023. The 2010 phase III failure
of Semagacestat, a γ-secretase inhibitor, was due to side effects of
blocking Notch signaling; however, some questions remain about
the design of the clinical trial and potential optimization of the drug
dosage [163]. Although chronic but partial lowering of γ-secretase
activity in heterozygous knockout mice does not cause a diseased
phenotype and might have been tolerated in humans, the trial
design opted for short peaks of complete γ-secretase inhibition in
the brain alternated with periods of normal activity. This proved to
wreak havoc on the ultradian oscillation of Notch signaling, as
corroborated by severe Notch phenotypes in complete γ-secretase
KO mice. Plasma concentrations of the drug were ~360-fold higher
than the IC50 for γ-secretase inhibition in cell culture, and the side
effects on the skin, gastrointestinal system, and weight loss due to
very high intermittent drug dosage might also have contributed to
 Proteases: Pivot Points in Functional Proteomics 357

poor performance in cognitive tests. With these caveats, there are

still unexplored options for the development of γ-secretase inhibi-
tors as drugs for cognitive decline.
Other proteases have been implicated in neurogenerative dis-
eases. Several MMPs cleave APP in vitro, raising the question
whether they also do so in vivo, and if there is a correlation with
circulating MMPs and their inhibitors in AD. Expression of
MMP-9 and the tissue inhibitors (TIMPs) was found to be elevated
in postmortem AD brain tissue [164]. Significantly higher levels of
MMP-9, but not of MMP-2 or the TIMPs were found in the
plasma of AD patients, suggesting that MMP-9 may contribute to
AD. Caspase-6 is found in non-apoptotic brain tissue of Hunting-
ton’s disease and AD patients, indicating a function other than its
executioner role [58]. Caspase-6 is implicated in axonal degenera-
tion and neuronal loss in both diseases, and it cleaves tau, CREB-
binding protein (CBP) which regulates transcription in cortical
neurons, and NF-κB. Hence selective caspase-6 inhibitors may
have therapeutic potential. CREB is indispensable for synaptic
plasticity, and its impaired activation contributes to AD
[165]. CREB is a substrate of the neutral, cytosolic cysteine prote-
ase calpain, and inhibition of this protease restores synaptic plastic-
ity in a mouse model of familial AD. Calpain also cleaves tau
protein, and upregulation or decreased degradation of the endoge-
nous calpain inhibitor calpastatin has been a therapeutic goal in
AD. Defective mitochondrial proteases can cause neuronal cell
death and axonal dysfunction [166], and the human proteases
m-AAA, the serine protease HTRA2 (high temperature require-
ment) and the rhomboid protease PARL have been identified in
neurodegenerative processes. Two human m-AAA isoenzymes are
differentially involved in neurodevelopment and protection against
neurodegeneration, by preventing accumulation of misfolded poly-
peptides, and regulating mitochondrial protein synthesis, transport
and proteolytic control of gatekeeping functions to prevent Ca2+
overload in the neuron [167]. Mutations in m-AAA cause heredi-
tary spastic paraplegia and spinocerebellar ataxia. HTRA2 and
PARL increase the susceptibility of neurons to apoptotic cell
death. HTRA2 is involved in caspase-dependent apoptosis and in
Parkinson’s disease [168], but the role of PARL is still controver-
sial. Posttranslational modification of proteins by small ubiquitin-
related modifier (SUMO) can be reversed by SUMO-specific pro-
tease 2 (SENP2), and accumulation of SUMO-conjugated proteins
is observed in patients with neurodegeneration. A knockout mouse
model confirmed that disruption of this mitochondrial protease
causes neuronal cell death [169]. Finally, the link between neuro-
degeneration and dysregulated complement activity has been firmly
established. Acute brain injury triggers uncontrolled complement
activation, flooding of the injury site with inflammatory anaphyla-
toxins and phagocytes, and blood brain barrier (BBB) damage
358 Ingrid M. Verhamme et al.

[170]. However, normal complement function plays a role in brain

development (wiring), and brain homeostasis and repair during
adulthood. Therapeutic approaches of complement modulation
will therefore depend on its acute, subacute, and chronic nature
of activation, and will require selective targeting of complement
components.

3.6 Autoimmune Upregulation and activation of pro-inflammatory cytokines and

Diseases chemokines, uncontrolled endogenous protease activity, inflamma-
tion and antibodies/T lymphocytes against “self” antigens are hall-
marks of autoimmune diseases [171]. Chemokines recruit
leukocytes to release MMP-9 that generates peptides with immu-
nodominant epitopes. These epitopes are presented to autoreactive
T lymphocytes and stimulate B cells to produce autoantibodies.
The “Remnant Epitopes Generate Autoimmunity” (REGA)
model, based on cytokine, chemokine, and protease action, has
been validated for multiple sclerosis, rheumatoid arthritis, and
diabetes. According to this model, potential strategies for disease
treatment may involve the use of anti-inflammatory cytokines, and
the inhibition of pro-inflammatory and protease-inducing cyto-
kines and chemokines. MMP-9 cleavage of collagen in rheumatoid
arthritis and of insulin in autoimmune pancreatitis was found to
generate remnant epitopes. Inflammasomes are large macromolec-
ular complexes involved in activation of procaspase-1. Caspase-1
proteolytically activates the precursors of the pro-inflammatory
cytokines IL-1β, IL-18, and IL33, and has been implicated in
various autoimmune diseases. IL-1β blockade in autoimmune dis-
eases can be accomplished with IL-1 receptor antagonists, neutra-
lizing monoclonal antibodies, and the injectable IL-1β inhibitor
Rilonacept [172]. Clinical trials of Pralnacasan, an oral caspase-1
inhibitor for treatment of rheumatoid arthritis, were halted in 2003
after liver toxicity was observed in animal studies (Table 13).
B cells contribute to autoimmune diseases by secretion of
autoantibodies, presentation of autoantigen, and inflammatory
cytokine secretion. Antibody therapy with Rituximab targets
CD20 on the B cell surface, triggering cell death, and is used for
B cell depletion to treat rheumatoid arthritis, idiopathic thrombo-
cytopenic purpura, pemphigus vulgaris, and myasthenia gravis. The
recent discovery that the intramembrane signal peptide peptidase-
like protease SPPL2A promotes B cell differentiation by cleavage of
CD74 suggested that SPPL2A may be a suitable target for inhibi-
tion in the treatment of autoimmune diseases [82]. Major histo-
compatibility complex (MHC) class II-mediated priming of T and
B lymphocytes occurs in systemic lupus erythematosus (SLE) and
lupus nephritis. The cysteine protease cathepsin S degrades CD74
during MHC II assembly with antigenic peptide in antigen-
presenting cells, and cathepsin S inhibition might be therapeutic
in SLE [173]. In some cases, deficiency of protease activity is
 Proteases: Pivot Points in Functional Proteomics 359

Table 13
Autoimmune diseases: activities and Uniprot ID codes (where applicable) of proteases, protease
inhibitors, and cofactors associated with autoimmune diseases

Protease/inhibitor Uniprot Action

MMP-9 P14780 Cleaves collagen IV and V and fibronectin
Implicated in neovascularization in malignant
gliomas
Caspase-1 (Interleukin-1 converting P29466 Activates interleukin 1β and interleukin
enzyme) CASP1 18, initiating inflammation
Activates gasdermin D, initiating lytic cell
death
Activated by incorporation into inflammasome
complex, initiated by NOD-like receptors or
AIM-1 like receptors
Inhibited by CARD only proteins (COPs),
COPs prevent formation of inflammasome
SPPL2A (Signal peptide peptidase-like 2A) Q8TCT8 Cleaves type II membrane signal peptides,
such as tumor necrosis factor alpha (TNF),
the Fas antigen ligand (FASLG), and
Cluster of Differentiation 74 (CD74)
Initiation of innate immune response through
CD74 activation implicated in autoimmune
diseases
Cathepsin S P25774 Cleaves proteins into peptides for presentation
as antigens in macrophages, B-lymphocytes,
microglia, and dendritic cells
ADAMTS13 (a disintegrin and Q76LX8 Degrades von Willebrand factor, adversely
metalloproteinase with a thrombospondin affects clot formation
type 1 motif, member 13)

associated with autoimmune disease. The plasma of thrombotic

thrombocytopenic purpura (TTP) patients contains unusually
large forms of von Willebrand factor multimers. Most TTP cases
arise from autoantibody-mediated inhibition or accelerated clear-
ance of ADAMTS13 [174]. Highly similar anti-ADAMTS13 auto-
antibodies were found in unrelated TTP patients, suggesting that
this autoimmune response is antigen-driven.

3.7 Proteases, Infectious organisms employ their own arsenal of proteases for
Inhibitors, and propagation and virulence, such as HIV protease [175], Trypano-
Cofactors in Infectious soma cruzi cruzipain [176], Porphyromonas gingivalis gingipains
Organisms [177], and Bacillus anthracis lethal factor [178]. Several bacterial,
viral, protozoan, and fungal proteases trigger inflammation by
activating the intrinsic coagulation pathway [102], or act as pro-
coagulants by non-canonical, direct activation of prothrombin
[179]. Large panels of small molecule inhibitors of the proteasome
360 Ingrid M. Verhamme et al.

in pathogenic organisms are currently being screened for potential

therapeutic benefit and minimal toxicity toward the cellular
machinery of the host [180]. Bacterial infections are associated
with increased thrombotic risk; however, this correlation is not
restricted to pathogenic bacteria. Subtilisin, produced by the
non-virulent bacterium Bacillus subtilis, can cleave prothrombin
to an active thrombin-like species that converts fibrinogen to fibrin
[181]. Deregulation of the intestinal microbiota is typical in IBS,
and excess protease production by commensal enteric bacteria has
been proposed to promote adherence to and invasion of intestinal
epithelial cells, activate protease-activated receptors (PARs), disrupt
the intestinal barrier, and facilitate bacterial interaction with
immune cells, leading to inflammation [182] (Table 14).
Pathogens may use mechanisms other than direct proteolytic
activity to enhance their virulence or facilitate dissemination and
propagation. The streptococcal and staphylococcal cofactors strep-
tokinase (SK) and staphylocoagulase (SC) are not enzymes them-
selves, and respectively bind and activate host plasminogen and
prothrombin in a non-proteolytic fashion by inserting their
N-termini into the zymogen activation pocket. This triggers a
conformational change that forms the active site in the zymogen
[101, 183]. The cofactor complexes with the zymogens as well as
the active proteases are very tight, and refractive to endogenous
serpins that inactivate plasmin and thrombin, thus adding to the
bacterial virulence. The prothrombinlSC and thrombinlSC com-
plexes cleave host fibrinogen to form fibrin barriers, protecting the
pathogen from the host immune system. Upon activation of free
host plasminogen to plasmin by the plasminogenlSK complex
[183], the tighter binding plasminlSK complex is formed, and
degrades the host extracellular matrix to facilitate pathogen inva-
sion and dissemination. Numerous streptococcal strains also
increase their invasiveness by recruiting host plasminogen and plas-
min to bacterial cell wall M-proteins [184]. Von Willebrand factor-
binding protein (VWbp) is another conformational prothrombin
activator secreted by S. aureus, and belongs to the family of staphy-
lococcal and streptococcal homologs named zymogen activator and
adhesion proteins (ZAAPs), based on the SC structure [185]. Sta-
phylokinase (SAK) bears no sequence similarity to SK, but shares a
similar domain fold. It does not activate plasminogen conforma-
tionally, but forms a tight plasminlSAK complex that cleaves plas-
minogen as a substrate [186]. The skizzle (SkzL) protein, secreted
by Streptococcus agalactiae, has moderate sequence identity to SK
and SAK [187]. SkzL binds host plasminogen, and enhances its
activation by the plasminogen activators uPA and single chain tPA,
and plasma clot lysis by these plasminogen activators. S. agalactiae
pathogenesis likely includes SkzL to enhance bacterial spreading
through fibrinolytic enhancement. These are prime examples of
Table 14
Proteases, inhibitors and cofactors in infectious organisms: activities and Uniprot ID codes (where
applicable) of proteases, protease inhibitors, and cofactors associated with infectious organisms

Protease/inhibitor Uniprot Action

HIV protease (HIV-1) P04585 Cleaves viral polyprotein into individual proteins,
including itself
Critical for viral replication, prominent drug target
Trypanosoma cruzi cruzipain P25779 cysteine protease expressed by Trypanosoma cruzi,
vital to the parasitic protozoa’s life cycle
Porphyromonas gingivalis P28784 Bacterial thiol proteases
gingipains Degrade host tissue proteins and cytokines
Bacillus anthracis lethal factor Anthrax protein which degrades mitogen-activated
protein kinase kinase, disrupting function of
mitogen-activated protein kinases (MAPKs)
Subtilisin Nonspecific bacterial protease known to activate
thrombin
Staphylocoagulase (1 and 2) P07767 and Activates prothrombin through binding, not a
P17855 protease
Streptokinase (SK) Activates plasmin
Von Willebrand factor-binding A0A1D4Z3F9 Staphylococcus protein promotes clot formation
protein
Staphylokinase (SAK) P68802 Staphylococcus protein, plasmin-SAK complex
activates plasminogen
skizzle (SkzL) Q8DZH4 Streptococcus protein, enhances activation of
plasminogen by uPA and sc-tPA
ADAMTS7 Q9UKP4 Degrades cartilage oligomeric matrix protein
(COMP)
Implicated in cancer, arthritis, and coronary artery
disease
Required for influenza A virus replication
Carboxypeptidase E P16870 Cleaves C-terminal arginine of lysine residue
Processes most neuropeptides and peptide
hormones
Required for influenza A virus replication
Dipeptidyl peptidase 3 Q9NY33 Degrades angiotensin, Leu-enkephalin, and
Met-enkephalin
Implicated in ovarian cancer
Required for influenza A virus replication
Macrophage stimulating P26927 Unknown, sequence homology with hepatocyte
1 protease (macrophage growth factor
stimulating protein) Required for influenza A virus replication
Neurotrypsin PRSS12 Cleaves agrin
Multidomain serine protease expressed in the
nervous system
Required for influenza A virus replication
Ubiquitin-specific protease 14 P54578 Proteasome-associated deubiquitinase, prevents
ubiquitin digestion
Prevents degradation of prion protein
362 Ingrid M. Verhamme et al.

pathogens exerting virulence by hijacking the host coagulation and

fibrinolytic systems [188].
The human host proteases ADAM-TS7, carboxypeptidase E,
dipeptidyl peptidase 3, macrophage stimulating 1 protease, and
neurotrypsin are required for influenza A virus replication, and
are under control of eight host miRNAs regulating gene expression
during virus replication [189]. These host genes and microRNAs
may provide new therapeutic targets. The ubiquitin-specific prote-
ase 14 (USP14), a deubiquitinating enzyme, prevents degradation
of prion protein by rescuing it from the proteasome, and may be a
suitable target in the development of therapeutic strategies for
prion diseases [190].
Porphyromonas gingivalis is prevalent in periodontitis, a risk
factor for oral and gastric tract tumors, and also lung cancer, as
recently identified in a follow-up of the Atherosclerosis Risk in
Communities (ARIC) study [191]. P. gingivalis gingipains are
cysteine proteases associated with this type of chronic inflamma-
tion, and they are the only bacterial proteases that degrade
SPINK6, a Kazal-type inhibitor of various human kallikreins in
skin and oral epithelium. Loss of this proteolytic control has been
suggested as a link between periodontal disease and tumor
development [177].

4 Proteolysis-Related Processes as Drug Targets

4.1 Overexpressed or As of 2010, an estimated 5–10% of all drugs under development

Impaired Endogenous were targeted toward proteases [192], many of them small mole-
Proteolytic Activity cules designed to block the protease active site. Among past and
present commercially successful protease inhibitors are blood pres-
sure regulators (e.g., captopril and aliskiren) which respectively
inhibit the metalloprotease angiotensin-converting enzyme
(ACE), and the aspartic protease renin, by competitive binding to
the protease active site; dipeptidyl peptidase-4 inhibitors (e.g.,
sitagliptin) to combat type 2 diabetes; the threonine protease
inhibitor bortezomib as a cancer drug directed against the protea-
some; the direct oral anticoagulants (DOACs), thrombin and factor
Xa inhibitors (argatroban, dabigatran, apixaban, rivaroxaban, edox-
aban) that bind tightly and reversibly to the protease active site;
tight-binding hirudin-based thrombin inhibitors (lepirudin, desir-
udin, bivalirudin) for patients with heparin sensitivity.
Several endogenous protease, cofactor, and inhibitor deficien-
cies are treated by augmentation therapy. Hemophilia A is the
deficiency of factor VIII, the essential cofactor of factor IXa to
activate factor X; and hemophilia B patients lack functional factor
IX. Both deficiencies prevent the formation of the intrinsic Xase
complex that is responsible for generation of the majority of active
factor Xa, with as end-result the impairment of clot formation.
 Proteases: Pivot Points in Functional Proteomics 363

Intravenous replacement with plasma-derived or recombinant

fVIII and factor IX requires frequent injections, although prepara-
tions with longer half-life are being developed. Gene therapy for
hemophilia B, based on in vivo gene transfer with adeno-associated
viral (AAV) vectors to the liver has been in clinical trials for 16 years,
with partial success due to cellular immune responses [193]. How-
ever, as of December 2017 the results of two small cohort studies
are promising: 52 weeks after infusion of a single intravenous dose
of an AAV5 vector encoding factor VIII, no cellular immune
response, liver toxicity, or inhibitory antibodies were observed in
a clinical trial for treatment of severe hemophilia A [194]; and in a
small-scale hemophilia B patient study, a high level of expression of
functional factor IX was seen after a single injection of an AAV
vector containing the hyperfunctional factor IX Padua gene, target-
ing the liver [195]. Deficiency of factor XI, also known as
hemophilia C, is a rare bleeding disorder, often seen in Ashkenazi
Jewish populations, and does not cause bleeding in the joints.
Tranexamic acid is administered to control traumatic bleeding
incidents and during dental procedures, whereas fresh frozen
plasma or recombinant factor XI may be used during surgery.
In sepsis, the systemic response by the host to pathogenic
invasion triggers activation of inflammatory and coagulation path-
ways and inhibition of fibrinolysis. In this regard, administration of
recombinant human-activated protein C (drotrecogin alpha acti-
vated, DAA) as an anticoagulant was deemed a useful strategy, and
in 2001 it became the first biologic approved for treatment of
severe sepsis [196]. Although a first trial indicated reduction in
mortality, later trials failed to confirm these findings, and DAA
was withdrawn in 2011. Observational trials consistently showed
a benefit while randomized trials did not. The difficulties associated
with obtaining reproducibility in these trials may be attributed to a
variety of reasons: differences in acute illness of patient subgroups,
perhaps as a result of conscious or subconscious patient selection;
midway amendment through the first trial, changing inclusion/
exclusion criteria, the type of placebo and the drug formulation,
the combination of which favored the use of DAA and led to early
termination; and differences in the timing and appropriateness of
antibiotic administration and fluid resuscitation. New drug devel-
opment for targeting severe sepsis will undoubtedly benefit from
targeting pathophysiologic pathways characterized by specific bio-
markers rather than heterogeneous patient populations grouped by
clinical phenotypes, and DAA may yet be found beneficial for well-
defined target groups.
Dysfunctional or poorly expressed serpins, inhibitors of serine
proteases, cause a variety of severe diseases. COPD, emphysema,
cystic fibrosis, liver disease and panniculitis due to functional α1-PI
deficiency and accumulation of inhibitor polymers are alleviated by
intravenous administration of plasma-derived α1-PI. Experimental
364 Ingrid M. Verhamme et al.

approaches include aerosolized formulations of plasma or recombi-

nant inhibitor, and direct delivery to the lung is expected to cir-
cumvent short half-life issues plaguing the intravenous
formulation. However, no clinical trial reports are available to
date. Intravenous recombinant α1-PI formulations are in the exper-
imental stage, and conjugation with polyethylene glycol may delay
rapid renal clearance [89]. It has been recognized that α1-PI inhi-
bits proteases other than elastase and trypsin, namely proteinase-3,
kallikreins 7 and 14, matriptase, caspase-3, and the metallopepti-
dase ADAM17 [197]. This opens up new avenues for modulating
the activities of these proteases in disease states. Recombinant
human antithrombin (Atryn) is purified from the milk of transgenic
goats, and used to avoid perioperative and peripartum clotting
complications in patients with hereditary antithrombin deficiency.
It is not indicated for treatment of thromboembolic events in
these patients. Its glycosylation profile differs from that of plasma-
derived antithrombin, with increased heparin affinity as a result.
The modification ensures efficient inhibition of elevated thrombin
and factor Xa.
C1-inhibitor is a serpin targeting C1 esterase of the comple-
ment system, and it is also the physiological inhibitor of kallikrein,
and factors XIIa and XIa of the contact activation pathway of
coagulation. Both inherited and acquired C1-inhibitor deficiency
can lead to life-threatening angioedema [198]. Inherited, hetero-
zygous deficiency results in lack of transcription, translation or
secretion, or in expression of a mutated, dysfunctional inhibitor.
Acquired deficiency is the result of inhibitor depletion due to
autoantibody formation or accelerated consumption in lymphopro-
liferative diseases. Elevated kallikrein activity causes unregulated
cleavage of high molecular weight kininogen and release of brady-
kinin, the mediator of angioedema. Acute attacks of angioedema
are treated with C1-inhibitor concentrate from plasma, recombi-
nant inhibitor, and the kallikrein inhibitor ecallantide. Prophylactic
treatment with the antifibrinolytic agents ε-aminocaproic acid and
tranexamic acid regulate the fibrinolytic system which is continu-
ously activated in autoimmune angioedema. Aprotinin (Trasylol),
or bovine pancreatic trypsin inhibitor, is a Kunitz-type inhibitor of
kallikrein and plasmin. It was used to treat laryngeal edema until its
temporary withdrawal from the market in 2007 due to reports of
increased death risk as a bleeding preventative during cardiac sur-
gery. As aprotinin was derived from bovine lung tissue, concerns for
allergic reactions and bovine spongiform encephalopathy prompted
its discontinuation in Italy. In 2012 the European Medicines
Agency proposed to lift the ban, and aprotinin is currently mar-
keted by Nordic.
The proteases of the complement system have increasingly
been recognized as potentially attractive points of interference for
mitigation of inflammatory diseases. Recent developments in
 Proteases: Pivot Points in Functional Proteomics 365

complement therapeutics focus on the proteases of the initiation

pathways, with C1-inhibitor targeting C1r/s and mannan-binding
lectin serine protease (MASP) (e.g., Cinryze, Berinert, Cetor,
Ruconest), and antibodies targeting C1q, C1s, C2, MASP-2 and
-3 (ANX-005, TNT009, OMS721, CLG561, NM9401) [199].

4.2 Protease Undesired properties or side effects of many of these therapies

Inhibitors as Drugs: underscore the need for continued mechanism-based drug design.
Some Caveats In 2014 the renin inhibitor aliskiren was placed on the list of drugs
to avoid, due to severe side effects in patients with diabetes and
kidney impairment [200]. Hirudin derivatives are an attractive
alternative for treatment of patients with heparin hypersensitivity
or thrombocytopenia; however, they have a short half-life. Hiru-
dins are cleared through the kidneys, and dose adjustment is
required in patients with renal impairment. Some small molecule
drugs have limited bioavailability and solubility, and their efficiency
may be mitigated by resistance mutations in the target proteases.
Fast-acting DOACs are at least as effective as warfarin, with reduced
risk for intracranial bleeding, and are prescribed for stroke preven-
tion in atrial fibrillation, thromboprophylaxis in hip or knee replace-
ment surgery, and for treatment and secondary prevention of
venous thromboembolic disease [201]. Unlike with vitamin K
antagonists, no routine monitoring of coagulation is needed. A
recent disadvantage was the lack of antidotes for direct factor Xa
inhibitors in case of traumatic bleeding. Activated prothrombin
complex concentrates and recombinant activated factor VIIa have
been proposed to reverse DOAC action; the monoclonal antibody
idarucizumab was approved by the FDA in 2015 as a dabigatran
antidote; and andexanet alfa, a factor Xa decoy, in May 2018 to
counteract apixaban and rivaroxaban.
MMP-dependent degradation of extracellular matrix proteins
is associated with angiogenesis and metastasis in cancer, and MMP
inhibitors were proposed as suitable anticancer drugs. The zinc ion
in MMPs was the first target, but small molecule peptidomimetic
inhibitors based on zinc-targeting warheads (e.g., batimastat) had
limited selectivity, failed to distinguish between different MMP
classes also involved in the Notch-, Wnt-, and NFκB-signaling
pathway, and were fraught with off-target side effects
[202, 203]. A novel class of small molecules blocking the hydro-
phobic S10 specificity pocket, exosites, and other MMP domains
yielded reasonably specific inhibitors for several MMPs; however,
their efficacy as second generation drugs has not yet been demon-
strated. None of the 50 completed clinical trials with MMP inhibi-
tors were successful, due to off-target toxicity or absence of efficacy.
A recent study reports effective allosteric prevention of pro-MMP-9
activation in a mouse neuroinflammatory model by a small, highly
selective heterocyclic chemical inhibitor [204]. This orally adminis-
tered compound does not prevent activation of the structurally
366 Ingrid M. Verhamme et al.

related pro-MMP-2, and does not inhibit catalytically active MMP-

1, -2, -3, -9, or -14. Although these findings are encouraging for
future drug development, the efficacy of this compound needs to
be tested in other models, relevant to cancer, fibrosis, and
neurodegeneration.
Indiscriminate targeting of overexpressed MMPs in certain
cancers may not always yield desired results [205, 206]. Pancreatic
ductal adenocarcinoma cells (PDAC) overexpress MMP-9, thought
to play a role in invasion and metastasis. However, systemic knock-
out of MMP-9 in a PDAC mouse model caused increased
interleukin-6 (IL-6) expression, and induced invasive growth and
STAT3 activation in PDAC cells via IL-6 receptor signaling. The
model system, animal genetic background and other experimental
conditions may influence the effect of MMP activity, resulting in
tumor-promoting, -inhibitory or null effect, as seen in various
mouse models of breast cancer. In the light of many controversial
experimental results, systemic MMP inhibitors should be used with
caution, and information obtained from clinical studies associating
specific MMPs with diseases profiles will be critical for identifying
the proper MMPs as therapeutic targets. The metalloprotease
ADAM10 has been associated with various disease states
[159]. In the brain, it cleaves APP via a non-amyloidogenic path-
way, with formation of the neuroprotective soluble ectodomain,
and decrease of the toxic Aβ fragment. It may also slow down
progression of fibrosis in chronic liver inflammation. However, it
acts as a sheddase for the cellular prion protein, potentially promot-
ing spreading, and increased activity is seen in synaptic dysfunction
linked to Huntington’s disease. Respectively, ADAM10 inhibition
would be therapeutic or detrimental in these pathologies.
ADAM10 is upregulated in various cancers, atherosclerosis, and
various autoimmune diseases, suggesting a potential benefit of
inhibition. The broad substrate specificity of ADAM10 and its
similarity with ADAM17, its systemic presence, and its divergent
effects in various diseases pose significant problems in targeting
ADAM10, either for upregulation or inhibition. Ideally, drugs
would be needed that regulate ADAM10 activity in a tissue and
substrate specific manner. Focusing on the interaction of ADAM10
with specific substrates or regulatory partner proteins may provide
some promise.
Efforts to develop γ-secretase inhibitors for treating Alzhei-
mer’s disease turned out to miss the mark, mainly due to the fact
that the mechanism of APP processing by γ-secretase was not fully
understood. Mutations in its presenilin domain were originally
interpreted to enhance γ-secretase activity; however, the
γ-secretase inhibitor Semagacestat showed worsening of familial
Alzheimer’s patients in a phase III clinical trial that was halted in
2010. In addition to issues with the design and dosage, this result
was in part also explained by a recent in vivo study showing that the
 Proteases: Pivot Points in Functional Proteomics 367

presenilin-1 mutations inactivate rather than enhance γ-secretase

activity, impair hippocampal memory and synaptic function, and
cause neurodegeneration [207]. γ-Secretase inhibitors have since
then been repurposed as potential cancer therapeutics due to their
inhibition of the Notch signaling pathway, upregulated in many
cancers. However the panel of known inhibitors shows a wide range
of activities toward cleavage of various other γ-secretase substrates,
and off-target interference is likely to cause major side effects, thus
limiting the long-term clinical usefulness of these inhibitors [208].
Osteoclasts express the cysteine protease cathepsin K, which
degrades type I collagen in bone. Selective inhibition of cathepsin K
increases bone mass, improves bone strength, decreases bone
resorption, and contributes to bone formation [209]. The small
molecule inhibitors relacatib, balicatib, and odanacatib were initi-
ally tested clinically as potential drugs for treatment of postmeno-
pausal osteoporosis, with odanacatib ultimately making it through
Phase II and III trials. The phase III trial was halted early after
reports of positive efficacy and safety; however, a more thorough
analysis discovered an increased risk of atrial fibrillation and stroke
[210]. Odanacatib development was discontinued in 2016, after
more than 12 years of research.
In the treatment of ischemic stroke, an active protease, rather
than an inhibitor, is used as a therapeutic. Tissue-type plasminogen
activator (tPA) was approved in 1996 for clot dissolution in ische-
mic stroke; however, its use is restricted to the first 3 h poststroke,
and carries an established risk for bleeding. One study attributed
the increased blood-brain barrier permeability to tPA-catalyzed
activation of platelet-derived growth factor-CC, with Mac-1 integ-
rin and LRP1 acting as cofactors in this reaction [211]. Another
study suggested increased MMP-9 activity as a potential cause
for increased bleeding risk [212], while yet another report
pointed toward a connection of tPA-induced bleeding with
hyperglycemia [213].
These are but a few examples of disappointing outcomes, or
interference of significant off-target and other side effects in
protease-related drug development. A recurring theme is that of
insufficient knowledge of the underlying biochemical mechanisms
and of the interconnectedness of protease and inhibitor activity in
the proteome network. With the increasing availability of large
online platforms and databases such as the National Center for
Biotechnology Information, RCSB Protein Databank, UniProt,
and MEROPS, discovery of such interconnectedness should
prove increasingly less challenging.

4.3 Targeting As antibiotic-resistant infections are on the rise, targeting proteases

Exogenous Proteolytic and protease-related processes in pathogenic bacteria may offer
Activity novel avenues for drug development. Proteases and protease cofac-
tors that function as virulence factors are obviously first choice
368 Ingrid M. Verhamme et al.

targets. Progress has been made in developing small molecule

compounds that inhibit the expression of SK in Group A strepto-
cocci [214]. In addition, the proteolytic complexes Lon, ClpXP,
HtrA, the proteasome, and signal peptidases are good candidates
for disruption of bacterial mechanisms necessary for survival and
pathogenicity [178]. Caseinolytic proteases (ClpPs) are conserved
multimeric complexes that are conformationally activated by bind-
ing of Hsp100 ATPases. This binding aligns the catalytic triad of
the proteases, and energy provided by ATP is used to unfold
protein substrates for entry into the pore and subsequent degrada-
tion in an energy-dependent manner. Several mechanisms for ClpP
deregulation have been proposed: (a) inhibitors such as phenyl
esters and β-lactones can directly interact with the catalytic residues
and halt protein degradation; (b) blocking of ATPase binding and
uncoupling of ClpP and Hsp100 ATPase activity by acyldepsipep-
tides (ADEPs) and macrocyclic peptides may result in continuous
ClpP activation and promiscuous protein degradation. However,
many of these compounds are not 100% efficient due to limitations
of stability, solubility, resorption, and half-life. Small molecule inhi-
bitors have limited specificity and may not distinguish between
bacterial protease complexes and their human orthologs. Recently
developed ADEP derivatives are active against methicillin-, vanco-
mycin-, and penicillin-resistant pathogens, and work well in combi-
nation therapy. The natural oligopeptide compounds cyclomarin A,
ecumicin, and lassomycin target Mycobacterium tuberculosis but do
not kill commensal members of the human microbiota. Encourag-
ing in vitro and in vivo results have also been reported for small
molecule inhibitors of B. anthracis lethal factor, and the cysteine
protease domains of Clostridium difficile toxins. Specific interfer-
ence with proteases required for bacterial viability offers an attrac-
tive possibility for developing a novel class of therapeutics, less
prone to trigger resistance.
Parasitic cysteine proteases may be attractive targets for treat-
ment of Chagas disease, African sleeping sickness, and leishmania-
sis. Recent progress was made in the design of more efficient
peptide inhibitors of cruzipain, by modifying the electrophilic war-
head group that forms a covalent bond with the catalytic cysteine
[215]. Classical nitrile group warheads are metabolically stable,
polar, and small, but proved to be less potent than oxime and
aldehyde warheads. Substitutions at the P1 and P3 residues alter
the inhibitory potency and provide a means of modulating specific-
ity. Ideally, a successful drug specifically recognizes the parasitic
proteases over the host caspases, calpains, and cathepsins by form-
ing specific non-covalent interactions adjacent to the active site, but
obtaining this level of selectivity is challenging. Covalent binding is
often irreversible, and one drawback of permanent attachment to
peptide fragments after protein degradation is the immunogenicity
of these fragments.
 Proteases: Pivot Points in Functional Proteomics 369

Currently ten FDA-approved HIV protease inhibitors are

approved, and nine available, with structures that were thought to
mimic the substrate transition state [216]. HIV protease inhibitors
show off-target interference with proteases required for maturation
of SREBP-1, a transcription factor that regulates gene expression in
lipogenesis, with as result lipodystrophy syndrome; and blocking of
glucose transporter-4 with as result insulin resistance; inhibition of
the proteasome, resulting in metabolic complications, increased ER
stress and autophagy; and caspase-dependent apoptosis, the discov-
ery of which triggered interest in HIV protease inhibitors as poten-
tial anticancer drugs. The emergence of HIV-1 strains that are
resistant to the current protease inhibitor drugs prompted the
design of novel compounds with broad-spectrum activity against
these variants [217, 218]. Small non-peptide molecules with sub-
stituted pyrrolidines, piperidines, and thiazolidines as P2-P3
ligands for binding to the S2-S3 specificity site, and flexible macro-
cyclic P1’-P2’ tethers were good candidates, with inhibition con-
stants (Ki) and IC50 values in the nanomolar range. Incorporation
of heteroatoms in the macrocyclic skeleton yielded inhibitors with
picomolar Ki and nanomolar IC50 antiviral activity. Biological
evaluation, structure-activity relationships, and X-ray studies of
the protease-inhibitor complexes validated the design approach,
illustrating the power of structure-based molecular design. Peptide
ketoamide-based NS3/4A serine protease inhibitors (boceprevir
and telaprevir) are used for treating genotype 1 hepatitis C
(HCV) [219]. In August 2017 Abbvie released Mavyret, a combi-
nation drug of glecaprevir/pibrentasvir targeting NS3/4A serine
protease activity and the NS5A replication complex, and suitable
for treatment of all HCV genotypes. However, complications may
arise in HCV/hepatitis B (HBV) co-infected patients who com-
pleted treatment with HCV direct-acting antivirals and were not
receiving HBV antiviral therapy. Fulminant hepatitis, hepatic fail-
ure, hepatitis flare, HBV reactivation, and death have been
reported.

5 The Future of Proteolysis-Related Drug and Diagnostic Development

5.1 Active Site Active site targeting is a main component in many established
Targeting, Exosite, and approaches of drug development for controlling protease activity.
Effector-Binding Sites However, small molecule inhibitors that are limited to interference
with the conserved catalytic machinery of an entire class of pro-
teases may have severe limitations due to their broad-spectrum
activity, resulting in off-target protease inhibition. This was illu-
strated by the failure of many small molecule and zinc-targeting
MMP inhibitors in clinical trials. Early irreversible inhibitors used
active site targeting of nucleophilic proteases with electrophilic
alkylating agents such as diazo- or halo-ketone warheads; however,
370 Ingrid M. Verhamme et al.

the necessity of attaching a sizeable peptide to the warhead for

specific protease binding proved impractical as large libraries of
peptide warheads were required to identify effective inhibitors.
The development of sitagliptin, an inhibitor of dipeptidyl peptidase
4 (DPP4) in type 2 diabetes, is a success story after a few consecu-
tive setbacks [220]. DPP-4 inhibitors threo- and allo-isoleucyl
thiazolidide initially showed significant animal toxicity due to
off-target reaction with DPP8 and DPP9. Alpha amino compounds
related to isoleucyl thiazolidide proved nonselective, and structure-
activity screening identified a highly selective and rapidly metabo-
lized beta-amino acid piperazine series. Bicyclic derivatization
resulted in triazolopiperazine compounds that had suitable preclin-
ical pharmacokinetic properties. Optimization led to the discovery
of the highly selective sitagliptin.
Pitfalls in inhibitor design based on substrate interaction with
the active site are illustrated by the development of transition state
analogs against HIV protease. Enzyme transition states are very
short-lived, on the femto- to picosecond timescale, but binding of
transition state analogs converts these to a stable thermodynamic
state. Kinetic isotope effects and computational chemistry identify
which chemical steps are involved in transition state binding. Typi-
cally, these analogs can bind up to millions of times tighter than
substrates, making them attractive compounds in drug develop-
ment [221]. The HIV-1 protease–substrate complex has three
transition states with partial bonds in the reaction coordinate, and
two intermediates with equilibrated bonds. The high-energy inter-
mediates bind tightly to the protease, as do inhibitors designed as
mimics of these intermediates. Ten FDA-approved competitive
HIV-1 protease inhibitors, with Saquinavir as first and prototypical
drug, were originally considered transition state analogs because
they have an sp3 center to mimic the geometry of the transition
state, but they were later actually found to be intermediate mimics.
Ile84Val and Leu90Met and several other mutations in HIV prote-
ase were symptoms of emerging drug resistance toward these inhi-
bitors [222], prompting investigators to take a closer look at the
protease transition states. The crystal structures of the transition
states of both native and protease-inhibitor-resistant HIV-1 pro-
teases showed that they are chemically and structurally identical,
which means that resistance is due to changes outside the true
transition state [223]. Mimicking specific chemical features of the
true transition state may solve this resistance problem.
The active site of proteases forms a groove that accommodates
several substrate residues adjacent to the scissile P1-P1’ bond. Sub-
strates exhibit complementarity with the protease residues S4-S3-
S2-S1-S1’-S2’-S3’-S4’ in the binding site, causing a favorable binding
interaction. The architecture of the S1 or specificity pocket often
defines the nature of substrate cleavage, e.g., hydrophobic and
aromatic P1 substrate residues for chymotrypsin-like proteases,
 Proteases: Pivot Points in Functional Proteomics 371

basic P1 residues for trypsin-like proteases, small aliphatic P1 resi-

dues for elastase-like proteases, and hydrophobic bonds for aspar-
tate proteases. Cysteine proteases prefer bulky nonpolar residues at
the P2 position. The MMP substrate specificity is more involved,
with the S1’ pocket selectively accommodating the substrate residue
immediately after the scissile bond [224]. MMP-1 and -7 have
small S1’ pockets preferring small hydrophobic residues, whereas
MMP-2, -3, -8, -9, and -13 have large pockets and bind a diverse
array of amino acids. Additional allosteric, exosite, and effector-
binding site interactions are expected to contribute significantly to
exclusive selection of the target protease. Structural conservation of
the specificity site throughout a protease family with diverse cata-
lytic properties and biological targets may pose a problem in
designing specific drugs, and a promising alternative was developed
by targeting zymogen activation rather than the active protease
[204]. A highly selective compound that allosterically inhibits
MMP-9 activation by binding to a pocket near the zymogen cleav-
age site may be a first viable drug candidate.
Identification of protease subsite preferences can be achieved
by positional scanning to identify the best fit. Peptides spanning the
active site cleft carry a fluorophore and an internal quencher, and
preferential cleavage of peptide libraries is determined from fluo-
rescence yields upon proteolytic removal of the quencher [192].
High-throughput screening (HTS) and fragment-based screening
do not require previous knowledge of substrate specificity and may
yield rapid results, but need appropriate filtering by functional
activity to eliminate nonselective reactions. Fragment-based screen-
ing using NMR, mass spectrometry, or differential scanning fluori-
metry identifies moderate to weak binders that can be optimized
into more potent inhibitors. The use of X-ray crystallographic
structural information coupled with in silico drug design is getting
a boost from rapidly developing high-throughput X-ray crystalliza-
tion and structure determination, using TRAP screens with the
most successful crystallization conditions [225]. High-throughput
crystallographic screening of brominated fragment libraries, based
on anomalous scattering to localize bromine, successfully identified
targets for HIV protease, and detected novel binding sites in the
surface-exposed active site glycine-rich β-hairpin flap region and the
exosite region [226].
The importance of exosite interactions in protease inhibition is
illustrated by the thrombin inhibitor, hirudin, a polypeptide pro-
duced by the salivary glands of medicinal leeches such as Hirudo
medicinalis. Hematophagous animals have a need for a natural
anticoagulant to prevent their food from clotting. The crystal
structure of the hirudin–thrombin complex shows a globular
N-terminal domain making contact with the active site, and a
17-residue extended C-terminal chain wrapping around thrombin
exosite I, the anion-binding site (ABE I) that binds the thrombin
372 Ingrid M. Verhamme et al.

substrate, fibrinogen [227]. For this reason hirudins are sometimes

referred to as bivalent direct thrombin inhibitors [228]. This dual
interaction confers tight binding, and the classification as a slow
tight-binding inhibitor indicates that the off-rate for inhibitor dis-
sociation is extremely slow. Several recombinant hirudins are on the
market. They have a short half-life, but may be preferred to heparin
as an anticoagulant in patients with heparin-induced thrombocyto-
penia. Thrombin exosites also figure prominently in its irreversible
inactivation by the serpins antithrombin and heparin cofactor II
(HCII). Thrombin ABE I binds fibrinogen, and exosite II (ABE II)
binds heparin. Because heparin accelerates the irreversible inactiva-
tion of thrombin by endogenous serpins, it is sometimes called an
indirect anticoagulant. A long heparin template binds both throm-
bin via ABE II and antithrombin or HCII via their heparin-binding
site. This approximation causes a dramatic increase in the thrombin
inactivation rate. HCII provides an extra interaction by binding of
its N-terminus to thrombin ABE I. Fragment screening against
caspase-7 identified two small molecule noncompetitive inhibitors
with potential for drug development [229]. X-ray crystallography
showed allosteric binding at the caspase dimer interface, more than
17 Å removed from the active site. This recent finding illustrates
that allosteric control is yet another approach toward drug
development.
Lysosomal cysteine cathepsins are required for normal lipid
metabolism and cholesterol homeostasis, proper function of mito-
chondria, and clearance of apoptotic cells [230]. A deregulated
lipid metabolism, vascular inflammation, arterial remodeling, neo-
vascularization, autophagy, and necrosis are all hallmarks of athero-
sclerosis, and these processes are linked to upregulated cysteine
cathepsin activity. Because vascular lesions in atherosclerotic
patients may remain silent until the disease is well advanced, suit-
able biomarkers for these dysregulated pathways would be useful.
Detection of cysteine cathepsins in macrophages allows distinguish-
ing between stable and unstable lesions in excised carotid plaques,
and this information from the plaque microenvironment may be
harnessed in the development of molecular imaging. With this
noninvasive technique, localized cleavage or covalent retention of
specific protease substrates and inhibitors carrying fluorophores,
reporter groups, or contrast agents can be detected. Substrate-
based probes change their spectral properties upon cleavage, and
the commercial self-quenched poly-lysine cathepsin probe Prosense
has shown usefulness in preclinical cardiovascular imaging studies
for detection of vascular inflammation, macrophage concentration,
and cathepsin activity. Newer, lipidated cathepsin substrates show
promise due to their improved homing properties. Quenched
activity-based probes (qABPs) for cysteine cathepsin activity have
the quencher QSY21 attached to a near-infrared fluorophore
Cy5-labeled acyloxymethyl-ketone analog. The cell-permeable
 Proteases: Pivot Points in Functional Proteomics 373

probe covalently modifies the target cathepsin, resulting in loss of

the quenching group, and formation of a fluorescently labeled
target protease [231]. This type of probe has a low nonspecific
fluorescent background and has been successfully applied for
detecting cysteine cathepsin activity in tumor cells, and in preclini-
cal models of atherosclerosis. Reduced light penetrance is a major
issue, necessitating further development of multimodal ABPs for
PET-CT application [232]. This imaging approach has already
been used successfully to localize idiopathic pulmonary fibrosis in
patients, and may be adapted to detect rupture-prone atheroscle-
rotic lesions.

5.2 Indirect, Cytokines and chemokines are activated by uncontrolled protease

Mechanism-Based action in many immune pathologies, and these proteins and their
Targeting receptors may be potential therapeutic targets themselves. Cytokine
storms typically occur as a reaction of a healthy immune system
against new, highly pathogenic invaders, and are thought to be
responsible for many deaths in influenza pandemics. Cleavage of
the influenza hemagglutinin by host trypsin-like proteases is
required for infectivity, and IL-1β was identified as the major cyto-
kine that upregulates host trypsin expression and triggers formation
of more IL-1β [233]. An anti- IL-1β antibody successfully sup-
pressed upregulation of pro-inflammatory cytokines and trypsin in
a mouse model, and antibodies against IL-1β and its receptor have
been proposed as potential therapeutics. Drug targeting of non-
protease components of the complement network involve the acti-
vation and amplification pathways, with peptides (AMY-101,
APL-1 and -2), proteases (CB2782) and protein inhibitors
(AMY-201, Mirococept) targeting C3 and C3b; and inhibitors of
the terminal pathways, with antibodies (eculizumab), a tick saliva
protein (Coversin) and aptamers (Zimura), binding to C5 and
preventing C5 convertase activity [199]. The complement compo-
nent C5a, generated in the terminal pathway of complement acti-
vation, is a chemokine, and attracts leukocytes to the inflammatory
focus. Patients with paroxysmal nocturnal hemoglobinuria and
atypical hemolytic uremic syndrome exhibit uncontrolled comple-
ment activation, and the recombinant tick protein prevents release
of C5a and formation of C5b–9, the membrane attack complex
(MAC). The natural function of the saliva protein in Ornithodoros
moubata is suppression of the host immune response when the tick
is feeding. The therapeutic protein is also effective in patients with
C5 polymorphisms who are resistant to eculizumab.
Protease-activated receptors have long been considered poten-
tial drug targets, and the therapeutic applications of PAR-1 and
PAR-4 antagonists are discussed in a previous section. In preclinical
models, a PAR-2 antagonist inhibited tumor growth and the for-
mation of new blood vessels in cancer, and inflammation in rheu-
matoid arthritis and acute inflammation models, perhaps indicating
374 Ingrid M. Verhamme et al.

a potential link between inflammation and cancer [234, 235]. Stra-

tegies for developing suitable PAR antagonists include modified
peptidomimetics such as trans-cinnamoyl-YPGKF-NH2 that bind
but do not activate the receptor; low molecular weight heterocyclic
structures, e.g., 1-benzyl-3-(ethoxycarbonylphenyl)-indazole;
N-terminal palmitate-modified oligopeptides (pepducins) that
anchor the peptide to the cell membrane; and specific function-
blocking monoclonal antibodies [236]. The majority of these
efforts is ongoing, and focused on PAR-4 antagonists as novel
antiplatelet agents.

5.3 Drugging the Designing substrates, inhibitors, and activity-based probes with
Undruggable: A specificity toward a single protease is a major challenge when the
Familiar Theme, with protease shares a similar catalytic mechanism and substrate specific-
Some Fancy Targeting ity with other proteases in the same family, but is functionally
completely different [237]. The main approaches utilized to date
for optimizing complementarity with the protease specificity sites
include positional scanning synthetic combinatorial libraries with
coumarin-derived reporters (PS-SCL) [238], phage display, hybrid
combinatorial substrate libraries using unnatural amino acids that
allow more thorough scanning of the active site (HyCoSuL),
counter selection substrate libraries (CoSeSuL), internally
quenched fluorescent substrate or fluorescence resonance energy
transfer libraries (IQF or FRET), proteomics and exopeptidase
fingerprinting. HyCoSuL screening of the P3 and P4 positions,
respectively, with methionine sulfone and 2-amino-6-benzyloxy-
hexanoic acid allowed distinguishing between neutrophil elastase
and proteinase-3 for the first time, and tracking neutrophil elastase
activity in neutrophil traps [239]. Whereas PS-SCL, HyCoSuL, and
CoSeSuL allow determination of protease preferences only in prime
active site pockets, IQF can be used to refine complementarity to
both prime and non-prime pockets. The advantage of phage display
is its ability to generate large and diverse substrate arrays, up to 1010
peptides, and enrich for specificity after each cycle, which is not
feasible with chemical synthesis. The label-free nature of this
method requires individual kinetic analysis and reporter group
labeling of positive hits. Multiplex substrate profiling with liquid
chromatography–tandem mass spectrometry sequencing proved
successful for distinguishing neutral serine protease activity in
human neutrophil extracellular traps toward a large and diverse
tetradecapeptide panel, and ranking granzyme B substrate effi-
ciency, using label-free quantitation of precursor-ion abundance
[240]. The above techniques have also allowed differentiation
between various diverse members of the granzyme, kallikrein, cas-
pase, metalloproteinase, exopeptidase, deubiquitinating and desu-
moylating protease families. A novel approach of profiling protease
specificity is currently being developed, combining yeast endoplas-
mic reticulum (ER) sequestration with next-generation sequencing
 Proteases: Pivot Points in Functional Proteomics 375

(YESS-NGS) [241]. A substrate library, targeted to the ER, is

exposed to ER proteases as it transports through the secretory
pathway, and cleaved/uncleaved substrates localize to the cell sur-
face. FACS analysis of cells labeled with fluorophore-conjugated
antibodies specifically detects substrate cleavage. Characterization
of proteolytic processing in secretory pathways may be useful to
detect changes in the secretome in various disease states.
Recent, targeted “intractable” protein degradation methodol-
ogy uses heterobifunctional chemistry for simultaneous binding of
proteins inside cells, which tags them for degradation by the cell’s
own ubiquitin-proteasome system. The small molecule drug candi-
date uses a protein-binding domain linked to a ubiquitin ligase-
binding domain, with the goal of eliminating pathological or defec-
tive intracellular proteins. This approach is currently developed by
Kymera. A variant of this method is targeted proteolysis of endoge-
nous proteins via the affinity-directed protein missile (AdPROM)
system [242], that harbors the von Hippel-Lindau (VHL) protein,
the substrate receptor of the Cullin2 (CUL2) E3 ligase complex,
tethered to polypeptide binders that selectively bind and recruit
endogenous target proteins to the CUL2-E3 ligase complex for
ubiquitination and proteasomal degradation. Synthetic monobo-
dies and a camelid-derived VHH nanobody were used in a feasibil-
ity model to target the tyrosine phosphatase SHP2 and the
inflammasome protein ASC for degradation. This method has
advantages over CRISPR/Cas9-mediated gene knockouts that are
irreversible and may not always be feasible, and over RNA interfer-
ence that requires prolonged treatment and may be incomplete.
Both methods may have off-target effects. A possible breakthrough
in nanomedicine may be the construction of a “DNA vault”, a
DNA origami nanodevice that locks up a single enzyme molecule,
and that can be opened and closed by DNA locks to regulate access
to substrate [243]. In a proof-of-principle model, chymotrypsin
was covalently anchored to an open DNA vault, and after closure a
FITC-casein substrate was added, together with the opening key or
a control key. Enzymatic activity was detected predominantly in the
reaction containing the opening key. This technology can be
refined to program natural enzymes to operate as signal amplifiers
for diagnostics applications and as delivery vehicles for therapeutic
applications.

5.4 Discovery of Conventional shotgun proteomics for biomarker discovery lack

Protease and Inhibitor sensitivity and selectivity because they focus mainly on quantitative
Biomarkers for rather than qualitative differences between the normal and the
Disease diseased state. These approaches are also ill-suited for detecting
pathologic posttranslational modifications. Novel MS-based prote-
omics techniques have the potential of identifying newly generated
N- and C-termini that are characteristic in disease-related proteoly-
sis. The TAILS and C-TAILS techniques enrich for protein N- and
C-terminal peptides by polymer-based removal of internal peptides
376 Ingrid M. Verhamme et al.

generated by tryptic digestion, and improve detection limits by

several orders of magnitude [154]. The differential processing of
APP protein in Alzheimer’s disease, and of the chemokines CCL7
and SDF1 in arthritis and HIV-associated dementia, respectively,
are prime examples of neo-terminus formation. N- and C-terminal
peptide removal may activate or inactivate chemokines, turn them
into receptor antagonists or change their receptor specificity. As a
result, simple quantitation of chemokine expression is not sufficient
to define the extent of the pathological inflammatory response, but
quantitation of neo-terminus formation is. A fraction of Aβ pep-
tides in Alzheimer’s patients is N-truncated, with a cyclized termi-
nal glutamate posttranslational modification (Aβ3(Pe)), and this
species may be a promising biomarker for early detection of Alzhei-
mer’s disease. A search of the TopFIND public database [244]
shows that many FDA-approved biomarker proteins have multiple
different N- and C-termini, potentially affecting their biological
activity and “visibility” in current assays. Specific disease-related
protease assays may be confounded by endogenous and exogenous
inhibitors, and varying concentrations of cofactors and competing
proteases in biological samples. A combination of targeted MS and
neo-terminal-directed antibody assays may vastly improve the
development of reliable biomarkers.
Novel functional proteomics techniques such as bead-based
proteome enrichment combined with the 2D-electrophoresis-
based Protein Elution Plate (PEP) also prove extremely useful for
rapid, convenient profiling of functional protease activity in physi-
ological and pathological patient samples [245]. This versatile plate
reader technology uses specific protease substrates with various
degrees of sensitivity and allows functional analysis of the protease
landscape beyond the information available from protein abun-
dance measurements.
The pursuit of biomarker development within a Functional
Proteomics context is addressed elsewhere in this book (“Making
the Case for Functional Proteomics” and “Methods to Monitor the
Functional Subproteomes of SERPIN Protease Inhibitors”). Some
of the observations of those sections are re-viewed here through the
lens of proteolytic activity. Historically, biomarkers that drive drug
and/or diagnostic discovery comprise a single entity, be it a metab-
olite, waste product or protein. In no small part that pattern reflects
technology limitations of the time, not a preferred indicator of
biological functionality. Further, of the approximately 2000
in vitro protein assays approved by the US Food and Drug Admin-
istration, only seven assays reference protease activity (and some of
those are redundant for a given assay). This apparent lack of empha-
sis also points to limitations in proteolytic assay procedures, espe-
cially in media such as body fluids. Finally, as is abundantly apparent
from the information presented in this chapter, the activity of a
single protease is an unreliable indicator of physiological activity. A
wider net must be cast.
 Proteases: Pivot Points in Functional Proteomics 377

For any functionally defined biological process, e.g., apoptosis

or pathogen response, multiple proteolytic events occur in (rela-
tively) linear sequences or in (parallel) cascades. Thus, any proposed
therapeutic, regardless of selectivity at the level of a single entity,
modulates multiple outcomes. And, any diagnostic must reflect
that multi-entity process. The odds that a single proteolytic event
or single intermediate or end product will define that process are
slim indeed. The logic seems inescapable that biomarker develop-
ment of the future must focus on biological networks, not single
entities. Pioneering efforts are occurring as this is written.
An example: one diagnostic in development [246] for cancer
“Stroma Liquid Biopsy™” demonstrates a network (or pattern
[247])-based analysis of LC/MS data, connecting multiple net-
works or pathways that rely profoundly on proteolytic activity
(coagulation, complement, and inflammation). The discovery and
development sequence for such biomarker development is
addressed elsewhere in this book (“Making the Case . . .”). Applica-
tion of the diagnostic in the clinic relies on selection of a test
“subproteome” as described in “Methods to Monitor the Func-
tional Subproteomes of SERPIN Protease Inhibitors.” It’s noted
that not only does the “Stroma Liquid Biopsy™” apparently differ-
entiate serum from cancer patients, the pattern on which it is based
informs cancer biology. Further, the identified subproteome
(s) become a relevant test bed for screening new therapeutics.
The future is integrated discovery of network-based biomarkers
that support therapeutic and diagnostic discovery while informing
both biology and disease. Detection of protease activity per se in
complex media will likely require development of new methods.

5.5 A Two-Protein This chapter presents, in text and tabulated formats, the diversity of
Wrap-Up proteases and protease-related activity. Perhaps the most compel-
ling overall observation is the degree of overlap among physiologi-
cal processes and disease states at the level of individual proteases.
This is most readily seen in the tabulated summary of proteases,
inhibitors and activity. A thumbnail analysis shows that half of all
listed proteases are active in multiple physiological processes
and/or associated with multiple disease states. Chief among the
multi-process activities are the cathepsins, followed by the caspases
and MMPs. The chapter closes with an examination of the multiple
associations with physiological processes and disease states of two
related proteins: the protease thrombin and the proteolysis product
fibrin (gene expression products, prothrombin and fibrinogen).
Textbooks cite the cleavage of fibrinogen by thrombin to form
fibrin as a central event in formation of blood clots formed in
response to injury. This common response is certainly beneficial.
However, other actions and associations of thrombin and fibrin are
of a darker or unexpected nature (Fig. 2). In addition to its partici-
pation in blood clot formation, thrombin also triggers proteases
that attenuate clot formation [20]. Thus thrombin (and
378 Ingrid M. Verhamme et al.

One Protease, One Substrate:

Impact on Physiology and Disease
Hemostasis Innate Immune System
Clot Formation Clot Dissolution Complement Activation

Cardiovascular Cancer
(pro)Thrombin Pathway Fiber
Metabolic
Thrombosis Bleeding Fibrin(ogen) Activation Component

Bacterial Self Defense Neurodegeneration

Hijacked Fibrinogen Neurotoxic Plaque Component

Fig. 2 Multiple Physiological and Pathological Roles of the (pro)Thrombin - Fibrin(ogen) Interaction

prothrombin) is associated with both thrombosis and clot

dissolution. Elevated fibrin(ogen) levels track CV disease severity
[248], and elevated levels of fibrin are found in diabetic patients
[249]. In cancer, thrombin activates key pathways and fibrin is
found in tumors. Thrombin is often cited as a neurotoxin [250]
and is implicated in both Alzheimer’s disease and peripheral neu-
ropathy. Fibrin is found in AD plaques and may be oligomerized by
amyloid beta [251]. Fibrinogen is hijacked by bacteria and through
nonenzymatic prothrombin activation forms fibrin barriers to
evade cellular defenses [101, 185]. Finally, thrombin cleaves com-
plement C3 [252], thus igniting the innate immune system inde-
pendent of any triggers associated with infection. This latter event
may explain, in part, thrombin’s apparent and destructive ability to
trigger inflammatory and immune responses. Proteolytic events
govern the life cycle of all proteins, as was noted in the Introduction
to this chapter.
The clear lesson of the contents of this chapter is that proteoly-
sis is central to the functioning of the organism, indeed can reshape
the organism. Proteolytic activity is essential to multiple physiolog-
ical processes and, specific proteases are central to multiple pro-
cesses. It is not surprising, therefore, that disease is also strongly
connected to proteolysis. Indeed, even at the resolution of a single
protease, activity is not always to the benefit of the organism. As was
demonstrated in the paragraph immediately above, what might be
called the side effects of proteolysis can prove quite harmful. Of
particular interest is the tantalizing correlation of proteolytic activ-
ity and chronic disease.
 Proteases: Pivot Points in Functional Proteomics 379

Within the context of Functional Proteomics, whether the scale

is that of reshaping individual proteins or determining the state of
the organism, proteolytic events are foundational.

Acknowledgements

I.M.V. is supported by NIH/NHLBI grants R01 HL071544 and

R01 HL130018.

References

1. Fruton JS (2002) A history of pepsin and 49:593–626. https://doi.org/10.1146/

related enzymes. Q Rev Biol 77(2):127–147 annurev.bi.49.070180.003113
2. Northrop JH, Kunitz M, Herriott RM (1938) 11. Fortelny N, Cox JH, Kappelhoff R, Starr AE,
Crystalline enzymes. Columbia Univ. Press, Lange PF, Pavlidis P, Overall CM (2014) Net-
New York work analyses reveal pervasive functional reg-
3. Neurath H (1999) Proteolytic enzymes, past ulation between proteases in the human
and future. Proc Natl Acad Sci U S A 96 protease web. PLoS Biol 12(5):e1001869.
(20):10962–10963 https://doi.org/10.1371/journal.pbio.
4. Rawlings ND, Barrett AJ, Finn R (2016) 1001869
Twenty years of the MEROPS database of 12. Rechsteiner M, Rogers SW (1996) PEST
proteolytic enzymes, their substrates and inhi- sequences and regulation by proteolysis.
bitors. Nucleic Acids Res 44(D1): Trends Biochem Sci 21(7):267–271
D343–D350. https://doi.org/10.1093/ 13. Lopez-Otin C, Bond JS (2008) Proteases:
nar/gkv1118 multifunctional enzymes in life and disease. J
5. Kappelhoff R, Puente XS, Wilson CH, Seth A, Biol Chem 283(45):30433–30437. https://
Lopez-Otin C, Overall CM (2017) Overview doi.org/10.1074/jbc.R800035200
of transcriptomic analysis of all human pro- 14. Chakraborti S, Chakraborti T, Dhalla NS
teases, non-proteolytic homologs and inhibi- (eds) (2017) Proteases in human diseases.
tors: organ, tissue and ovarian cancer cell line Springer Singapore, New York
expression profiling of the human protease 15. Macfarlane RG (1964) An enzyme cascade in
degradome by the CLIP-CHIP DNA micro- the blood clotting mechanism, and its func-
array. Biochim Biophys Acta 1864(11 Pt tion as a biochemical amplifier. Nature
B):2210–2219. https://doi.org/10.1016/j. 202:498–499
bbamcr.2017.08.004 16. Gailani D, Broze GJ Jr (1993) Factor XI acti-
6. Perez-Silva JG, Espanol Y, Velasco G, Ques- vation by thrombin and factor XIa. Semin
ada V (2016) The Degradome database: Thromb Hemost 19(4):396–404. https://
expanding roles of mammalian proteases in doi.org/10.1055/s-2007-993291
life and disease. Nucleic Acids Res 44(D1): 17. Gailani D, Renne T (2007) Intrinsic pathway
D351–D355. https://doi.org/10.1093/ of coagulation and arterial thrombosis. Arter-
nar/gkv1201 ioscler Thromb Vasc Biol 27(12):2507–2513.
7. Rawlings ND, Salvesen G (2013) Handbook https://doi.org/10.1161/ATVBAHA.107.
of proteolytic enzymes, 3rd edn. Elsevier/AP, 155952
Amsterdam 18. Gailani D, Renne T (2007) The intrinsic path-
8. Turk B, Turk D, Turk V (2012) Protease way of coagulation: a target for treating
signalling: the cutting edge. EMBO J 31 thromboembolic disease? J Thromb Haemost
(7):1630–1643. https://doi.org/10.1038/ 5(6):1106–1112. https://doi.org/10.1111/
emboj.2012.42 j.1538-7836.2007.02446.x
9. Gettins PG, Olson ST (2016) Inhibitory ser- 19. Nesheim M, Bajzar L (2005) The discovery of
pins. New insights into their folding, poly- TAFI. J Thromb Haemost 3(10):2139–2146.
merization, regulation and clearance. https://doi.org/10.1111/j.1538-7836.
Biochem J 473(15):2273–2293. https://doi. 2005.01280.x
org/10.1042/BCJ20160014 20. Bode W (2006) The structure of thrombin: a
10. Laskowski M Jr, Kato I (1980) Protein inhi- janus-headed proteinase. Semin Thromb
bitors of proteinases. Annu Rev Biochem
380 Ingrid M. Verhamme et al.

Hemost 32(Suppl 1):16–31. https://doi. 31. Ehlers MR (2014) Immune-modulating

org/10.1055/s-2006-939551 effects of alpha-1 antitrypsin. Biol Chem 395
21. Huntington JA (2008) How Na+ activates (10):1187–1193. https://doi.org/10.1515/
thrombin—a review of the functional and hsz-2014-0161
structural data. Biol Chem 389 32. Boya P (2012) Lysosomal function and dys-
(8):1025–1035. https://doi.org/10.1515/ function: mechanism and disease. Antioxid
BC.2008.113 Redox Signal 17(5):766–774. https://doi.
22. Di Cera E (2007) Thrombin as procoagulant org/10.1089/ars.2011.4405
and anticoagulant. J Thromb Haemost 5 33. Schmidt M, Finley D (2014) Regulation of
(Suppl 1):196–202. https://doi.org/10. proteasome activity in health and disease. Bio-
1111/j.1538-7836.2007.02485.x chim Biophys Acta 1843(1):13–25. https://
23. Trouw LA, Pickering MC, Blom AM (2017) doi.org/10.1016/j.bbamcr.2013.08.012
The complement system as a potential thera- 34. Goldberg AL (2005) Nobel committee tags
peutic target in rheumatic disease. Nat Rev ubiquitin for distinction. Neuron 45
Rheumatol 13(9):538–547. https://doi. (3):339–344. https://doi.org/10.1016/j.
org/10.1038/nrrheum.2017.125 neuron.2005.01.019
24. Sim RB, Laich A (2000) Serine proteases of 35. Tanaka K (2009) The proteasome: overview
the complement system. Biochem Soc Trans of structure and functions. Proc Jpn Acad Ser
28(5):545–550 B Phys Biol Sci 85(1):12–36
25. Cooper NR, Muller-Eberhard HJ (1970) The 36. Antalis TM, Shea-Donohue T, Vogel SN,
reaction mechanism of human C5 in immune Sears C, Fasano A (2007) Mechanisms of dis-
hemolysis. J Exp Med 132(4):775–793 ease: protease functions in intestinal mucosal
26. Dobo J, Szakacs D, Oroszlan G, Kortvely E, pathobiology. Nat Clin Pract Gastroenterol
Kiss B, Boros E, Szasz R, Zavodszky P, Gal P, Hepatol 4(7):393–402. https://doi.org/10.
Pal G (2016) MASP-3 is the exclusive 1038/ncpgasthep0846
pro-factor D activator in resting blood: the 37. Alloy AP, Kayode O, Wang RY, Hockla A,
lectin and the alternative complement path- Soares AS, Radisky ES (2015) Mesotrypsin
ways are fundamentally linked. Sci Rep-UK has evolved four unique residues to cleave
6. https://doi.org/10.1038/srep31877 trypsin inhibitors as substrates. J Biol Chem
27. Sciascia S, Radin M, Yazdany J, 290(35):21523–21535. https://doi.org/10.
Tektonidou M, Cecchi I, Roccatello D, 1074/jbc.M115.662429
Dall’Era M (2017) Expanding the therapeutic 38. Giebeler N, Zigrino P (2016) A disintegrin
options for renal involvement in lupus: eculi- and metalloprotease (ADAM): historical over-
zumab, available evidence. Rheumatol Int 37 view of their functions. Toxins (Basel) 8
(8):1249–1255. https://doi.org/10.1007/ (4):122. https://doi.org/10.3390/
s00296-017-3686-5 toxins8040122
28. de Koning PJ, Kummer JA, de Poot SA, 39. Rodriguez D, Morrison CJ, Overall CM
Quadir R, Broekhuizen R, McGettrick AF, (2010) Matrix metalloproteinases: what do
Higgins WJ, Devreese B, Worrall DM, they not do? New substrates and biological
Bovenschen N (2011) Intracellular serine roles identified by murine models and proteo-
protease inhibitor SERPINB4 inhibits gran- mics. Biochim Biophys Acta 1803(1):39–54.
zyme M-induced cell death. PLoS One 6(8): https://doi.org/10.1016/j.bbamcr.2009.
e22645. https://doi.org/10.1371/journal. 09.015
pone.0022645 40. Freitas-Rodriguez S, Folgueras AR, Lopez-
29. Soriano C, Mukaro V, Hodge G, Ahern J, Otin C (2017) The role of matrix metallopro-
Holmes M, Jersmann H, Moffat D, teinases in aging: Tissue remodeling and
Meredith D, Jurisevic C, Reynolds PN, beyond. Biochim Biophys Acta 1864(11 Pt
Hodge S (2012) Increased proteinase A):2015–2025. https://doi.org/10.1016/j.
inhibitor-9 (PI-9) and reduced granzyme B bbamcr.2017.05.007
in lung cancer: mechanism for immune eva- 41. Brew K, Nagase H (2010) The tissue inhibi-
sion? Lung Cancer 77(1):38–45. https://doi. tors of metalloproteinases (TIMPs): an
org/10.1016/j.lungcan.2012.01.017 ancient family with structural and functional
30. Biancheri P, Di Sabatino A, Corazza GR, diversity. Biochim Biophys Acta 1803
MacDonald TT (2013) Proteases and the (1):55–71. https://doi.org/10.1016/j.
gut barrier. Cell Tissue Res 351(2):269–280. bbamcr.2010.01.003
https://doi.org/10.1007/s00441-012- 42. Coughlin SR (2005) Protease-activated
1390-z receptors in hemostasis, thrombosis and
 Proteases: Pivot Points in Functional Proteomics 381

vascular biology. J Thromb Haemost 3 52. Kuida K (2000) Caspase-9. Int J Biochem
(8):1800–1814. https://doi.org/10.1111/j. Cell Biol 32(2):121–124
1538-7836.2005.01377.x 53. Blasche S, Mortl M, Steuber H, Siszler G,
43. Fender AC, Rauch BH, Geisler T, Schror K Nisa S, Schwarz F, Lavrik I, Gronewold TM,
(2017) Protease-activated receptor par-4: an Maskos K, Donnenberg MS, Ullmann D,
inducible switch between thrombosis and vas- Uetz P, Kogl M (2013) The E. coli effector
cular inflammation? Thromb Haemost 117 protein NleF is a caspase inhibitor. PLoS One
(11):2013–2025. https://doi.org/10.1160/ 8(3):e58937. https://doi.org/10.1371/jour
TH17-03-0219 nal.pone.0058937
44. Takamori N, Azuma H, Kato M, 54. Li P, Nijhawan D, Wang X (2004) Mitochon-
Hashizume S, Aihara K, Akaike M, drial activation of apoptosis. Cell 116
Tamura K, Matsumoto T (2004) High plasma (2 Suppl):S57–59, 52 p following S59
heparin cofactor II activity is associated with 55. Denault JB, Eckelman BP, Shin H, Pop C,
reduced incidence of in-stent restenosis after Salvesen GS (2007) Caspase 3 attenuates
percutaneous coronary intervention. Circula- XIAP (X-linked inhibitor of apoptosis
tion 109(4):481–486. https://doi.org/10. protein)-mediated inhibition of caspase
1161/01.CIR.0000109695.39671.37 9. Biochem J 405(1):11–19. https://doi.
45. Nierodzik ML, Karpatkin S (2006) Thrombin org/10.1042/BJ20070288
induces tumor growth, metastasis, and angio- 56. Creagh EM (2014) Caspase crosstalk: integra-
genesis: Evidence for a thrombin-regulated tion of apoptotic and innate immune signal-
dormant tumor phenotype. Cancer Cell 10 ling pathways. Trends Immunol 35
(5):355–362. https://doi.org/10.1016/j. (12):631–640. https://doi.org/10.1016/j.
ccr.2006.10.002 it.2014.10.004
46. Asanuma K, Wakabayashi H, Okamoto T, 57. Wang XJ, Cao Q, Liu X, Wang KT, Mi W,
Asanuma Y, Akita N, Yoshikawa T, Zhang Y, Li LF, LeBlanc AC, Su XD (2010)
Hayashi T, Matsumine A, Uchida A, Sudo A Crystal structures of human caspase 6 reveal a
(2013) The thrombin inhibitor, argatroban, new mechanism for intramolecular cleavage
inhibits breast cancer metastasis to bone. self-activation. EMBO Rep 11(11):841–847.
Breast Cancer 20(3):241–246. https://doi. https://doi.org/10.1038/embor.2010.141
org/10.1007/s12282-012-0334-5 58. Graham RK, Ehrnhoefer DE, Hayden MR
47. McIlwain DR, Berger T, Mak TW (2013) (2011) Caspase-6 and neurodegeneration.
Caspase functions in cell death and disease. Trends Neurosci 34(12):646–656. https://
Cold Spring Harb Perspect Biol 5(4): doi.org/10.1016/j.tins.2011.09.001
a008656. https://doi.org/10.1101/ 59. Bartel A, Gohler A, Hopf V, Breitbach K
cshperspect.a008656 (2017) Caspase-6 mediates resistance against
48. Li P, Nijhawan D, Budihardjo I, Srinivasula Burkholderia pseudomallei infection and
SM, Ahmad M, Alnemri ES, Wang X (1997) influences the expression of detrimental cyto-
Cytochrome c and dATP-dependent forma- kines. PLoS One 12(7):e0180203. https://
tion of Apaf-1/caspase-9 complex initiates doi.org/10.1371/journal.pone.0180203
an apoptotic protease cascade. Cell 91 60. Sollberger G, Strittmatter GE,
(4):479–489 Garstkiewicz M, Sand J, Beer HD (2014)
49. Zhivotovsky B, Samali A, Gahm A, Orrenius S Caspase-1: the inflammasome and beyond.
(1999) Caspases: their intracellular localiza- Innate Immun 20(2):115–125. https://doi.
tion and translocation during apoptosis. Cell org/10.1177/1753425913484374
Death Differ 6(7):644–651. https://doi.org/ 61. Duclos C, Lavoie C, Denault JB (2017) Cas-
10.1038/sj.cdd.4400536 pases rule the intracellular trafficking cartel.
50. Li P, Zhou L, Zhao T, Liu X, Zhang P, Liu Y, FEBS J 284(10):1394–1420. https://doi.
Zheng X, Li Q (2017) Caspase-9: structure, org/10.1111/febs.14071
mechanisms and clinical application. Oncotar- 62. Julien O, Wells JA (2017) Caspases and their
get 8(14):23996–24008. https://doi.org/ substrates. Cell Death Differ 24
10.18632/oncotarget.15098 (8):1380–1389. https://doi.org/10.1038/
51. Vu NT, Park MA, Shultz JC, Goehe RW, Hoe- cdd.2017.44
ferlin LA, Shultz MD, Smith SA, Lynch KW, 63. Aziz M, Jacob A, Wang P (2014) Revisiting
Chalfant CE (2013) hnRNP U enhances caspases in sepsis. Cell Death Dis 5:e1526.
caspase-9 splicing and is modulated by https://doi.org/10.1038/cddis.2014.488
AKT-dependent phosphorylation of hnRNP
L. J Biol Chem 288(12):8575–8584. https:// 64. Pemberton CJ (2014) Signal peptides: new
doi.org/10.1074/jbc.M112.443333 markers in cardiovascular disease? Biomark
382 Ingrid M. Verhamme et al.

Med 8(8):1013–1019. https://doi.org/10. nuclear translocation of pro-caspase-1

2217/bmm.14.64 mediated by its prodomain. J Biol Chem 273
65. Morocz M, Zsigmond E, Toth R, Enyedi MZ, (37):23621–23624
Pinter L, Haracska L (2017) DNA-dependent 74. Kamada S, Kikkawa U, Tsujimoto Y, Hunter
protease activity of human Spartan facilitates T (2005) Nuclear translocation of caspase-3 is
replication of DNA-protein crosslink-contain- dependent on its proteolytic activation and
ing DNA. Nucleic Acids Res 45 recognition of a substrate-like protein(s). J
(6):3172–3188. https://doi.org/10.1093/ Biol Chem 280(2):857–860. https://doi.
nar/gkw1315 org/10.1074/jbc.C400538200
66. Stingele J, Habermann B, Jentsch S (2015) 75. Hill JW, Poddar R, Thompson JF, Rosenberg
DNA-protein crosslink repair: proteases as GA, Yang Y (2012) Intranuclear matrix metal-
DNA repair enzymes. Trends Biochem Sci loproteinases promote DNA damage and apo-
40(2):67–71. https://doi.org/10.1016/j. ptosis induced by oxygen-glucose deprivation
tibs.2014.10.012 in neurons. Neuroscience 220:277–290.
67. Vaz B, Popovic M, Newman JA, Fielden J, https://doi.org/10.1016/j.neuroscience.
Aitkenhead H, Halder S, Singh AN, 2012.06.019
Vendrell I, Fischer R, Torrecilla I, 76. Eguchi T, Calderwood SK, Takigawa M,
Drobnitzky N, Freire R, Amor DJ, Lockhart Kubota S, Kozaki KI (2017) Intracellular
PJ, Kessler BM, McKenna GW, Gileadi O, MMP3 promotes HSP gene expression in col-
Ramadan K (2016) Metalloprotease laboration with chromobox proteins. J Cell
SPRTN/DVC1 orchestrates replication- Biochem 118(1):43–51. https://doi.org/10.
coupled DNA-protein crosslink repair. Mol 1002/jcb.25607
Cell 64(4):704–719. https://doi.org/10. 77. Stepanova V, Jayaraman PS, Zaitsev SV,
1016/j.molcel.2016.09.032 Lebedeva T, Bdeir K, Kershaw R, Holman
68. Maskey RS, Flatten KS, Sieben CJ, Peterson KR, Parfyonova YV, Semina EV, Beloglazova
KL, Baker DJ, Nam HJ, Kim MS, Smyrk TC, IB, Tkachuk VA, Cines DB (2016)
Kojima Y, Machida Y, Santiago A, van Deur- Urokinase-type plasminogen activator (uPA)
sen JM, Kaufmann SH, Machida YJ (2017) promotes angiogenesis by attenuating
Spartan deficiency causes accumulation of proline-rich homeodomain protein (PRH)
Topoisomerase 1 cleavage complexes and transcription factor activity and de-repressing
tumorigenesis. Nucleic Acids Res 45 vascular endothelial growth factor (VEGF)
(8):4564–4576. https://doi.org/10.1093/ receptor expression. J Biol Chem 291
nar/gkx107 (29):15029–15045. https://doi.org/10.
69. Butler LR, Densham RM, Jia J, Garvin AJ, 1074/jbc.M115.678490
Stone HR, Shah V, Weekes D, Festy F, 78. Antalis TM, Bugge TH, Wu Q (2011)
Beesley J, Morris JR (2012) The proteasomal Membrane-anchored serine proteases in
de-ubiquitinating enzyme POH1 promotes health and disease. Prog Mol Biol Transl Sci
the double-strand DNA break response. 99:1–50. https://doi.org/10.1016/B978-0-
EMBO J 31(19):3918–3934. https://doi. 12-385504-6.00001-4
org/10.1038/emboj.2012.232 79. Friis S, Sales KU, Schafer JM, Vogel LK,
70. Pinto-Fernandez A, Kessler BM (2016) Kataoka H, Bugge TH (2014) The protease
DUBbing cancer: deubiquitylating enzymes inhibitor HAI-2, but not HAI-1, regulates
involved in epigenetics, DNA damage and matriptase activation and shedding through
the cell cycle as therapeutic targets. Front prostasin. J Biol Chem 289
Genet 7:133. https://doi.org/10.3389/ (32):22319–22332. https://doi.org/10.
fgene.2016.00133 1074/jbc.M114.574400
71. Enari M, Sakahira H, Yokoyama H, Okawa K, 80. Bardou O, Menou A, Francois C, Duitman
Iwamatsu A, Nagata S (1998) A caspase- JW, von der Thusen JH, Borie R, Sales KU,
activated DNase that degrades DNA during Mutze K, Castier Y, Sage E, Liu L, Bugge TH,
apoptosis, and its inhibitor ICAD. Nature Fairlie DP, Konigshoff M, Crestani B, Bor-
391(6662):43–50 ensztajn KS (2016) Membrane-anchored ser-
72. Venkatesh S, Lee J, Singh K, Lee I, Suzuki CK ine protease matriptase is a trigger of
(2012) Multitasking in the mitochondrion by pulmonary fibrogenesis. Am J Respir Crit
the ATP-dependent Lon protease. Biochim Care Med 193(8):847–860. https://doi.
Biophys Acta 1823(1):56–66. https://doi. org/10.1164/rccm.201502-0299OC
org/10.1016/j.bbamcr.2011.11.003 81. Le Gall SM, Szabo R, Lee M, Kirchhofer D,
73. Mao PL, Jiang Y, Wee BY, Porter AG (1998) Craik CS, Bugge TH, Camerer E (2016)
Activation of caspase-1 in the nucleus requires Matriptase activation connects tissue factor-
 Proteases: Pivot Points in Functional Proteomics 383

dependent coagulation initiation to epithelial antitrypsin deficiency and lung cancer risk: a
proteolysis and signaling. Blood 127 case-control study in never-smokers. J Thorac
(25):3260–3269. https://doi.org/10.1182/ Oncol 10(9):1279–1284. https://doi.org/
blood-2015-11-683110 10.1097/JTO.0000000000000609
82. Verhelst SHL (2017) Intramembrane pro- 91. Soderberg D, Segelmark M (2016) Neutro-
teases as drug targets. FEBS J 284 phil extracellular traps in ANCA-associated
(10):1489–1502. https://doi.org/10.1111/ vasculitis. Front Immunol 7:256. https://
febs.13979 doi.org/10.3389/fimmu.2016.00256
83. Dusterhoft S, Kunzel U, Freeman M (2017) 92. Denadai-Souza A, Ribeiro CM, Rolland C,
Rhomboid proteases in human disease: Thouard A, Deraison C, Scavone C,
Mechanisms and future prospects. Biochim Gonzalez-Dunia D, Vergnolle N, Avellar
Biophys Acta 1864(11 Pt B):2200–2209. MCW (2017) Effect of tryptase inhibition
https://doi.org/10.1016/j.bbamcr.2017. on joint inflammation: a pharmacological
04.016 and lentivirus-mediated gene transfer study.
84. Saita S, Nolte H, Fiedler KU, Kashkar H, Arthritis Res Ther 19. https://doi.org/10.
Venne AS, Zahedi RP, Kruger M, Langer T 1186/s13075-017-1326-9
(2017) PARL mediates Smac proteolytic mat- 93. Leskinen MJ, Lindstedt KA, Wang Y, Kova-
uration in mitochondria to promote apopto- nen PT (2003) Mast cell chymase induces
sis. Nat Cell Biol 19(4):318–328. https:// smooth muscle cell apoptosis by a mechanism
doi.org/10.1038/ncb3488 involving fibronectin degradation and disrup-
85. Ranganathan P, Weaver KL, Capobianco AJ tion of focal adhesions. Arterioscler Thromb
(2011) Notch signalling in solid tumours: a Vasc Biol 23(2):238–243
little bit of everything but not all the time. 94. He A, Shi GP (2013) Mast cell chymase and
Nat Rev Cancer 11(5):338–351. https://doi. tryptase as targets for cardiovascular and met-
org/10.1038/nrc3035 abolic diseases. Curr Pharm Des 19
86. Chauhan S, Mandal P, Tomar RS (2016) Bio- (6):1114–1125
chemical analysis reveals the multifactorial 95. Shirai T, Hilhorst M, Harrison DG, Goronzy
mechanism of histone H3 clipping by chicken JJ, Weyand CM (2015) Macrophages in vas-
liver histone H3 protease. Biochemistry 55 cular inflammation—from atherosclerosis to
(38):5464–5482. https://doi.org/10.1021/ vasculitis. Autoimmunity 48(3):139–151.
acs.biochem.6b00625 https://doi.org/10.3109/08916934.2015.
87. Vossaert L, Meert P, Scheerlinck E, Glibert P, 1027815
Van Roy N, Heindryckx B, De Sutter P, 96. Sendler M, Maertin S, John D, Persike M,
Dhaenens M, Deforce D (2014) Identifica- Weiss FU, Kruger B, Wartmann T, Wagh P,
tion of histone H3 clipping activity in human Halangk W, Schaschke N, Mayerle J, Lerch
embryonic stem cells. Stem Cell Res 13 MM (2016) Cathepsin B activity initiates apo-
(1):123–134. https://doi.org/10.1016/j. ptosis via digestive protease activation in pan-
scr.2014.05.002 creatic acinar cells and experimental
88. Deraison C, Bonnart C, Vergnolle N (2018) pancreatitis. J Biol Chem 291
Proteases. In: Cavaillon J-M, Singer M (eds) (28):14717–14731. https://doi.org/10.
Inflammation: from molecular and cellular 1074/jbc.M116.718999
mechanisms to the clinic. Wiley-VCH, Wein- 97. Kayode O, Huang Z, Soares AS, Caulfield
heim, Germany, pp 727–766 TR, Dong Z, Bode AM, Radisky ES (2017)
89. Chotirmall SH, Al-Alawi M, McEnery T, Small molecule inhibitors of mesotrypsin
McElvaney NG (2015) Alpha-1 proteinase from a structure-based docking screen. PLoS
inhibitors for the treatment of alpha-1 anti- One 12(5):e0176694. https://doi.org/10.
trypsin deficiency: safety, tolerability, and 1371/journal.pone.0176694
patient outcomes. Ther Clin Risk Manag 98. Rolland-Fourcade C, Denadai-Souza A,
11:143–151. https://doi.org/10.2147/ Cirillo C, Lopez C, Jaramillo JO,
TCRM.S51474 Desormeaux C, Cenac N, Motta JP,
90. Torres-Duran M, Ruano-Ravina A, Parente- Larauche M, Tache Y, Berghe PV,
Lamelas I, Abal-Arca J, Leiro-Fernandez V, Neunlist M, Coron E, Kirzin S, Portier G,
Montero-Martinez C, Pena C, Castro- Bonnet D, Alric L, Vanner S, Deraison C,
Anon O, Golpe-Gomez A, Gonzalez-Barcala Vergnolle N (2017) Epithelial expression
FJ, Martinez C, Guzman-Taveras R, and function of trypsin-3 in irritable bowel
Provencio M, Mejuto-Marti MJ, Fernandez- syndrome. Gut 66(10):1767–1778. https://
Villar A, Barros-Dios JM (2015) Alpha-1 doi.org/10.1136/gutjnl-2016-312094
384 Ingrid M. Verhamme et al.

99. Ricklin D, Lambris JD (2013) Complement II. J Biol Chem 285(11):8278–8289.

in immune and inflammatory disorders: https://doi.org/10.1074/jbc.M109.
pathophysiological mechanisms. J Immunol 005967
190(8):3831–3838. https://doi.org/10. 110. Raghuraman A, Mosier PD, Desai UR (2010)
4049/jimmunol.1203487 Understanding dermatan sulfate-heparin
100. Hua Y, Nair S (2015) Proteases in cardiome- cofactor II interaction through virtual library
tabolic diseases: pathophysiology, molecular screening. ACS Med Chem Lett 1
mechanisms and clinical applications. Bio- (6):281–285. https://doi.org/10.1021/
chim Biophys Acta 1852(2):195–208. ml100048y
https://doi.org/10.1016/j.bbadis.2014.04. 111. Tollefsen DM, Maimone MM, McGuire EA,
032 Peacock ME (1989) Heparin cofactor II acti-
101. Friedrich R, Panizzi P, Fuentes-Prior P, vation by dermatan sulfate. Ann N Y Acad Sci
Richter K, Verhamme I, Anderson PJ, 556:116–122
Kawabata S, Huber R, Bode W, Bock PE 112. Aihara K, Azuma H, Takamori N,
(2003) Staphylocoagulase is a prototype for Kanagawa Y, Akaike M, Fujimura M,
the mechanism of cofactor-induced zymogen Yoshida T, Hashizume S, Kato M,
activation. Nature 425(6957):535–539. Yamaguchi H, Kato S, Ikeda Y, Arase T,
https://doi.org/10.1038/nature01962 Kondo A, Matsumoto T (2004) Heparin
102. Weidmann H, Heikaus L, Long AT, cofactor II is a novel protective factor against
Naudin C, Schluter H, Renne T (2017) The carotid atherosclerosis in elderly individuals.
plasma contact system, a protease cascade at Circulation 109(22):2761–2765. https://
the nexus of inflammation, coagulation and doi.org/10.1161/01.CIR.0000129968.
immunity. Biochim Biophys Acta 1864(11 Pt 46095.F3
B):2118–2127. https://doi.org/10.1016/j. 113. Polderdijk SG, Adams TE, Ivanciu L, Camire
bbamcr.2017.07.009 RM, Baglin TP, Huntington JA (2017)
103. Zamolodchikov D, Renne T, Strickland S Design and characterization of an
(2016) The Alzheimer’s disease peptide APC-specific serpin for the treatment of
beta-amyloid promotes thrombin generation hemophilia. Blood 129(1):105–113.
through activation of coagulation factor XII. J https://doi.org/10.1182/blood-2016-05-
Thromb Haemost 14(5):995–1007. https:// 718635
doi.org/10.1111/jth.13209 114. Panizzi P, Boxrud PD, Verhamme IM, Bock
104. Esmon CT, Vigano-D’Angelo S, D’Angelo A, PE (2006) Binding of the COOH-terminal
Comp PC (1987) Anticoagulation proteins C lysine residue of streptokinase to plasmin
and S. Adv Exp Med Biol 214:47–54 (ogen) kringles enhances formation of the
105. Bertina RM, Koeleman BP, Koster T, Rosen- streptokinase.plasmin(ogen) catalytic com-
daal FR, Dirven RJ, de Ronde H, van der plexes. J Biol Chem 281(37):26774–26778.
Velden PA, Reitsma PH (1994) Mutation in https://doi.org/10.1074/jbc.C600171200
blood coagulation factor V associated with 115. Verhamme IM, Bock PE (2008) Rapid-
resistance to activated protein C. Nature 369 reaction kinetic characterization of the path-
(6475):64–67. https://doi.org/10.1038/ way of streptokinase-plasmin catalytic com-
369064a0 plex formation. J Biol Chem 283
106. Kujovich JL (1993) Factor V Leiden throm- (38):26137–26147. https://doi.org/10.
bophilia. In: Adam MP, Ardinger HH, Pagon 1074/jbc.M804038200
RA et al (eds) GeneReviews(R). University of 116. Verhamme IM, Bock PE (2014) Rapid bind-
Washington, Seattle, WA ing of plasminogen to streptokinase in a cata-
107. Lane D Antithrombin mutation database lytic complex reveals a three-step mechanism.
108. Verhamme IM, Olson ST, Tollefsen DM, J Biol Chem 289(40):28006–28018. https://
Bock PE (2002) Binding of exosite ligands doi.org/10.1074/jbc.M114.589077
to human thrombin. Re-evaluation of alloste- 117. Weiss D, Sorescu D, Taylor WR (2001)
ric linkage between thrombin exosites I and Angiotensin II and atherosclerosis. Am J Car-
II. J Biol Chem 277(9):6788–6798. https:// diol 87(8A):25C–32C
doi.org/10.1074/jbc.M110257200 118. Kossmann S, Lagrange J, Jackel S, Jurk K,
109. Sarilla S, Habib SY, Kravtsov DV, Ehlken M, Schonfelder T, Weihert Y,
Matafonov A, Gailani D, Verhamme IM Knorr M, Brandt M, Xia N, Li H, Daiber A,
(2010) Sucrose octasulfate selectively acceler- Oelze M, Reinhardt C, Lackner K, Gruber A,
ates thrombin inactivation by heparin cofactor Monia B, Karbach SH, Walter U, Ruggeri
 Proteases: Pivot Points in Functional Proteomics 385

ZM, Renne T, Ruf W, Munzel T, Wenzel P 127. Crocker SJ, Pagenstecher A, Campbell IL
(2017) Platelet-localized FXI promotes a vas- (2004) The TIMPs tango with MMPs and
cular coagulation-inflammatory circuit in more in the central nervous system. J Neu-
arterial hypertension. Sci Transl Med 9 rosci Res 75(1):1–11. https://doi.org/10.
(375). https://doi.org/10.1126/ 1002/jnr.10836
scitranslmed.aah4923 128. Brucher BL, Jamall IS (2016) Somatic muta-
119. Camare C, Pucelle M, Negre-Salvayre A, Sal- tion theory—why it’s wrong for most cancers.
vayre R (2017) Angiogenesis in the athero- Cell Physiol Biochem 38(5):1663–1680.
sclerotic plaque. Redox Biol 12:18–34. https://doi.org/10.1159/000443106
https://doi.org/10.1016/j.redox.2017.01. 129. Brucher BL, Jamall IS (2014) Epistemology
007 of the origin of cancer: a new paradigm. BMC
120. Wilson WRW, Anderton M, Choke EC, Cancer 14:331. https://doi.org/10.1186/
Dawson J, Loftus IM, Thompson MM 1471-2407-14-331
(2008) Elevated plasma MMP1 and MMP9 130. Schuliga M (2015) The inflammatory actions
are associated with abdominal aortic aneu- of coagulant and fibrinolytic proteases in dis-
rysm rupture. Eur J Vasc Endovasc 35 ease. Mediators Inflamm 2015:437695.
(5):580–584. https://doi.org/10.1016/j. https://doi.org/10.1155/2015/437695
ejvs.2007.12.004 131. Fan J, Ning B, Lyon CJ, Hu TY (2017) Cir-
121. Xue L, Borne Y, Mattisson IY, Wigren M, culating peptidome and tumor-resident pro-
Melander O, Ohro-Melander M, teolysis. Enzyme 42:1–25. https://doi.org/
Bengtsson E, Fredrikson GN, Nilsson J, 10.1016/bs.enz.2017.08.001
Engstrom G (2017) FADD, caspase-3, and 132. Guo Z, Jin X, Jia H (2013) Inhibition of
caspase-8 and incidence of coronary events. ADAM-17 more effectively down-regulates
Arterioscler Thromb Vasc Biol 37 the Notch pathway than that of gamma-
(5):983–989. https://doi.org/10.1161/ secretase in renal carcinoma. J Exp Clin Can-
ATVBAHA.117.308995 cer Res 32:26. https://doi.org/10.1186/
122. Musante L, Tataruch D, Gu D, Liu X, 1756-9966-32-26
Forsblom C, Groop PH, Holthofer H 133. Jackson HW, Defamie V, Waterhouse P, Kho-
(2015) Proteases and protease inhibitors of kha R (2017) TIMPs: versatile extracellular
urinary extracellular vesicles in diabetic regulators in cancer. Nat Rev Cancer 17
nephropathy. J Diabetes Res 2015:289734. (1):38–53. https://doi.org/10.1038/nrc.
https://doi.org/10.1155/2015/289734 2016.115
123. Zhao Z, Yang P, Eckert RL, Reece EA (2009) 134. LaRocca G, Aspelund T, Greve AM,
Caspase-8: a key role in the pathogenesis of Eiriksdottir G, Acharya T, Thorgeirsson G,
diabetic embryopathy. Birth Defects Res B Harris TB, Launer LJ, Gudnason V, Arai AE
Dev Reprod Toxicol 86(1):72–77. https:// (2017) Fibrosis as measured by the bio-
doi.org/10.1002/bdrb.20185 marker, tissue inhibitor metalloproteinase-1,
124. Augstein P, Bahr J, Wachlin G, Heinke P, predicts mortality in Age Gene Environment
Berg S, Salzsieder E, Harrison LC (2004) Susceptibility-Reykjavik (AGES-Reykjavik)
Cytokines activate caspase-3 in insulinoma Study. Eur Heart J 38(46):3423–3430.
cells of diabetes-prone NOD mice directly https://doi.org/10.1093/eurheartj/ehx510
and via upregulation of Fas. J Autoimmun 135. Lee JY, Kong G (2016) Roles and epigenetic
23(4):301–309. https://doi.org/10.1016/j. regulation of epithelial-mesenchymal transi-
jaut.2004.09.006 tion and its transcription factors in cancer
125. Trompet S, Pons D, Kanse SM, de Craen AJ, initiation and progression. Cell Mol Life Sci
Ikram MA, Verschuren JJ, Zwinderman AH, 73(24):4643–4660. https://doi.org/10.
Doevendans PA, Tio RA, de Winter RJ, Slag- 1007/s00018-016-2313-z
boom PE, Westendorp RG, Jukema JW 136. Otsuki T, Fujimoto D, Hirono Y, Goi T,
(2011) Factor VII activating protease poly- Yamaguchi A (2014) Thrombin conducts
morphism (G534E) is associated with epithelialmesenchymal transition via protea-
increased risk for stroke and mortality. Stroke seactivated receptor1 in human gastric cancer.
Res Treat 2011:424759. https://doi.org/10. Int J Oncol 45(6):2287–2294. https://doi.
4061/2011/424759 org/10.3892/ijo.2014.2651
126. Turner RJ, Sharp FR (2016) Implications of 137. Bawa-Khalfe T, Lu LS, Zuo Y, Huang C,
MMP9 for blood brain barrier disruption and Dere R, Lin FM, Yeh ET (2012) Differential
hemorrhagic transformation following ische- expression of SUMO-specific protease 7 var-
mic stroke. Front Cell Neurosci 10:56. iants regulates epithelial-mesenchymal
https://doi.org/10.3389/fncel.2016.00056
386 Ingrid M. Verhamme et al.

transition. Proc Natl Acad Sci U S A 109 Biol 24(2):125–126. https://doi.org/10.

(43):17466–17471. https://doi.org/10. 1016/j.chembiol.2017.01.007
1073/pnas.1209378109 147. Weyburne ES, Wilkins OM, Sha Z, Williams
138. Schmidt N, Irle I, Ripkens K, Lux V, Nelles J, DA, Pletnev AA, de Bruin G, Overkleeft HS,
Johannes C, Parry L, Greenow K, Amir S, Goldberg AL, Cole MD, Kisselev AF (2017)
Campioni M, Baldi A, Oka C, Kawaichi M, Inhibition of the proteasome beta2 site sensi-
Clarke AR, Ehrmann M (2016) Epigenetic tizes triple-negative breast cancer cells to
silencing of serine protease HTRA1 drives beta5 inhibitors and suppresses Nrf1 activa-
polyploidy. BMC Cancer 16:399. https:// tion. Cell Chem Biol 24(2):218–230.
doi.org/10.1186/s12885-016-2425-8 https://doi.org/10.1016/j.chembiol.2016.
139. Johnson JJ, Miller DL, Jiang R, Liu Y, Shi Z, 12.016
Tarwater L, Williams R, Balsara R, Sauter ER, 148. Reis ES, Mastellos DC, Ricklin D,
Stack MS (2016) Protease-activated receptor- Mantovani A, Lambris JD (2018) Comple-
2 (PAR-2)-mediated Nf-kappaB activation ment in cancer: untangling an intricate rela-
suppresses inflammation-associated tumor tionship. Nat Rev Immunol 18(1):5–18.
suppressor MicroRNAs in oral squamous cell https://doi.org/10.1038/nri.2017.97
carcinoma. J Biol Chem 291 149. Rutkowski MJ, Sughrue ME, Kane AJ, Mills
(13):6936–6945. https://doi.org/10.1074/ SA, Parsa AT (2010) Cancer and the comple-
jbc.M115.692640 ment cascade. Mol Cancer Res 8
140. Zhang W, Wang S, Wang Q, Yang Z, Pan Z, (11):1453–1465. https://doi.org/10.1158/
Li L (2014) Overexpression of cysteine 1541-7786.MCR-10-0225
cathepsin L is a marker of invasion and metas- 150. Zhu L, Jaamaa S, Af Hallstrom TM, Laiho M,
tasis in ovarian cancer. Oncol Rep 31 Sankila A, Nordling S, Stenman UH, Koisti-
(3):1334–1342. https://doi.org/10.3892/ nen H (2013) PSA forms complexes with
or.2014.2967 alpha1-antichymotrypsin in prostate. Prostate
141. Dian D, Heublein S, Wiest I, Barthell L, 73(2):219–226. https://doi.org/10.1002/
Friese K, Jeschke U (2014) Significance of pros.22560
the tumor protease cathepsin D for the biol- 151. DiScipio RG (1982) The activation of the
ogy of breast cancer. Histol Histopathol 29 alternative pathway C3 convertase by human
(4):433–438. https://doi.org/10.14670/ plasma kallikrein. Immunology 45
HH-29.10.433 (3):587–595
142. Cohen I, Kayode O, Hockla A, Sankaran B, 152. Caine GJ, Stonelake PS, Lip GY, Kehoe ST
Radisky DC, Radisky ES, Papo N (2016) (2002) The hypercoagulable state of malig-
Combinatorial protein engineering of proteo- nancy: pathogenesis and current debate. Neo-
lytically resistant mesotrypsin inhibitors as plasia 4(6):465–473. https://doi.org/10.
candidates for cancer therapy. Biochem J 1038/sj.neo.7900263
473(10):1329–1341. https://doi.org/10. 153. Amiral J, Seghatchian J (2017) Monitoring of
1042/BJ20151410 anticoagulant therapy in cancer patients with
143. Salameh MA, Radisky ES (2013) Biochemical thrombosis and the usefulness of blood acti-
and structural insights into mesotrypsin: an vation markers. Transfus Apher Sci 56
unusual human trypsin. Int J Biochem Mol (3):279–286. https://doi.org/10.1016/j.
Biol 4(3):129–139 transci.2017.05.010
144. Tanabe LM, List K (2017) The role of type II 154. Huesgen PF, Lange PF, Overall CM (2014)
transmembrane serine protease-mediated sig- Ensembles of protein termini and specific pro-
naling in cancer. FEBS J 284 teolytic signatures as candidate biomarkers of
(10):1421–1436. https://doi.org/10.1111/ disease. Proteomics Clin Appl 8
febs.13971 (5-6):338–350. https://doi.org/10.1002/
145. Zoratti GL, Tanabe LM, Hyland TE, prca.201300104
Duhaime MJ, Colombo E, Leduc R, 155. Kang JH, Korecka M, Toledo JB, Troja-
Marsault E, Johnson MD, Lin CY, nowski JQ, Shaw LM (2013) Clinical utility
Boerner J, Lang JE, List K (2016) Matriptase and analytical challenges in measurement of
regulates c-Met mediated proliferation and cerebrospinal fluid amyloid-beta(1-42) and
invasion in inflammatory breast cancer. Onco- tau proteins as Alzheimer disease biomarkers.
target 7(36):58162–58173. https://doi.org/ Clin Chem 59(6):903–916. https://doi.org/
10.18632/oncotarget.11262 10.1373/clinchem.2013.202937
146. Rolfe M (2017) The holy grail: solid tumor 156. Janelidze S, Stomrud E, Palmqvist S,
efficacy by proteasome inhibition. Cell Chem Zetterberg H, van Westen D, Jeromin A,
 Proteases: Pivot Points in Functional Proteomics 387

Song L, Hanlon D, Tan Hehir CA, Baker D, 165. Siklos M, BenAissa M, Thatcher GR (2015)
Blennow K, Hansson O (2016) Plasma beta- Cysteine proteases as therapeutic targets: does
amyloid in Alzheimer’s disease and vascular selectivity matter? A systematic review of cal-
disease. Sci Rep 6:26801. https://doi.org/ pain and cathepsin inhibitors. Acta Pharm Sin
10.1038/srep26801 B 5(6):506–519. https://doi.org/10.1016/
157. Roher AE, Kokjohn TA, Clarke SG, Sierks j.apsb.2015.08.001
MR, Maarouf CL, Serrano GE, Sabbagh 166. Martinelli P, Rugarli EI (2010) Emerging
MS, Beach TG (2017) APP/Abeta structural roles of mitochondrial proteases in neurode-
diversity and Alzheimer’s disease pathogene- generation. Biochim Biophys Acta 1797
sis. Neurochem Int 110:1–13. https://doi. (1):1–10. https://doi.org/10.1016/j.
org/10.1016/j.neuint.2017.08.007 bbabio.2009.07.013
158. Evin G, Li QX (2012) Platelets and Alzhei- 167. Konig T, Troder SE, Bakka K, Korwitz A,
mer’s disease: potential of APP as a bio- Richter-Dennerlein R, Lampe PA, Patron M,
marker. World J Psychiatry 2(6):102–113. Muhlmeister M, Guerrero-Castillo S,
https://doi.org/10.5498/wjp.v2.i6.102 Brandt U, Decker T, Lauria I, Paggio A,
159. Wetzel S, Seipold L, Saftig P (2017) The Rizzuto R, Rugarli EI, De Stefani D, Langer
metalloproteinase ADAM10: A useful thera- T (2016) The m-AAA protease associated
peutic target? Biochim Biophys Acta 1864 with neurodegeneration limits MCU activity
(11 Pt B):2071–2081. https://doi.org/10. in mitochondria. Mol Cell 64(1):148–162.
1016/j.bbamcr.2017.06.005 https://doi.org/10.1016/j.molcel.2016.08.
160. Bu XL, Xiang Y, Jin WS, Wang J, Shen LL, 020
Huang ZL, Zhang K, Liu YH, Zeng F, Liu 168. Strauss KM, Martins LM, Plun-Favreau H,
JH, Sun HL, Zhuang ZQ, Chen SH, Yao XQ, Marx FP, Kautzmann S, Berg D, Gasser T,
Giunta B, Shan YC, Tan J, Chen XW, Dong Wszolek Z, Muller T, Bornemann A,
ZF, Zhou HD, Zhou XF, Song W, Wang YJ Wolburg H, Downward J, Riess O, Schulz
(2017) Blood-derived amyloid-beta protein JB, Kruger R (2005) Loss of function muta-
induces Alzheimer’s disease pathologies. Mol tions in the gene encoding Omi/HtrA2 in
Psychiatry. https://doi.org/10.1038/mp. Parkinson’s disease. Hum Mol Genet 14
2017.204 (15):2099–2111. https://doi.org/10.1093/
161. Budd Haeberlein S, O’Gorman J, Chiao P, hmg/ddi215
Bussiere T, von Rosenstiel P, Tian Y, Zhu Y, 169. Fu J, Yu HM, Chiu SY, Mirando AJ, Mar-
von Hehn C, Gheuens S, Skordos L, Chen T, uyama EO, Cheng JG, Hsu W (2014) Dis-
Sandrock A (2017) Clinical development of ruption of SUMO-specific protease 2 induces
aducanumab, an anti-abeta human monoclo- mitochondria mediated neurodegeneration.
nal antibody being investigated for the treat- PLoS Genet 10(10):e1004579. https://doi.
ment of early Alzheimer’s disease. J Prev org/10.1371/journal.pgen.1004579
Alzheimers Dis 4(4):255–263. https://doi. 170. Orsini F, De Blasio D, Zangari R, Zanier ER,
org/10.14283/jpad.2017.39 De Simoni MG (2014) Versatility of the com-
162. Honig LS, Vellas B, Woodward M, Boada M, plement system in neuroinflammation, neuro-
Bullock R, Borrie M, Hager K, Andreasen N, degeneration and brain homeostasis. Front
Scarpini E, Liu-Seifert H, Case M, Dean RA, Cell Neurosci 8:380. https://doi.org/10.
Hake A, Sundell K, Poole Hoffmann V, 3389/fncel.2014.00380
Carlson C, Khanna R, Mintun M, 171. Descamps FJ, Van den Steen PE, Nelissen I,
DeMattos R, Selzler KJ, Siemers E (2018) Van Damme J, Opdenakker G (2003) Rem-
Trial of solanezumab for mild dementia due nant epitopes generate autoimmunity: from
to Alzheimer’s disease. N Engl J Med 378 rheumatoid arthritis and multiple sclerosis to
(4):321–330. https://doi.org/10.1056/ diabetes. Adv Exp Med Biol 535:69–77
NEJMoa1705971 172. Dinarello CA, Simon A, van der Meer JW
163. De Strooper B (2014) Lessons from a failed (2012) Treating inflammation by blocking
gamma-secretase Alzheimer trial. Cell 159 interleukin-1 in a broad spectrum of diseases.
(4):721–726. https://doi.org/10.1016/j. Nat Rev Drug Discov 11(8):633–652.
cell.2014.10.016 https://doi.org/10.1038/nrd3800
164. Lorenzl S, Albers DS, Relkin N, Ngyuen T, 173. Rupanagudi KV, Kulkarni OP, Lichtnekert J,
Hilgenberg SL, Chirichigno J, Cudkowicz Darisipudi MN, Mulay SR, Schott B,
ME, Beal MF (2003) Increased plasma levels Gruner S, Haap W, Hartmann G, Anders HJ
of matrix metalloproteinase-9 in patients with (2015) Cathepsin S inhibition suppresses sys-
Alzheimer’s disease. Neurochem Int 43 temic lupus erythematosus and lupus nephri-
(3):191–196 tis because cathepsin S is essential for MHC
388 Ingrid M. Verhamme et al.

class II-mediated CD4 T cell and B cell 182. Carroll IM, Maharshak N (2013) Enteric bac-
priming. Ann Rheum Dis 74(2):452–463. terial proteases in inflammatory bowel dis-
https://doi.org/10.1136/annrheumdis- ease- pathophysiology and clinical
2013-203717 implications. World J Gastroenterol 19
174. Schaller M, Vogel M, Kentouche K, (43):7531–7543. https://doi.org/10.3748/
Lammle B, Kremer Hovinga JA (2014) The wjg.v19.i43.7531
splenic autoimmune response to ADAMTS13 183. Boxrud PD, Verhamme IM, Bock PE (2004)
in thrombotic thrombocytopenic purpura Resolution of conformational activation in
contains recurrent antigen-binding CDR3 the kinetic mechanism of plasminogen activa-
motifs. Blood 124(23):3469–3479. https:// tion by streptokinase. J Biol Chem 279
doi.org/10.1182/blood-2014-04-561142 (35):36633–36641. https://doi.org/10.
175. Sadiq SK, Noe F, De Fabritiis G (2012) 1074/jbc.M405264200
Kinetic characterization of the critical step in 184. Chandrahas V, Glinton K, Liang Z, Donahue
HIV-1 protease maturation. Proc Natl Acad DL, Ploplis VA, Castellino FJ (2015) Direct
Sci U S A 109(50):20449–20454. https:// host plasminogen binding to bacterial surface
doi.org/10.1073/pnas.1210983109 M-protein in pattern D strains of streptococ-
176. Duschak VG, Couto AS (2009) Cruzipain, cus pyogenes is required for activation by its
the major cysteine protease of Trypanosoma natural coinherited SK2b protein. J Biol
cruzi: a sulfated glycoprotein antigen as rele- Chem 290(30):18833–18842. https://doi.
vant candidate for vaccine development and org/10.1074/jbc.M115.655365
drug target. A review. Curr Med Chem 16 185. Panizzi P, Friedrich R, Fuentes-Prior P,
(24):3174–3202 Bode W, Bock PE (2004) The staphylocoagu-
177. Plaza K, Kalinska M, Bochenska O, Meyer- lase family of zymogen activator and adhesion
Hoffert U, Wu Z, Fischer J, Falkowski K, proteins. Cell Mol Life Sci 61
Sasiadek L, Bielecka E, Potempa B, Kozik A, (22):2793–2798. https://doi.org/10.1007/
Potempa J, Kantyka T (2016) Gingipains of s00018-004-4285-7
porphyromonas gingivalis affect the stability 186. Parry MA, Zhang XC, Bode I (2000) Molec-
and function of serine protease inhibitor of ular mechanisms of plasminogen activation:
kazal-type 6 (SPINK6), a tissue inhibitor of bacterial cofactors provide clues. Trends Bio-
human kallikreins. J Biol Chem 291 chem Sci 25(2):53–59
(36):18753–18764. https://doi.org/10. 187. Wiles KG, Panizzi P, Kroh HK, Bock PE
1074/jbc.M116.722942 (2010) Skizzle is a novel plasminogen- and
178. Culp E, Wright GD (2017) Bacterial pro- plasmin-binding protein from Streptococcus
teases, untapped antimicrobial drug targets. agalactiae that targets proteins of human fibri-
J Antibiot (Tokyo) 70(4):366–377. https:// nolysis to promote plasmin generation. J Biol
doi.org/10.1038/ja.2016.138 Chem 285(27):21153–21164. https://doi.
179. Chang AK, Kim HY, Park JE, Acharya P, Park org/10.1074/jbc.M110.107730
IS, Yoon SM, You HJ, Hahm KS, Park JK, 188. Verhamme IM, Panizzi PR, Bock PE (2015)
Lee JS (2005) Vibrio vulnificus secretes a Pathogen activators of plasminogen. J
broad-specificity metalloprotease capable of Thromb Haemost 13(Suppl 1):S106–S114.
interfering with blood homeostasis through https://doi.org/10.1111/jth.12939
prothrombin activation and fibrinolysis. J 189. Meliopoulos VA, Andersen LE, Brooks P,
Bacteriol 187(20):6909–6916. https://doi. Yan X, Bakre A, Coleman JK, Tompkins SM,
org/10.1128/JB.187.20.6909-6916.2005 Tripp RA (2012) MicroRNA regulation of
180. Bibo-Verdugo B, Jiang Z, Caffrey CR, human protease genes essential for influenza
O’Donoghue AJ (2017) Targeting protea- virus replication. PLoS One 7(5):e37169.
somes in infectious organisms to combat dis- https://doi.org/10.1371/journal.pone.
ease. FEBS J 284(10):1503–1517. https:// 0037169
doi.org/10.1111/febs.14029 190. Homma T, Ishibashi D, Nakagaki T, Fuse T,
181. Pontarollo G, Acquasaliente L, Peterle D, Mori T, Satoh K, Atarashi R, Nishida N
Frasson R, Artusi I, De Filippis V (2017) (2015) Ubiquitin-specific protease 14 modu-
Non-canonical proteolytic activation of lates degradation of cellular prion protein. Sci
human prothrombin by subtilisin from Bacil- Rep 5:11028. https://doi.org/10.1038/
lus subtilis may shift the procoagulant- srep11028
anticoagulant equilibrium toward thrombo- 191. Michaud DS, Lu J, Peacock-Villada AY, Bar-
sis. J Biol Chem 292(37):15161–15179. ber JR, Joshu CE, Prizment AE, Beck JD,
https://doi.org/10.1074/jbc.M117. Offenbacher S, Platz EA (2018) Periodontal
795245 disease assessed using clinical dental
 Proteases: Pivot Points in Functional Proteomics 389

measurements and cancer risk in the ARIC 202. Tallant C, Marrero A, Gomis-Ruth FX (2010)
study. J Natl Cancer Inst. https://doi.org/ Matrix metalloproteinases: fold and function
10.1093/jnci/djx278 of their catalytic domains. Biochim Biophys
192. Drag M, Salvesen GS (2010) Emerging prin- Acta 1803(1):20–28. https://doi.org/10.
ciples in protease-based drug discovery. Nat 1016/j.bbamcr.2009.04.003
Rev Drug Discov 9(9):690–701. https://doi. 203. Gomis-Ruth FX (2017) Third time lucky?
org/10.1038/nrd3053 Getting a grip on matrix metalloproteinases.
193. Herzog RW (2015) Hemophilia gene ther- J Biol Chem 292(43):17975–17976. https://
apy: caught between a cure and an immune doi.org/10.1074/jbc.H117.806075
response. Mol Ther 23(9):1411–1412. 204. Scannevin RH, Alexander R, Haarlander TM,
https://doi.org/10.1038/mt.2015.135 Burke SL, Singer M, Huo C, Zhang YM,
194. Rangarajan S, Walsh L, Lester W, Perry D, Maguire D, Spurlino J, Deckman I, Carroll
Madan B, Laffan M, Yu H, Vettermann C, KI, Lewandowski F, Devine E,
Pierce GF, Wong WY, Pasi KJ (2017) AAV5- Dzordzorme K, Tounge B, Milligan C,
factor VIII gene transfer in severe hemophilia Bayoumy S, Williams R, Schalk-Hihi C,
A. N Engl J Med. https://doi.org/10.1056/ Leonard K, Jackson P, Todd M, Kuo LC,
NEJMoa1708483 Rhodes KJ (2017) Discovery of a highly selec-
195. George LA, Sullivan SK, Giermasz A, Rasko tive chemical inhibitor of matrix
JEJ, Samelson-Jones BJ, Ducore J, Cuker A, metalloproteinase-9 (MMP-9) that allosteri-
Sullivan LM, Majumdar S, Teitel J, McGuinn cally inhibits zymogen activation. J Biol
CE, Ragni MV, Luk AY, Hui D, Wright JF, Chem 292(43):17963–17974. https://doi.
Chen Y, Liu Y, Wachtel K, Winters A, org/10.1074/jbc.M117.806075
Tiefenbacher S, Arruda VR, van der Loo 205. Grunwald B, Vandooren J, Gerg M,
JCM, Zelenaia O, Takefman D, Carr ME, Ahomaa K, Hunger A, Berchtold S,
Couto LB, Anguela XM, High KA (2017) Akbareian S, Schaten S, Knolle P, Edwards
Hemophilia B gene therapy with a high- DR, Opdenakker G, Kruger A (2016) Sys-
specific-activity factor IX variant. N Engl J temic ablation of MMP-9 triggers invasive
Med 377(23):2215–2227. https://doi.org/ growth and metastasis of pancreatic cancer
10.1056/NEJMoa1708538 via deregulation of IL6 expression in the
196. Lai PS, Thompson BT (2013) Why activated bone marrow. Mol Cancer Res 14
protein C was not successful in severe sepsis (11):1147–1158. https://doi.org/10.1158/
and septic shock: are we still tilting at wind- 1541-7786.MCR-16-0180
mills? Curr Infect Dis Rep 15(5):407–412. 206. Radisky ES, Raeeszadeh-Sarmazdeh M,
https://doi.org/10.1007/s11908-013- Radisky DC (2017) Therapeutic potential of
0358-9 matrix metalloproteinase inhibition in breast
197. Janciauskiene SM, Bals R, Koczulla R, cancer. J Cell Biochem 118(11):3531–3548.
Vogelmeier C, Kohnlein T, Welte T (2011) https://doi.org/10.1002/jcb.26185
The discovery of alpha1-antitrypsin and its 207. Xia D, Watanabe H, Wu B, Lee SH, Li Y,
role in health and disease. Respir Med 105 Tsvetkov E, Bolshakov VY, Shen J, Kelleher
(8):1129–1139. https://doi.org/10.1016/j. RJ 3rd (2015) Presenilin-1 knockin mice
rmed.2011.02.002 reveal loss-of-function mechanism for familial
198. Carugati A, Pappalardo E, Zingale LC, Alzheimer’s disease. Neuron 85(5):967–981.
Cicardi M (2001) C1-inhibitor deficiency https://doi.org/10.1016/j.neuron.2015.
and angioedema. Mol Immunol 38 02.010
(2-3):161–173 208. Ran Y, Hossain F, Pannuti A, Lessard CB,
199. Ricklin D, Lambris JD (2016) New mile- Ladd GZ, Jung JI, Minter LM, Osborne BA,
stones ahead in complement-targeted ther- Miele L, Golde TE (2017) gamma-Secretase
apy. Semin Immunol 28(3):208–222. inhibitors in cancer clinical trials are pharma-
https://doi.org/10.1016/j.smim.2016.06. cologically and functionally distinct. EMBO
001 Mol Med 9(7):950–966. https://doi.org/
10.15252/emmm.201607265
200. Towards better patient care: drugs to avoid in
2014 (2014). Prescrire Int 23(150):161–165 209. Duong le T, Leung AT, Langdahl B (2016)
Cathepsin K inhibition: a new mechanism for
201. Adcock DM, Gosselin R (2015) Direct Oral the treatment of osteoporosis. Calcif Tissue
Anticoagulants (DOACs) in the Laboratory: Int 98(4):381–397. https://doi.org/10.
2015 Review. Thromb Res 136(1):7–12. 1007/s00223-015-0051-0
https://doi.org/10.1016/j.thromres.2015.
05.001 210. Drake MT, Clarke BL, Oursler MJ, Khosla S
(2017) Cathepsin K inhibitors for
390 Ingrid M. Verhamme et al.

osteoporosis: biology, potential clinical utility, biological evaluation of novel macrocyclic

and lessons learned. Endocr Rev 38 HIV-1 protease inhibitors involving the
(4):325–350. https://doi.org/10.1210/er. P10 -P20 ligands. Bioorg Med Chem Lett 27
2015-1114 (21):4925–4931. https://doi.org/10.1016/
211. Su EJ, Cao C, Fredriksson L, Nilsson I, j.bmcl.2017.09.003
Stefanitsch C, Stevenson TK, Zhao J, 219. McCauley JA, Rudd MT (2016) Hepatitis C
Ragsdale M, Sun YY, Yepes M, Kuan CY, virus NS3/4a protease inhibitors. Curr Opin
Eriksson U, Strickland DK, Lawrence DA, Pharmacol 30:84–92. https://doi.org/10.
Zhang L (2017) Microglial-mediated 1016/j.coph.2016.07.015
PDGF-CC activation increases cerebrovascu- 220. Thornberry NA, Weber AE (2007) Discovery
lar permeability during ischemic stroke. Acta of JANUVIA (Sitagliptin), a selective dipepti-
Neuropathol 134(4):585–604. https://doi. dyl peptidase IV inhibitor for the treatment of
org/10.1007/s00401-017-1749-z type 2 diabetes. Curr Top Med Chem 7
212. Lakhan SE, Kirchgessner A, Tepper D, Leo- (6):557–568
nard A (2013) Matrix metalloproteinases and 221. Schramm VL (2013) Transition States, analo-
blood-brain barrier disruption in acute ische- gues, and drug development. ACS Chem Biol
mic stroke. Front Neurol 4:32. https://doi. 8(1):71–81. https://doi.org/10.1021/
org/10.3389/fneur.2013.00032 cb300631k
213. Hafez S, Coucha M, Bruno A, Fagan SC, 222. Mitsuya H, Maeda K, Das D, Ghosh AK
Ergul A (2014) Hyperglycemia, acute ische- (2008) Development of protease inhibitors
mic stroke, and thrombolytic therapy. Transl and the fight with drug-resistant HIV-1 var-
Stroke Res 5(4):442–453. https://doi.org/ iants. Adv Pharmacol 56:169–197. https://
10.1007/s12975-014-0336-z doi.org/10.1016/S1054-3589(07)56006-0
214. Sun H, Xu Y, Sitkiewicz I, Ma Y, Wang X, 223. Kipp DR, Hirschi JS, Wakata A, Goldstein H,
Yestrepsky BD, Huang Y, Lapadatescu MC, Schramm VL (2012) Transition states of
Larsen MJ, Larsen SD, Musser JM, Ginsburg native and drug-resistant HIV-1 protease are
D (2012) Inhibitor of streptokinase gene the same. Proc Natl Acad Sci U S A 109
expression improves survival after group A (17):6543–6548. https://doi.org/10.1073/
streptococcus infection in mice. Proc Natl pnas.1202808109
Acad Sci U S A 109(9):3469–3474. https:// 224. Overall CM (2002) Molecular determinants
doi.org/10.1073/pnas.1201031109 of metalloproteinase substrate specificity:
215. Silva DG, Ribeiro JFR, De Vita D, Cianni L, matrix metalloproteinase substrate binding
Franco CH, Freitas-Junior LH, Moraes CB, domains, modules, and exosites. Mol Bio-
Rocha JR, Burtoloso ACB, Kenny PW, technol 22(1):51–86. https://doi.org/10.
Leitao A, Montanari CA (2017) A compara- 1385/MB:22:1:051
tive study of warheads for design of cysteine 225. Skarina T, Xu X, Evdokimova E, Savchenko A
protease inhibitors. Bioorg Med Chem Lett (2014) High-throughput crystallization
27(22):5031–5035. https://doi.org/10. screening. Methods Mol Biol
1016/j.bmcl.2017.10.002 1140:159–168. https://doi.org/10.1007/
216. Lv Z, Chu Y, Wang Y (2015) HIV protease 978-1-4939-0354-2_12
inhibitors: a review of molecular selectivity 226. Tiefenbrunn T, Forli S, Happer M,
and toxicity. HIV AIDS (Auckl) 7:95–104. Gonzalez A, Tsai Y, Soltis M, Elder JH,
https://doi.org/10.2147/HIV.S79956 Olson AJ, Stout CD (2014) Crystallographic
217. Ghosh AK, Brindisi M, Nyalapatla PR, fragment-based drug discovery: use of a bro-
Takayama J, Ella-Menye JR, Yashchuk S, minated fragment library targeting HIV pro-
Agniswamy J, Wang YF, Aoki M, Amano M, tease. Chem Biol Drug Des 83(2):141–148.
Weber IT, Mitsuya H (2017) Design of novel https://doi.org/10.1111/cbdd.12227
HIV-1 protease inhibitors incorporating 227. Rydel TJ, Tulinsky A, Bode W, Huber R
isophthalamide-derived P2-P3 ligands: Syn- (1991) Refined structure of the hirudin-
thesis, biological evaluation and X-ray struc- thrombin complex. J Mol Biol 221
tural studies of inhibitor-HIV-1 protease (2):583–601
complex. Bioorg Med Chem 25
(19):5114–5127. https://doi.org/10.1016/ 228. Warkentin TE (2004) Bivalent direct throm-
j.bmc.2017.04.005 bin inhibitors: hirudin and bivalirudin. Best
Pract Res Clin Haematol 17(1):105–125.
218. Ghosh AK, Sean Fyvie W, Brindisi M, https://doi.org/10.1016/j.beha.2004.02.
Steffey M, Agniswamy J, Wang YF, Aoki M, 002
Amano M, Weber IT, Mitsuya H (2017)
Design, synthesis, X-ray studies, and
 Proteases: Pivot Points in Functional Proteomics 391

229. Vance NR, Gakhar L, Spies MA (2017) Allo- Drag M (2014) Design of ultrasensitive
steric tuning of caspase-7: a fragment-based probes for human neutrophil elastase through
drug discovery approach. Angew Chem Int hybrid combinatorial substrate library
Ed Engl 56(46):14443–14447. https://doi. profiling. Proc Natl Acad Sci U S A 111
org/10.1002/anie.201706959 (7):2518–2523. https://doi.org/10.1073/
230. Weiss-Sadan T, Gotsman I, Blum G (2017) pnas.1318548111
Cysteine proteases in atherosclerosis. FEBS J 240. O’Donoghue AJ, Eroy-Reveles AA, Knudsen
284(10):1455–1472. https://doi.org/10. GM, Ingram J, Zhou M, Statnekov JB, Gre-
1111/febs.14043 ninger AL, Hostetter DR, Qu G, Maltby DA,
231. Lee S, Xie J, Chen X (2010) Activatable Anderson MO, Derisi JL, McKerrow JH,
molecular probes for cancer imaging. Curr Burlingame AL, Craik CS (2012) Global
Top Med Chem 10(11):1135–1144 identification of peptidase specificity by mul-
232. Ren G, Blum G, Verdoes M, Liu H, Syed S, tiplex substrate profiling. Nat Methods 9
Edgington LE, Gheysens O, Miao Z, Jiang H, (11):1095–1100. https://doi.org/10.1038/
Gambhir SS, Bogyo M, Cheng Z (2011) nmeth.2182
Non-invasive imaging of cysteine cathepsin 241. Li Q, Yi L, Hoi KH, Marek P, Georgiou G,
activity in solid tumors using a 64Cu-labeled Iverson BL (2017) Profiling protease specific-
activity-based probe. PLoS One 6(11): ity: combining yeast ER sequestration screen-
e28029. https://doi.org/10.1371/journal. ing (YESS) with next generation sequencing.
pone.0028029 ACS Chem Biol 12(2):510–518. https://doi.
233. Indalao IL, Sawabuchi T, Takahashi E, Kido org/10.1021/acschembio.6b00547
H (2017) IL-1beta is a key cytokine that 242. Fulcher LJ, Hutchinson LD, Macartney TJ,
induces trypsin upregulation in the influenza Turnbull C, Sapkota GP (2017) Targeting
virus-cytokine-trypsin cycle. Arch Virol 162 endogenous proteins for degradation
(1):201–211. https://doi.org/10.1007/ through the affinity-directed protein missile
s00705-016-3093-3 system. Open Biol 7(5). https://doi.org/10.
234. Kelso EB, Lockhart JC, Hembrough T, 1098/rsob.170066
Dunning L, Plevin R, Hollenberg MD, Som- 243. Grossi G, Dalgaard Ebbesen Jepsen M,
merhoff CP, McLean JS, Ferrell WR (2006) Kjems J, Andersen ES (2017) Control of
Therapeutic promise of proteinase-activated enzyme reactions by a reconfigurable DNA
receptor-2 antagonism in joint inflammation. nanovault. Nat Commun 8(1):992. https://
J Pharmacol Exp Ther 316(3):1017–1024. doi.org/10.1038/s41467-017-01072-8
https://doi.org/10.1124/jpet.105.093807 244. Lange PF, Huesgen PF, Overall CM (2012)
235. Vergnolle N (2009) Protease-activated recep- TopFIND 2.0—linking protein termini with
tors as drug targets in inflammation and pain. proteolytic processing and modifications
Pharmacol Ther 123(3):292–309. https:// altering protein function. Nucleic Acids Res
doi.org/10.1016/j.pharmthera.2009.05. 40(Database issue):D351–D361. https://
004 doi.org/10.1093/nar/gkr1025
236. French SL, Hamilton JR (2016) Protease- 245. Wang X, Davies M, Roy S, Kuruc M (2015)
activated receptor 4: from structure to func- Bead based proteome enrichment enhances
tion and back again. Br J Pharmacol 173 features of the protein elution plate (PEP)
(20):2952–2965. https://doi.org/10.1111/ for functional proteomic profiling. Proteomes
bph.13455 3(4):454–466. https://doi.org/10.3390/
237. Kasperkiewicz P, Poreba M, Groborz K, Drag proteomes3040454
M (2017) Emerging challenges in the design 246. Zheng H, Roy S, Soherwardy A, Rahman S,
of selective substrates, inhibitors and activity- Kuruc M (2017) Stroma liquid biopsy—pro-
based probes for indistinguishable proteases. teomic profiles for cancer biomarkers. Poster
FEBS J 284(10):1518–1539. https://doi. reprint first presented at NJ Cancer Retreat,
org/10.1111/febs.14001 May 25, 2017 New Brunswick, NJ, USA
238. Harris JL, Backes BJ, Leonetti F, Mahrus S, 247. Rifai N, Gillette MA, Carr SA (2006) Protein
Ellman JA, Craik CS (2000) Rapid and gen- biomarker discovery and validation: the long
eral profiling of protease specificity by using and uncertain path to clinical utility. Nat Bio-
combinatorial fluorogenic substrate libraries. technol 24(8):971–983. https://doi.org/10.
Proc Natl Acad Sci U S A 97(14):7754–7759. 1038/nbt1235
https://doi.org/10.1073/pnas.140132697 248. Koenig W (2003) Fibrin(ogen) in cardiovas-
239. Kasperkiewicz P, Poreba M, Snipas SJ, cular disease: an update. Thromb Haemost 89
Parker H, Winterbourn CC, Salvesen GS, (4):601–609
392 Ingrid M. Verhamme et al.

249. Dunn EJ, Ariens RA, Grant PJ (2005) The (2010) Alzheimer’s disease peptide beta-
influence of type 2 diabetes on fibrin structure amyloid interacts with fibrinogen and induces
and function. Diabetologia 48 its oligomerization. Proc Natl Acad Sci U S A
(6):1198–1206. https://doi.org/10.1007/ 107(50):21812–21817. https://doi.org/10.
s00125-005-1742-2 1073/pnas.1010373107
250. Grammas P, Martinez JM (2014) Targeting 252. Amara U, Rittirsch D, Flierl M, Bruckner U,
thrombin: an inflammatory neurotoxin in Klos A, Gebhard F, Lambris JD, Huber-Lang
Alzheimer’s disease. J Alzheimers Dis 42 M (2008) Interaction between the coagula-
(Suppl 4):S537–S544. https://doi.org/10. tion and complement system. Adv Exp Med
3233/JAD-141557 Biol 632:71–79
251. Ahn HJ, Zamolodchikov D, Cortes-Canteli-
M, Norris EH, Glickman JF, Strickland S
 Chapter 21

The Use of Combinatorial Hexapeptide Ligand Library (CPLL)

in Allergomics
Youcef Shahali, Hélène Sénéchal, and Pascal Poncet

Abstract
The recent progress of proteomic protocols led to more efficient protein extraction and concentration
procedures to remove nonprotein interfering compounds present in the starting material and to increase
the concentration of underrepresented proteins. Combinatorial hexapeptide ligand libraries (CPLL) were
recently applied to both plant- and animal-derived tissues for capturing the low- and very low-abundance
allergens. Several IgE-binding proteins which were previously absent or poorly represented by using
conventional proteomics tools have been detected and characterized through a CPLL-based approach. In
the present chapter, a protocol based on improved protein extraction and enrichment by CPLL, allowing
the immunochemical characterization of several “hidden allergens” in cypress pollen, is described in detail.

Key words Hexapeptide ligand libraries, Low-abundance allergens, Pollen allergens, Mass spectrom-
etry, Proteomics

1 Introduction

Over the last decade, the study of allergens, defined as allergomics,

allergenomics, or IgE immunoproteomics, increases in relevance
and precision with the development of proteomics tools allowing
the discovery of trace allergens as well as allergenic proteins present
in lower concentrations in various biological samples [1–4]. It is
estimated that no more than 30% of expressed proteins are detect-
able using standard analytical methods [1]. A complete knowledge
of the proteome compositions of complex biological samples is thus
the starting challenge for deciphering allergenic proteins to which
people are exposed and sensitized. Although most known allergens
have an average concentration and could be identified without the
use of enrichment tools, the presence of high-abundance proteins
in extracts from allergenic sources often preclude proper detection
of the entire allergen repertoire responsible for allergic sensitization
[2, 3]. This is particularly important knowing that low- and very

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_21, © Springer Science+Business Media, LLC, part of Springer Nature 2019

393
394 Youcef Shahali et al.

low-abundance allergens can generate strong immunogenic reac-

tions in spite of their low concentration [3–7].
To pave the way for the identification of underrepresented
allergens in allergenic extracts, bead-based combinatorial hexapep-
tide ligand library (CPLL) has been successfully applied to allergo-
mics studies [3–11]. This method has the advantage of being well
compatible with a large panel of profiling techniques, including all
types of gel electrophoresis and mass spectrometry-based identifi-
cations such as MALDI and SELDI or LC-MSMS [12]. Hexapep-
tide library coupled to carrier beads provides non-covalent binding
sites for a broad range of proteins due to their diverse physicochem-
ical properties resulting from their differential combination of
amino acids. As each specific combination is limited to a given
amount of bead volume, hexapeptides having an affinity to high-
abundant proteins are rapidly saturate, whereas low-abundance
proteins are quantitatively trapped and concentrated (see Fig. 1).
Thus, this method circumvents the drawbacks of the depletion or
chromatographic fractionation methods where proteins of interest
could be removed in an uncontrolled manner or diluted and dena-
tured prior to analyses. Therefore, CPLL represent an efficient tool
to get access to low-abundant IgE-binding proteins and a method
of choice for detecting trace allergens in food. Over the last few
years, peptide libraries help to the discovery of novel allergens in
various extracts. One of the first allergenic sources investigated
using CPLL was cow’s milk whey [5]. In this pioneer study, a
number of novel IgE-binding proteins were evidenced after
CPLL treatment, especially a polymorphic immunoglobulin
[5]. The same approach was fruitfully applied by the same authors
to explore the allergen repertoire of other allergy-causing sub-
stances such as latex [6] and maize [7] leading to the identification
of new IgE-binding proteins. Other authors adopted the CPLL
technology for the detection of peanut [8] and casein [9] traces in
baked cookies and wines, respectively. The use of CPLL in allergo-
mics studies also led to the detection of fungal allergens present in
blood of patients with invasive aspergillosis [10] and novel allergen
candidates in hen’s egg [11].
The CPLL technology has been used by our team for the
in-depth exploration of the cypress pollen (CP) allergen composi-
tion [3]. The analysis of CP proteome was known to be difficult due
to the relatively poor protein content of CP extracts and the inher-
ent characteristics of its matrix containing high amounts of inter-
fering compounds such as carbohydrates and pigments
[13, 14]. Protein enrichment by CPLL was performed at three
different pHs (see Fig. 2) for the capture of the largest number of
proteins [15]. Sample treatment using CPLL led to an impressive
extension of the number of identified proteins (see Fig. 3) and the
characterization of novel candidate allergens (see Fig. 4). This study
showed that individual reactions of allergic patients against this
 The Use of Combinatorial Hexapeptide Ligand Library (CPLL) in Allergomics 395

Fig. 1 Schematic illustration of the enrichment process of low-abundant protein species using ProteoMiner
beads. A high number of hexapeptide ligands are used in a single analysis which independently bind specific
proteins until they reach their maximum capacity. Hexapeptides having an affinity to high-abundant proteins
(e.g., P1 and P2) are rapidly saturate, whereas low-abundance proteins (e.g., P5 and P6) are quantitatively
trapped and concentrated. High-abundant proteins exceeding the capacity of the beads are removed

pollen are not necessarily against dominant known proteins. Here,

the CPLL-based sample preparation protocol used in this study is
described in detail.

2 Materials

2.1 Equipment 1. Combinatorial peptide bead libraries are generally provided as

an aqueous solution containing preserving agents against
potential bacterial contaminations. This material could also be
available as powdered dry beads. In both cases, CPLL beads
need to be conditioned prior to use. Prepacked spin-columns
filled with hexapeptide-coupled beads for different starting
protein amounts as well as bulk beads for customized applica-
tions are available under the commercial trademark ProteoMi-
ner™ (Bio-Rad Laboratories, Hercules, CA, USA). In this
study, we used bulk beads in order to capture CP proteins at
three different pHs for the enrichment of the largest number of
proteins as previously described [4].
2. Vortex.
396 Youcef Shahali et al.

Cypress pollen
protein extraction

PUN extract PBS extract

CPLL, pH 7.0 CPLL, pH 7.0

CPLL, pH 4.0 CPLL, pH 4.0

Supernatant CPLL, pH 9.2 CPLL, pH 9.2 Supernatant
discarded discarded

CPLL blend CPLL blend

1-TUC
Protein elutions 2- Distilled water Protein elutions
3- SDS+2-Me
4- UCA

PUN eluate pool PBS eluate pool

Fig. 2 Schematic experimental procedure of pollen extraction, CPLL treatment, and protein elution from the
ProteoMiner beads (adapted from Shahali et al.) [3]. Since the pH influences the protein capture by CPLL,
protein enrichment by CPLL was performed at three different pHs. In order to ensure complete protein
stripping from the beads, four elution steps were sequentially performed

Fig. 3 Overlapping Venn diagram of the total proteins detected by LC-MS/MS analysis. LC-MS analysis was
performed in control, untreated PUN and PBS extracts (blue circle) and in pooled eluates (yellow circle) from
hexapeptide ligand libraries (from PUN and PBS). The green represented common proteins found in the two
types of extraction products. Adapted from Shahali et al. [3]
 The Use of Combinatorial Hexapeptide Ligand Library (CPLL) in Allergomics 397

Fig. 4 Two-dimensional IgE immunoblotting probed with the serum of a cypress pollen allergic patient
(adapted from Shahali et al.) [3]. PBS control extract: a; PBS eluate: c; PUN control extract: b; PUN eluate:
d. Novel IgE-binding were identified in CPLL eluates (outlined in boxes)

3. Refrigerated centrifuge with gravitational force of at least

10,000  g.
4. A set of micropipette.
5. Sonication device.
6. Tube rotator.
7. Magnetic stirrer.

2.2 Pollen Proteins 1. Cupressus sempervirens pollen was supplied by Allergon AB

Extraction (Ängelholm, Sweden).
2. Chemicals composing the PUN buffer: 10 mM phosphate
buffer pH 7.06 containing 3 M urea and 0.2% NP-40 (Non-
idet™ 40) were from Sigma-Aldrich (St Louis, MO, USA).
3. Chemicals composing the phosphate buffered saline (PBS):
150 mM NaCl, 7.8 mM Na2HPO4 and 0.51 mM KH2PO4
pH 7.4 were all from Sigma-Aldrich.
4. Both ethylene diamine tetra acetic acid (EDTA) and phenyl
methane sulfonyl fluoride (PMSF) as anti-protease cocktail
were supplied by Sigma-Aldrich.

2.3 Hexapeptide 1. Ammonium sulfate for protein precipitation was supplied by

Ligand Library Sample Sigma-Aldrich.
Preparation 2. Bulk ProteoMiner™ beads were from Bio-Rad Laboratories.
398 Youcef Shahali et al.

3. Thiourea, urea, 3-[3-cholamidopropyl dimethylammonio]-1-

propansulfonate (CHAPS), acid acetic, 2-mercaptoethanol
(2-Me), ammonia, and sodium dodecyl sulfate (SDS) used for
sample elution were all from Sigma-Aldrich (see Note 1).
4. Elution 1: 2 M thiourea, 7 M urea, and 2% CHAPS in distilled
water (TUC).
5. Elution 2: distilled water.
6. Elution 3: 6% SDS, 2% 2-Me in distilled water (SDS + 2-Me).
7. Elution 4: 8 M urea, 2% CHAPS, and 5% acetic acid pH 3.3
(UCA).
8. Quick Start Bradford Protein Assay from Bio-Rad was used for
protein determination of their concentration.

3 Method

3.1 Pollen Protein 1. Slurry ten times 5 g of pollen (total 50 g) in PUN buffer (see
Extraction Note 2).
2. Shake tubes containing the pollen suspension for 2–3 min to
wet the pollen grains.
3. Sonicate the suspension for 20 s at 4  C.
4. Place the tube on a rotator and gently shake the suspension
overnight at room temperature.
5. Centrifuge the suspension at 10,000  g and 4  C during
15 min.
6. Pipet the supernatant into a new tube, label PUN extract
and date.
7. Add 1 mL of PUN buffer on remaining pollen pellets.
8. Homogenize and centrifuge the suspension at 10,000  g and
4  C during 15 min.
9. Pool the supernatant with the PUN extract previously obtained
in step 6. The overall volume of the PUN extract was 100 mL.
10. Add 30 mL of PBS pH 7.5 to each pollen pellet.
11. Place the tube on a rotator and shake the suspension overnight
at room temperature.
12. Collect the supernatants by centrifugation at 10,000  g and
4  C during 15 min.
13. Pool all PBS extracts. The resulting total volume was 125 mL.

3.2 Removal of 1. Add ammonium sulfate to both PUN and PBS extracts under
Interfering Substances stirring to reach up to 90% saturation or, respectively, 66 g (for
Prior to CPLL Analysis 100 mL) and 82.5 g (for 125 mL).
2. Gently agitate the two mixtures at 4  C overnight.
 The Use of Combinatorial Hexapeptide Ligand Library (CPLL) in Allergomics 399

3. Centrifuge the two mixtures during 30 min at 18,000  g at

4  C.
4. Discard the supernatants.
5. Dissolve the pellets of PUN extract in 10 mL of PUN buffer.
6. Dissolve the pellet of PBS extract in 9 mL of PBS.
7. Add EDTA and PMSF up to 1 mM concentration for both.
8. Dialyze (cutoff 3500 Da) PUN solution against 3 M urea
overnight.
9. Dialyze PBS solution against PBS (1 L) overnight at 4  C (see
Note 3).
10. After dialyzing, add again to the same concentration as before
dialysis protease inhibitors (PMSF and EDTA).
11. Perform Bradford assay on recovered PUN and PBS extracts
and keep them at 20  C before use for analysis.

3.3 Sample 1. Wash three times one hundred microliters of ProteoMiner

Treatment with CPLL beads with 200 μL PUN solution (see Note 4).
2. Centrifuge the suspension at 1000  g for 30–60 s at room
temperature to remove the PUN washing solution.
3. After elimination of washing solution excess, add 30 mL of
PUN protein extract to the beads (see Note 5).
4. Check the pH at 7.0 and shake the mixture under stirring for
2 h at room temperature.
5. Separate beads by centrifugation at 1000  g for 30–60 s at
15  C.
6. Transfer the supernatants to a new tube, label and store at 4  C
until use for a subsequent treatment in different pH (see Note
6).
7. Wash the beads with distilled water and then separate by cen-
trifugation at 1000  g for 30–60 s.
8. Discard excess water and store the bead pellet at 4  C for
further use (see Note 7).
9. Mix the supernatant collected in the step 6 with a second
sample of 100 μL prewashed ProteoMiner beads.
10. Adjust the pH of the suspension to 4.0 by addition of acetic
acid 1 M dropwise.
11. Shake the mixture under stirring for 2 h at room temperature.
12. Separate beads by centrifugation at 1000  g for 30–60 s at
15  C.
13. Transfer the supernatants to a new tube, label, and store at 4  C
until use for a subsequent treatment in different pH.
400 Youcef Shahali et al.

14. Wash the beads with distilled water and then separate the bead
pellet by centrifugation at 1000  g for 30–60 s at 15  C.
15. Discard excess water and store the bead pellet at 4  C for
further use.
16. Mix the supernatant collected in the step 13 with a third
sample of 100 μL prewashed ProteoMiner beads.
17. Adjust the pH of the suspension to 9.25 by dropwise addition
of 4 M ammonia.
18. Gently shake the resulting suspension at room temperature for
2 h.
19. Separate beads by centrifugation at 1000  g for 30–60 s and
discard the supernatant.
20. Wash the beads once with distilled water and then separate the
bead pellet by centrifugation at 1000  g for 30–60 s at 15  C.
21. Discard excess water and store the bead pellet at 4  C for
further use.

3.4 Protein elution 1. Blend together the three bead aliquots of 100 μL each (col-
from CPLL Beads lected from steps 8, 15, and 21 of the CPLL treatment).
2. Rapidly wash once the beads with 600 μL of distilled water and
discard supernatants.
3. Add 500 μL of TUC elution solution to 300 μL bead pellets.
4. Gently shake the mixture for 2 h at room temperature.
5. Centrifuge the mixture at 1000  g for 30–60 s and collect the
supernatant comprising desorbed proteins (eluate “a”).
6. Wash the bead pellet once with 300 μL distilled water and
collect by centrifugation the supernatant (eluate “b”).
7. Submit the resulting bead pellet to the third elution by adding
500 μL of a solution composed of 6% SDS and 2%
2-Mercaptoethanol (SDS+ 2-Me).
8. Gently shake the mixture for 2 h at room temperature.
9. Centrifuge the mixture at 1000  g for 30–60 s and collect the
supernatant (eluate “c”).
10. In order to ensure complete protein stripping from the beads,
perform a fourth elution step by adding the UCA solution to
the bead pellet.
11. Gently shake the mixture for 2 h at room temperature.
12. Centrifuge the mixture at 1000  g for 30–60 s and collect the
supernatant (eluate “d”).
13. Pool the four eluates together (PUN eluate).
14. Adjust the pH to 7.0 and dialyze (cutoff 3500 Da) PUN eluate
against 3 M urea overnight (see Note 8).
 The Use of Combinatorial Hexapeptide Ligand Library (CPLL) in Allergomics 401

15. Perform the Bradford–Lowry standard spectrophotometric

method and store eluates at 20  C for further proteomic
analysis (see Note 9).

4 Notes

1. It is recommended to use freshly prepared buffer when con-

ducting all experiments.
2. Because of its particular structural features and physicochemi-
cal composition, cypress pollen is one of the most difficult
pollens to analyze in terms of protein content and therefore
allergens. In aqueous media with pH 7.5 [15] the external wall
(exine) cracks in a few minutes under the effect of swelling of
the intine (inner wall) particularly rich in polysaccharide. Few
proteins are then extracted in aqueous conditions. The dry
milling may be a good alternative for the extraction of cypress
pollen proteins and to generate fragments of smaller sizes for
experiments and ultrastructural analysis of immunoreactivity.
Good results were achieved by using Minilys homogenizer and
the Precellys kit: 03961-1-003 (1.4 mm ceramic beads) from
Bertin Instruments (Montigny-le-Bretonneux, France) [15].
3. In this step the ionic strength of PBS extracts could be reduced
by dialysis against a solution containing a lower concentration
of sodium chloride, for instance 50 mM instead of 150 mM.
This condition improves the ability of the beads to absorb
proteins and is especially advised when the concentration of
the protein is very low.
4. Beads should be washed with the same solution used for spe-
cific protein extraction. After vortexing, the suspension is cen-
trifuged to remove the excess of supernatant and the bead-
preserving agents. Repeat the washing procedure three times
to be sure that all undesired preserving agents are removed. For
powdered dry beads, it is recommended to slurry 100 mg of
dry beads in 2 mL methanol for 30 min under shaking and then
add 2 mL of phosphate buffer pH 7.5. The rehydration is
performed overnight at room temperature. The rehydrated
beads are then washed extensively, with the same buffer used
for protein extraction as described above.
5. The optimal protein concentration of treated extracts should
be between 1 and 10 mg/mL. For a treatment with 100 μL of
hexapeptide, the total amount of protein should be larger than
50 mg.
6. Since the pH influences largely the protein capture by CPLLs,
it is possible to capture proteins from a biological sample at
different pHs in order to increase the efficacy of the CPLL
402 Youcef Shahali et al.

enrichment [16]. This approach is particularly recommended

for biological samples with low protein content.
7. After protein absorption by CPLL, beads can be stored at 4  C
for a few days before desorbing and protein elution. In this
case, the beads should be equilibrated prior to elution at the
same temperature at which the protein capture has been
carried out.
8. In parallel 100 μL of ProteoMiner beads was washed with
200 μL PBS solution. PBS extract (12 mL) was treated using
the same protocol described above for the PUN extract. Here
also the four eluates (a, b, c, and d) were pooled together (PBS
eluate), neutralized, desalted, lyophilized, and stored at
20  C after the Bradford protein assay.
9. For 2-DE separation, the desired volume of non-treated sam-
ple and the mixture of all eluates were solubilized in the “2-DE
sample buffer” (TUC, 40 mM Tris-acetate) to a final concen-
tration of 2 mg/mL protein.

Acknowledgments

Collaboration with Drs. Egisto Boschetti and Pier Giorgio Righ-

etti, the pioneer developers of the CPLL technology, is gratefully
acknowledged and was at the basis of this protocol and its adapta-
tion to allergomics studies.

References

1. Boschetti E, Righetti PG (2013) 6. D’Amato A, Bachi A, Fasoli E, Boschetti E,

Low-abundance proteome discovery: state of Peltre G, Sénéchal H et al (2010) In-depth
the art and protocols. Newnes exploration of Hevea brasiliensis latex prote-
2. Shahali Y, Sutra JP, Peltre G, Charpin D, ome and “hidden allergens” via combinatorial
Sénéchal H, Poncet P (2010) IgE reactivity to peptide ligand libraries. J Proteome
common cypress (C. sempervirens) pollen 73:1368–1380
extracts: evidence for novel allergens. World 7. Fasoli E, Pastorello EA, Farioli L, Scibilia J,
Allergy Organ J 3:229–234 Aldini G, Carini M et al (2009) Searching for
3. Shahali Y, Sutra JP, Fasoli E, D’Amato A, Righ- allergens in maize kernels via proteomic tools. J
etti PG, Futamura N et al (2012) Allergomic Proteome 72:501–510
study of cypress pollen via combinatorial pep- 8. Pedreschi R, Nørgaard J, Maquet A (2012)
tide ligand libraries. J Proteome 21:101–110 Current challenges in detecting food allergens
4. Righetti PG, Fasoli E, D’Amato A, Boschetti E by shotgun and targeted proteomic
(2014) The “dark side” of food stuff proteo- approaches: a case study on traces of peanut
mics: the CPLL-marshals investigate. Foods allergens in baked cookies. Nutrients
3:217–237 4:132–150
5. D’Amato A, Bachi A, Fasoli E, Boschetti E, 9. D’Amato A, Kravchuk AV, Bachi A, Righetti
Peltre G, Sénéchal H et al (2009) In-depth PG (2010) Noah’s nectar: the proteome con-
exploration of cow’s whey proteome via com- tent of a glass of red wine. J Proteome
binatorial peptide ligand libraries. J Proteome 73:2370–2377
Res 8:3925–3936 10. Fekkar A, Pionneau C, Brossas JY, Marinach-
Patrice C, Snounou G, Brock M, Mazier D
 The Use of Combinatorial Hexapeptide Ligand Library (CPLL) in Allergomics 403

(2012) DIGE enables the detection of a puta- between microarray and immunoblot for the
tive serum biomarker of fungal origin in a comparative evaluation of IgE repertoire of
mouse model of invasive aspergillosis. J Prote- French and Italian cypress pollen allergic
ome 75:2536–2549 patients. Folia Biologica (Prague) 60:192
11. Martos G, López-Fandiño R, Molina E (2013) 14. Danti R, Della Rocca G, Calamassi R, Mori B,
Immunoreactivity of hen egg allergens: influ- Mariotti Lippi M (2011) Insights into a hydra-
ence on in vitro gastrointestinal digestion of tion regulating system in Cupressus pollen
the presence of other egg white proteins and grains. Ann Bot 108:299–306
of egg yolk. Food Chem 136:775–781 15. Shahali Y (2011) Etude analytique de l’allergie
12. Hartwig S, Lehr S (2012) Combination of au pollen de cyprès: aspects moléculaires et
highly efficient hexapeptide ligand library- particulaires, Thesis Université Paris VI, Pierre
based sample preparation with 2D DIGE for et Marie Curie, Paris, France, p 220
the analysis of the hidden human serum/ 16. Fasoli E, Farinazzo A, Sun CJ, Kravchuk AV,
plasma proteome. Methods Mol Biol Guerrier L, Fortis F et al (2010) Interaction
854:169–180 among proteins and peptide libraries in prote-
13. Shahali Y, Nicaise P, Brazdova A, Charpin D, ome analysis: pH involvement for a larger cap-
Scala E, Mari A et al (2014) Complementarity ture of species. J Proteome 73:733–742
 Chapter 22

Efficient Extraction and Digestion of Gluten Proteins

Haili Li, Keren Byrne, Crispin A. Howitt, and Michelle L. Colgrave

Abstract
Coeliac disease (CD) is a T-cell mediated autoimmune disorder triggered by ingestion of cereal gluten
found in wheat (gliadins and glutenins), barley (hordeins), and rye (secalins). As the only treatment for CD
is a lifelong gluten-free diet, the measurement of gluten in raw ingredients and processed food products is
critical to protecting people with CD or gluten intolerance. The most commonly employed method is the
enzyme-linked immunosorbent assay (ELISA), but more recently mass spectrometry has been employed
wherein the extracted gluten proteins are digested to peptides that are then directly measured. To achieve
the goal of accurate gluten quantitation, gluten must be efficiently extracted from the ingredient or food
matrix and then digested to yield the peptides that are monitored by LC-MS. In this chapter, a rapid,
simple, and reproducible protocol for extraction and digestion of gluten proteins is described.

Key words Flour, Gluten, Digestion, Trypsin, Chymotrypsin, Mass spectrometry

1 Introduction

Coeliac disease (CD) is a disease of the small intestine that occurs in

genetically susceptible subjects triggered by the ingestion of cereal
gluten proteins. The only treatment is strict adherence to a lifelong
gluten-free diet [1, 2]. Gluten is the collective name for a class of
proteins found in wheat, rye, and barley that are the elicitors of CD,
an autoimmune disorder with a prevalence of 1% worldwide
[3–5]. Establishing accurate methods for gluten measurement is
of critical importance to the health of those affected by CD or non-
coeliac gluten sensitivity (NCGS) [6]. Enzyme-linked immunosor-
bent assays (ELISAs) are the currently accepted method for the
detection of gluten in food [7]. One drawback of ELISAs is that
they cannot adequately quantify gluten that has been highly hydro-
lyzed [8]. A number of approaches have been developed with mass
spectrometry (MS) showing great promise in gluten measurement
owing to its specificity, sensitivity, ability to multiplex and identify
hydrolyzed gluten [2, 9–13]. The successful application of bottom-

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_22, © Springer Science+Business Media, LLC, part of Springer Nature 2019

405
406 Haili Li et al.

up proteomics to the analysis of gluten critically depends upon the

efficiency and reproducibility of proteolytic digestion of gluten
from the grain or highly processed food products.
In this chapter, a rapid, simple, and reproducible method for
extraction and digestion of gluten proteins is described. Using this
method, raw ingredients and processed food products may be
analyzed enabling detection and relative quantitation using liquid
chromatography-mass spectrometry (LC-MS/MS). This method
has been found to be applicable to all food matrices tested to date.

2 Materials

Prepare all solutions using MilliQ water (prepared by purifying

deionized water to attain a resistance of 18 MΩ-cm at 25  C) and
analytical grade reagents. Prepare and store all reagents at room
temperature (unless indicated otherwise). Follow all waste disposal
regulations when disposing of waste materials and check Material
Safety Data Sheets (MSDS) for reagents prior to use. All reagents are
prepared immediately prior to use. The isopropanol (IPA) is HPLC
grade. All other reagents are the highest possible commercial grade
available (see Note 1). The volumes described below are suitable for
the analysis of 10 samples with four replicates per sample.

2.1 Chemicals 1. Extraction buffer: 55% isopropanol (IPA)/2% dithiothreitol

(DTT) (see Note 2). Prepare 200 μL per sample (replicate).
Prepare 10 mL of 55% IPA/2% DTT by combining 5.5 mL IPA
with 4.5 mL water and 200 mg DTT.
2. Urea (UA) buffer: 8 M urea in 0.1 M Tris–HCl (see Note 3).
Prepare 1 mL per 1 sample (replicate). Prepare 50 mL of UA
buffer by weighing 24 g of urea and dissolving in 45 mL of water.
To this solution, add 5 mL of 1 M Tris–HCl, pH 8.5 (see Note 4).
3. Iodoacetamide (IAM) solution: 0.05 M IAM in UA (see Note
5). Prepare 0.1 mL per 1 sample (replicate). Prepare 5 mL by
weighing 46.2 mg and dissolving in UA buffer.
4. Ammonium bicarbonate: 50 mM NH4HCO3 in water, pH 8.0
(see Note 6). Prepare 0.5 mL per 1 sample (replicate).
5. Trypsin: 0.25 mg/mL in 50 mM ammonium bicarbonate, 1 mM
CaCl2 (see Note 7). Prepare 0.2 mL per sample (replicate).
6. Chymotrypsin: 0.25 mg/mL in 50 mM ammonium bicarbon-
ate, 1 mM CaCl2 (see Note 8). Prepare 0.2 mL per sample
(replicate).

2.2 Equipment 1. 10 kDa MWCO filters (e.g., Millipore: catalog number

UFC5010BK).
2. Benchtop centrifuge (e.g., Eppendorf, model: 5415R), tem-
perature 25  C.
 Gluten Extraction and Digestion 407

3. Wet chamber with a rack for eppendorf tubes.

4. Thermal mixer set to 50  C.

2.3 HPLC 1. HPLC buffer A: 0.1% formic acid, 99.9% water. Mix by inver-
Components sion (see Note 9).
2. HPLC buffer B: 0.1% formic acid, 90% acetonitrile, 9.9% water.
Mix by inversion.

3 Methods

3.1 Gluten Extraction 1. Weigh 20 mg of flour (or milled food product) into a 1.5 mL
eppendorf tube.
2. Add 200 μL of 55% IPA/2% DTT and vortex the tube until
flour is properly mixed with solution (see Note 10).
3. Place the tube in an ultrasonic bath for 5 min at room
temperature.
4. Put the tube in a dry block heater at 50  C for 30 min (see Note 11).
5. Centrifuge the suspension for 10 min at 20,800  g.
6. Transfer the supernatant, containing the gluten, into a fresh tube.

3.2 Protein Digestion 1. Transfer 100 μL of gluten extract to a 10 kDa MWCO filter,
add 100 μL of UA buffer, and centrifuge at 20,800  g for
15 min.
2. Wash the protein by addition of 200 μL of UA buffer to the
filter unit. Centrifuge at 20,800  g for 15 min.
3. Add 100 μL of IAM solution and incubate at room tempera-
ture for 20 min in the dark (see Notes 5 and 12).
4. Centrifuge the filter units at 20,800  g for 15 min.
5. Add 200 μL of UA to the filter unit and centrifuge at
20,800  g for 15 min to remove excess IAM. Discard the
flow-through from the collection tube.
6. Exchange the buffer (see Note 13) by adding 200 μL of 50 mM
ammonium bicarbonate to the filter unit and centrifuge at
20,800  g for 15 min. Repeat.
7. Transfer the filter units to new collection tubes. Digest the
protein by adding 200 μL of 0.25 mg/mL trypsin or chymo-
trypsin (in 50 mM ammonium bicarbonate, 1 mM CaCl2) and
mix briefly at low speed (400 rpm). Incubate the units in a wet
chamber at 37  C overnight (~18 h).
8. Centrifuge the filter units at 20,800  g for 15 min to collect
the digested peptides. Wash the filter by adding 200 μL of
50 mM ammonium bicarbonate and centrifuge the filter units
at 20,800  g for 15 min.
9. Lyophilize the filtrate in a vacuum centrifuge and store at
20  C until analysis.
408 Haili Li et al.

3.3 Assessment of 1. Reconstitute samples in 100 μL of 1% formic acid immediately

Digestion Efficiency prior to analysis.
2. The peptide fractions (5.0 μL) can be chromatographically
separated on a reverse-phase (RP) high-performance liquid
chromatography (HPLC) system. In this example, we describe
the use of a Shimadzu Nexera UHPLC system. The HPLC
eluate can be directly coupled to a mass spectrometer. In this
example, we describe the use of a QTRAP 6500 MS/MS
(SCIEX, Redwood City, CA, USA).
3. The peptides were separated on a Phenomenex Kinetex C18
(1.7 μm, 100 Å, 150 mm  2.1 mm) column at a flow rate of
400 μL/min. A linear gradient from 5% to 45% solvent B over
10 min was employed followed by 45–80% B over 1 min, a
1 min hold at 80% B, return to 5% B over 0.1 min, and a 3 min
re-equilibration.
4. Relative quantitation was achieved using scheduled multiple
reaction monitoring (MRM) scanning experiments (Table 1)
using a 40 s detection window for each MRM transition and a
0.3 s cycle time (see Note 14). The ion spray voltage was set to
5500 V, the curtain gas was set to 35 psi, ion source gas 1 and
2 (GS1 and GS2) to 40 and 50 psi, and the heated interface was
set to 500  C (see Note 15). Spectra were acquired using the
manufacturer’s rolling collision energy (CE) based on the size
and charge of the precursor ion for optimum peptide fragmen-
tation (see Note 16).

Table 1
Gluten peptide markers used in the assessment of digestion efficiency. All transitions for a given
peptide are summed to give the peak area

Uniprot RT Q3 m/z CE
# Peptide sequence accession (min) Q1 m/z (z) (fragment, z) (V)
G1 ELQESSLEAC(cam)R Barley: I6TRS8 3.15 661.296 (2+) 735.325 (y6, 1+) 33.1
Wheat: P10387 822.357 (y7, 1+)
Rye: Q94IL1 951.400 (y7, 1+)
G2 AQQLAAQLPAMC Barley: I6TRS8 4.90 729.361 (2+) 747.343 (y6, 1+) 36.5
(cam)R Wheat: P08489 946.439 (y8, 1+)
Rye: D3XQB7 1017.476 (y9, 1+)
G2* AQQLAAQLPAM(ox)C Barley: I6TRS8 4.00 737.361 (2+) 763.343 (y6, 1+) 36.9
(cam)R Wheat: P08489 962.439 (y8, 1+)
Rye: D3XQB7 1033.476 (y9, 1+)
RT retention time (min), Q1 precursor ion m/z with charge, z, Q3 product ion m/z with fragment ion assignment and
charge, z, CE collision energy (V)
 Gluten Extraction and Digestion 409

5. Peaks were integrated using MultiQuant v3.0 (SCIEX) (see

Note 17) wherein all three transitions were required to
co-elute at the same retention time (RT, min) with a signal-
to-noise (S/N) > 3 for detection and a S/N > 5 and intensity
>1000 counts per second (cps) for quantitation. Figure 1
shows an example LC-MS chromatogram for the detection of
the peptides in barley (a), wheat (b), and rye (c).

Fig. 1 LC-MRM-MS analysis of gluten-derived peptides commonly derived from

barley (a), wheat (b), and rye (c). The peptides are ELQESSLEACR (G1),
AQQLAAQLPAMCR (G2), and its oxidized form AQQLAAQLPAM(ox)CR (G2*)
410 Haili Li et al.

6. The peak areas of the three MRM transitions monitored were

summed and the data for the four replicates should be assessed
by examining the mean, standard deviation (SD), and coeffi-
cient of variation (CV), wherein the CV should be <10%.

4 Notes

1. Isopropanol (propan-2-ol, IPA) is highly flammable (keep away

from naked flames) and is toxic by inhalation, ingestion, or in
contact with skin. DTT and ammonium bicarbonate are harm-
ful by ingestion—do not breathe dust. All chemicals used here
should be used in well-ventilated areas and avoid contact by
using personal protective equipment.
2. Always prepare fresh extraction buffer.
3. Urea is a powerful protein denaturant that disrupts noncova-
lent bonds in the proteins. Urea is used to increase the solubil-
ity of proteins. Urea is a hygroscopic material (absorbs water)
and as such we recommend purchasing and storing it in small
quantities to avoid large discrepancies in the concentration of
prepared solutions.
4. A stock solution of 1 M Tris–HCl may be prepared and stored
for 1 month. Prepare by weighing 78.8 g of Tris–HCl and
dissolving in 450 mL of water. Adjust the pH of the solution
to 8.5 using HCl and then make up to a final volume of
500 mL.
5. IAM is light sensitive and should be prepared immediately prior
to use and stored wrapped in foil until use.
6. Ammonium bicarbonate may be prepared and stored for
1 month. Prepare by weighing 1.98 g and dissolving in
450 mL of water. Adjust the pH of the solution to 8.0 using
HCl and then make up to a final volume of 500 mL. Both
trypsin and chymotrypsin are maximally active over the pH
range 7–9.
7. The recommended range for protein digestion by trypsin is
100:1 to 20:1 with incubation at 30–37  C. A protein-to-
enzyme ratio of 20:1 is employed for gluten digestion owing
to its recalcitrant nature.
8. Chymotrypsin is both activated and stabilized by Ca2+. The
recommended range for protein digestion by trypsin is 200:1
to 20:1 with incubation at 30–37  C. A protein-to-enzyme
ratio of 20:1 is employed for gluten digestion owing to its
recalcitrant nature.
9. Formic acid is a strong acid, is corrosive, and may cause burns
by inhalation, ingestion, or contact. Use in a fume hood with
 Gluten Extraction and Digestion 411

appropriate personal protective equipment. The HPLC buffers

should be degassed if the HPLC to be used is not fitted with a
degassing module. This can be achieved by sonication of the
solutions in an ultrasonic bath for 10 min.
10. The mixing step is critical to gluten extraction. It is important
to vortex mix the sample immediately after solvent addition to
wet the flour and avoid clumping of flour.
11. The extraction step also serves to reduce the disulfide bonds
between the cysteines (intra- and intermolecular linkages). It is
preferable to use a thermal mixer for the extraction step at a
low speed, e.g., 400 rpm.
12. The addition of IAM serves to irreversibly alkylate the cysteines
preventing reoxidation.
13. It is important to wash away excess reagents. IAM can modify
additional sites, for example, tyrosine. Urea interferes with
protein digestion above concentrations of 2 M.
14. MRM methods comprise lists of precursor ion m/z values that
represent the peptides (also known as the Q1 ions) and product
ion m/z values that represent the peptide fragment ions (also
known as the Q3 ions). Mass selection at two levels renders
greater specificity to MRM methods.
15. The source conditions (gas and voltage settings) used are
dependent on the instrument, the flow rate, and the solvents
employed. The parameters defined here may act as a starting
point but should be optimized using standards to achieve the
best sensitivity.
16. The collision energy (CE) is increased in a linear manner as the
m/z increases for each charge state. Doubly charged ions
require a higher CE than triply charged ions.
17. A number of software packages that allow peak integration are
available. MultiQuant™ is available from SCIEX as a licensed
product; however, there are software packages available free of
charge, such as Skyline as available on https://skyline.ms.

Acknowledgments

This work was supported by a fellowship to HL “International

Training for High-level Talent in 2016” (YUWAIZHUAN
[2016] No.8) from Foreign Experts Bureau of Henan Province.

References
1. Guandalini S, Assiri A (2014) Celiac disease: a 2. Colgrave ML, Byrne K, Blundell M et al
review. JAMA Pediatr 168(3):272–278 (2016) Comparing multiple reaction monitor-
ing and sequential window acquisition of all
412 Haili Li et al.

theoretical mass spectra for the relative quanti- 9. Colgrave ML, Byrne K, Blundell M et al
fication of barley gluten in selectively bred bar- (2016) Identification of barley-specific peptide
ley lines. Anal Chem 88(18):9127–9135 markers that persist in processed foods and are
3. Stamnaes J, Sollid LM (2015) Celiac disease: capable of detecting barley contamination by
autoimmunity in response to food antigen. LC-MS/MS. J Proteome 147:169–176
Semin Immunol 27(5):343–352 10. Colgrave ML, Goswami H, Byrne K et al
4. Fasano A, Berti I, Gerarduzzi T et al (2003) (2015) Proteomic profiling of 16 cereal grains
Prevalence of celiac disease in at-risk and not- and the application of targeted proteomics to
at-risk groups in the United States: a large detect wheat contamination. J Proteome Res
multicenter study. Arch Intern Med 163 14(6):2659–2668
(3):286–292 11. Fiedler KL, McGrath SC, Callahan JH et al
5. Lionetti E, Gatti S, Pulvirenti A et al (2015) (2014) Characterization of grain-specific pep-
Celiac disease from a global perspective. Best tide markers for the detection of gluten by mass
Pract Res Clin Gastroenterol 29(3):365–379 spectrometry. J Agric Food Chem 62
6. Catassi C, Bai JC, Bonaz B et al (2013) (25):5835–5844
Non-celiac gluten sensitivity: the new frontier 12. Gomaa A, Boye J (2015) Simultaneous detec-
of gluten related disorders. Nutrients 5 tion of multi-allergens in an incurred food
(10):3839–3853 matrix using ELISA, multiplex flow cytometry
7. Koerner TB, Abbott M, Godefroy SB et al and liquid chromatography mass spectrometry
(2013) Validation procedures for quantitative (LC-MS). Food Chem 175:585–592
gluten ELISA methods: AOAC allergen com- 13. Sealey-Voyksner JA, Khosla C, Voyksner RD
munity guidance and best practices. J AOAC et al (2010) Novel aspects of quantitation of
Int 96(5):1033–1040 immunogenic wheat gluten peptides by liquid
8. Thompson T, Mendez E (2008) Commercial chromatography-mass spectrometry/mass
assays to assess gluten content of gluten-free spectrometry. J Chromatogr A 1217
foods: why they are not created equal. J Am (25):4167–4183
Diet Assoc 108(10):1682–1687
 Chapter 23

Glycosylation Profiling of Tumor Marker in Plasma Using

Bead-Based Immunoassay
Hongye Wang, Zheng Cao, Hu Duan, and Xiaobo Yu

Abstract
As one of the most important posttranslational modifications, glycosylation plays critical roles in protein
folding, trafficking, cell differentiation, immune recognition, etc. The alteration of glycosylation is closely
associated with the pathological processes during and after caner development, and thus holds great value in
cancer detection. In this chapter, we describe a protocol on the glycosylation profiling of tumor marker in
plasma using bead-based immunoassay with CA125 as a model, including bead coupling, coupling control,
glycosylation assay, as well as the plasma screening for breast cancer patients. This protocol can be used to
profile the glycosylation of protein markers in clinical plasma or serum samples for different human cancers.

Key words Posttranslational modification, Glycosylation, Tumor marker, Plasma, Lectin, Bead-based
immunoassay

1 Introduction

As a major public health issue in the world, cancer has remained as

the leading cause of death in China [1]. The early detection is
probably the optimal approach to increase the opportunity for the
treatment and control of this disease [2–4]. However, the sensitiv-
ity and specificity of most cancer biomarkers are still limited and it is
urgent to identify new biomarkers that can improve the cancer
detection [5, 6]. Glycosylation is one of the most important post-
translational modifications that plays critical roles in protein fold-
ing, trafficking, cell differentiation, immune recognition, etc.
[7]. The alteration of glycosylation is closely associated with the
pathological processes during and after the cancer development.
Many tumor protein markers have been found that contain altered
glycosylation expression, such as CA19-9, CA15-3, and AFP-L3
[8, 9]. Thus, the method that can detect and quantify glycosylation

Hongye Wang and Zheng Cao contributed equally to this work.

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_23, © Springer Science+Business Media, LLC, part of Springer Nature 2019

413
414 Hongye Wang et al.

on tumor markers would hold great value in improving cancer

diagnosis and prognosis [6, 10, 11].
Lectins preferably bind to a specific terminal mono- or polysac-
charide structure and have been widely used as affinity reagent for
the capture and detection of glycosylated proteins [12]. Chen et al.
developed a multiplexed assay that enables the detection of protein
expression and glycosylation in high-throughput on antibody
microarrays using protein detection antibody and lectins, respec-
tively. The results indicated that the changes of protein glycosyla-
tion on MUC1 and CEA are associated to pancreatic cancer
[13]. Using similar strategy, Li et al. coupled the antibody to
magnetic beads and detected the expression and glycosylation of
serological TIMP-1 (tissue inhibitor of metallopeptidase 1) and
DPP-4 (membrane metallo-endopeptidase and dipeptidyl pepti-
dase-IV) using biotinylated antibody and lectins, respectively. The
system provides a useful tool for the validation of glycoproteins as
biomarkers for cancers [14].
Our research is focused on the glycosylation of tumor markers
in the improvement of cancer detection, prognosis, as well as
therapeutic treatment [6, 15]. Using CA125 as a model, here we
describe a protocol on the glycosylation profiling of tumor marker
in plasma using bead-based immunoassay, including beads cou-
pling, coupling control, glycosylation assay, as well as the plasma
screening for breast cancer patients (Fig. 1). This assay described in
this chapter can be used to profile the glycosylation of protein
markers in clinical plasma or serum samples for different human
cancers.

2 Materials

2.1 Reagents 1. Beads: Luminex MegPlex beads (Luminex Inc., Austin, TX,
USA).
2. Activation buffer: 0.1 M NaH2PO4 (Sigma-Aldrich, St. Louis,
MO, USA), pH 6.2.
3. EDC solution: 50 mg/mL EDC (1-ethyl-3-[3-dimethylami-
nopropyl] carbodiimide hydrochloride, Thermo Fisher Scien-
tific, IL, USA).
4. Sulfo-NHS solution: 50 mg/mL sulfo-NHS (Thermo Fisher
Scientific, IL, USA).
5. Coupling buffer: 50 mM MES, pH 5.0 (Sigma-Aldrich,
St. Louis, MO, USA).
6. Assay/Washing buffer: 0.05% (v/v) Tween 20 in 0.01 M PBS.
7. Blocking buffer: PBS-TBN, 1% BSA in washing buffer.
8. Lectin: Biotinylated Aleuria Aurantia Lectin (AAL), Biotiny-
lated Phaseolus vulgaris Erythroagglutinin (PHA-E),
 Glycosylation Profiling of Tumor Marker in Plasma Using Bead-Based Immunoassay 415

y
od
ra d
ib
ke te
nt

Quantification
ar la
Add Cy3 labeled

m tiny

Protein
streptavidin

or io
m b
tu d
ti- Ad
An
Add Cy3 labeled
1 streptavidin
d c tin
Ad d-le
l a te
Beads
t iny
coupling Plasma bio

Detections of Protein
bio
tiny Add

Glycosylation
Magnetic Anti-tumor late
beads marker d-le Add Cy3 labeled
ctin
2 streptavidin

Ad
capture

db
antibody

iot
iny
lat
ed
-le
Add Cy3 labeled

cti
n3
streptavidin

Fig. 1 Schematic illustration of bead-based immunoassay on glycosylation profiling of tumor marker in

plasma. The antitumor marker antibody is coupled to the magnetic beads modified with carboxyl groups
through EDC/NHS chemistry, which is then incubated with the plasma or serum from cancer patients. The
tumor proteins captured to the beads can be detected by either biotinylated anti-protein antibody for protein
quantification or biotinylated lectins for glycosylation profiling, respectively. At last, the fluorescent signal can
be detected after incubation with Cy3-labeled streptavidin through Luminex instrument or flow cytometry

Biotinylated Vicia villosa Lectin (VVL, VVA) (Vector Labora-

tories Inc., Burlingame, CA, USA).
9. Cy3-Streptavidin (Jackson ImmunoResearch Inc., West Grove,
PA, USA).
10. Recombinant Human CA125/MUC16 Protein (R&D sys-
tems Inc., Minneapolis, MN, USA).
11. Anti-CA125/MUC16 mAb (Abcam Inc., Cambridge, MA,
USA).
12. Goat anti-Mouse IgG-Alexa555 (Thermo Fisher Scientific, IL,
USA).

3 Methods

3.1 Beads Coupling 1. Resuspend the stock beads by vortex for 30 s and followed with
sonication for 20 s (see Note 1).
2. Take 100 μL of 1.25  107 beads into a microcentrifuge tube.
Put the tube into a magnetic separator and let it stay for 1 min
(see Note 2).
3. Remove the supernatant and resuspend the beads in 100 μL
dH2O by vortex for 30 s.
416 Hongye Wang et al.

4. Keep the tube in a magnetic separator for 1 min and remove the
supernatant without disturbing the beads.
5. Resuspend the beads in 80 μL activation buffer by vortex for
30 s.
6. Prepare 50 mg/mL Sulfo-NHS and EDC diluted in dH2O.
7. Add 10 μL 50 mg/mL Sulfo-NHS to the beads and mix gently
by vortex.
8. Add 10 μL 50 mg/mL EDC to the beads and mix gently by
vortex.
9. Incubate for 20 min at room temperature with gentle mixing.
10. Put the tube into a magnetic separator for 1 min and remove
the supernatant.
11. Wash the beads with 250 μL coupling buffer twice.
12. Resuspend the beads in 100 μL washing buffer and add 5 μg
anti-CA125 antibody. Incubate for 2 h at room temperature
with gentle mixing (see Note 3).
13. Wash the coupled beads with 500 μL blocking buffer, then
incubate for 30 min at room temperature.
14. Wash the blocked beads with 500 μL washing buffer and
resuspend the beads in 1000 μL washing buffer.
15. Count the number of beads using BD FACSVerse (see Note 4).

3.2 Quality Control of 1. Dilute the mouse anti-CA125 antibody-coupled beads to a

Anti-CA125 Antibody- final concentration of 50 beads/μL with assay buffer and add
Coupled Magnetic 50 μL to a 96-well plate (pre-blocked with PBS-TBN buffer) in
Beads duplicate.
2. Incubate beads with Alexa555 conjugated goat anti-Mouse
IgG with 0.125, 0.25, 0.5, 1, 2, 4 μg/mL for 30 min at
room temperature in a 96-well plate (pre-blocked with
PBS-TBN buffer), with assay buffer as a blank control (see
Note 5).
3. Put the 96-well plate on a magnetic separator for 3 min and
remove the supernatant.
4. Wash the beads with 150 μL washing buffer twice and resus-
pend the beads in 100 μL washing buffer.
5. Submit the resulting beads to a Luminex200 instrument for
the fluorescent measurement. The representative result of cou-
pling control is shown in Fig. 2.

3.3 Detection of 1. Dilute the coupled beads to a final concentration of 50 beads/μ

CA125’s Expression in L with assay buffer and add 50 μL to a 96-well plate
Plasma Using (pre-blocked with PBS-TBN buffer) with duplicate.
Sandwich 2. Incubate beads with a series of concentrations of CA125 pro-
Immunoassay tein in matrix for 2 h at room temperature in 96-well plate
(pre-blocked with PBS-TBN buffer) in duplicate.
Glycosylation Profiling of Tumor Marker in Plasma Using Bead-Based Immunoassay 417

Fig. 2 Quality control of antibody-coupled magnetic beads. MFI is the median

fluorescence intensity of 100 counted magnetic beads

Fig. 3 Dose-response curve for CA125 detection using bead-based sandwich

immunoassay. MFI is the median fluorescence intensity of 100 counted
magnetic beads

3. Put the 96-well plate on a magnetic separator for 3 min and

remove the supernatant.
4. Wash the beads with 150 μL washing buffer three times.
5. Add 100 μL CA125 detection antibody (4 μg/mL) and incu-
bate for 1 h at room temperature.
6. Wash the beads with 150 μL washing buffer three times.
7. Add 100 μL goat anti-mouse IgG-Alexa555 (4 μg/mL) and
incubate for 1 h at room temperature.
8. Wash the beads with 150 μL washing buffer twice and resus-
pend the beads in 100 μL washing buffer.
9. Submit the resulting beads to a Luminex200 instrument for
the fluorescent measurement. The detection of different con-
centrations of CA125 using sandwich immunoassay is shown in
Fig. 3.
418 Hongye Wang et al.

3.4 Glycosylation 1. The plasma samples are thawed at 4  C and centrifuged for
Profiling of CA125 in 10 min at 14,000  g (see Note 6).
Plasma for Breast 2. Dilute the coupled beads to a final concentration of 50 beads/μ
Cancer Patients L with assay buffer and add 50 μL to a 96-well plate
(pre-blocked with PBS-TBN buffer) in duplicate.
3. Add 50 μL plasma sample and mix with Eppendorf MixMate at
390  g. Incubate for 2 h at room temperature.
4. Wash the beads with 150 μL washing buffer three times.
5. Add 100 μL biotinylated-lectin(2 μg/ml) and incubate for 1 h
at room temperature.
6. Wash the beads with 150 μL washing buffer three times.
7. Add 100 μL Cy3-Streptavidin (4 μg/mL) to each well and
incubate at room temperature. For 1 h on an Eppendorf Mix-
Mate at 390  g.
8. Wash the beads with 150 μL washing buffer twice and resus-
pend the beads in 150 μL washing buffer.
9. Submit the resulting beads to a Luminex200 instrument for
the fluorescent measurement (see Note 7). The representative
results of plasma screening are shown in Fig. 4, in which
differential profiling of CA125 glycosylation can be observed
with the lectins of AAL, PHA-E, and VVA.

4 Notes

1. This step is essential to prevent beads aggregation.

2. Minimize exposure time, as prolonged exposure to light might
result in fluorescent quenching. During incubation, the
96-well plate should be covered with aluminum foil.
3. Do not make use of solution containing Tris or other amine-
based buffers, which could decrease coupling efficiency.
4. The number of coupled beads can also be counted by cell
counter or hemocytometer.
5. During incubation, 96-well plate should be put on the shaker
to reduce the precipitation and aggregation of the beads.
6. We highly recommend that the plasma or serum samples are
thawed at 4  C, and followed with centrifugation for 10 min at
14,000  g to remove the precipitation in samples.
7. This method is suitable for the detection of different glycosyla-
tion on tumor protein markers. The color-coded beads enable
researchers to screen up to 100 parameters in a single
experiment.
 Glycosylation Profiling of Tumor Marker in Plasma Using Bead-Based Immunoassay 419

Fig. 4 Glycosylated profiling of CA125 in the plasma of breast cancer patients using biotinylated lectins. All
plasma samples were collected with the written informed consent under the approval of intuitional review
board (IRB) from Beijing Proteome Research Center

Acknowledgments

This project is financially supported from the National Natural

Science Foundation of China (81673040), the State Key Labora-
tory of Proteomics (SKLP-O201504 and SKLP-K201505) to X.Y.

References
1. Chen W et al (2016) Cancer statistics in China, multi-society task force on colorectal cancer,
2015. CA Cancer J Clin 66:115–132 and the American College of Radiology. CA
2. Levin B et al (2008) Screening and surveillance Cancer J Clin 58:130–160
for the early detection of colorectal cancer and 3. Cuzick J et al (2014) Prevention and early
adenomatous polyps, 2008: a joint guideline detection of prostate cancer. Lancet Oncol
from the American Cancer Society, the US 15:e484–e492
420 Hongye Wang et al.

4. Wang D, Yang L, Zhang P, et al (2017) AAgA- 10. Hanash SM, Pitteri SJ, Faca VM (2008)
tlas 1.0: a human autoantigen database. Mining the plasma proteome for cancer bio-
Nucleic acids research 45 (D1): D769–D776 markers. Nature 452:571–579
5. Hanash SM, Baik CS, Kallioniemi O (2011) 11. Wu L, Qu X (2015) Cancer biomarker detec-
Emerging molecular biomarkers—blood- tion: recent achievements and challenges.
based strategies to detect and monitor cancer. Chem Soc Rev 44:2963–2997
Nat Rev Clin Oncol 8:142–150 12. Syed P et al (2016) Role of lectin microarrays in
6. Yu X, Schneiderhan-Marra N, Joos TO (2010) cancer diagnosis. Proteomics 16:1257–1265
Protein microarrays for personalized medicine. 13. Chen S et al (2007) Multiplexed analysis of
Clin Chem 56:376–387 glycan variation on native proteins captured by
7. Wang JR et al (2017) A method to identify antibody microarrays. Nat Methods 4:437–444
trace sulfated IgG N-glycans as biomarkers for 14. Li D, Chiu H, Chen J, Zhang H, Chan DW
rheumatoid arthritis. Nat Commun 8:631 (2013) Integrated analyses of proteins and
8. Li D, Mallory T, Satomura S (2001) AFP-L3: a their glycans in a magnetic bead-based multi-
new generation of tumor marker for hepatocel- plex assay format. Clin Chem 59:315–324
lular carcinoma. Clin Chim Acta 313:15–19 15. Yu X, Petritis B, Duan H, Xu D, LaBaer J
9. Kirwan A, Utratna M, O’Dwyer ME, Joshi L, (2018) Advances in cell-free protein array
Kilcoyne M (2015) Glycosylation-based serum methods. Expert Rev Proteomics 15:1–11
biomarkers for cancer diagnostics and prognos-
tics. Biomed Res Int 2015:490531
 Chapter 24

Protein-Specific Analysis of Invertebrate Glycoproteins

Alba Hykollari, Daniel Malzl, Iain B. H. Wilson, and Katharina Paschinger

Abstract
N-Glycans are posttranslational modifications of proteins attached to the amide side chains of asparagine
residues, with possible heterogeneity due to different structures being possible at the same glycosylation
site. In contrast to the mammalian systems, invertebrate N-glycosylation presents a challenge in analysis as
there exist unfamiliar epitopes and a high degree of structural and isomeric variation between different
species. A simple analytical approach to analyze N-glycans on specific glycoproteins is presented, which
involves a combination of tryptic peptide mass spectrometry and “off-line” RP-HPLC MALDI-TOF
MS/MS complemented by blotting to recognize specific epitopes. An additional N-glycan enrichment
and labeling step can facilitate the analysis of single structures and even provide isomeric separation of
N-glycans from specific proteins.

Key words Glycosylation, Glycoproteomics, Mass spectrometry, “off-line” MALDI-TOF MS(MS)

Abbreviations

DTT Dithiothreitol
HRP Horseradish peroxidase
MALDI-TOF MS matrix-assisted laser-desorption/ionization time-of-flight mass spectrometry
NPGC Nonporous graphitized carbon
PC Phosphorylcholine
PE Phosphoethanolamine
RP-HPLC Reversed phase high pressure liquid chromatography
SDS-PAGE Sodium dodecyl sulfate polyacrylamide gel electrophoresis

Full data on the N-glycans of honeybee royal jelly will appear in a forthcoming paper: Hykollari et al. 2018.
Isomeric separation and recognition of anionic and zwitterionic N-glycans from royal jelly glycoproteins.

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_24, © Springer Science+Business Media, LLC, part of Springer Nature 2019

421
422 Alba Hykollari et al.

1 Introduction

Among the posttranslational modifications, the various forms of

glycosylation represent a big challenge in terms of analysis. Most
commonly, glycans are N- or O-linked to amino- or hydroxyl-side
chains of proteins, although C- and S-linkages are also known [1].
Probably N-glycans are the most studied forms and are known from
bacteria, archaea, and almost all eukaryotes; in the latter case,
asparagine residues are modified with an oligosaccharide via a core
N-acetylglucosamine residue [2]. Certainly, N-glycans from mam-
mals are quite well studied, whereas for invertebrate organisms the
N-glycan structures and their functions remain rather unknown. In
contrast to their simple body shape or size, recent studies on
N-glycosylation have proven that invertebrate organisms synthesize
complicated N-glycomes, which compete in terms of complexity
with those of vertebrates [3–6].
Due to the high variety of N-glycan core and antennal mod-
ifications in invertebrates, their analysis is challenging and more
time consuming, as most of the available bioinformatics tools are
based on mammalian structures and so have limited utility when
considering invertebrate glycopeptide and N-glycan data. On the
other hand, invertebrate glycoproteins are biomedically relevant
either due to their immunogenicity (e.g., in venoms), immuno-
modulatory activity and relevance as vaccine targets (e.g., parasite
glycoproteins) or in the use of invertebrate cell lines (e.g., for
baculovirus-based systems) for production of recombinant bio-
pharmaceuticals [7–9]. Thus, there is a need to adequately deter-
mine invertebrate N-glycan structures at the glycomic and
glycoprotein levels. Clearly, glycan annotations on the basis of
m/z alone are insufficient and are often misleading. Orthogonal
proofs are therefore necessary, including the use of specific detec-
tion reagents, MS/MS fragmentation, chemical or exoglycosidase
treatments or reference to in-depth glycomic analyses from the
same organism [10, 11].
The analysis of N-glycans is often required from either purified
single proteins or low amounts of biological material. Here we
describe procedures for protein-specific N-glycan purification,
enrichment, and analysis, successfully used with invertebrate glyco-
proteins. The protein is first screened for N-glycan epitopes with
Western blot, whereby the affinity of specific antibodies, lectins and
pentraxins, can give first impressions as to the modifications of the
oligosaccharides attached to proteins (e.g., core fucose, terminal
galactose, phosphorylcholine, phosphoethanolamine). The tryptic
digest of the protein and subsequent peptide mass spectrometry
fingerprinting help to identify the protein in silico and provide
(glyco)peptides for further analysis. Peptide:N-glycosidases can
then be used to cleave the N-glycans from the peptides prior to
 Invertebrate Glycoprotein Analysis 423

mass spectrometric analyses; as amounts allow, these can be fluo-

rescently labeled and subject to HPLC and MS/MS analyses. Thus,
we can go beyond the typical glycoproteomic procedures in order
to more firmly define N-glycan structures on specific glycoproteins.

2 Materials

2.1 Equipment 1. Probe sonifier, e.g., Branson Sonifier 250.

2. Porcelain mortar with pestle; tight fitting glass homogenizer
(Wheaton; customize as required).
3. Vacuum centrifuge (e.g., Speedvac, Thermo).
4. Microcentrifuge (e.g., Heraeus, Thermo).
5. Lyophilizer (Labconco).
6. Mini Protean® Tetra cell and Power Pac power supply
(Bio-Rad).
7. Trans blot SD semi dry electrophoretic transfer cell (Bio-Rad).
8. Glass columns of 1 cm diameter and 50 cm length (Bio-Rad).
9. Multifunctional microtiter plate reader (e.g., Infinite M200,
Tecan); black 96-well microtiter plates, e.g., Microfluor™ 1 or
LumiNunc (Thermo).
10. HPLC liquid chromatograph with fluorescence detector (e.g.,
Shimadzu Nexera); reverse phase chromatography column,
e.g., Ascentis® Express RP-Amide (150 mm
46 mm,
2.7 μm, Supelco).
11. MALDI-TOF-TOF-MS: Autoflex Speed or UltrafleXtreme
MALDI-TOF-TOF; appropriate MALDI polished or ground
steel target plate (Bruker Daltonics, Billerica, MA). Alternatives
are available commercially from Shimadzu or Applied Biosystems.

2.2 Reagents, 1. 2
SDS-PAGE reducing sample buffer containing 200 mg

Buffers, and Columns SDS, 154 mg DTT, 5 mL stacking gel buffer, 3.6 mL 87%
(See Notes 1 and 2) glycerol (make up to 10 mL with water, then add a few crystals
of bromophenol blue).
2.2.1 Disruption of
Biological Material and
SDS-PAGE Sample
Preparation

2.2.2 SDS-PAGE and 1. 12% SDS-PAGE gel (using 40% acrylamide stock from
Western Blotting Bio-Rad, diluted with either stacking gel buffer with 0.5 M
Tris/HCl pH 6.8 or separation gel buffer with 1.5 M Tris/
HCl pH 8.8).
2. SDS-PAGE running buffer (25 mM Tris, 192 mM glycine,
0.1% SDS; components from either VWR or Roth).
424 Alba Hykollari et al.

3. Western blotting transfer buffer (25 mM Tris, 192 mM glycine,

10% methanol; VWR or Roth).
4. SDS-PAGE protein standard ladder (e.g., Thermo
PageRuler™).
5. Nitrocellulose membrane (BioTrace™ NT from Pall Life
Science).
6. Extra thick blotting paper (Bio-Rad).
7. 0.5% (w/v) Ponceau S (Sigma) in 1% (v/v) acetic acid solution.
8. Membrane washing buffer: Tris buffered saline (TBS, i.e.,
0.1 M Tris/HCl, pH 7.4, 0.1 M NaCl; typically made as a
tenfold concentrated stock) with 0.05% Tween (Sigma).
9. Membrane blocking and antibody/lectin dilution buffer: Tris
buffered saline with 0.05% Tween and 0.5% BSA (Roth).
10. Primary, secondary antibodies, lectins and pentraxins (Sigma
or Vector Laboratories) (see Table 1).
11. SigmaFAST BCIP/NBT or SigmaFAST 3,30 -diaminobenzi-
dine tetrahydrochloride tablets (Sigma), dissolved in 10 mL
and 5 mL, respectively.

2.2.3 Tryptic Peptide 1. Colloidal Coomassie Blue staining solution: 0.02% (w/v) Coo-
Mapping massie Brilliant Blue G-250 (Bio-Rad), 5% aluminum sulfate-
(14-18)-hydrate [e.g., Al2(SO4)3.16H2O (Roth)], ethanol
96% (VWR), phosphoric acid 85% (Roth). Weigh in 100 g of
aluminum sulfate and dissolve it in 1500 mL of water; add
200 mL ethanol and mix well; add 0.4 g of Coomassie Brilliant
Blue G-250 and mix well for at least 30 min; add slowly 47 mL
of phosphoric acid and mix well; make up to 2000 mL with
water (see Note 3).
2. Acetonitrile LC-MS grade (VWR), ammonium bicarbonate
(Roth), water HPLC super gradient grade (VWR), iodoaceta-
mide (Sigma), dithiothreitol (Roth), trifluoroacetic acid
(Fluka).
3. Sequencing grade modified trypsin dissolved to 0.1 mg/mL in
50 mM acetic acid (Promega); typically cleaves after Arg and
Lys residues.

2.2.4 N-Glycome 1. Peptide:N-glycosidase F (PNGase F, recombinant from Flavo-

Release and Analysis bacterium meningosepticum; Sigma).
2. Peptide:N-glycosidase A (either purified native from almond
meal, PNGase A from Sigma, or recombinant Endo H-treated
from Oryza sativa and expressed in Pichia pastoris, PNGase Ar
from NEB).
3. For PNGase F: 100 mM McIlvaine phosphate/citrate buffer
(pH 7.5) or 50 mM ammonium hydrogen carbonate (pH 8;
 Invertebrate Glycoprotein Analysis 425

Table 1
List of selected antibodies, lectins, and pentraxins for N-glycan epitope screening (see Note 4)

Dilution Epitope [13, 14] Supplier

Antibody (1st)
Anti-HRP from rabbit, 10 mg/mL 1:10000 Core α1,3-Fuc/ Sigma
core β1,2-Xyl
Anti-PC (TEPC-15 mouse IgA), 10 mg/mL 1:200 PC-Hex(NAc) Sigma
Antibody (2nd)
Anti-rabbit IgG from goat conjugated with alkaline 1:2000 Vector Labs
phosphatase
Anti-mouse IgA from goat conjugated with alkaline 1:1000 Sigma
phosphatase
Pentraxin
C-reactive Protein (CRP) from human plasma (CaCl2 1:200 PC-Hex(NAc) MP Biochemicals
2.5 mM added)
Amyloid P component from human serum (SAP) 1:200 PE-Hex(NAc) Sigma
Pentraxin recognition
Anti-human C-reactive protein from rabbit 1:1000 Dako
Anti-Amyloid P human serum component IgG from 1:1000 Calbiochem
rabbit (anti SAP)
Lectin
Biotinylated Aleuria aurantia lectin 1:1000 Core α1,6-Fuc/ Vector Labs
Lex
Biotinylated wheat germ agglutinin 1:1000 β1,4HexNAc/ Vector Labs
α2,3Sia
Biotinylated peanut agglutinin 1:1000 Galβ1,3GalNAc Vector Labs
Lectin recognition
Anti-biotin from goat conjugated with alkaline 1:10000 Sigma
phosphatase

mixture of ammonium carbonate and ammonium hydrogen

carbonate).
4. For PNGase A: 50 mM ammonium acetate (pH 5).
5. 1–3 mL solid-phase extraction column and frits (Supelco)
6. Acetonitrile, isopropanol, acetic acid, water (as above).
7. Dowex AG® 50 W-X8 200–400 mesh H+ form (Bio-Rad,
biotechnology grade; washed serially with 0.1 M NaOH,
water, 0.1 M HCl, water, 1 M ammonium acetate and water)
426 Alba Hykollari et al.

and pre-equilibrated with 2% acetic acid prior to usage; C18

material (Lichroprep, Merck); nonporous graphitized carbon
material (NPGC; e.g., Supelco ENVICarb™).
8. MALDI matrices: α-cyanocinnamic acid (ACH, Sigma;
10 mg/mL in 0.1% trifluoroacetic acid/50% acetonitrile) or
6-aza-thiothymine (ATT, Sigma; 3 mg/mL ATT dissolved in
50% ethanol).
9. Glycan labeling: 2-aminopyridine (PA, >99% purity, Sigma),
sodium cyanoborohydride (95% purity, Sigma), hydrochloric
acid (37% HCl, Roth).
10. Gel filtration: Sephadex G-15 and G-25 coarse
(GE Healthcare).

2.2.5 Proteome and 1. Matrix Science web server (www.matrixscience.com/cgi).

N-Glycan Data Analysis 2. ProteinProspector (prospector.ucsf.edu/prospector/cgi-bin).
Database Search Programs

Peptide/Protein Utility 1. Theoretical peptide mass calculator (www.expasy.org).

Program

N-Glycan Analysis 1. Glycoworkbench (www.glycoworkbench.org).

2. FlexAnalysis Bruker software.

3 Methods (See Flow Chart and Example Data in Figs. 1 and 2)

3.1 Sample The purification procedure of the (glyco)protein of interest

Preparation and depends on the biological material which can be whole organisms,
Glycoepitope cells, tissues, cyst fluid, semi-purified proteins or secreted (glyco)
Recognition proteins in culture media or buffer.
3.1.1 Sample 1. Heat inactivate the biological material in boiling water for
Preparation for 10 min. After cooling the sample (cells, worms, royal jelly,
Glycoprotein Analysis (See etc.), homogenize the material using a sonifier or mortar and
Notes 1 and 2) pestle or tight fitting glass homogenizer; for tissues or fungal
mycelium, lyophilize after heat inactivation and ground to a
fine powder in liquid nitrogen. Large volumes of proteinaceous
liquid samples are concentrated with an additional precipitation
step (e.g., methanol). Alternatively, after homogenization,
purify a subset of glycoproteins by an additional enrichment
step using affinity chromatography (if available), e.g., agarose
immobilized monoclonal antibodies [12].
2. Prior to SDS-PAGE, precipitate an aliquot of the samples with
a fivefold excess volume of methanol, incubate at –80  C for 1 h
and centrifuge for 10 min at 4  C, 21,000
g. Dry the protein
pellet at 65  C for several minutes to evaporate excessive
 Invertebrate Glycoprotein Analysis 427

Fig. 1 A potential glycome and glycoproteomic workflow. Starting from biological material, proteins can be
separated by SDS-PAGE prior to Western blotting or peptide map fingerprinting. The peptides and glycopep-
tides are analyzed directly by mass spectrometry; the glycans are released by an N-glycanase such as PNGase
Ar and purified by two rounds of solid-phase extraction prior to mass spectrometry and/or HPLC. Glycans
(examples from honeybee royal jelly) are depicted according to the Symbol Nomenclature for
Glycans, whereby circles, squares, triangles and diamonds respectively represent hexose (here Man or
Gal), N-acetylhexosamine (GalNAc or GlcNAc), deoxyhexose (Fuc) or hexuronic acids (GlcA); S, sulfate; PE,
phosphoethanolamine
428 Alba Hykollari et al.

Fig. 2 Example of a glycoproteomic study of honeybee glycoprotein MRJP1 found in royal jelly. The flow chart
scheme shows the glycoproteomic workflow for the biological sample; the letters refer to steps exemplified by
the following data. (a) N-Glycan epitope detection by Western blotting of the royal jelly glycoproteins after
incubation with anti-HRP antibodies. (b) Tryptic peptide mapping of one of the major royal jelly glycoproteins
as measured with MALDI TOF MS in positive ion mode. (c) Free N-glycans from MRJP1 measured with MALDI-
TOF MS after deglycosylation with PNGase Ar with the [M + Na]+ ions annotated with abbreviations of the form
HxNy, where H is hexose and N is N-acetylhexosamine. (d) RP-HPLC chromatogram of reductively aminated
N-glycans from royal jelly glycoprotein MRJP1 with fractions annotated with the detected glycan m/z values
and calibrated in terms of glucose units. (e) MALDI-TOF MS/MS data of fractionated MRJP1 glycans, whereby
[M + H]+ ions were fragmented and key B and Y ions are annotated. Note that the precursor ions for the H6N4
and H4N5PE structures cannot be separated (Δm/z ¼ 2), but the zoom shows that Y ions derived from both
are present; the structure of the PA (pyridylamino) reducing-terminal label as well as the symbol nomenclature
are shown beneath the flowchart

methanol and redissolve the pellet in 20 μL SDS-PAGE sample

buffer. In addition, heat treat the mixture for 10 min at 95  C,
and after cooling, centrifuge again for 5 min at room tempera-
ture, 21,000
g.

3.1.2 SDS-PAGE and For initial screening of the N-glycan epitopes, approx. 2 μg of
Western Blotting proteins are subject to SDS-PAGE under reducing conditions,
followed by protein transfer to a nitrocellulose membrane (Western
blotting).
1. Check the quality of the successful transfer by incubating the
membrane with Ponceau S staining solution for 1 min. After
 Invertebrate Glycoprotein Analysis 429

de-staining with water (protein bands will stain red), block the
membrane with Tris buffered saline containing 0.05% Tween
and 0.5% BSA for 1 h at room temperature under smooth
shaking.
2. Wash the membrane three times using Tris buffered saline with
0.05% Tween (washing buffer).
3. Incubate with biotinylated lectins, pentraxins or primary anti-
bodies in blocking/dilution buffer for 60 min (see Table 1 and
Note 4).
4. Wash the membrane again thrice as above and incubate with
the relevant peroxidase or alkaline phosphatase conjugated
secondary antibodies in blocking/dilution buffer for 60 min.
5. Again wash the membrane three times as above.
6. Develop the Western blots respectively for peroxidase or phos-
phatase conjugates with either SigmaFAST 3,30 -diaminobenzi-
dine tetrahydrochloride or SigmaFAST BCIP/NBT (dissolve
tablets first in water). Chemiluminescence or other detection
methods can also be used.

3.2 Tryptic Peptide 1. For the peptide mass fingerprinting identification of proteins
Mapping (See Note 5) with MALDI-TOF MS, apply 10 μg of protein to the
SDS-PAGE and stain with Coomassie Blue.
2. After de-staining the gel with water, excise the protein bands in
small pieces on glass plates using a clean scalpel.
3. Wash/destain the gel pieces twice with 50% acetonitrile in
water and successively once with 1:1 0.1 M ammonium bicar-
bonate/acetonitrile and 100% acetonitrile only, prior to drying
in a Speedvac.
4. In addition, reduce the gel pieces with 10 mM DTT for 1 h at
56  C and alkylate for 45 min at room temperature with
iodoacetamide (55 mM in 0.1 M ammonium bicarbonate) in
the dark. Subject the gel pieces to a second round of serial
washing (twice 50% acetonitrile, 1:1 0.1 M ammonium bicar-
bonate/acetonitrile and 100% acetonitrile only) and drying in a
Speedvac.
5. For proteolytic digestion, cover the gel pieces with a 1:2 mix-
ture of 0.1 M ammonium bicarbonate /trypsin (100 ng/μL)
and incubate overnight at 37  C.
6. Extract the peptides at room temperature three times using
acetonitrile/water/trifluoroacetic acid in a ratio of 660:330:1
(v/v/v). Dry the enriched glycopeptides using a vacuum cen-
trifuge and redissolve them in 5 μL water, prior to spotting on a
target plate for MALDI-TOF MS analysis.
7. Spot 0.5 μL of the peptides before applying the matrix (either
α-cyanocinnamic acid (ACH) or 6-aza thiothymine (ATT).
430 Alba Hykollari et al.

The peptides are typically measured in the positive ion mode

and 2000 shots are summed for MS and 4000 for MSMS. The
spectra are processed with the manufacturer’s software; for the
Bruker Flexanalysis software, this includes the SNAP algo-
rithms with corresponding signal to noise thresholds.

3.3 (Glyco)Peptide 1. Predict/identify the corresponding proteins with, e.g., the

Analysis (See Note 6) MASCOT program (Matrix Science web server) or MS-Fit
(ProteinProspector server) using the peptide masses obtained
from tryptic digest and MALDI-TOF MS results. Use one of
the sequence databases available online such as Swissprot or
Uniprot. In parallel, the list of theoretical peptide masses can be
generated by online software (e.g., MS-digest at prospector.
ucsf.edu or web.expasy.org/peptide-mass).
2. Verify the selected “peptide-hits” with the sequences of the
single masses when subject to MALDI-TOF MS/MS. In
order to obtain optimal sequence coverage, allow a mass toler-
ance of 0.5 Da, one missed cleavage site and consider all fixed
modifications (e.g., carbamidomethylation of Cys residues if
alkylated) and potentially known contaminants for mass
spectrometric fingerprint analysis (e.g., human keratin).
Include in the results the protein accession number, number
of successfully assigned peptides and the percentage of
sequence coverage, the software version, number of database
entries, and number of species selected for the software search.
Glycosylated peptides will not be identified unless subject to
PNGase digestion, whereby Asn residues will be converted
to Asp (Δm/z ¼ +1 Da); controls or digestion in 18O-H2O
may be necessary to assess for non-PNGase-mediated
deamination.

3.4 N-Glycome 1. The peptide:N-glycosidase F (PNGase F) can release N-glycans

Release and Analysis from undigested proteins. Denature approximately 8 μg of
protein first in 10 μL 0.5% SDS in water for 5 min at 95  C.
3.4.1 N-Glycome
Alternatively, recombinant PNGase Ar from rice may be partly
Release from Intact
effective and release also any core α1,3-fucosylated structures,
Glycoproteins (See Note7)
which commonly occur in invertebrates and plants.
2. After cooling, add 3 μL of 100 mM McIlvaine phosphate/
citrate buffer pH 7.5 and 2 μL of PNGase F to the sample
prior to incubation for 2 days at 37  C.
3. Mix approximately 2 μg of either glycosylated or deglycosy-
lated protein with 2
SDS-PAGE buffer and after heat dena-
turation and a short centrifugation step, apply both samples to
SDS-PAGE and Western blotting to estimate the degree of
deglycosylation and N-glycan epitope removal.
 Invertebrate Glycoprotein Analysis 431

3.4.2 N-Glycome 1. After protein identification, heat treat the glycopeptides to

Release from inactivate the protease and incubate 90% of the sample with
Glycopeptides (See Notes either PNGase F (see Subheading 3.4.1, Step 2) or PNGase Ar
7, 8 and 9) (see Note 7).
2. Optimal conditions for the PNGase Ar activity (use approxi-
mately 5 U/reaction) are 20 mM ammonium acetate buffer,
pH 5 for 2 days at 37  C.
3. Purify the released N-glycans using two different columns
packed with Lichroprep C18/Dowex AG 50 and nonporous
graphitized carbon/Lichroprep C18 (see Note 8). Wash first
the Lichroprep C18/Dowex AG 50 column with 2% acetic acid
and 60% isopropanol and equilibrate with 2% acetic acid. Apply
the glycopeptide sample after acidifying with 10% acetic acid
and collect immediately the unbound released N-glycans in the
flow-through and wash fractions (three column volumes of 2%
acetic acid).
4. Apply the flow-through/wash from the Lichroprep/Dowex
column directly to a nonporous graphitized carbon/Lichro-
prep C18 column (prewashed and pre-equilibrated with first
100% acetonitrile then water). After sample application, wash
the column with water and elute the N-glycans with 40%
acetonitrile containing 0.1% trifluoroacetic acid. Due to the
presence of TFA, this sample contains a mixed pool of the
neutral and anionic N-glycans.
5. Lyophilize the purified N-glycans overnight and after dissol-
ving them in water, spot an aliquot for MALDI-TOF MS/MS
analysis with 6-azathiothymine (ATT); regarding acquisition
and interpretation of mass spectra, refer to Notes 6 and 9. In
comparison to peptides, higher laser power and detector gain
settings are necessary to detect glycans. For a more detailed
analysis, label the N-glycans by reductive amination using
2-aminopyridine and in addition subject them to HPLC and
MALDI-TOF MS analysis as described below.
6. Fluorescent labeling is performed as follows: dissolve 100 mg
2-aminopyridine in 76 μL concentrated HCl and 152 μL water;
add 80 μL of this solution to the dried glycan sample, prior to
incubation in boiling water for 15 min. Then prepare a solution
of 4.4 mg of sodium cyanoborohydride in a mixture of 9 μL of
the aforementioned 2-aminopyridine solution and 13 μL
water; add 4 μL of this cyanoborohydride-aminopyridine solu-
tion to the sample and continue the incubation overnight at
90  C.
7. Removal of excess labeling reagent is performed immediately
the following day by gel filtration. Dilute the sample in 1.5 mL
of 0.5% acetic acid (i.e., no more than 5% of the gel filtration
column volume), apply to a 30 mL Sephadex G-15 column
432 Alba Hykollari et al.

(1
40 cm) equilibrated in 0.5% acetic acid, and collect 1.5 mL

fractions. Transfer aliquots of fractions (80 μL) to a 96 F black
plate and detect fluorescence in a microtiter plate reader (exci-
tation/emission: 320/400 nm). Pool fluorescent glycans elut-
ing before the excess labeling reagent and lyophilize.
8. Dissolve dried sample by washing the flask four-times with
20 μL of water and transfer to a microcentrifuge tube;
re-lyophilize as required and analyze an aliquot by MALDI-
TOF MS.
9. Inject the major portion of the sample onto an Ascentis®
Express RP-Amide column pre-equilibrated with 100 mM
ammonium acetate (pH 4; buffer A); elute at 0.8 mL/min
using a linear gradient of 30% (v/v) MeOH (buffer B) from
0% B up to 35% B over 35 min (higher percentages of B
generate higher pressure). The glycans are detected by fluores-
cence using excitation/emission wavelengths of 320/400 nm
and the column is calibrated in terms of glucose units with a
fluorescently labeled oligoglucose standard (partial dextran
hydrolysate). Collect fractions based on fluorescence intensity
and lyophilize prior to another round of MALDI-TOF MS and
MS/MS to identify the glycans in the fractions (for example,
data, refer to Fig. 2). Normal phase or non-fused core reversed
phase columns can also be used [15].

4 Notes.

1. The quality of water and other reagents (acetonitrile, metha-

nol, isopropanol) used for analytical purposes should be high
and free of ionic and microbial contaminants.
2. In general, contaminants should be avoided; to prevent analysis
of “foreign” components from the food/nutrition source or
media (e.g., fetal calf serum), the material (whole organisms or
cells) should be washed several times before the heat treatment
and homogenization. After collection, the biological material
should be stored at 80  C, if not immediately homogenized.
To prevent hydrolysis of the anionic or zwitterionic residues
(e.g., phosphate, sialic acid, PC, or PE), the samples should be
heat treated only in water and not in acidic buffers; however,
heat inactivation is necessary to prevent degradation of the
glycans by endogenous glycosidases. For small amounts of
biological samples, also a lysis buffer supplemented with prote-
ase inhibitor cocktail (Sigma) can be used prior to SDS-PAGE.
The methanol precipitation step after cell lysis helps to desalt
the sample and so avoids smearing upon electrophoresis.
 Invertebrate Glycoprotein Analysis 433

3. Colloidal Coomassie aggregates and tiny blue dots are visible.

Make sure that the staining solution is mixed well (e.g., with a
magnetic mixer) before each use.
4. Results obtained from antibody or lectin binding are no struc-
tural proof of the N-glycans on the glycoprotein as their speci-
ficities are sometimes wide or not fully determined. Positive
and negative controls and pull-downs to “pre-clear” endoge-
nous biotinylated proteins, as well as Western blots with and
without lectins/antibodies (i.e., just secondary reagents) or
after glycosidase digestions should be considered for data inter-
pretation. The “mini-description” of the epitopes in Table 1 is
based on determination of binding of the antibodies, lectins or
pentraxins to standard ligands; these determinations are by no
means exhaustive as invertebrate standards are rarely tested
[13, 16]. Nevertheless, anti-horseradish peroxidase is valuable
for screening of core β1,2-xylose and core α1,3-fucose [17],
but the anti-xylose and anti-fucose components of the antisera
are difficult to properly separate. Phosphorylcholine
(PC) epitopes can be detected with either the TEPC-15 anti-
body or by human C-reactive protein [12].
5. The peptides measured “off-line” by MALDI-TOF MS can be
sometimes suppressed by contaminant ions generating from
the protease itself (e.g., trypsin), which is in part autohydro-
lyzed. It is recommended to generate online the theoretical
peptide masses of the protease as well as the target protein
using the MS-digest software.
6. The method described here is a simple and initial procedure for
glycoprotein identification by off-line peptide mass fingerprint-
ing and (glyco)peptide analysis of the selected proteins before
and after PNGase F or A digestion followed by MALDI-TOF
MS. For qualitative/quantitative peptide studies several
“online” methods such as LC-ESI MS/MS can also be
employed. Invertebrate N-glycomes can dramatically differ
from those of mammalian systems, so the N-glycan assignment
on the defined glycopeptides should be based at least on
MS/MS data analysis as compositions based on mass alone
can be misleading: for instance, a difference of 324 Da can
either correspond to two hexoses or one
methylaminoethylphosphonate-modified HexNAc as seen,
e.g., in molluscs. Also, a difference of 176 Da may be either a
methylated hexose or a glucuronic acid [11]. Nevertheless,
mass differences of 146, 162, or 203 can suggest the presence
of fucose, hexose, and N-acetylhexosamine residues. There are
various bioinformatics tools for automated glycopeptide and
glycan identification and the following software can be applied
for glycopeptide MS: “GlycoMod,” “GlycoX,” “Glyco-
pepDB,” “Massy tools,” and “GlycoSpectrumScan” and
434 Alba Hykollari et al.

MSMS “GlycoMiner,” “Protein Prospector,” “GlycopepID,”

“GlycoMasterDB,” and many more [18]. As these are generally
applied to mammalian glycomes and glycoproteomes, caution
is required when using search engines to annotate invertebrate
glycans. For publication, consider the MIRAGE guidelines for
presentation of glycomic data [19] as well as use of the dia-
grammatic Symbol Nomenclature for Glycans [20].
7. PNGase F can release N-glycans from both glycoproteins and
glycopeptides, whereas recombinant PNGase Ar still works
best on peptides. PNGase F does not release N-glycans with
core α1,3-fucose modification (but does release core α1,6-
fucosylated or β1,3-mannosylated structures), while recombi-
nant PNGase Ar can release substituted core α1,3-fucosylated
glycans [6]. The degree of protein deglycosylation can be
monitored with SDS-PAGE (reduced size of the protein after
deglycosylation) and Western blotting (reduced or abolished
N-glycan epitope binding).
8. After PNGase F or A digest of the glycopeptides one aliquot of
the sample should be analyzed with MALDI-TOF MS to verify
deglycosylated peptides and potential “occupation” of an
N-glycosylation site of the protein. The released glycopeptides
should be acidified with 10% acetic acid before Dowex cation
exchange chromatography. For N-glycan recovery protocol
after PNGase F or A release, refer to our recent protocol on
“Analysis of invertebrate and protist N-glycans” [15]. For
O-glycosylation, there is no single universal de-O-glycosylation
enzyme available; O-glycanase has a restricted substrate speci-
ficity and will not remove most extended GalNAc-Ser/Thr
(mucin-type) or other O-glycan structures.
9. The released N-glycans should be measured in positive and
negative ion mode for the identification of potential anionic
residues as sulfate (+80 Da), phosphate (+80 Da), glucuronic
acid (+176 Da), phosphoethanolamine (+123 Da), and ami-
noethylphosphonate (+107 Da; +121 Da if methylated)
[11]. Sialic acids are rare in invertebrates and have been only
convincingly proven in Drosophila or in echinoderms [21, 22],
but are absent, e.g., from nematodes.

Acknowledgments

This work was supported by the Austrian Fonds zur Förderung der
wissenschaftlichen Forschung (FWF; grants P26662, P25058, and
P23922 to A.H., K.P, and I.B.H.W.).
 Invertebrate Glycoprotein Analysis 435

References
1. Spiro RG (2002) Protein glycosylation: nature, antibodies using remodeled glycoproteins.
distribution, enzymatic formation, and disease Anal Biochem 386:133–146
implications of glycopeptide bonds. Glycobiol- 14. Mikolajek H, Kolstoe SE, Pye VE, Mangione P,
ogy 12:43R–56R Pepys MB, Wood SP (2011) Structural basis of
2. Aebi M (2013) N-linked protein glycosylation ligand specificity in the human pentraxins,
in the ER. Biochim Biophys Acta C-reactive protein and serum amyloid P com-
1833:2430–2437 ponent. J Mol Recognit 24:371–377
3. Schiller B, Hykollari A, Yan S, Paschinger K, 15. Hykollari A, Paschinger K, Eckmair B, Wilson
Wilson IBH (2012) Complicated N-linked gly- IBH (2017) Analysis of invertebrate and pro-
cans in simple organisms. Biol Chem Hoppe tist N-glycans. Methods Mol Biol
Seyler 393:661–673 1503:167–184
4. Eckmair B, Jin C, Abed-Navandi D, Paschinger 16. Purohit S, Li T, Guan W, Song X, Song J,
K (2016) Multi-step fractionation and mass Tian Y, Li L, Sharma A, Dun B, Mysona D,
spectrometry reveals zwitterionic and anionic Ghamande S, Rungruang B, Cummings RD,
modifications of the N- and O-glycans of a Wang PG, She JX (2018) Multiplex glycan
marine snail. Mol Cell Proteomics 15:573–597 bead array for high throughput and high con-
5. Stanton R, Hykollari A, Eckmair B, Malzl D, tent analyses of glycan binding proteins. Nat
Dragosits M, Palmberger D, Wang P, Wilson Commun 9:258
IBH, Paschinger K (2017) The underestimated 17. Paschinger K, Rendić D, Wilson IBH (2009)
N-glycomes of lepidopteran species. Biochim Revealing the anti-HRP epitope in Drosoph-
Biophys Acta 1861:699–714 ila and Caenorhabditis. Glycoconj J 26:
6. Yan S, Vanbeselaere J, Jin C, Blaukopf M, 385–395
Wols F, Wilson IBH, Paschinger K (2018) 18. Tsai PL, Chen SF (2017) A brief review of
Core richness of N-glycans of Caenorhabditis bioinformatics tools for glycosylation analysis
elegans: a case study on chemical and enzy- by mass spectrometry. Mass Spectrom (Tokyo)
matic release. Anal Chem 90:928–935 6:S0064
7. Tretter V, Altmann F, Kubelka V, M€arz L, 19. York WS, Agravat S, Aoki-Kinoshita KF,
Becker WM (1993) Fucose α1,3-linked to the McBride R, Campbell MP, Costello CE,
core region of glycoprotein N-glycans creates Dell A, Feizi T, Haslam SM, Karlsson N,
an important epitope for IgE from honeybee Khoo KH, Kolarich D, Liu Y, Novotny M,
venom allergic individuals. Int Arch Allergy Packer NH, Paulson JC, Rapp E,
Immunol 102:259–266 Ranzinger R, Rudd PM, Smith DF, Struwe
8. Prasanphanich NS, Mickum ML, Heimburg- WB, Tiemeyer M, Wells L, Zaia J, Kettner C
Molinaro J, Cummings RD (2013) Glycocon- (2014) MIRAGE: the minimum information
jugates in host-helminth interactions. Front required for a glycomics experiment. Glyco-
Immunol 4:240 biology 24:402–406
9. Geisler C, Mabashi-Asazuma H, Jarvis DL 20. Varki A, Cummings RD, Aebi M, Packer NH,
(2015) An overview and history of glyco- Seeberger PH, Esko JD, Stanley P, Hart G,
engineering in insect expression systems. Darvill A, Kinoshita T, Prestegard JJ, Schnaar
Methods Mol Biol 1321:131–152 RL, Freeze HH, Marth JD, Bertozzi CR,
10. Hykollari A, Malzl D, Yan S, Wilson IBH, Etzler ME, Frank M, Vliegenthart JF,
Paschinger K (2017) Hydrophilic interaction Lutteke T, Perez S, Bolton E, Rudd P,
anion exchange for separation of multiply mod- Paulson J, Kanehisa M, Toukach P, Aoki-
ified neutral and anionic Dictyostelium Kinoshita KF, Dell A, Narimatsu H, York W,
N-glycans. Electrophoresis 38:2175–2183 Taniguchi N, Kornfeld S (2015) Symbol
nomenclature for graphical representations of
11. Paschinger K, Wilson IBH (2016) Analysis of glycans. Glycobiology 25:1323–1324
zwitterionic and anionic N-linked glycans from
invertebrates and protists by mass spectrome- 21. Aoki K, Perlman M, Lim JM, Cantu R, Wells L,
try. Glycoconj J 33:273–283 Tiemeyer M (2007) Dynamic developmental
elaboration of N-linked glycan complexity in
12. Paschinger K, Gonzalez-Sapienza GG, Wilson the Drosophila melanogaster embryo. J Biol
IBH (2012) Mass spectrometric analysis of the Chem 282:9127–9142
immunodominant glycan epitope of Echino-
coccus granulosus antigen Ag5. Int J Parasitol 22. Miyata S, Sato C, Kumita H, Toriyama M, Vac-
42:279–285 quier VD, Kitajima K (2006) Flagellasialin: a
novel sulfated α2,9-linked polysialic acid glyco-
13. Iskratsch T, Braun A, Paschinger K, Wilson protein of sea urchin sperm flagella. Glycobiol-
IBH (2009) Specificity analysis of lectins and ogy 16:1229–1241
 Chapter 25

The Use of Proteomics Studies in Identifying Moonlighting

Proteins
Constance Jeffery

Abstract
Proteomics studies that characterize hundreds or thousands of proteins in parallel can play an important
part in the identification of moonlighting proteins, proteins that perform two or more distinct and
physiologically relevant biochemical or biophysical functions. Functional assays, including ligand-binding
assays, can find a surprising second function for a protein that was previously identified as performing a
different function, for example, a DNA-binding ability for an enzyme in amino acid metabolism. The results
of large-scale assays of protein–protein interactions, gene knockouts, or subcellular protein localizations, or
bioinformatics analysis of amino acid sequences and three-dimensional structures, can also be used to
predict that a protein has additional functions, but in these cases it is important to use biochemical and
biophysical methods to confirm the protein can perform each function.

Key words Moonlighting proteins, Multifunctional proteins, Protein function prediction,

Proteomics

1 Introduction

The goal of many proteomics studies is to identify protein func-

tions. Complicating this task is the ability of a single protein to have
different functions in different cellular processes, with different
ligands or different protein partners, in different cell types,
and/or in different subcellular locations. Hundreds of proteins
have been identified as these moonlighting proteins, which com-
prise a subset of multifunctional proteins that perform two or more
distinct and physiologically relevant biochemical or biophysical
functions that are not due to gene fusions, multiple RNA splice
variants, or pleiotropic effects [1]. Some of the first moonlighting
proteins to be identified were taxon-specific crystallins [2, 3], pro-
teins that are found in high concentration in the lens of the eye but
function as enzymes in other cell types. For example, zeta-crystallin
from the guinea pig lens is identical to the enzyme quinone oxido-
reductase [4]. Over 300 moonlighting proteins are described in the

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_25, © Springer Science+Business Media, LLC, part of Springer Nature 2019

437
438 Constance Jeffery

online MoonProt Database [5]. As a group, the known moonlight-

ing proteins perform a large variety of functions and combinations
of functions and don’t share sequence or structural motifs or other
physical characteristics that enable easy identification. Although
interpreting the results of proteomics projects might be compli-
cated by the presence of moonlighting proteins, the diversity of
moonlighting proteins means that these large-scale projects to
characterize proteins, without a prior hypothesis about each pro-
tein’s function, can be the best way for finding more moonlighting
proteins and their multiple functions.

2 Methods

Proteomics experiments can be used to help identify proteins with

multiple functions in both direct and indirect ways. Projects based
on a functional assay can identify a second function for a protein
that already has a known function. Projects that test other protein
characteristics can also be used to suggest that some proteins have a
second function, although they might or might not provide infor-
mation about what the function is. Bioinformatics analyses used
alone or in combination with proteomics projects can be used to
suggest which other proteins might also have multiple functions.

2.1 Experimental The proteomics methods that have been the most useful in finding
Methods proteins with multiple functions include those that test binding to a
specific molecule, protein–protein interactions, the results of gene
knockout experiments, and cellular localization.

2.1.1 Binding Studies Proteomics studies that involve screening hundreds or thousands of
proteins to find those that bind to DNA, extracellular matrix, or
other macromolecules have identified several dozen proteins that
were already known to have a different function. This is not too
surprising because dozens of the known moonlighting proteins
have at least one function that involves binding to another mole-
cule—as a cell surface receptor for a soluble ligand or the extracel-
lular matrix, as a secreted ligand binding to a receptor on another
cell, or as a DNA- or RNA-binding protein.
The use of microarrays of proteins, DNA oligonucleotides, or
RNA oligonucleotides enables screening of vast numbers of pro-
teins to find those that bind to a chosen macromolecule. Hall and
coworkers screened a microarray of yeast proteins for binding to
DNA oligonucleotides and identified mitochondrial Arg5,6
(N-acetylglutamate kinase/N-acetylglutamyl-phosphate reduc-
tase), an enzyme in the arginine biosynthetic pathway, as having
DNA-binding activity. Complementary chromatin immunoprecip-
itation experiments and gene deletion experiments confirmed that
 Moonlighting Proteins in Proteomics 439

Arg5,6 is a transcription regulator for several specific nuclear and

mitochondrial genes [6].
Assays to identify proteins that bind to a specific protein also
identified several dozen yeast and bacterial proteins that perform
one function when expressed inside the cell and a second function
when displayed on the cell surface. A proteomic study of cell wall
proteins from the pathogenic fungus Candida albicans was used to
identify eight cytosolic proteins (phosphoglycerate mutase, alcohol
dehydrogenase, thioredoxin peroxidase, catalase, transcription
elongation factor, glyceraldehyde-3-phosphate dehydrogenase,
phosphoglycerate kinase, and fructose bisphosphate aldolase) as
cell surface receptors for host plasminogen [7]. A similar study of
the intestinal “pro-biotic” bacterium Bifidobacterium lactis identi-
fied the cytosolic enzymes bile salt hydrolase, glutamine synthetase,
and phosphoglycerate mutase and the chaperone DnaK also to bind
host plasminogen when expressed on the cell surface [8].

2.1.2 Protein–Protein Proteomics-scale studies of protein–protein interactions, such as

Interactions yeast 2-hybrid assays, often yield results that are more complex
than expected, with a single protein being found to interact with
proteins acting in multiple biochemical pathways, molecular
machines or multiprotein complexes. These results are sometimes
interpreted as being due to false positives, but interacting with
multiple groups of proteins from different cellular processes is a
common characteristic of moonlighting proteins, and these pro-
tein–protein interaction results could be due to physiologically
relevant interactions [9]. In a study of human proteins, Chapple
and coworkers combined protein–protein interaction information
with analysis of the protein functional annotations to identify
430 proteins that they described as being extreme multifunctional
proteins [10]. Further analysis of protein–protein interaction net-
works from humans and other species could lead to predictions of
additional proteins that might interact with different groups of
protein partners to perform different functions, along with sugges-
tions of the types of additional functions based on the identities of
the interacting proteins. Follow-up testing through biochemical or
biophysical assays would be needed to confirm the observed inter-
actions are due to a second function and not due to false positives,
or to proteins that interact with multiple proteins as part of a single
function, such as in a signaling pathway, or proteins that perform
the same function in different cellular locations.

2.1.3 Gene Knockouts When a single protein participates in several cellular processes,
deletion of the gene encoding the protein can result in phenotypes
that are more complex than can be explained by the loss of a single
function. Several labs made use of yeast genetics to look for
enzymes for which replacement of the wild-type enzyme with a
440 Constance Jeffery

catalytic deficient mutant does not recapitulate the results of the

complete gene knockout. Because the mutant protein exhibits only
part of the deletion phenotype, the wild-type protein must have a
second function. The S. cerevisiae Bat2 transaminase in sugar and
amino acid metabolism and the isoleucine/valine biosynthetic
enzymes Ilv1 and Ilv2 were found to have second functions in
addition to their respective catalytic functions [11]. The Alt1 ala-
nine transaminases from two other yeast species, Lacchancea kluy-
veri (LkAlt1) and Kluyveromyces lactis (KlAlt1), were also found to
be moonlighting proteins [12].

2.1.4 Expression Proteomics projects that determine a protein’s cellular location

Patterns/Cellular (s) can be used to suggest that it might have multiple functions.
Localization Many of the known moonlighting proteins perform their different
functions in different subcellular locations or cell types. For exam-
ple, several dozen cytosolic proteins, like the plasminogen-binding
proteins mentioned above, have a second function as a receptor or
adhesin on the cell surface in bacteria, humans, and many other
species. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH)
was the first cytoplasmic protein found to be attached to the surface
of pathogenic streptococci [13], and several dozen cytosolic
enzymes in multiple species have also been found to be displayed
on the cell surface, where they play roles in signaling, adhesion, or
acquiring nutrients. Studies to identify all the proteins on the cell
surface of dozens of bacterial species, through cell fractionation and
isolation of proteins followed by identification through mass spec-
trometry, found that many other cytoplasmic proteins are also
attached to the cell surface, and some may be additional moon-
lighting proteins [14, 15]. Other methods of studying protein
localization can also be scaled up and used in proteomics studies.
In a recent study using antibody-based immunofluorescence of
12,003 human proteins in 30 subcellular structures and 13 orga-
nelles, about half of the proteins were found in more than one
compartment and might also include candidates for moonlighting
proteins [16].
The taxon-specific crystallins mentioned above clearly have two
functions because these enzymes are found in a high concentration
in the lens of the eye where their catalytic substrates are not found,
and the known intracellular/cell surface moonlighting proteins
have been tested through binding studies to confirm the presence
of a second function. In other cases, finding a protein in a subcellu-
lar location where it is not expected to perform its known function
can suggest that the protein has a second function, but experimen-
tal evidence that the protein is performing a different function in
each location is needed to confirm that the protein is multifunc-
tional. For example, proteins that move between cellular compart-
ments as part of a signaling pathway would not be considered
moonlighting proteins.
 Moonlighting Proteins in Proteomics 441

2.2 Bioinformatics The lack of common sequence or structural characteristics among

Analysis the moonlighting proteins has made it difficult to develop a univer-
sal computational method to predict that a protein has more than
one function, but several labs are developing bioinformatics meth-
ods by using collections of known moonlighting proteins, such as
the MoonProt Database, as a positive control set. Unfortunately, a
true negative control set of proteins that don’t have multiple func-
tions is not available because it’s currently not possible to know if a
protein has only one function or if it has additional functions that
have not yet been identified.
Large-scale searches of the literature and database annotation,
including searches for proteins with diverse GO terms in UniProt
[17, 18], have shown some success in identifying proteins with
multiple functions in diverse processes [19, 20]. Searches for
amino acid sequence or structural motifs known to be associated
with specific protein functions can help identify proteins that have
motifs corresponding to multiple functions. For example, the use of
an X-ray crystal structure revealed the Streptomyces coelicolor alba-
flavenone synthase also has a terpene synthase active site [21]. How-
ever, the use of motifs as a tool for prediction of function is limited
by the lack of known motifs for many classes of functions, for
example, protein–protein interactions. As more information
becomes available about sequence motifs, protein–protein interac-
tion surfaces, the constellations of amino acids that make up cata-
lytic sites, and the three-dimensional structures of moonlighting
proteins, the ability of these methods to find the multiple functions
of moonlighting proteins might be improved [22, 23].
Some recent methods combine analysis of protein sequences
and structures with information about the results from proteomics
projects (protein–protein interactions, cellular locations, etc.)
[24, 25]. Because of the challenges in identifying moonlighting
proteins, these combined methods might have the most success in
the future.

3 Conclusions

The presence of moonlighting proteins adds to the complexity of

the results of proteomics studies, but these large-scale methods are
valuable for identifying more examples of these proteins, which are
otherwise usually found through serendipity. One caution, how-
ever, is that only the proteomics methods that directly test for a
specific function, such as the plasminogen-binding assays men-
tioned above, provide evidence that the protein performs that
function. The results of the other proteomics and bioinformatics
methods described herein can usually only be used to predict that a
protein has a second function. It is necessary to use biochemical and
biophysical methods to confirm that a protein performs both
functions.
442 Constance Jeffery

References
1. Jeffery CJ (1999) Moonlighting proteins. Microbiol 8:1150. https://doi.org/10.3389/
Trends Biochem Sci 24(1):8–11. PMID: fmicb.2017.01150
10087914 13. Pancholi V, Fischetti VA (1992) A major sur-
2. Piatigorsky J, Wistow GJ (1989) Enzyme/ face protein on group a streptococci is a glycer-
crystallins: gene sharing as an evolutionary aldehyde-3-phosphate-dehydrogenase with
strategy. Cell 57:197–199 multiple binding activity. J Exp Med
3. Wistow GJ, Kim H (1991) Lens protein 176:415–426
expression in mammals: taxon specificity and 14. Olaya-Abril A, Jiménez-Munguı́a I, Gómez-
the recruitment of crystallins. J Mol Evol Gascón L, Rodrı́guez-Ortega MJ (2014) Sur-
32:262–269 fomics: shaving live organisms for a fast proteo-
4. Huang QL, Russell P, Stone SH, Zigler JS Jr mic identification of surface proteins. J
(1987) Zeta-crystallin, a novel lens protein Proteome 97:164–176. https://doi.org/10.
from the Guinea pig. Curr Eye Res 1016/j.jprot.2013.03.035
6:725–732. PMID: 3595182 15. Wang W, Jeffery CJ (2016) An analysis of sur-
5. Mani M, Chen C, Amblee V, Liu H, Mathur T, face proteomics results reveals novel candidates
Zwicke G, Zabad S, Patel B, Thakkar J, Jeffery for intracellular/surface moonlighting proteins
CJ (2015) MoonProt: a database for proteins in bacteria. Mol BioSyst 12:1420–1431
that are known to moonlight. Nucleic Acids 16. Thul PJ, Åkesson L, Wiking M, Mahdessian D,
Res 43:D277–D282 Geladaki A, Ait Blal H, Alm T, Asplund A,
6. Hall DA, Zhu H, Zhu X, Royce T, Gerstein M, Björk L, Breckels LM, B€ackström A,
Snyder M (2004) Regulation of gene expres- Danielsson F, Fagerberg L, Fall J, Gatto L,
sion by a metabolic enzyme. Science Gnann C, Hober S, Hjelmare M,
306:482–484 Johansson F, Lee S, Lindskog C, Mulder J,
7. Crowe JD, Sievwright IK, Auld GC, Moore Mulvey CM, Nilsson P, Oksvold P,
NR, Gow NA, Booth NA (2003) Candida Rockberg J, Schutten R, Schwenk JM, Siverts-
albicans binds human plasminogen: identifica- son Å, Sjöstedt E, Skogs M, Stadler C, Sullivan
tion of eight plasminogen-binding proteins. DP, Tegel H, Winsnes C, Zhang C,
Mol Microbiol 47:1637–1651. PMID: Zwahlen M, Mardinoglu A, Pontén F, von
12622818 Feilitzen K, Lilley KS, Uhlén M, Lundberg E
(2017) A subcellular map of the human prote-
8. Candela M, Bergmann S, Vici M, Vitali B, ome. Science 356:eaal3321. https://doi.org/
Turroni S, Eikmanns BJ, Hammerschmidt S, 10.1126/science.aal3321
Brigidi P (2007) Binding of human plasmino-
gen to Bifidobacterium. J Bacteriol 17. Consortium GO (2015) Gene ontology con-
189:5929–5936. https://doi.org/10.1128/ sortium: going forward. Nucleic Acids Res 43:
JB.00159-07 D1049–D1056
9. Gómez A, Hernández S, Amela I, Piñol J, 18. UniProt Consortium (2015) UniProt: a hub
Cedano J, Querol E (2011) Do protein- for protein information. Nucleic Acids Res 43:
protein interaction databases identify moon- D204–D212
lighting proteins? Mol BioSyst 7:2379–2382. 19. Khan IK, Bhuiyan M, Kihara D (2017)
https://doi.org/10.1039/c1mb05180f DextMP: deep dive into text for predicting
10. Chapple CE, Robisson B, Spinelli L, Guien C, moonlighting proteins. Bioinformatics 33:
Becker E, Brun C (2015) Extreme multifunc- i83–i91. https://doi.org/10.1093/bioinfor
tional proteins identified from a human protein matics/btx231
interaction network. Nat Commun 6:7412. 20. Pritykin Y, Ghersi D, Singh M (2015)
https://doi.org/10.1038/ncomms8412 Genome-wide detection and analysis of multi-
11. Espinosa-Cantú A, Ascencio D, Herrera- functional genes. PLoS Comput Biol 11:
Basurto S, Xu J, Roguev A, Krogan NJ, e1004467. https://doi.org/10.1371/journal.
DeLuna A (2018) Protein moonlighting pcbi.1004467
revealed by noncatalytic phenotypes of yeast 21. Zhao B, Lei L, Vassylyev DG, Lin X, Cane DE,
enzymes. Genetics 208:419–431. https://doi. Kelly SL, Yuan H, Lamb DC, Waterman MR
org/10.1534/genetics.117.300377 (2009) Crystal structure of albaflavenone
12. Escalera-Fanjul X, Campero-Basaldua C, monooxygenase containing a moonlighting
Colón M, González J, Márquez D, González terpene synthase active site. J Biol Chem
A (2017) Evolutionary diversification of ala- 284:36711–36719. https://doi.org/10.
nine transaminases in yeast: catabolic speciali- 1074/jbc.M109.064683
zation and biosynthetic redundancy. Front 22. Khan I, Chitale M, Rayon C, Kihara D (2012)
Evaluation of function predictions by PFP,
 Moonlighting Proteins in Proteomics 443

ESG, and PSI-BLAST for moonlighting pro- 24. Khan IK, Kihara D (2016) Genome-scale pre-
teins. BMC Proc 6(Suppl 7):S5. https://doi. diction of moonlighting proteins using diverse
org/10.1186/1753-6561-6-S7-S5 protein association information. Bioinformat-
23. Hernández S, Franco L, Calvo A, Ferragut G, ics 32:2281–2288. https://doi.org/10.1093/
Hermoso A, Amela I, Gómez A, Querol E, bioinformatics/btw166
Cedano J (2015) Bioinformatics and moon- 25. Khan I, McGraw J, Kihara D (2017) MPFit:
lighting proteins. Front Bioeng Biotechnol computational tool for predicting moonlight-
3:90. https://doi.org/10.3389/fbioe.2015. ing proteins. Methods Mol Biol 1611:45–57.
00090 https://doi.org/10.1007/978-1-4939-7015-
5_5
 Chapter 26

Two-Dimensional Biochemical Purification for Global

Proteomic Analysis of Macromolecular Protein Complexes
Reza Pourhaghighi and Andrew Emili

Abstract
A high-resolution two-dimensional (2-D) proteomic fractionation technique for the systematic purification
and subsequent mass spectrometry-based identification of endogenous protein macromolecular complexes
is described. The method hyphenates preparative isoelectric focusing (IEF) with mixed-bed ion exchange
chromatography (IEX) to efficiently separate cell- or tissue- derived soluble protein mixtures, allowing for
more effective and less biased physiochemical characterization of stable multiprotein assemblies. After
comprehensive 2D fractionation of cell-free lysates, each fraction is subjected to quantitative tandem
mass spectrometry (MS/MS) and subsequent computational analysis to map high-confidence protein–-
protein interactions (PPIs). Herein, the experimental component (workflow protocols) for this global
“interactome” network mapping platform is described.

Key words Protein–protein interaction, Protein complexes, Isoelectric focusing, High-performance

liquid chromatography (HPLC), Biochemical separation, Ion exchange chromatography, Fraction-
ation, Nanoflow liquid chromatography tandem mass spectrometry (nLC-MS/MS)

1 Introduction

Since stable macromolecular assemblies are responsible for many, if

not most, of the key biochemical processes operating inside living
cells, the comprehensive experimental analysis of multiprotein
complexes represents a significant goal of the field of systems (net-
work) biology. To date, several methodologies for systematic large-
scale analysis of the composition (physically associated compo-
nents) of cellular multiprotein complexes and protein–protein
interaction networks have been reported [1–4]. In this context,
we have developed a flexible platform for the global study of
endogenous protein complexes from diverse cells and tissue sam-
ples based on the extensive biochemical pre-fractionation of native
protein assemblies prior to in-depth quantitative nLC/MS/MS-
based detection [5]. Recently, in order to further improve the
analytical dynamic range, we have devised complementary

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_26, © Springer Science+Business Media, LLC, part of Springer Nature 2019

445
446 Reza Pourhaghighi and Andrew Emili

Fig. 1 Two-dimensional macromolecular complex profiling platform

biochemical separation methods that enable rapid and efficient

fractionation of soluble cell-free mixtures to isolate stably asso-
ciated protein complexes with the highest possible resolution with-
out disturbing macromolecular integrity [5]. These include a novel
hyphenated 2-D separation workflow based on hyphenation of
non-denaturing preparative IEF with orthogonal IEX-based
HPLC separations. In this approach, native soluble protein com-
plexes are first gently extracted from a biological specimen and then
selectively enriched by IEF over pH range of 5–8 into five fractions,
which are each then subsequently subjected to a more extensive
salt-gradient mixed-bed IEX-based fractionation. The collected
protein factions are precipitated and digested into peptides with
trypsin, in order to identify stably associated interacting proteins
which reproducibly co-elute through 2-D IEF-IEX protein separa-
tions platform, resulted fractions are quantitative analyzed by tan-
dem mass spectrometry.
In this chapter, we provide a detailed experimental protocol
and an illustrative example showing the application of this 2-D
proteomic fractionation technique to resolve protein assemblies in
the microbe E. coli for a comprehensive assessment of the microbial
interactome analysis. Figure 1 illustrates the described protein
complex profiling platform. Key steps and troubleshooting issues
are explained, while complementary computational data analysis
strategies are described in [6].

2 Materials

It is essential to consult the appropriate Material Safety Data Sheets

and the institutional Environmental Health and Safety Office for
proper handling of the potentially hazardous material used in this
protocol. Use of analytical (HPLC) grade water and solvents to
prepare buffers and reagents is required.
 Two-Dimensional Biochemical Purification for Global Proteomic Analysis. . . 447

2.1 Protein 1. Lysis buffer: Modified B-PER protein extraction buffer

Extraction from E. coli (Thermo Scientific) supplemented with 10% v/v glycerol,
Cells 0.5 mM Dithiothreitol (DTT), 0.2 mg/mL Lysozyme,
2 μL/mL DNase I, and Ethylenediaminetetraacetic acid
(EDTA)-free protease inhibitor (Thermo Scientific) (see Note
1).
2. Bradford reagent: Store in dark at 4  C.

2.2 Non-denaturing 1. MicroRotofor cell (Bio-Rad).

Isoelectric Focusing 2. High voltage power supply (see Note 2).
3. Vacuum source and tubing.
4. MicroRotofor focusing chamber (Bio-Rad).
5. Sealing tape.
6. 3-mL syringe.
7. Anodic and cathodic ion exchange membranes (each one
piece).
8. Anodic electrolyte solution: 0.1 M H3PO4.
9. Cathodic electrolyte solution: 0.1 M NaOH.
10. Carrier ampholytes (Bio-Lyte), pH 5-8 (Bio-Rad).
11. Glycerol.

2.3 HPLC 1. HPLC-IEX Column: Pre-packed mixed-bed PolyCATWAX

Fractionation column (PolyLC Inc.) 200  2.1 mm, 5 μm, 1000-Å (see
Note 3).
2. HPLC MES mobile phase-A: 10 mM 2-(N-morpholino)etha-
nesulfonic acid (MES) buffer pH 6, 5% Glycerol and 0.01%
NaN3 to inhibit bacterial growth.
3. HPLC MES mobile phase-B: 10 mM MES buffer pH 6, 1.5 M
NaCl, 5% Glycerol, 0.01% NaN3.
4. HPLC Tris mobile phase-A: 10 mM tris(hydroxymethyl)ami-
nomethane (Tris) buffer pH 8, 5% Glycerol and 0.01% NaN3.
5. HPLC Tris mobile phase-B: 10 mM Tris buffer pH 8, 1.5 M
NaCl, 5% Glycerol, 0.01% NaN3.

2.4 Sample 1. Microcentrifuge with temperature control.

Preparation and 2. Trichloroacetic acid (TCA). Store at 4  C.
Protein Digestion
3. Acetone. Store at 20  C.
4. Incubating shaker with temperature control.
5. DTT stock solution: Dissolve 7.7 mg DTT in water to obtain a
final concentration of 0.5 M (see Note 4).
448 Reza Pourhaghighi and Andrew Emili

6. Iodoacetamide (IAA) stock solution: Dissolve 9.2 mg of IAA in

500 μL of 50 mM NH4CO3 pH 8 solution to obtain a IAA
stock solution of 0.1 M (see Note 5).
7. Sequencing grade trypsin (Promega).

2.5 nLC-MS/MS 1. nLC system coupled online with a high-resolution tandem MS

Analysis system.
2. 1% Formic Acid (FA).
3. nLC mobile phase-A: 0.1% FA.
4. nLC mobile phase-B: 80% ACN, 0.1% FA.

3 Methods

Timing is critical throughout the protocol. Work quickly and con-

sistently. Try to minimize the time especially before the protein
sample is fractionated. Keep sample on ice unless otherwise stated.

3.1 Soluble Protein 1. Pellet E. coli cells by centrifugation at 5000  g for 10 min (see
Extraction Note 6).
2. Add 4 mL of modified B-PER protein extraction buffer per one
gram of cell pellet and mix gently until the pellet is fully
dissolved.
3. Lightly mix the lysate for 30 min at 4  C using a rotatory
shaker.
4. Centrifuge lysate at 15,000  g for 10 min at 4  C to separate
soluble proteins from insolubles and the cell debris.
5. Remove and filter the supernatant with 0.45 μm centrifugal
filters according to manufacturer’s recommended time and
speed.
6. Use Bradford protein assay to measure the protein concentra-
tion in lysate obtained.

3.2 Isoelectric 1. Place the MicroRotofor unit in cold room (4  C) at least

Focusing (IEF) 15 min before beginning the focusing run.
2. Equilibrate the ion exchange anode (red) and cathode (black)
membranes overnight in 0.1 M H3PO4 and 0.1 M NaOH
solutions, respectively.
3. The sample volume for the separation chamber in a MicroRo-
tofor unit is ~2.5 mL. Prepare IEF sample at a protein concen-
tration of ~2 mg/mL. Add carrier ampholyte and glycerol to
protein sample solution to the final concentrations of 2 and
10% v/v, respectively (see Note 7).
Two-Dimensional Biochemical Purification for Global Proteomic Analysis. . . 449

4. Rinse the equilibrated anodic and cathodic membranes with

deionized water and securely mount them at two ends of the
focusing chamber.
5. Anodic and cathodic electrode assemblies are color coded in
red and black. Assemble them at corresponding ends of the
focusing chamber. Align the vents on electrode assemblies with
one row of ports on the focusing chamber and tight the
threaded sleeve around the assemblies.
6. The other row of ports of the focusing chamber which are not
aligned with the vents on the electrode chambers are used for
harvesting the fractionated material. Seal with a strip of sealing
tape before loading the sample.
7. Using a 3-mL syringe, gradually load the sample solution
through the centermost loading port of the focusing chamber.
8. Make sure that all the channels of focusing chamber are filled
with sample solution and there is no bubble inside them (see
Note 8).
9. If outside surface of the focusing chamber is wet, gently dry it
and seal the row of focusing chamber used for sample loading
with a strip of sealing tape. Make sure all the ports are fully
covered and the tape does not overlap with the harvesting port
sealing tape.
10. Place the focusing assembly on MicroRotofor unit with the
vents on the electrode chambers facing up.
11. Using two syringes, add 6 mL 0.1 M H3PO4 through the vent
hole of the anodic electrolyte chamber (left/red) and 6 mL
0.1 M NaOH through the vent hole of the cathodic electrolyte
chamber (right/black).
12. Close the cooling block cover. Make sure the focusing chamber
and the sealing tapes can rotate without any obstruction.
13. Place the green lid on top of the chassis and turn on the power
switch.
14. Attach the electrodes from the MicroRotofor lid to the power
supply and perform focusing at 1 W constant power. Typical
run is usually completed in 2–3 h (see Note 9).
15. After the IEF run is complete, collect the fractions as quickly as
possible and avoid sample diffusion.
16. Turn the power supply and the MicroRotofor unit off.
17. Apply a vacuum to the MicroRotofor chassis.
18. Open the cooling block cover and gently remove the sealing
tape from the sample loading ports (top row).
19. Place the focusing assembly in the harvesting station with the
row of sample loading ports facing up. Press down it firmly so
450 Reza Pourhaghighi and Andrew Emili

that the needles could pass sealing tape and penetrate the
harvesting ports.
20. After the fractions are aspirated into the harvesting tray, turn
off the vacuum source and remove the harvesting tray.
21. Transfer the fractions to 1.5 mL tubes and store them at 4  C.

3.3 Ion Exchange NOTE: No sample clean-up and/or buffer exchange step is
Chromatography required prior to IEX-HPLC separation of IEF fractions collected.
(IEX-HPLC) Depending of the pH of the IEF fractions, Tris–HCl pH 8 (for
acidic IEF fractions with the pH up to 7) or MES pH 6 buffer (for
basic fractions with pHs above 7) are used for IEX separations of
IEF focused protein complexes. During IEX separation, proteins
can be eluted from the IEX-HPLC column with a salt (NaCl)
gradient and recovered in biologically active forms.
1. Equilibrate the IEX-HPLC column by running two blank
gradients using the buffer systems preferred before running
the protein sample. Likewise, re-equilibrate the column after
each gradient with at least 30 min running 100% buffer-A.
2. Set the injection volume of the HPLC method to inject the
entire sample collected from IEF chamber into the IEX column
(normally between 200 and 250 μL).
3. The recommended flowrate for a 2.1-mm i.d. column as
described here is 0.2 mL/min.
4. A typical gradient as below could be employed for IEX-HPLC
separation:
(a) 100% A from 0 to 3 min, followed by a shallow gradient to
10% B from 3 to 45 min, a linear gradient to 35% B from
45 to 65 min and then a gradient to 100% B from 65 to
80 min followed by an isocratic hold at 100% B until
90 min.
5. Monitor the protein elution by UV absorption signal at
280 nm.
6. Set the method to collect the fractions with 2-min intervals
during IEX separation run. Depending on the chromatogram,
the collected fractions could be further merged to reduce the
total number of fractions for subsequent analysis.
The HPLC-IEX parameters described are summarized in
Table 1. Figure 2 shows a representative IEX-HPLC chromato-
gram recorded after running combined IEF fractions 1 and
2 (pH 5–5.6) using a Tris pH 8 buffer system and using the
parameters described in Table 1.
 Two-Dimensional Biochemical Purification for Global Proteomic Analysis. . . 451

Table 1
IEX-HPLC parameters

Flow rate 0.2 mL/min

Time (min) LC gradient (%B)
0–3 0
3–45 0–10
45–65 10–35
65–80 35–100
80–90 100
Detection 280 nm
Fraction collection intervals 2 min
UV absorbance at 280 nm(mAu)

800
600
400
200
0

10

20

30

40

50

60

70

80

90

Retention Time(min)

Fig. 2 Representative chromatogram of IEX-HPLC fractionated E. coli protein extract, generated after
combining IEF pre-fractions 1 and 2 (pH 5–5.6). Detailed experimental parameters are listed in Table 1

3.4 Trichloroacetic 1. If fractions are collected in a 96-well plate, carefully transfer

Acid (TCA) them into individual 1.5-mL tubes.
Precipitation 2. Precipitate the proteins by adding 10% v/v cold TCA to each
tube, briefly vortex, and incubate them at 4  C overnight.
3. Centrifuge the protein samples at 15,000  g at 4  C for
30 min to pellet the proteins.
4. Remove supernatant gently leaving protein pellet intact. Con-
sider that the protein pellet might not be visible in dilute
fractions.
5. Add 200 μL ice-cold acetone to wash the white protein pellet.
Briefly vortex and incubate the sample for an hour at 20  C
(see Note 10).
452 Reza Pourhaghighi and Andrew Emili

6. Spin the samples with 15,000  g at 4  C for 30 min.

7. Repeat the acetone wash steps for one more time. Add acetone,
incubate, and centrifuge the sample as before.
8. Remove supernatant and leave the protein pellet to air-dry for
about 30 min.

3.5 Trypsin Digestion 1. Dissolved each dried pellet in 90 μL of 5 mM DTT, 50 mM

NH4CO3 pH 8 solution and incubate the sample for 15 min at
50  C with gentle agitation (see Note 11).
2. Bring the protein solution to room temperature. Add 10 μL of
100 mM IAA solution to reach a final concentration of 10 mM
and incubate the samples for 15 min in the dark with gentle
agitation.
3. In order to quench possible excess of IAA, add 1 μL DTT from
0. 5 M DTT stock solution.
4. Add sequencing grade trypsin at a 1:50 enzyme:protein ratio
and incubate the samples overnight at 37  C with gentle
agitation.
5. Quench the digestion by acidifying the samples by adding
formic acid to a final concentration of 1% (v/v).
6. Use a vacuum centrifuge to lyophilize the peptides to dryness.
Then dissolve the dried samples in 1% formic acid for
subsequent LC-MS/MS analysis (see Note 12).

3.6 LC-MS/MS 1. Set the method to inject about 1–2 μg for digested protein
samples into the nanoLC column for MS analysis.
2. A 60-min LC gradient as outlined below is generally appropri-
ate for nLC-MS/MS analysis of the fractions obtained: A linear
gradient from 5% to 30% B from 0 to 46 min and then a
gradient to 100% B from 46 to 50 min followed by an isocratic
hold at 100% B until 60 min.
3. The parameters for MS analysis of peptide fractions depend on
the type and performance specifications of MS instrument
used. As an illustrative example, the recommended MS para-
meters for a 60-min method on an Orbitrap Q-Exactive HF
instrument are listed in Table 2.

3.7 Computational Search all MS/MS spectra against an appropriate fasta file using a
Proteomics Analysis search engine (e.g., MaxQuant). The calculated related intensity of
proteins in each IEX fraction can then be used to calculate similarity
profiles, to predict protein associations and co-complex member-
ships, using computational algorithms (e.g., machine learning clas-
sifiers) and statistical filtering to identify high-confidence physical
interactions among the co-eluting proteins as described in [6].
 Two-Dimensional Biochemical Purification for Global Proteomic Analysis. . . 453

Table 2
nLC-MS/MS parameters

Time (min) LC gradient (%B)

0–46 5–30
46–50 30–100
50–60 100–100
Full-MS
Microscans 1
Resolution 60,000
Automatic gain control target 3e6
Maximum ion time 70 ms
Number of scans 1
Scan range 300–1650 m/z
dd-MS2
Microscans 1
Resolution 15,000
Automatic gain control target 1e5
Maximum ion time 25 ms
Loop count 15
Isolation window 1.4 m/z
Normalized collision energy 27
dd setting
Charge exclusion Unassigned, 1
Exclude isotopes On
Dynamic exclusion 6s

4 Notes

1. The indicated values are final concentrations of each reagent in

B-PER buffer. Prepare the lysis buffer freshly before use and
keep it on ice.
2. The power supply must be capable of power control at 1 W
constant power. If constant power mode is not available, it
must be programmable to run multiple step constant voltage
methods and capable of supplying up to 1000 V as well as
operating at low currents.
454 Reza Pourhaghighi and Andrew Emili

3. Alternatively, an IEX column with 4.6 i.d. could be used.

However, experimental parameters like sample loading capacity
and gradient flow rate should be adjusted accordingly.
4. DTT is susceptible to oxidation and should be prepared freshly
before use. Keep the DTT solution on ice and store the remain-
ing at 20  C for later use.
5. IAA is sensitive to light. Prepare it freshly and keep in the dark.
6. The bacterial cell pellet could be stored frozen at 80  C. The
method descried in this chapter is also applicable to extracts
generated from frozen cell pellets.
7. The final concentration of ampholyte in sample solution
depends on the protein concentration and may be increased
up to 3% to maintain protein solubility. The pH and concentra-
tion of ampholyte is critical and is recommended to be opti-
mized according to sample and for each fractionation
experiments.
8. Air bubbles will disturb the electric field which result to a poor
IEF separation. Make sure to remove all air bubbles by aspirat-
ing the sample from a channel and reloading it.
9. During an IEF run under constant power, the voltage increases
over time. The run is typically complete when the voltage
stabilizes. Allow the run to continue for 30 min after this
point before harvesting the fractions.
10. Incubate the acetone in 20  C for at least 1 h before adding to
sample.
11. Make sure that the pH of the solution is above 7.5 to avoid
alkylation of lysine and histidine.
12. Using small C18 tips (Ziptips) for further cleaning digested
samples is recommended.

References
1. Krogan NJ, Cagney G, Yu H et al (2006) Global 4. Kristensen AR, Gsponer J, Foster LJ (2012) A
landscape of protein complexes in the yeast Sac- high-throughput approach for measuring tem-
charomyces cerevisiae. Nature 440:637–643. poral changes in the interactome. Nat Methods
https://doi.org/10.1038/nature04670 9:907–909. https://doi.org/10.1038/nmeth.
2. Tarassov K, Messier V, Landry CR et al (2008) 2131
An in vivo map of the yeast protein interactome. 5. Havugimana PC, Hart GT, Nepusz T et al
Science 320:1465–1470. https://doi.org/10. (2012) A census of human soluble protein com-
1126/science.1153878 plexes. Cell 150:1068–1081. https://doi.org/
3. Uetz P, Giot L, Cagney G et al (2000) A com- 10.1016/j.cell.2012.08.011
prehensive analysis of protein-protein interac- 6. Zhong Ming Hu L, Goebels F, Wan C, et al
tions in Saccharomyces cerevisiae. Nature EPIC: elution profile-based inference of protein
403:623–627. https://doi.org/10.1038/ complex membership. Nat Methods
35001009
 Chapter 27

A Data Analysis Protocol for Quantitative Data-Independent

Acquisition Proteomics
Sami Pietil€a, Tomi Suomi, Juhani Aakko, and Laura L. Elo

Abstract
Data-independent acquisition (DIA) mode of mass spectrometry, such as the SWATH-MS technology,
enables accurate and consistent measurement of proteins, which is crucial for comparative proteomics
studies. However, there is lack of free and easy to implement data analysis protocols that can handle the
different data processing steps from raw spectrum files to peptide intensity matrix and its downstream
analysis. Here, we provide a data analysis protocol, named diatools, covering all these steps from spectral
library building to differential expression analysis of DIA proteomics data. The data analysis tools used in
this protocol are open source and the protocol is distributed at Docker Hub as a complete software
environment that supports Linux, Windows, and macOS operating systems.

Key words Proteomics, Mass spectrometry, DDA, DIA, SWATH-MS, Spectral library, Data analysis

1 Introduction

The current method of choice for large-scale identification and

quantification of proteins is liquid chromatography tandem mass
spectrometry (LC-MS/MS) [1]. In addition to data-dependent
acquisition (DDA) mode of mass spectrometry, there is an
increased interest in data-independent acquisition (DIA) mode,
such as Sequential Windowed Acquisition of All Theoretical Frag-
ment Ion Mass Spectra (SWATH-MS) [2].
DIA has been suggested to combine the advantage of the high-
throughput of DDA proteomics with the benefit of the high repro-
ducibility of targeted analysis, such as selective reaction monitoring
(SRM) [2, 3]. In SWATH-MS, all precursors generated from a
sample are systematically fragmented within a predetermined
mass-to-charge ratio (m/z) and retention-time range. Since the
spectra are generated without explicit association between peptide
precursors and corresponding fragments, a spectral library is used
for the identification of peptides from the data. For building the

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3_27, © Springer Science+Business Media, LLC, part of Springer Nature 2019

455
456 € et al.
Sami Pietila

spectral library, data from samples produced by mass spectrometry

in DDA mode can be used.
Here, we provide a comprehensive data analysis protocol,
named diatools, together with its open source implementation for
analyzing DIA data. The protocol covers all steps from raw spec-
trum files to a final result of differentially expressed proteins, with
focus on SWATH-MS data. After installation of the required soft-
ware and preparation of folder structure for the data (Subheading
3.1), the raw mass spectrum files are converted to required open
formats (Subheading 3.2) and a database FASTA is constructed,
which should contain sequences of all possible proteins that can be
potentially found from the whole data set (Subheading 3.3). Sub-
heading 3.4 then discusses the optional customization of the para-
meters of the protocol and Subheading 3.5 illustrates how to run
the protocol to build the spectral library, to produce an intensity
matrix of the identified peptides for each sample, and to perform
differential expression analysis between sample groups. The spectral
library is built as described by Schubert et al. [4]. The SWATH-MS
data is processed using the OpenSWATH software [5] including a
TRIC alignment step [6].
The diatools protocol is distributed as a Docker image at
Docker Hub (compbiomed/diatools). Docker [7] is a software
technology that provides light-weight virtualized software environ-
ment enabling easy implementation of the data analysis
protocol. The protocol supports Linux, Windows, and macOS
operating systems, with the exception that Windows is needed to
convert the raw spectrum files to open formats.

2 Materials

The raw mass spectrum files produced by mass spectrometers are

typically in proprietary vendor-specific formats, which need to be
converted to open formats before data analysis. The format conver-
sion of the raw mass spectrum files can be done with the Proteo-
Wizard software [8] on a Windows platform. Otherwise, our
diatools data analysis protocol and all the required software are
distributed as a Docker image and, therefore, can be run on any
platform that supports Docker.
The diatools Docker image is available at Docker Hub (comp-
biomed/diatools). Additionally, the source code of our implemen-
tation is released under open source General Public Licence (GPL)
3.0 and can be downloaded from GitHub https://github.com/
computationalbiomedicine/diatools.git. The Docker image is
based on Ubuntu 17.04 operating system and it contains multiple
proteomics tools, including OpenMS version 2.3 [9], Trans-
Proteomics Pipeline (TPP) version 5.0 [10], msproteomicstools
version 0.6.0, and ProteoWizard version 3.0.11252 [8]. The
 Quantitative DIA Data Analysis Protocol 457

downstream statistical analyses are performed using R. The R 3.3.2

environment with the appropriate packages are also included in the
Docker image.
For running the diatools data analysis protocol, we recommend
having at least 128 GB of RAM depending on the number of
samples and the size of the sequence database (FASTA). The
Docker image requires at least 30 GB of free disk space. Addition-
ally, mass spectrometry raw data take typically a lot of space and,
therefore, depending on the data, multiple terabytes of disk space
might be required to store the input files.
The diatools protocol assumes that Biognosys iRT kit peptides
are used in the laboratory protocol. The default settings are for the
Q Exactive HF mass spectrometer (Thermo Fisher Scientific, Wal-
tham, Massachusetts, USA) but data from other instruments can be
used as well by adjusting the parameters accordingly.

3 Methods

This chapter describes the steps of our diatools data analysis proto-
col, including data conversion, peptide identification, quantifica-
tion, and differential expression analysis for the acquired data. A
schematic illustration of the protocol is shown in Fig. 1.

3.1 Software installation and

data folder structure preparation

3.2 Conversion of raw mzXML mzML

data les to open format (DDA) (SWATH-MS)

3.3 Construction of the Sequences

sequence database FASTA FASTA

optional

3.4 Peptide search

parameters

optional

mzXML mzML Sequences 3.5 Running the data Peptide

Peptide Differential
(DDA) (SWATH-MS) FASTA analysis protocol intensity
Intensity expression
matrix
matrix analysis

Fig. 1 The diatools data analysis protocol for quantitative data-independent acquisition proteomics data.
Protocol steps are shown with their corresponding section numbers and their input/output files. Optional steps
are marked with dashed line
458 € et al.
Sami Pietila

3.1 Software Install the ProteoWizard software to the Windows machine that is
Installation and Data used to convert the raw data files to an open format. The Proteo-
Folder Structure Wizard software can be downloaded from http://proteowizard.
Preparation sourceforge.net/.
Install Docker on the machine on which the data analysis
protocol will be run. Docker installation package for Linux, Win-
dows, and macOS can be downloaded from www.docker.org. If the
Linux distribution is Ubuntu, RedHat, or CentOS Linux, the
Docker can be installed from their respective software repositories
(see Note 1). For Windows, the latest version 10 is recommended.
On Windows, allow Docker to access the needed drive (for exam-
ple, C:) from the Docker settings.
Once Docker is installed, download the data analysis environ-
ment with the command:

docker pull compbiomed/diatools

On the machine where the data analysis is done, create a folder

called dataset with the following subfolders under it:
l config
l DDA
l DIA
l ref
l out

3.2 Conversion of Use the ProteoWizard tool to convert the raw data files to an open
Raw Data Files to Open format.
Format To convert the raw DDA files to mzXML format, open Win-
dows Command Prompt and go to the folder containing the DDA
raw data. Run qtofpeakpicker from ProteoWizard to pick peaks and
to convert the raw files to mzXML format:

FOR %i IN (*.raw) DO "\Program Files\ProteoWizard\ProteoWizard

3.0.11252\qtofpeakpicker.exe" --resolution=2000 --area=1
--threshold=1 --smoothwidth=1.1 --in %i --out %~ni.mzXML

The default settings in the protocol are according to those by

Schubert et al. [4]. If the ProteoWizard install location or version is
different from the present protocol, modify the command
accordingly.
To convert the raw SWATH-MS files to mzML format, use the
MSConvert program from the ProteoWizard software with the
following options:
l Output format: mzML
l Extension: empty
 Quantitative DIA Data Analysis Protocol 459

l Binary encoding precision: 64bit

l Write index: checked
l TPP compatibility: checked
l Use zlib compression: unchecked
l Package in gzip: unchecked
l Use numpress linear compression: unchecked
l Use numpress short logged float compression: unchecked
l Use numpress short positive integer compression: unchecked
l Only titleMaker filter
Copy the DDA mzXML files to the dataset/DDA folder and
the SWATH-MS mzML files to the dataset/DIA folder.

3.3 Construction of Create a sequence database in FASTA format that consists of all
the Sequence proteins that may exist in the sample set under analysis. The FASTA
Database FASTA File file is used to construct the spectral library by searching the DDA
files against it. Create a FASTA file that contains the following
protein or peptide sequences:
l Proteins of interest (for example Swiss-Prot Human)
l IRT peptides1 (Biognosys|iRT-Kit_WR_fusion)
l Peptides related to lysis (Uniprot ID: Q7M135)
l Digestion enzyme (typically Trypsin (Uniprot ID: P00761))
l Possible contaminants
Do not generate decoy sequences to the FASTA file manually.
They are automatically generated by the protocol by reversing the
peptide/protein sequences. Copy the FASTA file to the dataset/ref
folder and name it as sequences.fasta.

3.4 Peptide Search The default parameters of the protocol are for the nanoflow HPLC
Parameters system (Easy-nLC1200, Thermo Fisher Scientific) coupled to the
Q Exactive HF mass spectrometer (Thermo Fisher Scientific)
equipped with a nano-electrospray ionization source. The device
and lab protocol specific default settings are listed below:
l Precursor mass tolerance: 10 ppm
l Fragment ion tolerance: 0.02 Da
l Cleavage site: Trypsin_P
l Fixed modification: Carbamidomethyl (C)
l Variable modification: Oxidation (M)
If another type of instrument is used, these settings need to be
customized (see Note 2).

1
https://biognosys.com/media.ashx/irtfusion.fasta
460 € et al.
Sami Pietila

3.5 Running the Data Open terminal prompt and set the working directory to dataset/out
Analysis Protocol folder, where the LOCALPATH refers to the path to the previously
created folder structure:

cd /LOCALPATH/dataset/out

Run the data analysis protocol with the following command:

docker run --rm \

-v /LOCALPATH/dataset/:/dataset \
--workdir /dataset/out \
-u $(id -u):$(id -g) \
compbiomed/diatools \
/opt/diatools/dia-pipeline.py \
--in-DDA-mzXML ../DDA/*.mzXML \
--in-DIA-mzML ../DIA/*.mzML \
--db ../ref/sequences.fasta \
--use-comet \
--use-xtandem

On a Windows platform, the path to the dataset is given in the

following form: “-v//c/LOCALPATH/dataset:/dataset”
where c is the drive letter. On Linux platform, Docker might be
available only to superusers. In that case, add sudo command
before the docker command.
To perform the optional differential expression analysis
between sample groups, the groups must be provided using an
additional parameter in the command:

--design-file <designFilename>

The design file must be defined as a tab-separated file (see

Table 1 for an example), where the column Filename refers to the
SWATH-MS filename of a sample, the column Condition is the
group to which the sample belongs, the column BioReplicate refers
to the biological replicate, and the column Run to the MS run.
By default, the false discovery rate (FDR) used by the diatools
protocol for the peptide identifications is 0.01. However, the FDR
threshold can be adjusted by the user (see Note 3). The number of
parallel processing threads used is four by default, but the user can
choose different numbers of threads according to hardware
resources (see Note 4). For the downstream differential expression
analysis, the data are median normalized and differential expression
analysis is performed for all possible pairs of sample groups using
the PECA R-package [11] available from Bioconductor [12].
Once the data analysis run has completed successfully, the
output folder dataset/out contains two tab-separated data files:
DIA-peptide-matrix.tsv and DIA-protein-matrix.tsv. These files
contain the peptides and proteins with their respective intensity
 Quantitative DIA Data Analysis Protocol 461

Table 1
Example design file

Filename Condition BioReplicate Run

Sample1.mzML Treatment 1 1
Sample2.mzML Treatment 2 2
Sample3.mzML Treatment 3 3
Sample4.mzML Control 1 4
Sample5.mzML Control 2 5
Sample6.mzML Control 3 6

values for each sample. The files can be opened with MS Excel or
with LibreOffice Calc. The output folder contains also files of
intermediate results written by various external tools run by the
protocol as well as a log.txt file which includes details on the run.
The log can be used for troubleshooting if the run fails.
If the optional differential expression analysis is performed,
those results are stored as tab-separated files in the dataset/out
folder with the compared groups as filenames. For each identified
protein, the result files contain the protein name, the value of the
test statistic (t), the number of peptides per protein (n), the signifi-
cance p-value (p), and the estimated false discovery rate (p.fdr). In
addition to performing the differential expression analysis by run-
ning the diatools protocol, it is possible to perform the differential
expression analysis separately using the peptide intensity file pro-
duced by the protocol (see Note 5).

4 Notes

1. Docker installation under Ubuntu, RedHat, or CentOS. If

the analysis is done on Ubuntu, RedHat, or CentOS Linux
distributions, the Docker can be installed from the software
repository.
In Ubuntu, use the following shell command:

apt-get install docker.io

For Ubuntu, it is also convenient to add user to a system group

called docker, which makes it possible to run the Docker with-
out a sudo command.
In RedHat/CentOS, use the following commands:
yum install docker
systemctl enable docker
systemctl start docker
462 € et al.
Sami Pietila

2. Customization of peptide search parameters. The default

parameters of the diatools protocol are for the nanoflow
HPLC system (Easy-nLC1200, Thermo Fisher Scientific) cou-
pled to the Q Exactive HF mass spectrometer (Thermo Fisher
Scientific) equipped with a nano-electrospray ionization
source. If another type of instrument is used, the settings
need to be customized by editing the Comet and X!Tandem
search engine parameters. This can be done by modifying the
Comet and X!Tandem configuration files comet.params.tem-
plate and xtandem_settings.xml, respectively, which are
distributed with the protocol. Copy the modified files to the
dataset/config folder and give the following extra parameters
when running the protocol:

--comet-cfg-template config/comet.params.template
--xtandem-cfg-template config/xtandem_settings.xml

3. Adjusting the false discovery rate (FDR) for the peptide

identifications. By default, the protocol uses 0.01 as a FDR
threshold for the spectral library building. For the TRIC align-
ment step, 0.01 is used as target and 0.05 as max threshold.
These values can be adjusted by adding the following extra
parameters when running the diatools protocol:

--library-FDR
--feature-alignment-FDR

For instance, the following parameter instructs to use 0.05 as

the FDR threshold for spectral library building:

--library-FDR 0.05

Similarly, the following parameter instructs to use 0.01 as target

and 0.02 as max threshold for the TRIC alignment:

--feature-alignment-FDR 0.01 0.02

4. Adjusting the number of parallel processing threads. Cur-

rently, the protocol uses a maximum of four threads by default
to process the data. If the protocol is run on a high-end desktop
or on a server, the number of threads can be increased to
correspond to the CPU core count. It speeds up the analysis,
but also increases the amount of consumed RAM. For example,
the following parameter increases the thread count to 20:

--threads 20
 Quantitative DIA Data Analysis Protocol 463

5. Separate differential expression analysis using the peptide

intensity file. In addition to performing the differential expres-
sion analysis by running the diatools protocol, it is possible to
perform the analysis separately using the peptide intensity file
produced by the protocol (peptide-intensity-matrix.tsv).
First, the peptide intensity data are transformed into a suitable
format using the R/Bioconductor package SWATH2stats. To
install SWATH2stats, open R and enter:

source(“https://bioconductor.org/biocLite.R”)
biocLite("SWATH2stats")

To read in the peptide intensity data, the following commands

can be used:

library(data.table)
library(SWATH2stats)
data <- data.frame(fread(file=" peptide-intensity-
matrix.tsv",
sep=’\t’, header=TRUE))

Next, get rid of unneeded columns (line 2), remove the iRT
peptides that are used for retention time normalization (line 3),
filter out rows corresponding to multiple proteins (line 4), and
shorten the protein names (line 5):

data$run_id <- basename(data$filename)

data <- reduce_OpenSWATH_output(data)
data <- data[grep(’iRT’, data$ProteinName, invert=TRUE),]
data <- data[grep(’^1/’, data$ProteinName),]
data$ProteinName <- sapply(strsplit(data$ProteinName, "\cr|"),
function(x) unlist(x)[2])

To define the sample groups, read in the design matrix, after

which the data can be converted to a suitable format for the
statistical analysis with PECA:

design <- read.table(’design.txt’, sep="\t", header=TRUE,

stringsAsFactors=FALSE)
data <- sample_annotation(data, design)
data <- convert4PECA(data)

PECA determines differential protein expression using directly

the peptide-level measurements, instead of the common practice of
using pre-calculated protein-level values. Differential expression
statistic is first calculated for each measured peptide, after which
the protein-level significance is estimated by taking the number of
464 € et al.
Sami Pietila

identified peptides per protein into account. To install PECA, open

R and enter:

source(“https://bioconductor.org/biocLite.R”)
biocLite("PECA")

For DIA data, the reproducibility-optimized test statistic

(ROTS) [13] is the suggested statistic to be used within PECA
[14]. This is done by setting the test parameter as rots when
calling the function. The following commands perform the differ-
ential expression analysis for all possible pairs of sample groups:

library(PECA)
comb <- combn(unique(design$Condition),2)
for(i in 1:ncol(comb)) {
group1 <- paste(design$Condition,
design$BioReplicate,sep="_")
[design$Condition==comb[1,i]]
group2 <- paste(design$Condition,
design$BioReplicate,sep="_")
[design$Condition==comb[2,i]]
peca.out <- PECA_df(data, group1, group2,
id="ProteinName", normalize="median",
test="rots", progress=TRUE)
write.table(peca.out,
file=paste("PECA_",comb[1,i],
"-",comb[2,i],".txt",sep=""),
sep="\t", quote=FALSE, row.names=TRUE, col.names=NA)
}

References
1. Aebersold R, Mann M (2003) Mass targeted analysis of data-independent acquisi-
spectrometry-based proteomics. Nature tion MS data. Nat Biotechnol 32:219–223
422:198–207 6. Röst HL, Liu Y, D’Agostino G et al (2016)
2. Gillet LC, Navarro P, Tate S et al (2012) Tar- TRIC: an automated alignment strategy for
geted data extraction of the MS/MS spectra reproducible protein quantification in targeted
generated by data-independent acquisition: a proteomics. Nat Methods 13:777–783
new concept for consistent and accurate prote- 7. Merkel D (2014) Docker: lightweight Linux
ome analysis. Mol Cell Proteomics 11: containers for consistent development and
O111.016717 deployment. Linux J
3. Huang Q, Yang L, Luo J et al (2015) SWATH 8. Chambers MC, Maclean B, Burke R et al
enables precise label-free quantification on pro- (2012) A cross-platform toolkit for mass spec-
teome scale. Proteomics 15:1215–1223 trometry and proteomics. Nat Biotechnol
4. Schubert OT, Gillet LC, Collins BC et al 30:918–920
(2015) Building high-quality assay libraries 9. Sturm M, Bertsch A, Gröpl C et al (2008)
for targeted analysis of SWATH MS data. Nat OpenMS—an open-source software frame-
Protoc 10:426–441 work for mass spectrometry. BMC Bioinfor-
5. Röst HL, Rosenberger G, Navarro P et al matics 9:163
(2014) OpenSWATH enables automated,
 Quantitative DIA Data Analysis Protocol 465

10. Deutsch EW, Mendoza L, Shteynberg D et al analysis with bioconductor. Nat Methods
(2010) A guided tour of the trans-proteomic 12:115–121
pipeline. Proteomics 10:1150–1159 13. Elo LL, Filén S, Lahesmaa R et al (2008)
11. Suomi T, Corthals GL, Nevalainen OS et al Reproducibility-optimized test statistic for rank-
(2015) Using peptide-level proteomics data ing genes in microarray studies. IEEE/ACM
for detecting differentially expressed proteins. Trans Comput Biol Bioinform 5:423–431
J Proteome Res 14:4564–4570 14. Suomi T, Elo LL (2017) Enhanced differential
12. Huber W, Carey VJ, Gentleman R et al (2015) expression statistics for data-independent
Orchestrating high-throughput genomic acquisition proteomics. Sci Rep 7:5869
 INDEX

A mitochondrial membrane proteins........................... 58

porcine myocardium ................................................. 58
Acetylation....................................................198–202, 206 protein samples.......................................................... 62
Adenylate energy charge (AEC).......................... 287, 291 SDS ......................................................................57, 65
ADP ...................................................................... 287, 291
Bifidobacterium lactis .................................................... 439
AFP-L3 .......................................................................... 413 Biochemical separation ................................................. 446
Alanine transaminases ................................................... 440 BioID
AlbuSorb™ .................................................................43, 46
FRET technology.................................................... 144
AlbuSorb™ PLUS ........................................................... 43 generation................................................................ 146
AlbuVoid™ ................................................... 43, 46, 49, 52 HEK293 cells .......................................................... 148
AlbuVoid™ LC-MS On-Bead......................................... 49
insoluble protein complexes ................................... 145
Alcohol dehydrogenase................................................. 439 limitation ................................................................. 144
Allergenomics ................................................................ 393 materials.......................................................... 145, 146
Allergens ........................................................................ 133
PDL methods .......................................................... 144
Allergomics ........................................................... 393, 394 PPIs.......................................................................... 143
Alpha-1-Antitrypsin (AAT) .............................. 42–44, 46, pull down assay and LC-mass analysis .......... 147, 149
47, 53, 54 validation ........................................................ 146, 148
ALS ................................................................................ 324
Biological networks......................................................... 16
Alternative splicing........................................................ 107 Biomarker development.............................. 329, 376, 377
Alzheimer’s disease ...................................... 36, 324, 336, Biomarker discovery............................................. 108, 110
341, 350, 366, 376, 378
Biomarker screening, see Protein microarray
AMP............................................................. 284, 287, 291 Biomarkers.................................................. 329, 340, 350,
Antibiotic-resistance...................................................... 280 353, 356, 375–377, 413, 414
Antibodies ..................................................................... 109
Biotech Support Group ............................................ 15
Antibody-based enrichment ................................ 199, 201 cancer screening ........................................................ 70
Antibody microarrays.................................................... 414 CTC ........................................................................... 69
Antivenom ....................................................154–156, 158 ctDNA ....................................................................... 69
Apolipoproteins........................................... 297, 300, 302
definition ................................................................... 11
Assay development and validation ...................... 270–272 extracellular vesicles .................................................. 69
Assembled arrays FDA ........................................................................... 12
capture ..................................................................... 109
function ..................................................................... 11
reverse-phase .................................................. 109, 111 in vitro diagnostic tests ............................................. 12
ATP .............................................................. 284, 287, 291 micro-RNAs .............................................................. 69
Autoimmune diseases ......................................... 316, 321,
nucleotides................................................................. 13
323, 337, 340, 356–359, 366 serum ......................................................................... 70
therapy selection........................................................ 11
B
urine analysis.............................................................. 11
Bacteria ................................................................. 123–130 Biomass conversion .............................................. 160, 175
Bead-based immunoassay ............................................. 415 Biomedical research ...................................................... 265
Benzyldimethyl-n-hexadecylammonium chloride (16-BAC) Biotin ligase BirA .......................................................... 144
chemical structure ..................................................... 58 Boronic acid-based chemical enrichment ........... 203, 204
Fenton reaction ......................................................... 61 Both tPA ........................................................................ 319
Ferguson plot analysis............................................... 57 Bottom-up proteomics ................................................. 225
Membrane proteins................................................... 57 Byonic ............................................................................ 232

Xing Wang and Matthew Kuruc (eds.), Functional Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1871,
https://doi.org/10.1007/978-1-4939-8814-3, © Springer Science+Business Media, LLC, part of Springer Nature 2019

467
 FUNCTIONAL PROTEOMICS: METHODS AND PROTOCOLS
468 Index
C CAZymes in microbial secretomes
biomass conversion ................................................. 160
C1 inhibitor........................................................ 42, 44, 47 biotechnological applications ................................. 159
C18 peptide fractionation .............................................. 94 culture plates
C3/C5-convertase ........................................................ 321 bacteria inoculation........................................... 167
CA125 .................................................................. 414–419 casting plate with membrane................... 166, 167
CA15-3 .......................................................................... 413 fungi inoculation ............................................... 167
CA19-9 .......................................................................... 413 membrane filter ........................................ 160, 161
Calpains ......................................330, 339, 347, 357, 368 protein extraction and sample
Calpastatins.................................................. 330, 355, 357 preparation.......................................... 167, 169
Cancer sampling............................................................. 168
angiogenesis and metastasis.................................... 363 data integration and heat map generation............. 162
atherosclerosis ......................................................... 346 plate-based method................................................. 160
biomarkers ............................................................... 353 polysaccharides ........................................................ 159
caspase-8 .................................................................. 351 prediction
cathepsins B ............................................................. 351 database ............................................................. 162
cell lysosomes .......................................................... 321 dbCAN web server................................... 162, 163
cell proliferation ............................................. 341, 352 protein secretion (see Protein secretion)
cellular micro-environment .................................... 341 protein extraction (see Protein extraction)
chronic inflammation .............................................. 341 quantitative proteomics data
colorectal ................................................................. 336 calculation, secretome enrichment................... 173
disease prognosis ..................................................... 353 Perseus, heat map.............................168, 171, 172
epigenetic................................................................. 341 research .............................................................. 168
epitopes.................................................................... 353 sample ................................................................ 173
extracellular matrix.................................................. 353 spreadsheet application ..................................... 170
hemostasis................................................................ 354 technique ........................................................... 168
inflammation ........................................................... 352 transparency....................................................... 168
inflammatory breast cancer (IBC).......................... 352 Cell migration ............................................ 324, 326–328,
lung ................................................321, 329, 337, 362 331, 336
mesotrypsin ............................................................. 351 Cell painting ..............................................................20, 21
MMPs ............................................................. 341, 366 Cell proliferation ........................................ 315, 324, 326,
ofatumumab ............................................................ 352 328, 331, 335, 336, 341, 345, 351–353
PC-3 cells................................................................. 335 Cells and tissue samples ................................................ 266
plasma ............................................................. 351, 353 Chain of translatability..............................................20, 38
procoagulant............................................................ 354 Chemical proteomics .................................................... 199
proliferation ............................................................. 353 Chromatin immunoprecipitation ................................. 438
protease inhibitors................................. 349, 350, 353 Chymotrypsin.............................................. 406, 407, 409
proteasome .............................................................. 352 Chymotrypsinogen .............................................. 324, 325
rituximab ................................................................. 352 Clinical Laboratory Improvement Amendments
signaling.......................................................... 341, 353 (CLIA) ........................................................... 13
thyroid ..................................................................... 329 Coeliac disease (CD)..................................................... 405
treatment ........................................................ 321, 330 Collagenases .................................................................. 326
zymogens ................................................................. 341 Collision induced dissociation (CID) .......................... 230
Cancer diagnosis .......................................................70, 71 Colonic biopsies ............................................................ 127
Candida albicans .......................................................... 439 Colonic lavage ............................................................... 127
Capture arrays ............................................................... 109 Combinatorial hexapeptide ligand library (CPLL)
Carbohydrate-Active Enzymes (CAZymes), see CAZymes allergomics............................................................... 394
in microbial secretomes IgE-binding proteins .............................................. 394
Cardiovascular disease.........................124, 339, 342–344 preparation protocol ............................................... 395
Caspase-8 .................................................... 314, 331, 344, protein elution................................................ 400, 401
346, 347, 351 protein enrichment ................................................. 394
Catalase .......................................................................... 439 treatment ............................................... 394, 396, 399
Cathepsins ..................................314, 321, 330, 337, 340 Companion diagnostic.................................13, 22–24, 36
Cathepsins B.................................................................. 351 Complement system ............................................ 318, 322
 FUNCTIONAL PROTEOMICS: METHODS AND PROTOCOLS
Index 469
Computational analysis DNA repair ..........................................315, 329, 331, 332
data......................................................... 118, 119, 121 DNA replication .......................................... 315, 329, 332
image........................................................................ 118 DNA/RNA binding protein. ....................................... 438
Computational proteomics analysis ............................. 452 Double-one dimensional electrophoresis (D1-DE)
Conformational variants ...........................................44, 45 allergomics application............................................ 134
Critical micelle concentration (CMC) ........................... 58 IEF and SDS-PAGE................................................ 136
Crosstalk ..............................................197, 200, 206, 207 IEF preparation .............................................. 136, 137
Cryomolds ............................................................ 255, 257 IgE immunoblots .................................................... 139
Cryotome .................................................... 255, 257, 262 materials.......................................................... 134, 135
Cytosolic proteins ................................................ 266, 268 preparation ............................................ 136, 138, 139
Cytosolic proteolysis ............................................ 331, 334 principles.................................................................. 135
screening .................................................................. 134
D DPP-4 (membrane metallo-endopeptidase and
Data analysis ........................................226, 229, 232, 233 dipeptidyl peptidase-IV).............................. 414
Drug development ..................... 363, 366–368, 370, 372
Data independent acquisition (DIA)
DDA ........................................................................ 458 Drug discovery
FDR ......................................................................... 462 binary interaction model .......................................... 21
cell-killers ................................................................... 19
General Public Licence (GPL) ............................... 456
mass spectrometry................................................... 457 FDA ........................................................................... 20
reproducibility-optimized test statistic isoforms ..................................................................... 21
(ROTS) ........................................................ 464 oligomers ................................................................... 21
PDD..................................................................... 19–22
spectral library ................................................ 455, 459
SWATH-MS ............................................................ 455 SNPs .......................................................................... 19
Data mining..................................................198, 202–206 TDD ....................................................................19, 20
Drug targets
Data-dependent acquisition (DDA) .......... 455, 458, 459
Data-independent acquisition (DIA) anti-AD .................................................................... 336
LC-MS/MS............................................................. 455 chymase and tryptase .............................................. 346
exogenous proteolytic activity .............. 366, 368, 369
Digestion .................................................... 313–316, 376,
406–409, 411 identification.............................................................. 71
Dimethyl isotopic labelling........................................... 187 non-protease components ...................................... 370
Dimethylation .....................................227, 236, 238, 240 protease inhibitors.......................................... 363–367
proteases ......................................................... 316, 341
Disease
Alzheimer’s ....................................336, 366, 376, 378
E
assessment................................................................ 329
autoimmune ......................................... 321, 337, 340, E. coli .................................................................... 446–448
356, 359, 366 Eculizumab........................................................... 321, 370
cardiovascular and metabolic.................339, 342–344 Effectors............................. 314, 316, 320, 329, 331, 347
chronic ..................................................................... 378 Electrolyte........................................................................ 72
degenerative............................................................. 340 Electron paramagnetic resonance
epigenetics ............................................................... 334 spectroscopy (EPR) ....................................... 32
Huntington’s........................................................... 366 Electrophoresis system.................................................... 62
immune.................................................................... 321 Electrospray ionization (ESI) ....................................... 181
inflammation .................................................. 338, 339 Enamelysin (MMP-20) ................................................. 326
inflammatory bowel ................................................ 321 Enrich post-translational modifications .............. 198–200
liver .......................................................................... 363 Enterokinase .................................................................. 325
lymphoproliferative ................................................. 364 Enzyme-linked immunosorbent assays
metabolic ................................................................. 340 (ELISAs) .........................................29, 30, 405
neurodegenerative................ 324, 329, 350, 355–357 Epidermal growth factor (EGF)................. 316, 324, 336
PAR4 signaling ........................................................ 328 Epigenetics ................................................. 8, 26, 35, 316,
Parkinson’s .............................................................. 336 332, 334, 337, 341, 351, 352
prognosis ................................................................. 353 Epilysin (MMP-28)....................................................... 326
proteases ......................................................... 316, 364 Epitopes ......................................321, 329, 339, 353, 356
thromboembolic ..................................................... 363 Error tolerant search ............................................ 239, 242
DNA processing ................................................... 329, 332 Exoglycosidase .............................................................. 422
 FUNCTIONAL PROTEOMICS: METHODS AND PROTOCOLS
470 Index
Extracellular matrix (ECM)................................ 324–327, MS........................................................................29, 30
334, 338, 339, 341, 343, 346, 349, 350, 353, NMR....................................................................31, 32
360, 363 PDD............................................................................. 3
Extracellular regulated protein kinase (ERK).............. 266 PM (see Precision Medicine (PM))
protein activity........................................................... 33
F protein localization ................................................... 32
Factor VII ................. 314, 318, 335, 342, 344, 347, 363 proteomics methods appraisal ............................28, 29
Factor Xa .................. 318, 319, 335, 342, 360, 363, 364 PTMs ..................................................... 27, 29, 35, 37
quantitation .........................................................30, 31
Factor XII ............................................................. 319, 364
Factor XIIIa ................................................................... 319 Sep-Sci/MS .........................................................29, 30
False discovery rate (FDR) .................................. 460, 462 specification ................................................................. 2
spin labeling............................................................... 34
Fecal samples ................................................126–128, 130
Fenton reaction ............................................................... 61 systems biology .................................................2, 7–10
Flour ..................................................................... 407, 411 Uniprot profile .......................................................... 27
X-ray crystallography ................................................ 31
Fluorescence resonance emission transfer
(FRET) technology ..................................... 144 Y2H............................................................................ 34
Fluorescent-protein tagging (FP) ..................... 28, 32, 33 Fungi.............................................................................. 123
Folate receptor (FR) ..................................................... 266
G
Fractionation ............................................... 446, 447, 454
Fructose bisphosphate aldolase .................................... 439 Galileo Biosciences.......................................................... 62
Functional proteomics ..............................................44, 45 Gas chromatography coupled to mass
biomarker spectrometry (GC-MS) ............................... 280
definition ............................................................. 11 Gastrointestinal (GI)..................................................... 123
FDA ..................................................................... 12 Gel electrophoresis........................................................ 394
function ............................................................... 11 Gelatinases ..................................................................... 326
functional networks and pathways ........................3 Gene expression ................................................. 14, 26, 35
nucleotides........................................................... 13 Gene knockouts ................................................... 438, 439
therapy selection.................................................. 11 Genome
urine analysis........................................................ 11 COL1A1 ...................................................................... 3
bottom-up approaches.............................................. 27 HBB, HBA1, and HBA2............................................ 3
diagnostic/prognostic discovery HGP............................................................................. 5
biomarker............................................................. 15 proteome ..................................................................... 5
Differentiation Set V0 ......................................... 16 PRSS1 .......................................................................... 3
LC-MS/MS......................................................... 14 PTM............................................................................. 5
pattern diagnostics .............................................. 14 Glucose-6-Phosphate Dehydrogenase
protein-based....................................................... 14 (G-6-PDH) .................................................... 72
“risk assessment” tests ........................................ 17 Glutamine synthetase.................................................... 439
serum proteome .................................................. 15 Gluten .......................................................... 405, 407, 409
systems biology ................................................... 17 Glyceraldehyde 3-phosphate dehydrogenase
drug discovery (GAPDH) .................................................... 440
binary interaction model .................................... 21 Glycoepitope recognition .................................... 426–429
cell-killers ............................................................. 19 Glycolytic enzyme activity
FDA ..................................................................... 20 data transformation and analysis .............................. 78
isoforms ............................................................... 21 electrolyte .................................................................. 72
oligomers ............................................................. 21 G-6-PDH .................................................................. 72
PDD............................................................... 19–22 hexokinase activity..................................................... 72
SNPs .................................................................... 19 IEF .......................................................................75, 76
TDD ..............................................................19, 20 IPG.......................................................................72, 76
EPR ......................................................................32, 34 MS.............................................................................. 79
FP ............................................................................... 32 NADP ........................................................................ 72
genome ................................................................ 3–5, 7 P24............................................................................. 78
HPP ........................................................................... 27 PEP ......................................................................72, 78
IF................................................................................ 32 PEP Universal Protein Purification kit .................... 74
“known” interacting pairs ........................................ 33 protein purity confirmation ...................................... 79
 FUNCTIONAL PROTEOMICS: METHODS AND PROTOCOLS
Index 471
protein staining ......................................................... 73 Immobilized enzyme reactor (IMER) ........................296,
protein transfer ....................................................77, 78 297, 304–306
sample treatment ....................................................... 75 Immobilized metal affinity chromatography
SDS-PAGE gel .......................................................... 71 (IMAC) ............................................... 199, 206
Glycopeptide .......................................422, 429, 430, 434 Immobilized pH gradient (IPG).................................... 72
Glycoproteomics .................................................. 422, 427 Immune regulation .............................................. 318, 322
Glycosidases................................................................... 321 Immunoassay........................................................ 414–417
Glycosylation .............................................. 199–201, 203, Immunofluorescence (IF).........................................28, 32
204, 206, 413, 414, 418, 421, 434 Immunohistochemical staining ........................... 254, 263
Guinea pig lens.............................................................. 437 Immunoproteomics ............................................. 133, 393
Gut microbiome Infectious organisms ............................................ 359–362
adult gastrointestinal tract ...................................... 123 Inflammation .............................................. 314, 316, 321,
challenges................................................................. 128 325, 328, 330, 335, 337–341, 345–347, 352,
diet ........................................................................... 124 356, 359, 362, 366, 370, 372, 377
early childhood........................................................ 124 Inflammatory bowel disease (IBD) ..................... 124, 125
HMP ........................................................................ 124 Inflammatory breast cancer (IBC) ............................... 352
human ...................................................................... 124 In-gel digestion ........................................... 228, 229, 247
inflammation, diseases ............................................ 125 In-solution tryptic digestion ........................................ 268
low abundance ........................................................ 124 Intramembrane proteolysis......................... 331, 334, 335
mass spectrometry instrumentation ....................... 130 Intranuclear proteolysis ....................................... 331, 334
metaproteomics, complex samples................ 125, 126 Iodoacetamide (IAA) ............................................. 70, 267
Pie chart................................................................... 129 Ion chromatogram curve.............................................. 189
sample collection ............................................ 127, 128 Ion exchange chromatography (IEX-HPLC) .... 449, 451
shotgun proteomics approach ................................ 126 Isobaric peptide labels.......................................... 188, 189
software tools ................................................. 128–130 Isobaric Tag for Relative and Absolute
Quantification (iTRAQ).............................. 188
H Isoelectric focusing (IEF) ....................................... 56, 57,
70, 71, 446, 448, 449, 454
Heat shock proteins 27 (HSP27) ................................ 266
Heavy stable isotopes .................................................... 266 Isoforms........................................................................... 21
Hemocytometer ................................................... 266, 267 Isotope dilution MS (IDMS) ....................................... 295
Hemostasis .................................316, 328, 339, 341, 354 Isotope-coded Affinity Tags (ICAT)............................ 187
Hexapeptide ligand libraries ................................ 396, 398
K
Hexokinase activity ...................................................72, 74
High performance liquid chromatography Kallikreins ..................................................................42, 48
(HPLC)..................................... 297, 298, 301, Kluyveromyces lactis (KlAlt1) ........................................ 440
408, 411, 446, 447, 449 KUMASI staining .....................................................65, 67
High-density microarray, see Protein microarray
Higher energy collisional dissociation (HCD)............ 230 L
High-performance liquid chromatography Label-free proteomics ................................................... 254
(HPLC)............................................... 406, 407
Label-free quantitative shotgun proteomics
High-resolution mass spectrometry............................. 285 LC-MS/MS............................................................. 190
Human Microbiome Project (HMP)........................... 124 processing ................................................................ 189
Human Proteome Project (HPP) .................................. 26 protein identification .............................................. 189
Human serum ...................................................... 298, 301
Lacchancea kluyveri (LkAlt1) ....................................... 440
Huntington’s disease .......................................... 324, 331, Laser microdissection system ....................................... 255
339, 344, 355, 357, 366 LCM human pancreatic islets....................................... 256
Hydrazide chemistry ................................... 199, 203, 204
LC-MSMS ..................................................................... 394
Hydrophobic proteins .................................................... 56 Leakproof gel casting system.......................................... 62
Lectin affinity chromatography ........................... 203, 204
I
Lectins................................................................... 414, 418
IgE ............................................................... 393, 394, 397 Ligands ....................................................... 314, 316, 320,
IgE binding proteins............................................ 133, 140 321, 324, 328, 329, 331, 332, 369
IMER-LC-MS/MS.............................296, 297, 299, 303 Lipids ................................................................................. 7
 FUNCTIONAL PROTEOMICS: METHODS AND PROTOCOLS
472 Index
Liquid chromatography, see Staphylococcus aureus MaxQuant ............................................................ 232, 261
Liquid chromatography-electrospray ionization- Mean spectral intensity (MSI) ........................................ 90
tandem mass spectrometry Median effective dose (ED50) ..................................... 156
(LC-ESI-MS/MS)....................................... 225 Median lethal dose (LD50) ................................. 155, 156
Liquid chromatography-tandem mass spectrometry Membrane proteins............................................ 56–58, 60
(LC-MS/MS) ...................................... 14, 254, Mesotrypsin ......................................................... 324, 325,
260, 265, 295, 452, 455 340, 350, 351
Low-abundance allergens ............................................. 393 Metabolic disease .........................................339, 342–344
Metabolic disorders....................................................... 124
M Metabolic enzyme ........................................................... 71
Magnetic beads ........................................... 414, 416, 417 Metabolites ..................................................................7, 12
Metabolomics, see Staphylococcus aureus
MALDI, see Matrix-assisted laser desorption and
ionization (MALDI) Metal-ion based enrichment......................................... 199
MALDI-Imaging MS (MALDI-IMS) ................ 190, 191 Metallo-elastase (MMP-12) ......................................... 326
Metaproteomics analysis, gut microbiome,
MALDI-MS.......................................................... 225, 246
Mascot ............................................... 226, 228, 232, 239, see Gut microbiome
241, 242, 244, 245 Methicillin resistance............................................ 279, 281
Microarray
Mass spectrometry (MS)...................................29, 30, 79,
198, 200–204, 216–217, 394, 405, 422, 430, protein (see Protein microarray)
455, 457 Microbiome .................. 6, 7, 14, 26 See Gut microbiome
biological samples ................................................... 190 Microbiota ..................................................................... 3, 7
Microdissected...................................................... 254, 259
challenges, proteome measures ..................... 179, 180
components ............................................................. 180 MicroRotofor ....................................................... 447–449
dimethylation .......................................................... 227 Mitochondria................................................................... 60
Mitochondria isolation buffer (MIB) ............................ 58
error tolerant searching .......................................... 239
ESI ........................................................................... 181 MMP-1 ................................................326, 346, 366, 371
fractionation ............................................................ 182 MMP-2 and -9 ...........................326, 327, 331, 341, 347
MMP-3, -10, and -11 .......................................... 326, 327
in-gel digestion........................................................ 229
MALDI .................................................................... 181 Moonlighting proteins, see Proteomics
MALDI-IMS .................................................. 190, 191 MoonProt ............................................................. 438, 441
Mascot ..................................................................... 226 MRM, see Multiple reaction monitoring (MRM)
MSConvert GUI ........................................................... 249
mass tolerant searches ............................................. 241
Orbitrap ................................................................... 228 Mucosal lavage ..................................................... 127, 128
protein identification Multifunctional proteins ...................................... 437, 439
Multiple post translational modifications .................... 200
bottom-up approach ................................ 181, 182
top-down approach.................................. 183, 184 Multiple reaction monitoring (MRM) ................. 30, 184
Proteome Discoverer .............................................. 236 Murine double minute.................................................. 266
proteotypic peptide ................................................. 184
N
PTMs ....................................................................... 230
QTOF ...................................................................... 233 Nano-electrospray ionization tandem mass
quantitative proteomics .......................................... 185 spectrometry (nano-ESI-LC-
selective reaction monitoring ................................. 184 MS/MS) ........................................................ 89
shotgun proteomics ................................................ 183 Nanoflow liquid chromatography tandem mass
(see also Shotgun proteomics) spectrometry (nLC-MS/MS) ....................448,
trypsin ...................................................................... 227 452, 453
Unimod ................................................................... 226 Neurodegenerative diseases ................................ 324, 329,
Mass tolerant search.....................................239, 241–243 350, 355–357
Matrilysin ....................................................................... 326 Neutralization potency (P) ........................................... 156
Matrilysin-2 (MMP-26)................................................ 326 New Liberty Proteomics (NLP)....................................... 9
Matrix metalloproteinases (MMPs) ............................314, N-glycans ............................................................. 421, 426,
321, 326, 331, 339–341, 346, 357, 363, 428, 430, 431, 434
366, 376 N-glycomes........................................................... 422, 433
Matrix-assisted laser desorption and ionization Normalization, data ...................................................... 119
(MALDI) ................................... 181, 190, 394 Normalized potency (n-P)................................... 156, 158
 FUNCTIONAL PROTEOMICS: METHODS AND PROTOCOLS
Index 473
Nuclear magnetic resonance spectroscopy separation......................................97–98, 100–101
(NMR) ....................................... 28, 31, 32, 36 SDS-PAGE .............................................................. 100
Nucleotides........................................................... 7, 13, 22 Plasma ...............................318, 321, 329, 335, 346, 347,
NuGel™ ........................................................................... 43 350, 351, 353, 356, 357, 359, 360, 363, 364,
414, 416, 418
O Plasminogen ...................................... 316, 319, 320, 331,
Obesity........................................................................... 124 338, 343, 345, 360, 361, 439–441
Ofatumumab ................................................................. 352 Pleiotropic effects.......................................................... 437
Pollen ...................................................394, 396, 398, 401
“Off-line” MALDI-TOF MS ....................................... 433
18
O labelling.................................................................. 187 Polyacrylamide gel electrophoresis
Oligomers ........................................................................ 21 (PAGE).................................56, 57, 59, 60, 64
Polyethylene glycol (PEG) ............................................. 94
Online digestion............................................................ 297
Orbitrap ....................................................... 228, 230, 233 Polymerase chain reaction (PCR) ................................ 179
Post-translational modifications
P (PTMs)............... 4–6, 28, 107, 226, 230, 413
PPI, see Protein-protein interactions (PPI)
Palmitoylation .....................................199, 201, 203, 204 Precision Medicine (PM)..........................................23–26
Parkinson’s disease ..............................324, 336, 355, 357 Pre-Human Genome Project (HGP)............................... 5
Pathogenic organisms ................................................... 123 Pre-procalcitonin........................................................... 329
PDL, see Proximity-dependent labeling (PDL) Procarboxypeptidases.................................................... 325
PEAKS ........................................................................... 232 Procoagulant .......................................316, 320, 354, 359
PEP Universal Protein Purification kit .......................... 74 Proelastases .................................................................... 325
Peptidase.............................................................. 315, 329, Programmed cell death................................315, 326–331
332, 352, 356, 368 Prolipases ....................................................................... 325
Peptide fractionation......................................98, 101–102 proMMP-2 ...................................................................... 42
Peptide spectral library ................................................... 94 Propeptide ...........................................313, 314, 326, 334
Peptides .............................................................. 3, 7, 9, 21 Prostate cancer ................................................................ 42
Peptide-spectrum matches............................................ 182 Prostate specific antigen (PSA) ...................................... 42
Perseus ........................................................................... 261 Protease degradome...................................................... 315
pFind.............................................................................. 232 Protease inhibitors ............................................ 16, 41–43,
p-Glycoprotein (P-gp) .................................................. 266 316, 319, 326, 341, 353, 360, 363, 366, 367,
Phenotypic drug discovery (PDD) ...........................3, 18, 369, 370, 377
19, 21, 22, 26 Protease promoter....................................... 314, 334, 351
assay ........................................................................... 22 Proteases
cellular level ............................................................... 20 diseases
to TDD................................................................19, 20 autoimmune ...................................................... 356
Phosphatases............................................ 71, 77, 321, 375 cancer (see Cancer)
Phosphoglycerate kinase ............................................... 439 cardiovascular and metabolic diseases .............. 339
Phosphoglycerate mutase ............................................. 439 epigenetics ................................................ 334, 337
Phospholipases .............................................................. 321 infectious organisms.........................359, 360, 362
Phosphorylation ......................... 197, 199–202, 207, 218 inflammation ............................................ 337, 340
Plant proteome neurodegenerative.....................................350–358
acetone/TCA/phenol extraction ......................95, 99 stroke ................................................................. 339
extraction .............................................................95, 99 physiological and regulatory roles................. 314, 316
homogenization ..................................................94, 98 protein digestion ............................................ 313–315
in-gel digestion...................................97–98, 100–101 regulation of physiological processes
in-solution digestion ........................................ 97, 100 cell migration and proliferation................324–328
PEG fractionation ........................................ 94, 95, 99 complement system and immune
peptide regulation.............................................318–323
desalting................................................ 96, 98, 101 DNA replication, repair and
fractionation ................................ 96, 98, 101–102 processing ........................................... 329, 333
protein hemostasis..................................................316–320
digestion .............................................................. 95 intramembrane proteolysis ............................... 335
extraction ............................................................. 95 intranuclear proteolysis ..................................... 331
 FUNCTIONAL PROTEOMICS: METHODS AND PROTOCOLS
474 Index
programmed cell death ..................................... 326 assembled arrays ...................................................... 109
protein secretion ............................................... 329 bead arrays ............................................................... 109
proteolytic processing ....................................... 321 computational analysis (see Computational
tissue remodeling .............................................. 324 analysis)
transmembrane proteolysis ...................... 334, 335 computational resources ......................................... 113
Proteasome ................................ 314, 317, 321, 329–331, customize design ............................................ 113, 114
350, 352, 359, 360, 362, 368, 369, 375 experimental workflow............................................ 116
Protein high-density............................................................. 108
Differentiation Set V0 ............................................... 16 image acquisition
digestion .................................................................... 89 experimental workflow...................................... 117
functionality............................................................... 21 materials............................................................. 113
HGP............................................................................. 4 planar arrays.................................................... 108, 109
identification........................................................27, 87 reverse-phase arrays................................................. 109
interaction............................................................22, 27 self-assembled arrays ............................................... 110
quantitation ............................................................... 90 Protein purification ......................................................... 74
separation.............................................................84, 86 Protein quantification .......................................... 266, 274
Uniprot ........................................................................ 4 Protein secretion ......................................... 315, 329, 332
visualization ............................................................... 86 fungi......................................................................... 165
Protein complexes ......................................................... 145 in silico prediction ................................................... 163
biochemical separation............................................ 445 lipoproteins..................................................... 164, 165
computational proteomics analysis ........................ 452 machine learning approaches.................................. 163
E. coli ....................................................................... 447 materials................................................................... 160
fractionation ............................................................ 447 nonclassical .............................................................. 166
HPLC ...................................................................... 446 SignalP ............................................................ 163–165
IEF ........................................................................... 448 WoLF PSORT ......................................................... 165
IEX-HPLC .............................................................. 449 Protein staining ............................................................... 73
LC-MS/MS............................................................. 452 ProteinPilot ................................................................... 232
nLC-MS/MS .......................................................... 448 Protein–protein interactions (PPIs) ................... 143–145,
non-denaturing isoelectric focusing....................... 447 150, 438, 439, 445
PPIs .......................................................................... 445 Proteoforms............................................. 44, 52, 183, 296
soluble protein extraction ....................................... 448 Proteolysis
TCA ......................................................................... 451 complexity ............................................................... 316
trypsin digestion...................................................... 452 cytosolic ................................................................... 331
2-D IEF-IEX ........................................................... 446 drug and diagnostic development
2-D proteomic fractionation .................................. 446 active site targeting ........................................... 368
Protein Data Bank (PDB) .............................................. 31 exosite and effector binding sites ..................... 371
Protein elution plate (PEP) ............................................ 71 inhibitor biomarkers .................................375–377
glycolytic enzyme activity ......................................... 72 mechanism-based targeting .............................. 370
protein purification ................................................... 74 drug targets (see Drug targets)
protein transfer .......................................................... 77 intramembrane ........................................................ 335
Protein fractionation, see Plant proteome organelle membranes .............................................. 329
Protein function prediction .......................................... 437 regulatory mechanisms ........................................... 316
Protein interactions............................................... 5, 9, 22, transmembrane............................................... 334, 335
28, 33, 34, 37 Proteolytic enzymes ............................................. 314, 315
Protein isoforms (proteoforms) ................................... 183 Proteolytic processing.......................................... 321–325
Protein kinases................................................................. 71 Proteome Discoverer .......................................... 228, 232,
Protein markers ........................................... 413, 414, 418 233, 236, 239, 248, 249
Protein microarray Proteomics................................................... 107, 108, 393
array assays bioinformatics analysis ............................................ 441
experimental workflow...................................... 116 D1-DE (see Double-one dimensional
materials.................................................... 112, 113 electrophoresis (D1-DE))
array printing expression patterns/cellular localization ............... 440
experimental workflow...................................... 114 gene knockouts ....................................................... 439
materials.................................................... 111, 112 MS (see Mass spectrometry (MS))
 FUNCTIONAL PROTEOMICS: METHODS AND PROTOCOLS
Index 475
protein–protein interactions ................................... 439 Serine suicidal protease inhibitors .................................. 42
(see also Targeted proteomics) SERPIN
Proteomics methods appraisal ..................................28, 29 AAT............................................................................ 44
Proteotypic peptide....................................................... 184 AlbuSorb™ ...........................................................43, 46
ProteoWizard .............................................. 249, 456, 458 AlbuSorb™ PLUS ..................................................... 43
Proximity-dependent labeling (PDL) .......................... 144 AlbuVoid™ ...........................................................43, 46
PTM crosstalk ............................................................... 206 biomarkers .................................................... 42, 44, 53
PTM sites.............................................................. 198, 207 function ........................................................ 42–44, 48
LC-MS .................................................................47, 52
Q mechanisms .................................................. 42, 44, 53
Q-Orbitrap mass spectrometer..................................... 283 NuGel™ ..................................................................... 43
protease...................................................................... 43
QTOF ................................................................... 230, 233
Quantitation ............................................... 184, 186, 187, proteoforms ............................................................... 53
189, 191, 200, 201, 211, 217 PSA ............................................................................ 42
RCL .....................................................................42, 44
Quantitative.......................................................... 295, 297
Quantitative proteomics SERPINA1 ................................................................ 42
gel-based methods .................................................. 185 SERPINC1 ................................................................ 42
SERPINF2................................................................. 42
protein expression ................................................... 185
Quantitative shotgun proteomics SERPING1................................................................ 42
label-free (see Label-free quantitative shotgun tissue kallikreins......................................................... 42
proteomics) Serpins................................ 316, 319, 337, 360, 363, 372
Serum.................................................................... 414, 418
labelling
18
O labelling...................................................... 187 Shotgun proteomics............................................. 126, 130
dimethyl isotopic labelling................................ 187 data-dependent and independent .......................... 183
protein identification .............................................. 182
ICAT .................................................................. 187
isobaric peptide ........................................ 188, 189 quantitative (see Quantitative shotgun proteomics)
SILAC ................................................................ 186 Signaling .................................... 316, 321, 324–328, 335,
339–341, 345, 351–353, 356, 363, 366, 367
super-SILAC...................................................... 186
Quinone oxidoreductase .............................................. 437 Signaling pathways ...................................... 198, 201, 216
Single-nucleotide polymorphisms (SNPs)..................... 19
R Snake venom
ESI-LC-MS/MS ....................................................... 89
Reactive center loop (RCL)............................... 42, 44–46 peptides extraction and desalting ............................. 86
Reversed-phase high performance liquid chromatography protein
(RP-HPLC) .......................................... 84, 229 digestion ........................................................85, 89
Reverse-phase arrays...................................................... 109 identification........................................................ 87
Rituximab ............................................................. 352, 356 and peptides......................................................... 84
RNA splice variants ....................................................... 437 quantitation ......................................................... 90
separation............................................................. 86
S
visualization ......................................................... 86
Saccharomyces cerevisiae................................200–204, 440 RP-HPLC .................................................................. 84
Secretomics, see CAZymes in microbial secretomes SDS-PAGE ................................................... 84, 85, 87
SELDI............................................................................ 394 separation................................................................... 91
Selected/multiple reaction monitoring Snake venom proteomes
(SRM/MRM)....................................... 93, 265 antivenom (see Antivenom)
Selective reaction monitoring (SRM) ................. 184, 455 complexity ............................................................... 153
Self-assembled arrays..................................................... 110 death and survival ................................................... 156
Sequential window acquisition of all theoretical evolution, clinical syndrome................................... 154
spectra (SWATH) .......................................... 94 lethality determination............................................ 155
Sequential windowed acquisition of all theoretical lethality neutralization ................................... 156, 157
fragment ion mass spectra materials................................................................... 154
(SWATH-MS) ........... 184, 455, 456, 458–460 median effective dose.............................................. 156
SEQUEST ..................................................................... 232 median lethal dose .................................................. 156
Serial enrichment of PTMs........................................... 199 toxicity ..................................................................... 154
 FUNCTIONAL PROTEOMICS: METHODS AND PROTOCOLS
476 Index
Sodium dodecyl sulphate polyacrylamide gel Three-dimensional structures....................................... 441
electrophoresis (SDS-PAGE) .....................423, Threonine proteases............................................. 315, 360
426, 427, 429–431 Thrombomodulin ...................................... 316, 318–320,
Sodium dodecylsulfate (SDS)............................ 56, 57, 60 336, 342, 349
Spectral counting .......................................................... 189 Tight-binding tissue inhibitors of metalloproteinases
Spectral library.....................................455, 456, 459, 462 (TIMPs) ....................................................... 326
Stable isotope-labeled (SIL) ............................... 295, 296, TIMP-1 (tissue inhibitor of metallopeptidase 1) ........ 414
298, 300, 302, 304, 307 TMT, see Tandem Mass Tags (TMT)
Stable isotopic labelling of amino acids in Toxin-based drug discovery ........................................... 84
culture (SILAC)........................................... 186 Transcription elongation factor.................................... 439
Staphylococcus aureus Transferrin (Tr) ............................................................. 266
bacteria sampling and metabolism quenching ...... 285 Transferrin receptor (TfR)............................................ 266
bacterial culture ....................................................... 282 Transmembrane domains (TMDs) ..........................56, 57
bacterial pre-culture and culture ............................ 284 Transmembrane proteolysis........................ 331, 334, 335
bacterial strains ........................................................ 281 Trichloroacetic acid (TCA)........................................... 451
GC-MS .................................................................... 280 Trypsin ........................................................ 227, 230, 238,
intracellular metabolites................................. 285–287 406, 407, 409
LC-MS analysis........................................................ 283 Trypsin digestion ................................................... 85, 452
metabolite extraction ..................................... 282–283 Tryptic peptide mapping ............................ 424, 428, 429
MRSA ...................................................................... 279 Tumor markers.............................................................. 414
Streptococci ................................................................... 440 Two dimensional differential gel electrophoresis
Streptomyces coelicolor .................................................... 441 (2D-DIGE).................................................. 126
Stroke.......................................... 331, 339–348, 363, 367 Two-dimensional electrophoresis
Stromelysins................................................................... 326 (2-DE).........................................133–135, 140
Strong anion exchange (SAX) ............................. 229, 249 2-D gel electrophoresis.............................................56, 70
Strong cation exchange (SCX) peptide 2-D gel separation........................................................... 72
fractionation.......................................... 94, 102 2D IEF/SDS PAGE ....................................................... 57
Succinylation .......................................199, 201, 202, 205 2-D IEF-IEX ................................................................. 446
Sulfatases........................................................................ 321 2-D proteomic fractionation ........................................ 446
Surrogate peptide selection .......................................... 269 2-D separation............................................................... 446
SwissProt ....................................................................... 233
Systems biology............................................................. 265 U
Drug discovery ............................................................ 8 Ubiquitination..............................................197, 199–201
molecular level........................................................... 10 Unexpected modifications .................................. 226, 228,
NLP ............................................................................. 9
230, 232, 233, 239
Unimod ....................................................... 226, 239, 245
T
Uniprot .............................................................................. 4
Tandem mass spectrometry (MS/MS) ................ 84, 126, Urokinase............................................................. 319, 320,
181, 186–188, 190 324, 339, 344, 347, 349
Tandem mass tag labeling ............................................ 208
Tandem Mass Tags (TMT)........................................... 188 V
Target-based drug discovery (TDD) .......................18, 19 Validation..................................................... 226, 242, 245
Targeted proteomics Venom proteome ................................................. 153, 154
cell viability test ....................................................... 266
Venom separation............................................................ 91
LC-MS/MS............................................................. 265 Venom toxicity .............................................................. 154
MS/MS instrument ................................................ 267 Venomics ......................................................................... 83
protein extraction.................................................... 268
protein extraction buffers ....................................... 266 W
SRM/MRM ............................................................ 265
surrogate peptide selection..................................... 269 Wash Buffer AVWB......................................................... 50
tissue homogenization ............................................ 266 Western blotting.................................................. 422, 423,
tryptic digestion buffer ........................................... 267 428, 430, 431, 434
Thioredoxin peroxidase ................................................ 439
 FUNCTIONAL PROTEOMICS: METHODS AND PROTOCOLS
Index 477
X Z
X!Tandem ............................................................. 232, 462 Zeta-crystallin................................................................ 437
X-ray crystallography .....................................31, 315, 372 Zymogens ..................................................... 41, 314, 326,
341, 360, 371
Y
Yeast two-hybrid (Y2H) ..................................28, 34, 439