Methods in

Molecular Biology 2247

Arnaud Poterszman Editor

Multiprotein
Complexes
Methods and Protocols
METHODS IN MOLECULAR BIOLOGY

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

For further volumes:

http://www.springer.com/series/7651
For over 35 years, biological scientists have come to rely on the research protocols and
methodologies in the critically acclaimed Methods in Molecular Biology series. The series was
the first to introduce the step-by-step protocols approach that has become the standard in all
biomedical protocol publishing. Each protocol is provided in readily-reproducible step-by-
step fashion, opening with an introductory overview, a list of the materials and reagents
needed to complete the experiment, and followed by a detailed procedure that is supported
with a helpful notes section offering tips and tricks of the trade as well as troubleshooting
advice. These hallmark features were introduced by series editor Dr. John Walker and
constitute the key ingredient in each and every volume of the Methods in Molecular Biology
series. Tested and trusted, comprehensive and reliable, all protocols from the series are
indexed in PubMed.
 Multiprotein Complexes

Methods and Protocols

Edited by

Arnaud Poterszman
Institut de Génétique et de Biologie Moléculaire et Cellulaire, Center for Integrated Biology, Integrated
Structural Biology Department, Equipe labellisée Ligue Contre le Cancer, CNRS UMR 7104 - Inserm U
1258, University of Strasbourg, Illkirch, France
Editor
Arnaud Poterszman
Institut de Génétique et de Biologie
Moléculaire et Cellulaire, Center for
Integrated Biology, Integrated
Structural Biology Department, Equipe
labellisée Ligue Contre le Cancer
CNRS UMR 7104 - Inserm U 1258
University of Strasbourg
Illkirch, France

ISSN 1064-3745 ISSN 1940-6029 (electronic)

Methods in Molecular Biology
ISBN 978-1-0716-1125-8 ISBN 978-1-0716-1126-5 (eBook)
https://doi.org/10.1007/978-1-0716-1126-5

© Springer Science+Business Media, LLC, part of Springer Nature 2021

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,
expressed 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 imprint is published by the registered company Springer Science+Business Media, LLC, part of Springer
Nature.
The registered company address is: 1 New York Plaza, New York, NY 10004, U.S.A.
Preface

Molecular complexes of interacting proteins govern virtually all biological processes such as
metabolism, cell signaling, DNA repair, or gene expression. Macromolecular assemblies are
also of great biomedical relevance as factors that perturb biomolecular interaction networks
underlie a number of diseases, and deliberate inhibition of protein–protein interactions is an
increasingly common strategy in drug discovery initiatives. Unraveling their functions and
mechanisms of action is often only potentially accessible through a detailed structural
description and the integration of dynamic information. This volume of the Methods in
Molecular Biology series aims to provide the scientific community with strategies and detailed
protocols for the preparation of macromolecular complexes and their characterization in
view of structural analysis.
Protein engineering and production are essential tools for structural, biophysical, and
functional studies as well as for biotechnology and medical applications. Strategies to
prepare proteins and protein complexes have tremendously been improved in part thanks
to recent structural genomic programs. Yet, no universal solution has been implemented,
and the production and/or reconstitution of protein complexes remains a major bottleneck.
This is in particular the case for complexes composed of many subunits which are often
incompletely characterized. Additional difficulties result from their low natural abundance
or their versatile nature, in part because regulation often involves the formation of transient
complexes with low binding constants and in part because their composition varies with the
physiological context.
The first section of this book focuses on sample preparation. While Chapters 1 and
2 concentrate on strategies for recombinant expression of multiprotein complexes in pro-
karyotic and eukaryotic hosts, Chapter 3 illustrates how genome editing with the CRISPR-
Cas9 system can be used to precisely modify protein coding genes in mammalian cells; this
offers the possibility to replace any coding gene by a reshaped version fused to an affinity tag
protein or to a fluorescent reporter, enabling the characterization of endogenous macromo-
lecular complexes expressed under near physiological conditions. In the first section, the
production of recombinant antibodies and artificial binding proteins which emerge as key
reagents in multiprotein complex research, as inhibitors of protein–protein interactions to
modulate activity, as probes for cellular imaging or to facilitate structural determination is
also discussed. Chapter 4 details the production of recombinant antibodies in different
formats, Chapter 5 the intervening removable affinity tag (iRAT) system for the production
of recombinant antibody fragments, and Chapter 6 the isolation of artificial binding proteins
(Affimer reagents) based on a non-antibody scaffold.
The second section of the book details a set of biophysical methods that can provide
useful indicators for sample optimization and often complement structural information
obtained with core technologies for structure determination (X-ray crystallography, nuclear
magnetic resonance, and cryo-electron microscopy) by quantitative solution data, helping to
understand how biological systems function. Three chapters focus on techniques for the
biophysical characterization of biological macromolecules and their complexes. Chapter 7
details interaction measurements of protein–DNA complexes by isothermal titration calo-
rimetry (ITC) and microscale thermophoresis (MST), Chapter 8 the use of the switch-
SENSE technology for the analysis of enzyme kinetics, and Chapter 9 the sedimentation

v
vi Preface

velocity methods for the characterization of protein heterogeneity and protein affinity
interactions. Another set of articles describe mass spectrometry (MS)-based approaches
for macromolecular complex analysis. Chapter 10 focuses on the applications of native
mass spectrometry for the characterization of multiprotein complexes ranging from 16 to
801 kDa, Chapter 11 on hydrogen/deuterium exchange mass spectrometry for analyzing
protein–DNA interactions, and Chapter 12 on integrative mass spectrometry-based
approaches for modeling macromolecular assemblies.
Although high‐resolution structure determination using X-ray crystallography or single‐
particle cryo‐electron microscopy (cryo‐EM) is now producing a rapid stream of break-
throughs in structural biology, the preparation of suitable crystals or hydrated frozen
samples on EM grids is often quite challenging. Purified samples, intact and structurally
homogeneous in the test tube, may not crystallize or survive the standard methods of
preparing thin aqueous films on grids. In the case of cryo-EM, optimization of sample
stability and extensive screening of parameters for grid preparation are often required to
collect high-quality datasets. The two last chapters of this section address sample optimiza-
tion. Chapter 13 provides detailed protocols for preparing negatively stained and hydrated
frozen EM-grid while Chapter 14 addresses solubilization screening using membrane
proteins as model system.
Finally, the third section of this book addresses the characterization of multiprotein
complexes in a cellular environment using state-of-the-art imaging technologies and in vivo
approaches. Chapter 15 describes practical aspects of super-resolution imaging and
Chapter 16 multi-color FRET-FLIM microscopy in live cells. Chapters 17 and 18 present
applications of directed evolution systems and of context-specific and proximity-dependent
labeling using the BioID technology.
This book is expected to be used not only by structural/molecular biologists who need
to prepare multi-components complexes for their own applications but also by scientists
from other fields who are working on macromolecular assemblies from other standpoints
and need an overview of state-of-art approaches. I am especially thankful to all the authors
for their great contributions, devoting their valuable time to the preparation of the manu-
scripts. I am also indebted to the Series Editor John M. Walker, to the editorial staff
members of Springer and to Marie Christine Poterszman for their kind support in making
this book publishable. I hope that this volume provides a useful overview preparation and
structural analysis of macromolecular complexes and fills a need for well-described hands-on
protocols.

Illkirch, France Arnaud Poterszman

Contents

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
Contributors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix

PART I RECOMBINANT EXPRESSION AND ISOLATION FROM ENDOGENOUS SOURCE

1 Production of Multi-subunit Membrane Protein Complexes . . . . . . . . . . . . . . . . . 3

Burak V. Kabasakal, Qiyang Jiang,
and Christiane Schaffitzel
2 Production of Multiprotein Complexes Using the Baculovirus
Expression System: Homology-Based and Restriction-Free
Cloning Strategies for Construct Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Paola Rossolillo, Olga Kolesnikova, Karim Essabri,
Gala Ramon Zamorano, and Arnaud Poterszman
3 Tagging Proteins with Fluorescent Reporters Using the CRISPR/Cas9
System and Double-Stranded DNA Donors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Sylvain Geny, Simon Pichard, Alice Brion,
Jean-Baptiste Renaud, Sophie Jacquemin,
Jean-Paul Concordet, and Arnaud Poterszman
4 Validation of the Production of Antibodies in Different Formats
in the HEK 293 Transient Gene Expression System . . . . . . . . . . . . . . . . . . . . . . . . . 59
Jens König, Michael Hust, and Joop van den Heuvel
5 The Intervening Removable Affinity Tag (iRAT) System
for the Production of Recombinant Antibody Fragments . . . . . . . . . . . . . . . . . . . . 77
Norimichi Nomura, Yayoi Nomura,
Yumi Sato, and So Iwata
6 Isolation of Artificial Binding Proteins (Affimer Reagents)
for Use in Molecular and Cellular Biology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Anna A. S. Tang, Christian Tiede,
Michael J. McPherson, and Darren C. Tomlinson

PART II BIOPHYSICAL CHARACTERIZATION AND SAMPLE OPTIMIZATION

7 Measurements of Protein–DNA Complexes Interactions

by Isothermal Titration Calorimetry (ITC)
and Microscale Thermophoresis (MST). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Amandine Gontier, Paloma F. Varela, Clément Nemoz,
Virginie Ropars, Magali Aumont-Nicaise,
Michel Desmadril, and Jean-Baptiste Charbonnier
8 switchSENSE Technology for Analysis of DNA Polymerase Kinetics . . . . . . . . . . 145
Guillaume Bec and Eric Ennifar
9 Sedimentation Velocity Methods for the Characterization
of Protein Heterogeneity and Protein Affinity Interactions . . . . . . . . . . . . . . . . . . . 155
Christine Ebel and Catherine Birck
vii
viii Contents

10 Hands on Native Mass Spectrometry Analysis

of Multi-protein Complexes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
Stéphane Erb, Sarah Cianférani, and Julien Marcoux
11 Studying Protein–DNA Interactions by Hydrogen/Deuterium Exchange
Mass Spectrometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
Ruzena Filandrova, Daniel Kavan,
Alan Kadek, Petr Novak, and Petr Man
12 Integrative Mass Spectrometry–Based Approaches for Modeling
Macromolecular Assemblies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Andy M. Lau and Argyris Politis
13 Optimization of Sample Preparation for the Observation
of Macromolecular Complexes by Electron (cryo-)Microscopy . . . . . . . . . . . . . . . 243
Alexandre Frechard, Grigory Sharov,
Maximilien Werderer, and Patrick Schultz
14 Solubilization and Stabilization of Native Membrane Proteins
for Drug Discovery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
Vincent Corvest and Anass Jawhari

PART III ANALYSIS IN A CELLULAR CONTEXT

15 Practical Aspects of Super-Resolution Imaging and Segmentation

of Macromolecular Complexes by dSTORM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
Leonid Andronov, Jean-Luc Vonesch,
and Bruno P. Klaholz
16 Multicolor FRET-FLIM Microscopy to Analyze Multiprotein
Interactions in Live Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
Abdullah Ahmed, Jennifer Schoberer,
Emily Cooke, and Stanley W. Botchway
17 Context-Specific and Proximity-Dependent Labeling
for the Proteomic Analysis of Spatiotemporally Defined
Protein Complexes with Split-BioID . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
Cinthia Amaya Ramirez, Stefanie Egetemaier,
and Julien Béthune
18 A Directed Evolution System for Lysine Deacetylases . . . . . . . . . . . . . . . . . . . . . . . 319
Martin Spinck, Maria Ecke,
Damian Schiller, and Heinz Neumann

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
Contributors

ABDULLAH AHMED • Central Laser Facility, Science and Technology Facilities Council
(STFC) Rutherford Appleton Laboratory, Research Complex at Harwell, Oxford, UK;
Plant Cell Biology, Biological and Medical Sciences, Oxford Brookes University, Oxford, UK
LEONID ANDRONOV • Centre for Integrative Biology (CBI), Institut de Génétique et de
Biologie Moléculaire et Cellulaire (IGBMC), Centre National de la Recherche Scientifique
(CNRS), UMR 7104, Institut National de la Santé et de la Recherche Médicale
(INSERM), U1258, Université de Strasbourg, Illkirch, France
MAGALI AUMONT-NICAISE • Institute for Integrative Biology of the Cell (I2BC), Université
Paris-Saclay, CEA, CNRS, Gif-s-Yvette, France
GUILLAUME BEC • Institut de Biologie Moléculaire et Cellulaire, Centre National de la
Recherche Scientifique (CNRS), Université de Strasbourg, Strasbourg, France
JULIEN BÉTHUNE • Hamburg University of Applied Sciences, Hamburg, Germany; Cluster of
Excellence CellNetworks, Heidelberg University, Heidelberg, Germany; Heidelberg
University Biochemistry Center (BZH), Heidelberg, Germany
CATHERINE BIRCK • Institut de Génétique et de Biologie Moléculaire et Cellulaire (IGBMC),
Centre National de la Recherche Scientifique (CNRS), UMR 7104, Institut National de
la Santé et de la Recherche Médicale (INSERM), U1258, Université de Strasbourg, Illkirch,
France
STANLEY W. BOTCHWAY • Central Laser Facility, Science and Technology Facilities Council
(STFC) Rutherford Appleton Laboratory, Research Complex at Harwell, Oxford, UK;
Plant Cell Biology, Biological and Medical Sciences, Oxford Brookes University, Oxford, UK
ALICE BRION • Laboratoire Structure et Instabilité des Génomes, Muséum National
d’Histoire Naturelle (MNHN), Institut National de la Santé et de la Recherche Médicale
(INSERM), U1154, Centre National de la Recherche Scientifique (CNRS), UMR7196,
Paris, France
JEAN-BAPTISTE CHARBONNIER • Institute for Integrative Biology of the Cell (I2BC),
Université Paris-Saclay, CEA, CNRS, Gif-s-Yvette, France
SARAH CIANFÉRANI • Laboratoire de Spectrométrie de Masse BioOrganique, Institut
Pluridisciplinaire Hubbert Curien (IPHC), Centre National de la Recherche Scientifique
(CNRS), UMR 7178, Université de Strasbourg, Strasbourg, France
JEAN-PAUL CONCORDET • Laboratoire Structure et Instabilité des Génomes, Muséum
National d’Histoire Naturelle (MNHN), Institut National de la Santé et de la Recherche
Médicale (INSERM), U1154, Centre National de la Recherche Scientifique (CNRS),
UMR7196, Paris, France
EMILY COOKE • Central Laser Facility, Science and Technology Facilities Council (STFC)
Rutherford Appleton Laboratory, Research Complex at Harwell, Oxford, UK
VINCENT CORVEST • CALIXAR, Lyon, France
MICHEL DESMADRIL • Institute for Integrative Biology of the Cell (I2BC), Université Paris-
Saclay, CEA, CNRS, Gif-s-Yvette, France
CHRISTINE EBEL • Institut de Biologie Structurale (IBS), Univ Grenoble Alpes,
Commissariat à l’énergie atomique et aux énergies alternatives (CEA), Centre National de
la Recherche Scientifique (CNRS), Grenoble, France

ix
x Contributors

MARIA ECKE • Department of Structural Biochemistry, Max-Planck-Institute of Molecular

Physiology, Dortmund, Germany
STEFANIE EGETEMAIER • Cluster of Excellence CellNetworks, Heidelberg University,
Heidelberg, Germany; Heidelberg University Biochemistry Center (BZH), Heidelberg,
Germany
ERIC ENNIFAR • Institut de Biologie Moléculaire et Cellulaire, Centre National de la
Recherche Scientifique (CNRS), Université de Strasbourg, Strasbourg, France
STÉPHANE ERB • Laboratoire de Spectroméérie de Masse BioOrganique, Universitéde
Strasbourg, CNRS, IPHC UMR7178, Strasbourg, France
KARIM ESSABRI • Institut de Génétique et de Biologie Moléculaire et Cellulaire (IGBMC),
Centre National de la Recherche Scientifique (CNRS), UMR 7104, Institut National de
la Santé et de la Recherche Médicale (INSERM), U1258, Université de Strasbourg, Illkirch,
France
RUZENA FILANDROVA • Institute of Microbiology of the Czech Academy of Sciences, Prague,
Czech Republic; Faculty of Sciences, Charles University, Prague, Czech Republic
ALEXANDRE FRECHARD • Institut de Génétique et de Biologie Moléculaire et Cellulaire
(IGBMC), Integrated Structural Biology, Equipe labellisée Ligue Contre le Cancer, Centre
National de la Recherche Scientifique (CNRS), UMR 7104, Institut National de la Santé
et de la Recherche Médicale (INSERM), U1258, Université de Strasbourg, Illkirch, France
SYLVAIN GENY • Laboratoire Structure et Instabilité des Génomes, Muséum National
d’Histoire Naturelle (MNHN), Institut National de la Santé et de la Recherche Médicale
(INSERM), U1154, Centre National de la Recherche Scientifique (CNRS), UMR7196,
Paris, France
AMANDINE GONTIER • Institute for Integrative Biology of the Cell (I2BC), Université Paris-
Saclay, CEA, CNRS, Gif-s-Yvette, France; Structural Biology and Biophysics (SBB), Bio
Structure and Biophysics (BSB), Sanofi, Vitry-sur-Seine, France
MICHAEL HUST • Technische Unversit€ a t Braunschweig, Institut für Biochemie, Biotechnologie
und Bioinformatik, Abteilung Biotechnologie, Braunschweig, Germany; YUMAB GmbH,
Braunschweig, Germany
SO IWATA • Department of Cell Biology, Graduate School of Medicine, Kyoto University,
Kyoto, Japan
SOPHIE JACQUEMIN • Institut de Génétique et de Biologie Moléculaire et Cellulaire (IGBMC),
Equipe labellisée Ligue Contre le Cancer, Centre National de la Recherche Scientifique
(CNRS), UMR 7104, Institut National de la Santé et de la Recherche Médicale
(INSERM), U1258, Université de Strasbourg, Illkirch, France
ANASS JAWHARI • Independent Advisor, Lyon, France
QIYANG JIANG • Viva Biotech (Shanghai) Limited, Shanghai, China
BURAK V. KABASAKAL • School of Biochemistry, University of Bristol, Bristol, UK; Institute of
Accelerator Technologies, Ankara University, Gölbaçı, Ankara, Turkey
ALAN KADEK • Institute of Microbiology of the Czech Academy of Sciences, Prague, Czech
Republic; Heinrich Pette Institute, Leibniz Institute for Experimental Virology, Hamburg,
Germany
DANIEL KAVAN • Institute of Microbiology of the Czech Academy of Sciences, Prague, Czech
Republic
BRUNO P. KLAHOLZ • Centre for Integrative Biology (CBI), Institut de Génétique et de
Biologie Moléculaire et Cellulaire (IGBMC), Centre National de la Recherche Scientifique
(CNRS), UMR 7104, Institut National de la Santé et de la Recherche Médicale
(INSERM), U1258, Université de Strasbourg, Illkirch, France
 Contributors xi

OLGA KOLESNIKOVA • Institut de Génétique et de Biologie Moléculaire et Cellulaire

(IGBMC), Département of Integrated Structural Biology, Equipe Labellisée Ligue, Centre
National de la Recherche Scientifique (CNRS), UMR 7104, Institut National de la Santé
et de la Recherche Médicale (INSERM), U1258, Université de Strasbourg, Illkirch, France;
Structural and Computational Biology Unit, EMBL Heidelberg, Heidelberg, Germany
JENS KÖNIG • Department of Structure and Function of Proteins, Helmholtz Zentrum für
Infektionsforschung GmbH, Braunschweig, Germany
ANDY M. LAU • Department of Chemistry, King’s College London, London, UK
PETR MAN • Institute of Microbiology of the Czech Academy of Sciences, Prague, Czech
Republic
JULIEN MARCOUX • Institut de Pharmacologie et de Biologie Structurale (IPBS), Université
de Toulouse, CNRS, UPS, Toulouse, France
MICHAEL J. MCPHERSON • Astbury Centre for Structural and Molecular Biology, School of
Molecular and Cellular Biology, University of Leeds, Leeds, UK
CLÉMENT NEMOZ • Institute for Integrative Biology of the Cell (I2BC), Université Paris-
Saclay, CEA, CNRS, Gif-s-Yvette, France
HEINZ NEUMANN • Department of Structural Biochemistry, Max-Planck-Institute of
Molecular Physiology, Dortmund, Germany; Department of Chemical Engineering and
Biotechnology, University of Applied Sciences, Darmstadt, Germany
NORIMICHI NOMURA • Department of Cell Biology, Graduate School of Medicine, Kyoto
University, Kyoto, Japan
YAYOI NOMURA • Department of Cell Biology, Graduate School of Medicine, Kyoto University,
Kyoto, Japan
PETR NOVAK • Institute of Microbiology of the Czech Academy of Sciences, Prague, Czech
Republic
SIMON PICHARD • Institut de Génétique et de Biologie Moléculaire et Cellulaire (IGBMC),
Equipe labellisée Ligue Contre le Cancer, Centre National de la Recherche Scientifique
(CNRS), UMR 7104, Institut National de la Santé et de la Recherche Médicale
(INSERM), U1258, Université de Strasbourg, Illkirch, France
ARGYRIS POLITIS • Department of Chemistry, King’s College London, London, UK
ARNAUD POTERSZMAN • Institut de Génétique et de Biologie Moléculaire et Cellulaire, Center
for Integrated Biology, Integrated Structural Biology Department, Equipe labellisée Ligue
Contre le Cancer, CNRS UMR 7104 - Inserm U 1258, University of Strasbourg, Illkirch,
France
CINTHIA AMAYA RAMIREZ • Cluster of Excellence CellNetworks, Heidelberg University,
Heidelberg, Germany; Heidelberg University Biochemistry Center (BZH), Heidelberg,
Germany
JEAN-BAPTISTE RENAUD • Laboratoire Structure et Instabilité des Génomes, Muséum National
d’Histoire Naturelle (MNHN), Institut National de la Santé et de la Recherche Médicale
(INSERM), U1154, Centre National de la Recherche Scientifique (CNRS), UMR7196,
Paris, France
VIRGINIE ROPARS • Institute for Integrative Biology of the Cell (I2BC), Université Paris-
Saclay, CEA, CNRS, Gif-s-Yvette, France
PAOLA ROSSOLILLO • Institut de Génétique et de Biologie Moléculaire et Cellulaire (IGBMC),
Centre National de la Recherche Scientifique (CNRS), UMR 7104, Institut National de
la Santé et de la Recherche Médicale (INSERM), U1258, Université de Strasbourg, Illkirch,
France
xii Contributors

YUMI SATO • Department of Cell Biology, Graduate School of Medicine, Kyoto University,
Kyoto, Japan
CHRISTIANE SCHAFFITZEL • School of Biochemistry, University of Bristol, Bristol, UK
DAMIAN SCHILLER • Department of Structural Biochemistry, Max-Planck-Institute of
Molecular Physiology, Dortmund, Germany
JENNIFER SCHOBERER • Department of Applied Genetics and Cell Biology, University of
Natural Resources and Life Sciences, BOKU Vienna, Vienna, Austria
PATRICK SCHULTZ • Institut de Génétique et de Biologie Moléculaire et Cellulaire (IGBMC),
Integrated Structural Biology, Equipe labellisée Ligue Contre le Cancer, Centre National
de la Recherche Scientifique (CNRS), UMR 7104, Institut National de la Santé et de la
Recherche Médicale (INSERM), U1258, Université de Strasbourg, Illkirch, France
GRIGORY SHAROV • Institut de Génétique et de Biologie Moléculaire et Cellulaire (IGBMC),
Integrated Structural Biology, Equipe labellisée Ligue Contre le Cancer, Centre National
de la Recherche Scientifique (CNRS), UMR 7104, Institut National de la Santé et de la
Recherche Médicale (INSERM), U1258, Université de Strasbourg, Illkirch, France; EM
Facility, MRC Laboratory of Molecular Biology, Cambridge, UK
MARTIN SPINCK • Department of Structural Biochemistry, Max-Planck-Institute of
Molecular Physiology, Dortmund, Germany
ANNA A. S. TANG • Astbury Centre for Structural and Molecular Biology, School of Molecular
and Cellular Biology, University of Leeds, Leeds, UK
CHRISTIAN TIEDE • Astbury Centre for Structural and Molecular Biology, School of Molecular
and Cellular Biology, University of Leeds, Leeds, UK
DARREN C. TOMLINSON • Astbury Centre for Structural and Molecular Biology, School of
Molecular and Cellular Biology, University of Leeds, Leeds, UK
JOOP VAN DEN HEUVEL • Department of Structure and Function of Proteins, Helmholtz
Zentrum für Infektionsforschung GmbH, Braunschweig, Germany
PALOMA F. VARELA • Institute for Integrative Biology of the Cell (I2BC), Université Paris-
Saclay, CEA, CNRS, Gif-s-Yvette, France
JEAN-LUC VONESCH • Centre for Integrative Biology (CBI), Institut de Génétique et de
Biologie Moléculaire et Cellulaire (IGBMC), Centre National de la Recherche Scientifique
(CNRS), UMR 7104, Institut National de la Santé et de la Recherche Médicale
(INSERM), U1258, Université de Strasbourg, Illkirch, France
MAXIMILIEN WERDERER • Institut de Génétique et de Biologie Moléculaire et Cellulaire
(IGBMC), Integrated Structural Biology, Equipe labellisée Ligue Contre le Cancer, Centre
National de la Recherche Scientifique (CNRS), UMR 7104, Institut National de la Santé
et de la Recherche Médicale (INSERM), U1258, Université de Strasbourg, Illkirch, France
GALA RAMON ZAMORANO • Institut de Génétique et de Biologie Moléculaire et Cellulaire
(IGBMC), Département of Integrated Structural Biology, Equipe Labellisée Ligue, Centre
National de la Recherche Scientifique (CNRS), UMR 7104, Institut National de la Santé
et de la Recherche Médicale (INSERM), U1258, Université de Strasbourg, Illkirch, France;
Department of Life and Medical Sciences, University of Hertfordshire, Hatfield, UK
 Part I

Recombinant Expression and Isolation from Endogenous

Source
 Chapter 1

Production of Multi-subunit Membrane Protein Complexes

Burak V. Kabasakal, Qiyang Jiang, and Christiane Schaffitzel

Abstract
Membrane proteins constitute an important class of proteins for medical, pharmaceutical, and biotechno-
logical reasons. Understanding the structure and function of membrane proteins and their complexes is of
key importance, but the progress in this area is slow because of the difficulties to produce them in sufficient
quality and quantity. Overexpression of membrane proteins is often restricted by the limited capability
of translocation systems to integrate proteins into the membrane and to fold them properly. Purification of
membrane proteins requires their isolation from the membrane, which is a further challenge. The choice of
expression system, detergents, and purification tags is therefore an important decision. Here, we present a
protocol for expression in bacteria and isolation of a seven-subunit membrane protein complex, the
bacterial holo-translocon, which can serve as a starting point for the production of other membrane protein
complexes for structural and functional studies.

Key words Membrane protein complex, Expression in bacteria, Endogenous host, Complex purifica-
tion, Translocon

1 Introduction

1.1 Aim of the Study Membrane proteins represent more than 25% of the proteome of all
cells. They mediate the cell’s interaction with its environment, i.e.,
transmission of signals, cell adhesion, and transport across mem-
branes. Membrane proteins are of prime pharmaceutical interest as
they constitute ~50% of known potential drug targets. In addition
to this, membrane proteins are the natural entry and/or anchoring
points for infectious agents. Dysfunctional membrane proteins are
the basis of many disorders such as cystic fibrosis and Alzheimer’s.
Understanding the function and molecular structure of this class of
proteins is of key interest.
As in the cytosol, the majority of proteins in the membrane
occur in and function in complexes. For instance in budding yeast,
membrane proteins were found to interact on average with two to
three interaction partners [1]. Due to their association with cellular
membranes, expression and purification of membrane proteins and
their complexes are often a challenging task. Progress is often

Arnaud Poterszman (ed.), Multiprotein Complexes: Methods and Protocols, Methods in Molecular Biology, vol. 2247,
https://doi.org/10.1007/978-1-0716-1126-5_1, © Springer Science+Business Media, LLC, part of Springer Nature 2021

3
4 Burak V. Kabasakal et al.

slowed down by the requirement to optimize protein expression

and to subsequently identify conditions to solubilize the membrane
proteins without compromising integrity and activity. While the
optimal expression and purification conditions needs to be opti-
mized for each new protein and complex, standard protocols as
presented here offer a good starting point for subsequent optimi-
zation by variation of conditions.

1.2 Recombinant E. coli has traditionally been, and still is, the most popular expres-
Expression sion system for proteins. This is because of low costs, low expendi-
in Escherichia coli ture of time, and the fact that it is less labor-intense than other
expression systems.

1.2.1 Choice of Promoter When expressing membrane proteins, saturation of the E. coli
and Strain membrane protein translocation and folding machinery should be
avoided. This would lead to protein misfolding, aggregation, and
formation of inclusion bodies. Very strong promoters such as the
T7 promoter frequently lead to these problems. Weaker promoters
synthesize membrane proteins at more moderate levels. Examples
include T7/lac hybrid promoters or the tightly regulated arabinose
promoter, both of which are often used for membrane protein
expression [2].
Addressing the same issues, tailor-made strains have been
developed for membrane protein production in E. coli. The T7
RNA polymerase-based expression strain BL21λ(DE3) and its
mutant strains C41λ(DE3) and C43λ(DE3) [3] are the most pop-
ular strains for expressing membrane proteins in E. coli [2]. C41
and C43 contain mutations in the lacUV5 promoter region of the
gene encoding for T7 RNA polymerase leading to lower expression
of the T7 RNA polymerase [4] and consequently slower transcrip-
tion and translation rates of the proteins under the control of T7
promoter. The same principle applies to the E. coli BL21-AI strain
(Invitrogen) which expresses T7 RNA polymerase under the con-
trol of a tightly regulated arabinose promoter.

1.2.2 Other Auto-induction media are commonly used for improved membrane
Considerations protein production [5]. They can be used for all IPTG-inducible
expression systems. Protein production is induced in auto-
induction media at high cell density when the glucose in the
media is depleted, which is during the mid/late log phase of
growth. Lactose uptake then leads to the production of allolactose
which causes the release of the lac repressor and induction of T7
RNA polymerase from the lacUV5 promoter.
For membrane protein complexes, a common strategy is the
co-expression of the proteins from their natural or artificial operons
[6] (see example in Fig. 1a). Co-expression of all subunits of a
complex allows complex assembly in vivo in the cell. This is partic-
ularly important for membrane protein complexes, where
Fig. 1 The production of the holo-translocon complex, comprising seven membrane proteins. (a) The
expression vector for the holo-translocon was generated using the ACEMBL system [13]. It comprises two
artificial operons encoding for YidC, SecD, and SecF (operon 1) and for SecY, SecE, and SecG (operon 2), as
well as the gene encoding for YajC with a C-terminal CBP-tag. The hexa-histidine tags on SecD, YidC, and
SecE are indicated as red triangles. The position of the CBP-tag is indicated with a purple triangle. Origins of
replication of donor (oriR6Kγ) and acceptor (oriBR322) vectors are indicated in black. Arabinose (ara) and trc
promoters are shown as grey arrows and terminators as black rectangles. Antibiotic resistance genes are
colored blue (Ap ampicillin, Kn kanamycin, Cm chloramphenicol). (b) Scheme showing the general workflow
for holo-translocon production and parameters that require optimization when a new membrane protein
complex is produced. (c) Coomassie-stained SDS PAGE gel of holo-translocon, analyzing the membrane
fraction, the detergent-solubilized protein fraction (Ni-NTA input), the flow-through from Ni-NTA, wash
fractions, and the eluate peak fraction (HS high salt). The bands for holo-translocon subunits are marked.
(d) Size exclusion chromatogram of holo-translocon and SDS PAGE of the peak fractions. The blue and red
curves show the absorption at 280 nm and 260 nm, respectively. Upper inset: Coomassie-stained SDS PAGE
showing the peak fractions; below: Western blot using an antibody directed against the CBP-tag to detect
YajC-CBP
6 Burak V. Kabasakal et al.

reconstitution of the complexes from the purified, detergent-

solubilized subunits is often not possible. Co-expression of the
subunits of a complex from one polycistronic messenger RNA
supports a balanced expression. Unexpectedly, E. coli operons
were shown to be translated with the proper stoichiometry even
when the subunit stoichiometry significantly deviates from one
copy each per complex [7]. Accordingly, the order of genes in a
synthetic operon needs to be experimentally optimized if the pro-
tein production is not balanced [8].

1.2.3 Recombinant When expressing eukaryotic proteins in E. coli, codon optimization

Membrane Protein of the corresponding gene is recommended in order to account for
Expression the E. coli codon usage bias and thus avoid low or unreliable
expression. Overexpression of recombinant membrane proteins
can be optimized by fusing a green fluorescent protein (GFP) to
the membrane protein. Using fluorescence detection, GFP serves as
a marker for membrane protein localization, quantity of folded
protein, and the efficiency of solubilization from the membrane
by detergents [9, 10].
In this chapter, we describe the production of the E. coli holo-
translocon, a seven-subunit membrane protein complex comprising
SecY, SecE, SecG, SecD, SecF, YajC, and YidC. It is one of the
largest recombinant membrane protein complexes produced to
date. The holo-translocon complex comprises all proteins known
to be important for protein translocation into and across the E. coli
plasma membrane [11, 12]. For the balanced expression of the
holo-translocon, the ACEMBL system was used [13, 14]. ACEMBL
comprises acceptor and donor vectors which have multi-integration
elements for individual genes or polycistrons. Donor vectors differ
from acceptor vectors and other commonly used plasmids in that
they comprise a conditional origin of replication (oriR6Kγ) and
therefore need to be propagated in strains with the phage R6Kγ
pir gene. All vectors have a LoxP site, but different antibiotic
resistances. Donor and acceptor vectors can be fused by Cre recom-
binase and the donor–acceptor fusion plasmid can be selected in a
conventional pir¯ E. coli strain using the appropriate combination
of antibiotics [13].
For the expression of the holo-translocon, the conserved SecY,
SecE, and SecG subunits were encoded by a synthetic polycistronic
operon on a donor vector (Fig. 1a). A second donor vector carried
the gene for YajC, which is the only nonessential subunit of the
holo-translocon [15]. These two donor vectors were fused to an
acceptor vector with a polycistron encoding for YidC, SecD, and
SecF. Hexa-histidine tags were fused to YidC, SecD, and SecG and
a calmodulin-binding peptide to YajC (Fig. 1a). Double-affinity
purification via these tags results in pure holo-translocon complex,
comprising one copy of each subunit [15, 16].
 Production of Multi-subunit Membrane Protein Complexes 7

1.3 Recombinant Expression of recombinant eukaryotic membrane protein in E. coli

Expression frequently results in protein misfolding and aggregation. This can
in Eukaryotic be due to differences in the lipid composition, a different chaperone
Expression Systems repertoire, or the lack of post-translational modifications required
to obtain folded, active membrane protein. In these cases, eukary-
otic expression systems that are more closely related to the natural
source of the protein have to be used. Unfortunately, membrane
protein expression in mammalian cells often results in moderate or
low protein yields. This is attributed to a limited capability of
mammalian transport systems to import additional proteins or to
replace its own proteins within the membranes of a particular
subcellular compartment. GFP fusion to the membrane protein of
interest is less useful for quality control and for optimizing expres-
sion in mammalian cells because in contrast to E. coli, the GFP will
fold independently from the membrane protein fused to it. Thus,
the GFP fusion method detects folded proteins as well as misfolded
proteins which are retained in the endoplasmic reticulum
membrane [17].
The majority of membrane proteins produced for structural
and functional analysis to date are from baculovirus insect cells
[18]. Baculovirus insect cell expression has been particularly suc-
cessful for production of G protein-coupled receptors (GPCRs),
leading to the first crystal structures of this class of membrane
proteins [19, 20]. The MultiBac system [21] has the same modular
architecture of donor/acceptor vectors as described for the
ACEMBL system for E. coli expression above. This supports the
production of membrane protein complexes in insect cells, such as
the active γ-secretase complex comprised of four membrane protein
subunits [22].

1.4 Overexpression Isolation of membrane protein complexes from the natural host
in the Endogenous used to be restricted to membrane protein machines which are
Host highly abundant in the cell, such as complexes for photosynthesis,
respiration, and oxidative phosphorylation [23]. However, with the
advent of miniaturization and automatization of biochemical, bio-
physical, and structural analyses, less abundant membrane protein
complexes can now be studied.
Purification of endogenous complexes is significantly facilitated
if a purification tag is fused to one of the subunits, which is possible
if the source can be genetically modified. The tagged subunit is
ideally an integral subunit of the complex. The chosen affinity
purification tag should be compatible with the presence of deter-
gents in the buffer (frequent choices include His-tags, 3xFLAG-
tag, and CBP-tag).
The E. coli holo-translocon production described below uses
the endogenous host, thus providing the required chaperone rep-
ertoire and the physiological lipid composition. The purified com-
plex has been shown to be active in co- and post-translational
8 Burak V. Kabasakal et al.

translocation and successfully used for biophysical and structural

analyses by electron cryo-microscopy and small-angle neutron scat-
tering [14–16]. A scheme for the general workflow of holo-
translocon production and parameters to be optimized is presented
in Fig. 1b.

2 Materials

Prepare all solutions using double-distilled water and analytical

grade reagents. Store all stock solutions at 4  C.

2.1 Expression 1. pACEMBL HTL3 plasmid [15] (Fig.1a).

of HTL 2. Chemically competent or electrocompetent E. coli BL21 Star
(DE3) cells (Thermo Fisher) or E. coli C43λ(DE3) cells [3] are
prepared according to published protocols [24, 25].
3. Prepare antibiotic stock solutions: 100 mg/mL ampicillin,
25 mg/mL chloramphenicol (dissolved in EtOH), and
50 mg/mL kanamycin.
4. Prepare LB medium: 10 g/L Bacto tryptone, 5 g/L yeast
extract, 10 g/L NaCl (verify pH 7.5, or adjust with NaOH),
and autoclave.
5. Prepare LB agar: Add 15 g/L Bacto agar to 1 L LB medium
and autoclave.
6. Incubator 37  C.
7. Shaker 37  C.
8. Spectrophotometer.
9. Baffled 5 L Erlenmeyer flasks.
10. Prepare Terrific Broth: 24 g/L yeast extract, 20 g/L Bacto
tryptone, 4 mL/L glycerol, 0.017 M KH2PO4, 0.072 M
K2HPO4, and autoclave.
11. Prepare stock solutions of 10% L-arabinose by dissolving 1 g L-
arabinose in 10 mL of water. Filter with 0.22 μm syringe filter
and store aliquots at 4  C. 1 M IPTG (isopropyl β-D-1-thioga-
lactopyranoside) is prepared by dissolving 2.38 g IPTG in
10 mL of water. Filter with a 0.22 μm syringe filter.
12. Centrifuge, e.g., Sorvall Lynx 6000 with F9 6x1000 Lex rotor
(Thermo Fisher).

2.2 Membrane 1. French Press (Sim Aminco, Spectronic Instruments) or Cell

Preparation Disrupter/Microfluidizer (Microfluidics).
2. Centrifuge, e.g., Sorvall RC6 and SS-34 rotor (Sorvall).
3. Prepare 1 L HSGM buffer: 20 mM HEPES-KOH pH 8.0
(20 mL of 1 M stock solution), 130 mM NaCl (26 mL of
 Production of Multi-subunit Membrane Protein Complexes 9

5 M stock solution), 5 mM MgOAc2 (5 mL of 1 M stock

solution), 10% glycerol.
4. cOmplete protease inhibitor cocktail tablets, EDTA-free
(Roche).
5. Prepare 100 mM PMSF (phenylmethylsulfonyl fluoride) stock
solution: add 17.4 mg of PMSF per 1 mL of isopropanol and
freeze 20 μL aliquots and store at 20  C.
6. Ultracentrifuge, Ti70 rotor, and centrifugation tubes (Beck-
man Coulter).
7. Dounce homogenizer.

2.3 Detergent 1. Prepare 10% w/v DDM (n-dodecyl-β-D-maltoside, Anatrace)

Solubilization stock solution in water. Dissolve by gently agitating; do not
vortex.
2. Head-over-tail rotator or equivalent.
3. Ultracentrifuge, Ti70 rotor, and centrifugation tubes (Beck-
man Coulter).

2.4 NiNTA Affinity 1. Ni-NTA agarose resin (Qiagen).

Chromatography 2. Gravity flow column (e.g., MoBiTec, #10141).
3. Prepare 100 mL Ni-NTA buffer: 20 mM HEPES-KOH
pH 8.0 (2 mL of 1 M stock solution), 130 mM NaCl
(2.6 mL of 5 M stock solution), 5 mM MgOAc2 (0.5 mL of
1 M stock solution), 10% glycerol, 10 mM imidazole (1 mL of
1 M stock solution, adjusted to pH 8.0), 0.1% DDM (1 mL of
10% stock solution).
4. Prepare 50 mL high salt buffer: 20 mM HEPES-KOH pH 8.0
(1 mL of 1 M stock solution), 500 mM NaCl (5 mL of 5 M
stock solution), 5 mM MgOAc2 (250 μL of 1 M stock solu-
tion), 10% glycerol, 10 mM imidazole (0.5 mL of 1 M stock
solution, adjusted to pH 8.0), 0.1% DDM (0.5 mL of 10%
stock solution).
5. Prepare 100 mL Ni-NTA Wash buffer: 20 mM HEPES-KOH
pH 8.0 (2 mL of 1 M stock solution), 130 mM NaCl (2.6 mL
of 5 M stock solution), 5 mM MgOAc2 (0.5 mL of 1 M stock
solution), 10% glycerol, 50 mM imidazole (5 mL of 1 M stock
solution, adjusted to pH 8.0), 0.1% DDM (1 mL of 10% stock
solution).
6. Prepare 100 mL Ni-NTA elution buffer: 20 mM HEPES-
KOH pH 8.0 (2 mL of 1 M stock solution), 130 mM NaCl
(2.6 mL of 5 M stock solution), 5 mM MgOAc2 (0.5 mL of
1 M stock solution), 10% glycerol, 300 mM imidazole (30 mL
of 1 M stock solution, adjusted to pH 8.0), 0.1% DDM (1 mL
of 10% stock solution).
7. Nanodrop spectrophotometer (Thermo Scientific).
10 Burak V. Kabasakal et al.

2.5 Desalting Step 1. 5 mL HiTrap Desalting column prepacked with Sephadex

G-25 resin (GE Healthcare).
2. AKTA protein purification system or equivalent.
3. Prepare 100 mL CBP-binding buffer: 50 mM HEPES-KOH
pH 8.0 (5 mL of 1 M stock solution), 130 mM NaCl (2.6 mL
of 5 M stock solution), 10% glycerol, 2 mM CaCl2 (200 μL of
1 M stock solution), 0.03% DDM (0.3 mL of 10% stock
solution).

2.6 Calmodulin 1. Calmodulin affinity resin (Agilent Technology).

Affinity 2. Gravity flow column (e.g., MoBiTec, #10141).
Chromatography
3. Prepare 100 mL CBP-binding buffer: 50 mM HEPES-KOH
pH 8.0 (5 mL of 1 M stock solution), 130 mM NaCl (2.6 mL
of 5 M stock solution), 10% glycerol, 2 mM CaCl2 (200 μL of
1 M stock solution), 0.03% DDM (0.3 mL of 10% stock
solution).
4. Prepare 100 mL CBP washing buffer: 50 mM HEPES-KOH
pH 8.0 (5 mL of 1 M stock solution), 130 mM NaCl (2.6 mL
of 5 M stock solution), 10% glycerol, 0.2 mM CaCl2 (20 μL of
1 M stock solution), 0.03% DDM (0.3 mL of 10% stock
solution).
5. Prepare 25 mL CBP elution buffer: 50 mM HEPES-KOH
pH 8.0 (1.25 mL of 1 M stock solution), 400 mM NaCl
(2 mL of 5 M stock solution), 3% glycerol, 2 mM EGTA
(100 μL of 0.5 M stock solution), 0.03% DDM (0.3 mL of
10% stock solution).
6. NanoDrop spectrophotometer (Thermo Scientific).

2.7 Sample 1. Amicon Ultra centrifugal filter units, molecular weight cut-off
Concentration 100 kDa, Ultra-15 (Amicon).
2. Centrifuge, e.g., Heraeus Megafuge 16R.
3. Prepare 10 mL S6 buffer: 20 mM HEPES-KOH pH 8.0
(0.2 mL of 1 M stock solution), 130 mM NaCl (0.26 mL of
5 M stock solution), 5 mM MgOAc2 (50 μL of 1 M stock
solution), 0.03% DDM (30 μL of 10% stock solution).

2.8 Size Exclusion 1. Superose 6 Increase 10/300 GL column (GE Healthcare).

Chromatography 2. AKTA protein purification system or equivalent.
3. Prepare 250 mL S6 buffer: 20 mM HEPES-KOH pH 8.0
(5 mL of 1 M stock solution), 130 mM NaCl (6.5 mL of
5 M stock solution), 5 mM MgOAc2 (1.25 mL of 1 M stock
solution), 0.03% DDM (0.75 mL of 10% stock solution).
 Production of Multi-subunit Membrane Protein Complexes 11

3 Methods

Once the proteins are solubilized from the isolated membranes, it is

important to work at 4  C and proceed quickly with the protein
purification to avoid protein aggregation and degradation.

3.1 Expression 1. Transform the pACEMBL HTL3 plasmid (Fig. 1a) into chem-
of HTL ically competent or electrocompetent E. coli BL21 Star (DE3)
(SecY-SecE-SecG + or E. coli C43λ(DE3) cells [3] and select for growth on LB agar
YidC-SecD-SecF plates with ampicillin, chloramphenicol, and kanamycin at
+ YajC) 37  C overnight.
2. Inoculate 100 mL preculture in LB medium supplemented
with 50 μg/mL ampicillin, 20 μg/mL chloramphenicol, and
30 μg/mL kanamycin with one colony from the LB agar plate
and grow at 37  C overnight by shaking at 180 rpm.
3. Inoculate 1 L of Terrific Broth with 50 μg/mL ampicillin,
20 μg/mL chloramphenicol, and 30 μg/mL kanamycin in a
5 L baffled Erlenmeyer flask with approximately 20 mL pre-
culture to reach an initial optical density (OD600nm) of 0.05 at
600 nm. Shake the E. coli cultures at 37  C and 150 rpm.
Typical culture volumes for holo-translocon preparations are
6–12 L.
4. When the cell cultures reach an OD600 of 1.2–1.5 (mid-log
phase), induce protein expression by the addition of 21 mL of
10% L-arabinose (0.2% final concentration) and 0.5 mL of
1 M IPTG.
5. After 3 h of protein expression (37  C, 150 rpm), the cells are
harvested by centrifugation for 20 min at 5000  g (4  C) (see
Note 1).
6. Discard the supernatant and determine the weight of the cell
pellet (see Note 2).

3.2 Membrane 1. Thaw ~10 g E. coli HTL3 cell pellet on ice.

Preparation 2. Add 30 mL of HSGM buffer and one tablet EDTA-free prote-
ase inhibitor cocktail and 0.5 mM PMSF.
3. Resuspend the cells by pipetting.
4. Open the cells by two passages through a French press cell at
1000 psi (see Note 3).
5. Centrifuge the cell lysate at 16,000  g for 30 min (at 4  C).
6. Transfer the supernatant into an ultracentrifugation tube and
discard the pellet.
7. Centrifuge the cleared lysate for 2 h in a Ti70 rotor at
45,000 rpm (149,000  g) at 4  C.
12 Burak V. Kabasakal et al.

8. Discard the supernatant and resuspend the brown-yellow pellet

(membranes) in ~5 mL HSGM buffer using a Dounce homog-
enizer. Add cOmplete EDTA-free protease inhibitor cocktail
and 0.5 mM PMSF (see Note 4).
9. Analyze the membrane fraction using a 15% acrylamide gel to
confirm the presence of SecD (67.5 kDa), YidC (63 kDa), SecY
(50 kDa), SecF (35 kDa), SecE (15 kDa), SecG (14 kDa). The
presence of YajC-CBP (14.7 kDa) needs to be confirmed by
Western blotting using an anti-CBP antibody (see Fig. 1c, lane
1). If the Coomassie stained gel is difficult to interpret, the
presence of SecD, YidC, and SecE can be confirmed by Western
blotting using an anti-His antibody. SecD and SecE are
encoded by polycistrons and part of SecYEG and SecDF pro-
tein complexes. Therefore, the other complex subunits are
likely to be also present in the membrane.

3.3 Detergent 1. Add DDM to a final concentration of 1.5% w/v to the mem-
Solubilization branes (see Note 5).
2. Gently agitate the sample at 4  C (in the cold room) for 2 h by
head-over-tail rotation.
3. Centrifuge solubilized membranes for 1 h in a Ti70 rotor at
45,000 rpm (149,000  g) at 4  C (see Note 6).

3.4 NiNTA Affinity 1. Equilibrate 5 mL Ni-NTA resin with ten column volumes of
Chromatography (All Ni-NTA buffer (see Note 7 for using a prepacked column).
Steps at 4  C) 2. Add the Ni-NTA resin to the solubilized membrane superna-
tant (from step 3 of Subheading 3.3).
3. Gently agitate for 1 h by head-over-tail rotation.
4. Transfer the sample and beads into a gravity flow column.
Collect the flow-through.
5. Wash with at least 20 column volumes of Ni-NTA buffer (see
Note 8, to change the detergent to LMNG at this step).
6. Wash with three to five column volumes of high salt buffer.
7. Wash with at least five column volumes of Ni-NTA buffer.
8. Wash with 20 column volumes of Ni-NTA wash buffer.
9. Verify that there is no A280nm signal in the wash fraction using a
NanoDrop spectrophotometer before proceeding.
10. Elute with Ni-NTA elution buffer and collect the eluate in
1.5–2 mL tubes.
11. Determine the protein concentration of each fraction by
A280nm measurements using a NanoDrop spectrophotometer
(see Note 9).
12. Add 0.1 mM PMSF and cOmplete protease inhibitor cocktail
to the protein-containing fractions.
 Production of Multi-subunit Membrane Protein Complexes 13

13. Analyze the input, flow-through, wash, and eluate fractions by

SDS PAGE using a 15% acrylamide gel (see Fig. 1c for a typical
result).
14. Pool the protein-containing fractions.

3.5 Desalting Step 1. Equilibrate a 5 mL desalting column with CBP-binding buffer

(at 4  C) using the AKTA system (see Notes 10 and 11).
2. Inject 3 mL of Ni-NTA eluate (use combined protein-
containing fractions from step 14 of Subheading 3.4) onto
the desalting column (see Note 12).
3. Collect 1 mL fractions and pool protein-containing fractions.
Repeat this procedure until all holo-translocon-containing elu-
ate fractions from step 14 of Subheading 3.4 are in
CBP-binding buffer.

3.6 Calmodulin 1. Equilibrate 5 mL calmodulin affinity resin with CBP-binding

Affinity buffer.
Chromatography 2. Add the pooled fractions from step 3 of Subheading 3.5 to the
(at 4  C) resin.
3. Incubate overnight at 4  C and gently agitate.
4. Transfer the sample into a gravity flow column.
5. Wash the column with at least ten column volumes of CBP
washing buffer.
6. Verify by NanoDrop measurements that there is no A280nm
signal in the last wash fraction before proceeding to the
next step.
7. Elute with five column volumes of CBP elution buffer and
collect fractions.
8. Determine the concentration of each fraction by NanoDrop
A280nm measurements (see Note 9).
9. Analyze the input, flow-through, wash, and eluate fractions by
SDS PAGE using a 15% acrylamide gel.
10. Pool holo-translocon-containing fractions.

3.7 Sample 1. Equilibrate a 100 kDa molecular weight cut-off concentrator

Concentration with S6 Buffer.
2. Load the holo-translocon-containing fractions from step 10 of
Subheading 3.6 into the concentrator.
3. Centrifuge at 3500  g, 4  C until the sample volume is
reduced to 500 μL (monitor volume regularly).
14 Burak V. Kabasakal et al.

3.8 Size Exclusion 1. Equilibrate a Superose 6 column with water and then with two
Chromatography column volumes of S6 Buffer (see Note 13).
(Optional) 2. Load 500 μL of sample from step 3 of Subheading 3.7 onto the
column.
3. Collect 0.5 mL fractions.
4. Analyze the protein-containing fractions on a 15% SDS
PAGE gel.
Pool the peak fractions (Fig. 1d), if required concentrate to
~1 mg/mL using a concentrator with 100 kDa molecular weight
cut-off. Freeze aliquots in liquid nitrogen and store the purified
holo-translocon at 80  C.

4 Notes

1. For each membrane protein complex, the growth temperature

and the optimal time point to harvest the cells have to be
determined experimentally. It is recommended to also test
different strains (Fig. 1b).
2. It is possible to freeze the cells in liquid nitrogen and store the
cells at 80  C at this step.
3. Alternatively, two passages through a microfluidizer or cell
disrupter at 25 kPsi can be used to open the cells.
4. It is possible to freeze the dissolved membranes in liquid nitro-
gen and store at 80  C after this step.
5. For each new membrane protein complex, detergent screens
have to be tested to identify a suitable mild, non-denaturing,
solubilizing detergent [26], ideally followed by an activity assay
(Fig. 1b). Different screens have been reported [27, 28], vari-
ous commercial detergent screens are available (e.g., Qiagen,
Jena Bioscience).
6. If the centrifuge tube is not completely filled, it is recom-
mended to either fill the Ti70 centrifugation tubes to the top
with buffer or to reduce the centrifugation speed.
7. Alternatively, use a HisTrap FF column (GE Healthcare) cou-
pled to a peristaltic pump for loading and washing the sample.
Prior to loading, the supernatant from ultracentrifugation is
passed through a 0.22 μM filter to further remove any large
particles. During the washing step, the color of the column will
change from brown to almost light blue over the course of
about 1 h. During this time, detergent exchange (to LMNG for
example) or supplementation of any additive (e.g., cholesteryl
hemisuccinate) can be performed. For elution, the column is
connected to an AKTA chromatography system.
 Production of Multi-subunit Membrane Protein Complexes 15

8. It is possible to exchange the detergent to Lauryl Maltose

Neopentyl Glycol (LMNG, Anatrace) at this step. To do this,
replace DDM in the Ni-NTA Wash buffer with 0.1% w/v
LMNG and in all other buffers with 0.03% w/v LMNG. In
the final size exclusion chromatography step, the concentration
of LMNG in the S6 buffer can be further lowered to 0.002%
w/v LMNG. The critical micelle concentration of LMNG is
0.001% w/v in water.
9. The molecular weight of holo-translocon is 240 kDa and the
extinction coefficient is ~500,000 M 1 cm 1 at 280 nm.
10. Equilibrate the desalting column during the Ni-NTA wash
steps.
11. The maximum pressure of the desalting column is 0.3 MPa.
12. Very important: Quickly proceed from step 14 of Subheading
3.4 to the desalting step to avoid aggregation of the holo-
translocon sample in the Ni-NTA elution buffer due to high
imidazole concentration.
13. Superose 6 maximum pressure is 1.5 MPa; recommended flow
rate is 0.4 mL/min.

Acknowledgments

This work is supported through a BBSRC Responsive Mode Grant

(BB/P000940/1) and a Wellcome Trust Investigator Grant
(210701/Z/18/Z). We would like to thank Kyle Powers, Flint
Stevenson-Jones, and Josh Bufton for critically reading the
manuscript.

References

1. Babu M, Vlasblom J, Pu S et al (2012) Interac- 5. Studier FW (2005) Protein production by

tion landscape of membrane-protein com- auto-induction in high density shaking cul-
plexes in Saccharomyces cerevisiae. Nature tures. Protein Expr Purif 41:207–234
489:585–589 6. Zorman S, Botte M, Jiang Q et al (2015)
2. Hattab G, Warschawski DE, Moncoq K et al Advances and challenges of membrane-protein
(2015) Escherichia coli as host for membrane complex production. Curr Opin Struct Biol
protein structure determination: a global anal- 32:123–130
ysis. Sci Rep 5:12097 7. Li GW, Burkhardt D, Gross C et al (2014)
3. Miroux B, Walker JE (1996) Over-production Quantifying absolute protein synthesis rates
of proteins in Escherichia coli: mutant hosts that reveals principles underlying allocation of cel-
allow synthesis of some membrane proteins lular resources. Cell 157:624–635
and globular proteins at high levels. J Mol 8. Smolke CD, Keasling JD (2002) Effect of gene
Biol 260:289–298 location, mRNA secondary structures, and
4. Wagner S, Klepsch MM, Schlegel S et al (2008) RNase sites on expression of two genes in an
Tuning Escherichia coli for membrane protein engineered operon. Biotechnol Bioeng
overexpression. Proc Natl Acad Sci U S A 80:762–776
105:14371–14376
16 Burak V. Kabasakal et al.

9. Drew D, Lerch M, Kunji E et al (2006) Opti- expression in insect and mammalian cells. Nat
mization of membrane protein overexpression Biotechnol 23:567–575
and purification using GFP fusions. Nat Meth- 19. Akermoun M, Koglin M, Zvalova-Iooss D et al
ods 3:303–313 (2005) Characterization of 16 human G
10. Kawate T, Gouaux E (2006) Fluorescence- protein-coupled receptors expressed in
detection size-exclusion chromatography for baculovirus-infected insect cells. Protein Expr
precrystallization screening of integral mem- Purif 44:65–74
brane proteins. Structure 14:673–681 20. Rasmussen SG, Choi HJ, Rosenbaum DM et al
11. Duong F, Wickner W (1997) Distinct catalytic (2007) Crystal structure of the human beta2
roles of the SecYE, SecG and SecDFyajC sub- adrenergic G-protein-coupled receptor. Nature
units of preprotein translocase holoenzyme. 450:383–387
EMBO J 16:2756–2768 21. Fitzgerald DJ, Berger P, Schaffitzel C et al
12. Scotti PA, Urbanus ML, Brunner J et al (2000) (2006) Protein complex expression by using
YidC, the Escherichia coli homologue of mito- multigene baculoviral vectors. Nat Methods
chondrial Oxa1p, is a component of the Sec 3:1021–1032
translocase. EMBO J 19:542–549 22. Khan I, Krishnaswamy S, Sabale M et al (2018)
13. Bieniossek C, Nie Y, Frey D et al (2009) Auto- Efficient production of a mature and functional
mated unrestricted multigene recombineering gamma secretase protease. Sci Rep 8:12834
for multiprotein complex production. Nat 23. Mesa P, Deniaud A, Montoya G et al (2013)
Methods 6:447–450 Directly from the source: endogenous prepara-
14. Komar J, Botte M, Collinson I et al (2015) tions of molecular machines. Curr Opin Struct
ACEMBLing a multiprotein transmembrane Biol 23:319–325
complex: the functional SecYEG-SecDF-YajC- 24. Inoue H, Nojima H, Okayama H (1990) High
YidC Holotranslocon protein secretase/inser- efficiency transformation of Escherichia coli
tase. Methods Enzymol 556:23–49 with plasmids. Gene 96:23–28
15. Schulze RJ, Komar J, Botte M et al (2014) 25. Dower WJ, Miller JF, Ragsdale CW (1988)
Membrane protein insertion and proton- High efficiency transformation of E. coli by
motive-force-dependent secretion through the high voltage electroporation. Nucleic Acids
bacterial holo-translocon SecYEG-SecDF- Res 16:6127–6145
YajC-YidC. Proc Natl Acad Sci U S A 26. Linke D (2014) Explanatory chapter: choosing
111:4844–4849 the right detergent. Methods Enzymol
16. Botte M, Zaccai NR, Lycklama JA et al (2016) 541:141–148
A central cavity within the holo-translocon sug- 27. Kubicek J, Block H, Maertens B et al (2014)
gests a mechanism for membrane protein inser- Expression and purification of membrane pro-
tion. Sci Rep 6:38399 teins. Methods Enzymol 541:117–140
17. Thomas JA, Tate CG (2014) Quality control in 28. Lantez V, Nikolaidis I, Rechenmann M et al
eukaryotic membrane protein overproduction. (2015) Rapid automated detergent screening
J Mol Biol 426:4139–4154 for the solubilization and purification of mem-
18. Kost TA, Condreay JP, Jarvis DL (2005) Bacu- brane proteins and complexes. Eng Life Sci
lovirus as versatile vectors for protein 15:39–50
 Chapter 2

Production of Multiprotein Complexes Using

the Baculovirus Expression System: Homology-Based
and Restriction-Free Cloning Strategies for Construct
Design
Paola Rossolillo, Olga Kolesnikova, Karim Essabri,
Gala Ramon Zamorano, and Arnaud Poterszman

Abstract
Most cellular processes are mediated by multi-subunit protein complexes which have attracted major
interest in both academia and industry. Recombinant production of such entities in quantity and quality
sufficient for functional and structural investigations may be extremely challenging and necessitate specific
technologies. The baculovirus expression vector system is widely used for the production of eukaryotic
multiprotein complexes, and a variety of strategies are available to assemble transfer vectors for the
generation of recombinant baculoviruses. Here we detail applications of homology-based cloning techni-
ques for one-step construction of dual promoter baculovirus transfer plasmids and of restriction-free
(RF) cloning for the modification of existing constructs.

Key words Baculovirus insect cell expression, Multiprotein complex, Homology-based cloning,
Sequence and ligation–independent cloning (SLIC), restriction-free (RF) cloning

1 Introduction

Most processes in living systems are mediated by multi-subunit

protein complexes which have attracted major interest in both
academia and industry. Recombinant production of such entities
in quantity and quality sufficient for functional and structural
investigations requires specific technologies. In favorable cases,
components of a complex can be expressed separately and assem-
bled in vitro. This approach is unfortunately not always applicable
as isolated components of a complex are often poorly soluble or
misfolded in the absence of interacting partners. Complex samples,
including self-assembling multi-protein complexes, proteins requir-
ing specific interacting partners, or post-translational modifications

Arnaud Poterszman (ed.), Multiprotein Complexes: Methods and Protocols, Methods in Molecular Biology, vol. 2247,
https://doi.org/10.1007/978-1-0716-1126-5_2, © Springer Science+Business Media, LLC, part of Springer Nature 2021

17
18 Paola Rossolillo et al.

frequently require the simultaneous expression of several proteins

foreign to the host cell.
With the baculovirus insect cell expression system, protein
targets can be expressed from multiple viruses carrying a single
foreign gene each, from a single baculovirus carrying multiple
foreign genes, or from a combination of the two approaches. For
routine production of complex systems, it is now widely accepted
that the use of multigene expression constructs is extremely effi-
cient. Progress in our understanding of the baculovirus biology
together with the development of new tools for DNA manipulation
have allowed to streamline protocols for the generation of viruses,
expression screening, and large-scale productions (see refs. 1, 2 for
recent reviews). A major breakthrough was accomplished through
the development of the Multibac technology, which directs
co-expression of multiple genes from a single virus under the
control of multiple copies of the late p10 promoter and the very
late polyhedron (PH) promoter [3, 4].
Preparation of multigene expression constructs is a multistep
process with a first round consisting in the assembly of single or
dual expression cassettes. In the absence of prior knowledge about
the proteins of interest, experiments with viruses containing the
single or dual expression cassettes will provide valuable information
on expression level/solubility of individual subunits; moreover,
different affinity tags can be tested. At this stage, coinfection experi-
ments deliver first insights into the protein–protein interaction
network and the architecture of the complex.
In the second step, expression cassettes are combined to gener-
ate multigene transfer vectors and produce the corresponding bacu-
loviruses (Fig 1a). This can be achieved by using a combination of
restriction-ligation and in vitro Cre-mediated plasmid fusion [5–7]
as well as with restriction-free technologies such as In-Fusion [8] or
USER (Uracil-Specific Excision Reagent) cloning [9] (Fig. 1b).
We have previously described optimized protocols to generate,
evaluate, and amplify recombinant viruses, as well as to perform
expression screening and large-scale production of difficult-to-
express proteins and complexes [10, 11]. In this chapter, we focus
on the assembly of multigene expression constructs and detail
applications of homology-based cloning techniques (sequence
and ligation–independent cloning (SLIC) and similar commercially
available technologies such as In-Fusion, Gibson, or NEBuilder) to
assemble pairs of promoters to create dual expression plasmids. We
also provide a simple protocol for the modification of existing
constructs based on restriction-free (RF) cloning approach. These
protocols were implemented in our laboratory to generate con-
structs for the reconstitution and characterization of human multi-
protein complexes that include the CAK and pTefb kinase
complexes as well as the ten subunit transcription/DNA repair
factor TFIIH (see for example [12–14]).
 Production of Multiprotein Complexes Using the Baculovirus Expression. . . 19

Fig. 1 Construction of multigene transfer vectors for the expression of multiprotein complexes. (a) DNA
elements comprising genes of interest, promoters, and terminators are combined with plasmid backbone
modules into multigene expression transfer vectors (left). These plasmids are transferred into the viral genome
through Tn7 transposition in bacterial BH10Bac cells or by homologous recombination after co-transfection in
insect cells (right). Recombinant viruses produce proteins of interest (here under the control of the PH
promoter) together with a reporter gene (here mCherry under the P10 promoter). (b) To assemble multigene
constructs, individual or dual expression cassettes can be excised by digestion with a pair of restriction
endonucleases or amplified by PCR and inserted via compatible restriction sites or homology-based cloning
into the multiplication module of a progenitor plasmid. The top panel shows an example in which an
expression cassette containing genes C and D is cloned into a plasmid containing genes A and
B. Alternatively, multigene constructs can be assembled using Cre-mediated fusion via LoxP sites (green
circles). Acceptors have a regular replication origin (ori ColE1), whereas donors have a conditional origin
derived from phage (ori R6Kγ); the two plasmids also have different antibiotic resistance. The bottom panel
illustrates the Cre-mediated fusion of an acceptor plasmid containing a dual expression cassette bearing
genes A and B with a donor vector bearing gene E
20 Paola Rossolillo et al.

2 Materials

2.1 Working You need access to standard equipment for molecular biology
Environment including PCR machines, agarose gel electrophoresis apparatus,
and Instruments UV trans-illuminator, refrigerated benchtop centrifuge, nanodrop
for DNA quantification, fridge, and freezers (
20  C and
80  C).
Prepare all solutions using ultrapure water (prepared by purify-
ing deionized water, to attain a sensitivity of 18 MΩ-cm at 25  C)
and analytical grade reagents. Unless indicated otherwise, prepare
and store all reagents at room temperature. Diligently follow all
waste disposal regulations when disposing waste materials.

2.2 Bacterial Strains 1. Luria–Bertani (LB) medium with antibiotics added at the fol-
and Plasmids lowing concentrations: ampicillin (100 μg/mL), kanamycin
(30 μg/mL), spectinomycin (90 μg/mL), and chlorampheni-
col (15 μg/mL).
2. E. coli DH5α, TOP10, XL-1 blue, or Pir1 (see Note 1) compe-
tent cells (chemically or electrocompetent cells).
3. Dual expression transfer vectors (see Table 1). These plasmids
are available from Addgene.
4. Templates for amplification of the target genes. When a starting
plasmid containing the cDNA of your gene of interest is not
present, chemical synthesis is often an advantageous option for
generating cDNAs. In any case, the removal of restriction sites
that might be used for the transfer of the expression cassettes is

Table 1
Dual promoter plasmids for baculovirus expression systems

Plasmid Method Resistance Size Origin References

pFastBac Dual Tn7 transposition Ampicillin, 5.2 kbp pUC ori None
gentamicin
pFL Tn7 transposition Ampicillin, 5.8 kbp ColE1 [6]
gentamicin
pKL Tn7 transposition Kanamycin, 4.9 kbp pBR322 [5]
gentamicin
pAC8_MF Homology Ampicillin 6.9 kbp pUC ori Kolesnikova
recombination et al. (in prep)
pUCDMa Cre-mediated fusion Chloramphenicol 3.0 kbp R6Kγ [5]
pSPLa Cre-mediated fusion Spectinomycin 2.9 kbp R6Kγ [6]
These plasmids contain the expression cassette detailed in Fig. 2a and with the exception of pFastBac Dual all possess a
LoxP site. Vectors pFastBac Dual, pFL, and pKL use Tn7 transposition while pAC8_MF relies on homology recombina-
tion to integrate the expression cassette in the viral DNA
a
Vectors pUCDM and pSPL cannot be used directly to generate recombinant viruses but can be fused to pFL, pKL, and
pAC8_MF through using Cre-mediated reactions
 Production of Multiprotein Complexes Using the Baculovirus Expression. . . 21

Fig. 2 Dual expression plasmids. (a) Expression cassette from the transfer vector pFastBac Dual, where
promoters (p10 and PH), cloning sites to insert the genes of interest (A and B) and terminators (HSV Tk and
SV40 polyA) are indicated. (b) Representation of the transfer vectors pFastBac Dual, pKL, and pAC8_MF. The
expression cassettes of pKL and pAC8_MF are flanked by a pair of unique restriction sites (AvrII and PmeI) and
contain a multiplication module (M) composed of the BstZ17I, SpeI, or NruI recognition sequences. These can
be used to assemble constructs for the co-expression of three and more cDNAs. Plasmids pKL and pAC8_MF
also contain a LoxP site (blue circle). Elements for the transfer into the viral genome (Tn7-R and Tn7-L or
AcMNPV sequences), a replication unit (ColE1) and antibiotic resistance genes (AmpR and KmR) are also
indicated

strongly suggested. Try in particular to eliminate PmeI, AvrII,

BstZ17I, NruI, and SpeI recognition sequences from your
starting plasmids or cDNAs as these sites are essentials for the
creation of multigene vectors (Fig. 2).

2.3 Reagents 1. Plasmid containing the target cDNAs and dual expression
for the Preparation transfer vectors (see Subheading 2.2).
of the Linearized 2. Synthetic oligonucleotides for PCR amplification designed as
Vector and DNA detailed in Subheading 3.2 (see Table 2). Dry oligonucleotides
Fragments are usually resuspended at 100 μM and further diluted at
10 μM.
3. Phusion Polymerase (ThermoScientific™) with 5xHF or GC
buffer and DMSO or another high-fidelity polymerase.
4. dNTPs mix. Prepare and aliquot 10 mM dNTP stock solutions
in ultrapure water.
5. DpnI (10 u/μL) restriction enzyme and corresponding buffer.
6. If the plasmid backbone is prepared by digestion: BamHI,
XbaI, XhoI, NheI (10 u/μL) restriction enzymes, and the
corresponding buffers.
 22

Table 2
Oligonucleotides for the amplification of the dual promoter region, the plasmid backbone, and the two target cDNAs and their cloning

Region to amplify Primer name Primer sequence

Native constructs (pFastBac Dual, pKL. or pAC8_MF)
Dual promoter module Prom-F (PH) GGTGGATCCGCGCCCGATGG
Paola Rossolillo et al.

Prom-R (p10) GGTGGCTCGAGATCCCGGGTG

Plasmid backbone Backbone-R (PH) TCTAGAGCCTGCAGTCTCG
Backbone-F (p10) GCTAGCAGCTGATGCATAG
Gene under pH promoter pH-Gene-START-F CATCGGGCGCGGATCCACC/ATG. . .first 15–21 nt of the gene (BamHI+Kozak)
pH-Gene-STOP-R CGAGACTGCAGGCTCTAGA/STOP. . .RC(last 15–21 nt of the gene) (XbaI)
Gene under p10 promoter p10-Gene-START-F CACCCGGGATCTCGAGCCACC/ATG. . .first 15–21 nt of the gene (XhoI+Kozak)
p10-Gene-STOP-R CTATGCATCAGCTGCTAGC/STOP. . .RC(last 15–21 nt of the gene) (NheI)
C-terminal histidine-tagged construct (pKL-PH—3C-10His-Cter)
Plasmid backbone Backbone-3C-R (PH) TCTAGACTGGAAGTTCTGTTTC
Gene under pH pH-Gene-10His-3C R GAAACAGAACTTCCAGTCTAGA. . .RC(last 15–21 nt of the gene) (XbaI)
The NheI, XhoI, BamHI, and XbaI restriction sites are underlined. Note that oligonucleotides Prom-F and Prom-R contain a kozak consensus sequence (CCACC in bold).
pKL-PH-3C-10His-Cter is a pKL derivative designed to co-express a cleavable C-terminally histidine-tagged protein under the control of the PH promoter with an untagged
protein under the control of p10. For this vector, the gene under PH promoter is amplified using the pH-Gene-START-F/pH-Gene-10His-3C-R primer pair and the plasmid
backbone with the Backbone-3C-R (PH)/Backbone-F (p10) pair. The gene under p10 promoter is amplified with the p10-Gene-START-F/p10-Gene-STOP-R pair. RC stands for
reverse and complement (see Figs. 4 and 5)
 Production of Multiprotein Complexes Using the Baculovirus Expression. . . 23

2.4 Reagents 1. T4 DNA Polymerase (New England Biolabs).

for a SLIC Reaction 2. 10 NEB 2.1 buffer (Tris-HCl 100 mM, MgCl2 100 mM,
NaCl 500 mM, BSA 1 mg/mL, pH 7.9).
3. 10 mM dCTP.
4. 10 Ligation buffer (e.g., Tris-HCl 500 mM, MgCl2 100 mM,
DDT 100 mM, pH 7.5) or any DNA ligase buffer for
annealing.
5. 1 μg of linearized vector and of each amplified DNA fragment
to be assembled prepared as described in Subheading 3.3.

2.5 Reagents for an 1. In-Fusion or Gibson assembly kit.

In-Fusion Reaction 2. 1 μg of linearized vector and of each amplified DNA fragment
to be assembled prepared as described in Subheading 3.3.

2.6 Solution 1. Phusion Polymerase (ThermoScientific™) with 5xHF or GC

and Media buffer and DMSO or another high-fidelity polymerase.
for Restriction-Free 2. dNTPs mix. Prepare and aliquot 10 mM dNTP stock solutions
Cloning in ultrapure water.
DpnI (10 u/μL) restriction enzyme and corresponding
buffer.

3 Methods

Here we detail the preparation of dual expression plasmids using

homology-based cloning and the modification of existing con-
structs. We focus on an expression cassette composed of two diver-
gent baculovirus late promoters (PH and p10), each followed by a
multiple cloning site (MSC1 and MSC2) and a transcription termi-
nation signal (SV40 pA and HSV-TK pA) (Fig. 2a). This cassette
organization is found in the transfer vector FastBac Dual and in its
Multibac derivatives which include the acceptor vectors pFL and
pKL and the donor plasmids pUCDM and pSPL (Table 1). It was
also inserted into the backbone of the transfer vector pBacPack8
(plasmid pAC8_MF) to enable the generation of recombinant
baculoviruses by homologous recombination directly in insect
cells (Fig. 2b). In a typical experiment, the plasmid backbone is
combined with the promoter module (comprising the PH and p10
promoters) and cDNAs encoding the two target genes, in a four-
fragment assembly reaction (Fig. 3). The first cDNA is typically
inserted between the BamHI and XbaI sites of MCS1 and the
second between the XhoI and NheI sites of MSC2 (Fig. 2a).
Following a general outline of homology-based technologies
for fragment assembly in Subheading 3.1, we detail primer selec-
tion in Subheading 3.2 and provide protocols for the preparation
24 Paola Rossolillo et al.

Fig. 3 Homology-based assembly of dual expression vectors. The plasmid backbone, the promoter module
(comprising the PH and p10 promoters), and the cDNAs encoding the two target genes are amplified by PCR
using primers designed to generate end-terminal homology and combined in a single-step four-fragment
assembly reaction

and assembly of the linearized vector and DNA fragments in Sub-

headings 3.3–3.5. In Subheading 3.6, the use of restriction-free
(RF) cloning for modification of existing constructs is illustrated.

3.1 Principles Several seamless cloning technologies are now available to assemble
of Homology-Based multiple DNA fragments into a destination plasmid without intro-
Technologies ducing undesired nucleotides or scar sequences between vector and
for Fragment insert. Homology-based techniques—Gibson Assembly [15],
Assembly In-Fusion Cloning (Clontech™, [16]), and SLIC [17]—rely on
PCR to add homologous sequence overhangs to the DNA frag-
ments to be assembled adjacently and on an exonuclease activity to
expose single-stranded overhangs that will anneal with complemen-
tary overlapping fragments.
The Gibson assembly technique [15] relies on an enzyme mix
composed of the T5 exonuclease, which chews back the 50 ends of
DNA fragments, generating long overhangs and allowing the
 Production of Multiprotein Complexes Using the Baculovirus Expression. . . 25

single-stranded regions with homology to anneal; a DNA polymer-

ase, to fill in the gaps; and a DNA ligase, to seal the nicks between
the annealed and filled-in sequences.
SLIC (sequence and ligation–independent cloning) [17] and
In-Fusion Cloning (Clontech™, [16]) are based on the ligation-
free cloning of overlapping PCR products treated by a DNA poly-
merase (the T4 and the vaccinia virus DNA polymerases, respec-
tively), which possesses a 30 to 50 proofreading exonuclease activity.
The polymerase chews back one strand of each fragment, creating
single-stranded compatible ends. These overhangs are annealed
in vitro, and the recombinant circular construct is rescued in E. coli.

3.2 Primer Selection The assembly of a dual expression transfer plasmid involves the
for Homology-Based combination of four DNA fragments with overlapping ends
Cloning which are in general obtained by PCR amplification: the vector
backbone, the dual promoter module, and the two cDNAs
(Fig. 3). Four pairs of oligonucleotides are required.
(a) The region of the plasmid including the divergent promoters
spanning from XhoI to BamHI sites is amplified with primers
named prom-F and prom-R, both including the restriction site
in their sequence and containing a Kozak consensus in their
50 end (see Table 2 for the primer sequences).
(b) The two genes are amplified by a PCR performed with the
primer pairs pH-gene-START-F/pH-gene-STOP-R and p10-
gene-START-F/p10-gene-STOP-R. They are composed of a
50 part corresponding to the cloning junction followed by 30
part complementary to the insert target (see step 4 below).
(c) Finally, the vector backbone, from XbaI to NheI sites, is
amplified with primers Backbone-F and Backbone-R. Note
that, alternatively to PCR amplification, the linearized vector
can be obtained by XbaI/NheI restriction digest followed by
gel purification (see Subheading 3.3).
Construct design and selection of primers for PCR amplifica-
tion are detailed using an example where a dual expression cassette
for the co-expression of eGFP and mCherry is assembled (Fig. 4
and Table 2). We strongly advise using a DNA cloning software
(i.e., Serial cloner, ApE, SnapGene) to define the desired con-
structs, select a set of specific primers, and carefully simulate the
cloning procedure in silico before starting any experiment.
1. Identify plasmids/cDNAs from which the sequences encoding
the proteins of interest will be amplified and carefully verify
their nucleotide sequence. Do not hesitate to re-sequence if
cDNAs were obtained from external sources.
26 Paola Rossolillo et al.

Fig. 4 Plasmid and primer sequences at junctions of a dual expression cassette for co-expression of eGFP and
mCherry. The eGFP and mCherry cDNAs were cloned into pKL (plasmid available from Addgene # Plasmid
#110741). PCR primers were designed to generate DNA fragments with 19- to 21-bp-long overhangs and
maintain the NheI, XhoI, BamHI, and XbaI restriction sites. The 30 region of the eGFP is detailed in panel (a), the
dual promoter region with the 50 region of the eGFP, the p10 and PH promoters, and the 50 region of the
mCherry cDNA in panel (b) and the 30 region of the mCherry cDNA in panel (c). The plasmid backbone is
amplified with the oligonucleotide pair Backbone-R/Backbone-F and the dual-promoter module with the
 Production of Multiprotein Complexes Using the Baculovirus Expression. . . 27

2. Decide on the constructs to be generated and select the transfer

vector accordingly. Consider, for example, the presence and
position of affinity tags.
3. Design a pair of generic primers to amplify the vector backbone
(Backbone-F and Backbone-R) and another one for the dual
promoter module (Prom-F and Prom-R). Whenever possible,
oligonucleotides Prom-F and Prom-R contain a kozak consen-
sus sequence (CCACCATGG, where ATG encodes the initia-
tor Methionine) and are perfectly complementary to their
vector template.
4. Design pairs of specific primers to amplify the target cDNAs
(pH-gene-START-F, pH-gene-STOP-R, p10-gene-START-F
and p10-gene-STOP-R). They should be composed of a 50
part corresponding to the cloning junction (which provides a
homology to the vector sequence for the homology-based
cloning reaction) followed by a 30 part complementary to the
insert target (which provides a priming region for the PCR).
The START primers include the first 15–21 nt of the cDNA,
preceded by a sequence complementary to that of the oligonu-
cleotide used to amplify the promoter region. The STOP pri-
mers include the last 15–21 nt of the gene, followed by a STOP
codon and the reverse complement of the primers used for the
backbone amplification.
5. Simulate the assembly of the expected transfer plasmids in silico
and carefully check that the tags, if any, are inserted in phase
with the genes of interest.
6. Check the propensity of the primers to form secondary struc-
tures or dimerize. Structures with a ΔG greater than
6 kcal/
mole suggest that redesign may be required, particularly for
those at the 30 end, for example, by increasing the length of the
50 -tail.
7. Order the oligonucleotides.
By default, oligonucleotides are chosen to generate PCR frag-
ments with 20 nucleotide-overlapping ends. They are usually
designed taking into account the length and base composition of
each primer pair, so that the Tm of the primers/templates are well
matched and close to 60  C (see Note 2). To minimize the forma-
tion of secondary structures, instead of using long oligonucleotides
(e.g., to append an affinity or detection tag), do not hesitate to use
nested PCRs.
ä

Fig. 4 (continued) Prom-R/Prom-F pair. The eGFP and mCherry cDNAs are amplified with the p10-gene-
START-F/p10-gene-STOP-R and the pH-gene-START-F/pH-gene-STOP-R. Note that oligonucleotides for the
amplification of the “promoter” cassette and of the 50 end of the cDNAs are designed to contain a Kozak
consensus CCACCATGG sequence where ATG (in bold) encodes the initiator codon (see Table 2)
28 Paola Rossolillo et al.

If an affinity or detection tag needs to be fused to one of the

target genes, the tag coding sequence can be added to the gene
with nested PCRs, or a modified transfer plasmid whose multiple
cloning site already contains the tag sequence can be used. The
plasmid and primer sequences at junctions of a dual expression
cassette for co-expression of eGFP and mCherry fused to a
C-terminal histidine tag are shown Fig. 5.

3.3 Preparation The linearized vector and the DNA fragments to be assembled are
of the Linearized prepared by PCR, except when the expression cassette is to be
Vector and of the DNA cloned into large transfer vectors such as pAC8_pMF derivatives.
Fragments 1. Set up 50 μL PCR to amplify the promoter and backbone
regions of the plasmid and the two genes (Table 3). Use
1–50 ng of template DNA if you amplify your fragments
from a plasmid. Use 100–200 ng of template if you amplify
your genes from cDNA.
2. Run the PCR in a thermocycler by using the conditions sug-
gested by the polymerase manufacturer. We typically denature
the template DNA at 98  C for 30 s, perform 25 cycles of
amplification (denaturation at 98  C for 10 s, annealing at the
calculated Tm for 15 s, and elongation at 72  C for 30 s in the
case of a 1 kb gene) followed by a final incubation at 72  C for
5 min.
3. Load 1/10 of the PCR volume on a 1% agarose gel in parallel
with a molecular weight marker to verify that the PCR amplifi-
cation resulted in a single band of the expected size.
4. If the fragments are amplified from a plasmid, add ten units of
the DpnI restriction enzyme to the PCR mix and incubate at
37  C for 1 h. DpnI will digest the (methylated) plasmid
template.
5. Purify the PCR on PCR clean-up columns following the
instruction manual, elute bound DNA in 30 μL, and
determine the concentration of the extracted DNA
spectrophotometrically.
We usually prepare between 1 and 2 μg of each fragment. Do
not hesitate to optimize PCR conditions (polymerase buffer,
annealing temperature, addition of DMSO, ) or redesign PCR
primers if yield or PCR specificity are not sufficient (see Note 3).
If DNA fragments are to be cloned into large transfer vectors
such as pAC8_pMF derivatives, the backbone DNA is prepared by
XbaI/NheI restriction and gel purification:
1. Digest 3–5 μg of plasmid in a total volume of 30 μL for 3 h at
37  C with ten units of NheI-HF and XbaI in CutSmart Buffer
(New England Biolabs).
Fig. 5 Plasmid and primer sequences at junctions of a dual expression cassette for co-expression of eGFP and
mCherry fused to a C-terminal histidine tag. The eGFP and mCherry cDNAs were cloned into the pKL derivative
pKL-PH-3C-10His-Cter. PCR primers were designed to generate DNA fragments with 19- to 21-bp-long
overhangs and maintain the NheI, XhoI, BamHI, and XbaI restriction sites. The 30 region of the eGFP cDNA is
detailed in panel (a), the dual promoter region with the 50 region of the eGFP cDNA, the p10 and PH promoters,
and the 50 region of the mCherry cDNA in panel (b), and the 30 region of the mCherry cDNA in panel (c). The
plasmid backbone is amplified with the oligonucleotide pair Backbone-3C-R/Backbone-F and the dual-
promoter module with the Prom-R/Prom-F pair. The eGFP and mCherry cDNAs are amplified with the p10-
gene-START-F/p10-gene-STOP-R and the pH-gene-START-F/pH-gene-10His-3C-R. The amino acid sequence
recognized by the prescission protease (3C) is LEVLFQ!GP (see Table 2)
30 Paola Rossolillo et al.

Table 3
Composition of the PCR mixes for amplification of the plasmid backbone, of the dual-promoter
module, and of the two cDNAs

Plasmid Promoter cDNA 1 under pH cDNA 2 under p10

backbone module promoter promoter
PCR pre-mix 25 μL 25 μL 25 μL 25 μL
Plasmid 10–50 ng 10–50 ng 10–50 ng 10–50 ng
template
cDNA 1 10–50 ng of plasmid or
100–200 ng of total
cDNA
cDNA 2 10–50 ng of plasmid or
100–200 ng of total
cDNA
Backbone-F 20 pmoles
Backbone-R 20 pmoles
Prom-F 20 pmoles
Prom-R 20 pmoles
pH-gene- 20 pmoles
START-F 20 pmoles
pH-gene-
STOP-R
p10-gene- 20 pmoles
START-F 20 pmoles
p10-gene-
STOP-R
H2O qsp 50 μL qsp 50 μL qsp 50 μL qsp 50 μL
For a 50 μL reaction, we typically mix 25 μL of PCR pre-mix, 3 μL of plasmid template at 10 ng/μL, and 2 μL of forward
and 2 μL of reverse primer at 10 μM each. The PCR pre-mix is prepared by mixing 10 μL of 5 polymerase buffer, 1 μL of
10 mM dNTP mix, one unit of Phusion DNA polymerase, and water to a 25 μL volume

2. Load 1/10 of the reaction volume on a 1% agarose gel with a

molecular weight marker to verify that the digestion is com-
plete and resulted in the expected products.
3. Treat with Calf Alkaline Phosphatase: add one unit of Calf
Alkaline Phosphatase (New England Biolabs) per microgram
of DNA in the restriction reaction. Increase the reaction vol-
ume to 50 μl and add 1/10 of the volume of CIAP buffer.
Incubate at 37  C for 1 h. The Alkaline Phosphatase will
eliminate phosphate groups from DNA extremities and will
prevent the re-circularization of the plasmid.
4. Purify the linearized plasmid using clean-up columns following
the instruction manual, elute bound DNA in 30 μL, and deter-
mine the concentration of the extracted DNA
spectrophotometrically.
 Production of Multiprotein Complexes Using the Baculovirus Expression. . . 31

3.4 SLIC Reaction As mentioned above, different approaches can be used to assemble
DNA fragments into a linearized vector containing overlapping
ends. Sequence and ligation–independent cloning (SLIC), in con-
trast to most other homology-based cloning strategies, does not
necessitate specific kits and only requires T4 DNA polymerase to be
tested first.
1. Set up the T4 DNA polymerase treatment for the four PCR
products in separate tubes. In each tube, mix 1 μg of DNA,
3 μL of 10 NEB 2.1 buffer, and 0.5 U of T4 DNA Polymer-
ase in a final volume of 30 μL. Incubate the reaction for 30 min
at room temperature (see Note 4).
2. Stop the exonuclease reaction by adding 3 μL of a 10 mM
dCTP solution (1/10 of the reaction volume) and leave on
ice. In the absence of dNTPs, the T4 DNA polymerase has a 30
to 50 exonuclease activity and “chews” the DNA leaving a
single-stranded region. Addition of dCTP stops the exonucle-
ase activity and shifts it to a polymerase activity at the first
occurrence of this nucleotide.
3. Prepare an annealing mix containing the polymerase-treated
vector backbone, the dual promoter unit, and the two cDNAs
in a 1:2:2:2 ratio using 250 ng (0.15 pmoles) of a 3 kb vector
and appropriate amount of inserts (see Note 5), in a total
volume of 20 μL. Incubate the annealing mix at 37  C for
30 min and leave on ice until transformation or store at

20  C. A negative control where one of the insert is omitted
can be included to evaluate background.
4. Use 10 μL of the annealing reaction to transform 100 μL of
E. coli competent cells (do not forget to use Pir1 cells for
plasmids with an R6Kγ replication origin). Plate the transfor-
mation mixture on LB agar plates containing the appropriate
antibiotic and incubate overnight at 37  C. If a control experi-
ment lacking one of the fragments was included, significantly
more colonies should be obtained in the presence of all inserts
than in the negative control.
5. Purify plasmid DNA from 6 to 12 colonies for restriction
analysis and sequencing. We typically first digest 500 ng of
purified plasmid with BamHI/XbaI and XhoI/NheI (both in
CutSmart Buffer) to control the presence of the desired inserts
(see Note 6). In the second step, two positive clones are
selected, and their expression cassettes are sequenced (see
sequencing oligos in Table 4).

3.5 Gibson Since SLIC is a simple and cost-effective technology, it is usually

Assembly®, tested by default in our laboratory. However, the assembly of some
NEBuilder® HiFi DNA constructs may prove more challenging than others, and it does not
Assembly, or always yield successful constructs through SLIC. In these cases, we
In-Fusion® HD Cloning try commercial technologies such as NEBuilder HiFi (NEB) or
32 Paola Rossolillo et al.

Table 4
List of primers for colony PCR analysis and sequencing

Region to sequence Primer name Primer sequence

0
5 end of cDNA 1 (pH) PH-F TAAAATGATAACCATCTCGC
30 end of cDNA 1 (pH) SV40 p(A)-R TTTAAAGCAAGTAAAACCTC
50 end of cDNA 2 (p10) p10-F CCCAACACAATATATTATAG
0
3 end of cDNA 2 (p10) HSV TK p(A)-R CACCCGTGCGTTTTATTCTGTC

In-fusion HD (Takara) cloning. NEBuilder HiFi cloning kit uses

the same technology as Gibson Assembly, although with higher
efficiency and better performance especially on short (<250 bp)
fragments.
These technologies are available as user-friendly cloning kits,
which provide the reagents necessary for the assembly of two or
more DNA fragments in a ligation-free reaction. However, prelim-
inary work is still needed, as for the SLIC reaction, to produce
DNA fragments with homologous sequence overhangs. A guide for
primer design is described in Subheading 3.2, and the preparation
of all DNA fragments, including linearization of the vector back-
bone, is described in Subheading 3.3.
Similar to the SLIC reaction, the purity of the fragments is
critical for their successful assembly. PCR and restriction digestion
products should be cleaned up or gel purified, if necessary, to
remove undesired by-products (see Note 7) before proceeding to
cloning.
1. On ice, set up a 20 μL reaction containing 1 master mix and
all DNA fragments in the selected molar ratio. We typically
work with 100 ng of vector and a molar ratio of 1:2:2:2, vector
to inserts (see Note 5), when assembling the dual promoter
unit and two cDNA fragments. A negative control reaction, in
which one of the fragments is omitted, may be set up in parallel.
2. Incubate the reaction at 50  C for 60 min. The samples can be
kept on ice until transformation or stored at
20  C.
3. Use 2–5 μL of the assembled product to transform 50 μL of
E. coli competent cells (Pir1 cells for plasmids with an R6Kγ
replication origin). Plate the transformation mixture on LB
agar plates containing the appropriate antibiotic and incubate
overnight at 37  C. If a control experiment was included,
significantly more colonies should be obtained in the presence
of all inserts than in the negative control.
 Production of Multiprotein Complexes Using the Baculovirus Expression. . . 33

4. Purify plasmid DNA from 6 to 12 colonies for restriction

analysis and sequencing. We typically first digest 500 ng of
purified plasmid with BamHI/XbaI and XhoI/NheI (both in
CutSmart Buffer) to control the presence of the desired inserts
(see Note 6). In the second step, two positive clones are
selected, and their expression cassettes are sequenced (Table 4).

3.6 Restriction-Free Preparation of multiprotein complexes has its own challenges and
(RF) Cloning specificities. Typically, constructs might need to be modified by
for the Modification mutation, truncation, or deletion of low complexity regions and
of the Existing the position or nature of affinity tags might have to be modified. In
Constructs the example shown in Fig. 6, RF cloning is used to introduce a
cleavable affinity tag at the 30 extremity of a cDNA in a complex
expression construct.
Restriction-free (RF) cloning, developed for the introduction
of a foreign DNA into a plasmid at any predetermined position, is
based on the overlap extension site-directed mutagenesis and popu-
larized under the name QuickChange mutagenesis (Stratagene™)
[18–20]. The sequence of interest is PCR-amplified using two
hybrid primers designed with complementarity to the desired insert
and the destination plasmid (Fig. 7a). The double-stranded PCR
product is used as a mega-primer for the second amplification
reaction (Fig. 7b). In this step, each of the DNA strands anneals
to the destination vector at the predetermined position and is
extended in a linear amplification reaction that leads to the forma-
tion of a double-stranded nicked plasmid. The parental DNA is
then removed by DpnI treatment (Fig. 7c), and the newly synthe-
sized plasmid, containing the DNA insert, is then transformed into
E. coli cells, in which the nicked DNA is ligated by endogenous
enzymatic activity.
1. Collect the plasmid/cDNA required for the first PCR and the
construct that you wish to modify, carefully verify their
sequences, and design the desired construct.
2. Design a pair of hybrid primers with a 50 part complementary to
the destination plasmid followed by a 30 part complementary to
the insert and compatible melting temperatures (see Note 2).
Oligonucleotides typically have a 18–25 bp overlap (Tm#
50  C) with the destination vector and 18–25 bp overlap
(Tm# 50  C) with the insert of interest.
3. Set up the first PCR (50 μL) to amplify the fragment to be
introduced using high-fidelity polymerase according to the
manufacturer’s instructions (see Note 8). Use 1–50 ng of tem-
plate DNA containing the sequence of interest if you amplify
your fragment from a plasmid. Use 100–200 ng of template if
you amplify your fragment from cDNA.
34 Paola Rossolillo et al.

Fig. 6 Modification of an existing construct. In this example, RF cloning is used to introduce a cleavable affinity
tag (3C-twinstrep) at the 30 extremity of the MAT1 cDNA which is part of a multigene expression construct. The
sequence of the modified plasmid where the existing part (in blue) and the insert (in green) are shown. The
hybrid primers MAT1-TwinStrep-FW and MAT1-TwinStrep-RV used to amplify the insert (3C cleavage site and
Twin-Strep tag) and introduce 20–22 bp overhangs matching the MAT1 gene at the insertion point (immedi-
ately before the Stop codon) are also displayed

4. Run the PCR in a thermocycler by using the conditions sug-

gested by the polymerase manufacturer. We typically denature
the template DNA at 98  C for 30 s, perform 25 cycles of
amplification (denaturation at 98  C for 30 s, annealing at the
calculated Tm for 15 s and elongation at 72  C for 60 s in the
case of a 1 kb gene) followed by a final incubation at 72  C for
10 min.
5. Load 1/10 of the PCR volume on a 1% agarose gel in parallel
with a molecular weight marker to verify that the PCR amplifi-
cation resulted in a single band of the expected size.
 Production of Multiprotein Complexes Using the Baculovirus Expression. . . 35

Fig. 7 Principle of site-directed insertion by RF cloning. (a) A first PCR amplifies the sequence to be inserted
(green, double-stranded) and introduces overhangs at each end (light blue strands), which overlap the
insertion site at the destination vector. The primers contain a target-specific sequence and an 18–25 bp
extension overlapping the desired insertion sites at the destination vector. The double-stranded PCR product is
then used as a megaprimer for the second amplification reaction. (b) In this step, each of the DNA strands
anneals to the destination vector and extends in a linear amplification reaction using the destination vector
(navy blue circumferences) as template, leading to a double-stranded nicked plasmid. (c) The parental DNA is
then removed by DpnI treatment, and newly synthesized plasmid, containing the DNA insert, is transformed
into Escherichia coli cells where the nicked DNA is sealed by endogenous repair activities. (d) Agarose gel
electrophoresis of the product from the first (left panel) and second (right panel) PCRs. The amplified insert
(lane 2, 132 bp) was used as a mega-primer to generate the desired vector (lane 4, 5800 bp). The Phusion
DNA polymerase was omitted in the negative control (lanes 1 and 3). Note that the 132-bp fragment, visible in
the negative control (lane 3), has been consumed in the presence of DNA polymerase (lane 4)

6. Purify the PCR product using PCR clean-up columns follow-

ing the instruction manual, elute bound DNA in 30 μL, and
determine the concentration of the extracted DNA
spectrophotometrically.
7. Set the second 50 μL PCR with 100–200 ng of the megapri-
mer, 20 ng of the recipient plasmid, and 0.8 μL of Phusion
polymerase (Thermo scientific™). Run the PCR in a thermo-
cycler by using the conditions suggested by the polymerase
manufacturer. As a negative control, set up the same reaction
omitting DNA polymerase.
8. 1/10 of the second PCR can be loaded on an agarose gel with a
molecular weight marker to verify that the PCR amplification
resulted in a band of the expected size, but this step is optional
(see Note 9).
9. Add 1 μL of DpnI restriction enzyme directly to the second
PCR and incubate 1 h at 37  C. DpnI enzyme will digest and
destroy the plasmid template still present in the PCR.
10. Use 2 μL of the DpnI-treated PCR product from the second
reaction to transform 50 μL of E. coli competent cells (Pir1
cells for plasmids with an R6Kγ replication origin). Plate the
36 Paola Rossolillo et al.

transformation mixture on LB agar plates containing the

appropriate antibiotic and incubate overnight at 37  C. If all
steps were successful, there will be no or little colonies in the
negative control and at least one order of magnitude more
colonies in the experiment.
11. Purify plasmid DNA from six colonies for restriction analysis
and sequencing.

4 Notes

1. E. coli strain Pir1 (F-Δlac169 rpoS(Am) robA1 creC510

hsdR514 endA recA1 uidA(ΔMluI)::pir-116) or Pir2
(F-Δlac169 rpoS(Am) robA1 creC510 hsdR514 endA recA1
uidA(ΔMluI)::pir) is for use with vectors that contain the R6Kγ
origin of replication. The pir gene encodes the replication
protein π, which is required to replicate and maintain plasmids
containing the R6Kγ origin. Other strains including BW23473
or BW23474 can also be used.
2. Refer to general guidelines for primer selection [21] and/or
use a primer design tool available from vector mapping soft-
ware or web-services (SnapGene or NEBuilder for the
planning of homology-based cloning projects and RF-Clon
ing.org for designing hybrid RF cloning primers). Also check
the propensity of the primers to form dimers or secondary
structures. Structures with a ΔG greater than
6 kcal/mole
suggest that redesign may be required.
3. It is important to have a sharp, specific PCR band for the SLIC
reaction. Gel purification of PCR products considerably
reduces the amount and the purity of the DNA. Large and
difficult-to-amplify cDNAs (more than 2 kb) can be splitted
into smaller fragments which are easier to prepare and can be
assembled more efficiently by SLIC.
4. Incubation time controls the formation of compatible single-
stranded ends and is thus a critical parameter for T4 polymerase
treatment. Do not hesitate to optimize the amount of T4 DNA
polymerase and incubation time when a new batch of polymer-
ase is used or when the overhang length between PCR pro-
ducts and vector is changed.
5. To calculate the amount of each DNA fragment to be used at a
given ratio and knowing the starting amount of vector, use the
formula:
insert ðbpÞ
Insert ðngÞ ¼ ratio   vector ðngÞ
vector ðbpÞ
 Production of Multiprotein Complexes Using the Baculovirus Expression. . . 37

For example, in a multiple fragment reaction, where we choose

a vector:insert ratio of 1:2, using 250 ng of a vector 3 kb long,
we would calculate the amount of 500-bp-long insert as:

500
Insert ðngÞ ¼ 2   250 ¼ 83:3 ng of insert
3000
6. As an alternative to restriction digestion, the presence of the
correct inserts can be checked by PCR or colony-PCR and
amplification across two or more assembled fragments, for
example, by using one primer that hybridizes on the coding
sequence and another on the plasmid backbone.
7. When using the In-Fusion reaction, the step of DNA purifica-
tion may be avoided if the PCR products show one clean band
of the expected size after gel electrophoresis. These samples
may be treated with Cloning Enhancer in the same PCR tube
and used directly to set up the In-Fusion reaction. Add 2 μL of
Cloning Enhancer to 5 μL of the PCR product and incubate at
37  C for 15 min, followed by 80  C for 15 min. Use a
thermocycler for accurate temperature control.
8. It is important to have a sharp, specific PCR band as a mega-
primer. The eventual gel purification of the PCR should be
avoided since it can impact the efficiency of the following PCR
amplification reaction. It is crucial to optimize the PCR condi-
tions until a specific band is obtained.
9. Agarose gel electrophoresis after the second PCR is optional.
Even if the discrete band of the high molecular weight PCR
product is not clearly visible, there might be enough material to
obtain several colonies. You can proceed immediately to E. coli
transformation step after DpnI treatment.

Acknowledgments

This work was supported by the Centre National pour la Recherche

Scientifique (CNRS), the Institut National de la Santé et de la
Recherche Médicale (INSERM), Université de Strasbourg (UdS),
the Association pour la Recherche sur le Cancer (ARC), the Ligue
nationale contre le cancer, the Institut National du Cancer (INCa;
INCA 9378), and the Agence National pour la Recherche
(ANR-12-BSV8-0015-01 and ANR-10-LABX-0030-INRT under
the program Investissements d’Avenir ANR-10-IDEX-0002-02).
Instruct (R&D Project Funding), Instruct-ULTRA as part of the
European Union’s Horizon 2020 (Grant ID 731005), and
Instruct-ERIC are also acknowledged.
38 Paola Rossolillo et al.

References
1. Contreras-Gomez A, Sanchez-Miron A, using the baculovirus expression system. Meth-
Garcia-Camacho F, Molina-Grima E, Chisti Y ods Mol Biol 1261:91–114
(2014) Protein production using the 12. Abdulrahman W, Iltis I, Radu L, Braun C,
baculovirus-insect cell expression system. Bio- Maglott-Roth A, Giraudon C, Egly JM,
technol Prog 30(1):1–18 Poterszman A (2013) ARCH domain of
2. Martinez-Solis M, Herrero S, Targovnik AM XPD, an anchoring platform for CAK that con-
(2019) Engineering of the baculovirus expres- ditions TFIIH DNA repair and transcription
sion system for optimized protein production. activities. Proc Natl Acad Sci U S A 110(8):
Appl Microbiol Biotechnol 103(1):113–123 E633–E642
3. Berger I, Poterszman A (2015) Baculovirus 13. Kobbi L, Demey-Thomas E, Braye F, Proux F,
expression: old dog, new tricks. Bioengineered Kolesnikova O, Vinh J, Poterszman A, Ben-
6(6):316–322 saude O (2016) An evolutionary conserved
4. Gupta K, Tolzer C, Sari-Ak D, Fitzgerald DJ, Hexim1 peptide binds to the Cdk9 catalytic
Schaffitzel C, Berger I (2019) MultiBac: site to inhibit P-TEFb. Proc Natl Acad Sci U
baculovirus-mediated multigene DNA cargo S A 113(45):12721–12726
delivery in insect and mammalian cells. Viruses 14. Radu L, Schoenwetter E, Braun C, Marcoux J,
11(3):198 Koelmel W, Schmitt DR, Kuper J, Cianferani S,
5. Berger I, Fitzgerald DJ, Richmond TJ (2004) Egly JM, Poterszman A, Kisker C (2017) The
Baculovirus expression system for heterolo- intricate network between the p34 and p44
gous multiprotein complexes. Nat Biotechnol subunits is central to the activity of the tran-
22(12):1583–1587 scription/DNA repair factor TFIIH. Nucleic
6. Fitzgerald DJ, Berger P, Schaffitzel C, Acids Res 45(18):10872–10883
Yamada K, Richmond TJ, Berger I (2006) Pro- 15. Gibson DG, Young L, Chuang RY, Venter JC,
tein complex expression by using multigene Hutchison CA 3rd, Smith HO (2009) Enzy-
baculoviral vectors. Nat Methods 3 matic assembly of DNA molecules up to several
(12):1021–1032 hundred kilobases. Nat Methods 6
7. Trowitzsch S, Bieniossek C, Nie Y, Garzoni F, (5):343–345
Berger I (2010) New baculovirus expression 16. Irwin CR, Farmer A, Willer DO, Evans DH
tools for recombinant protein complex produc- (2012) In-fusion(R) cloning with vaccinia
tion. J Struct Biol 172(1):45–54 virus DNA polymerase. Methods Mol Biol
8. Weissmann F, Petzold G, VanderLinden R, 890:23–35
Huis In’t Veld PJ, Brown NG, Lampert F, 17. Li MZ, Elledge SJ (2007) Harnessing homol-
Westermann S, Stark H, Schulman BA, Peters ogous recombination in vitro to generate
JM (2016) biGBac enables rapid gene assembly recombinant DNA via SLIC. Nat Methods 4
for the expression of large multisubunit protein (3):251–256
complexes. Proc Natl Acad Sci U S A 113(19): 18. Bryksin AV, Matsumura I (2010) Overlap
E2564–E2569 extension PCR cloning: a simple and reliable
9. Zhang Z, Yang J, Barford D (2016) Recombi- way to create recombinant plasmids. Biotech-
nant expression and reconstitution of multi- niques 48(6):463–465
protein complexes by the USER cloning 19. Unger T, Jacobovitch Y, Dantes A,
method in the insect cell-baculovirus expres- Bernheim R, Peleg Y (2010) Applications of
sion system. Methods 95:13–25 the Restriction Free (RF) cloning procedure
10. Osz-Papai J, Radu L, Abdulrahman W, Kolb- for molecular manipulations and protein
Cheynel I, Troffer-Charlier N, Birck C, expression. J Struct Biol 172(1):34–44
Poterszman A (2015) Insect cells-baculovirus 20. van den Ent F, Lowe J (2006) RF cloning: a
system for the production of difficult to express restriction-free method for inserting target
proteins. Methods Mol Biol 1258:181–205 genes into plasmids. J Biochem Biophys Meth-
11. Abdulrahman W, Radu L, Garzoni F, ods 67(1):67–74
Kolesnikova O, Gupta K, Osz-Papai J, 21. Bustin SA, Mueller R, Nolan T (2020) Para-
Berger I, Poterszman A (2015) The produc- meters for successful PCR primer design.
tion of multiprotein complexes in insect cells Methods Mol Biol 2065:5–22
 Chapter 3

Tagging Proteins with Fluorescent Reporters Using

the CRISPR/Cas9 System and Double-Stranded DNA Donors
Sylvain Geny, Simon Pichard, Alice Brion, Jean-Baptiste Renaud,
Sophie Jacquemin, Jean-Paul Concordet, and Arnaud Poterszman

Abstract
Macromolecular complexes govern the majority of biological processes and are of great biomedical
relevance as factors that perturb interaction networks underlie a number of diseases, and inhibition of
protein–protein interactions is a common strategy in drug discovery. Genome editing technologies enable
precise modifications in protein coding genes in mammalian cells, offering the possibility to introduce
affinity tags or fluorescent reporters for proteomic or imaging applications in the bona fide cellular context.
Here we describe a streamlined procedure which uses the CRISPR/Cas9 system and a double-stranded
donor plasmid for efficient generation of homozygous endogenously GFP-tagged human cell lines. Estab-
lishing cellular models that preserve native genomic regulation of the target protein is instrumental to
investigate protein localization and dynamics using fluorescence imaging but also to affinity purify asso-
ciated protein complexes using anti-GFP antibodies or nanobodies.

Key words CRISPR/Cas9, Genome editing, Double-stranded DNA donors, Fluorescent protein

1 Introduction

Molecular complexes of interacting proteins govern virtually all

biological processes such as metabolism, cell signaling, DNA repair,
and gene expression. Macromolecular assemblies are also of great
biomedical relevance as their dysfunctions underlie a number of
diseases, and deliberate inhibition of protein–protein interactions is
an increasingly common strategy in drug discovery [1–3]. To fully
understand their biological roles, it is essential to study the struc-
ture and function of intact protein assemblies. Although advanced
recombinant protein technologies are available to reconstitute mul-
tiprotein complexes composed of ten or more subunits, many
protein complexes are difficult to obtain using recombinant meth-
ods. An additional hurdle is that the subunit composition of com-
plexes is not always known well enough to proceed to

Arnaud Poterszman (ed.), Multiprotein Complexes: Methods and Protocols, Methods in Molecular Biology, vol. 2247,
https://doi.org/10.1007/978-1-0716-1126-5_3, © Springer Science+Business Media, LLC, part of Springer Nature 2021

39
40 Sylvain Geny et al.

reconstitution of functional entities and therefore to structure

determination.
The CRISPR/Cas9 system has revolutionized many fields of
life sciences by making it possible to modify the genome sequence
with unprecedented efficiency and precision [4, 5]. For example, it
is now possible to insert fusion tags at endogenous loci of mamma-
lian cells, providing an efficient way to undertake affinity purifica-
tion of macromolecular complexes and/or visualize their
distribution and dynamics in a cellular context [6, 7]. We have
recently detailed a simplified protocol for CRISPR/Cas9-mediated
gene tagging in human cell lines using chemically modified single-
stranded oligonucleotides encoding a small affinity tag as donor
template and a co-selection strategy targeting the ATP1A1 gene
[8] (to be published in MiB, Editor R Owens). Here we describe a
procedure based on the use of the CRISPR/Cas9 system and
double-stranded DNA donors for tagging the protein of interest
with a fluorescent reporter and detail the generation of an U2-OS::
XPB-GFP
knocked-in cell line expressing a XPB-GFP fusion protein.
XPB is a subunit of the TFIIH complex, essential in initiation of
DNA transcription by RNA polymerase II and DNA repair by
nucleotide excision repair [9, 10]. This XPB-GFP-tagged U2-OS
cell line was instrumental to establish a partnership between TFIIH
and the histone acetyl transferase GCN5 and the impact of TFIIH
on GCN5 activity with important consequences on gene expression
and chromatin structure [11].

2 Materials

Procedures described here need access to standard equipment for

molecular biology (PCR amplification, agarose gel analysis, bacteria
transformation, access to a DNA sequencing/synthesis service,
etc.), cell culture (cryo-container and liquid nitrogen source, tem-
perature- and CO2-controlled incubator, laminar flow hood, cen-
trifuge with adaptor for 15 and 50 mL tubes, cell counter), cell
microscopy (fluorescence microscope with 63
or higher magnifi-
cation objectives, etc.), and protein analysis (refrigerated micro-
centrifuge, a small-scale ultrasonic homogenizer, nano-UV spec-
trometer, protein gel electrophoresis, and Western blotting transfer
system, etc.). Basic knowledge in these fields is expected.
Experiments detailed below were performed in U2-OS cells
(ATCC HTB-96). These adherent cells are grown at 37  C, 5%
CO2 in McCoy’s 5a or DMEM-based culture medium (see Note 1).
For experiments in Subheadings 3.2–3.5, in addition to specific
materials, you will need:
 Tagging Proteins with Fluorescent Reporters Using the CRISPR/Cas9 System. . . 41

1. T75 cell culture flasks (75 cm2) (Corning, cat. no. 431464 U).
2. McCoy’s 5a or DMEM medium supplemented with 10% fetal
bovine serum (FBS), 4 mM GlutaMAX™ (Gibco, cat.
no. 35050061) and 1% PenStrep (Gibco, cat. no. 15140-122).
3. Phosphate-buffered saline (PBS).
4. Trypsin–EDTA solution (0.025% w/v Trypsin, 0.05 mM
EDTA in 1
PBS) (Gibco, cat. no. R001100).

2.1 Reagents 1. Oligonucleotides for preparing (sgRNA-FW, sgRNA-RV) and

for the Preparation sequencing (U6-seq) the single guide RNA expression plas-
of sgRNA Expression mids (sgRNA) (Table 1).
Plasmids 2. Guide RNA expression plasmid: MLM3636 (Addgene plasmid
#43860).
3. T4 DNA ligase with buffer for the preparation of the guide
RNA expression plasmid (New England BioLabs, cat.
no. M0202S).
4. BsmBI restriction enzyme (New England BioLabs, cat.
no. R0580S).

Table 1
XPB locus

U-2 OSXPB::GFP
Chromosome/gene Chr. 2/XPB
Homology arms 50 Chr.2: 127257599–127257872 (hg 38)
30 Chr.2: 127257596–127256813 (hg 38)
sgRNA 50 TAGGAAATGATGCTTAGGCAggg 30
Selection marker Puromycin
sgRNA-FW 50 ACACCG TAGGAAATGATGCTTAGGC G 30
sgRNA-RV 50 AAAAC TGCCTAAGCATCATTTCCTA CG
U6-seq 50 CAGGGTTATTGTCTCATGAGCGG-30
XPB-FW 50 AGACAGTAAGCGATCTGTAAACA 30
XPB-RV 50 ACCCCACTCCCCAAAAAGTT 30
XPB-FW2 50 TCCTCTTCTTTCAGGTGTGGA 30
GFP-RV 50 GAACTTCAGGGTCAGCTTGC 30
Puro-FW 50 GCAACCTCC CCTTCTACGAG 30
XPB-RV2 50 GCGAATATGCCTTATGTGTG 30
Target sequence for the sgRNA used for editing the C-terminus of the XPB gene. Online web applications such as
CRISPOR (http://crispor.tefor.net/) aimed at optimizing CRISPR knock-in tag experimentation identify and rank
potential guide RNAs in an input sequence. Oligonucleotides for sense and antisense strands designed with CRISPOR
software for cloning into MLM3636 (sgRNA-FW and sgRNA-RV) as well as for genomic screening (XPB-FW, RV) are
detailed. Efficient transcription from the U6 promoter requires a 50 G (underlined)
42 Sylvain Geny et al.

5. LB and LB agar medium, ampicillin, competent DH5α E. coli

cells (New England BioLabs, cat. no. C2987I).
6. Plasmid DNA Mini and Midiprep purification kits (Macherey-
Nagel™ NucleoSpin Plasmid QuickPure™ Kit and Macherey-
Nagel™ NucleoBond™ Xtra Midi).

2.2 Reagents 1. Expression plasmid for SpCas9: JDS246 (Addgene plasmid

for Nucleofection #43861).
and Assessment 2. Expression plasmid for sgRNA (Addgene plasmid #43860)
of Gene Editing (from Subheading 3.1, step 4).
2.2.1 Nucleofection 3. AMAXA nucleofector machine® (Lonza) and consumables
with AMAXA Nucleofector including cuvettes (Lonza, cat. no. AAB-1001), pipettes, and
Machine solution V kit (Lonza, cat. no. VVCA-1003).
4. Tissue culture plate, 6 wells (Eppendorf, cat.
no. EP0030720121).

2.2.2 Pool and Clone 1. QuickExtract DNA solution (Viagen, cat. no. 302C).
Analysis 2. Oligonucleotides for PCR amplification and sequencing (see
Table 1).
3. Phusion high-fidelity DNA polymerase with buffer and dNTPs
(New England BioLabs, cat. no. M053OS).
4. PCR isolation kit (Macherey-Nagel™ NucleoSpin™ Gel and
PCR Clean-up Kit).

2.3 Reagents In addition to the reagents described in Subheading 2.2.1

for Gene Editing
1. Donor plasmid (from Subheading 3.3, step 2).
and FACS Sorting
2. Puromycin dihydrochloride (InvivoGen, cat. no. ant-pr-1).
2.3.1 Nucleofection
and Antibiotic Selection

2.3.2 FACS Sorting Sorting of GFP-fluorescent cells was carried out on a Melody FACS
(BD), and analysis of GFP-positive cells was carried on an Accuri
C6 analyzer (BD). You will need:
1. Cytometry Falcon tubes (Becton Dickinson).
2. Tissue culture plate, 96 wells (Eppendorf, cat.
no. EP0030730119).

2.4 Reagents In addition to reagents described in Subheading 2.2.2, you will

for the Validation need:
of Selected Clones
1. Tissue culture plates, 6 and 96 wells (Eppendorf, cat.
2.4.1 Genotyping no. EP0030720121, EP0030730119).
of Single-Cell Clones 2. Oligonucleotides from the inserted sequence and from the
target chromosomal locus.
 Tagging Proteins with Fluorescent Reporters Using the CRISPR/Cas9 System. . . 43

2.4.2 Western Blot 1. 150-mm Petri dishes (Falcon® 150 mm TC-treated dishes,
Analysis Corning, cat. no. 353025).
2. Cell scrapers (Dominique Dutscher, cat. no. 353085).
3. PBS containing 30% w/v glycerol (Sigma Aldrich, cat.
no. G5516).
4. RIPA buffer: 20 mM Tris-HCl or HEPES, pH 7.5, 120 mM
KCl, 1% NP-40, 0.1% SDS, 1 mM EDTA, 0.5%
Na-deoxycholate, supplemented with protein inhibitor cocktail
(Roche™) and 0.5 mM 1,4-dithreothiol (DTT) (Sigma
Aldrich).
5. Lysis buffer: 20 mM Tris or HEPES, 250 mM KCl, NP-40
0.05% supplemented with protein inhibitor cocktail (Roche™)
and 0.5 mM DDT.
6. Protein assay dye reagent (Bio-Rad, cat. no. 5000006).
7. Laemmli buffer 4
: 60 mM Tris-HCl, pH 6.8, 10% glycerol,
2% SDS, 0.0005% Bromophenol Blue, 355 mM
β-mercaptoethanol.
8. SDS PAGE gels, PVDF transfer membranes, 3MM Whatman
paper.
9. TBST: 20 mM Tris/HCl pH 7.4, 150 mM NaCl, 0.1% Tween-
20.
10. Non-fat dry milk or BSA powder.
11. Western blotting detection reagents (Amersham ECL Prime,
cat no. 3030-931).
12. Primary antibodies against the target protein or against the
GFP tag and corresponding secondary antibodies coupled to
horseradish peroxidase (HRP). For the detection of XPB pro-
tein, we used a mouse monoclonal anti-XPB antibody (1B3,
MABE1123, Sigma-Aldrich) and a donkey anti-mouse anti-
body coupled with HRP.

2.4.3 Fluorescence 1. Glass-bottom dishes (Glass Bottom Dish 35 mm, Clinis-

Microscopy ciences/Ibidi, cat. no. 81218-200).
2. Epifluorescence inverted microscope with filters compatible
with DAPI/Hoechst and the fluorescent reporter used (here
GFP).
3. Paraformaldehyde (PFA) 16% ultrapure methanol free (Euro-
medex, cat. no. 15710).
4. Hoechst (Sigma Aldrich, cat. no. 32670).
5. Antibodies to confirm the expression of the target protein or
for co-localization experiments.
44 Sylvain Geny et al.

3 Methods

Targeted gene knock-in is achieved through the following steps.

First, the Cas9 endonuclease is directed by a guide RNA to a
specific target site in the genome to generate a double-stranded
break (DSB). The DSB is then repaired with a donor template
through a homology-directed repair (HDR) pathway [12]. Below,
we first describe the design of the single-guide RNAs and donor
plasmid template to edit the genome (Subheadings 3.1–3.3) and
the clonal isolation of CRISPR/Cas9-modified cells (Subheading
3.4). Selected clones are finally characterized at the genomic and
protein levels (Subheading 3.5) (Fig. 1).
To facilitate the generation of homozygous knock-in cells, we
favored a tagging strategy based on the use of a donor plasmid
containing 300–1000 bp homology arms flanking the eGFP-2A-
PURO coding sequence [13] (Fig. 2). This strategy allows to
combine transient antibiotic selection with fluorescence-assisted
cell sorting (FACS) to isolate CRISPR/Cas9-modified cells [14].

3.1 Design The design of the sgRNA and HDR donor depends on the nature
of the sgRNAs and position of the tag (see Note 2). As the efficiency of HDR
insertion varies with the distance from the DSB, the DSB (gener-
ated 3 bp upstream of the PAM sequence corresponding to the
sgRNA) should be as close as possible to the insertion site of tag.
For example, to tag a protein at its C-terminal end, the DSB should
be as close as possible to the stop codon of the gene of interest
(Fig. 3).

sgRNAs Design DONOR Design Validation of

Efficiency assesment Gene targetting and cloning (FACS) selected clones
Sections 3.1, 3.2 Sections 3.3, 3.4 Section 3.5

Cas nuclease guide RNA donor template

synthetic
Cas protein ssDNA Nucleofection
crRNA:tracRNA
in vitro
Cas mRNA PCR product
transcribed sgRNA

Cas plasmid sgRNA plasmid plasmid

Lentiviral Cas Lentiviral sgRNA

Fig. 1 Experimental workflow for CRISPR/Cas9-mediated genome editing. Several sources of Cas nuclease,
guide RNA and donor template can be used. We detailed a protocol that utilizes electroporation (nucleofection)
for the delivery of plasmidic reagents (donor, Cas9, sgRNA)
 Tagging Proteins with Fluorescent Reporters Using the CRISPR/Cas9 System. . . 45

sgRNA

Chr.2

STOP
XPB exon Last exon

STOP
5’ HA GFP P2A Puro 3’ HA

~700 nt ~1500nt ~700 nt

Chr.2 - GFP P2A Puro

STOP
XPB exon Last exon

Fig. 2 Schematic of the tagging strategy for the XPB locus. To introduce a GFP tag at the C-terminal end of the
XPB gene, we select an sgRNA to generate a CRISPR/Cas9-induced DSB in vicinity of the stop codon of the
XPB gene. The donor plasmid contains the eGFP-2A-Puro coding sequence flanked by respectively 300 and
800 bp homology arms (50 and 30 HA). The eGFP-2A-Puro coding sequence is inserted at the position of the
XPB STOP codon and in frame with the last exon of the target protein

1. Retrieve the nucleotide sequence of the gene of interest (GOI)

and focus on the site of tag insertion. The UCSC genome
browser is a convenient source for this step (genome.ucsc.edu).
2. Screen the genomic regions of interest using an online CRISPR
design tool such as CRISPOR (http://crispor.tefor.net/) to
identify and rank the guide sequences (Table 1). We recom-
mend testing several sgRNAs with cut sites less than 20 bp away
from the insertion site to find an optimal sgRNA. Guide RNAs
for the human genome can also be visualized directly in the
UCSC genome browser with pre-calculated data from CRIS-
POR (select “Full” display mode of the “CRISPR targets” in
the “Genes and Gene Predictions” list of features).
3. Order the sense and antisense oligonucleotides for cloning the
sgRNA sequences into an sgRNA expression plasmid such as
MLM3636. Note that the oligonucleotides contain overhangs
for ligation into the pair of BsmBI sites with the sense and
antisense sequences matching the genomic target. The oligo-
nucleotide sequences can be accessed by clicking on the “PCR/
cloning primers” button in CRISPOR. If you have used the
UCSC genome browser for sgRNA selection, you can directly
transfer the sgRNA sequence to CRISPOR by a hyperlink
feature or simply by copy-pasting to the CRISPOR input page.
46 Sylvain Geny et al.

A
Cas9

TAGGAAATGATGCTTAGGCA PAM

ATCCTTTACTACGAATCCGT
||||||||||||||||||||
UAGGAAAUGAUGCUUAGGCA

sgRNA

B Cas9 cleavage site

5’ Homology Arm PAM 3’ Homology Arm

+ CCGCTCTTCAAGCGCTTTAGGAAA TGATGCTTAG GCAGGG TACTTCGTTCAAGACCGGCGCTT...

locus |P |L |F |K | R |F |R |K| *
- GGCGAGAAGTTCGCGAAATCCTTT ACTACGAATC CGTCCC ATGAAGCAAGTTCTGGCCGCGAA...

5 Homology Arm 3’ Homology Arm

CCGCTGTTCAAGAGATTTCGGAAAGGTGGTGGC... ...CGGTGCCTGA TACTTCGTTCAAGACCG...
Repair +
donnor |P |L | F |K | R | | | | |
F| R K G G GFP P2A Puro | | |
G A *
- GGCGACAAGTTCTCTAAAGCCTTTCCACCACCG... ...GCCACGGACT ATGAAGCAAGTTCTGGC...

Fig. 3 Design of sgRNAs and HDR donor sequences. (a) Schematic representation of the Cas9 protein (gray)
bound to sgRNA (red) and targeted to genomic DNA (black). The PAM sequence (GGG with overbar), the Cas9
cleavage sites (black triangles), and the stop codon of the XPB gene (TGA, in cyan) are indicated. A ribbon
representation of the Cas9 nuclease from Streptococcus pyogenes bound to PAM-containing DNA target is
shown in the right panel (PDB code, 4UN3; [30]). (b) Repair donor designed to insert the eGFP-2A-Puro coding
sequence (all in frame with the last exon of the target protein) upstream of the XPB STOP codon (TGA).
Nucleotides targeted by the sgRNA are in bold. After integration of the tag, the sequence is significantly
modified (red line above the + strand), preventing cleavage of the repaired locus by Cas9

4. Generate the oligo duplex, clone into the BsmBI predigested

plasmid MLM3636, sequence the resulting plasmid. Trans-
form and amplify the guide expression plasmid in DH5 com-
petent cells using a Midi or Maxi Prep kit depending on the
amount of plasmid required.

3.2 Delivery As knock-in using double-stranded donor template has relatively

of CRISPR Reagents low efficiency, successful HDR-mediated experiments require
and Assessment experimental optimization. In particular, it is recommended to
of Gene Editing control and optimize delivery of CRISPR reagents by evaluating
the levels of gene editing in absence of donor template. In some
cases, GFP fluorescence can be used to directly characterize and
optimize HDR events, but low to very low levels of fluorescent-
tagged proteins driven by endogenous promoters can be
 Tagging Proteins with Fluorescent Reporters Using the CRISPR/Cas9 System. . . 47

challenging to detect. An attractive alternative consists in delivering

the nuclease with a single strand donor designed to fuse a HiBiT
peptide to the target protein. The abundance of HiBiT-tagged
proteins in transfected cells is proportional to the proportion of
edited gene copies and is readily detected using the Nano-Glo
HiBiT luciferase-based detection system (see ref. 8).
In this section, we detail the delivery of CRISPR/Cas9
reagents by nucleofection (see Note 3) and PCR-based analysis of
the cleavage efficiency of guide RNAs.

3.2.1 Nucleofection 1. Thaw and maintain U2-OS cells using your favorite protocol
(see Note 1). We usually split cells 1:5 every 3–4 days, verify
that the viability is always higher than 95% and limit the num-
ber of passages to 20.
2. Two or three days before nucleofection, seed two 75-cm2 flasks
and incubate at 37  C, 5% CO2.
3. When 50–70% confluence is reached, detach the cells with
trypsin, spin-down in a 15-mL tube at 90 g at room tempera-
ture for 10 min, remove the supernatant and re-suspend the
cells in PBS to a concentration of 1–3
106 cells/mL.
4. Aliquot 106 cells into 1.5-mL microcentrifuge tubes (one tube
per guide RNA to be tested plus an additional tube for a
negative control), pellet the cells, discard supernatant
completely, and resuspend cells in Nucleofector® solution
(100 μL per sample).
5. Prepare the DNA mixes containing 2 μg of guide sgRNA
plasmid (sgXPB) and 2 μg of plasmid expressing the SpCas9
protein (JDS246) in a total volume less than 10 μL.
6. Mix the DNA solution with 100 μL of cell suspension and
transfer to a cuvette. Process the samples quickly to avoid
storing the cells longer than 10 min in Nucleofector®
Solution V.
7. Insert the cuvette into the Nucleofector®, select the cell-type-
specific program X-001 and press the start button. Using the
provided pipette, immediately remove the sample from the
cuvette and transfer into the six-well plate containing 2 mL of
culture medium.
8. Incubate the six-well plates at 37  C, 5% CO2 and proceed to
analysis or collect the cells and store the cell pellet at 20  C.

3.2.2 PCR-Based CRISPR/Cas9 cleavage is usually assessed by amplifying the target

Analysis of the Cleavage region by PCR and analyzing the mutation rate in the resulting
Efficiency for the sgRNA product. This can be performed using various assays such as the
polymerase chain reaction (PCR)/restriction enzyme (RE) assay,
T7 endonuclease I (T7EI) assay, Surveyor nuclease assay or with
TIDE [15–18].
48 Sylvain Geny et al.

Table 2
Composition of the PCR mix used to amplify the genomic site XPB

Genomic DNA (50 ng/μL) 1 μL

dNTPs (25 mM) 0.4 μL
DNA polymerase (2000 U/mL) 0.5 μL
Buffer PCR 5
10 μL
Forward Primer (10 μM) 2.5 μL
Reverse Primer (10 μM) 2.5 μL
H2O 33.1 μL
Use the following PCR program: 98  C/30 s; 98  C/10 s; 65  C/20 s; 72  C/20 s
(30 cycles); 72  C/5 min; 4  C/hold

1. Three days after transfection, wash cells with PBS, extract

genomic DNA (typically from 1
105 cells) using QuickEx-
tract DNA extraction kit or similar according to the manufac-
turer’s recommendations. Dilute genomic DNA at a final
concentration around 50 ng/μL.
2. Design a pair of PCR primers (XPB-FW and XPB-RV) which
hybridize on either side of the target site to amplify a ~500 bp
stretch of DNA. The sgRNA cut site should ideally be located
~200 bp downstream from one of the primers for convenient
analysis by DNA sequencing.
3. Amplify the locus of interest by PCR (see Table 2), run 2 μL of
PCR products on 1% agarose gel TBE 0.5
at 100 V for
90 min and visualize DNA using a UV transilluminator. A
sharp single band of the expected size should be visible in
control samples. Additional bands may be detected in test
samples corresponding to insertion and/or deletion.
4. Purify the PCR products using an appropriate PCR purification
kit and measure its concentration.
A wide panel of assays can be used to analyze PCR products and
verify the efficacy of programmable nucleases. The enzymatic Sur-
veyor and T7 endonuclease I cleavage assays [19, 20] are com-
monly used, but these approaches are semiquantitative and their
sensitivity is limited [15]. We favor the TIDE assay that accurately
quantifies editing efficacy and simultaneously identifies the pre-
dominant types of insertions and deletions (indels) in the pool of
treated cells [21].
5. Send the PCR products (control and experimental samples) for
Sanger sequencing and retrieve the sequence trace files in .ab1
or .scf format.
 Tagging Proteins with Fluorescent Reporters Using the CRISPR/Cas9 System. . . 49

6. Upload the trace files and the guide RNA sequence (20 nt) into
the TIDE web tool (available at http://tide.nki.nl and https://
deskgen.com) and perform the analysis with default
parameters.
Use the TIDE assay to compare the efficiency of the different
guides and select the best experimental conditions for further
experiments. An example of TIDE analysis is provided on the
TIDE website.

3.3 Design Several types of DNA repair template can be utilized in

of Plasmid Donor HDR-mediated CRISPR/Cas9 gene editing. Donor plasmids,
with homology arms of 300–1000 bp allow for large insertions in
the 1–2 kbp range. Other types of DNA repair templates such as
linear DNAs with short homology arms obtained by PCR or gene
synthesis can also be used for inserting fluorescent labels or other
protein tags such as the Halo or Snap tags [22, 23]. Linear DNA,
however, may concatemerize before insertion [24], which may
perturb gene expression or make the modification unstable during
cell replication. Different types of donor DNA will give similar or
different KI efficiencies depending on the gene targeted and cell
line used and may be tested if needed.
This section details the design and preparation of the donor
plasmid which requires the introduction of the coding sequence for
the insertion flanked by two homology arms of a few hundred bases
(typically between 0.3 and 1.0 kbp).
1. Define the DNA fragment that you wish to introduce, which in
this example case consists in the eGFP-P2A-PURO coding
sequence where eGFP, the 2A self-cleaving peptide (P2A),
and the puromycin N-acetyl transferase coding sequence are
separated by flexible glycine/serine-rich linkers (Fig. 4). Refer-
ences [25, 26] provide useful information on the use of 2A
peptides in a polycistronic vector and the design of protein
linkers, respectively. Note that when using 2A self-cleaving
peptide, antibiotic resistance is expressed from the target
gene, which minimizes the selection of clones with
non-targeted integration. However, in some cases, expression
levels of the target gene may be insufficient to provide antibi-
otic resistance, and it may be necessary to provide antibiotic
resistance from a constitutive promoter; higher numbers of
clones with random integration of the donor plasmid into the
cell genome may be isolated [27].
2. Retrieve the sequence of a fragment encompassing upstream
and downstream homology arms of approximatively 700 bp
and replace the STOP codon of the GOI by the eGFP-P2A-
PURO coding sequence. Carefully verify that the eGFP-P2A-
PURO coding sequence is in frame with the last exon of the
50 Sylvain Geny et al.

ggtggtggcggttcaggcggaggtggctctggcggtggcggatcggccATGGTGAGCAAGGGCGAGGAGCTGTTCACCGG

GGTGGTGCCCATCCTGGTCGAGCTGGACGGCGACGTAAACGGCCACAAGTTCAGCGTGTCCGGCGAGGGCGAGGGCGATG

CCACCTACGGCAAGCTGACCCTGAAGTTCATCTGCACCACCGGCAAGCTGCCCGTGCCCTGGCCCACCCTCGTGACCACC

CTGACCTACGGCGTGCAGTGCTTCAGCCGCTACCCCGACCACATGAAGCAGCACGACTTCTTCAAGTCCGCCATGCCCGA

AGGCTACGTCCAGGAGCGCACCATCTTCTTCAAGGACGACGGCAACTACAAGACCCGCGCCGAGGTGAAGTTCGAGGGCG

ACACCCTGGTGAACCGCATCGAGCTGAAGGGCATCGACTTCAAGGAGGACGGCAACATCCTGGGGCACAAGCTGGAGTAC

AACTACAACAGCCACAACGTCTATATCATGGCCGACAAGCAGAAGAACGGCATCAAGGTGAACTTCAAGATCCGCCACAA

CATCGAGGACGGCAGCGTGCAGCTCGCCGACCACTACCAGCAGAACACCCCCATCGGCGACGGCCCCGTGCTGCTGCCCG

ACAACCACTACCTGAGCACCCAGTCCGCCCTGAGCAAAGACCCCAACGAGAAGCGCGATCACATGGTCCTGCTGGAGTTC

GTGACCGCCGCCGGGATCACTCTCGGCATGGACGAGCTGTACAAGtacagcggtcgcgactctagagtcgacggttctGG

TAAGCAAACTTTGAATTTTGACCTTCTTAAGCTTGCGGGAGACGTCGAGTCCAACCCCGGGCCCgcatgcaagcttcagc

tgaagcttaccATGACCGAGTACAAGCCCACGGTGCGCCTCGCCACCCGCGACGACGTCCCCAGGGCCGTACGCACCCTC

GCCGCCGCGTTCGCCGACTACCCCGCCACGCGCCACACCGTCGATCCGGACCGCCACATCGAGCGGGTCACCGAGCTGCA

AGAACTCTTCCTCACGCGCGTCGGGCTCGACATCGGCAAGGTGTGGGTCGCGGACGACGGCGCCGCGGTGGCGGTCTGGA

CCACGCCGGAGAGCGTCGAAGCGGGGGCGGTGTTCGCCGAGATCGGCCCGCGCATGGCCGAGTTGAGCGGTTCCCGGCTG

GCCGCGCAGCAACAGATGGAAGGCCTCCTGGCGCCGCACCGGCCCAAGGAGCCCGCGTGGTTCCTGGCCACCGTCGGCGT

CTCGCCCGACCACCAGGGCAAGGGTCTGGGCAGCGCCGTCGTGCTCCCCGGAGTGGAGGCGGCCGAGCGCGCCGGGGTGC

CCGCCTTCCTGGAGACCTCCGCGCCCCGCAACCTCCCCTTCTACGAGCGGCTCGGCTTCACCGTCACCGCCGACGTCGAG

GTGCCCGAAGGACCGCGCACCTGGTGCATGACCCGCAAGCCCGGTGCCTGA

Fig. 4 eGFP-2A-Puro nucleotide sequence. cDNAs encoding gGFP and the puromycin resistance genes are in
green and blue, respectively, the P2A peptide in red and linker regions in black

target protein and that the guide RNA or PAM sequence has
been modified to prevent cleavage by Cas9. If that is not the
case, introduce point mutations to avoid unwanted mutations
being introduced after integration.
Having precisely defined the donor sequence, different options
can be considered to prepare the donor plasmid.
3. Purchasing the designed sequence from your preferred supplier
is the easiest approach. Alternatively, you can amplify the
eGFP-P2A-PURO coding sequence from an existing plasmid,
the homology arms from genomic DNA, and assemble the
three fragments into your preferred cloning vector using
restriction enzyme-free technologies (see Note 4).
 Tagging Proteins with Fluorescent Reporters Using the CRISPR/Cas9 System. . . 51

3.4 Gene Targeting When the CRISPR/Cas9 reagents are ready, you can proceed to
and Isolation of Gene genome editing with CRISPR/Cas9 to generate cell lines that
Edited Cells by FACS express an endogenously tagged protein. The strategy detailed
Sorting below relies on antibiotic selection combined with fluorescence-
activated cell sorting (FACS) for enrichment in HDR during
genome editing. FACS is an efficient technology which allows to
simultaneously measure the knock-in efficiency by counting fluo-
rescently positive cells and sort isolated cells into 96-well plates. As
cell sorting relies on the expression levels of the fluorescent-tagged
proteins driven by endogenous promoters, isolation of positive
clones can be challenging for proteins expressed at low levels.

3.4.1 Nucleofection 1. Proceed as detailed in Subheading 3.2.1 with a master mix

and Antibiotic Selection containing 2 μg of the selected guide sgRNA plasmid
(sgXPB), 2 μg of plasmid expressing the SpCas9 protein
(JDS246), and 6 μg of donor plasmid in a total volume lower
than 10 μL.
2. Three days after transfection, replace the culture medium and
add puromycin to a concentration of 5 μg/mL and further
incubate for an additional week. The use of puromycin selec-
tion can be avoided if the signal of the fluorescently tagged
protein is sufficiently strong and the proportion of fluorescent
cells is sufficient for FACS sorting.

3.4.2 Isolation of Gene Proceed to FACS sorting 7 days after the beginning of the puro-
Edited Cells by FACS mycin treatment (or 3 days after nucleofection in absence of antibi-
Sorting otic selection). Use non-transfected cells as negative control for
gating.
1. Prepare 96-well plates containing 100 μL of growth medium in
each well.
2. Wash cells with PBS, add 1 mL of trypsin and transfer the
resuspended cells in a 1.5-mL microcentrifuge tube containing
0.5 mL of medium.
3. Centrifuge the cell suspension for 5 min at 90
g at room
temperature, remove the supernatant, resuspend the cells in
1 mL PBS containing 2% FBS, and filter the cells through a
50-μm cell strainer into a sterile flow cytometry tube.
4. Turn on the cytometer, run the cleaning/calibration proce-
dures and load settings to detect GFP. Refer to the user manual
of your instrument for detailed procedures.
5. Run the non-transfected cells to check the cell sorter para-
meters and establish gates to select single live cells (Fig. 5).
For single cells, plot forward scatter linear (FCS) on the y-axis
and the fluorescent channel (GFP-FL) on the x-axis. Ideally, set
a background threshold where less than 0.1% of the
non-transfected control cells are counted as GFP positive.
52 Sylvain Geny et al.

A B

105 105

104 104
FSC

FSC
103 103

102 102

102 103 104 105 102 103 104 105

GFP-FL GFP-FL

Fig. 5 FACS analysis of wild-type and modified U2-OS cells. Non-transfected U2-OS cells (a) and CRISPR/Cas9
modified cells in which a GFP tag is fused to the C-terminus of XPB (b) were analyzed for the expression of a
green fluorescent protein (b). Both the plots show the green fluorescence intensity (GFP-FL) versus the forward
scatter (FSC). U2-OS wild-type cells served as a negative control (blue dots) and define background signal;
gates for the GFP-positive cells (green dots) were set accordingly. The GFP-positive cells were sorted into
96-well plates to obtain single clones expressing XPB-eGFP

6. Repeat with the cell population to be sorted and verify that the
CRISPR/Cas9-modified cells can be differentiated from
parental cells. Run the transfected cells to quantify the amount
of fluorescently labeled cells and define the collection gate. The
fluorescent signal depends on expression levels of the target
gene and can be relatively dim (Fig. 6b).
7. Sort the positive cells into the 96-well plates and incubate at
37  C, 5% CO2.
8. After ~5 days, identify 24 single-cell clones for validation and
keep them growing. Maintain the cells in 96-well plates until
you can proceed to validation. Split cells by trypsinization when
required.

3.5 Validation This section details: (1) genotyping of the knock-in cell clones by
of Selected Clones PCR to verify that the fluorescent marker was correctly inserted at
the target locus and to test for homozygosity, (2) Western blot
analysis to verify that the GFP fusion protein is effectively
expressed, and (3) fluorescence microscopy to validate expression
and localization of the tagged protein.

3.5.1 Validation 1. Split cells from each positive well into two 96-well plates. The
at the Genomic Level day after, spin down and wash the cells in one of the plates
with PBS.
 Tagging Proteins with Fluorescent Reporters Using the CRISPR/Cas9 System. . . 53

A B

FP
-G
PB
:X
OS
OS

U2
U2
250

130 XPB-GFP

70 XPB

55

35

25

1 2

Fig. 6 Validation of engineered cell lines. (a) Western blot analysis of engineered
U2-OS cells in which wild-type XPB was replaced by C-terminal GFP-tagged
version (U2-OSXPB::GFP, lane 2). Unmodified U2-OS cells were used as control
(U2-OS, lane 1) (Adapted from Sandoz et al. 2019). (b) GFP fluorescence of living
U2-OSXPB::GFP cells observed using an inverted microscope equipped with a 63

immersion objective

2. Extract genomic DNA directly in the wells using 25 μL of

commercially available DNA lysis buffer.
3. Proceed to PCR amplification from genomic DNA of each
clone with the primer pairs XPB-FW/XPB-RV or XPB-FW/
GFP-RV following the conditions used for PCR analysis of the
pool of transfected cells. Run 2 μL of the PCR on 1% agarose
TBE 1
gel at 100 V for 1 h. Visualize the PCR products using
a UV transilluminator (see Note 5).
4. Sequence the PCR fragments of clones to ensure correct tag
sequence insertion.

3.5.2 Western Blot 1. Seed five 150-mm Petri dishes containing 25 mL of growth
Analysis medium and incubate at 37  C, 5% CO2 until a confluency of
80% is reached.
2. Wash the plate with PBS containing 30% w/v glycerol, detach
the cells with a scraper, and after centrifugation at 1000
g for
10 min at 4  C, snap freeze the pellet in liquid nitrogen and
store at 80  C. We usually prepare batches of 15
106 and
25
106 cells.
3. Resuspend 15
106 cells in 150 μL of RIPA buffer and
incubate for 10 min by pipetting up and down.
54 Sylvain Geny et al.

4. Sonicate two times for 30 s on ice. We use a Bioruptor™

sonication system (amplitude 30 and 0.5 s pulse on ice)
(optional).
5. Centrifuge for 15 min at 14,000
g at 4  C, collect the
supernatant, and estimate the total protein concentration
using a Bradford assay. A concentration of 3–4 mg/mL is
expected.
6. Heat 20 μg of total proteins from the soluble extract mixed
with 2
Laemmli buffer at 95  C for 5 min and centrifuge
samples at 10,000
g for 30 s to bring down the condensate
and remove insoluble debris.
7. Load the centrifuged sample on an SDS polyacrylamide gel,
electrophorese, and transfer proteins from the gel matrix to a
nitrocellulose or PVDF membrane using your favorite device.
8. Block the membrane for 1 h at room temperature or overnight
at 4  C using 3% w/v dry skimmed milk or BSA solution in
PBS, incubate the membrane with an appropriate dilution of a
primary antibody directed against the affinity tag or the tagged
subunit in the same buffer for 1 h at room temperature or
overnight at 4  C.
9. Wash the membrane three times in TBST for 5 min each,
incubate with the recommended dilution of conjugated sec-
ondary antibody in TBST at room temperature for 1 h and
develop the Western blot.
The Western blot analysis of protein lysates from the parental
and from a modified cell line where XPB and XPB-GFP are detected
with an anti-XPB antibody is shown in Fig. 6a. As expected in the
case of a homozygous modification, the 90 kDa wild-type XPB
protein (lane 1) is replaced by the 130 kDa XPB-GFP fusion (lane
2). A heterozygous modification would result in the detection of
both a 90 kDa and a 130 kDa protein.

3.5.3 Fluorescence We typically select four positive clones and include the
Microscopy non-modified cell line as control. The target protein being fused
to GFP, expression of a fluorescent fusion protein can be checked
using an epifluorescence inverted microscope. Live or fixed cell
imaging can be performed.
1. Seed glass-bottom dishes/plates with clones that have been
tested positive until 70–80% confluency is reached.
2. For live cell imaging, change the medium to a phenol red-free
medium and observe cells using an inverted epifluorescence
microscope with adapted filters (i.e., an FITC filter for GFP
detection). Maintain cells at 37  C in the incubation chamber
of the microscope. Record images with a 63
/1.4 NA
 Tagging Proteins with Fluorescent Reporters Using the CRISPR/Cas9 System. . . 55

immersion objective. Cells can be counter-stained with

Hoechst (0.1 μg/mL) to observe subcellular localization of
the fusion protein.
3. For imaging of fixed cells, wash the cells with PBS, fix them
with 4% PFA in PBS and again twice with PBS. Record images
with a 63
/1.4 NA immersion objective. Although fixed cells
can be stored for a few weeks at 4  C, try to observe them
1–2 days after fixation. Cells can be counter-stained with anti-
bodies to confirm expression of the target protein or for
co-localization experiments.
A live cell image of U2-OS::XPB-GFP cells in which the XPB gene
is fused to eGFP is shown in Fig. 6b.

4 Notes

1. Detailed information is available on the ATCC website

(https://www.lgcstandards-atcc.org/). A guide containing
general technical information for working with animal cells in
culture, including media, sub-culturing, cryopreservation, and
thawing is also accessible.
2. Affinity tags can be introduced at the N- or C- end of the target
protein or inserted into internal loops. It is wise to perform
extensive literature search for functionality studies and expres-
sion data in human cells as well as in other species such as
mouse, C. elegans, or yeast to avoid trying to tag an extremity
that would compromise protein function. In the absence of
information on the accessibility of the extremities of the target
protein, a C-terminal tag is generally preferred [28].
3. Nucleofection relies on permeabilization of the cellular mem-
branes by electroporation. Although Amaxa nucleofection is
often superior to other widely used transfection approaches,
lipid-based transfection with reagents such as Lipofectamine
3000™ or jet-CRISPR™ can also be used.
4. Homology-based cloning techniques including sequence and
ligation–independent cloning (SLIC) and commercially avail-
able technologies such as In-Fusion and Gibson assembly are
perfectly adapted to assemble three to four DNA fragments in a
single step reaction. These technologies are now routinely used
in many laboratories (see ref. [29] in this issue) and can be used
to rapidly generate the donor plasmid. The homology arms can
be amplified from genomic DNA and the insert from plasmids
which can be obtained from non-profit organizations (e.g., see
Addgene plasmids #112848, #52379). Note that a growing
number of companies also offer donor plasmid design and
preparation as a service.
56 Sylvain Geny et al.

5. With the XPB-FW2/XPB-RV2 primers which hybridize on

either side of the insertion site, both knock-in and
non-modified alleles should be amplified: A 2800 bp PCR
product is expected if the eGFP-P2A-PURO was successfully
inserted while a 1300 bp fragment should be observed for a
non-modified allele, allowing to discriminate between homo-
zygous, heterozygous, and non-modified clones. With the
XPB-FW/GFP-RV and Puro-FW/XPB-RV2 pairs which
amplify 540 bp and 1000 bp fragments only modified alleles
will be detected as the GFP-RV and Puro-RV2 oligonucleo-
tides are complementary to the eGFP-P2A-PURO cDNA from
the inserted sequence.

Acknowledgments

We thank IGBMC cell culture and imaging facilities for assistance

and fruitful discussions. This work was supported by the Centre
National pour la Recherche Scientifique (CNRS), the Institut
National de la Santé et de la Recherche Médicale (INSERM),
Université de Strasbourg (UdS), Association pour la Recherche
sur le Cancer (ARC), the Ligue nationale contre le cancer, Institut
National du Cancer (INCa; INCA 9378), Agence National pour la
Recherche (ANR-12-BSV8-0015-01 and ANR-10-LABX-0030-
INRT under the program Investissements d’Avenir ANR-10-
IDEX-0002-02), Instruct (R&D Project Funding), and Instruct-
ULTRA as part of the European Union’s Horizon 2020 (Grant ID
731005) by Instruct-ERIC.

References
1. Villicana C, Cruz G, Zurita M (2014) The 6. Dalvai M, Loehr J, Jacquet K, Huard CC,
basal transcription machinery as a target for Roques C, Herst P, Cote J, Doyon Y (2015)
cancer therapy. Cancer Cell Int 14(1):18 A scalable genome-editing-based approach for
2. Scott DE, Bayly AR, Abell C, Skidmore J mapping multiprotein complexes in human
(2016) Small molecules, big targets: drug dis- cells. Cell Rep 13(3):621–633
covery faces the protein-protein interaction 7. Dewari PS, Southgate B, McCarten K,
challenge. Nat Rev Drug Discov 15 Monogarov G, O’Duibhir E, Quinn N,
(8):533–550 Tyrer A, Leitner MC, Plumb C,
3. Bushweller JH (2019) Targeting transcription Kalantzaki M, Blin C, Finch R, Bressan RB,
factors in cancer—from undruggable to reality. Morrison G, Jacobi AM, Behlke MA, von
Nat Rev Cancer 19(11):611–624 Kriegsheim A, Tomlinson S, Krijgsveld J, Pol-
4. Doudna JA, Charpentier E (2014) Genome lard SM (2018) An efficient and scalable pipe-
editing. The new frontier of genome engineer- line for epitope tagging in mammalian stem
ing with CRISPR-Cas9. Science 346 cells using Cas9 ribonucleoprotein. elife 7
(6213):1258096 8. Geny S, Pichard S, Poterszman A, Concordet J
5. Jinek M, Chylinski K, Fonfara I, Hauer M, (in press) Gene tagging with the CRISPR-Cas9
Doudna JA, Charpentier E (2012) A program- system to facilitate macromolecular complex
mable dual-RNA-guided DNA endonuclease purification. Methods Mol Biol
in adaptive bacterial immunity. Science 337
(6096):816–821
 Tagging Proteins with Fluorescent Reporters Using the CRISPR/Cas9 System. . . 57

9. Compe E, Egly JM (2016) Nucleotide excision cells with zinc finger nucleases constructed via
repair and transcriptional regulation: TFIIH modular assembly. Genome Res 19
and beyond. Annu Rev Biochem 85:265–290 (7):1279–1288
10. Kolesnikova O, Radu L, Poterszman A (2019) 20. Guschin DY, Waite AJ, Katibah GE, Miller JC,
TFIIH: a multi-subunit complex at the cross- Holmes MC, Rebar EJ (2010) A rapid and
roads of transcription and DNA repair. Adv general assay for monitoring endogenous
Protein Chem Struct Biol 115:21–67 gene modification. Methods Mol Biol
11. Sandoz J, Nagy Z, Catez P, Caliskan G, Geny S, 649:247–256
Renaud JB, Concordet JP, Poterszman A, 21. Brinkman EK, Chen T, Amendola M, van
Tora L, Egly JM, Le May N, Coin F (2019) Steensel B (2014) Easy quantitative assessment
Functional interplay between TFIIH and of genome editing by sequence trace decompo-
KAT2A regulates higher-order chromatin sition. Nucleic Acids Res 42(22):e168
structure and class II gene expression. Nat 22. Paix A, Rasoloson D, Folkmann A, Seydoux G
Commun 10(1):1288 (2019) Rapid tagging of human proteins with
12. Yeh CD, Richardson CD, Corn JE (2019) fluorescent reporters by genome engineering
Advances in genome editing through control using double-stranded DNA donors. Curr Pro-
of DNA repair pathways. Nat Cell Biol 21 toc Mol Biol 129(1):e102
(12):1468–1478 23. Paix A, Schmidt H, Seydoux G (2016) Cas9-
13. Chu VT, Weber T, Wefers B, Wurst W, assisted recombineering in C. elegans: genome
Sander S, Rajewsky K, Kuhn R (2015) Increas- editing using in vivo assembly of linear DNAs.
ing the efficiency of homology-directed repair Nucleic Acids Res 44(15):e128
for CRISPR-Cas9-induced precise gene edit- 24. Auer TO, Del Bene F (2014) CRISPR/Cas9
ing in mammalian cells. Nat Biotechnol 33 and TALEN-mediated knock-in approaches in
(5):543–548 zebrafish. Methods 69(2):142–150
14. Sluch VM, Chamling X, Wenger C, Duan Y, 25. Liu Z, Chen O, Wall JBJ, Zheng M, Zhou Y,
Rice DS, Zack DJ (2018) Highly efficient scar- Wang L, Ruth Vaseghi H, Qian L, Liu J (2017)
less knock-in of reporter genes into human and Systematic comparison of 2A peptides for clon-
mouse pluripotent stem cells via transient anti- ing multi-genes in a polycistronic vector. Sci
biotic selection. PLoS One 13(11):e0201683 Rep 7(1):2193
15. Sentmanat MF, Peters ST, Florian CP, Con- 26. Chen X, Zaro JL, Shen WC (2013) Fusion
nelly JP, Pruett-Miller SM (2018) A survey of protein linkers: property, design and function-
validation strategies for CRISPR-Cas9 editing. ality. Adv Drug Deliv Rev 65(10):1357–1369
Sci Rep 8(1):888 27. Koch B, Nijmeijer B, Kueblbeck M, Cai Y,
16. Vouillot L, Thelie A, Pollet N (2015) Compar- Walther N, Ellenberg J (2018) Generation
ison of T7E1 and surveyor mismatch cleavage and validation of homozygous fluorescent
assays to detect mutations triggered by engi- knock-in cells using CRISPR-Cas9 genome
neered nucleases. G3 (Bethesda) 5 editing. Nat Protoc 13(6):1465–1487
(3):407–415 28. Kimple ME, Brill AL, Pasker RL (2013) Over-
17. Renaud JB, Boix C, Charpentier M, De Cian A, view of affinity tags for protein purification.
Cochennec J, Duvernois-Berthet E, Curr Protoc Protein Sci 73:19–23
Perrouault L, Tesson L, Edouard J, 29. Rossolillo P, Kolesnikova O, Essabri K, Ramon
Thinard R, Cherifi Y, Menoret S, Zamorano G, Poterszman A (in press) Produc-
Fontaniere S, de Croze N, Fraichard A, tion of multiprotein complexes using the Bacu-
Sohm F, Anegon I, Concordet JP, Giovannan- lovirus expression system: homology based and
geli C (2016) Improved genome editing effi- Restriction Free cloning strategies for con-
ciency and flexibility using modified struct design. Methods Mol Biol
oligonucleotides with TALEN and CRISPR-
Cas9 nucleases. Cell Rep 14(9):2263–2272 30. Anders C, Niewoehner O, Duerst A, Jinek M
(2014) Structural basis of PAM-dependent tar-
18. Shan Q, Wang Y, Li J, Gao C (2014) Genome get DNA recognition by the Cas9 endonucle-
editing in rice and wheat using the CRISPR/ ase. Nature 513(7519):569–573
Cas system. Nat Protoc 9(10):2395–2410
19. Kim HJ, Lee HJ, Kim H, Cho SW, Kim JS
(2009) Targeted genome editing in human
 Chapter 4

Validation of the Production of Antibodies in Different

Formats in the HEK 293 Transient Gene Expression System
Jens König, Michael Hust, and Joop van den Heuvel

Abstract
Mammalian cells are the most commonly used production system for therapeutic antibodies. Protocols for
the expression of recombinant antibodies in HEK293-6E cells in different antibody formats are described in
detail. As model, antibodies against Kallikrein-related peptidase 7 (KLK7) were used. KLK7 is a key player
in skin homeostasis and represents an emerging target for pharmacological interventions. Potent inhibitors
can not only help to elucidate physiological and pathophysiological functions but also serve as a new
archetype for the treatment of inflammatory skin disorders. Phage display-derived affinity-matured human
anti-KLK7 antibodies were converted to scFv-Fc, IgG, and Fab formats and transiently produced in the
mammalian HEK293-6E system. For the production of the corresponding antigen—KLK7—the baculo-
virus expression vector system (BEVS) and virus-free expression in Hi5 insect cells were used in a
comparative approach. The target proteins were isolated by various chromatographic methods in a one-
or multistep purification strategy. Ultimately, the interaction between anti-KLK7 and KLK7 was character-
ized using biolayer interferometry. Here, protocols for the expression of recombinant antibodies in
different formats are presented and compared for their specific features. Furthermore, biolayer interferom-
etry (BLI), a fast and high-throughput biophysical analytical technique to evaluate the kinetic binding
constant and affinity constant of the different anti-KLK7 antibody formats against Kallikrein-related
peptidase 7 is presented.

Key words Transient gene expression, Recombinant antibody formats, scFv, scFv-Fc, Fab, IgG,
Biolayer interferometry, Kallikrein-related protease 7

1 Introduction

1.1 Aim of the Study The functional and structural analyses of mammalian protein and
protein complexes require simple, robust, and efficient protein
production systems. Virus-free transient gene expression in insect
cell lines and mammalian cell lines as well as the baculoviral expres-
sion vector system allows the purification of ample amounts of
high-quality protein [1, 2]. Furthermore, adequate biochemical
and biophysical methods are essential for the kinetic analysis of
the binding affinity. Biolayer interferometry is a simple method

Arnaud Poterszman (ed.), Multiprotein Complexes: Methods and Protocols, Methods in Molecular Biology, vol. 2247,
https://doi.org/10.1007/978-1-0716-1126-5_4, © Springer Science+Business Media, LLC, part of Springer Nature 2021

59
60 Jens König et al.

which can be used in high throughput to determine the kinetic

binding constant of interacting proteins [3].
Here, we describe the expression and binding studies of the
Kallikrein-related protease 7 (KLK7) with improved inhibitory
anti-KLK7 antibodies. The functional integrity of the skin epider-
mis is highly dependent on cutaneous protease activity. The human
kallikrein-related peptidases (KLKs) are key players in skin homeo-
stasis [4] and therefore important pharmacological therapeutic
targets [5].
In our approach, genetically engineered antibodies are investi-
gated for the potential as KLK inhibitors [6]. Biologics like recom-
binant antibodies have high specificity. Inhibitory antibodies have
broad application and high acceptance in the community and
industry as a therapeutic tool. Efficient screening for appropriate
recombinant antibodies involves the small single-chain variable
fragment format (scFv), which can be isolated using the in vitro
phage display technology [7, 8].
For therapeutic use, however, the IgG molecule is still the most
widely accepted format. It allows for bivalent binding and immune
effector functions, along with a generally high stability, long serum
half-life, and low immunogenicity [9–11]. Accordingly, fragments
are routinely converted to IgG format after completed scFv selec-
tion and engineering. Nevertheless, there are other recombinant
formats that should be considered depending on the application.
Previous studies have shown that the conversion into the IgG
format can affect antigen binding both positively and negatively
[12, 13]. So, a structural compromise like the IgG-like scFv-Fc
format might better preserve the binding properties of the selected
scFv fragment. Due to their smaller size, recombinant antigen-
binding fragments (like Fabs) of antibodies generally have better
tissue penetration, which may be particularly beneficial in applica-
tions for passing the skin barrier [14].
There are a variety of expression systems available today to
express the different antibody formats. Each has its own respective
advantages in terms of cost, ease of use, and the post-translational
modification profiles [15].

1.2 Expression Below, the set of expression vectors to generate different antibody
of Different formats and purification recombinant proteins expressed in
Recombinant Antibody HEK293-6E cells is described. The HEK293-6E cell line is espe-
Formats cially qualified for fast and simple protein production as this cell line
is compatible with the plasmid-based expression using polyethyle-
nimine (PEI) [1, 2, 16]. The VH and VL coding fragments are
cloned in specific vector pairs to be able to generate scFv, homo-
dimeric scFv-Fc, heterodimeric Fab, or tetrameric IgG antibodies
[9–11] (see Fig. 1)
 Expression of Recombinant Antibodies in Different Formats 61

Fig. 1 Recombinant antibody formats compared to IgG. IgGs are large heterotetrameric molecules and depend
on homodimerization of the Fc region (fragment crystallizable) of two identical heavy chains (HCs) and the
subsequent assembly of two identical light chains (LCs) via disulfide linkages. The modular structure of
multiple immunoglobulin domains allows for recombinant antibody formats with preserved antigen binding in
a less complex dimeric (Fab, scFv-Fc) or single-chain structure (scFv). Red and grey: heavy chain; blue: light
chain; orange: artificial peptide linker

1.3 Analysis Binding affinities of antibodies to their target protein (antigen) can
of the Binding Kinetics be determined by the dip and read biosensor technology of the
of the Antibodies OctetRed System. This technology uses biolayer interferometry
and KLK7 Using (BLI) to measure protein–protein interactions [3]. The data are
Biolayer generated in real time, allowing kinetic analysis as well as quantita-
Interferometry (BLI) tive analysis of bound protein. Here, we describe the application to
determine the binding affinity KD of immobilized biotinylated-
KLK7 to streptavidin sensors to the different antibody formats.
After initial equilibration of the biosensor, the ligand (KLK7) is
tightly bound to the tip. Binding of protein results in a change of
the thickness of the biolayer at the tip of the biosensor. The increase
in thickness of the biolayer results in a change of the interference
between two reflected beams from the internal reference layer and
the surface of the tip used for specific binding to the protein
samples. The binding of protein is detected and reported as a
change in wavelength. In this way, the association and dissociation
of the antibody to the immobilized KLK7 can be followed in time
for a range of concentrations using parallel tips. The kinetic con-
stants can be calculated from these curves presuming a 1:1 interac-
tion stoichiometry, which allows proper fitting of the curves.

2 Materials

In the first part, we present an overview of the vectors used, to

enable the expression of the scFv fragment derived from the phage
display analysis into different antibody formats. Furthermore, we
62 Jens König et al.

describe the necessary materials needed for fast and efficient protein
expression of the materials required for kinetic analysis of the
binding of the different antibody format to the KLK7 target pro-
tein by biolayer interferometry (BLI).

2.1 Expression All cloning steps were performed using standard restriction enzyme
Vectors digestion with subsequent ligation. Alternatively, the sequence-
and ligation-independent cloning (SLIC) technique according to
Li et al. [17] was used to generate the expression clones. DNA
preparation for transient gene expression was done using the Pure-
Yield Plasmid Midiprep kit (Promega) or the endotoxin-free Plas-
mid DNA Gigaprep kit (Qiagen). The construction of the cloning
vectors is described in detail in J€ager et al. [16].
1. pCSE2.6-hIgG1-Fc-LUP37-A10 (Expression vector for the
synthesis of the antibody LUP37-A10 in the scFv-Fc format).
2. pCSE2.6-hIgG1-Fc-LUP37-B10 (Expression vector for the syn-
thesis of the antibody LUP37-B10 in the scFv-Fc format).
3. pCSE2.6-hIgG1-Fc-LUP37-C11 (Expression vector for the syn-
thesis of the antibody LUP37-C11 in the scFv-Fc format).
4. pCSE2.6-hIgG1-Fc-LUP37-D11 (Expression vector for the
synthesis of the antibody LUP37-D11 in the scFv-Fc format).
5. pCSL3l (Expression vector for the synthesis of antibody light
chains. VL domains were cloned into this vector for production
of IgG and Fab antibodies).
6. pCSEH1c (Expression vector for the synthesis of antibody
heavy chains. VH domains were cloned into this vector for
production of IgG antibodies).
7. pCSE2.5-Fab-h-HIS-XP (Expression vector for the synthesis of
antibody heavy chains lacking the domains CH2 and CH3. VH
domains were cloned into this vector for the production of
antibodies in the Fab format).
8. pUC57-His8-proKLK-3C-hKLK7 (Cloning vector obtained
from GenScript carrying the synthetic codon optimized DNA
sequence His8-proKLK-3C-hKLK7).
9. pOpIE2- His8-proKLK-3C-hKLK7 (The sE2-dHVR1
sequence of pOpIE2-sE2-dHVR1 was exchanged with the
insert His8-proKLK-3C-hKLK7 to get a Hi5 expression vector
for human proKLK7).
10. pTTo/eGFPq (Control vector for the determination of trans-
fection efficiency in HEK293-6E cells via flow cytometry).
11. pOpiE2-eGFP-HA (Control vector for the determination of
transfection efficiency in Hi5 cells via flow cytometry).
 Expression of Recombinant Antibodies in Different Formats 63

Fig. 2 Vector cloning for transient scFv-Fc, IgG, and Fab expression in HEK293-6E. (a). Schematic overview of
the cloning strategies for the generation of pCSL3l-LUP37-A10-VL, pCSL3l-LUP37-B10-VL, pCSL3l-LUP37-
C11-VL, pCSL3l-LUP37-D11-VL, pCSEH1c-LUP37-All-VH, and pCSE2.5-Fab-h-HIS-LUP37-All-VH (“All” in the
plasmid names means that all four different LUP isolates (A10, B10, C11 and D11) were cloned in the same
way) starting from the corresponding pCSE2.6-hIgG1-Fc-LUP37 vectors. Shown are the expression cassettes
consisting of signal peptide (green), Ig domains, and artificial peptide linker (orange), as well as the particular
restriction sites used for cloning. (b) Analysis of colony-PCR products via agarose gel electrophoresis. Black
arrows indicate the expected lengths of positive amplificates

The transfer of the scFv gene fragment into the vector

pCSE2.6-hIgG1-Fc-XP [16, 18] can be done by simple restriction
digestion and ligation using the isolated scFv phage DNA from the
library. However, for the generation of the other formats, the VH
and VL DNA fragments have to be individually isolated by PCR
amplification with the addition of new compatible cloning sites.
The successful cloning is shown in Fig. 2.

2.2 Cell Lines 1. The HEK293-6E cell line was licensed from National Research
for Protein Production Council (NRC), Biotechnological Research Institute (BRI),
Montreal, Canada.
2. The Trichoplusia ni Hi5 insect cell line (BTI-Tn-5B1-4) was
isolated by the Boyce Thompson Institute for Plant Research,
Ithaca, USA.
3. The Spodoptera frugiperda Sf21 (DSMZ #ACC119) insect cell
line (IPLB-Sf21-AE) was used for amplification of the
baculovirus.
64 Jens König et al.

2.3 Cell Culture 1. 125 mL up to 1 L polycarbonate shake flasks (Corning).

Conditions 2. Climo-shaker ISF1-XC CO2 orbital shaker with 50 mm orbit
(Kühner).
3. Complete cultivation medium for HEK293-6E: FreeStyle F17
(Gibco, Life Technologies/Thermo Scientific) supplemented
with 7.5 mM glutamine, 0.1% Pluronic-F68, and 25 μg/mL
G418. Prepare by adding the following components to 1 L
fresh F17 medium: 40 mL of 200 mM stock solution of L-
glutamine, 0.5 mL of 50 mg/mL G418-solution, and 10 ml of
10% (w/v) Pluronic-F68 (20 g in 200 mL of water and filter
through 0.2 μm). Mix well and store at 4
C.
4. Complete cultivation medium for Hi5 insect cells:
EX-CELL 405.
5. Complete cultivation medium for Sf21 insect cells:
EX-CELL 420.

2.4 Transfection Prepare all solutions using ultrapure water and analytical grade
Reagents reagents. All reagents will be sterile filtered and stored at 4
C
and Additional (unless otherwise mentioned).
Chemicals for Protein 1. 1 mg/mL 25 kDa Polyethylenimine (PEI_25), linear (Poly-
Production sciences #23966): Dissolve 0.05 g of PEI in 50 mL of water
and heat until it is completely dissolved. Store aliquots at 70


C.
2. Use 15-mL polystyrene tubes for preparing DNA/PEI_25-
complexes.
3. 20% (w/v) Tryptone N1 (Organotechnie S.A.S., La Cour-
neuve, France): Weigh 100 g Tryptone N1. Make up to
500 mL with water.
4. 300 g/L Glucose: Weigh 150 g glucose and make up to
500 mL with water.
5. 75 mM Valproic acid (20). Dissolve 1.2 g of valproic acid
sodium salt (MW 155.19) and make up to 100 mL in F17
medium (supplemented).
6. Guava EasyCyte™ Mini (Merck, Millipore).
7. CASY Cell Counter (Innovatis).

2.5 Biolayer Prepare all solutions using ultrapure water and analytical grade
Interferometry reagents. All reagents are sterile filtered and stored at 4
C (unless
otherwise mentioned).
1. 1 mM EZ-Link Sulfo-NHS-LC-LC-Biotin solution (Thermo
Fisher).
2. Zeba spin desalting columns (Thermo Fisher).
3. OctetRed 96 System (Pall, Fortebio).
 Expression of Recombinant Antibodies in Different Formats 65

4. Streptavidin (SA) Dip and Read Biosensors (Pall, Fortebio).

5. 96-well flat bottom polypropylene microplate, black.
6. PBS (1): 138 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4,
2mM KH2PO4, pH7.4).
7. Kinetics buffer (1 PBS, pH 7.4) optionally with 0.1% BSA
and 0.2% Tween 20.
8. 100 mM 10 glycine buffer pH 1.5.

3 Methods

Carry out all procedures at room temperature unless otherwise

specified. Use sterile equipment, flasks, and centrifuge tubes
throughout the experiment.

3.1 Transient Protein The protocol is described in Karste et al. [1]. The employed vectors
Expression contain the OriP to ensure optimal long-term expression levels.
in HEK293-6E Cells Here the expression is described for 500 mL.
1. [Day 3, e.g., Friday] Prepare a 400 mL culture containing 0.3
 106 HEK293-6E c/mL. Incubate the culture for 72 h at
37
C, 110 rpm, and 5% CO2.
2. [Day 0, e.g., Monday] Count the cells of the preparatory culture
and prepare a 250 mL culture containing 1.8  106 HEK293-
6E c/mL by centrifuging the required volume of the cell
suspension at 180  g for 4 min. Discard the supernatant
carefully and resolve the cell pellet in 250 mL fresh F17. Place
the cells back into the incubator (37
C, 90 rpm, and 5% CO2)
until use (see Note 1).
3. Prepare the DNA solution in 12.5 mL F17 in a 50-mL corning
flask (see Note 2). Hereto, mix 237 μg of the target plasmid
(s) containing your gene(s) of interest with 13 μg control
plasmid pTTo-GFPq coding, e.g., for eGFP. Mix by agitating
the tube briefly.
4. Prepare the PEI_25 solution in 12.5 mL F17 in a 50-mL
Corning flask (see Note 2). Hereto, pipette 625 μL PEI of
the 1 mg/mL PEI_25 stock solution into the. Mix by vortex-
ing briefly.
5. Mix the PEI_25 solution with the DNA solution directly (see
Note 3).
6. Incubate the PEI_25–DNA solution for 15 min at RT to
preform PEI–DNA complexes.
7. Pipette the PEI–DNA complexes (25 mL) to the cells and mix
gently. Incubate the cells until harvest at 37
C, 90 rpm, and 5%
CO2.
66 Jens König et al.

8. [Day 1–7, e.g., Tuesday to Monday] Take samples daily and

determine transfection efficiency in the flow cytometer
(Guava EasyCyte) or/and determine target protein expression
by a suitable technique (SDS-PAGE, slot blot, or
Western blot).
9. [Day 2, e.g., Wednesday] Add 250 mL fresh F17 to the cells as
well as 12.5 mL of the TN1-Stock (20%) ¼ 0.5%.
10. [Day 3, e.g., Thursday] Add 7.5 mL of the glucose stock
(300 g/L) ¼ 4.5 g/L.
11. [Day 4, e.g., Friday] Add 25 mL of the valproic acid stock
(75 mM) ¼ 3.75 mmol/L.
12. [Day 7, e.g., Monday] Harvest secreted target proteins by
centrifuging the cell suspension for 15 min at 1000  g and
store the supernatant at 4
C until purification. Depending on
the protein, it might be required to add protease inhibitor. For
intracellular proteins, snap freeze the cell pellet and store at
80
C until cell lysis and purification.
The cloned expression vectors were transfected single or in
pairs into HEK293-6E cells, and the culture was grown for
168 h. The viability of the cells remained >80%, and the efficiency
of transfection varied from 20% to 80% for the different experi-
ments (see Fig. 3). All the supernatants contained sufficient anti-
body for subsequent purification of the protein samples. The
quality of the purified antibodies was validated by SDS-PAGE.
Total volumetric yield and specific cellular yield of the recombinant
antibodies are presented in Fig. 4.

3.2 Transient Protein The protocol is described in Karste et al. MiMB [1]. Here, we show
Expression in Hi5 Cells the expression and purification of the secreted KLK7 in Hi5 cells in
500 mL scale. The employed vectors contain the optimal expres-
sion cassette with the strong OpIE2 promoter (see Note 4, [20]).
1. [Day 1, e.g., Friday] Prepare a 150 mL preculture starting at
0.3–0.4  106 Hi5 c/mL. Incubate the culture for 72 h at
27
C and 110 rpm (see Note 5).
2. [Day 0, e.g., Monday] Count the cells of the preparatory culture
and prepare a 100 mL culture containing 5  106 Hi5 c/mL by
centrifuging the required volume of the cell suspension at
180  g for 4 min. Discard the supernatant and resolve the
cell pellet in 100 mL fresh EX-CELL 405.
3. Incubate the culture at 27
C and 110 rpm for 1 h.
4. Mix 425 μg of the expression plasmid and 25 μg of the
control plasmid, e.g., pOpIE2-eGFP-HA, coding for eGFP
(see Note 6).
 Expression of Recombinant Antibodies in Different Formats 67

Fig. 3 Transient antibody production in HEK293-6E cells. (a) Schematic vector map of the mammalian
HEK293-6E expression vector with exemplary expression cassette (found in pTT5 or pFlpBtMII [2]). (b)
Schematic illustration of corresponding gene expression cassettes used for the production of recombinant
antibodies in the scFv-Fc, Fab, and IgG format. (c/d) Monitoring of growth, viability, and transfection efficiency
(GFP-positive cells) over time. Volume doubling and supplementations at 48, 72, and 96 h post-transfection
are indicated by black arrows. Mean values are presented

5. Pipette the DNA mix directly to the prepared cells and mix
gently (see Note 7).
6. Immediately pipette 4 mL PEI of the 1 mg/mL PEI stock
solution to the cells and mix gently.
7. Incubate the culture at 27
C and 110 rpm for 3 up to 5 h.
8. Add 400 mL fresh EX-CELL 405 (see Note 8).
9. [Day 1–3, Tuesday to Thursday] Take samples daily, monitor
cell numbers and viability, and determine transfection efficiency
in the cytometer. If cells reach a density above 3  106 c/mL,
adjust the concentration back to 2  106 c/mL by adding fresh
culture medium.
10. [Day 3, Thursday] Harvest secreted target proteins by centrifu-
ging the cell suspension 15 min at 1000  g and store the
supernatant at 4
C after sterile filtration (0.45 μm) until
purification.
68 Jens König et al.

Fig. 4 Yields of antibody production in HEK293-6E cells. (a) Stained protein gel of purified scFv-Fc, IgG, and
Fab antibodies. Samples of 3.5 μg/lane were separated for 50 min on a 12% SDS-PAGE gel at 160 V and
analyzed via Coomassie staining. (b, c) Column charts of volumetric and specific cellular recombinant protein
yields with SEM error bars. Yield was calculated on the basis of the total amount of purified protein and final
production volume. For the calculation of the specific cellular yield, maximal cell number and cultivation time
were additionally considered
 Expression of Recombinant Antibodies in Different Formats 69

3.3 Baculoviral The protocol describes the expression and purification of the
Expression of hKLK7 secreted KLK7 in Hi5 cells in 500 mL scale using the BEVS
in Hi5 Cells technology (see Note 9).
1. [Day 3, Friday] Prepare a 250 mL preculture starting with
0.3–0.4  106 Hi5 c/mL. Incubate the culture for 72 h at
27
C and 110 rpm (see Note 5).
2. [Day 0, e.g., Monday] Count the cells of the preparatory culture
and prepare a 500 mL culture containing 1.0–1.5  106 Hi5
c/mL by diluting the required volume of the cell suspension
into fresh EX-CELL 405 medium. Infect the cells at a MOI of
0.5–2.0 with the baculoviral stock solution (see Note 10). Mix
gently and incubate the culture at 27
C and 110 rpm for 72 h.
3. [Day 1–3, Tuesday to Thursday] Take samples daily, monitor
cell numbers, viability, and cell diameter in the CASY Cell
Counter to determine the progress of the baculoviral infection.
If cells reach a density above 3  106 c/mL, adjust the concen-
tration back to 1  106 c/mL by adding fresh EX-CELL
405 culture medium.
4. [Day 3, Thursday] Harvest secreted target proteins 48 h after
total proliferation stop by centrifuging the cell suspension for
15 min at 1000  g and store the supernatant at 4
C after
sterile filtration (0.45 μm) until purification. Prevent microbial
growth by adding 0.05% sodium azide.

3.4 Preparation 1. [Day 3, Friday] Prepare a 150 mL Sf21 preculture starting with
of the Baculoviral 0.3–0.4  106 c/mL in EX-CELL 420 medium. Incubate the
Stock Solution culture for 72 h at 27
C and 110 rpm (see Note 5).
2. [Day 0, e.g., Monday] Count the cells of the preparatory culture
and prepare a 500 mL culture containing 0.5  106 Hi5 c/mL
by diluting the required volume of the cell suspension into
fresh EX-CELL 420 medium. Infect the cells with a MOI of
0.2 with the baculoviral transfection supernatant or use 5% v/v
of a nontitered viral stock. Mix gently and incubate the culture
at 27
C and 110 rpm for 4 days.
3. [Day 1–4, Tuesday to Friday] Monitor the cell number, viabil-
ity, and the cell diameter (CASY cell counter). The cells should
be split to 1  106 c/mL by adding fresh EX-CELL
420 medium once they reach 2  106 c/mL.
4. [Day 4, e.g., Friday] The viral supernatant was harvested
48 hours after an increase in cell diameter of 3 μm was
observed.

3.5 Protein Purification and maintenance of the columns and chromatography

Purification by Affinity systems were done according to the protocols recommended by the
Chromatography manufacturing company.
70 Jens König et al.

1. [HisTrap Excel column chromatography] The His-tagged Fab

antibodies and proKLK7 were isolated using the ÄKTA start
system from (GE Healthcare) using HisTrap Excel columns for
direct loading of cell culture supernatants. The pre-filtered
(0.45 μm) supernatants were loaded onto a 1 mL HisTrap
Excel column equilibrated with buffer A (20 mM sodium
phosphate buffer, pH 7.4, 500 mM NaCl). Washing was
done using buffer A containing 10–30 mM imidazole. The
protein was eluted using buffer A containing 250 mM imidaz-
ole. Eluted fractions were analyzed by SDS-PAGE electropho-
resis. Protein-containing fractions were pooled and dialyzed
against PBS using Spectra/Por dialysis tubing (Spectrum,
MWCO 6-8000) to remove imidazole from the sample. The
purification of KLK7 from virus-free transient gene expression
in Hi5 resulted in higher amounts and higher purity of recom-
binant protein compared to the BEVS expressed KLK7. (see
Fig. 5).
2. [Protein A column chromatography] The semi-automated Pro-
finia Protein purification system was used for the purification of
the Fc-tag containing recombinant antibody formats (IgG,
scFv-Fc) using the customized “Protein A and Desalting”

Fig. 5 Affinity chromatography. The cell culture supernatants were filtrated (0.45 m) and loaded onto the
HisTrap Excel column. (a, b) Comparison of the chromatograms of KLK7 affinity purification of the production
via BEVS and virus-free expression in Hi5 cells (c, d, respectively). Peak fractions were detected by
SDS-PAGE. 20 μL samples of the collected 1 mL fractions were applied per lane of a 12% SDS-gel and
separated for 50 min at 160 V. Black arrows indicate the bands resembling pro-KLK7. The imidazole gradient
is shown on the right axis in % B (100% B corresponds to 500 mM Imidazole)
 Expression of Recombinant Antibodies in Different Formats 71

method. The pre-filtered (0.45 μm) supernatants were loaded

in manual mode (for volumes >50 mL) onto a 1 mL
pre-packed protein A cartridge. Following washing, the anti-
bodies were eluted by pH shift and directly loaded onto a
10 mL P6 cartridge for desalting. The protein fraction was
stored at 4
C for further analysis

3.6 Biolayer Protocols were set up using the Acquisition and Analysis User
Interferometry Guide for the OctetRed 96 (Fortebio, Pall) and according to the
methods described in detail in [19].
1. Purified pro-KLK7 samples were activated for 16 h at 4
C
using 0.02 U/g bovine Enterokinase (Sigma).
2. The activated KLK7 was biotinylated using 1 mM EZ-link
Sulfo-NHS-LC-LC-Biotin (Thermo Fisher). The protein was
cross-linked in at molar ratio of 3:1 (linker: KLK7) for 30 min
at room temperature.
3. Non-reacted biotinylating reagent was removed by gel filtra-
tion in a Zeba spin desalting column (Thermo Fisher) accord-
ing to the manufacturer’s protocol. The biotinylated KLK7 was
diluted to 12.5 μg/mL. Antibodies were diluted in a range of
concentration from 1 to 500 nM. The sample plate was
prepared as described in Fig 6. Each well was filled with
200 μL of buffer or sample.
4. Prepare a plate with eight streptavidin biosensors in row 1 and
preincubate the tips in 200 μL PBS for 10 min. Start the
acquisition software (8.2) of the OctetRed 96 and set up and
kinetic experiment with regeneration of the tips according to
the Octet Data Acquisition User Guide. Detailed protocols are
described by Kamaraswamy and Tobias [19].
5. The setup for the biolayer interferometric kinetic measure-
ments is described in detail in Table 1. Preincubated streptavi-
din biosensors were equilibrated for 60 s in PBS before the
biotine-KLK7 was immobilized on the sensor for
300 s. Baseline, association, and dissociation to the antibody
were monitored over time as shown in Table 1. The tips were
subsequently regenerated by a threefold cycle of regeneration
in 10 mM 1 glycine buffer, pH 1.5 followed by neutralization
in PBS. The tips were reused to measure the remaining
KLK7—anti-KLK7 antibody kinetics.
6. The data were processed using the Octet Data Analysis soft-
ware package (8.2). The association step was aligned to the
baseline. The relative signal of antibody binding to the immo-
bilized KLK7 is shown in Fig 7. Make sure that the antibody
does not show nonspecific binding to the tips by monitoring
the signal of a tip which is not charged with biotinylated KLK7.
72 Jens König et al.

Fig. 6 Plate map diagram and principle of KLK7—anti-KLK7 kinetic measurements. Plate map: column 1, 3,
5, 7, 9, 12 (gray ¼ PBS), column 2 (blue ¼ Biotin-KLK7 (12.5 μg/mL)), column 6, 8, 10 (red ¼ antibody dilution
series (1.1-500 nM)), well H4, H6, H8 (orange ¼ reference for subtraction of the background), column
11 (green ¼ glycine buffer for regeneration). Biolayer interferometry principle: Top-emitting light is reflected
at the internal reference layer and the outer reflection surface of the biosensor tip. Binding molecules increase
the optical thickness and cause a shift of the resultant wave due to changes in the wave interference pattern.
This shift in wavelength is measured in real-time

Table 1
Settings for the biolayer interferometric kinetic measurements

Sample plate column

Step Step name Time [s] Flow [rpm] Run 1 Run 2 Run 3
1 Equilibration 60 1000 1 1 1
2 Loading 300 1000 2 2 2
3 Baseline 60 1000 3 3 3
4 Association 600 1000 4 6 8
5 Dissociation 600 1000 5 7 9
6 Regeneration 30 1000 11,12 11,12 11,12
The experiment was repeated after regeneration of the tips using the different antibody formats in column 4, 6, and
8 respectively
 Expression of Recombinant Antibodies in Different Formats 73

Fig. 7 Biolayer interferometry-based affinity ranking for recombinant scFv-Fc, IgG, and Fab antibodies.
Streptavidin Dip and Read tips were loaded with biotinylated KLK7 and assayed for their response to various
anti-KLK7 antibody levels. (a) Nonlinear fit of grouped association and dissociation curves based on a 1:1
binding model for the three recombinant antibody formats of LUP37-C11. (b) Summary of apparent affinity and
rate constants in a grouped column chart with error bars representing the SEM. Antibodies with insufficient fits
(X2: > 3 and R2: < 0.95) are displayed in gray

7. Select parameters for the kinetic analysis and determination of

the affinity constant as described in [19]. After curve-fitting,
the lower concentrations are used for determining the kinetic
and affinity constants (Table 2). The results shown in Table 2
indicate that the apparent affinities of the three scFv-Fc anti-
bodies (A10, C11 and D11) are in a similar middle picomolar
range. Conversion into the IgG format showed similar affinities
for the antibodies, only C11 improved a factor of 19 in affinity.
74 Jens König et al.

Table 2
Summary of antibody yields and apparent affinity to KLK7

Antibody Volume yield (mg/L) KD (nM) kon (1/M·s) koff (1/s)

scFv-Fc-LUP37-A10 7.36 0.304 1.45  10 5
4.39  105
scFv-Fc-LUP37-B10 7.42 5.53 9.84  104 5.44  104
scFv-Fc-LUP37-C11 8.10 0.231 1.32  105 3.05  105
scFv-Fc-LUP37-D11 6.96 0.179 1.34  105 2.39  105
IgG-LUP37-A10 0.40 0.247 1.08  105 2.67  105
IgG-LUP37-B10 0.47 – – –
IgG-LUP37-C11 1.05 0.012 3.68  105 4.43  106
IgG-LUP37-D11 2.02 0.419 6.74  104 2.82  105
Fab-LUP37-A10 1.33 18.8 7.95  104 1.50  103
Fab-LUP37-B10 1.75 – – –
Fab-LUP37-C11 4.55 27.8 1.05  105 2.91  103
Fab-LUP37-D11 4.47 25.3 1.56  105 3.96  103

However, the Fab format substantially reduced the affinity by

two orders of magnitude. On account of the chosen orienta-
tion of the assay, the affinity constant determination can be
affected by avidity effects. See Note 11.

4 Notes

1. Replacing used culture medium for fresh F17 (without antibi-

otic), ensuring a sufficient amount of nutrients to keep the cells
growing until the addition of fresh culture medium at day 2.
2. Using polystyrene tubes instead of polypropylene tubes is
important to ensure that the PEI-DNA complexes do not
stick to the tube what would lead to a lower transfection rate
due to a lower total amount of available complex.
3. Directly mix the DNA with PEI to form stable PEI-DNA
complexes. Be sure that the DNA is not precipitating. Turbid-
ity or visible aggregates indicate that the quality of the DNA
preparation is not sufficient. Make sure the DNA sample is not
contaminated with chromosomal E. coli DNA.
4. The optimal combination in our facility comprises the OpIE2
promoter, the IE1 terminator, and a FlashBac compatible
backbone [20].
 Expression of Recombinant Antibodies in Different Formats 75

5. This step must ensure that the cells are still in the exponential
growth phase, and this will increase the transfection efficiency,
baculoviral infection, and protein expression vastly.
6. Using higher amounts of DNA (up to 3 μg per 1  106 cells)
leads to higher yields if the PEI amount is increased accord-
ingly and the PEI:DNA ratio of 4:1 is kept stable. Further
increasing the amount of DNA will decrease cell viability due
to the required higher PEI concentration, as PEI is known to
be cytotoxic in high concentrations. This will decrease the
overall productivity.
7. Direct transfection without preforming the DNA-PEI com-
plexes led to the highest transfection rates and protein levels
in our lab. In contrast, Shen et al. observed no differences
between direct transfection or transfection after preforming
the DNA-PEI complexes [21]. The difference might be caused
by different performance of other batches of Hi5 cells. How-
ever, a direct transfection is easier and faster. Thus, the
described direct transfection is the optimal method in our
hands.
8. Diluting with fresh media is important to keep the cells in the
exponential growth phase until the time of harvest.
9. Optimized protocols are validated and reviewed in the bench-
marking study by Stolt-Bergner et al. [22].
10. For optimal expression results, the titer of the viral stock solu-
tion should be determined by a plaque assay. Additionally, the
best MOI for expression should be determined in a small-scale
expression analysis prior to scaling up by varying the MOI from
0.2 to 2.0.
11. Avidity effects can be caused by a multivalent molecule, like an
antibody, when the antigen is immobilized. For accurate deter-
mination of the kinetic constants, the orientation of the assay
must also be reverted by using anti-human IgG Fc Capture
(AHC) biosensors for the immobilization of the IgG and
Fc-tagged antibody format using KLK7 as analyte. The Fab
format cannot be analyzed in this way and needs direct chemi-
cal cross-linking to suitable tips.

Acknowledgments

We greatly thank Luciano Puzer from the Universidade Federal do

ABC, Sao Paulo, Brazil for his valuable scientific input and the
biomaterials he made available for this research project. We also
thank Daniela Gebauer, Nadine Konisch, and Anke Samuels for
their excellent technical support during the experiments. This work
was supported by the Helmholtz Protein Sample Production Facil-
ity (PSPF).
76 Jens König et al.

References
1. Karste K, Bleckmann M, van den Heuvel J 13. Wenzel EV, Bosnak M, Tierney R et al (2020)
(2017) Not limited to E. coli: versatile expres- Human antibodies neutralizing diphtheria
sion vectors for mammalian protein expression. toxin in vitro and in vivo. Sci Rep 10:571
Methods Mol Biol 1586:313–324 14. Li Z, Krippendorff B-F, Sharma S et al (2016)
2. Meyer S, Lorenz C, Baser B et al (2013) Multi- Influence of molecular size on tissue distribu-
host expression system for recombinant pro- tion of antibody fragments. mAbs 8:113–119
duction of challenging proteins. PLoS One 8 15. Khan KH (2013) Gene expression in mamma-
(7):e68674 lian cells and its applications. Adv Pharm Bull
3. Kamat V, Rafique A (2017) Designing binding 3:257–263
kinetic assay on the bio-layer interferometry 16. J€ager V, Büssow K, Wagner A et al (2013) High
(BLI) biosensor to characterize antibody- level transient production of recombinant anti-
antigen interactions. Anal Biochem 536:16–31 bodies and antibody fusion proteins in
4. Egawa G, Kabashima K (2018) Barrier dys- HEK293 cells. BMC Biotechnol 13:52
function in the skin allergy. Allergol Int 17. Li MZ, Elledge SJ (2012) SLIC: a method for
67:3–11 sequence- and ligation-independent cloning.
5. Sotiropoulou G, Pampalakis G (2012) Target- Methods Mol Biol 852:51–59
ing the kallikrein-related peptidases for drug 18. Fühner V, Heine PA, Helmsing S et al (2018)
development. Trends Pharmacol Sci Development of neutralizing and
33:623–634 non-neutralizing antibodies targeting known
6. Goettig P, Magdolen V, Brandstetter H (2010) and novel epitopes of TCDB of Clostridioides
Natural and synthetic inhibitors of kallikrein- difficile. Front Microbiol 9:2908
related peptidases (klks). Biochimie 19. Kumaraswamy S, Tobias R (2015) Label-Free
92:1546–1567 kinetic analysis of an antibody-antigen interac-
7. Russo G, Meier D, Helmsing S et al (2018) tion using biolayer interferometry. Methods
Parallelized antibody selection in microtiter Mol Biol 1278:165–182
plates. Methods Mol Biol 1701:273–284 20. Bleckmann M, Schürig M, Chen FF et al
8. Frenzel A, Hust M, Schirrmann T (2016) (2016) Identification of essential genetic bacu-
Phage display-derived antibodies in clinical loviral elements for recombinant protein
development and therapy. mAbs 8:1177–1194 expression by transactivation in Sf21 insect
9. Strohl WR (2018) Current progress in innova- cells. PloS One 11:e0149424
tive engineered antibodies. Prot Cell 9:86–120 21. Shen X, Pitol AK, Bachmann V et al (2015) A
10. Frenzel A, Hust M, Schirrmann T (2013) simple plasmid-based transient gene expression
Expression of recombinant antibodies. Front method using High Five cells. J Biotechnol
Immunol 4:217 216:67–75
11. Ha J-H, Kim J-E, Kim Y-S (2016) Immuno- 22. Stolt-Bergner P, Benda C, Bergbrede T et al
globulin Fc heterodimer platform technology: (2018) Baculovirus-driven protein expression
From design to applications in therapeutic anti- in insect cells: a benchmarking study. J Struct
bodies and proteins. Front Immunol 7:394 Biol 203:71–80
12. Thie H, Toleikis L, Li J et al (2011) Rise and
fall of an anti-MUC1 specific antibody. PLoS
One 6:e15921
 Chapter 5

The Intervening Removable Affinity Tag (iRAT) System

for the Production of Recombinant Antibody Fragments
Norimichi Nomura, Yayoi Nomura, Yumi Sato, and So Iwata

Abstract
Fv and Fab antibody fragments are versatile co-crystallization partners that aid in the structural determina-
tion of otherwise “uncrystallizable” proteins, including human/mammalian membrane proteins. Accessi-
ble methods for the rapid and reliable production of recombinant antibody fragments have been long
sought. In this chapter, we describe the concept and protocols of the intervening removable affinity tag
(iRAT) system for the efficient production of Fv and Fab fragments in milligram quantities, which are
sufficient for structural studies. As an extension of the iRAT system, we also provide a new method for the
creation of genetically encoded fluorescent Fab fragments, which are potentially useful as molecular devices
in various basic biomedical and clinical procedures, such as immunofluorescence cytometry, bioimaging,
and immunodiagnosis.

Key words Recombinant antibody, Polyprotein, Secretory expression, Fv fragment, Fab fragment,
Crystallography, Genetically encoded fluorescent antibody, Therapeutic antibody, Biosimilar,
Immunodiagnostics

1 Introduction

Crystallography of biomedically relevant proteins often remains

stagnant because of challenges in preparing diffraction-quality crys-
tals. In most cases, their molecular flexibility, conformational het-
erogeneity, or polydispersal characteristics in solution hinder
protein crystallization. Antibody-aided crystallography is a power-
ful strategy to manage this problem and can be applied to various
membrane proteins as well as heavily glycosylated and/or multi-
domain soluble proteins [1–7]. Fv and Fab antibody fragments
assist crystallization by increasing the hydrophilic surface area avail-
able for rigid crystal lattice formation. The bound antibody frag-
ments reduce the inherent protein flexibility and conformational
heterogeneity, thereby increasing the chance of successful crystalli-
zation of difficult target proteins. In addition to crystallography,
Fab and Fv fragments have also been proven to be powerful as

Arnaud Poterszman (ed.), Multiprotein Complexes: Methods and Protocols, Methods in Molecular Biology, vol. 2247,
https://doi.org/10.1007/978-1-0716-1126-5_5, © Springer Science+Business Media, LLC, part of Springer Nature 2021

77
78 Norimichi Nomura et al.

fiducial markers for single-particle cryo-electron microscopy (cryo-

EM) to aid in structure determination [8, 9].
The preparation of milligram quantities of purified Fv and Fab
fragments is the most critical prerequisite once the strategies for
antibody-aided structural studies are adopted. Recently, we devel-
oped an efficient production strategy for recombinant Fv antibody
fragments and named it the “iRAT system” [10]. A primary feature
of the iRAT system is the secretory expression of a single synthetic
polyprotein consisting of a variable light (VL) domain, an interven-
ing removable affinity tag (iRAT), and a variable heavy (VH)
domain using the Gram-positive bacterium Brevibacillus choshinen-
sis. This ensures 1:1 stoichiometric expression of the VL and VH
domains from the monocistronic construct, followed by proper
folding and assembly of the two domains, each of which contains
an intrachain disulfide bond. A polyhistidine-maltose binding pro-
tein (His6-MBP) cassette, embedded within the intervening region
of the two variable domains, serves as an affinity purification tag
that can be removed by tobacco etch virus (TEV) protease cleavage
during the purification process to yield tag-free Fv fragments suit-
able for crystallization trials (Fig. 1a). We have validated the iRAT-
based production of multiple Fv fragments, including a crystalliza-
tion chaperone for a mammalian membrane protein as well as

Fig. 1 Concept of the iRAT production system. (a) iRAT-synthetic polyproteins

with the built-in MBP facilitate production of Fvs. (b) iRAT-synthetic polyproteins
with the built-in mCherry and long Gly/Ser-rich tethering linkers facilitate
production of Fabs or Fvs. The resulting polyproteins are recovered from the
culture supernatant of Brevibacillus or Sf9, purified by IMAC and excised with
TEV-His6 (indicated by the “scissors” icon). After the removal of the iRAT portion
and TEV-His6 by reverse IMAC, non-tagged antibody fragments are further
purified by using SEC. The antibody light chain-derived domains are red, iRAT-
MBP is green, iRAT-mCherry is magenta, and the antibody heavy chain-derived
domains are blue. The polyprotein N-terminus, polyprotein C-terminus, and CDR
surface are marked. The TEV cleavage site is indicated by the “cs” in orange
boxes. Notably, iRAT-mCherry polyproteins can be utilized as genetically
encoded fluorescent antibody fragments if not treated with TEV
 Production of Recombinant Antibody Fragments via the iRAT system 79

FDA-approved therapeutic antibodies. The resulting Fv fragments

are functionally active and are crystallized in complex with the
target proteins [4, 10–12].
Another important feature of the iRAT system is its high exten-
sibility. Antibodies in different modalities can be produced via the
original and modified iRAT system. Beyond the production of Fv
fragments, we have succeeded in producing Fab fragments through
a process of reorganization of the iRAT segment. Interestingly,
when a polyhistidine-mCherry (His6-mCherry) cassette is
incorporated within the long intervening region, tethering the
light and heavy chains of Fab fragments, the resultant polyproteins
fold and assemble into functional fluorescent Fab fragments
(Fig. 1b). The iRAT concept could provide promising clues to
design and create antibody-derived artificial molecular devices
that are useful in various basic biomedical and clinical procedures,
such as fluorescence-activated cell sorting (FACS), fluorescence-
linked immunosorbent assay (FLISA), immunohistochemistry
(IHC), fluorescent bioimaging, and immunodiagnostics.
In this chapter, we present reliable and broadly applicable iRAT
protocols to prepare milligram quantities of Fv and Fab antibody
fragments using Brevibacillus and Sf9 insect cells, respectively, with
an average duration. Anticipated results are also shown, taking
examples of the iRAT-mediated production of certolizumab Fv
and Fab. Certolizumab (Cimzia®) is an FDA-approved humanized
monoclonal antibody that targets tumor necrosis factor α (TNFα)
and is widely used to treat rheumatoid arthritis and Crohn’s disease.
Biosimilars of such known therapeutic antibodies for research use
can be easily prepared via the iRAT system. Both Fv production
using Sf9 cells and Fab production using Brevibacillus can be
performed similarly, although the detailed protocols are not
described here due to space limitations.

2 Materials

2.1 Molecular 1. PrimeSTAR Max Premix (2
) (Takara Bio/Clontech).

Biology 2. Gibson Assembly Master Mix (2
) (New England Biolabs).
3. TaKaRa Ex Taq (5 U/μL) (Takara Bio/Clontech).
4. 10
Ex Taq Buffer.
5. dNTPs mixture (2.5 mM each).
6. Wizard SV gel and PCR clean-up system (Promega).
7. MinElute PCR purification kit (Qiagen).
8. QIAquick spin miniprep kit (Qiagen).
9. Thermal cycler.
10. NanoDrop spectrophotometer.
80 Norimichi Nomura et al.

2.2 Protein 1. Brevibacillus choshinensis HPD31-SP3 (Takara Bio/Clontech).

Expression Using 2. TM medium: 10 g/L glucose, 10 g/L polypeptone, 5 g/L
Brevibacillus Cells meat extract, 2 g/L yeast extract, 10 mg/L FeSO47H2O,
10 mg/L MnSO44H2O, 1 mg/L ZnSO47H2O, pH 7.0.
3. MT medium: 10 g/L glucose, 10 g/L polypeptone, 5 g/L
meat extract, 2 g/L yeast extract, 10 mg/L FeSO47H2O,
10 mg/L MnSO44H2O, 1 mg/L ZnSO47H2O, 4.1 g/L
MgCl2, pH 7.0.
4. MT/neomycin agar plate: 10 g/L glucose, 10 g/L polypep-
tone, 5 g/L meat extract, 2 g/L yeast extract, 10 mg/L
FeSO47H2O, 10 mg/L MnSO44H2O, 1 mg/L
ZnSO47H2O, 4.1 g/L MgCl2, 10 mg/L neomycin, 15 g/L
agar, pH 7.0.
5. 2SY/neomycin medium: 20 g/L glucose, 40 g/L Bacto Soy-
tone, 5 g/L yeast extract, 0.15 g/L CaCl22H2O, 50 mg/L
neomycin.
6. PGH buffer: 1 mM HEPES-KOH (pH 7.0), 15% (w/v) glyc-
erol, 15% (w/v) PEG6000.
7. pBIM2 vector (Fig. 2).
8. Synthetic or natural antibody-encoding cDNA of interest (see
Note 1).
9. BIM2-F primer: 50 -ACTAGTGAAAATTTATATTTTCAAGG
TCACCATCACCATCACCATTCCTCTGGTGGCGGT.
10. BIM2-R primer: 50 -AAGCTTCGATTGGAAGTACAGGTTT
TCGGAAGATCTAGAGGAACCACCCCCACCCCCGAG.
11. pNY326-F primer: 50 -GAATTCGGTACCCCGGGTTCGA.
12. pNY326-R primer: 50 -
GGATCCTGCAGCGAAAGCCATGGGAGC.
13. BacillusExp-F primer: 50 -CACGCGCTTGCAGGATTCGG.
14. BacillusExp-R primer: 50 - CAATGTAATTGTTCCCTACCT
GC.
15. GenePulser electroporator (Bio-Rad).
16. 0.2-cm electroporation cuvette.
17. 125-mL baffled bottom Erlenmeyer flask.
18. 500-mL baffled bottom Erlenmeyer flask.
19. 2.5-L Tunair baffled shaker flask.
20. Orbital shaker incubator.
21. Centrifuge with adaptors for 1-L, 250-mL, 50-mL, and 15-mL
tubes.
Production of Recombinant Antibody Fragments via the iRAT system 81

Fig. 2 Structure of the iRAT-based Fv expression vector pBIM2. (a) Schematic

illustration of the region containing the promoter, secretion signal, and iRAT-
82 Norimichi Nomura et al.

2.3 Protein 1. E. coli TOP10 chemically competent cells (Thermo Fisher

Expression Using Sf9 Scientific).
Cells 2. E. coli DH10Bac chemically competent cells (Thermo Fisher
and Baculoviruses Scientific).
3. SOC medium: 20 g/L tryptone, 5 g/L yeast extract, 10 mM
NaCl, 2.5 mM KCl, 10 mM MgCl2, 10 mM MgSO4, 20 mM
glucose.
4. LB/ampicillin plate: 10 g/L tryptone, 5 g/L yeast extract,
5 g/L NaCl, 100 mg/L ampicillin, 15 g/L agar.
5. LB/ampicillin medium: 10 g/L tryptone, 5 g/L yeast extract,
5 g/L NaCl, 100 mg/L ampicillin.
6. LB/KTGIB agar plate: 10 g/L tryptone, 5 g/L yeast extract,
5 g/L NaCl, 50 mg/L kanamycin, 10 mg/L tetracycline,
7 mg/L gentamicin, 40 mg/L IPTG, 50 mg/L X-Gal, 15 g/
L agar.
7. LB/KTG medium: 10 g/L tryptone, 5 g/L yeast extract, 5 g/
L NaCl, 50 mg/L kanamycin, 10 mg/L tetracycline, 7 mg/L
gentamicin.
8. pFBGP67-iRATmC (Fig. 4).
9. IRAT70-F primer: 50 -ACTAGTGAAAATTTATATTTTCAAG
GTCACCATCACCATCACCATGGTGGCTCCTCTGGT.
10. IRAT70-R primer: 50 - AAGCTTCGATTGGAAGTACA
GGTTTTCACCACCACTACTACTGGAACCCGAAGAGC
CGGA.
11. FB-F1 primer: 50 - TAAAAGCTTGTCGAGAAGTACTAGAG
GATCATAATCAGCCATACCACA.
12. gp67-R primer: 50 - CGCCGCAAAGGCAGAATGCGCCGC
CGCCGCCAAAAGCACATATAAAAC.
13. SFseq-F primer: 50 -ACTGTTTTCGTAACAGTTTTGTAATA.
14. SFseq-R primer: 50 - GTGGTATGGCTGATTATGATCCTCT
AGTACTTCT.
15. Sf9 cells adapted for suspension growth.
16. PSFM-J1 medium (Wako).
17. Fetal bovine serum (FBS).

Fig. 2 (continued) polyprotein construct. (b) Universal PCR primers for the
construction of expression plasmid. (c) Sequence corresponding to the region
shown in (a). P5–35 and P5–10 represent the 35 and 10 box of the promoter
P5, respectively. SD1 and SD2 are Shine-Dalgarno sequences. The Sec secre-
tion signal, TEV recognition sequence, His6 tag, and MBP are shown in gray,
orange, yellow, and green, respectively. Gray arrowheads indicate the signal
cleavage site and TEV cleavage sites
 Production of Recombinant Antibody Fragments via the iRAT system 83

18. PSFM-J1/FBS-PS medium: PSFM-J1 medium, 2% FBS,

50,000 U/L penicillin G, 50 mg/L streptomycin.
19. 1
Grace’s medium (Thermo Fisher Scientific).
20. FuGENE HD transfection reagent (Promega).
21. LUNA automated cell counter and cell counting slides
(Thermo Fisher Scientific).
22. Trypan blue: 0.4% solution in PBS, pH 7.2.
23. Six-well tissue plate with flat bottom.
24. 125-mL Erlenmeyer flask (Vent filter cap/Baffle bottom)
(Corning).
25. 3-L Fernbach flask (Vent filter cap/Baffle bottom) (Corning).
26. Orbital shaker incubator.
27. Centrifuge with adaptors for 1-L and 50-mL tubes.
28. Temperature-controlled room or incubator set at 27  C.

2.4 Protein 1. Ammonium sulfate.

Purification 2. TBS: 10 mM Tris-HCl (pH 7.5), 150 mM NaCl.
and Characterization
3. Buffer A: 10 mM Tris-HCl (pH 8.0), 150 mM NaCl, 20 mM
imidazole.
4. Buffer B: 10 mM Tris-HCl (pH 8.0), 150 mM NaCl, 250 mM
imidazole.
5. Buffer C: 10 mM Tris-HCl (pH 8.0), 150 mM NaCl, 10 mM
imidazole.
6. Buffer D: 10 mM Tris-HCl (pH 8.0), 150 mM NaCl, 500 mM
imidazole.
7. Ni-NTA Superflow resin (Qiagen).
8. Ni Sepharose excel resin (GE Healthcare).
9. Glass Econo-column (Bio-Rad).
10. HisTrap HP column (5 mL) (GE Healthcare).
11. HiLoad16/60 Superdex200 column (GE Healthcare).
12. Dialysis tubing (MWCO 10 K) (Spectra/Por).
13. Amicon Ultra-15 (MWCO 10 K) centrifugal concentrator
(Merck).
14. Amicon Ultra-15 (MWCO 30 K) centrifugal concentrator
(Merck).
15. His6-tagged TEV protease (TEV-His6) (see Note 2).
16. Chromatograhy system (e.g., BioLogic DuoFlow system
(Bio-Rad)).
84 Norimichi Nomura et al.

3 Methods

3.1 Original iRAT Fv fragments (with a molecular weight of ~25 kDa) are heterodi-
System meric molecules consisting of two domains, VL and VH, that assem-
for the Production ble through hydrophobic interactions. Each of the VL and VH
of Recombinant Fv domains contains an intrachain disulfide bond for stabilization.
Fragments These structural properties require a sophisticated apparatus for
secretion, folding, and assembly, as well as an oxidizing environ-
ment for disulfide bond generation. The expression of Fv fragments
via the B. choshinensis Sec-dependent secretion pathway ensures the
formation of the intrachain disulfide bonds to stabilize the struc-
ture. A reason to include MBP in the iRAT segment is that the
distance between the N- and C-termini (43 Å) is similar to the
distance between the C-terminus of VL and the N-terminus of VH
(approximately 35–40 Å) for most Fv fragments.

3.1.1 Preparation 1. Day 1: Inoculate a single colony of B. choshinensis into 10 mL of

of Brevibacillus TM medium and grow overnight at 37  C and 200 rpm.
Electrocompetent Cells 2. Day 2: Inoculate 100 mL of TM medium in a baffled bottom
500-mL Erlenmeyer flask with 1 mL of overnight culture and
incubate at 37  C and 200 rpm until an OD600nm of 3.0–4.0
(see Note 3).
3. Centrifuge at 5000
g for 10 min at 4  C.
4. Resuspend the cell pellet in 20 mL of PGH buffer.
5. Centrifuge at 8000
g for 20 min at 4  C.
6. Remove the supernatant thoroughly and resuspend the cell
pellet carefully in 2 mL of PGH buffer.
7. Dispense into 100 μL aliquots and flash-freeze with liquid
nitrogen.
8. Store at 80  C.

3.1.2 Construction The pBIM2 vector (Fig. 2a) for the expression of Fv fragments is
of the Brevibacillus designed based on the backbone of the Brevibacillus plasmid
Expression Plasmid pNY326. The VL-insertion site, iRAT cassette, and VH-insertion
site are tandemly arranged in this order and are introduced down-
stream of the constitutively active P5 promoter and Sec secretion
signal sequence. The iRAT-MBP cassette encodes the N-terminal
TEV cleavage site, His6 tag, a 9-residue Gly/Ser-rich linker, the
MBP coding sequence (residues 27–392 from UniProt ID:
P0AEX9), a 25-residue Asn/Gly/Ser-rich linker, and a
C-terminal TEV cleavage site (Fig. 2c). Residues 1–107 (Kabat
numbering scheme) of the VL domain and 1–113 of the VH domain
derived from monoclonal antibodies are used for Fv fragment
expression.
 Production of Recombinant Antibody Fragments via the iRAT system 85

Table 1
Preparation of PCR tubes

Component Volume Final concentration

H2O 23.4 μL –
PrimeSTAR Max Premix (2
) 25 μL 1

Forward primer (100 pmol/μL) 0.3 μL 0.6 μM
Reverse primer (100 pmol/μL) 0.3 μL 0.6 μM
Template (5 ng/μL) 1 μL 0.1 ng DNA/μL
Total 50 μL

Table 2
PCR primer template combination

PCR product

Name Size Forward primer Reverse primer Template

VL 360 bp F1 R1 Native or synthetic cDNA encoding VL
iRAT-MBP 1300 bp BIM2-F BIM2-R pBIM2 vector
VH 380 bp F2 R2 Native or synthetic cDNA encoding VH
VEC-NY 3400 bp pNY326-F pNY326-R pBIM2 vector

1. Day 1: Perform PCR amplifications of VL, the iRAT-MBP

segment, VH, and the pNY326 vector backbone (VEC-NY).
Add the appropriate components to four 0.2 mL PCR tubes as
indicated in Table 1. The PCR primer–template combinations
are listed in Table 2. The primers for the VL and VH domains
include 50 overhangs (27 bp) complementing the upstream and
downstream sequences to the pBIM2 vector junctions. Mix the
contents gently, and place the tubes in a thermal cycler.
2. Run the PCR according to the following parameters: (1) 2 min
at 95  C; (2) 30 cycles of a sequence composed of 10 s at
98  C/5 s at 56  C/5 s/kbp at 72  C; and (3) 10 min at 72  C.
3. Check the PCR products by loading an aliquot (5 μL) onto a
1% agarose gel with a DNA ladder in a separate lane. After
ethidium bromide staining, observe the gel by UV
illumination.
4. Purify the remaining PCR product (45 μL) using a Wizard SV
gel and PCR clean-up system, according to the manufacturer’s
instructions. Use the PCR products directly for purification.
DNA extraction from the agarose gel slice is not necessary.
86 Norimichi Nomura et al.

Table 3
Gibson assembly reaction

Component Size of DNA fragment Amount per reaction

VL 360 bp 0.5 pmola
iRAT-MBP 1300 bp 0.5 pmola
VH 380 bp 0.5 pmola
VEC-NY 3400 bp 0.16 pmola
Gibson Assembly Master Mix (2
) --- 10 μL
Sterile MilliQ water --- To make 20 μL total volume
Total 20 μL
pmol ¼ (weight in ng)
1000/(base pairs
650 Da)
a

5. Measure the DNA concentration using a NanoDrop

spectrophotometer.
6. Set up the Gibson assembly reaction (20 μL) as in Table 3.
7. Incubate for 60 min at 50  C.
8. Add 80 μL of sterile MilliQ water to make a total volume of
100 μL.
9. Purify the reaction product using a MinElute PCR purification
kit, according to the manufacturer’s instructions. Carefully
elute the DNA with 10 μL of sterile MilliQ water.
10. Thaw 100 μL of Brevibacillus choshinensis electrocompetent
cells on ice (approximately 10 min), and mix the cells by flick-
ing gently.
11. Transfer the cells to a chilled 1.5-mL centrifuge tube. Add
10 μL of the purified Gibson assembly reaction product.
12. Carefully transfer the cell/DNA mix into a chilled 0.2-cm
cuvette without introducing bubbles, and ensure that the
cells deposit across the bottom of the cuvette. Electroporate
using the following conditions for the Bio-Rad GenePulser
electroporator: 1.5 kV, 1000 Ω, and 25 μF. The typical time
constant is ~19 ms.
13. Immediately add 1 mL of pre-warmed MT medium to the
cuvette, gently mix by pipetting up and down twice, and then
transfer to the 17 mm
100 mm round-bottomed
culture tube.
14. Shake (200 rpm) at 37  C for 1 h.
15. Concentrate the cells via centrifugation as appropriate, and
then spread 100 μL of cells onto a pre-warmed MT/neomycin
agar plate.
 Production of Recombinant Antibody Fragments via the iRAT system 87

Table 4
Colony PCR mix

Component Volume/reaction Final concentration

H2O 3.4 μL –
10
Ex Taq Buffer 1 μL 1

dNTPs mixture (2.5 mM each) 1 μL 250 μM each
BacillusExp-F (1 pmol/μL) 2 μL 0.2 μM
BacillusExp-R (1 pmol/μL) 2 μL 0.2 μM
TaKaRa Ex Taq (5 U/μL) 0.1 μL 0.05 U/μL
Brevibacillus cell lysate 0.5 μL –
Total 10 μL

16. Incubate the plate overnight at 37  C.

17. Day 2: Perform colony PCR to check the size of the insert
DNA. Aliquot 100 μL of sterile MilliQ water into 0.2 mL PCR
tubes (see Note 4).
18. To create a stock of each individual colony, pick a small colony
using a sterile pipette tip, streak the pipette tip onto another
MT/neomycin plate, and grow at 37  C for more than 6 h.
19. Dip the same pipette tip into MilliQ water in a 0.2-mL
PCR tube.
20. Boil the Brevibacillus cells on a thermal cycler by incubating the
tubes at 95  C for 5 min.
21. Centrifuge the boiled samples at 6000
g for 1 min at room
temperature.
22. Use 0.5 μL of the supernatant to prepare the colony PCR mix
(10 μL) as in Table 4.
23. Run the PCR according to the following parameters: (1) 5 min
at 95  C; (2) 30 cycles of a sequence composed of 30 s at
94  C/1 min at 55  C/1 min at 72  C; and (3) 5 min at 72  C.
24. Load 10 μL of each colony PCR product directly onto a 1%
agarose gel, alongside an appropriate DNA ladder. After ethi-
dium bromide staining, observe the gel by UV illumination.
The expected size of the colony PCR product is approximately
2200 bp.
25. Select several positive colonies and use them to inoculate 6 mL
of 2SY/neomycin medium. Shake the cultures overnight at
30  C and 200 rpm.
26. Day 3: Add 1 mL of each saturated overnight culture to 0.5 mL
of 50% glycerol in a 2-mL cryovial and gently mix. Freeze the
88 Norimichi Nomura et al.

glycerol stock tube at 80  C. The stock is now stable for

years, as long as it is kept at 80  C.
27. Extract and purify the plasmid DNA from the remaining 5 mL
of each culture using the QIAquick spin miniprep kit, accord-
ing to the manufacturer’s instructions. Measure the DNA con-
centration using a NanoDrop spectrophotometer.
28. Using the vector primers, BacillusExp-F and BacillusExp-R,
perform sequence analysis to confirm that the nucleotide
sequence of the plasmids is correct.

3.1.3 Secretory 1. Day 1: (Pre-culture) Open the tube of the glycerol stock of
Expression Using Brevibacillus choshinensis harboring a positive pBIM2 expres-
Brevibacillus Cells sion plasmid and scrape some of the frozen bacteria off using a
sterile loop. Inoculate the bacteria into 50 mL of 2SY/neomy-
cin medium in a 125-mL baffled-bottom Erlenmeyer flask and
grow for 40–55 h at 30  C and 200 rpm.
2. Day 3: (Large-scale culture) Inoculate 1 L of 2SY/neomycin
medium into a 2.5-L Tunair baffled shaker flask with 10 mL of
saturated pre-culture.
3. Grow the cells for 48–60 h at 30  C and 200 rpm.
4. Day 5: Centrifuge the culture at 6000
g for 15 min at 4  C.
Recover the supernatant and discard the cell pellet.

3.1.4 Purification 1. Day 1: (Ammonium sulfate precipitation) To the 1 L of Brevi-

of Tag-Free Fv Fragments bacillus culture supernatant, add 390 g of ammonium sulfate
powder slowly but steadily with thorough mixing to adjust to a
final concentration of 60% saturation. Do not allow clumps
to form.
2. Allow a precipitate to form for 1 h at 4  C with stirring.
3. Recover the precipitate by centrifugation at 10,000
g for
20 min at 4  C.
4. Dissolve the precipitate by adding 10 mL of TBS.
5. Dialyze overnight against 1 L of TBS at 4  C.
6. Centrifuge the dialyzed sample at 15,000
g for 10 min at
4  C. Recover the supernatant and discard the precipitate.
7. Day 2: (Immobilized metal affinity chromatography (IMAC))
Use 5 mL of Ni-NTA resin (10 mL of 50% slurry) per 1 L of
Brevibacillus culture, pour the slurry into a glass Econo-
Column, and equilibrate with five column volumes (CVs) of
buffer A.
8. Add the supernatant to the column and collect the flow-
through fraction.
9. Wash the column with 10 CVs (50 mL) of buffer A.
Production of Recombinant Antibody Fragments via the iRAT system 89

10. Elute the polyprotein of interest from the column by adding

three CVs (15 mL) of buffer B and collect the eluate into a
50-mL Falcon tube.
11. Measure the protein concentration of each fraction using a
NanoDrop spectrophotometer.
12. (TEV cleavage and dialysis) Add an appropriate amount (typi-
cally at a protease to target protein ratio of 1:100 (w/w) or
3 mg) of TEV-His6 to the elution fraction.
13. Transfer the mixture into dialysis tubing, and dialyze overnight
at 4  C against 2 L of TBS.
14. Day 3: (Reverse IMAC) Equilibrate a 5 mL HisTrap HP col-
umn with 25 mL of buffer C.
15. Remove the dialysate from the tubing and inject it into the
column using a 10 mL syringe.
16. Collect the flow-through fraction (FT).
17. Wash the column three times, each wash containing 1 CV
(5 mL) of buffer C and collect as W1, W2, and W3 fractions,
respectively.
18. Elute the bound material (excised iRAT-MBP segment, TEV--
His6, and contaminating proteins) from the column by adding
three CVs (15 mL) of buffer D and collect as E fraction.
19. Measure the protein concentration of each fraction using a
NanoDrop spectrophotometer.
20. Mix the FT and W1 fractions and concentrate the protein using
a Amicon Ultra-15 (MWCO 10 K) centrifugal concentrator to
a final volume of 5 mL.
21. (Size-exclusion chromatography (SEC)) Inject the 5 mL of
concentrated protein sample onto a HiLoad16/60 Super-
dex200 column pre-equilibrated with TBS at a flow rate of
1 mL/min.
22. Collect 2 mL fractions. Analyze the UV trace, and pool the
peak fractions.
23. Perform SDS-PAGE to check the purity.
24. Measure the protein concentration of each fraction using a
NanoDrop spectrophotometer to estimate the protein yield.
25. Use an aliquot to evaluate the biophysical and/or functional
proprieties of the purified tag-free Fv fragment (Fig. 3 as a
typical example). Ideally, binding assays, such as surface plas-
mon resonance (SPR), isothermal titration calorimetry (ITC),
and ELISA, should provide valuable information.
26. Concentrate the remaining purified Fv fragment using an Ami-
con Ultra-15 (MWCO 10 K) centrifugal concentrator to an
appropriate protein concentration (e.g., 10 mg/mL).
90 Norimichi Nomura et al.

Fig. 3 Certolizumab Fv produced via the Brevibacillus-iRAT system. (a) Typical profile in size exclusion
chromatography of the certolizumab Fv-TNFα complex in the presence of excess certolizumab Fv on a
Superdex200 Increase 10/300GL column (blue line). As a reference, TNFα alone was separated on the same
column (green line). Elution volumes of protein standards are indicated at the top. Peak Fv-T certolizumab
Fv-TNFα trimer complex, peak T free TNFα trimer, and peak Fv free certolizumab Fv. (b) SDS-PAGE analysis of
the corresponding peak fractions in (a). The theoretical molecular masses of the certolizumab Fv and TNFα
trimer are 24.6 kDa (VL: 11.6 kDa, VH: 13.0 kDa) and 51.9 kDa (17.3 kDa protomer
3), respectively

27. Re-measure the protein concentration of each fraction using a

NanoDrop spectrophotometer.
28. Prepare 100 μL aliquots and flash-freeze with liquid nitrogen.
29. Store at 80  C.

3.2 Extended iRAT Fab fragments (with a molecular weight of ~50 kDa) are hetero-
System dimeric molecules consisting of two chains, each of which has two
for the Production domains: the light chain variable domain (VL) linked to the light
of Recombinant Fab chain constant domain (CL) and the heavy chain variable domain
Fragments and Their (VH) linked to the heavy chain constant domain 1 (CH1). Fab
Fluorescent fragments contain five disulfide bonds (except IgA and IgE which
Derivatives have six disulfides) and need to be targeted to the Gram-negative
bacterial periplasm, the Gram-positive bacterial extracellular space,
or the eukaryotic endoplasmic reticulum. The linear distance
between the C-terminus of the Fab light chain (FabL) and
N-terminus of the Fab heavy chain (FabH) for most Fab fragments
is approximately 65 Å. Appropriate long flexible linkers are needed
for successful “molecular surgery” to anastomose the two termini
without violating the overall fold and biological function of the Fab
fragments. According to Koerber et al. a 70-residue Gly/Ser-rich
linker (sc70) is available to express functional single-chain Fab
[13]. Here we provide an example of which the mCherry red
fluorescent protein, instead of MBP, is installed into an intermedi-
ate point of the sc70 linker to create a new iRAT segment. This
construct yields a polyprotein that works as a genetically encoded
 Production of Recombinant Antibody Fragments via the iRAT system 91

fluorescent Fab with complete target-binding activity before

removal of the iRAT segment as described in the Subheading
3.2.3 (see Fig. 5). Once the iRAT segment is excised by TEV
protease, a large amount of tag-free Fab fragment can be prepared
as described in the Subheading 3.2.4 (see Fig. 6). Notably, this
newly innovated iRAT segment has a moonlighting function as a
fluorescent tag (see Note 5). The nested mCherry facilitates esti-
mation of the expression levels of antibody fragments by simply
applying the crude culture supernatant to fluorescence size-
exclusion chromatography (FSEC).
iRAT-mediated Fab production can be performed using either
Brevibacillus or Sf9 insect cells (see Note 6). When Sf9 expression is
adopted, the culture supernatant can be directly loaded onto a
column for IMAC using Ni Sepharose excel resin, which saves
time and labor for pretreatment, such as ammonium sulfate precip-
itation and dialysis.

3.2.1 Construction The pFBgp67-iRATmC vector (Fig. 4a) for the expression of Fab
of the Baculovirus Strain fragments is designed based on the backbone of the pFastBac1, an
entry vector to produce Bac-to-Bac baculoviruses. The FabL inser-
tion site, iRATmC cassette, and FabH insertion site are tandemly
arranged in this order and are introduced downstream of the con-
stitutively active PPH promoter and gp67 secretion signal sequence.
The iRATmC cassette encodes the N-terminal TEV cleavage site,
His6 tag, a 35-residue Gly/Ser/Thr-rich linker, the mCherry cod-
ing sequence (residues 3–236 from UniProt ID: X5DSL3), a
35-residue Gly/Ser-rich linker, and a C-terminal TEV cleavage
site (Fig. 4c).
1. Day 1: Perform PCR amplifications of the FabL, iRATmC,
FabH, and pFastBac vector backbone containing the gp67
signal sequence (VEC-FB). Add the appropriate components
to four 0.2-mL PCR tubes as in Table 5. The PCR primer-
template combinations are listed in Table 6. The primers for
FabL and FabH domains include 50 overhangs (27 bp) comple-
menting the upstream and downstream sequences to the
pFBgp67-iRATmC vector junctions. Mix the contents gently,
and place the tubes in a thermal cycler.
2. Run the PCR according to the following parameters: (1) 2 min
at 95  C; (2) 30 cycles of a sequence composed of 10 s at
98  C/5 s at 56  C/5 s/kbp at 72  C; and (3) 10 min at 72  C.
3. Check the PCR products by loading an aliquot (5 μL) on a 1%
agarose gel with a DNA ladder in a separate lane. After ethi-
dium bromide staining, observe the gel by UV illumination.
4. Purify the PCR product (45 μL) using a Wizard SV gel and
PCR clean-up system, according to the manufacturer’s instruc-
tions. Use the PCR products directly for purification. DNA
extraction from the agarose gel slice is not necessary.
92 Norimichi Nomura et al.

Fig. 4 Structure of the Bac-to-Bac entry vector pFBgp67-iRATmC for the iRAT-based Fab expression. (a)
Schematic illustration of the region containing the promoter, secretion signal, and iRAT-polyprotein construct.
(b) Universal PCR primers for construction of expression plasmid. (c) Sequence corresponding to the region
shown in (a). The gp67 secretion signal, TEV recognition sequence, His6 tag, and mCherry are shown in gray,
orange, yellow, and purple, respectively. Gray arrowheads indicate the signal cleavage site and TEV cleavage
sites. The sc70-derived linker regions are shown in green
 Production of Recombinant Antibody Fragments via the iRAT system 93

Table 5
PCR amplification

Component Volume Final concentration

H2O 23.4 μL –
PrimeSTAR Max Premix (2
) 25 μL 1

Forward primer (100 pmol/μL) 0.3 μL 0.6 μM
Reverse primer (100 pmol/μL) 0.3 μL 0.6 μM
Template (5 ng/μL) 1 μL 0.1 ng DNA/μL
Total 50 μL

Table 6
PCR primer–template combinations

PCR product
Forward Reverse
Name Size primer primer Template
FabL 700 bp F1 R1 Native or synthetic cDNA encoding Fab light
chain
iRATmC 980 bp IRAT70-F IRAT70-R pFBgp67-iRATmC
FabH 700 bp F2 R2 Native or synthetic cDNA encoding Fab heavy
chain
VEC-FB 4800 bp FB-F1 gp67-R pFBgp67-iRATmC

5. Measure the DNA concentration using a NanoDrop

spectrophotometer.
6. Set up the Gibson assembly reaction (20 μL) as in Table 7.
7. Incubate for 60 min at 50  C.
8. Add 60 μL of sterile MilliQ water to the reaction product.
9. Aliquot 10 μL of the diluted reaction product, and chill in a
1.5-mL centrifuge tube.
10. Thaw the E. coli TOP10 competent cells (100 μL) on ice.
11. Add 100 μL of competent cells to the DNA. Mix gently by
pipetting up and down to mix the cells and DNA. Do not
vortex.
12. Place the mixture on ice for 30 min.
13. Heat shock at 42  C for 30 s.
14. Place the mixture on ice for 2 min.
15. Add 900 μL of pre-warmed SOC medium to the tube.
94 Norimichi Nomura et al.

Table 7
Gibson assembly reaction

Component Size of DNA fragment Amount per reaction

FabL 700 bp 0.5 pmola
iRATmC 980 bp 0.5 pmola
FabH 700 bp 0.5 pmola
VEC-FB 4800 bp 0.16 pmola
Gibson Assembly Master Mix (2
) – 10 μL
Sterile MilliQ water – To make a final volume of 20 μL
Total 20 μL
pmol ¼ (weight in ng)
1000/(base pairs
650 Da)
a

Table 8
Colony PCR mix

Component Volume/reaction Final concentration

H2O 3.9 μL ––
10
Ex Taq Buffer 1 μL 1

dNTPs mixture 1 μL 250 μM each
(2.5 mM each)
SFseq-F (1 pmol/μL) 2 μL 0.2 μM
SFseq-R (1 pmol/μL) 2 μL 0.2 μM
TaKaRa Ex Taq (5 U/μL) 0.1 μL 0.05 U/μL
Total 10 μL

16. Incubate the tube at 37  C for 60 min with vigorously shaking

(250 rpm).
17. Spread 100 μL of the cells and DNA mixture onto an LB/am-
picillin plate.
18. Incubate overnight at 37  C.
19. Day 2: Perform colony PCR to check the size of the insert
DNA. Prepare the colony PCR mix (10 μL) in 0.2 mL PCR
tubes as in Table 8.
20. To create a stock of each individual colony, pick a small colony
using a sterile pipette tip, streak the pipette tip onto another
LB/ampicillin plate, and grow at 37  C for more than 6 h.
Production of Recombinant Antibody Fragments via the iRAT system 95

21. Run the PCR according to the following parameters: (1) 2 min
at 94  C; (2) 25 cycles of a sequence composed of 30 s at
94  C/1 min at 55  C/1 min at 72  C; and (3) 5 min at 72  C.
22. Load 10 μL of each colony PCR product directly onto a 1%
agarose gel, alongside an appropriate DNA ladder. After ethi-
dium bromide staining, observe the gel by UV illumination.
The expected size of the colony PCR product is approximately
2500 bp.
23. Select several positive colonies, and use them to inoculate 6 mL
of LB/ampicillin medium. Shake the cultures overnight at
37  C and 250 rpm.
24. Day 3: Add 1 mL of each saturated overnight culture to 0.5 mL
of 50% glycerol in a 2-mL screw top tube or cryovial and gently
mix. Freeze the glycerol stock tube at 80  C. The stock is now
stable for years, as long as it is kept at 80  C.
25. Extract and purify the plasmid DNA from the remaining 5 mL
of each culture using a QIAquick spin miniprep kit, according
to the manufacturer’s instructions. Measure the DNA concen-
tration using a NanoDrop spectrophotometer.
26. Using the vector primers, SFseq-F and SFseq-R, perform
sequence analysis to confirm that the nucleotide sequence of
the plasmids is correct.
27. Day 4: Chill 1 μL of the positive pFBgp67-iRATmC-based
entry plasmid in a 1.5-mL centrifuge tube.
28. Thaw the E. coli DH10Bac competent cells (10 μL) on ice.
29. Add 10 μL of DH10Bac competent cells to the plasmid DNA.
Mix gently by pipetting up and down to mix the cells and
DNA. Do not vortex.
30. Place the mixture on ice for 30 min.
31. Heat shock at 42  C for 45 s.
32. Place the mixture on ice for 2 min.
33. Add 90 μL of pre-warmed SOC medium to the tube.
34. Incubate the tube at 37  C for 4 h with vigorous shaking
(250 rpm).
35. Dilute the transformation culture with SOC medium as appro-
priate, and then spread 100 μL of cells onto an LB/KTGIB
agar plate.
36. Incubate the plate for 2 days at 37  C.
37. Day 6: Pick several white colonies, and use them to inoculate
6 mL of LB/KTG medium. Shake the cultures overnight at
37  C and 250 rpm.
96 Norimichi Nomura et al.

38. Day 7: Add 1 mL of each saturated overnight culture to 0.5 mL

of 50% glycerol in a 2-mL cryovial and gently mix. Freeze the
glycerol stock tube at 80  C. The stock is now stable for
years, as long as it is kept at 80  C.
39. Extract and purify the bacmid DNA from the remaining 5 mL
of each culture using a QIAquick spin miniprep kit, according
to the manufacturer’s instructions. Measure the DNA concen-
tration using a NanoDrop spectrophotometer. The recombi-
nant bacmid DNA is now ready for transfection into insect cells
to generate the baculovirus.
40. Take an aliquot of a properly maintained Sf9 stock culture,
count cells, and determine their viability (see Note 7).
41. Seed 1
106 of Sf9 cells in 2 mL of PSFM-J1/FBS-PS
medium per well of six-well plate.
42. Incubate the cells at 27  C until they attach (~15 min).
43. While waiting for the cells to attach, add 6 μL of FuGENE
transfection reagent to 140 μL of 1
Grace’s medium in a 1.5-
mL centrifuge tube for each transfection. Add 1 μg of bacmid
DNA to the FuGENE/1
Grace media mixture, incubate for
15 min at 27  C, and then add 1
Grace’s medium to make a
total volume of 500 μL.
44. Remove 2 mL of the supernatant from the seeded cells and
replace with 500 μL of the FuGENE/bacmid DNA mixture
dropwise onto the cells. Ensure even dispersal.
45. Incubate the cells for 4 h at 27  C, aspirate the transfectant-
DNA suspension, add 2 mL of fresh PSFM-J1/FBS-PS
medium, and return to 27  C (make sure to put the plates in
a sealable plastic bag to prevent strong evaporation of the
reagent/medium).
46. Incubate the cells for 96 h at 27  C.
47. Day 11: Collect the supernatant containing P0 virus (~2 mL
from each well), and filter the medium containing P0 virus into
a 2-mL centrifuge tube using a 3-mL syringe fitted with a small
0.2-μm filter. This virus is a stock of P0 virus that should be
stored at 4  C, protected from light.
48. Prepare a 125-mL Erlenmeyer flask (Vent filter cap/Baffle
bottom) containing 50 mL of Sf9 cells in the PSFM-J1/FBS-
PS medium at a density of 1.5
106 cells/mL in the exponen-
tial growth phase.
49. Infect with 500 μL of P0 virus.
50. Incubate the cells for 96 h at 27  C with agitation (120 rpm).
51. Day 15: Transfer the suspension into a 50-mL tube, and cen-
trifuge for 10 min at 2380
g at 4  C.
52. Collect the supernatant into a fresh 50-mL tube.
 Production of Recombinant Antibody Fragments via the iRAT system 97

53. Store at 4  C protected from light. This virus is the P1 virus

stock. P1 is sufficient for initial protein expression studies. If
large volumes of virus are required, amplify P1 to obtain P2.
54. Determine the titer of the P1 virus using the endpoint dilution
assay.

3.2.2 Secretory 1. Day 1: Expansion of Sf9 cells should be prepared in advance, so

Expression Using Sf9 Insect that a sufficient number of cells is available on this day. To
Cells/Baculoviruses expand Sf9 cells, determine the total number of cells and
percent viability using a LUNA automated cell counter and
Trypan blue exclusion, and ensure that the density of the cells
is 1.5
106 cells/mL. Based on the volume of cells needed
(0.5–2 L), calculate the volume of the medium needed to
dilute the culture. For 0.5 L, 7.5
108 cells and ~ 0.4 L of
PSFM-J1/FBS-PS medium will be needed and a 3-L Fernbach
flask (Vent filter cap/Baffle bottom).
2. Add P1 virus at a multiplicity of infection (MOI) of 1 to infect
Sf9 cells at a density of 1.5
106 cells/mL (see Note 8).
3. Incubate the cells at 27  C with agitation (120 rpm).
4. Days 2–4: Monitor the mCherry-derived fluorescence in the
culture supernatant over the elapsed time. Take an aliquot of
the Sf9 culture, spin down the cells using a table top centrifuge,
and use 200 μL of the supernatant to measure the fluorescence
(excitation: 587 nm; emission maximum: 610 nm).
5. Day 4: Collect the culture supernatant 96 h after infection by
centrifugation for 15 min at 6000
g at 4  C. Discard the cell
pellet. Adjust the pH of the supernatant to ~8.0 by adding
1/20 volume of 1 M Tris-HCl (pH 8.0).

3.2.3 Purification 1. Day 1: (IMAC) Use 20 mL of Ni Sepharose excel resin (40 mL

of Genetically Encoded of 50% slurry) per 1 L of Sf9 culture, pour the slurry into a glass
Fluorescent Fab Fragments Econo-Column, and equilibrate with five CVs of buffer A.
2. Add the Sf9 culture supernatant directly to the column, and
collect the flow-through fraction.
3. Wash the column with 10 CVs of buffer A.
4. Elute the polyprotein of interest from the column by adding
three CVs of buffer B and collect the eluate.
5. Measure the protein concentration of the eluate using a Nano-
Drop spectrophotometer.
6. Concentrate the protein using an Amicon Ultra-15 (MWCO
30 K) centrifugal concentrator to the volume of 5 mL.
7. (SEC) Inject the 5 mL of concentrated protein sample onto a
HiLoad16/60 Superdex200 column pre-equilibrated with
TBS at a flow rate of 1 mL/min.
98 Norimichi Nomura et al.

Fig. 5 iRAT-based production and characterization of mCherry-engineered certolizumab Fab. (a) The con-
centrated final product from the Subheading 3.2.3 was analyzed by SEC on a Superdex200 Increase 10/300GL
column. (b) FSEC profile. The fractions collected in (a) were subjected to mCherry fluorescence measurement
(excitation: 587 nm, emission maximum: 610 nm). (c) SDS-PAGE analysis of the peak fraction (labeled by
Fab*) of SEC in (a). The existence of multiple extra bands could be attributed to DsRed-specific fragmentation
 Production of Recombinant Antibody Fragments via the iRAT system 99

8. Day 2: Collect 2 mL fractions. Analyze the UV trace, and pool

the peak fractions.
9. Perform SDS-PAGE to check the purity.
10. Measure the protein concentration of each fraction using a
NanoDrop spectrophotometer to estimate the protein yield.
11. Use an aliquot to evaluate the biophysical and/or functional
properties of the purified mCherry-engineered Fab fragment
(Fig. 5 as a typical example). Ideally, binding assays, such as
SPR, ITC, or ELISA, should provide valuable information.
12. Concentrate the remaining purified fluorescent Fab fragment
using an Amicon Ultra-15 (MWCO 30 K) centrifugal concen-
trator to an appropriate protein concentration.
13. Re-measure the protein concentration of each fraction using a
NanoDrop spectrophotometer.
14. Prepare 100 μL aliquots and flash-freeze with liquid nitrogen.
15. Store at 80  C.

3.2.4 Purification 1. Day 1: (IMAC) After collecting the Sf9 culture supernatant,
of Tag-Free Fab Fragments perform the same procedures as steps 1–5 in Subheading
3.2.3.
2. (TEV cleavage and dialysis) Add an appropriate amount (typi-
cally at a protease to target protein ratio of 1:100 (w/w) or
3 mg) of TEV-His6 to the elution fraction.
3. Transfer the mixture into dialysis tubing, and dialyze overnight
at 4  C against 2 L of TBS.
4. Day 2: (Reverse IMAC) Equilibrate a 5-mL HisTrap HP col-
umn with 25 mL of buffer C.
5. Remove the dialysate from the tubing and inject it into the
column using a 10-mL syringe.
ä

Fig. 5 (continued) caused by boiling before SDS-PAGE as described in [18]. (d) SEC profile of the mCherry-
engineered certolizumab Fab-TNFα complex in the presence of excess engineered Fab on a Superdex200
10/300GL column (blue line). As a reference, TNFα alone was separated on the same column (green line).
Elution volumes of protein standards are indicated at the top. Peak Fab*-T engineered certolizumab Fab-TNFα
trimer complex, peak T free TNFα trimer, and peak Fab* free engineered certolizumab Fab. (e) FSEC profile.
The fractions collected in (d) were subjected to mCherry fluorescence measurement. Red line: mixture of the
engineered certolizumab Fab and TNFα trimer; pink line: TNFα trimer alone. (f) SDS-PAGE analysis of the
corresponding peak fractions in (d). (g) In gel fluorescence analysis of mCherry-engineered certolizumab Fab.
The same peak fractions as in (f) were added 1:1 with buffer containing 50 mM Tris-HCl (pH 7.6), 50 mM DTT,
5% glycerol, 5% SDS, 5 mM EDTA, and 0.02% bromophenol blue. Without boiling the sample, electrophoresis
was performed on a Novex WedgeWell 10% Tris-glycine gel (Thermo Fisher Scientific) at 100 V for 130 min.
For mCherry, protein bands were observed on a LED transilluminator. The theoretical molecular masses of the
red fluorescent certolizumab Fab and TNFα trimer are 81.2 kDa and 51.9 kDa (17.3 kDa protomer
3),
respectively
100 Norimichi Nomura et al.

6. Collect the flow-through fraction (FT).

7. Wash the column three times, each wash containing one CV
(5 mL) of buffer C and collect as W1, W2, and W3 fractions,
respectively.
8. Elute the bound material (excised fluorescent iRATmC seg-
ment, TEV-His6 and contaminating proteins) from the column
by adding three CVs (15 mL) of buffer D and collect as the E
fraction.
9. Measure the protein concentration of each fraction using a
NanoDrop spectrophotometer.
10. Mix the FT and W1 fractions and concentrate the protein using
an Amicon Ultra-15 (MWCO 30 K) centrifugal concentrator
to a volume of 5 mL.
11. (SEC) Inject the 5 mL of concentrated protein sample onto a
HiLoad16/60 Superdex200 column pre-equilibrated with
TBS at a flow rate of 1 mL/min.
12. Day 3: Collect 2 mL fractions. Analyze the UV trace, and pool
the peak fractions.
13. Perform SDS-PAGE to check the purity.
14. Measure the protein concentration of each fraction using a
NanoDrop spectrophotometer to estimate the protein yield.
15. Use an aliquot to evaluate the biophysical and/or functional
properties of the purified tag-free Fab fragment (Fig. 6 as a
typical example). Ideally, binding assays, such as SPR, ITC, or
ELISA, should provide valuable information.
16. Concentrate the remaining purified Fab fragment using a Ami-
con Ultra-15 (MWCO 10 K) centrifugal concentrator to an
appropriate protein concentration (e.g., 10 mg/mL).
17. Re-measure the protein concentration of each fraction using a
NanoDrop spectrophotometer.
18. Prepare 100 μL aliquots, and flash-freeze with liquid nitrogen.
19. Store at 80  C.

4 Notes

1. Antibody sequencing from monoclonal hybridoma cell lines

can be performed as described previously [14]. Amino acid
sequence data for the known therapeutic antibodies can be
retrieved from DrugBank (http://www.drug-bank.ca/).
2. The TEV-His6 was prepared as described previously [15]. It is
not necessary to add any additional DTT to the TEV-His6 from
that already contained in its storage buffer.
 Production of Recombinant Antibody Fragments via the iRAT system 101

Fig. 6 Certolizumab Fab produced via the Sf9-iRAT system. (a) Typical profile in SEC of the certolizumab
Fab-TNFα complex in the presence of excess certolizumab Fab on a Superdex200 10/300GL column (blue
line). As a reference, TNFα alone was separated on the same column (green line). Elution volumes of protein
standards are indicated at the top. Peak Fab-T certolizumab Fab-TNFα trimer complex, peak T free TNFα
trimer, and peak Fab free certolizumab Fab. (b) SDS-PAGE analysis of the corresponding peak fractions in (a).
The theoretical molecular masses of the certolizumab Fab and TNFα trimer are 47.6 kDa (Fab light chain:
23.1 kDa, Fab heavy chain: 24.5 kDa) and 51.9 kDa (17.3 kDa protomer
3), respectively. (c) Crystal of the
certolizumab Fab apo. (d) Crystal of the certolizumab Fab-TNFα complex

3. Typically, it takes 6–7 h to reach the desired OD600nm.

4. We usually screen 12–24 neomycin-resistant colonies. The
positive rate is approximately 65–70%.
5. The built-in mCherry within the iRAT segment can be replaced
with another fluorescent protein such as EGFP or mBanana,
depending on the experimental design.
6. Bypassing the papain digestion process is one technical advan-
tage of the iRAT-mediated Fab production. In conventional
preparation for Fabs from whole IgG, papain digestion often
results in a combination of under- and over-digestion and
requires specific optimization depending on different IgG
subclasses.
102 Norimichi Nomura et al.

7. Maintenance and management of insect cell culture can be

performed as described previously [16].
8. We usually add ~1.5 mL of P1 baculovirus stock per 1 L of Sf9
culture.

Acknowledgments

We thank Shoichiro Horita, Hidetsugu Asada, and Tomoko

Uemura for sharing their experiences on Sf9-baculovirus expres-
sion. The creation of the iRAT-based fluorescent Fab was conceived
after being inspired by the pioneering work on multicolored fluo-
rescent scFv by Markiv et al. [17]. This work was funded by the
Strategic Basic Research Program from the Japan Science and
Technology Agency (JST), the Basis for Supporting Innovative
Drug Discovery and Life Science Research (BINDS) from the
Japan Agency of Medical Research and Development (AMED),
the Research on Development of New Drugs from the AMED,
and the Grants-in-Aid for Scientific Research from the Japan Soci-
ety for the Promotion of Science (JSPS) (Nos. 15K06968 and
18K05334).

References
1. Griffin L, Lawson A (2011) Antibody frag- 8. Wu S, Avila-Sakar A, Kim J et al (2012) Fabs
ments as tools in crystallography. Clin Exp enable single particle cryoEM studies of small
Immunol 165:285–291 proteins. Structure 20:582–592
2. Lieberman RL, Culver JA, Entzminger KC 9. Koehl A, Hu H, Maeda S et al (2018) Structure
et al (2011) Crystallization chaperone strate- of the μ-opioid receptor-Gi protein complex.
gies for membrane proteins. Methods Nature 558:547–552
55:293–302 10. Nomura Y, Sato Y, Suno R et al (2016) The
3. Hino T, Arakawa T, Iwanari H et al (2012) G- intervening removable affinity tag (iRAT) pro-
protein-coupled receptor inactivation by an duction system facilitates Fv antibody
allosteric inverse-agonist antibody. Nature fragment-mediated crystallography. Protein
482:237–240 Sci 25:2268–2276
4. Nomura N, Verdon G, Kang HJ et al (2015) 11. Horita S, Nomura Y, Sato Y et al (2016) High-
Structure and mechanism of the mammalian resolution crystal structure of the therapeutic
fructose transporter GLUT5. Nature antibody pembrolizumab bound to the human
526:397–401 PD-1. Sci Rep 6:35297
5. Asada H, Horita S, Hirata K et al (2018) Crys- 12. Ono M, Horita S, Sato Y et al (2018) Structural
tal structure of the human angiotensin II type basis for tumor necrosis factor blockade with
2 receptor bound to an angiotensin II analog. the therapeutic antibody golimumab. Protein
Nat Struct Mol Biol 25:570–576 Sci 27:1038–1046
6. Nagarathinam K, Nakada-Nakura Y, Parthier C 13. Koerber JT, Hornsby MJ, Wells JA (2015) An
et al (2018) Outward open conformation of a improved single-chain Fab platform for effi-
Major Facilitator Superfamily multidrug/H+ cient display and recombinant expression. J
antiporter provides insights into switching Mol Biol 427:576–586
mechanism. Nat Commun 9:4005 14. Fields C, O’Connell D, Xiao S et al (2013)
7. Toyoda Y, Morimoto K, Suno R et al (2019) Creation of recombinant antigen-binding
Ligand binding to human prostaglandin E molecules derived from hybridomas secreting
receptor EP4 at the lipid-bilayer interface. Nat specific antibodies. Nat Protoc 8:1125–1148
Chem Biol 15:18–26
 Production of Recombinant Antibody Fragments via the iRAT system 103

15. Lucast LJ, Batey RT, Doudna JA (2001) 17. Markiv A, Beatson R, Burchell J et al (2011)
Large-scale purification of a stable form of Expression of recombinant multi-coloured
recombinant tobacco etch virus protease. Bio- fluorescent antibodies in gor-/trxB- E. coli
techniques 30:544–546, 548, 550 passim cytoplasm. BMC Biotechnol 11:117
16. Abdulrahman W, Radu L, Garzoni F et al 18. Gross LA, Baird GS, Hoffman RC et al (2000)
(2015) The production of multiprotein com- The structure of the chromophore within
plexes in insect cells using the baculovirus DsRed, a red fluorescent protein from coral.
expression system. Methods Mol Biol Proc Natl Acad Sci U S A 97:11990–11995
1261:91–114
 Chapter 6

Isolation of Artificial Binding Proteins (Affimer Reagents)

for Use in Molecular and Cellular Biology
Anna A. S. Tang, Christian Tiede, Michael J. McPherson,
and Darren C. Tomlinson

Abstract
Artificial binding proteins have been validated as alternatives to antibodies in their use as research reagents
in molecular and cellular biology. For example, they have been used as inhibitors of protein–protein
interactions to modulate activity, to facilitate crystallization, and as probes for cellular imaging.
Phage display is a widely used approach for isolating target-specific binding reagents, and it has even been
used to isolate isoform-specific binding proteins and binders that can distinguish between highly homolo-
gous protein domains.
Here, we describe methods that have been employed in isolating highly specific artificial binding proteins
against a wide range of target proteins.

Key words Artificial binding protein, Affimer, Research reagents, Protein–protein interaction,
Inhibitor, Phage display, Panning, Phage ELISA

1 Introduction

Antibodies are the most widely used research reagents for protein
binding; however, they have numerous limitations and issues with
validation and reproducibility [1–3]. Consequently, artificial bind-
ing proteins such as DARPins [4], monobodies [5], and Affimer
reagents [6–9] have been developed as promising alternatives with
the added benefits of bacterial production for ease of manufactur-
ability and mammalian cell expression to study protein function.
Affimer reagents are based on two types of closely related scaf-
folds—a mammalian scaffold based on Stefin A [6, 7] and the
Adhiron scaffold based on a consensus sequence of phytocystatins
[8]. Both scaffolds were engineered to constrain two variable pep-
tide sequences for molecular recognition (Fig. 1) [8]. Affimer
reagents have been utilized as affinity reagents in various applica-
tions, for example, as detection reagents for diagnostics, as inhibi-
tors of protein–protein interactions to modulate activity, as

Arnaud Poterszman (ed.), Multiprotein Complexes: Methods and Protocols, Methods in Molecular Biology, vol. 2247,
https://doi.org/10.1007/978-1-0716-1126-5_6, © Springer Science+Business Media, LLC, part of Springer Nature 2021

105
106 Anna A. S. Tang et al.

Interacon Interacon
Loop 2 Loop 1

Fig. 1 The full-length Adhiron scaffold as determined by X-ray crystallography at

1.75 Å (PDB: 4N6T). The ligand binding sequences within each interaction loop
(VVAG in loop 1 and PWE in loop 2) are highlighted in dark blue, and these were
replaced with nine randomized amino acids each (excluding cysteine residues) in
the library design [8]

chaperones to facilitate crystallization, and as probes for cellular

imaging [8–30].
Phage display remains the most widely used approach for iso-
lating protein binding reagents, despite the growing number of
alternative methods such as ribosome display [31], mRNA display
[32], and cis display [33]. Phage display was first described by
George P. Smith in 1985 [34] and has since been adapted to display
large libraries of antibody fragments, peptides, and artificial binding
proteins. Large libraries of Affimer reagents have been produced
that display Affimer proteins on a truncated pIII coat protein on
filamentous (Ff) phage [8]. Affimer reagents that specifically bind a
wide range of target proteins have been successfully isolated from
these libraries by phage display.
Selection of Affimer reagents by phage display is usually con-
ducted over three panning rounds. Biotinylated target is immobi-
lized on a streptavidin- or NeutrAvidin-coated surface and
incubated with a population of Affimer-displaying phage. After
washing away unbound phage, bound phage is eluted and propa-
gated for subsequent rounds of panning. A typical method of
Affimer selection by phage display is depicted in Fig. 2.
The described method has been used to isolate isoform-specific
Affimer reagents [24, 25, 35]. The population of phage is first
incubated with the homologous protein to remove binders to the
isoform, prior to selection against the target protein (as depicted in
Fig. 2). We have also used this method of negative selection to
isolate specific binders of targets expressed as fusion proteins
 Isolation of Artificial Binding Proteins 107

Pan 1 Pan 2 Pan 3

1. Immobilisationof
biotinylated target
x4
Wash
Phage Wash Wash
library
x2

2. Incubation with
phage
Pre-pan Pre-pan Pre-pan Control
x3 (negative x4
Wash Wash Wash
selection)

3. Phage elution

4. Infection of ER2738 cells,

plating and phage propagation
Control

Phage ELISA
Streptavidin NeutrAvidin Streptavidin coated Phage Target Deselection
coated well coated well magnetic beads protein target

Fig. 2 Affimer selection by phage display conducted over three panning rounds. In each panning round,
immobilized target (red spots) is incubated with a population of phage (green bars). After washing away any
unbound phage, bound phage is eluted and propagated for subsequent rounds of panning. The population of
phage can also be preincubated with a deselection target, to remove binders to the deselection target, and this
is shown from Pan 2 onwards in the example depicted. After the final round of panning, selected phage clones
are tested for binding to target by phage ELISA. (Figure adapted from Tang AA et al. (2017) [35])

(difficult-to-express proteins are often expressed as fusion proteins

to improve solubility). The phage population can be pre-panned
against the fusion tag alone, prior to incubating with the fusion
protein. Alternatively, the method can also be adapted to isolate
cross-reactive binders, for example, binders of highly homologous
proteins or isoforms of proteins from different species. To do this,
the target protein is alternated between panning rounds.

2 Materials

2.1 Affimer Selection 1. Affimer phage library (phagemid display on pIII).

by Phage Display 2. Biotinylated target protein (1 μg required for each panning
round).
3. Deselection target (optional)—biotinylated/non-biotinylated.
4. ER2738 strain of Escherichia coli cells (Lucigen).
5. M13KO7 helper phage (New England BioLabs).
6. Pierce Streptavidin-coated, high binding capacity, eight-well
strips (ThermoFisher Scientific, catalog no. 15501).
108 Anna A. S. Tang et al.

7. Pierce NeutrAvidin-coated, high binding capacity, eight-well

strips (ThermoFisher Scientific, catalog no. 15508).
8. Dynabeads MyOne Streptavidin T1 (ThermoFisher Scientific,
catalog no. 65601 or 65602).
9. Eppendorf LoBind microcentrifuge tubes (Eppendorf, catalog
no. 0030 108.116).
10. Magnetic separation rack for 1.5-ml microcentrifuge tubes—
for example, DynaMag™-2 Magnet (ThermoFisher Scientific,
catalog no. 12321D).
11. PBS: Prepare a 10
stock by dissolving 80.0 g of NaCl
(MW ¼ 58.44), 2.0 g of KCl (MW ¼ 74.55), 14.4 g of
Na2HPO4 (MW ¼ 141.96), and 2.4 g of KH2PO4
(MW ¼ 136.09) in 800 ml of dH2O. Adjust the pH to 7.4
with HCl and add dH2O until a total volume of 1 l. Autoclave
to sterilize and store at room temperature. For a 1
stock,
dilute the 10
stock with dH2O and adjust the pH to 7.4
with HCl.
12. PBST: Dilute the 10
PBS to 1
with dH2O and 0.1% (v/v)
Tween 20, and adjust to pH 7.4 with HCl.
13. 10
Casein Blocking Buffer (Sigma-Aldrich, catalog
no. B6429).
14. 2
Blocking Buffer: Dilute 10
Casein Blocking Buffer five-
fold in PBS.
15. 2YT medium: Dissolve 16 g of Bacto tryptone, 10 g of Bacto
yeast extract, and 5 g of NaCl (MW ¼ 58.44) in 900 ml of
dH2O. Adjust the pH to 7.0 with NaOH and add dH2O until a
total volume of 1 l. Autoclave to sterilize and store at room
temperature.
16. Tetracycline hydrochloride (1000
stock): Dissolve tetracy-
cline hydrochloride (12 mg/ml) in 70% ethanol. Store at
20  C, protected from light. Working concentration is
12 μg/ml.
17. 0.2 M glycine, pH 2.2: Dissolve 1.5 g of glycine
(MW ¼ 75.07) in 80 ml of dH2O. Adjust the pH to 2.2 with
HCl and add dH2O until a total volume of 100 ml. Autoclave
to sterilize and store at room temperature.
18. 1 M Tris-HCl, pH 9.1: Dissolve 121.14 g of Tris base
(MW ¼ 121.14) in 800 ml of dH2O. Adjust the pH to 9.1
with HCl and add dH2O until a total volume of 1 l. The pH of
Tris buffers varies with temperature and concentration; there-
fore, adjust the pH at room temperature. Autoclave to sterilize
and store at room temperature.
19. Triethylamine: Dilute 14 μl of triethylamine (Sigma-Aldrich,
catalog no. T0886) in 986 μl of PBS, immediately before use.
 Isolation of Artificial Binding Proteins 109

20. 1 M Tris-HCl, pH 9.1: Dissolve 121.14 g of Tris base

(MW ¼ 121.14) in 800 ml of dH2O. Adjust the pH to 9.1
with HCl and add dH2O until a total volume of 1 l. The pH of
Tris buffers varies with temperature and concentration; there-
fore, adjust the pH at room temperature. Autoclave to sterilize
and store at room temperature.
21. 80% glycerol: Dilute glycerol with dH2O. Autoclave to sterilize
and store at room temperature.
22. Halt protease inhibitor cocktail (100
) (ThermoFisher Scien-
tific, catalog no. 87786).
23. Carbenicillin (500
stock): Dissolve carbenicillin (50 mg/ml)
in ddH2O. Filter-sterilize and store at 20  C. Working con-
centration is 100 μg/ml.
24. LB agar plates with carbenicillin: Add 32.0 g of LB agar (Len-
nox L agar) (ThermoFisher Scientific, catalog no. 22700) per
liter of dH2O into an appropriately sized bottle for autoclaving
and swirl to mix. Autoclave to sterilize and cool to ~50–55  C.
Add carbenicillin to a final concentration of 100 μg/ml and
swirl to mix. Flame the neck of the bottle, pour into petri
dishes, and allow to solidify. Store at 4  C for up to 2 months.
Alternatively, the formulation for 1 l of Lennox L agar is as
follows: 10 g of Bacto tryptone, 5 g of Bacto yeast extract, 5 g
of NaCl (MW ¼ 58.44), and 12 g of agar. Swirl to mix and
autoclave to sterilize.
25. Kanamycin (500
stock): Dissolve kanamycin (25 mg/ml) in
ddH2O. Filter-sterilize and store at 20  C. Working concen-
tration is 50 μg/ml.
26. PEG-NaCl precipitation solution: Dissolve 200 g of PEG 8000
and 146.1 g of NaCl (MW ¼ 58.44) in dH2O to a total volume
of 1 l. Autoclave to sterilize and store at room temperature.
27. TE buffer (10 mM Tris, 1 mM EDTA), pH 8.0: Prepare stock
solutions of 1 M Tris-HCl (pH 8.0) and 0.5 M EDTA
(pH 8.0). To prepare 1 M Tris-HCl (pH 8.0), dissolve
121.14 g of Tris base (MW ¼ 121.14) in 800 ml of dH2O. Ad-
just the pH to 8.0 with HCl (at room temperature) and add
dH2O until a total volume of 1 l. Autoclave to sterilize and
store at room temperature. To prepare 0.5 M EDTA (pH 8.0),
add 93.06 g of EDTA disodium salt (MW ¼ 372.24) to 400 ml
of dH2O. Adjust the pH to 8.0 with NaOH—the disodium salt
of EDTA will not fully dissolve until the pH of the solution is
adjusted to pH 8.0. Once dissolved, adjust the volume to
500 ml with dH2O. Autoclave to sterilize and store at room
temperature. For 1 l of TE buffer, mix 10 ml of 1 M Tris-HCl
(pH 8.0) and 2 ml of 0.5 M EDTA (pH 8.0) with 988 ml of
sterile ddH2O. Store at room temperature.
110 Anna A. S. Tang et al.

2.2 Screening 1. 96-well V-bottom deep well plates (Greiner Bio-One, catalog
Selected Phage Clones no. 786201).
by Phage ELISA 2. 2YT medium: Dissolve 16 g of Bacto tryptone, 10 g of Bacto
yeast extract, and 5 g of NaCl (MW ¼ 58.44) in 900 ml of
dH2O. Adjust the pH to 7.0 with NaOH and add dH2O until a
total volume of 1 l. Autoclave to sterilize and store at room
temperature.
3. Carbenicillin (500
stock): Dissolve carbenicillin (50 mg/ml)
in ddH2O. Filter-sterilize and store at 20  C. Working con-
centration is 100 μg/ml.
4. M13KO7 helper phage (New England BioLabs).
5. Kanamycin (500
stock): Dissolve kanamycin (25 mg/ml) in
ddH2O. Filter-sterilize and store at 20  C. Working concen-
tration is 50 μg/ml.
6. Pierce Streptavidin-coated, clear, 96-well plates (ThermoFisher
Scientific, catalog no. 15126).
7. PBS: Prepare a 10
stock by dissolving 80.0 g of NaCl
(MW ¼ 58.44), 2.0 g of KCl (MW ¼ 74.55), 14.4 g of
Na2HPO4 (MW ¼ 141.96) and 2.4 g of KH2PO4
(MW ¼ 136.09) in 800 ml of dH2O. Adjust the pH to 7.4
with HCl and add dH2O until a total volume of 1 l. Autoclave
to sterilize and store at room temperature. For a 1
stock,
dilute the 10
stock with dH2O and adjust the pH to 7.4
with HCl.
8. PBST: Dilute the 10
PBS to 1
with dH2O and 0.1% (v/v)
Tween 20, and adjust to pH 7.4 with HCl.
9. 10
Casein Blocking Buffer (Sigma-Aldrich, catalog
no. B6429).
10. 2
Blocking Buffer: Dilute 10
Casein Blocking Buffer five-
fold in PBS.
11. Biotinylated target protein (recommended 0.5–1.0 μg
per well).
12. Biotinylated deselection target (optional) (recommended
0.5–1.0 μg per well).
13. Anti-Fd-Bacteriophage-HRP (Seramun Diagnostica GmbH,
catalog no. A-020-1-HRP).
14. TMB (SeramunBlau® fast TMB/substrate solution) (Seramun,
catalog no. S-001-TMB).
15. 0.2 M H2SO4 (optional).
 Isolation of Artificial Binding Proteins 111

3 Methods

3.1 Affimer Selection Target proteins can be biotinylated using chemical conjugation
by Phage Display methods, such as EZ-Link NHS-Biotin (ThermoFisher Scientific,
catalog no. 20217), following the manufacturer’s instructions.
Chemical biotinylation may block the binding sites on target pro-
teins. Alternatively, target protein sequence can be tagged with a
biotin acceptor peptide (BAP) sequence (GLNDIFEAQ-
KIEWHE), which is enzymatically biotinylated when expressed in
E. coli in the presence of biotin ligase.

3.1.1 Panning Round 1 Day 1

1. Block the streptavidin-coated wells with 300 μl per well of
2
Blocking Buffer. Prepare four wells per selection—three
wells for pre-panning the phage and one well to immobilize
the target. Cover the wells with an adhesive plate seal and
incubate overnight at 37  C.
2. Set up an overnight culture of ER2738 E. coli cells. Pick a single
colony into 5 ml of 2YT medium with 12 μg/ml tetracycline
and incubate overnight in an orbital incubator at 37  C,
230 rpm.
Day 2
3. Wash the blocked wells 3
with 300 μl of PBST and replace
with 100 μl per well of 2
Blocking Buffer.
4. Immobilization of biotinylated target and deselection target
(if using) (see Note 1). Add 1 μg of biotinylated target to the
selection well (see Note 2). If required, add 1 μg of biotinylated
deselection target into pre-pan wells 2 and 3. Proceed immedi-
ately to step 5.
5. Pre-pan 1. To the first pre-pan well, add 5 μl of Affimer phage
library (~1012 colony-forming units). Cover with a plate seal
and incubate for 40 min at room temperature on a vibrating
platform shaker.
6. Pre-pan 2. Remove the blocking buffer/deselection target
from pre-pan well 2. If deselection target was used, wash the
well 6
with 300 μl of PBST. Transfer the phage from pre-pan
well 1 to pre-pan well 2. Cover with a plate seal and incubate
for 40 min at room temperature on the vibrating platform
shaker.
7. Pre-pan 3. Remove the blocking buffer/deselection target
from pre-pan well 3. If deselection target was used, wash the
well 6
with 300 μl of PBST. Transfer the phage from pre-pan
well 2 to pre-pan well 3. Cover with a plate seal and incubate
for 40 min at room temperature on the vibrating platform
shaker.
112 Anna A. S. Tang et al.

8. Selection. Remove the target from the selection well and wash
6
with 300 μl of PBST. Transfer the phage from pre-pan well
3 to the selection well. If required, add 1 μg of
non-biotinylated deselection target (see Note 3). Cover with
a plate seal and incubate for 2 h at room temperature on the
vibrating platform shaker.
9. Wash the selection well 27
with 300 μl of PBST, preferably
using an automated plate washer.
10. Elution 1. Add 100 μl of 0.2 M glycine, pH 2.2 to the selection
well and incubate for 10 min at room temperature. Neutralize
by adding 15 μl of 1 M Tris-HCl, pH 9.1, and transfer the
eluted phage to a microcentrifuge tube.
11. Elution 2. Dilute 14 μl of triethylamine with 986 μl of PBS.
Add 100 μl of the diluted triethylamine to the selection well
and incubate for 6 min at room temperature. Neutralize by
adding 50 μl of 1 M Tris-HCl, pH 7, and transfer the eluted
phage to the microcentrifuge tube containing elution 1.
12. Infection of ER2738 cells. Before eluting the phage, set up an
8 ml culture of ER2738 cells for each selection. Dilute the
overnight cultures of ER2738 cells in 2YT medium with
12 μg/ml tetracycline to an OD600 of ~0.2 and incubate at
37  C, 230 rpm for about 1 h until OD600 reaches ~0.6. Add
the eluted phage and incubate for 1 h at 37  C at up to 90 rpm.
Mix at least once during the incubation.
13. Plating the cells. Plate 1 μl of the phage-infected ER2738 cells
(diluted in 100 μl of 2YT media) onto LB agar plates with
100 μg/ml carbenicillin (the selection antibiotic for the phage
library). Centrifuge the remaining culture at 3000
g for
5 min. Pour off the supernatant and resuspend the cell pellet
in the residual supernatant. Plate the cells on LB agar with
100 μg/ml carbenicillin. Incubate the plates overnight at
37  C.
Day 3
14. To estimate the phage titer, count the colonies on the plate
containing 1 μl of phage-infected ER2738 cells. Multiply by
8000 to determine the total number per 8 ml of cells (usually
between 0.1 and 2
106).
15. Scrape the cells from the plate containing the remaining cul-
ture. To do this, add 5 ml of 2YT medium with 12 μg/ml
tetracycline and scrape using a disposable plastic spreader.
Transfer the scraped cells to a 50 ml centrifuge tube. Use an
additional 2 ml of medium to scrape off any remaining cells and
mix with the cells in the centrifuge tube. Measure the OD600
of a 1/10 dilution to determine the volume required for an
8 ml culture at OD600 0.2.
 Isolation of Artificial Binding Proteins 113

16. Dilute the appropriate amount of cells in 2YT medium with

12 μg/ml tetracycline for an 8 ml culture at OD600 0.2.
Incubate for about 1 h at 37  C, 230 rpm until an OD600 of
~0.6 is achieved.
17. Add 3.5
1010 M13KO7 helper phage (multiplicity of infec-
tion ~30) to the culture and incubate at 37  C, 90 rpm for 1 h.
18. Add 16 μl of 25 mg/ml kanamycin and incubate overnight at
25  C, 170 rpm.
Day 4
19. Centrifuge the phage-infected cultures at 3500
g for 10 min
and transfer the phage-containing supernatant to fresh tubes.
Remove the required amount of phage-containing supernatant
for panning round 2.
20. Add 2 ml of PEG-NaCl precipitation solution to the remaining
supernatant and incubate for at least 1 h or overnight at 4  C.
21. Centrifuge at 4800
g for 30 min to pellet the phage. Pour off
the supernatant (blotting the tube on tissue paper to remove all
of the supernatant) and resuspend the pellet in 320 μl of TE
buffer.
22. Transfer to a microcentrifuge tube and centrifuge at
16,000
g for 10 min. Transfer the phage-containing super-
natant to a fresh tube. Phage can be stored for several weeks at
4  C. For long-term storage, add an equal volume of 80%
glycerol, mix thoroughly and store at 80  C.

3.1.2 Panning Round 2 Magnetic separation and handling using Dynabeads MyOne Strep-
tavidin T1 magnetic beads can easily be automated on a wide variety
of liquid handling platforms. Panning Round 2 has been automated
on a KingFisher Flex magnetic particle processor (ThermoFisher
Scientific, catalog no. 5400630) and is described by Tang et al.
[35]. However, this method describes how the panning can be
done manually using a magnetic separation rack.
Day 1
1. Block the streptavidin-coated magnetic beads. Resuspend the
Dynabeads MyOne Streptavidin T1 magnetic beads thor-
oughly by vortexing and transfer 20 μl per selection to an
Eppendorf LoBind microcentrifuge tube. Block in a minimum
of 300 μl 2
Blocking Buffer (or 100 μl 2
Blocking Buffer
per 20 μl of streptavidin beads). Incubate overnight at room
temperature on a tube rotator to keep the beads in suspension.
2. Set up an overnight culture of ER2738 E. coli cells. Pick a single
colony into 5 ml of 2YT medium with 12 μg/ml tetracycline
and incubate overnight in an orbital incubator at 37  C,
230 rpm. If performing competitive panning, set this up at
the end of day 2.
114 Anna A. S. Tang et al.

Day 2
3. Centrifuge the blocked streptavidin beads at 800
g for 1 min.
Place the tube on the magnetic separation rack to immobilize
the beads. Remove the blocking buffer and replace with fresh
2
Blocking Buffer, resuspending the beads in 100 μl per 20 μl
of streptavidin beads.
4. Immobilization of biotinylated target. In an Eppendorf
LoBind microcentrifuge tube, add 1 μg of biotinylated target
to 200 μl of 2
Blocking Buffer and 50 μl of the blocked
streptavidin beads. Incubate at room temperature on a tube
rotator to keep the beads in suspension.
5. Immobilization of biotinylated deselection target (if using) for
pre-panning phage (see Note 1). Add 1 μg of biotinylated
deselection target to 200 μl of 2
Blocking Buffer and 50 μl
of the blocked streptavidin beads. Incubate for 1 h at room
temperature on the tube rotator. Centrifuge the tube at
800
g for 1 min. Wash the beads 6
with 500 μl of 2
Block-
ing Buffer using the magnetic separation rack to immobilize
the beads. Resuspend the beads in 50 μl of 2
Blocking Buffer.
6. Pre-pan 1. In an Eppendorf LoBind microcentrifuge tube, add
125 μl of phage-containing supernatant from panning round
1 to 125 μl of 2
Blocking Buffer and 25 μl of the blocked
streptavidin beads (from step 3 or step 5, as appropriate).
Incubate for 1 h at room temperature on the tube rotator.
7. Pre-pan 2. Centrifuge the pre-pan 1 tube at 800
g for 1 min
and place the tube on the magnetic separation rack to immobi-
lize the beads. Transfer the phage-containing supernatant to
the tube containing the remaining 25 μl of blocked streptavidin
beads (from step 3 or step 5, as appropriate). Incubate for 1 h
at room temperature on the tube rotator.
8. Centrifuge the tube containing the immobilized target at
800
g for 1 min. Wash the beads 6
with 500 μl of 2
Block-
ing Buffer using the magnetic separation rack to immobilize
the beads.
9. Selection. Centrifuge the pre-pan 2 tube at 800
g for 1 min
and place the tube on the magnetic separation rack to immobi-
lize the beads. Transfer the phage-containing supernatant to
the tube containing the immobilized target. If desired, add
1 μg of non-biotinylated deselection target (see Note 3). Incu-
bate for 1 h at room temperature on the tube rotator.
10. Centrifuge the tube at 800
g for 1 min. Wash the beads 6

with 500 μl of 2
Blocking Buffer using the magnetic separa-
tion rack to immobilize the beads. For standard panning, elute
the phage immediately as described in steps 13 and 14 (see
Note 4). For competitive panning, continue with the following
step (see Note 5).
 Isolation of Artificial Binding Proteins 115

11. Resuspend the washed beads in the following: 77 μl of PBS,

60 μl of 10
Blocking Buffer, 60 μl of 80% glycerol, and 3 μl of
Halt protease inhibitor cocktail (100
). If desired, add 1 μg of
non-biotinylated deselection target (see Note 3). Incubate for
22–24 h at room temperature on the tube rotator.
Day 2 or 3
12. Centrifuge the tube at 800
g for 1 min. Wash the beads 6

with 500 μl of 2
Blocking Buffer using the magnetic separa-
tion rack to immobilize the beads.
13. Elution 1. Resuspend the beads in 100 μl of 0.2 M glycine,
pH 2.2 to elute the phage. Incubate for 10 min at room
temperature, placing back on the magnet for the last minute.
Transfer the eluted phage to fresh tubes containing 15 μl of
1 M Tris-HCl, pH 9.1 to neutralize.
14. Elution 2. Dilute 14 μl of triethylamine with 986 μl of PBS.
Elute any remaining phage by resuspending the beads in 100 μl
of the diluted triethylamine. Incubate for 6 min at room tem-
perature, placing back on the magnet for the last minute.
Transfer the eluted phage to fresh tubes containing 50 μl of
1 M Tris-HCl, pH 7 to neutralize. Pool with the eluted phage
from Elution 1.
15. Infection of ER2738 cells. Before eluting the phage, set up an
8 ml culture of ER2738 cells for each selection. Dilute the
overnight cultures of ER2738 cells in 2YT medium with
12 μg/ml tetracycline to an OD600 of ~0.2 and incubate at
37  C, 230 rpm for about 1 h until OD600 reaches ~0.6. Add
the eluted phage and incubate for 1 h at 37  C at up to 90 rpm.
Mix at least once during the incubation.
16. Plating the cells. Centrifuge the phage-infected culture at
3000
g for 5 min. Pour off the supernatant and resuspend
the cell pellet in the residual supernatant. Plate the cells on LB
agar with 100 μg/ml carbenicillin. Incubate the plates over-
night at 37  C.
Day 3 or 4.
17. Propagate the phage from Panning Round 2 as described for
Panning Round 1 (see Subheading 3.1.1, steps 15–22).

3.1.3 Panning Round 3 Day 1

1. Block the NeutrAvidin-coated wells with 300 μl per well of
2
Blocking Buffer. Prepare six wells per selection—four wells
for pre-panning the phage, one for panning against the immo-
bilized target, and a negative control well for panning against
either a deselection target or an empty well. Cover the wells
with an adhesive plate seal and incubate overnight at 37  C.
116 Anna A. S. Tang et al.

2. Set up an overnight culture of ER2738 E. coli cells. Pick a single

colony into 5 ml of 2YT medium with 12 μg/ml tetracycline
and incubate overnight in an orbital incubator at 37  C,
230 rpm. If performing competitive panning, set this up at
the end of day 2.
Day 2
3. Wash the blocked wells 3
with 300 μl of PBST and replace
with 100 μl per well of 2
Blocking Buffer.
4. Immobilization of biotinylated target and deselection target
(if using) (see Note 6). Add 1 μg of biotinylated target to the
selection well (see Note 7). If required, add 1 μg of biotinylated
deselection target to the negative control well and pre-pan
wells 1, 2, 3, and 4. Cover with a plate seal and incubate for
10 min at room temperature on a vibrating platform shaker.
5. Pre-pan 1. Remove the blocking buffer/deselection target
from pre-pan well 1. If deselection target was used, wash the
well 6
with 300 μl of PBST. Add 200 μl of phage-containing
supernatant from panning round 2 and 20 μl of 10
Blocking
Buffer. Cover with a plate seal and incubate for 1 h at room
temperature on the vibrating platform shaker.
6. Pre-pan 2. Remove the blocking buffer/deselection target
from pre-pan well 2. If deselection target was used, wash the
well 6
with 300 μl of PBST. Transfer the phage from pre-pan
well 1 to pre-pan well 2. Cover with a plate seal and incubate
for 1 h at room temperature on the vibrating platform shaker.
7. Pre-pan 3. Remove the blocking buffer/deselection target
from pre-pan well 3. If deselection target was used, wash the
well 6
with 300 μl of PBST. Transfer the phage from pre-pan
well 2 to pre-pan well 3. Cover with a plate seal and incubate
for 1 h at room temperature on the vibrating platform shaker.
8. Pre-pan 4. Remove the blocking buffer/deselection target
from pre-pan well 4. If deselection target was used, wash the
well 6
with 300 μl of PBST. Transfer the phage from pre-pan
well 3 to pre-pan well 4. Cover with a plate seal and incubate
for 1 h at room temperature on the vibrating platform shaker. If
required, add 1 μg of non-biotinylated deselection target to the
pre-panned phage prior to selection (see Note 3).
9. Selection. Wash the selection well containing the immobilized
target and the negative control well 6
with 300 μl of PBST.
Transfer 100 μl of pre-panned phage from pre-pan well 4 into
the selection well and 100 μl into the negative control well.
Cover with a plate seal and incubate for 30 min at room
temperature on the vibrating platform shaker.
 Isolation of Artificial Binding Proteins 117

10. Wash the selection well and negative control well 27
with
300 μl of PBST, preferably using an automated plate washer.
For standard panning, elute the phage immediately as
described in steps 13 and 14 (see Note 4). For competitive
panning, continue with the following step (see Note 5).
11. Add the following to the selection well and negative control
well: 80 μl of 2
Blocking Buffer, 20 μl of 80% glycerol and 1 μl
of Halt protease inhibitor cocktail (100
). If desired, add 1 μg
of non-biotinylated deselection target (see Note 3). Cover with
a plate seal and incubate for 22–24 h at room temperature on
the vibrating platform shaker.
Day 2 or 3
12. Wash the selection well and negative control well 6
with
300 μl of PBST.
13. Elution 1. Add 100 μl of 0.2 M glycine, pH 2.2 to the selection
well and negative control well. Incubate for 10 min at room
temperature. Neutralize by adding 15 μl of 1 M Tris-HCl,
pH 9.1 and transfer the eluted phage to a
microcentrifuge tube.
14. Elution 2. Dilute 14 μl of triethylamine with 986 μl of PBS.
Add 100 μl of the diluted triethylamine to the selection well and
negative control well. Incubate for 6 min at room temperature.
Neutralize by adding 50 μl of 1 M Tris-HCl, pH 7 and transfer
the eluted phage to the microcentrifuge tube containing
elution 1.
15. Infection of ER2738 cells. Before eluting the phage, set up
5 ml cultures of ER2738 cells for each selection. Dilute the
overnight cultures of ER2738 cells in 2YT medium with
12 μg/ml tetracycline to an OD600 of ~0.2 and incubate at
37  C, 230 rpm for about 1 h until OD600 reaches ~0.6. Add
the eluted phage and incubate for 1 h at 37  C at up to 90 rpm.
Mix at least once during the incubation.
16. Plating the cells. Plate a range of volumes (e.g., 0.01 μl, 0.1 μl,
1 μl, 10 μl, and 100 μl) onto LB agar plates with 100 μg/ml
carbenicillin. Also, centrifuge and plate the remaining cells as
described for Panning Round 1 (see Subheading 3.1.1, step
13). For the negative control, select one volume to plate
(usually 10 μl). Incubate the plates overnight at 37  C.
Day 3 or 4
17. Compare the number of colonies on the negative control plate
with the selection plate containing the same volume of cells. A
successful screen shows enrichment of target-specific phage
with fewer colonies (less than half) on the negative control
plate. If background is still high on the negative control plate,
118 Anna A. S. Tang et al.

further rounds of panning should be performed (e.g., fourth

round of panning in streptavidin-coated wells). The target-
specific phage clones can be further screened by phage ELISA
(see Subheading 3.2).

3.2 Screening 1. Aliquot 200 μl per well of 2YT medium with 100 μg/ml
Selected Phage Clones carbenicillin into a 96-well V-bottom deep well plate using a
by Phage ELISA multichannel pipette and sterile reservoir.
3.2.1 Expressing 2. Pick individual colonies from the final panning round of phage
Selected Phage Clones display to inoculate into the wells—pick 48 colonies or what-
in a 96-Well Culture Plate ever the amount you want to test.
3. Incubate overnight at 37  C, 1050 rpm in an incubating
microplate shaker (e.g., Heidolph Incubator 1000 and
Titramax 1000).
4. Aliquot 200 μl per well of 2YT medium with 100 μg/ml
carbenicillin into a new 96-well V-bottom deep well plate.
5. Transfer 25 μl per well of the overnight cultures into this new
plate and incubate for 1 h at 37  C, 1050 rpm in the incubating
microplate shaker. It is important to keep the remaining culture
plate overnight (store at 4  C for up to a week or use it to
prepare fresh cultures for glycerol stocks).
6. Dilute M13K07 helper phage (titer ~1014/ml) 1/1000 in 2YT
medium with 100 μg/ml carbenicillin and add 10 μl per well to
the freshly grown cultures using a multichannel pipette. Incu-
bate for 30 min at 37  C, 450 rpm in the incubating microplate
shaker.
7. Dilute the kanamycin stock (25 mg/ml) 1/20 in 2YT medium
with 100 μg/ml carbenicillin and add 10 μl per well to the
phage-infected cultures using a multichannel pipette. Incubate
overnight at 25  C, 750 rpm in the incubating microplate
shaker.
8. Centrifuge the phage-infected culture plate at 3500
g for
10 min.
9. The supernatant contains the phage and can be removed
directly into the ELISA plate to test for binding to target.

3.2.2 Phage ELISA A phage ELISA is performed to test binding of selected phage
clones to target protein. The phagemid DNA from specific phage
binders can then be extracted and identified by DNA sequencing.
1. Block the streptavidin-coated 96-well plate with 300 μl per well
of 2
Blocking Buffer. Cover the wells with an adhesive plate
seal and incubate overnight at 37  C.
2. Wash the blocked wells 3
with 300 μl of PBST, preferably
using a 96-well plate washer.
 Isolation of Artificial Binding Proteins 119

3. Immobilization of biotinylated target (see Note 8) and deselec-

tion target (if using) (see Note 9). Dilute the biotinylated target
and deselection target (if using) to 10 μg/ml (or other appro-
priate concentration, see Note 10) with 2
Blocking Buffer.
Add 50 μl per well of diluted target to the first 6 columns of the
plate and 50 μl per well of deselection target (if using) or
2
Blocking Buffer to the last six columns of the plate. Incu-
bate for 1 h at room temperature on a vibrating platform
shaker.
4. Wash the wells 3
with 300 μl of PBST and add 10 μl per well
of 10
Blocking Buffer into all wells using a multichannel
pipette.
5. Addition of phage clones. Add 40 μl per well of phage-
containing supernatant so that each one is tested against the
target and a negative control well (e.g., phage clone A1 is
added to wells A1 and A7). Incubate for 1 h at room tempera-
ture on the vibrating platform shaker.
6. Wash the wells 3
with 300 μl of PBST, preferably using a
96-well plate washer.
7. Detection of phage with anti-Fd-bacteriophage antibody.
Dilute anti-Fd-bacteriophage-HRP 1/1000 in 2
Blocking
Buffer and add 50 μl per well. Incubate for 1 h at room
temperature on the vibrating platform shaker.
8. Wash the wells 10
with 300 μl of PBST, preferably using a
96-well plate washer.
9. TMB detection of HRP. Add 50 μl per well of TMB and allow
to develop. (Note the amount of time the plate is allowed to
develop—usually 2–3 min) Measure absorbance at 620 nm.
10. Optional: To stop the reaction, add 50 μl per well of 0.2 M
H2SO4. Measure absorbance at 450 nm.

4 Notes

1. A biotinylated deselection target can be immobilized for

pre-panning of phage.
2. The binding capacity of Pierce streptavidin-coated, high bind-
ing capacity, eight-well strips (ThermoFisher Scientific, catalog
no. 15501) is ~125 pmol D-biotin per well; therefore, the
biotinylated target is added in excess and can be reduced, if
required.
3. A non-biotinylated deselection target can be added to the
selection, if desired, for deselection of binders to highly homol-
ogous proteins or binders to fusion tags.
120 Anna A. S. Tang et al.

4. Use standard panning to isolate a more diverse repertoire of

binders.
5. Use competitive panning to isolate higher affinity or more
selective binders.
6. If a deselection target is used to pre-pan the phage, the negative
selection in panning round 3 should be against the deselection
target.
7. The binding capacity of Pierce NeutrAvidin-coated, high bind-
ing capacity, eight-well strips (ThermoFisher Scientific, catalog
no. 15508) is ~60 pmol D-biotin per well; therefore, the bio-
tinylated target is added in excess and can be reduced, if
required.
8. If the method was used to isolate cross-reactive binders by
alternating the target protein between panning rounds, the
phage clones should be tested for binding against each of the
different target proteins used.
9. If a deselection target was used during the selection process,
the phage clones should be tested for binding against the
deselection target as a negative control.
10. The binding capacity of Pierce Streptavidin-coated, clear,
96-well (ThermoFisher Scientific, catalog no. 15126) is
~10 pmol D-biotin per well. A concentration of 10 μg/ml of
biotinylated target is usually sufficient to test in a phage ELISA,
but can be optimized, if required.

Acknowledgments

We thank the Biomedical Health Research Centre, University of

Leeds for funding the BioScreening Technology Group, where
these methods were developed.

References
1. Bradbury A, Pluckthun A (2015) Reproduc- 5. Koide A et al (1998) The fibronectin type III
ibility: standardize antibodies used in research. domain as a scaffold for novel binding proteins.
Nature 518(7537):27–29 J Mol Biol 284(4):1141–1151
2. Baker M (2015) Reproducibility crisis: blame it 6. Woodman R et al (2005) Design and validation
on the antibodies. Nature 521 of a neutral protein scaffold for the presenta-
(7552):274–276. https://doi.org/10.1038/ tion of peptide aptamers. J Mol Biol 352
521274a (5):1118–1133
3. McLeod J, Ferrigno PK (2016) Antibody alter- 7. Stadler LKJ et al (2011) Structure-function
natives. Scientist 30:38–45 studies of an engineered scaffold protein
4. Binz HK et al (2003) Designing repeat pro- derived from Stefin A. II: development and
teins: well-expressed, soluble and stable pro- applications of the SQT variant. Protein Eng
teins from combinatorial libraries of consensus Des Sel 24:751–763
ankyrin repeat proteins. J Mol Biol 332 8. Tiede C et al (2014) Adhiron: a stable and
(2):489–503 versatile peptide display scaffold for molecular
 Isolation of Artificial Binding Proteins 121

recognition applications. Protein Eng Des Sel 23. Gersch M et al (2017) Mechanism and regula-
27:145–155 tion of the Lys6-selective deubiquitinase
9. Tiede C et al (2017) Affimer proteins are ver- USP30. Nat Struct Mol Biol 24(11):920–930
satile and renewable affinity reagents. elife 27 24. Hughes DJ et al (2017) Generation of specific
(6):24903 inhibitors of SUMO-1- and SUMO-2/3-
10. Johnson A et al (2012) Sensitive affimer and mediated protein-protein interactions using
antibody based impedimetric label-free assays Affimer (Adhiron) technology. Sci Signal 10
for C-reactive protein. Anal Chem 84 (505):eaaj2005
(15):6553–6560 25. Robinson JI et al (2018) Affimer proteins
11. King R et al (2015) Inhibition of complement inhibit immune complex binding to Fcgam-
C3 and fibrinogen interaction: a potential maRIIIa with high specificity through compet-
novel therapeutic target to reduce cardiovascu- itive and allosteric modes of action. Proc Natl
lar disease in diabetes. Lancet 26 Acad Sci U S A 115(1):E72–E81
(385):60372–60375 26. Zhurauski P et al (2018) Sensitive and selective
12. Fisher MJ et al (2015) Trivalent Gd-DOTA Affimer-functionalised interdigitated
reagents for modification of proteins. RSC electrode-based capacitive biosensor for Her4
Adv 5(116):96194–96200 protein tumour biomarker detection. Biosens
13. Kyle HF et al (2015) Exploration of the Bioelectron 108:1–8
HIF-1alpha/p300 interface using peptide and 27. Heidelberger JB et al (2018) Proteomic
Adhiron phage display technologies. Mol Bio- profiling of VCP substrates links VCP to
Syst 11:2738–2749 K6-linked ubiquitylation and c-Myc function.
14. Rawlings AE et al (2015) Phage display EMBO Rep 19(4):21
selected magnetite interacting Adhirons for 28. Lopata A et al (2018) Affimer proteins for
shape controlled nanoparticle synthesis. Chem F-actin: novel affinity reagents that label
Sci 6:5586–5594 F-actin in live and fixed cells. Sci Rep 8
15. Stadler LK et al (2014) The use of a neutral (1):018–24953
peptide aptamer scaffold to anchor BH3 pep- 29. Schlichthaerle T et al (2018) Site-specific label-
tides constitutes a viable approach to studying ing of Affimers for DNA-PAINT microscopy.
their function. Cell Death Dis 30(5):564 Angew Chem Int Ed Engl 57
16. Sharma R et al (2016) Label-free electrochem- (34):11060–11063
ical impedance biosensor to detect human 30. Klont F et al (2018) Affimers as an alternative
interleukin-8 in serum with sub-pg/ml sensi- to antibodies in an affinity LC-MS assay for
tivity. Biosens Bioelectron 80:607–613 quantification of the soluble receptor of
17. Weckman NE et al (2016) Comparison of the advanced glycation end-products (sRAGE) in
specificity and affinity of surface immobilised human serum. J Proteome Res 17
Affimer binders using the quartz crystal micro- (8):2892–2899
balance. Analyst 141(22):6278–6286 31. Hanes J, Pluckthun A (1997) In vitro selection
18. Arrata I et al (2017) Interfacing native and and evolution of functional proteins by using
non-native peptides: using Affimers to recog- ribosome display. Proc Natl Acad Sci U S A 94
nise alpha-helix mimicking foldamers. Chem (10):4937–4942
Commun 53(19):2834–2837 32. Wilson DS, Keefe AD, Szostak JW (2001) The
19. Koutsoumpeli E et al (2017) Antibody use of mRNA display to select high-affinity
mimetics for the detection of small organic protein-binding peptides. Proc Natl Acad Sci
compounds using a quartz crystal microbal- U S A 98(7):3750–3755
ance. Anal Chem 89(5):3051–3058 33. Odegrip R et al (2004) CIS display: in vitro
20. Wang W et al (2017) Ultraefficient selection of peptides from libraries of protein-
cap-exchange protocol to compact biofunc- DNA complexes. Proc Natl Acad Sci U S A 101
tional quantum dots for sensitive ratiometric (9):2806–2810
biosensing and cell imaging. ACS Appl Mater 34. Smith GP (1985) Filamentous fusion phage:
Interfaces 9(18):15232–15244 novel expression vectors that display cloned
21. Xie C et al (2017) Development of an Affimer- antigens on the virion surface. Science 228
antibody combined immunological diagnosis (4705):1315–1317
kit for glypican-3. Sci Rep 7(1):017–10083 35. Tang AA et al (2017) Isolation of isoform-
22. Michel MA et al (2017) Ubiquitin linkage- specific binding proteins (Affimers) by phage
specific affimers reveal insights into K6-linked display using negative selection. Sci Signal 10
ubiquitin signaling. Mol Cell 68(1):233–246 (505):eaan0868
 Part II

Biophysical Characterization and Sample Optimization

 Chapter 7

Measurements of Protein–DNA Complexes Interactions by

Isothermal Titration Calorimetry (ITC) and Microscale
Thermophoresis (MST)
Amandine Gontier, Paloma F. Varela, Clément Nemoz, Virginie Ropars,
Magali Aumont-Nicaise, Michel Desmadril, and
Jean-Baptiste Charbonnier

Abstract
Interactions between protein complexes and DNA are central regulators of the cell life. They control the
activation and inactivation of a large set of nuclear processes including transcription, replication, recombi-
nation, repair, and chromosome structures. In the literature, protein–DNA interactions are characterized by
highly complementary approaches including large-scale studies and analyses in cells. Biophysical approaches
with purified materials help to evaluate if these interactions are direct or not. They provide quantitative
information on the strength and specificity of the interactions between proteins or protein complexes and
their DNA substrates. Isothermal titration calorimetry (ITC) and microscale thermophoresis (MST) are
widely used and are complementary methods to characterize nucleo-protein complexes and quantitatively
measure protein–DNA interactions. We present here protocols to analyze the interactions between a DNA
repair complex, Ku70–Ku80 (Ku) (154 kDa), and DNA substrates. ITC is a label-free method performed
with both partners in solution. It serves to determine the dissociation constant (Kd), the enthalpy (ΔH ),
and the stoichiometry N of an interaction. MST is used to measure the Kd between the protein or the DNA
labeled with a fluorescent probe. We report the data obtained on Ku–DNA interactions with ITC and MST
and discuss advantages and drawbacks of both the methods.

Key words Nucleo-protein complexes, Thermodynamic parameters, Microcalorimetry, Fluorescence,

Thermophoresis, Double-Strand Break repair, NHEJ

1 Introduction

The molecular mechanisms in the nucleus are regulated by intricate

networks of protein–protein and protein–DNA interactions. The
word “interaction” itself is used in the literature for many purposes.
It defines associations characterized by large-scale studies with

Amandine Gontier and Paloma Fernández Varela contributed equally with all other contributors.

Arnaud Poterszman (ed.), Multiprotein Complexes: Methods and Protocols, Methods in Molecular Biology, vol. 2247,
https://doi.org/10.1007/978-1-0716-1126-5_7, © Springer Science+Business Media, LLC, part of Springer Nature 2021

125
126 Amandine Gontier et al.

two-hybrid or tap-tag approaches and also in cell analyses with

proximity ligation assays (PLA) or co-immunoprecipita-
tion [1, 2]. “Interactions” include also genetic analyses as well as
biophysical approaches and structural studies of complexes. This
large set of approaches describe the same event of “interaction”
and leads in some case to controversy. On the contrary, these
multiple analyses favor in many cases an integrate view of the
relevance of a given protein complex both in vitro and in the cell.
For example, protein interaction measurements in the cellular con-
text provide a characterization of the interactions in the presence of
potential competitors or of post-translational modifications at dif-
ferent steps of the cell cycle. Biophysical approaches nicely comple-
ment these data by confirming or not the direct interaction with
purified proteins and ligands. Biophysical measurements are central
to characterize the strength, the specificity, and the stoichiometry
of the interactions in a quantitative manner. Interactions measured
in cells and not confirmed in vitro can also be informative of the
role of an additional partner or the need for post-translational
modifications eventually absent on purified proteins used for in vitro
studies.
A large panel of biophysical approaches are used nowadays to
perform quality controls and interaction analyses on macromolecu-
lar samples (some of these are presented in other chapters of this
issue). Each methodology offers complementary information and
comes with specific requirements on sample preparation. We pres-
ent here protocols for Isothermal Titration Calorimetry (ITC) and
MicroScale Thermophoresis (MST) analyses of the interactions
between a protein complex and DNA substrates. The complex is
formed by the heterodimer Ku70–Ku80 (or Ku). Ku is a core factor
of the main DNA Double-Strand Break (DSB) repair pathway in
human, called Non-Homologous End-Joining (NHEJ). Ku recog-
nizes the DSB ends through its ring-shaped structure [3]. The
heterodimer has a high affinity, in the nanomolar range, for
double-strand DNA (15 bp minimum). The Kd of the Ku–DNA
interaction was initially characterized by gel shift assays (EMSA) [4]
and fluorescent anisotropy analyses [5]. Ku binds DNA in a
sequence-independent manner and in vitro one Ku molecule can
thread on DNA with one Ku bound every 15–20 bp of DNA
duplex [3, 6]. Interestingly, a super resolution microscopy study
showed that in cells only one or two Ku molecules bind to DSB
ends, suggesting a limited threading of Ku from the DNA DSB
ends in a cellular context [7]. Ku plays also a central role for the
recruitment of the enzymatic activities (nucleases, polymerases,
ligases) that process and ligate the DSB ends [8, 9]. Our laboratory
recently described at the molecular level the mechanism of recruit-
ment by Ku of some of the downstream NHEJ factors [10]. Here,
we present the measurements by ITC of the thermodynamic para-
meters of the interaction between Ku and DNA substrates, and we
Measurements of Protein–DNA Complexes Interactions by Isothermal. . . 127

report the analyses by MST of the same interaction using DNA

labeled with a fluorescent probe. We previously used ITC to char-
acterize protein–protein and protein–DNA interactions involved in
the NHEJ pathway [6, 11, 12] and in other DNA metabolism
pathways [13–15].
Isothermal titration calorimetry measures the heat generated or
consumed upon the interaction of two molecules, called hereafter
the receptor and the ligand [16, 17]. The molecular interaction
between the receptor and the ligand can be defined by the follow-



ing equation: ΔG ¼ RT lnKa ¼ RT lnKd ¼ ΔH – TΔS, where ΔG
(kcal/M) is the standard Gibbs free energy, R (kcal/M) the gas
constant, T (K) the temperature, Ka (M1) the equilibrium associ-
ation constant, and Kd (M) the equilibrium dissociation constant.
The second part indicates that the free energy is the sum of an
enthalpy term ΔH (kcal/M) and an entropic term TΔS (kcal/M).
Figure 1 represents a schema of the calorimeter. A constant power is
applied to the reference cell. A very small temperature difference is
maintained between the sample and reference cells thanks to a
feedback circuit that applies a variable power to the sample cell.
The injection of a small volume of ligand, loaded in the syringe,
leads to a heat exchange when the ligand interacts with the recep-
tor. Exothermic reactions generate heat and induce a decrease in
the feedback power. Endothermic reactions absorb heat and lead to
an increase of the feedback power. The instrument modifies the
power applied to the measurement cell according to the nature of
the interaction. If the reaction is exothermic, it reduces the power,
and we observe a negative peak in the thermogram (Fig. 1b, top).
On the contrary, if the reaction is endothermic, it increases the
power, and we observe a positive peak in the thermogram (as in
the Ku–DNA example presented here, Fig. 3b). The isotherm of
titration (Fig. 1b, bottom) is determined by integrating each peak
of the thermogram and by reporting the calculated values that
correspond to the enthalpy of the reaction. The x axis of the
isotherm is defined as the ratio of ligand/receptor molecules at
each step of the titration. The curve that best fits these points
gives, in the case of a single binding site, the ΔH, the Kd, and the
stoichiometry of the interaction (Fig. 1b, bottom). ITC has several
advantages: (1) it is widely used since limited optimization is
needed; (2) the interaction is performed without labeling the part-
ners; (3) both partners are in solution at opposite to methods where
one partner is linked to a surface (like SPR, BLI, switchSENSE);
(4) it is nondestructive, the complex receptor–ligand can be used
for other application (cristallisation for example) or re-purified on a
size exclusion chromatography or anion/cation exchange chroma-
tography after measurement to separate the receptor and the
ligand; (5) it has no molecular weight limitation. This method
also has several limitations: (1) it consumes important quantities
128 Amandine Gontier et al.

A B Time (min)
0 20 40 60 80 100
0.00

-0.05

μcal/sec
Syringe
(V = 300 μL)
-0.10

Adiabac -0.15
shield
0.00
Measurement cell Ka

kcal.mol-1 of
-3.00

injectant
(V = 1,4 mL)
ΔT
Reference cell ΔH
-7.00
n
-10.00
0 0.5 1.0 1.5 2.0 2.5
Molar rao

Fig. 1 (a) Schematic representation of an isothermal titration calorimeter (ITC). The reference and measure-
ment cells are positioned in an adiabatic shield. A constant power is applied to the reference cell and a
variable power is applied to the measurement cell to maintain a small constant temperature difference
between the two cells thanks to a feedback circuit. Interactions are between a receptor deposited in the
measurement cell and a small volume of ligand injected by the syringe. Most interactions generate heat
(exothermic reaction). This triggers a decrease in the feedback power as represented in (b, top panel).
Some interactions can consume heat leading to an increase in the feedback power (endothermic reaction).
The integrals of the peaks from the top panel enable to calculate the enthalpy exchange at each injection.
These values are reported versus the ratio of receptor and ligand. The fit of this curve is called the isotherm of
titration. It gives access to the enthalpy ΔH, the association constant Ka (and thus the Kd, Kd ¼ 1/Ka), and the
stoichiometry N

of protein and DNA in particular for Kd in the μM range; (2) it is

difficult to measure Kd below nanomolar range though under
nanomolar range displacement titration can be done [18].
Microscale thermophoresis (MST) is a more recent approach
for measuring the affinity constant (Kd) between a receptor and a
ligand [19, 20]. This method can use either the intrinsic fluores-
cence of the protein (for example Monolith NT.LabelFree, Nano-
temper) or the fluorescence signal (for example Monolith NT.115,
Nanotemper) of a probe linked to the receptor or to the ligand. A
temperature gradient is applied overtime by an IR (infra-red) laser
on a capillary containing the fluorescent receptor alone or with
increasing concentration of the ligand (Fig. 2a). The laser generates
a local heat on the capillary inducing the diffusion of the fluorescent
receptor away from the irradiated zone. The laser is turned off and
the receptor diffuses again to the depleted zone. The diffusion rate
of the receptor is highly sensitive to its hydration state. The forma-
tion of the receptor–ligand complex modifies locally the hydration,
the size, and the charge of the receptor that can lead to detectable
variation on the diffusion of the fluorescently labeled receptor in
 Measurements of Protein–DNA Complexes Interactions by Isothermal. . . 129

Fig. 2 (a) Microscale thermophoresis optic system. MST is measured in capillaries with a small sample volume
(~4 μl). The fluorescence within the capillary is excited and detected through an objective. A focused IR laser
is used to locally heat a defined sample volume. Thermophoresis of fluorescent molecules through the
temperature gradient is detected by the objective. (b) Typical binding experiment. The thermophoretic
movement of a fluorescent receptor (blue trace; unbound) changes upon binding to a non-fluorescent ligand
(blue to red traces; bound), resulting in different traces. The change in thermophoresis is expressed as the
change in the normalized fluorescence (Fnorm), which is defined as F1(hot)/F0(cold) (F-values correspond to
average fluorescence values between defined areas marked by the red and blue cursors, respectively). (c)
Titration with a non-fluorescent ligand results in a gradual change in thermophoresis, which is plotted as Fnorm
versus ligand concentration. When relevant, the resulting binding curve can be fitted with a single binding site
model and enables to define the Kd

the capillary. The fluorescence curves (called MST traces) are

measured in the presence of varying ligand concentrations, ranging
from 1/10 to 20-fold the expected Kd (Fig. 2b). The dissociation
constant (Kd) can be estimated by fitting with a single-site model,
when relevant, the normalized fluorescence against the ligand con-
centration (Fig. 2c). MST presents several advantages: (1) it
130 Amandine Gontier et al.

requires very small amounts of material—typically in the 100 ng to

1 μg range; (2) there is almost no limitation to molecular size or
mass; (3) the method is easy and fast to perform; and (4) it is an
immobilization free method. MST has also several limitations:
(1) one of the partners should be labeled with possible steric
hindrance with the interaction sites, and hydrophobic fluorescent
probes may cause some nonspecific binding; and (2) the theory
behind the measurement principle is less obvious than for other
biophysical approaches.

2 Materials

2.1 Equipment 1. MicroCal VP-ITC: The VP-ITC microcalorimeter (Malvern) is

for Isothermal Titration composed of two compartments or cells: one reference cell
Calorimetry (ITC) is filled with water or buffer and one measurement cell (volume
of 1.4 ml) is filled with the first partner called the receptor.
Both are maintained at equal temperature and are positioned in
an adiabatic environment. The ligand is introduced with small
injections (5–10 μl) in the measurement cell. If the ligand
interacts with the receptor, in most cases, some heat is gener-
ated or absorbed due to the interaction. This is detected by the
device that compensates by respectively reducing or increasing
the electric current in the measurement cell (Fig. 1). In some
rare cases, at a given temperature, an interaction can happen
without heat exchange (ΔH ¼ 0). In that case, changing the
temperature (typically 10
C) should unmask the interaction,
if it exists, since the enthalpy varies with the temperature.
2. The ITC-200 microcalorimeter which is equipped with a mea-
surement cell of 200 μl and a syringe of 40 μl requires smaller
quantities of sample. In the case of the Ku–DNA interaction,
presented here, the VP-ITC calorimeter provided better ther-
mograms and a better signal/noise ratio.
3. ThermoVac™ (Malvern): This device enables to degas the
buffers and samples before use and to pre-equilibrate at a
temperature close to the measurement temperature. It is also
used as pump for cleaning procedures.
4. VPViewer™ and Origin™ software packages: VPViewer is
used to command the VP-ITC and to follow the experiment.
Origin is used to analyze the raw data called thermograms.

2.2 Samples for ITC As will be detailed below, suitable concentrations for a titration
depend on the expected Kd between the ligand and the receptor
and on the enthalpy associated with the formation of the complex.
Quantities listed below are required to characterize the interaction
between the Ku70–Ku80 (Ku) heterodimer and small DNA
 Measurements of Protein–DNA Complexes Interactions by Isothermal. . . 131

duplexes (18 bp and 42 bp). The Kd expected from the literature is

in the nM range [4].
1. The Ku heterodimer is produced and purified as described in
[10]. The complex Ku is expressed in insect cells and purified
using Ni-NTA affinity purification (Thermo Fisher Scientific)
followed by an anion exchange chromatography using a
Resource-Q column (GE Healthcare Life Science) [10]. Insect
cell pellets from 1.2 l culture classically yield 80 mg purified Ku
at the end of the protocol, which is sufficient for about
200 titrations Ku vs DNA.
2. 18 nt or 42 nt DNA oligonucleotides and their complementary
strands are synthetized at the 1 μmole scale (Sigma Aldrich,
HPLC 95% pure). The sequence of the DNA 18 nt is
50 GTCTCACGCTCGGATAAC30 ; the sequence of the DNA
42 nt is 50 GAACGAAAACATCGGGTACGAGGACGAAG
ACTGACCACGACA30 . With a command at 1 μmole scale,
we obtained between 115 and 530 nmoles of oligonucleotides
that is sufficient for at least 35 titrations after hybridization.
3. A dialysis membrane with a molecular weight cut-off (MWCO)
of 30 kDa (Spectra/Por® molecular porous membrane tub-
ing); 2 l of ITC buffer (20 mM Tris-HCl, pH 8.0, 150 mM
NaCl, 5 mM β-mercaptoethanol) prepared with deionized
water (18 mΩ-cm1 at 25
C); Amicon® Ultra-4 30 kDa ultra-
filtration devices.

2.3 Equipment 1. Monolith NT.115: The Monolith NT.115 Blue/Green is

for Microscale equipped with two LEDs (LED1: excitation 470 nm, emission
Thermophoresis (MST) 520 nm; LED2: excitation 550 nm, emission 600 nm, Nano-
temper Technologies). The instrument is equipped with a sam-
ple tray that can process automatically up to 16 samples per run
in 10 min. MST is measured in capillaries with a total volume of
3–5 μl. Control software guides step-by-step planning and
assay setup. Binding values are obtained instantly, and results
are exportable. Affinity Analysis software enables advanced data
analysis like merging, grouping, or comparing data sets.
2. Capillaries: All capillaries are made of glass with infrared laser
quality. Standard Treated Capillaries (Nanotemper) are physi-
cally treated to ensure high-quality surface properties. Hydro-
phobic (Nanotemper) and Premium-Coated (Nanotemper)
Capillaries are covalently coated with a dense brush of specially
designed polymers to avoid molecule adsorption.

2.4 Samples for MST MST is based on the detection of temperature-induced changes in
fluorescence of a target as a function of the concentration of a
non-fluorescent ligand. In typical experiments, 5–500 nM of the
fluorescently labeled molecule is used. The maximum ligand
132 Amandine Gontier et al.

concentration should be 10–20 times that of the Kd. Quantities

listed below are required to characterize the Ku–DNA complex (the
expected Kd is in the nM range).
For experiments with labeled DNA and non-labeled Ku, we
use:
1. Ku heterodimer purified as described in Subheading 2.2. A
total volume of 20 μl at the higher concentration is needed
per run.
2. A dialysis membrane with a molecular weight cut-off (MWCO)
of 30 kDa (Spectra/Por® molecular porous membrane tub-
ing); 2 l of MST buffer (50 mM Tris-HCl, pH 7.4, 150 mM
NaCl, 10 mM MgCl2, 0.05% Tween-20).
3. Desalted 18 bp or 42 bp oligonucleotides labeled with a
50 -FAM and their complementary strands synthetized at the
1 μmole scale (see sequences above with a FAM fluorophore
added in 50 ) (Eurofins). One synthesis yielded about
100 nmoles of oligonucleotides, sufficient for 50 titrations
after hybridization.
For experiments with non-labeled DNA and labeled Ku, you
will need:
4. Ku heterodimer purified as described in Subheading 2.2. The
buffer of Ku is changed for a buffer that does not contain
primary amines, using the labeling buffer of the Monolith
Protein Labeling kit BLUE-NHS (Amine Reactive) from
Nanotemper (reference MO-L003), for which composition is
not provided, but the optimal pH is between 8 and 8.4. A
volume of 100 μl at 20 μM is required.
5. MST buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl,
10 mM MgCl2, 0.05% Tween-20).
6. A Nanotemper labeling kit containing fluorescence dyes that
are suited for MST measurements. The protocols are optimized
to ensure a good labeling, an efficient removal of free dye and
about a 1:1 ratio of labeled protein to dye. Two coupling
chemistries are available, one for primary amines (NHS, e.g.,
lysine labeling) and one for thiol groups (maleimide for cyste-
ine labeling). Second-generation dyes are now available (Tris-
NTA) but only for the instrument with red LED. Proteins with
a molecular weight greater than 5 kDa can be labeled.
7. Desalted 18 bp or 42 bp non-labeled oligonucleotides and
their complementary strands are synthetized at the 1 μmole
scale. The DNA sequences are the same as indicated above.
 Measurements of Protein–DNA Complexes Interactions by Isothermal. . . 133

3 Methods

To get an accurate Kd measurement, the receptor and the ligand

concentrations should be known accurately. This can be deter-
mined with the UV spectrum of both protein and DNA samples
and their extinction coefficient. The biophysical measurement of
the Kd by ITC or MST (as with many other biophysical approaches)
requires highly purified samples. This implies that quality control
(QC) analyses are performed before measurements and that the QC
indicates high purity (greater than 95%). Protein purity can be
determined using several methods including UV spectrum, SDS
PAGE gels, and analytical chromatography (size exclusion chroma-
tography or anion exchange) [21]. DNA purity can be analyzed by
using UV spectrum, acrylamide gels, or analytical anion exchange
chromatography. The receptor is commonly used at a fixed concen-
tration, and the ligand is used at various concentrations ranging
from 1/10 to 20-fold the expected Kd. The way to define the
concentrations used for the receptor in ITC and MST is described
below according to the specificities of these two methods.

3.1 Methods for ITC To get an accurate Kd measurement with ITC, the receptor con-
centration in the cell of the instrument and the ligand loaded in the
syringe should be well chosen. The receptor concentration appro-
priate for an ITC experiment can be estimated using the dimen-
sionless constant c, determined by the formula, c ¼ N  Ka  [Ro],
where N is the stoichiometry of the reaction (number of ligand
molecules bound by molecule of receptor), Ka is the association
constant (M1), and [Ro] the initial concentration of the receptor
in the sample cell (M). To obtain a well-defined S-shaped binding
isotherm with two plateau (Fig. 1b, bottom) and an accurate
measurement of the thermodynamic parameters of the equilibrium,
it is recommended to use an initial concentration of the receptor
[Ro] that results in a c value between 20 and 200 [22]. In the
absence of a priori information, typical starting concentrations are
1 μM for the receptor and 10 μM for the ligand. The ligand is
usually loaded in the syringe at a concentration ten times higher
than the receptor, so that at the end of the titration (typically
28 injections of 10 μl), the molar ratio of the ligand over the
receptor is 2.

3.1.1 Experimental In our example Ku–DNA, with an expected Kd of about 5 nM and a

Design stoichiometry N of 1, we started with a concentration of Ku in the
cell ([Ro]) of 1 μM corresponding to a c value of 200. The ligand
and the receptor are dialyzed against the same buffer to minimize
heat exchanges due to buffer differences.
134 Amandine Gontier et al.

3.1.2 Sample 1. The Ku heterodimer (receptor) is dialyzed three times against

Preparation for ITC ITC buffer for at least 2 h against a volume of buffer
corresponding to at least 30-fold the volume of the receptor
sample. Keep 50 ml of the last dialysis bath as reference buffer.
It will be used to rinse the measurement cells and the syringe
and to adjust the concentration of the receptor and ligands, if
necessary.
2. Preparation of the receptor for the measurement cell: Concen-
trate the dialyzed protein using an Ultra-4 (Amicon), measure
UV absorption at 280 nm (using for example a Nanodrop;
Thermo Fisher Scientific), and adjust the concentration to
1 μM (0.15 mg/ml). A minimum volume of 2.3 ml (1.4 ml
for the measurement cell plus 0.9 ml dead volume) is prepared.
3. Hybridization of DNA substrates (ligand). The oligonucleo-
tides are classically dissolved at 500 μM in water and equal
amounts of the complementary strands are mixed at a temper-
ature of 95
C for 5 min and cooled down overnight at room
temperature. The concentration is estimated using a Nanodrop
(Thermo Fisher Scientific) and UV absorption at 260 nm.
4. Verify hybridization: Inject 6 nmoles (10 μl at 0,6 mM) on a
Mini-Q-1 ml (Sigma-Aldrich) column to check if the DNA is
correctly hybridized and the proportion of the remaining
single-strand DNA. Perform a gradient on the Mini-Q of
20 ml from 100% buffer A (Tris 20 mM, pH 8) to 100% buffer
B (Tris 20 mM, pH 8, NaCl 1 M).
5. The DNA ligand is diluted at 10 μM using the last dialysis
buffer to limit heat exchange due to buffer differences of the
receptor and the ligand. These dilution heats may mask the
desired signal. Prepare 0.6 ml of ligand to be positioned in the
serynge.

3.1.3 ITC Measurements 1. The receptor (2.3 ml of Ku at 1 μM) and the ligand (0.6 ml
and Data Interpretation DNA at 10 μM) are degassed and equilibrated for 10 min at the
measurement temperature (here 20
C) using the
ThermoVac™.
2. The receptor is deposited in the measurement cell with a 2.5 ml
glass syringe. The sample is slowly loaded in the cell to avoid
bubbles. Injection is stopped when the liquid comes above a
visible level.
3. The ligand is loaded on the injection syringe (in open position)
using a 10-ml plastic syringe connected to the top of the ITC
syringe. The syringe is then purged and filled twice to remove
bubbles.
4. Start titration.
 Measurements of Protein–DNA Complexes Interactions by Isothermal. . . 135

18bp
A +

+ 42bp

B n)
Time (min) C Time (min)

-6 12 30 48 66 84 -6 12 30 48 66 84
0.02 0.06
18bp
μcal/sec

μcal/sec
0.04
0.01
0.02 42bp
0.00 0.00

-0.02
6.00 22.0
kcal.mol-1 of inj

kcal.mol-1 of inj

4.00 16.0

2.00 8.0

0.00 0.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0
Molar Ratio Molar Ratio

D
KD ΔH° -TΔS° ΔG°
DNA N
(nM) (kcal.mol-1) (kcal.mol-1) (kcal.mol-1)
18bp 0.97 ± 0.1 3.5 ± 0.8 5.1 ± 0.1 -16.6 ± 0.2 -11.5 ± 0.2
42bp 0.47 ± 0.1 2.4 ± 0.5 22.9 ± 0.3 -34.7 ± 0.5 -11.8 ± 0.1

Fig. 3 ITC measurements between Ku and DNA. (a) Crystal structure of Ku-DNA complex with a 14 bp DNA. Ku
has a ring shape structure. The DNA present in the crystal structure has an additional hairpin part that has
been removed for clarity (pdb code 1JEY). A schema is presented of the two interactions analyzed with Ku and
either a 18 bp DNA or 42 bp DNA. (b) ITC measurement between Ku and a 18 bp DNA. The interaction is
endothermic with positive peaks on the thermogram. The heat exchanges observed in this interaction are very
weak (<0.02 μcal/s) and underline the high sensitivity of the VP-ITC. The isotherm of titration enables to
define ΔH, Kd, and N of the reaction, reported in Table (d). One Ku molecule binds to an 18 bp DNA. (c) ITC
measurement between Ku and a 42 bp DNA hairpin. ΔH, Kd, and N of the reaction are reported in Table (d).
Two Ku molecules bind on the 42 bp DNA leading to a stoichiometry N close to 0.5

We analyzed the interactions of Ku with a 18 bp and a 42 bp

DNA duplex (Fig. 3b, c) and observed two endothermic reactions
(positive peaks) in both the cases. The interaction between Ku and
the 18 bp DNA shows a stoichiometry of 0.97  0.1 molecules of
18 bp DNA bound to one Ku, a Kd of 3.5  0.8 nM, and an
136 Amandine Gontier et al.

entropy-driven reaction with a positive unfavorable enthalpy. The

interaction with the 42 bp DNA gives a stoichiometry of 0.47  0.1
molecule of 42 bp DNA bound to one Ku in good agreement with
the expected positioning of two Ku molecules on such DNA. The
interaction between and the 42 bp DNA shows a Kd of
2.4  0.5 nM and a reaction also driven by the entropy. The heat
exchanges measured for the two interactions are very small (maxi-
mum value of heat exchange ~0.02 μcal/s for Ku–DNA (18 bp))
and highlight the high sensitivity of the VP-ITC.

3.2 Methods for MST We describe here general guidelines to measure interactions using
MST and we illustrate two experimental setups with a labeled DNA
and non-labeled Ku or a labeled Ku and non-labeled DNA. We will
call receptor the fluorescently labeled molecules and ligand the
non-labeled one hereafter. The first step of an MST experiment is
to find a suitable combination of the receptor concentration
(10–100 nM are usually sufficient) and excitation LED power.
The LED can be adjusted to 1–100%. For the binding experiment
itself, you will need ~200 μl at twice the concentration you have
established at this step (for one replicate measurement). To get an
accurate, well-defined Kd, it is important to use a high enough
concentration of the ligand to reach saturation. Saturation is a
hallmark of ligand-specific molecular interaction: if the interaction
cannot be saturated, i.e., the signal does not change anymore by
increasing the ligand concentration, it is not a specific interaction,
but a nonspecific adsorption effect. To reach saturation, it is recom-
mended that the highest concentration of ligand is equal or higher
than 20-fold the expected Kd value.
In a preliminary assay, we compare the fluorescence intensity of
the receptor after switching on the IR laser LED (MST power) in
the absence and in the presence of a high enough concentration of
the ligand (respective to the expected Kd). We perform this experi-
ment with at least four capillaries with and without ligand. The
signal is defined as the response amplitude Fnorm. It is the ratio
between fluorescence intensity after turning on the IR laser LED
(usually measured at the 2-second or 5-second mark, called “MST
on time”) divided by the fluorescence intensity measured before
turning on the IR laser LED. It is expressed as ‰. The noise floor is
the error (standard deviation) between replicates without ligand. If
it is above 8‰, it may reflect some protein aggregation that needs
optimization. This also determines what should be considered
binding (or not) of the ligand: binding is only confirmed, if the
change observed upon ligand binding is threefold larger than your
noise level established at this step. For example, for a typical noise
 Measurements of Protein–DNA Complexes Interactions by Isothermal. . . 137

level of 2‰ at “medium” MST power, the signal change upon

ligand binding should be at least 6‰ (e.g., a change from an
Fnorm value of 857‰ to 863‰). Both the noise and the amplitude
are displayed when reviewing the dose–response panel. In order to
measure the Kd, we incubate our (fluorescently labeled) receptor, at
a concentration below the Kd, with increasing concentrations of
non-labeled ligand. An easy way to do this is to perform a dilution
series, in the form of 1:1 dilutions. In the Monolith instruments,
we can use 16 concentrations; this will cover a concentration range
of 3  104 (215).

3.2.1 Sample Non-labeled Ku and DNA are prepared as described in Subheading

Preparation 3.1.2. First, always spin down the stocks for 5 min at 15,000  g to
remove large aggregates, which is the main source of noise. Always
prepare volumes of at least 20 μl of sample in the smallest micro-
reaction tubes to avoid surface absorption.

3.2.2 Fluorescence Before starting binding assays, several pretests are required to opti-
Pretests mize. Check the optimal concentration of the fluorescence mole-
cules. Ideal fluorescent intensities are between 200 and 1500
counts.
Check the capillary type that is optimal for our study (stan-
dards, hydrophobic, and premium). The aim is to obtain narrow,
shoulder-free, regular, and symmetrical fluorescent peaks.
Check the optimal buffer composition. BSA and detergents can
be used to improve sample homogeneity. pH and/or ionic strength
are also parameters that can be adjusted if needed. Reducing agents
freshly prepared can also be used when needed.

3.2.3 Preliminary Binding In these preliminary assays, we compare the fluorescent signals of
Assays the receptor in the absence and in the presence of the higher
concentration used for the ligand. The DNA solution (10 μl of
[DNA-FAM5] at 10 nM) is mixed with 10 μl buffer or with 10 μl of
Ku at 266 nM concentration. We analyzed fluorescence intensity
and homogeneity, and the signal to noise ratio by repeating the
measurement four times. MST traces should be smooth, which will
indicate the absence of aggregation. The signal is defined as the
response amplitude (difference of fluorescence amplitude between
receptor alone and receptor–ligand complex) and the noise is the
standard deviation of errors between replicates. A signal-to-noise
ratio less than 5 suggests a signal too weak to measure a Kd, a ratio
of 5 or more indicates a quite favorable case for a Kd measurement
and a ratio higher than 12 reflects optimal conditions to measure a
Kd with MST.
138 Amandine Gontier et al.

3.2.4 Kd Determination Prepare 16 small micro-tubes and transfer 20 μl of the highest

concentration of the ligand (non-labeled molecule) to the first
one. The Concentration Finder software tool can be used to simu-
late the interaction with different concentrations of ligand and
receptor. Fill 10 μl of the optimal assay buffers into the tubes
2–16. Use the same pipette tip and transfer 10 μl of tube 1 to
tube 2 and mix well by pipetting up and down several times. Do the
same with tube 3 to tube 16 to obtain a serial dilution. Remove
10 μl from tube 16 after mixing. Add 10 μl of receptor (labeled
molecule) to each tube and mix well by pipetting up and down
several times. Here change tip for every well to ensure the addition
of the same quantity in each tube and no contamination of the
labeled molecule that may be due to the pipetting. Incubate if
needed before loading into the capillaries. Most interactions reach
equilibrium after a few minutes, so additional incubation is often
not needed. When larger conformational changes are required from
both partners, slow kinetics can be observed, and longer incuba-
tions might be necessary. Fill 16 capillaries of the correct type and
place them in the instrument. Start the measurements.

3.2.5 Assays Ku is labeled with NT495 final concentration 20 nM. Titrant DNA
with Labeled Ku (and 18 bp is prepared at concentrations ranging from 123 nM to 3.9
Non-labeled DNA) pM by serial dilution. The buffer used is MST buffer and the
capillaries are premium capillaries. Measurements are performed
at 22
C, at 40% LED excitation and 40% or 60% MST power.
The signal to noise ratio is 3 in these conditions and does not enable
to measure a reliable Kd (Fig. 4a). The excitation power of 40% was
enough to have around 350 fluorescence counts. Increasing the
MST power increases the signal, but also the noise of the data.

3.2.6 Assays DNA is purchased with a FAM fluorophore in 50 of one 18 nt

with Labeled DNA (and single-strand DNA. The fluorescent probe is separated to the
Non-labeled Ku) 18 nt by three nucleotides to avoid interference with the binding
site (Sigma Aldrich, HPLC 95% pure). The stock DNA is prepared
at 230 μM with the same hybridization protocol as for ITC (see
above). The final concentration of DNA in MST assays is 10 nM.
Ku is prepared by serial dilution from 1330 nM to 80 pM. The
buffer used is the MST buffer. Standard capillaries are used as well
as low binding tubes and tips. Measurements are performed at 60%
(data not shown), 80%, and 100% LED excitation, 40% MST
power, and temperature of 22
C (Fig. 4b, d). The signal to noise
ratio at 60%, 80%, and 100% are 5.8, 8.3, and 17.7, respectively.
The Kd is poorly defined at 60% excitation power with a Kd esti-
mated of 100  500 nM (data not shown). We measured a more
precise Kd of 16.0  7.7 nM at 80% (standard error of regression of
1.04 and a reduced χ 2 of 0.95) (Fig. 4b). At 100% excitation power,
we measured the Kd with the smallest standard error, Kd of
10.8  2.6 nM (standard error of regression is 0.37 and reduced
χ 2 is 0.19) (Fig. 4c, d).
 Measurements of Protein–DNA Complexes Interactions by Isothermal. . . 139

A Ku-NT495 vs DNA 18bp B DNA-FAM5 vs Ku

MST power 40% 60% LED power 80%

10

' FNorm [%]

' FNorm [%]

5
0

-1 0

1.0E-12 1.0E-11 1.0E-10 1.0E-09 1.0E-08 1.0E-07 1.0E-06 1.0E-11 1.0E-10 1.0E-09 1.0E-08 1.0E-07 1.0E-06 1.0E-05 1.0E-04
Ligand Concentration Ligand Concentration

C DNA-FAM5 vs Ku D DNA-FAM5 vs Ku
MST trace LED power 100%
8

1.00
6
[Ku] ↗
Relative Fluorescence [-]

0.95
4
' FNorm [%]

0.90 2

0
0.85

0 5 10 15 20
MST experiment time [s] 1.0E-11 1.0E-10 1.0E-09 1.0E-08 1.0E-07 1.0E-06 1.0E-05 1.0E-04
Ligand Concentration

Fig. 4 MST measurements between Ku70–Ku80 and an 18 bp DNA. (a) Ku is labeled with NT-495 fluorescent
probe. The signal to noise ratio is too weak, and no Kd can be measured at 40% or 60% MST power. (b) The
18 bp DNA is labeled in 50 with a FAM fluorophore on one strand of the duplex. The measurements are
performed at 60%, 80%, and 100% excitation power. The signal to noise ratio and the fits are better defined at
100% excitation (d) than at 80% excitation. The MST traces observed at 100% excitation are shown in (c)

4 Additional Remarks

1. Before each ITC experiment, make sure that the measurement

cell and the syringe are clean. Clean the sample cell and the
syringe with 14% v/v decon and rinse them with, respectively,
1 l and 250 ml of water. A control ITC measurement is then
performed with 20 injections of 10 μl of water from the syringe
into the cell filled with water. The signal should be small
(DP < 0.01 μcal/s) and regular. The next user can then safely
use the device for his measurements. When needed, stronger
cleaning procedures could be used (see recommendations of the
manufacturer (Malvern)). The Hastelloy alloy that covers the
140 Amandine Gontier et al.

ITC cells is compatible with most buffers except strong acids.

The ITC instrument should be preferentially in a room with a
control temperature (about 20
C).
2. The buffer composition can influence the ITC signals. Reduc-
ing agents like DTT cause erratic baseline variations and TCEP
or β-mercaptoethanol are preferred to maintain reducing con-
ditions, when needed. The ionization properties of the buffer
can have an influence on the values measured by ITC in partic-
ular when the interface involves charge exchange. In that case,
it is recommended to compare the values obtained with differ-
ent buffers and at different ionic strengths (e.g., [23]).
3. If no heat exchange is observed during an interaction, it can
happen that the reaction exists but has a null enthalpy at the
temperature chosen. To evaluate this possibility, the ITC mea-
surements can be performed at different temperatures (e.g.,
10
C lower or higher).
4. The ITC gives access to the stoichiometry (N) of the interac-
tion. In many reactions, a value of 1 for N is expected, one
molecule of ligand bound by molecule of receptor. Even, in this
case, some deviation can be observed for several reasons:
(1) the presence of inactive receptors (unfolded, aggregated,
etc.) in the sample cell will lead to a lower N value, proportional
to the fraction of inactive receptor. The presence of inactive
ligands in the syringe will lead to a higher N value, proportional
to the fraction of inactive ligand. Similarly, errors on receptor
or ligand concentrations will have a direct impact on the
N value.
5. In parallel to the measurement of the receptor–ligand interac-
tion, it is important to perform the ITC reaction with the
ligand in the syringe and buffer in the cell. This control will
evaluate if the signals observed are really due to the interaction
and not to other event like the dissociation of the ligand when
it is introduced in the sample cell. This can happen in particular
when the ligand is a protein that can form oligomers and
dissociate when injected in the cell.
6. Kinetic information can be recovered from ITC measurement
using a method called kinITC that can be applied for some
interactions presenting asymmetric peaks in the
thermogram [24].
7. The range of Kd accessible by ITC is classically between 0.1 nM
and 10 μM. For lower Kd, the slope of the isotherm will be
sharp, preventing a precise measurement of the Kd. However,
the ITC can be used in these conditions to determine ΔH and
N of the interaction. At higher Kd, the measurement will often
need a large amount of receptors and ligands to have a correct
fit of the Kd.
Measurements of Protein–DNA Complexes Interactions by Isothermal. . . 141

8. For MST experiments, it is recommended to use low binding

tubes and tips. For the serial dilution, better results are
obtained without changing the tips from tube to tube. How-
ever, when adding the labeling molecule, it is recommended to
change the tip at each tube. Avoid making bubbles by pipetting
carefully and never vortex these small volumes to avoid dena-
turation of samples. Handle capillaries with care and never
touch at the center where the measurement is performed.
9. An SDS denaturation test (SD-test) has to be performed to
distinguish between fluorescence changes caused by an interac-
tion and those caused by nonspecific effects, e.g., loss of pro-
tein due to aggregation or adsorption to labware. This test
consists in the denaturation of all proteins contained in the
sample using a mix of SDS and DTT and heating to 95
C. In
this context, the receptor–ligand interaction is disrupted. After
denaturation, the change of fluorescence should not be
observed in the absence or in the presence of increasing con-
centration of ligands. If addition of ligands induces fluores-
cence change with denatured protein, this suggests some
interference of the ligand and some new experimental condi-
tions should be researched.
10. If not enough fluorescence is observed, one can increase the
concentration of the labeled receptor or increase the LED
power. A different strategy of labeling can be used. In cases of
absorption to the capillary, adding detergents or other amphi-
philic polymer additives will help.
11. Under certain conditions, MST can also be used to obtain the
stoichiometry of the interaction. For that, we need to know
precisely the concentration of receptor and ligand as well as the
Kd of their interaction. In this case, we need a much higher
concentration of the receptor at least 20-fold than the Kd.
12. Thermodynamic can be extracted thanks to the Van’t Hoff
ΔH
equation: ln K D ¼ RT þ ΔS
R . When plotting ln(KD) ¼ f
(1/T), one can obtain a straight line which can be fitted as a
linear equation where the slope is given by ΔH/R and the
intercept is ΔS/R.
13. It can be worth to label the other molecule and repeat the
experiment in the reverse way to confirm the result.
14. A positive control and a negative control are advisable to
confirm that there is no unspecific interaction or that the
conditions of the assay are optimal.
142 Amandine Gontier et al.

Acknowledgments

J.B.C. is supported by ARC program (SLS220120605310), ANR

(ANR-12-SVSE8-012), INCA DomRep (PLBIO 2012-280), and
CEFIPRA grant 5203C and by the French Infrastructure for
Integrated Structural Biology (FRISBI) (ANR-10-INBS-05).
A.G. is supported by a CIFRE PhD fellowship with Sanofi. We
thank Pierre Soule from Nanotemper Technologies for his avail-
ability and all the fruitful discussion. The experiments were per-
formed on the Platform PIM (Platform for measurements of
Interactions of Macromolecules) (https://www.i2bc.paris-saclay.
fr/spip.php?article280).

References
1. Gavin AC, Bosche M, Krause R et al (2002) 10. Nemoz C, Ropars V, Frit P et al (2018) XLF
Functional organization of the yeast proteome and APLF bind Ku80 at two remote sites to
by systematic analysis of protein complexes. ensure DNA repair by non-homologous end
Nature 415(6868):141–147 joining. Nat Struct Mol Biol 25(10):971–980
2. Uetz P, Giot L, Cagney G et al (2000) A com- 11. Ropars V, Drevet P, Legrand P et al (2011)
prehensive analysis of protein-protein interac- Structural characterization of filaments formed
tions in Saccharomyces cerevisiae. Nature 403 by human Xrcc4-Cernunnos/XLF complex
(6770):623–627 involved in nonhomologous DNA
3. Walker JR, Corpina RA, Goldberg J (2001) end-joining. Proc Natl Acad Sci U S A 108
Structure of the Ku heterodimer bound to (31):12663–12668
DNA and its implications for double-strand 12. Malivert L, Ropars V, Nunez M et al (2010)
break repair. Nature 412(6847):607–614 Delineation of the Xrcc4-interacting region in
4. Blier PR, Griffith AJ, Craft J, Hardin JA (1993) the globular head domain of cernunnos/XLF.
Binding of Ku protein to DNA. Measurement J Biol Chem 285(34):26475–26483
of affinity for ends and demonstration of bind- 13. Bacquin A, Pouvelle C, Siaud N et al (2013)
ing to nicks. J Biol Chem 268(10):7594–7601 The helicase FBH1 is tightly regulated by
5. Arosio D, Costantini S, Kong Y, Vindigni A PCNA via CRL4(Cdt2)-mediated proteolysis
(2004) Fluorescence anisotropy studies on the in human cells. Nucleic Acids Res 41
Ku-DNA interaction: anion and cation effects. (13):6501–6513
J Biol Chem 279(41):42826–42835 14. Dherin C, Gueneau E, Francin M et al (2009)
6. Tadi SK, Tellier-Lebegue C, Nemoz C et al Characterization of a highly conserved binding
(2016) PAXX is an accessory c-NHEJ factor site of Mlh1 required for exonuclease
that associates with Ku70 and has overlapping I-dependent mismatch repair. Mol Cell Biol
functions with XLF. Cell Rep 17(2):541–555 29(3):907–918
7. Britton S, Coates J, Jackson SP (2013) A new 15. Liberti SE, Andersen SD, Wang J et al (2011)
method for high-resolution imaging of Ku foci Bi-directional routing of DNA mismatch repair
to decipher mechanisms of DNA double- protein human exonuclease 1 to replication
strand break repair. J Cell Biol 202 foci and DNA double strand breaks. DNA
(3):579–595 Repair (Amst) 10(1):73–86
8. Chang HHY, Pannunzio NR, Adachi N, Lie- 16. Holdgate GA (2001) Making cool drugs hot:
ber MR (2017) Non-homologous DNA end isothermal titration calorimetry as a tool to
joining and alternative pathways to double- study binding energetics. BioTechniques 31
strand break repair. Nat Rev Mol Cell Biol 18 (1):164–170
(8):495–506 17. Krell T (2008) Microcalorimetry: a response to
9. Frit P, Ropars V, Modesti M, Charbonnier JB, challenges in modern biotechnology. Microb
Calsou P (2019) Plugged into the Ku-DNA Biotechnol 1(2):126–136
hub: the NHEJ network. Prog Biophys Mol 18. Velazquez-Campoy A, Freire E (2006) Isother-
Biol 147:62–76 mal titration calorimetry to determine
 Measurements of Protein–DNA Complexes Interactions by Isothermal. . . 143

association constants for high-affinity ligands. 22. Myszka DG, Abdiche YN, Arisaka F et al
Nat Protoc 1(1):186–191 (2003) The ABRF-MIRG’02 study: assembly
19. Asmari M, Ratih R, Alhazmi HA, El Deeb S state, thermodynamic, and kinetic analysis of an
(2018) Thermophoresis for characterizing bio- enzyme/inhibitor interaction. J Biomol Tech
molecular interaction. Methods 146:107–119 14(4):247–269
20. Jerabek-Willemsen M, Wienken CJ, Braun D, 23. Czarny B, Stura EA, Devel L et al (2013)
Baaske P, Duhr S (2011) Molecular interaction Molecular determinants of a selective matrix
studies using microscale thermophoresis. Assay metalloprotease-12 inhibitor: insights from
Drug Dev Technol 9(4):342–353 crystallography and thermodynamic studies. J
21. Raynal B, Lenormand P, Baron B, Hoos S, Med Chem 56(3):1149–1159
England P (2014) Quality assessment and opti- 24. Dumas P, Ennifar E, Da Veiga C et al (2016)
mization of purified protein samples: why and Extending ITC to kinetics with kinITC. Meth-
how? Microb Cell Factories 13:180 ods Enzymol 567:157–180
 Chapter 8

switchSENSE Technology for Analysis of DNA Polymerase

Kinetics
Guillaume Bec and Eric Ennifar

Abstract
The switchSENSE technology is a recent approach based on surface sensor chips for the analysis of
interactions of macromolecules. The technology relies on electro-switchable DNA nanolevers tethered at
one end on a gold surface via a sulfur linker and labeled with a Cy3 dye on the other end. The switchSENSE
approach is effective in the determination of a large panel of biophysical parameters such as binding kinetics,
dissociation constant, hydrodynamic radius, or melting temperature. In addition, it can also give access to
some enzymatic data such as nuclease or polymerase activity. Here we describe a DNA polymerase assay that
allows retrieving, in a single experimental set, association and dissociation rates, as well as the catalytic rate
of the enzyme.

Key words HIV reverse transcriptase, switchSENSE technology, Kinetics, Biosensor, DNA
polymerase

1 Introduction

Immobilization of molecules on solid supports is a fairly well-

known method that allows for very sensitive measurement of
molecular interaction, as for the very popular surface plasmon
resonance (SPR) approach [1]. Usually, biophysical properties
such as molecular size or hydrodynamic radius are inferred from
the molecular mobility under the influence of external forces.
switchSENSE technology is a combination of these two approaches
with the measurement of analyte adsorption on a layer of actuated
surface-bound fluorescent probe [2, 3]. Two main measurement
modes are accessible: (1) a static mode, where analyte binding is
measured thanks to fluorescence intensity variation, and (2) a
dynamic mode, where binding is detected through the change of
the oscillation rate of the probe actuated by an AC electric field.
These two alternative measurement modes give access to a wide
range of biophysical parameters as kON, kOFF, KD, hydrodynamic
diameter, or conformational changes (see for instance refs. 4–11).

Arnaud Poterszman (ed.), Multiprotein Complexes: Methods and Protocols, Methods in Molecular Biology, vol. 2247,
https://doi.org/10.1007/978-1-0716-1126-5_8, © Springer Science+Business Media, LLC, part of Springer Nature 2021

145
146 Guillaume Bec and Eric Ennifar

Master piece of the switchSENSE technology, the probe is

made by a single-stranded DNA bearing a fluorescent dye at one
extremity and attached on its opposite end to a gold-quenching
surface through a sulfur bond. This single-stranded DNA is
annealed to a complementary strand (RNA or DNA) that could
be used alone or conjugated coupled to a protein. Hybridization of
the complementary strand generates a rigid negatively charged
electro-switchable biosensor, also referred as nanolever (Fig. 1).
Making use of the DNA plasticity and of biochemical tools, a
wide variety of nanolevers could be adapted to the biological con-
text of the experiment: size and type of DNA may vary, even
allowing the use of complex DNA origami constructs [12], RNA
motifs can be easily adapted [13], various types of modified extrem-
ity can be used (protein, cofactor, NTA, biotin, etc.) [2], as well as
various dyes (Cy3, Cy5, etc.).
Here we describe a specific setup adapted from Langer and
colleagues [14] for polymerase activity characterization, using an
overhang DNA as nanolever. With this kind of substrate, we show
that it is possible to follow the entire biological reaction including
DNA polymerase association, polymerization, and dissociation rate
in one set of experiment.

2 Materials

2.1 Instrument 1. DRX 2400 instrument (Dynamic Biosensors).

and Accessories 2. Nanodrop ND-1000 spectrophotometer (Thermo Fisher) or
equivalent.
3. switchCONTROL, switchBUILD, and switchANALYSER
software (Dynamic Biosensors).
4. Biochip for enzymatic studies (ENZ-80-1-Y1-S, Dynamic Bio-
sensors) bearing an 80-mer DNA probe attached at the 30
extremity.
5. Passivation and regeneration solutions (Dynamic Biosensors,
SOL-PAS-1-5 and SOL-REG-12-1).
6. The P36T80 oligonucleotide, a 36-mer DNA complementary
to 30 end region of the 80-mer DNA probe attached to the
biochip (see Note 1).
7. Heparin sodium salt, physico-chemical grade (Sigma-Aldrich).
8. 10 Auxilary buffer (100 mM sodium phosphate, pH 7.0,
400 mM NaCl, 0.5% Tween 20, 0.5 mM EDTA).
9. Deoxynucleotide, 100 mM solutions (dATP, dGTP, dCGT,
and dTTP).
10. Ultrapure RNase-free water.
 Analysis of a DNA Polymerase Using switchSENSE Approach 147

Fig. 1 Schematic representation of the HIV-1 DNA polymerase switchSENSE experiment. (a) Illustration
showing the building of the DNA nanolever following by regeneration of the biosensor chip. (b) Scheme of
the switchSENSE measurement described here, containing three main steps: association, polymerization, and
dissociation

11. Ethanol absolute 99.9%.

12. 1.5 and 10 mL autosampler vials with Septa caps and insets.

2.2 Enzyme 1. Reverse transcriptase HIV-1, BH10 isolate (RT) was expressed
Preparation and purified as described previously [15] and stored as a sus-
pension in a 2 M (NH4)2SO4 solution.
148 Guillaume Bec and Eric Ennifar

2. Heparin solution stock (10 mg/mL) is diluted in ultrapure

RNase-free water.
3. 10 Running Buffer (100 mM potassium phosphate, pH 8.0,
800 mM NaCl, 60 mM MgCl2, 0.5% Tween 20).
4. Amicon Ultra-4 centrifugation filter unit 30,000 MWCO
(Millipore, Billerica, MA, USA).

3 Methods

3.1 Polymerase 1. Aliquots of HIV-1 reverse transcriptase (ammonium sulfate

Preparation stock solution) are centrifuged for 20 min at 15,000  g,
for switchSENSE 4  C. Protein pellets are resuspended for 1 h on ice in the 1
Experiment Running Buffer at final concentration of ~1 mg/mL.
2. The protein is then dialyzed against 1 Running Buffer using
an Amicon Ultra-4 by three successive 10-min centrifugations
at 7,500  g, 4  C.
3. After the last centrifugation step, the polymerase is concen-
trated at a final concentration of ~1 mg/mL.
4. Precise concentration of polymerase is obtained using two
independent measures with a nanodrop spectrophotometer
(ε280 ¼ 231,480 mol1 cm1 for pure HIV-1 reverse
transcriptase).

3.2 Solutions 1. 300 μL of RT diluted with 1 Running Buffer (ranging from

Preparation 0.03 to 4 nM, based on concentration estimated in the
previous step).
2. 250 mL of freshly diluted 1 Auxiliary Buffer and 1 Running
Buffer are filtrated on a 2-μm nitrocellulose filter, inserted in
the fluidic system of the DRX 2400 and primed using the
switchCONTROL command of the system software.
3. 300 μL of p36T80 oligonucleotide diluted in ultrapure water
at a final concentration of 500 nM.
4. 1 mL of regeneration solution.
5. 10 mL of freshly diluted 1 passivation solution.
6. All the solutions and buffers are arranged as shown in the
switch CONTROL software window.

3.3 Schematic 1. Building of the DNA nanolever is the first step of the experi-
Overview ment. This reaction is achieved by the hybridization of the
of the Experiment P36T80 oligonucleotide on the 80-mer DNA probe attached
to the chip (30 end), to generate a DNA duplex with a 44 bases
50 overhang (Fig. 1a, see Note 2).
 Analysis of a DNA Polymerase Using switchSENSE Approach 149

2. One concentration of polymerase in the 1 Reaction Buffer is

then injected into the fluidic system and the fluorescence varia-
tion recorded to retrieve the kON (Fig. 1b—association).
3. Without regeneration of the chip, a second injection is directly
performed with a 1 Reaction Buffer containing the same
concentration of polymerase (see Note 3) and an oversaturated
concentration of the four dNTP mixed nucleotides. The
increase in fluorescence signal is recorded over time to retrieve
the polymerization rate signal (Fig. 1b—polymerization, see
Note 4).
4. A last injection of 1 Reaction Buffer completed with heparin
(see Note 5) on the non-regenerated chip is used to retrieve the
kOFF (Fig. 1b—dissociation).

3.4 Experimental 1. The experimental workflow is designed using switchBUILD

Design software by making successive program blocks.
(a) In the first block, channel number, buffer type (P40), and
biochip (ENZ-80-1) are designed.
(b) A “Passivation” step is added as a second block (see Note
6).
(c) The third block is a “Kinetics” step, where the ligand is the
P36T80 oligonucleotide, and the analyte is a defined
concentration of polymerase (see Note 7), the measure-
ment mode set as “Static Mode” (see Note 8). Following
parameters are also defined in this block: temperature,
electrode number (one of the six available in each chan-
nel), association time (see Note 9), flow rate (see Note
10), uncheck option for “regeneration.” If it is desired to
change any condition in the experiment (temperature,
polymerase concentration, etc.), a new block of this type,
with the corresponding modified parameter, should be
inserted next. It is also possible to add a blank-run
block, where the analyte is replaced with 1 Reaction
Buffer alone: resulting curves could be used for baseline
correction if a significant drift of the fluorescence signal is
observed.
(d) A last “Standby” block is used at the end of the experi-
ment (see Note 11).
(e) Once completed, the switchBUILD file is saved, and sam-
ple vials are disposed as requested in the “Autosampler”
window of the software.
2. The previous file is loaded in the switchCONTROL software.
If no other modifications are needed (see Note 12), the experi-
ment could then start (typically 1 h 30 min for a complete cycle
with one polymerase concentration). Examples of data gener-
ated are provided on Fig. 2.
150 Guillaume Bec and Eric Ennifar

Fig. 2 Example of DNA polymerization experiment using four HIV-1 RT concentrations. Fluorescence signal
observed upon addition of HIV-1 RT onto DNA nanolevers (a). on- (b) and off (c) rates obtained using the
switchANALYSIS software package. Determination of the kpol using switchANALYSIS

3. Once completed, datasets generated during the “Kinetics” step

are loaded in the switchANALYSIS software package. For each
concentration of polymerase, three distinct kinetics could be
successively analyzed: kON, kPOL, and kOFF. If several concen-
trations of polymerase have been used, a global fit can be used
in order to significantly improve the accuracy of each fitted
value. Polymerization rates (in residue s1) are calculated con-
sidering a 44 base elongated DNA.
 Analysis of a DNA Polymerase Using switchSENSE Approach 151

4. Once all datasets are analyzed in the switchANALYSIS soft-

ware, a publication-quality image could be copied in the clip-
board. Raw datasets could also be exported in various file
formats (ASCII, Excel, etc.) compatible with other software
tools.

3.5 Consideration Polymerization rate is only fitted during the first 30 s of the reac-
About Obtained tion. As for classical studies in enzymology, this time window
Kinetics Parameters corresponds to the initial linear part allowing to measure initial
rate of the reaction, before the substrate becomes limiting.
In this experiment datasets, please note that the determined
kON value stands for the association of polymerase on a 50 protrud-
ing DNA strand and that the kOFF value stands for dissociation of
the polymerase from a fully extended blunt end DNA. Therefore,
no proper KD value (usually deduced from the ratio kkOFFON
) could be
calculated, because of the change in the nature of the substrate that
occurs during polymerization. KD of the polymerase for both forms
of the DNA could be calculated using a different experimental set
using the polymerase as an analyte and two complementary DNA
nanolevers of 36 or 80 nucleotides long as ligands.

4 Notes

1. P36T80 oligonucleotide (GAGTAACCTGAAGTGTGAGCA-

TAAAGAGTGATGTAA) was chemically synthetized at a
250 nmoles scale, followed by a standard desalting.
2. This construct constitutes the substrate of the polymerase
where P36T80 is the primer strand oligonucleotide, and tem-
plate strand is the 80-mer DNA probe attached on the chip on
its 30 end.
3. The enzyme concentration is kept constant in the fluidic system
during polymerization to maintain the balance of associated/
dissociated polymerase on the DNA substrate.
4. During this step, the fluorescence increases significantly
because the dye moves away from the fluorescence-quenching
surface as the polymerase converts the upper DNA part from a
floppy single- to a rigid double strand.
5. As for many polymerase-involved reactions, heparin is used as a
trap to ensure that the dissociated polymerase do not
re-associate on the DNA nanolever. The use of heparin allows
the same fluorescent signal amplitude to be recovered for dis-
sociation as for the association step.
152 Guillaume Bec and Eric Ennifar

6. This automatized standard procedure (using a commercially

available kit solution) is used to passivate the chip, avoiding
any unspecific binding of analytes of the biosurface.
7. The procedure to analyze three successive steps for polymerase
reaction is not usual and deviates from the pre-programmed
assay in the switchBUILD software. For this reason, we have
chosen to insert three successive injections, each starting
directly after the previous one, without regenerating the chip.
For this purpose, three successive dilutions have to be defined
manually, and the “Regeneration” option should be
unchecked. Instead of filling the tubes with three different
dilutions of polymerase (as requested in the predefined para-
meters), the first dilution tube will contain one concentration
of polymerase diluted with the 1 Reaction Buffer, the second
will contain the same concentration of polymerase and 250 μM
of each dNTP, and the third dilution tube will contain 1
Reaction Buffer supplemented with 0.1 mg/mL heparin.
8. Static mode means that a constant voltage (0.1 V in our
conditions) is applied on the chip, maintaining the nanolever
at a constant angle in the 1 Reaction Buffer. Any event
affecting this angle and/or the distance of the fluorophore
with the quenching surface of the biochip (protein binding,
polymerization, dissociation) will be an interpretable signal for
the experiment.
9. Completion of association, polymerization, and dissociation
steps are dependent of kinetics parameters of the considered
polymerase, concentration, and temperature. Typically, for the
HIV-1 RT polymerase at 25  C, these three steps are set to 5, 4,
3, and 2 min for the concentrations of 0.15, 0.45, 1.35, and
4.05 nM, respectively.
10. Flow rate is automatically set from the estimated kON value set
in the “predicted interaction value” tool implemented in the
switchSENSE software. For HIV-1 RT polymerase (kON ¼ 1.3
E+7 M1 s1 at 25  C), flow rate is automatically set to 50 μL/
min for all the steps.
11. This automatized standard procedure removes analytes and
ligands from the surface and defines conditions suitable for
chip storage.
12. The switchCONTROL software takes the simplicity of switch-
BUILD scripts and converts them into advanced commands
for the unattended operation of the DRX instrument.
 Analysis of a DNA Polymerase Using switchSENSE Approach 153

References
1. Rich RL, Myszka DG (2011) Survey of the bioorthogonal labeling in vitro and in living
2009 commercial optical biosensor literature. cells. Chembiochem 19:1949–1953
J Mol Recognit 24:892–914 9. Rueda FO, Bista M, Newton MD et al (2017)
2. Knezevic J, Langer A, Hampel PA et al (2012) Mapping the sugar dependency for rational
Quantitation of affinity, avidity, and binding generation of a DNA-RNA hybrid-guided
kinetics of protein analytes with a dynamically Cas9 endonuclease. Nat Commun 8:1610
switchable biosurface. J Am Chem Soc 10. Webster MW, Chen YH, Stowell JAW et al
134:15225–15228 (2018) mRNA deadenylation is coupled to
3. Langer A, Hampel PA, Kaiser W et al (2013) translation rates by the differential activities of
Protein analysis by time-resolved measure- Ccr4-not nucleases. Mol Cell 70:1089–1100.
ments with an electro-switchable DNA chip. e8
Nat Commun 4:2099 11. Webster MW, Stowell JA, Passmore LA (2019)
4. Alt A, Dang HQ, Wells OS et al (2017) RNA-binding proteins distinguish between
Specialized interfaces of Smc5/6 control similar sequence motifs to promote targeted
hinge stability and DNA association. Nat Com- deadenylation by Ccr4-Not. elife 8:e40670
mun 8:14011 12. Kroener F, Heerwig A, Kaiser W et al (2017)
5. Blocquel D, Li S, Wei N et al (2017) Alterna- Electrical actuation of a DNA origami nanole-
tive stable conformation capable of protein ver on an electrode. J Am Chem Soc
misinteraction links tRNA synthetase to 139:16510–16513
peripheral neuropathy. Nucleic Acids Res 13. Clery A, Sohier TJM, Welte T et al (2017)
45:8091–8104 switchSENSE: a new technology to study
6. Denichenko P, Mogilevsky M, Clery A et al protein-RNA interactions. Methods
(2019) Specific inhibition of splicing factor 118-119:137–145
activity by decoy RNA oligonucleotides. Nat 14. Langer A, Schraml M, Strasser R et al (2015)
Commun 10:1590 Polymerase/DNA interactions and enzymatic
7. Nemoz C, Ropars V, Frit P et al (2018) XLF activity: multi-parameter analysis with electro-
and APLF bind Ku80 at two remote sites to switchable biosurfaces. Sci Rep 5:12066
ensure DNA repair by non-homologous end 15. Freisz S, Bec G, Radi M et al (2010) Crystal
joining. Nat Struct Mol Biol 25:971–980 structure of HIV-1 reverse transcriptase bound
8. Ploschik D, Ronicke F, Beike H et al (2018) to a non-nucleoside inhibitor with a novel
DNA primer extension with cyclopropenylated mechanism of action. Angew Chem Int Ed
7-Deaza-20 -deoxyadenosine and efficient Engl 49:1805–1808
 Chapter 9

Sedimentation Velocity Methods for the Characterization

of Protein Heterogeneity and Protein Affinity Interactions
Christine Ebel and Catherine Birck

Abstract
Sedimentation velocity analytical ultracentrifugation is a powerful and versatile tool for the characterization
of proteins and macromolecular complexes in solution. The direct modeling of the sedimentation process
using modern computational strategies allows among others to assess the homogeneity/heterogeneity state
of protein samples and to characterize protein associations. In this chapter, we will provide theoretical
backgrounds and protocols to analyze the size distribution of protein samples and to determine the affinity
of protein–protein hetero-associations.

Key words Sedimentation velocity, Analytical ultracentrifugation, Protein heterogeneity, Sedimenta-

tion coefficient, Hetero-association, Protein interaction

1 Introduction

The characterization of biological samples to assess their homoge-

neity and to study the formation of protein complexes is of interest
in many areas of biological research. Getting a homogeneous sam-
ple is among others an essential requirement for the accuracy and
reproducibility of the experimental results, the correct interpreta-
tion of biophysical data, and high-resolution structural work. It will
also benefit protein interaction studies, as the presence of aggre-
gates may decrease the fraction of protein competent for binding or
lead to improper modeling of the interaction. Few methods allow
the characterization of protein homogeneity and protein binding
properties. It is the case of sedimentation velocity (SV) analytical
ultracentrifugation (AUC) which is a powerful biophysical tech-
nique commonly used to determine the size and shape of macro-
molecules in solution and to study protein interactions. It is an
absolute method based on first principles that benefits from con-
stant developments in analysis software. In SV, macromolecules
loaded in sector-shaped cells are separated at high centrifugal
force and their sedimentation profiles, the evolution of the protein

Arnaud Poterszman (ed.), Multiprotein Complexes: Methods and Protocols, Methods in Molecular Biology, vol. 2247,
https://doi.org/10.1007/978-1-0716-1126-5_9, © Springer Science+Business Media, LLC, part of Springer Nature 2021

155
156 Christine Ebel and Catherine Birck

concentration with time and radial position, are monitored in real

time using the optical system of the analytical ultracentrifuge.
Three optics are currently available, absorbance, interference, and
fluorescence optics which allow the detection of a variety of macro-
molecules in a broad range of concentrations. Compared to other
hydrodynamic techniques like dynamic light scattering (DLS) and
size exclusion chromatography, SV presents a higher hydrodynamic
resolution and a wider range of protein molecular weights to be
studied, from a few hundred daltons to several hundred-million
daltons, corresponding to peptides and large macromolecular com-
plexes, respectively. Furthermore, the presence of contaminants or
aggregates below 1% can be quantified using SV [1, 2]. The analysis
of the sedimentation profiles allows the determination of the sedi-
mentation and the diffusion coefficients of the macromolecules,
from the motion and shape of the sedimenting boundaries, and
the calculation of their molecular mass using the Svedberg equation
(see Subheading 1.1). For non-interacting species, these hydrody-
namic parameters are directly related to the size and shape of
macromolecules. SV can also contribute to our understanding of
macromolecular assemblies with the determination of the equilib-
rium constant governing the interaction and the size of the macro-
molecular complex, and a robust approach to obtain these
parameters is the isotherm analysis. Binding isotherms are derived
from sedimentation coefficients distributions of a concentration
series and fit with a hetero-association model to obtain the dissoci-
ation constant and sedimentation coefficients of the macromolecu-
lar complexes. Concerning the stoichiometry of the complex, it is
deduced from the quality of the fit using different association
models but may require a more complicated analysis using multi-
signal SV approach, especially in the case of multi-component
systems. The dynamic range in binding affinities that can be
explored in SV is notably high from picomolar to millimolar Kd
values due to the availability of the three optics and their different
sensitivity. It is noteworthy that the fluorescence detection system
that was recently developed allows now the study of protein inter-
actions at low picomolar concentrations [3].
The present chapter aims to provide a theoretical background
of SV, to detail the SV protocols to characterize the heterogeneity
of a protein, and to measure a binary protein–protein interaction.
Two case studies will be presented. As SV technology and expertise
are accessible to researchers via many research infrastructures (see
Note 1), an additional goal of this chapter is to contribute to make
known the SV applications and raise interest for this technique.
More deep descriptions of the theory and applications of SV can
be found in recent reviews [2, 4–7].
 Heterogeneity and Affinity Interaction Analysis by Sedimentation Velocity 157

1.1 Sedimentation Sedimentation velocity measures in a rotor spinning at high angular

Velocity Equations velocity, ω, in the centrifuge, the evolution of the weight concen-
tration, c, with time, t, and radial position, r. For each homoge-
neous ideal solute, and given the sector shaped cells used in
analytical ultracentrifugation, the transport is described by the
Lamm equation:


ð∂c=∂t Þ ¼ 1=r ∂=∂r r csω2 r  D ∂c=∂r , ð1Þ
where s and D are the sedimentation and diffusion coefficients of
the macromolecule. s is defined as the ratio of the macromolecule
velocity (cm s1) to the centrifugal field (ω2r in cm s2). s is
expressed in Svedberg unit S (1 S ¼ 1013 s). s and D are functions
of the molar mass M, the hydrodynamic radius RH (also referred to
as the Stokes radius RS), and the partial specific volume v of the
macromolecule. s and D also depend on the solvent density ρ and
viscosity η. The Svedberg equation relates s to RH (or D), M, and v:
s ¼ M ð1  ρv Þ=ðN A 6πηRH Þ ¼ M ð1  ρvÞD=RT ð2Þ
NA is Avogadro’s number and T the absolute temperature.
The Stokes–Einstein equation relates D to RH:
D ¼ RT =ðN A 6πηRH Þ ð3Þ
The ratio of RH to the minimum theoretical hydrodynamic
radius Rmin of non-hydrated volume, V, of the particle defines the
frictional ratio f/fmin:
V ¼ ð4=3ÞπRmin 3 ¼ M v=N A ð4Þ
RH ¼ ð f = f min Þ Rmin ð5Þ
The value of f/fmin depends on shape and hydration, but not,
for compact macromolecules, on the size or molar mass [8]. A
typical value is 1.25 for globular compact particles, 1.5 for a mod-
erately anisotropic shape and 1.8 for a significative anisotropic
shape (see e.g. [9]). Glycosylated proteins are characterized by
larger values, such 1.5–1.8, for a moderate asymmetrical shape
(see e.g. [10]). f/fmin can also be calculated from pdb files [11].
Because of its dependence on buffer viscosity and density, the
experimental s value (sexp) is often normalized to standard solution
conditions of water at 20  C (s20,w):



s 20,w ¼ s 1  ρ20,w v =ð1  ρvÞ η=η20,w ð6Þ
D is also often expressed as D20,w:


D 20,w ¼ D ðT 20 =T Þ η=η20,w ð7Þ
158 Christine Ebel and Catherine Birck

Non-ideality effects in concentrated samples influences s and

D [12]. The sedimentation and diffusion at infinite dilution, s0 and
D0, can be derived from the linear approximations:
s 1 ¼ s 0 1 ð1 þ ks c Þ ð8Þ
D ¼ D 0 ð1 þ kD c Þ ð9Þ

1.2 The c(s) Analysis During centrifugation, domination of the sedimentation term gives
rise to a migrating boundary that spreads with time because of an
opposing diffusional flow in response to the resulting concentra-
tion gradient [2, 13]. The time evolution of the radial concentra-
tion distribution during sedimentation is given by the Lamm
equation (Eq. 1). Numerical solutions to the Lamm equation are
used in many SV analysis programs to extract s and D from SV
experimental data.
The c(s) method implemented in the SV program SEDFIT
[14, 15] allows to obtain a high-resolution distribution of particle
according to their s-values. The c(s) analysis considers a large num-
ber of types of particles. It deconvolutes the effect of diffusion
broadening, assuming that all proteins have the same shape and
density (same v, f/fmin), which gives a relationship between s and D,
or, in an equivalent way, s and M, or s and RH (Eqs. 2–5). The
program simulates for each type of particle (each s-value), the set of
SV profiles, and allows to refine the value of f/fmin. It then deter-
mines the best combination of simulated SV profiles that fits the
experimental data. The resulting c(s) distribution shows peaks for
all sedimenting species in solution, assuming a constant shape.
Peaks of the c(s) plot can be easily integrated, providing their s-
value, their signal, and the percentage of each species.

1.3 Characterization It is noteworthy that SV data are highly sensitive to the presence of
of Protein trace aggregates, and a wide range of s-values can be covered in the c
Heterogeneity Using (s) distribution from a single experiment. The c(s) analysis allows to
SV determine if samples are heterogeneous. However, a c(s) plot with
only one peak does not mean that the sample is homogeneous. In
1.3.1 Non-interacting general, there are fewer peaks than species when particles are inter-
Versus Interacting System acting. For this reason, SV profiles should be acquired at different
concentrations. The nice superposition of the normalized c(s) plots
for protein samples at different concentrations indicates that the
protein does not self-associate. A shift in the peak position indicates
that the protein is in an equilibrium of association. When working
at high protein concentrations, above the mg/mL range, a slight
decrease of s-value is expected due to non-ideality effect (Eq. 8).
This is a complication in the analysis of weakly interacting
systems [12].
 Heterogeneity and Affinity Interaction Analysis by Sedimentation Velocity 159

1.3.2 Non-interacting When s-values are invariant with concentration, the sample may be
System Analysis considered as composed of non-interacting species, a peak in the c
(s) representing a species. The non-interacting species analysis then
allows the determination of independent values for s and D, thus
M and RH, if the system consists of a limited number of species
(typically one or two).

1.3.3 Example of a Bovine serum albumin (BSA) will be used as an example of

Non-interacting Protein non-interacting protein species. BSA is a 66-kDa protein, well-
Species known to form irreversible oligomers. The distribution of mono-
meric, dimeric, and trimeric species is commonly studied in AUC to
determine the instrument performance [16]. The SV data of two
BSA samples taken before or after a gel filtration step will be
analyzed to illustrate the behavior of heterogeneous versus homo-
geneous samples.

1.4 Characterization SV offers two significant advantages for the study of interacting
of Protein macromolecules: a relatively high hydrodynamic size-dependent
Hetero-association resolution allowing the separation of free and bound species and
Using SV the fact that during the experiment, the complexes will remain in a
bath of the free species. Thereby, dissociating complexes can
re-associate during the sedimentation process in a way that will
reflect their equilibrium and kinetic properties [17]. The standard
methods of analyzing SV data to obtain association constants
between interacting macromolecules are Lamm equation fitting
and c(s)-based isotherm analysis. Practically, the determination of
equilibrium and kinetic binding constants requires to run multiple
experiments at concentrations that cover a range of about 1/10 to
10 Kd. It is advisable to make dilution or titration series to prepare
samples. In this way, these values can be refined with added con-
straints linking the concentrations of different experiments. In
dilution series of stock mixtures, the molar ratio is constant and
can be fixed or refined as a single global parameter. In titration
series, one protein is kept at low constant concentration in each
sample while the concentration of the binding partner varies in a
wide range.

1.4.1 Lamm Equation Current computational strategies allow for direct boundary model-
Fitting to Characterize ing of SV data for reacting systems with a set of coupled Lamm
Protein Hetero-association equations describing all the species participating in the interaction
combined with information on the equilibrium association con-
stant and reaction kinetics [18, 19]. It is noteworthy that only
this approach allows to estimate the kinetic off-rate constant koff
of interactions (see Note 2). This direct boundary fitting method
has been implemented in SEDPHAT with different binding models
available. Compared to the SV isotherm analysis, this analysis can be
conducted when only a few SV data sets are available (minimum
160 Christine Ebel and Catherine Birck

two) and yield more precise parameter estimates about Kd, koff, and
scomplex. This approach has been demonstrated to be efficient in a
number of cases but has several limitations including the fact that
individual species need to be reasonably described as discrete spe-
cies with the correct molecular weights. This is not always the case,
for example, when impurities contribute to boundary broadening
and thus preclude a good model fitting and accurate results. Fur-
thermore, as this method will take some time to converge, it is not
the first method of choice, but may be conducted using parameter
values obtained with the isotherm analysis.

1.4.2 Isotherm Analysis For the construction of an isotherm, the first step is the c(s) analysis
to Characterize Protein of all SV data corresponding to the concentration series. In fact, the
Hetero-association c(s) analysis can also be applied to slowly as well as rapidly interact-
ing systems. For slowly interacting systems, the boundary pattern
directly reflects the population of different species, which can often
be hydrodynamically resolved. In the case of a bimolecular interac-
tion of the type A + B forming an AB complex with s-values such
that sA < sB < sAB, three boundaries will be observed corresponding
to free A, free B, and complex AB. For rapidly interacting systems,
the association-dissociation events during sedimentation lead to at
most two boundaries, one always sedimenting at the s-value of
either free A or free B, termed undisturbed boundary, and the
other at a composition-dependent s-value between sB and sAB,
termed the reaction boundary. When the integration limits encom-
pass all sedimenting species, the result is termed signal-weighted
average sedimentation coefficient, sw, which is directly related to the
overall transport of the interacting system and independent of the
kinetics of the interactions. When the integration limits encompass
only the reaction boundary, the result is termed sfast referring to the
sedimentation coefficient of the fast-sedimenting species. The
resulting sw and sfast isotherms can be combined and fitted with
mass action law models implemented in program SEDPHAT to
determine the binding constants and species size [19]. In addition
to these two sw and sfast isotherms, a population isotherm can be
constructed by integrating each c(s) peak and tabulating the
reported amplitudes, in signal units, and combined in the analysis.

1.4.3 Example of Protein We will use the association between the BTG2 factor and the poly
Hetero-association (A)-binding protein PABP C1 as model system to illustrate protein
hetero-association. These proteins were shown to interact directly
and their association sufficient in vitro and necessary in cellulo for
BTG2 to stimulate mRNA deadenylation. The BTG2/PABP inter-
action is also required for BTG2 to exert its antiproliferative
function [20].
 Heterogeneity and Affinity Interaction Analysis by Sedimentation Velocity 161

2 Materials

2.1 Protein Samples 1. BSA (SIGMA) was dissolved at 5 mg/mL in 20 mM Na-K

Phos 150 mM NaCl pH 7 and filtered at 45 μm. BSA1 sample
was obtained by five times dilution to 1 mg/mL. BSA2 sample
at 0.8 mg/mL was obtained from the main elution peak of
500 μL BSA stock sample injected at 0.5 mL/min on a Super-
ose 12 10/300 GL (GE Healthcare) size exclusion chromatog-
raphy column.
2. Recombinant BTG2 fused to glutathione S transferase
(GST-BTG2) and PAPB proteins were purified by affinity and
size exclusion chromatography [20]. Stocks of GST-BTG2
protein at 1.1 mg/mL and PABP at 5.1 mg/mL in buffer
10 mM K phosphate, pH 7.5, 100 mM NaCl, 0.5 mM TCEP
were used in this study (see Note 3). In the dilution series,
mixtures contain a fixed concentration of GST-BTG2 (5 μM)
and a range of seven PABP concentrations to give molar ratio
(PABP/GST-BTG2) of 1:2, 1:1, 2:1, 4:1, 8:1, 16:1, and 32:1.

2.2 Instrument 1. An analytical ultracentrifuge (Optima XLI Beckman) and rotor

and Sample Cells (8-hole AnTi-50, Beckman). Analytical ultracentrifugation cell
assemblies equipped with sapphire windows and two-channel
Titanium, Aluminum or Epon centerpieces of 1.5, 3, and/or
12 mm optical path length (Nanolytics or Beckman).
2. Programs: The program SEDNTERP created by D. Hayes,
T. Laue, J. Philo and available free (http://bitcwiki.sr.unh.
edu/index.php/Downloads) for calculating the parameters
relevant to SV analysis (see Note 4). The program SEDFIT
created by P. Schuck and available free (http://sedfitsedphat.
nibib.nih.gov), for the c(s) analysis of SV experiments [14]. The
program SEDPHAT created by P. Schuck and available free
(http://sedfitsedphat.nibib.nih.gov), for the global analysis of
different AUC [21] and other biophysical experiments
[22, 23], for the analysis of isotherms [17] and to generate
simulated data sets. The program GUSSI created by
C. Brautigam and available free (http://biophysics.swmed.
edu/MBR/software.html), for calculations (e.g., the integra-
tion of the c(s) peaks, the output of isotherm data files), and
publication quality illustrations of AUC experiments [24].

3 Methods

3.1 Data Acquisition 1. Prepare the analytical ultracentrifugation cells assemblies. Fill
the sample compartments with 400 μL (12 mm path length
centerpiece) or 100 μL (3 mm path length centerpiece) of
protein solutions and the solvent compartments with 420 or
162 Christine Ebel and Catherine Birck

110 μL of buffer (see Note 5). Close the cells and check that
cells that will be placed in opposite positions in the rotor have
the same weights (tolerance of 0.5 g). Place and align the cells
in the rotor holes.
2. Prepare the instrument: Put the rotor then the optical arm into
the analytical ultracentrifuge, and initiate vacuum and temper-
ature equilibration. Wait at least 2 h at the requested tempera-
ture before starting the SV experiment.
3. Check the absorbance signal in each cell. During temperature
equilibration, start the program ProteomeLab XL-I. Give a
title for each cell and a common folder for the collected data.
Start the AUC at 3000 rpm and acquire for each cell a radial
scan at 280 nm to check that the absorbance signal corresponds
to expectation.
4. Start the SV run: In the method window, enter 150 for the
number of scans. In the options window, enter 3 for the last
scans overlay and select stop XL after the last scan. Start the
centrifuge at the chosen speed (see Note 6) then start the
method scan. After the stop of the ultracentrifuge, cells are
disassembled, cleaned, and reassembled. Raw data are copied
for analysis. A detailed protocol can be found in [8].

3.2 Non-interacting We perform heterogeneity analysis of two BSA samples using both c
System Data Analysis (s) and non-interacting species models to illustrate the potential
with BSA Data Sets and limits of these approaches. The first sample, BSA1, which
contains aging aggregates is heterogeneous while the second,
BSA2, freshly prepared from size exclusion chromatography and
composed of monomeric BSA, is homogenous.
1. The c(s) analysis: Open the program SEDFIT and load a set of
SV scans corresponding to the whole sedimentation process
(Fig. 1). Set the meniscus with the red line (air-sample inter-
face) and the radial limits for the fit with the green lines.
Choose the continuous c(s) distribution model. In the parame-
ter window, enter the experimental values for partial specific
volume, density, and viscosity, change the confidence level to
0.68 (see Note 7) and mark the frictional ratio and meniscus
check boxes. Use the default values for all other parameters
(100 for resolution, 0–20 S for sedimentation coefficient
range, 1.2 for frictional ratio value) and press the run command
to perform a simulation of the sedimentation process using the
given parameters. The simulated data are displayed as lines with
the loaded scans, and a first c(s) distribution plot appears. Then
press the fit command to optimize the checked parameters
(frictional ratio and meniscus position). After convergence,
simulated data should fit very well to the experimental data
 Heterogeneity and Affinity Interaction Analysis by Sedimentation Velocity 163

Fig. 1 SV data and c(s) analysis of BSA. Superposition of the experimental and fitted sedimentation velocity
profiles for BSA before (a) and after (b) size exclusion chromatography—samples BSA1 and BSA2, respec-
tively. Residuals of the fit are displayed in gray or absorbance scales. Corresponding c(s) distributions are
shown below each SV data. Sedimentation velocity profiles were obtained at 280 nm, at 20  C, and
130,000  g (42,000 rpm) using double-sector centerpieces with 12 mm optical path length. Analysis was
performed with the first 100 profiles, collected during 350 min. The c(s) analysis was obtained considering
200 particles with s-values in the 0–15 S range and a partial specific volume of 0.73 mL/g solvent density of
1.005 g/mL and a viscosity of 1.02 cp. A confidence level of 0.68 was used for the regularization procedure.
The rmsd were 0.006 and 0.009 for the BSA1 and BSA2 samples, respectively. The fitted frictional ratio were
1.25 and 1.33, respectively

(criteria for the quality of the fit are low rmsd value, less than
0.01 signal units under most cases, and randomly distributed
residuals around zero) to get the true best-fit distribution c(s).
Integration of the peaks in the c(s) distribution will give the
values of the different molecular parameters. Detailed steps of
the c(s) analysis can be found in SEDFIT help website (www.
analyticalultracentrifugation.com/sedfit_help.htm). Figure 1,
panels A and B, shows a nice fit for the two BSA samples.
When BSA is diluted, the sample consist of a mixture of mono-
mers, dimers, trimers, and traces of higher oligomers which can
164 Christine Ebel and Catherine Birck

be easily quantified by integration of the peaks. After size

exclusion chromatography, the sample is homogeneous, with
only one species in solution at 4.3 S, corresponding to the
globular compact monomer. Note that to ascertain the fact
that this peak corresponds to a non-interacting species, we
should have measured the samples at different concentrations.
Here we obtain this conclusion because the main species in the
heterogeneous samples have the same s-value.
2. The non-interacting species analysis: in SEDFIT, choose the
non-interacting discrete species model. In the parameter win-
dow, enter estimate values for the total concentration, the
molar mass, and s-value of the species and mark the
corresponding check boxes. Press the run command then the
Fit command. Figure 2 shows the result of an analysis in terms
of one non-interacting species for the two samples. This choice
is done for pedagogical purpose, since the c(s) analysis clearly
showed that diluted BSA (without purification) is heteroge-
neous. Figure 2 shows that the fit for the homogeneous sample
is excellent: the residuals are randomly null. The calculated
molar mass is 65.9 kDa, in very good agreement with the
theoretical—and checked by mass spectrometry—value of
66.4 kDa. On the other hand, a non-cautious analysis could
describe a nearly good superposition of the experimental and
fitted SV profiles for the heterogeneous sample in Fig. 2a. But
the residuals show clearly non-random deviations. And,

Fig. 2 SV data and analysis of BSA using a non-interacting species model. Superposition of the experimental
and fitted SV profiles for BSA, before (a) and after (b) size exclusion chromatography—samples BSA1 and
BSA2, respectively—using one non-interacting species model. Residuals of the fit are displayed in gray or
absorbance scales. Experimental conditions are detailed in the legend of Fig. 1. Sedimentation coefficients of
4.7 and 4.3 S with the corresponding molar masses of 27.4 and 65.9 kDa and rmsd on fits of 0.018 and 0.009
were obtained for BSA1 and BSA2 samples, respectively
 Heterogeneity and Affinity Interaction Analysis by Sedimentation Velocity 165

because the model is inappropriate, the calculated molar mass

of 27.4 kDa is definitively wrong, half the theoretical value for
the monomer. This is because boundary spreading is inter-
preted by the program—in the framework of the incorrect
model—as caused by diffusion, while heterogeneity affects
more predominantly the shape of the boundaries. In conclu-
sion, the non-interacting species model provides valuable
results, but only when this model is appropriate.

3.3 Hetero- We illustrate here c(s) analysis in SEDFIT and integration of the
association Data sedimentation coefficient distributions in GUSSI to determine sw
Analysis and sfast values, reflecting the properties of the interacting system.
with BTG2-PABP Isotherm analysis in SEDPHAT.
Mixtures Data Sets 1. Perform a c(s) analysis in SEDFIT with SV data for each mix-
ture of the titration series and save the corresponding c(s)
distribution in GUSSI format. Open the program SEDFIT
and perform the c(s) analysis with SV data for the first mixture.
Excellent fit of the raw data is essential for the accuracy of
parameters determined in next steps. The c(s) distribution is
exported to GUSSI using Plot/Gussi c(s) plot, then accept the
GUSSI terms in the GUSSI window that appears, select File/
Save data only, choose the correct directory, give a name to the
file and save. A file *.gcofs is created. Then File/quit. Proceed
the same way for all SV data.
2. Superimpose all the c(s) distributions corresponding to the
titration series and transform experimental c(s) in corrected c
(s) distributions (Fig. 3). Open GUSSI, accept terms and press
OK in the GUSSI window with default c(s) selection. In the
GUSSI c(s) module window, select File/Add Distribution and
load the GUSSI file *.gcofs corresponding to the first c(s)
distribution (corresponding to the lowest concentration of
A). Repeat the loading procedure in ascending order of con-
centration for all the c(s) distributions. Select Axis/Standardi-
zation/Standardization, then Axis/Standardization/Modify
Standardization Parameters, and enter experimental conditions
in Standardization window for density, viscosity, partial specific
volume (values at experimental temperature and 20  C), and
temperature. Then press propagate to all distributions and
finally press commit. The experimental c(s) distributions are
transformed to corrected c(s) distributions. Then File/Save
GUSSI State to save the corrected c(s) superimposition file.
This *.gussi file can be recalled if needed.
3. Integrate all the superimposed corrected c(s) distributions to
generate the signal-weighted average sw isotherm file (Fig. 4).
In GUSSI, select Integration/Make Isotherm/Hetero/sw,
press no for the exclusion zone selection (see Note 8), and
166 Christine Ebel and Catherine Birck

Fig. 3 SV data and c(s) analysis of GST-BTG2: PABP mixtures. (a) Experimental SV data, fit, and residuals for a
mixture of 5 μM GST-BTG2 and 20 μM PABP. Individual data points are shown as circles and fits to
experimental data as lines. Residuals of the fit are displayed in absorbance scales. Sedimentation velocity
profiles were obtained at 280 nm, at 4  C, and 182,000  g (50,000 rpm) (b) Sedimentation coefficient
distributions c(s) derived from SV data of a titration series of 5 μM GST-BTG2 with 2.5, 5, 10, 20, 40, 80, and
160 μM PABP. The vertical dotted line indicates the s-value of GST-BTG2 determined in a separate experiment
(s ¼ 4.49 S). For better clarity of the reaction boundary, the y-axis was truncated at 2.1 value. To optimize
absorbance measurements, centerpieces with a 12 mm optical path length were used to analyze mixtures
with 2.5, 5, 10, 20, and 40 μM PABP, and centerpieces with a 3 mm optical path length for ones with 80 and
160 μM PABP. The sw isotherm was generated using integration of the c(s) overlays from 1 to 7 S and sfast
isotherm using integration from 3.2 to 7 S corresponding to the reaction boundary

4.5 5.8
5.6
4.0
5.4
sw fast (S)
sw (S)

3.5 5.2
5.0
3.0
4.8

2.5 4.6
residuals

residuals

0.04 0.03
0.01 –0.01
–0.03 –0.04
10-6 10-5 10-4 10-3 10-6 10-5 10-4 10-3
concentration (M) concentration (M)

Fig. 4 Binding isotherms of sedimentation coefficients sw and sfast. The two isotherms were extracted from the
sedimentation coefficient distributions shown in Fig. 3. Isotherm of sw is derived from the integration of all c(s)
peaks while sfast is from only the fast c(s) peak, corresponding to the reaction boundary. The global analysis of
both isotherms was performed in SEDPHAT using a hetero-association model with: M ¼ 80,860 Da,
s ¼ 4.48 S, and ε280nm ¼ 122,620 M1 cm1 for species A (GST-BTG2) and M ¼ 22,387 Da, s ¼ 2.20 S,
and ε280nm ¼ 15,930 M1 cm1 for species B (PABP). Starting values of logKa ¼ 5 and sAB ¼ 5.5 S yielded a
best-fit for Kd of 20 μM (95% confidence interval: 14–29 μM) with refined values for sA ¼ 4.49 S, sB ¼ 2.36 S,
and sAB ¼ 5.85 S. These values are in agreement with initial fitting values and observed peak positions in the c
(s) overlays (see Note 11)
 Heterogeneity and Affinity Interaction Analysis by Sedimentation Velocity 167

OK to the sw picking mode then pick twice to define the s-

values of the integration. In the sw isotherm constructor win-
dow, select save xp, save SEDPHAT, and fill the values of A and
B concentrations for each corrected c(s) distribution, the molar
extinction coefficient (xt component) of A and B, and values
for sA, sB, sAB, and log(Ka). Press Save, give a name to the file,
and save. In fact, 3 files with the same name but different
extensions will be created, the sw isotherm file (sw*.isotherm),
the SEDPHAT experiment file (sw*.xp), and the SEDPHAT
configuration file (sw*. sedphat).
4. Integrate all the superimposed corrected c(s) distributions to
generate EPT swfast isotherm file (isotherm of signal-weighted
s-value of the reaction boundary) (Fig. 4). In GUSSI, select
Integration/Make Isotherm/Hetero/EPT swfast, press OK to
the EPT swfast picking mode, then pick twice to define the s-
values of the integration of the reaction boundary. In the EPT
swfast isotherm constructor window, follow the same proce-
dure as for the sw isotherm constructor window to create
swfast* files.
5. Isotherm analysis in SEDPHAT (Fig. 4). Open SEDPHAT, in
Data/Load Experiment, open sw*.xp and swfast*.xp files,
enter the vbar value at 20  C, OK, then Data/Read Configu-
ration from file, open sw*.sedphat. In the SEDPHAT A+B <->
AB model window that pops up, enter the molar masses for A
and B, and select the parameters sAB and log(Ka), OK. Choose
Run/Global Run, if the initial values in the model are not too
fare from the experimental data, Fit/Global Fit (see Note 9).
Then, in Global Parameters, select sA and sB, then Fit/Global
Fit. Evaluate the quality of the fit and the refined values for sA,
sB, sAB, and logKa (see Note 10). Choose Statistics/Automatic
confidence interval, and search w projection method to calcu-
late the confidence intervals of logKa. Press OK in the two
confidence search windows that pop up, then uncheck the
logKa parameter. Press CANCEL in the edit experimental
parameters window and OK in the constrained parameter win-
dow. In the confidence search window that pops up, enter the
P value of 0.95 and accept all the default parameters. At the end
of the calculations, a message box appears with the confidence
limits of logKa.

4 Notes

1. For example, the European Instruct-Eric (https://instruct-

eric.eu/) and French FRISBI (http://frisbi.eu/) infrastruc-
tures provide open access to AUC instruments and expertise.
168 Christine Ebel and Catherine Birck

2. The effective particle theory (EPT) explains the behavior of

sedimentation boundary patterns arising in rapidly interacting
systems [25]. Zhao and colleagues [17] have showed that the
regime (slow or fast) of the association equilibrium in SV
depends on the kinetic off-rate constant (see Fig. 7 from 17).
An equilibrium with log(koff) of 1 can be considered infinitely
fast in the context of SV. When log(koff) ¼ 3, the equilibrium
is still fast but on the limit where kinetics should significantly
influence the boundary pattern. A value of 3.4 corresponds to
an intermediate regime, a value of 4 corresponds to the slow
regime but still influenced by kinetics, and a value of 6 means
an essentially infinitely slow regime on the SV time-scale.
3. On Day D—1 before SV experiment, GST-BTG2 was loaded
on a gel filtration column equilibrated in SV buffer to eliminate
traces of aggregates, and PABP was dialyzed overnight against
the same buffer. As TCEP is not very stable in phosphate
buffers, the SV buffer was prepared immediately before use.
Data showed that TCEP is 100% oxidized within 72 h in PBS
pH 7 and 50% oxidized within 72 h in PBS pH 8 compared to
only 20% oxidized within 3 weeks in Tris or Hepes buffers in a
broad range of pH.
4. The program SEDNTERP allows to calculate the molar mass,
partial specific volume, and extinction coefficient of proteins
from their amino acid compositions, and the solvent density
and viscosity from the solvent composition using tabulated
data. For unusual solvents, these latter parameters should be
measured experimentally using a density-meter and a viscosity-
meter.
5. Filling the sample compartment with a larger volume of sample
will allow to collect a larger number of scans and therefore
extract more information from the SV data. Up to 450 μL of
sample can be loaded in the cell compartment with a 12 mm
centerpiece. The volume of buffer has to match the volume of
sample in case of interference data acquisition and be equal or
higher in the case of absorbance data acquisition. There is no
need of a blank buffer sector in the case of fluorescence data
acquisition which allows both sectors of a double-sector cen-
terpiece to be loaded with different samples.
6. Using the AN-50 Ti rotor, the maximum rotor speed is
50,000 rpm (201,600  g). For proteins with M < 300 kDa,
SV runs at 50,000 rpm are recommended to optimize hydro-
dynamic resolution. For M > 300 kDa, SV runs at lower speed
(40,000 rpm or less) are performed to maximize the number of
scans that will be collected before complete pelleting of the
macromolecule [21, 26].
 Heterogeneity and Affinity Interaction Analysis by Sedimentation Velocity 169

7. Use P ¼ 0.5 for the analysis without regularization procedure.

Use P value of 0.68 or 0.95 (confidence level of one or two
standard deviations, respectively) for analysis with regulariza-
tion. Increasing the P value will decrease the resolution of the
distribution but increase the accuracy of its information. Sedi-
menting species that will disappear in the c(s) distribution at
higher P-value should be interpreted with care.
8. In the presence of a contaminant species that does not partici-
pate in the interaction, it is possible with the program GUSSI
to define an s-value range that will be excluded from the calcu-
lation of the sw values [24].
9. A poor fit may indicate that the interaction model is incorrect.
Before drawing such conclusion, carefully verify there is no
error in the entered parameters (molar mass, molar extinction
coefficients, sA and sB values).
10. The refined values for sA, sB, and sAB should be in agreement
with values observed in the c(s) distributions of individual
components and titration series. If parameters are not well
restrained by experimental data, refined values may be unreal-
istic. In this case, sA and sB should be fixed in the
refinement [19].
11. It is noteworthy that GST-BTG2 (species A) is a dimer in
solution, and we can expect two independent binding sites
for PABP (species B). However, due to the low affinity of the
complex and the concentrations used, it is not possible to
determine the sedimentation coefficient and Kd for the ABB
species.

Acknowledgments

The authors acknowledge the support and the use of resources of

the French Infrastructure for Integrated Structural Biology FRISBI
ANR-10-INBS-05 and of Instruct-ERIC: the platforms of the
Grenoble Instruct-ERIC center (ISBG: UMS 3518 CNRS-CEA-
UGA-EMBL) within the Grenoble Partnership for Structural Biol-
ogy (PSB) and the Strasbourg Instruct-ERIC center (Centre de
Biologie Intégrative, CBI) within IGBMC (CNRS UMR 7104-
Inserm U 1258-Université de Strasbourg). CE also acknowledges
the support of the COST action CA 15126 MOBIEU and GRAL
financed within the University Grenoble Alpes graduate school
(Ecoles Universitaires de Recherche) CBH-EUR-GS (ANR-17-
EURE-0003).
170 Christine Ebel and Catherine Birck

References
1. Arthur KK, Kendrick BS, Gabrielson JP (2015) 15. Dam J, Schuck P (2004) Calculating sedimen-
Guidance to achieve accurate aggregate quan- tation coefficient distributions by direct mod-
titation in biopharmaceuticals by SV-AUC. eling of sedimentation velocity concentration
Methods Enzymol 562:477–500 profiles. Methods Enzymol 384:185–212
2. Correia JJ, Stafford WF (2015) Sedimentation 16. Zhao H, Ghirlando R, Alfonso C, Arisaka F,
velocity: a classical perspective. Methods Enzy- Attali I, Bain DL, Bakhtina MM, Becker DF,
mol 562:49–80 Bedwell GJ, Bekdemir A, Besong TMD,
3. Zhao H, Mayer ML, Schuck P (2014) Analysis Birck C, Brautigam CA, Brennerman W,
of protein interactions with picomolar binding Byron O, Bzowska A, Chaires JB, Chaton CT,
affinity by fluorescence-detected sedimentation Cölfen H, Connaghan KD, Crowley KA,
velocity. Anal Chem 86:3181–3187 Curth U, Daviter T, Dean WL, Dı́ez AI,
4. Kingsbury JS, Laue TM (2011) Fluorescence- Ebel C, Eckert DM, Eisele LE, Eisenstein E,
detected sedimentation in dilute and highly England P, Escalante C, Fagan JA, Fairman R,
concentrated solutions. Methods Enzymol Finn RM, Fischle W, de la Torre JG, Gor J,
492:283–304 Gustafsson H, Hall D, Harding SE, Cifre
JGH, Herr AB, Howell EE, Isaac RS, Jao
5. Schuck P (2013) Analytical ultracentrifugation S-C, Jose D, Kim S-J, Kokona B, Kornblatt
as a tool for studying protein interactions. Bio- JA, Kosek D, Krayukhina E, Krzizike D, Kusz-
phys Rev 5:159–171 nir EA, Kwon H, Larson A, Laue TM, Le
6. Le Roy A, Wang K, Schaack B, Schuck P, Roy A, Leech AP, Lilie H, Luger K, Luque-
Breyton C, Ebel C (2015) AUC and small- Ortega JR, Ma J, May CA, Maynard EL,
angle scattering for membrane proteins. Meth- Modrak-Wojcik A, Mok Y-F, Mücke N,
ods Enzymol 562:257–286 Nagel-Steger L, Narlikar GJ, Noda M,
7. Unzai S (2018) Analytical ultracentrifugation Nourse A, Obsil T, Park CK, Park J-K, Pawelek
in structural biology. Biophys Rev 10:229–233 PD, Perdue EE, Perkins SJ, Perugini MA,
8. Salvay AG, Communie G, Ebel C (2012) Sedi- Peterson CL, Peverelli MG, Piszczek G,
mentation velocity analytical ultracentrifuga- Prag G, Prevelige PE, Raynal BDE,
tion for intrinsically disordered proteins. Rezabkova L, Richter K, Ringel AE,
Methods Mol Biol 896:91–105 Rosenberg R, Rowe AJ, Rufer AC, Scott DJ,
9. Tabarani G, Reina JJ, Ebel C, Vivès C, Lortat- Seravalli JG, Solovyova AS, Song R,
Jacob H, Rojo J, Fieschi F (2006) Mannose Staunton D, Stoddard C, Stott K, Strauss
hyperbranched dendritic polymers interact HM, Streicher WW, Sumida JP, Swygert SG,
with clustered organization of DC-SIGN and Szczepanowski RH, Tessmer I, Toth RT,
inhibit gp120 binding. FEBS Lett Tripathy A, Uchiyama S, Uebel SFW,
580:2402–2408 Unzai S, Gruber AV, von Hippel PH,
Wandrey C, Wang S-H, Weitzel SE, Wielgus-
10. Dach I, Olesen C, Signor L, Nissen P, le Kutrowska B, Wolberger C, Wolff M, Wright E,
Maire M, Møller JV, Ebel C (2012) Active Wu Y-S, Wubben JM, Schuck P (2015) A mul-
detergent-solubilized H+,K+-ATPase is a tilaboratory comparison of calibration accuracy
monomer. J Biol Chem 287:41963–41978 and the performance of external references in
11. Rocco M, Byron O (2015) Computing trans- analytical ultracentrifugation. PLoS One 10:
lational diffusion and sedimentation coeffi- e0126420
cients: an evaluation of experimental data and 17. Zhao H, Balbo A, Brown PH, Schuck P (2011)
programs. Eur Biophys J 44:417–431 The boundary structure in the analysis of rever-
12. Chaturvedi SK, Ma J, Brown PH, Zhao H, sibly interacting systems by sedimentation
Schuck P (2018) Measuring macromolecular velocity. Methods 54:16–30
size distributions and interactions at high con- 18. Brautigam CA (2011) Using Lamm-equation
centrations by sedimentation velocity. Nat modeling of sedimentation velocity data to
Commun 9:4415 determine the kinetic and thermodynamic
13. Lebowitz J, Lewis MS, Schuck P (2002) Mod- properties of macromolecular interactions.
ern analytical ultracentrifugation in protein sci- Methods 54:4–15
ence: a tutorial review. Protein Sci 19. Zhao H, Brautigam CA, Ghirlando R, Schuck
11:2067–2079 P (2013) Overview of current methods in sedi-
14. Schuck P (2000) Size-distribution analysis of mentation velocity and sedimentation equilib-
macromolecules by sedimentation velocity rium analytical ultracentrifugation. Curr
ultracentrifugation and lamm equation model- Protoc Protein Sci. Chapter 20, Unit20.12
ing. Biophys J 78:1606–1619
 Heterogeneity and Affinity Interaction Analysis by Sedimentation Velocity 171

20. Stupfler B, Birck C, Seraphin B, Mauxion F stoichiometry and affinity in SEDPHAT. Acta
(2016) BTG2 bridges PABPC1 RNA-binding Crystallogr D Biol Crystallogr 71:3–14
domains and CAF1 deadenylase to control cell 24. Brautigam CA (2015) Calculations and
proliferation. Nat Commun 7:10811 publication-quality illustrations for analytical
21. Balbo A, Minor KH, Velikovsky CA, Mariuzza ultracentrifugation data. Methods Enzymol
RA, Peterson CB, Schuck P (2005) Studying 562:109–133
multiprotein complexes by multisignal sedi- 25. Schuck P (2010) Sedimentation patterns of
mentation velocity analytical ultracentrifuga- rapidly reversible protein interactions. Biophys
tion. Proc Natl Acad Sci U S A 102:81–86 J 98:2005–2013
22. Zhao H, Piszczek G, Schuck P (2015) SED- 26. Demeler B (2010) Methods for the design and
PHAT--a platform for global ITC analysis and analysis of sedimentation velocity and sedimen-
global multi-method analysis of molecular tation equilibrium experiments with proteins.
interactions. Methods 76:137–148 Curr Protoc Protein Sci. Chapter 7, Unit 7.13
23. Zhao H, Schuck P (2015) Combining biophys-
ical methods for the analysis of protein complex
 Chapter 10

Hands on Native Mass Spectrometry Analysis

of Multi-protein Complexes
Stéphane Erb, Sarah Cianférani, and Julien Marcoux

Abstract
By maintaining intact multi-protein complexes in the gas-phase, native mass spectrometry provides their
molecular weight with very good accuracy compared to other methods (typically native PAGE or
SEC-MALS) (Marcoux and Robinson, Structure 21:1541–1550, 2013). Besides, heterogeneous samples,
in terms of both oligomeric states and ligand-bound species can be fully characterized. Here we thoroughly
describe the analysis of several oligomeric protein complexes ranging from a 16 ¼ kDa dimer to a 801-kDa
tetradecameric complex on different instrumental setups.

Key words Structural mass spectrometry, Noncovalent mass spectrometry, Oligomeric states, Stoi-
chiometry, Subcomplexes

1 Introduction

Mass spectrometry (MS) plays a pivotal role for the characterization

of multi-protein complexes. Routinely coupled with reversed-phase
liquid chromatography (LC-MS), it allows purity and homogeneity
assessment of biological complexes. Classical LC-MS workflow
requires organic solvents (usually H2O/acetonitrile) and acidic
conditions (e.g., acetic or formic acid) to achieve best
chromatographic resolutions along with optimal MS detection.
However, those experimental conditions induce dissociation of
multi-protein noncovalent complexes. In the early 1990s, MS
approaches performed in ammonium acetate buffer maintaining
quaternary structures of multi-protein complexes have been first
described [1–3], opening the way for a new field of application
called noncovalent or “native” MS [4, 5]. By transferring intact
noncovalent assemblies into the gas phase of the mass spectrome-
ter, native MS enables multi-protein complex stoichiometry assess-
ment thanks to accurate mass measurements. Mostly performed on
TOF and Q-TOF [6] instruments in its early years, it has recently
benefited from technological improvements to reach high-

Arnaud Poterszman (ed.), Multiprotein Complexes: Methods and Protocols, Methods in Molecular Biology, vol. 2247,
https://doi.org/10.1007/978-1-0716-1126-5_10, © Springer Science+Business Media, LLC, part of Springer Nature 2021

173
174 Stéphane Erb et al.

resolution native MS [7]. Native MS is currently implemented in

most structural biology departments to complement conventional
structural biophysical techniques like crystallography, electron
microscopy, NMR, or SAXS. The methodology is versatile as any
type of multi-protein complex can be analyzed, ranging from
homo-oligomeric proteins [4] to hetero-oligomers [8, 9] or even
protein/nucleic assemblies [10, 11] and membrane complexes
[12–14]. Latest methodological breakthroughs in the field include
the direct analysis of complexes from cell lysates [15] or from native
membranes [13] and the release of a new instrument enabling the
complete analysis of very large complexes with unprecedented
resolution [16].
In this chapter, we illustrate the possibilities of native MS on
different types of multi-protein complexes and different MS plat-
forms to assess stoichiometries. Data interpretation is usually per-
formed by comparing the masses of the individual subunits of the
complexes obtained by classical MS analysis in denaturing condi-
tions and the mass of the intact noncovalent multi-protein assembly
measured in native conditions.

2 Materials

Prepare all solutions using deionized water and analytical grade

reagents. Prepare and store all reagents at 4  C.

2.1 Desalting Ammonium acetate: 200 mM solution in water (see Note 1). Weigh
1.54 g ammonium acetate (>98%) and transfer to a 100-mL cylin-
der. Add water to a volume of 95 mL. Mix and adjust pH to 7.4
with ammonium hydroxide (see Note 2). Make up to 100 mL with
water.

2.2 Cesium Iodide Resuspend cesium iodide powder in 50/50 isopropanol/water

Calibration Mix (v/v) to a final concentration of 2 mg/mL.

2.3 Ammonium Resuspend ammonium hexafluorophosphate powder in 50/50 iso-

Hexafluorophosphate propanol/water (v/v) to a final concentration of 1 mg/mL.
Calibration Mix

2.4 GroEL GroEL must be freshly prepared.

Reconstitution
1. Prepare 9.445 mL of 1 M Tris-acetate by weighting 1.21 g of
Tris-base and top up to 10 mL with acetic acid.
2. Prepare 10 mL of reconstitution buffer, composed as follows:
200 μL Tris-acetate at 1 M previously prepared, 50 μL EDTA at
100 mM, 10.7 mg magnesium acetate tetrahydrate, 5.5 mg
ATP, 37.3 mg potassium chloride to reach final concentrations
of 20 mM, 0.5 mM, 5 mM, 1 mM and 50 mM, respectively.
 Native Mass Spectrometry 175

3. Prepare a 200 mM ammonium acetate solution pH 6.9 by

weighting 770.8 mg in 50 mL.
4. Reconstitute GroEL (Chaperonin 60 from Escherichia coli,
Sigma-Aldrich reference no. C7688) powder (1 mg) by adding
80 μL of refolding buffer and 20 μL of methanol directly into
the vial.
5. Shake for 2 h at room temperature.
6. Transfer the entire volume into an Eppendorf, add 100 μL
acetone, and agitate gently.
7. After 2 min, centrifuge for 1 min at 11,000  g: a white
precipitate should be present. Delete the supernatant. Then,
add 200 μL of refolding buffer and agitate gently until com-
plete dissolution.
8. Shake for 1 h at room temperature. Centrifuge for 1 min at
1,500  g and take the supernatant to perform the first desalt-
ing step as explained in Subheading 3.1 (on Zeba™ spin
column).
9. The second desalting step is performed as explained in Sub-
heading 3.2 with 10 cycles (each one takes between 8 and
15 min) at 1.9  g with a 50-kD MWCO membrane. Finally,
check the protein concentration by UV absorbance at 280 nm.
Typically, approximately 10 μM of 14-mer should be obtained
(ε ¼ 10,430 M1 cm1 for GroEL monomer at 280 nm).

3 Methods

3.1 Benchtop Size 1. Operate at +4  C.

Exclusion 2. You can use 6 K MWCO Micro Bio-Spin™ 6 (Biorad) or
Chromatography 0.5 mL 7 K MWCO Zeba™ (Thermo Fisher Scientific) spin
Desalting columns.
3. Remove the storage solution by centrifugation for 2 min at
1,000  g for Micro Bio-Spin™ columns and 1 min at
1,500  g for Zeba™ columns.
4. Equilibrate the columns by adding 500 μL (Micro Bio-Spin™)
or 300 μL (Zeba™) of 200 mM ammonium acetate solution,
centrifuge for 1 min at 1,000  g (Micro Bio-Spin™) or at
1,500  g (Zeba™) and discard the eluate.
5. Repeat step 3 three times.
6. Load between 20 and 75 μL (Micro Bio-Spin™) or 30 and
130 μL (Zeba™) of sample. Loading smaller volume may affect
recovery (see Note 3).
176 Stéphane Erb et al.

7. Place the column in a 1-mL Eppendorf tube and elute the

desalted proteins by centrifuging for 4 min at 1,000  g
(Micro Bio-Spin™) or 2 min at 1,500  g (Zeba™).

3.2 Ultrafiltration 1. Operate at +4  C.

Desalting 2. After selecting the appropriate MWCO of the ultrafiltration
device, equilibrate its membrane with 500 μL of 200 mM
ammonium acetate. Typical devices include Amicon™ Ultra
0.5 mL or Vivaspin™ 500 centrifugal concentrators (Merck).
Centrifuge at 10,000  g for 30 s, discard the eluate, and repeat
three times.
3. Load your sample (up to 500 μL).
4. Concentrate down to 50 μL at 10,000  g. Make up to 500 μL
with 200 mM ammonium acetate and repeat five times this
dilution/concentration cycle.

3.3 Direct Infusion of 1. In the interface setting panel, set the controller power On and
the Sample with the the temperature to 4  C (see Note 4).
TriVersa Nanomate™ 2. Load 10 μL of desalted protein sample to the 96- or 384-well
(Advion) plate.
3. In the Spray Optimization panel, set the Sample Volume to
5 μL, the gas pressure to 0.35 psi, the Voltage to apply to
1.60 kV and press “Deliver Sample” (see Note 5).

3.4 Calibration 1. Using the TriVersa Nanomate™ (see Subheading 3.3), deliver
5 μL of cesium iodide calibration mix (see Subheading 2.2).
3.4.1 TOF LCT (Upgraded
for High Mass Detection by 2. In MassLynx MS Tune window, manually set the following
MS Vision) parameters: Operate the instrument in positive mode with an
m/z range from 1,000 to 10,000. Apply 120 V and 5 V for the
sampling cone and extraction cone, respectively. The source
temperature is set at 90  C, but the capillary voltage and the gas
flows are not used in the TriVersa Nanomate™ interface setup.
3. Adjust the backing pressure to 6 mbar (see Note 6).
4. Start a 2-min acquisition with 4-s scans.
5. Average the signal over the 2 min in the Chromatogram panel
and in the Spectrum panel, select “Tools,” then “Make Cali-
bration,” and select “CsI_esi.ref” profile. Adjust calibration
points for the m/z range targeted and click on File\Save Spec-
trum, then click OK and finally accept the calibration after
closing the window.
6. Do not forget to load this calibration file (e.g., named CsI_-
date) before starting the first acquisition of the analyzed sam-
ple. To do so, click on “Acquire,” then select “calibration,” and
load the calibration file. Finally perform the acquisition.
 Native Mass Spectrometry 177

3.4.2 Q-TOF Synapt G2 1. Using the TriVersa Nanomate™ (see Subheading 3.3), deliver
and G2Si 5 μL of cesium iodide calibration mix (see Subheading 2.2).
2. In MassLynx MS Tune windows, manually set the following
parameters: Operate the instrument in positive and sensitivity
modes with an m/z range from 1,000 to 8,000.
3. Nanoflow+ panel: Sampling Cone: 150 V. Source Offset: 30 V.
Source Temperature: 80  C. The Capillary Voltage and the gas
flows are not used in the TriVersa Nanomate™ interface setup.
4. Instrument: Trap and Transfer Collision Energy are turned off
(4 V and 2 V, respectively, by default).
5. Save these parameters as an .ipr file.
6. Press “acquire” to start a 2-min TOF-MS acquisition in the
1,000–8,000 m/z range.
7. Average the signal over the 2 min in the Chromatogram panel
and in the Spectrum panel, select Process\Automatic Peak
Detection with the “Set Peak Detection Parameters Automati-
cally” option activated.
8. Click on File\Save Spectrum and click OK.
9. In the Acquity UPLC Console, click on Intellistart, select
“Create Calibration,” and click Start.
10. In the calibration profile editor, select the right Mass Calibra-
tion Profile (here CsI in the 1,000–8,000 m/z range), right
click, and reset it.
11. Right Click and edit the Mass Calibration Profile. Below the
Positive Polarity chart, click on edit. Select the right Reference
Compound (here CsI_1,000–8,000 (positive) with reference
masses ranging from 1,172.1450 Da to 7,927.2031 Da) (see
Note 7).
12. Browse to select the previous acquisition file. Click on History
and select the last AccMass2 line.
13. Close the four windows by clicking OK and close the Mass
Calibration Profile Editor.
14. Click Next, Next, and Start to launch the calibration.

3.4.3 Q-TOF Maxis II 1. Using the HESI source and the 500 μL Hamilton™ syringe,
deliver cesium iodide calibration mix (see Subheading 2.2) with
a flow rate of 3 μL/min.
2. In the MS Tune window, manually set the following
parameters:
Operate the instrument in positive mode with an m/z
range from 1,000 to 8,000. Apply 20 eV and 5 eV for is CID
and CE parameters, respectively. Furthermore, native experi-
ments involve adapting transfer and prepulse storage times (see
Note 8): set to 320 μs and 15 μs, respectively. Concerning the
178 Stéphane Erb et al.

source parameters, set the capillary voltage at 4 kV, the neb-

ulizer at 0.4 bar, the dry gas at 4.0 L/min, and the temperature
at 220  C.
3. To perform the calibration, go to the “Calibration” tab and
select the CsI reference file, then check that the calibration
mode is on “Enhanced Quadratic Mode.” Finally, click to
“Calibrate” and accept the calibration.

3.4.4 Exactive™ Plus 1. Using the HESI source and the 500 μL Hamilton™ syringe,
EMR Orbitrap deliver cesium iodide calibration mix (see Subheading 2.2) with
a flow rate of 10 μL/min.
2. In the MS Tune page, manually set the following parameters:
Operate the instrument in positive mode with an m/z
range from 1,000 to 20,000. Apply 25 eV and 100 eV for the
CID and CE parameters, respectively. The source temperature
is set at 250  C, the capillary voltage is set to 4 kV, and the gas
flow is set to 10 u.a. The trapping gas pressure is set to 7 u.a.
Set the ion optics (injection, inter, and bent flatapoles) at 4 V,
the nominal resolution at 17,500 and activate the EMR mode.
3. To perform the calibration, wait for a stable Total Ion Chro-
matogram (<12%), then select “Calibrate,” and check “EMR
MS Mass Calibration (pos)”.

3.4.5 Q-Exactive™ 1. Using the syringe pusher to deliver the ammonium hexafluor-
BioPharma Orbitrap ophosphate calibration mix at 5 μL/min.
2. In the Calibration Panel, select “HMR Mode calibration
(pos).”.

3.5 Native MS High mass detection on a modified ESI-TOF is illustrated by the

Analysis of E. coli RNA analysis of E. coli RNA polymerase, a 400 kDa hexameric enzyme
Polymerase on a consisting of a dimer of α subunits bound to single copies of β, β0 ,
Modified TOF LCT ω, and a σ-subunits [17]. Here, a complex purified using an estab-
lished procedure [18] is characterized using a simple TOF analyzer,
previously optimized for the transfer of high-molecular weight
species (see Note 9).
1. Desalt approximatively 100 μL of E. coli RNA polymerase at
30 μM in 500 mM ammonium acetate at pH 6.9 as explained in
Subheading 3.2 using a Vivaspin™ 100 kDa cut-off membrane
at 15,000  g and +4  C.
2. Infuse the sample at 5 μM with the TriVersa Nanomate™ (see
Subheading 3.3).
3. Start with the same parameters as the ones used for CsI calibra-
tion but set the cone voltage to 150 V and the extraction cone
to 50 V. Increase the backing pressure up to 7 mbar (see Note
6). Adjust the pressure inside the first hexapole ion guide to
 Native Mass Spectrometry 179

0.11 mbar using argon (available for the upgraded LCT) in

order to optimize the transmission of high molecular species
for native MS application.
4. Start a 2-min acquisition in the 1,000–20,000 m/z range.
5. The corresponding smoothed full mass spectrum is presented
in Fig. 1a and the corresponding deconvoluted spectrum in
Fig. 1b. In order to perform this deconvolution with UniDec
[19, 20] (http://unidec.chem.ox.ac.uk/), first export the spec-
trum list (Edit\Copy spectrum list), then paste it in a text editor
and save it as a .txt file. Import the .txt file in the UniDec
software. Process the data with the following parameters: m/z
range 1,000–12,000 Th; Subtract Curved: 0; Gaussian
Smoothing: 10; Bin Every: 10. Deconvolute the processed
spectrum with the following parameters: Charge Range:
20–50; Mass Range: 20,000–500,000 Da; Sample Mass
Every 1 Da; Peak FWHM 15 Th; Peak Shape Function: Gauss-
ian. Detect and label the deconvoluted species with the follow-
ing parameters: Peak Detection Range: 50; Peak Detection
Threshold: 0.1; Peak Normalization: Max.
As shown in Fig. 1, the pentameric α2ββ0 ω E. coli RNA
polymerase is detected at the expected molecular weight
(390,315 Da), together with some partially dissociated α sub-
unit (36,509 Da).

Fig. 1 Native MS analysis of the E. coli RNA polymerase complex. (a) Full MS spectrum of E. coli RNA
polymerase complex showing charged states 35+ to 41+ in the 9,000–12,000 m/z range corresponding to the
hetero-oligomer composed of five subunits (orange triangle) and 16+ to 33+ under m/z 2,500 corresponding
to the α-subunit (green circle). (b) Nondenaturing mass spectrum of E. coli RNA polymerase after deconvolu-
tion with UniDec highlights the presence of the hetero-oligomeric protein (390,315 Da) and the monomeric
α-subunit (36,509 Da)
180 Stéphane Erb et al.

3.6 Native MS The use of a Q-TOF Synapt G2 platform is illustrated by the

Analysis of the CDK- characterization of the CDK-activating kinase complex (CAK), a
Activating Kinase sub-complex of the transcription/DNA repair factor TFIIH
Complex on a Q-TOF [21, 22]. CAK is composed of a cyclin-dependent kinase (Cdk7),
Synapt G2 its associated cyclin (Cyclin H), and a third polypeptide known as
MAT1, which stabilizes the Cdk7/cyclin H pair and bridges the
kinase module of TFIIH to the core-complex composed of seven
additional subunits. Recombinant CAK was produced in insect cells
using the baculovirus expression system as described [23].
1. Desalt 20 μL of recombinant CAK at 20 μM in 200 mM
ammonium acetate supplemented with 0.02% n-dodecyl β-D-
maltoside (see Note 10), as explained in Subheading 3.1.
2. Connect the TriVersa NanomateTM as explained in Subhead-
ing 3.3.
3. In MassLynx MS Tune window manually set the following
parameters:
Operate the instrument in positive and sensitivity modes
with an m/z range from 1,000 to 10,000 (see Note 11).
Nanoflow+ panel: Sampling Cone: 175 V (see Note 12).
Extraction cone: 5 V. Source Temperature: 90  C.
Instrument: Increase the backing pressure up to 6 mbar (see
Note 6). Trap Collision Energy is turned on (75 V) and
Transfer Collision Energy is turned off (2 V by default) (see
Note 13). Trap gas flow (argon) is set to 4.5 mL/min.
Save these parameters as an .ipr file.
4. Press “acquire” to start a 2-min TOF-MS acquisition in the
1,000–10,000 m/z range.
5. Analyze data.
The corresponding averaged and smoothed (20 times) raw
data is represented in Fig. 2a. Manual or software-driven (see
Subheading 3.5, step 5) deconvolution of the data identifies

Fig. 2 Native MS spectrum of the recombinant CDK-activating kinase complex desalted on a Micro
Bio-SpinTM 6 column. MS spectrum (a) before and (b) after data processing; and (c) after deconvolution
with UniDec
 Native Mass Spectrometry 181

different species corresponding to the three monomers of the

CAK complex: namely, Cdk7 (MW: 40.1 kDa), Cyclin-H
(MW: 37.7 kDa), and MNAT1 (MW: 35.9 kDa) (Fig. 2b, c).
A species corresponding to the heterodimer Cdk7-Cyclin-H at
77.9 kDa is visible, confirming their direct interaction. Finally,
the physiologically active heterotrimeric CAK is identified at
113.9 kDa.

3.7 Protein/DNA The noncovalent interaction between the Estrogen-Related Recep-

Complex Analysis on a tor DNA binding domain (ERR-DBD) and its DNA response
Q-TOF Synapt G2 element [24] can be monitored by direct infusion on a Synapt G2
platform. This transcription factor belongs to the steroid hormone
nuclear receptor family and shares strong similarity in its
DNA-binding domain (DBD) with that of the estrogen receptor
(ER). In vitro, ERR binds with high affinity inverted repeat REs
with a 3-bp spacing (IR3), but in vivo, it preferentially binds to
single half-site response elements extended at the 50 -end by 3 bp.
1. Desalt 50 μL of ERRα DBD-BE26PSIR3 complex [24] at
28 μM on a Zeba™ spin column equilibrated with 200 mM
ammonium acetate at pH 7.1 as explained in Subheading 3.1.
2. Use the TriVersa Nanomate™ (see Subheading 3.3) coupled to
the mass spectrometer and infuse the protein complex at 5 μM.
3. In the Tune page, start with the parameters used for the CsI
calibration and adapt the cone voltage at 140 V, the extraction
cone at 5 V and the backing pressure at 6 mbar (see Note 6).
4. Start a 2-min acquisition in the 1,000–6,000 m/z range.
5. The average full MS spectrum obtained is presented in Fig. 3a.
The deconvoluted spectrum of the protein/DNA complex
(see Subheading 3.5, step 5) highlights a 1:1 stoichiometry for
the protein/DNA complex (Fig. 3). The ERR-DBD binds two
zinc atoms and one DNA duplex (27,892.5 Da for full-length
DNA). Other species corresponding to truncated DNA asso-
ciated with the protein with a 1:1 stoichiometry are also
detected (27,740.6 Da and 27,586.3 Da).

3.8 SEC Native MS Size-exclusion chromatography (SEC) is commonly used both as a

Coupling (ADH) on a Q- purification and for analytical purposes to monitor sample hetero-
TOF Synapt G2Si geneity. SEC is often coupled with UV Absorbance, Refractive
Index (RI), or Multi-Angle Laser Light Scattering (MALLS) to
estimate in particular the average molecular weight and the con-
centration of each component of the sample. SEC can also be
coupled with mass spectrometry (MS) to precisely measure the
molecular weight of each separated species. This approach is illu-
strated here by the characterization of yeast ADH oligomers on an
ACQUITY UPLC H-Class Bio coupled to a Q-TOF Synapt G2Si.
182 Stéphane Erb et al.

Fig. 3 Native MS analysis of the ERRα DBD-BE26PSIR3 complex. (a) Full MS spectrum of protein/DNA
complex. Two charge states distributions in the 2,000–4,000 m/z range are present, first corresponding to the
6+ and 7+ charge states of DNA alone in the 2,000–4,000 m/z range and second to the 9+ and 8+ charge
states of the protein/DNA complex in the 3,000–3,800 m/z range. (b) Zoom in the 3,000–3,600 m/z range
showing the protein/DNA complex 9+ and 8+ charge states revealing the presence of two truncated DNA
species. (c) The associated deconvolution of native mass spectrum with UniDec software highlights a 1:1
stoichiometry for the protein/DNA complex (27,892.5 Da) and confirms the presence of two truncated DNA
species which are also able to interact with the protein (27,740.6 Da and 27,586.3 Da)

1. Connect the ACQUITY UPLC H-Class Bio to the Synapt

G2Si.
2. Connect the ACQUITY UPLC Protein BEH SEC Column
(200 Å, 1.7 μm, 4.6 mm  150 mm) and equilibrate with
200 mM ammonium acetate at 300 μL/min.
3. Weigh 1.6 mg of alcohol dehydrogenase from Saccharomyces
Cerevisiae (Sigma-Aldrich). Resuspend in 280 μL of 200 mM
ammonium acetate to reach a tetrameric concentration of
40 μM.
 Native Mass Spectrometry 183

4. In MassLynx, set up an MS file with the following parameters:

Start Time: 0 min; End Time: 10 min; Polarity: Positive;
Analyser Mode: Sensitivity; Dynamic Range: Normal; Sensitiv-
ity: Normal; Low Mass: 1,000 Da; High Mass: 10,000 Da;
Scan Time: 1 s (Continuum); No Trap or Transfer CE.
5. In MassLynx, set up an Inlet file with the following parameters:
Flow rate 200 μL/min with 100% solvent A (200 mM ammo-
nium acetate) for 10 min.
6. In MassLynx MS Tune windows, manually set the following
parameters:
Nanoflow+ panel: Capillary: 3 kV; Sampling Cone: 150 V.
Source Offset: 80 V. Source Temperature: 100  C. Cone Gas:
50 L/h; Nano Flow Gas: 0 Bar; Purge Gas: 100 L/h.
Instrument: Trap and Transfer Collision Energy are turned
off (4 V and 2 V, respectively by default).
Save these parameters as an MS tune file (.ipr file).
7. In MassLynx, create a sample list with the corresponding MS
file, Inlet file, and MS tune file. Specify the sample position
(Bottle) and the volume to be injected (10 μL).
8. Start the acquisition.
The chromatogram shows two main peaks at 6.14 min and
7.32 min (Fig. 4a), corresponding to the tetramer (Fig. 4c) and
monomer (Fig. 4b) of the ADH, respectively.

Fig. 4 SEC-Native MS analysis of Saccharomyces cerevisiae ADH. (a) Total Ion Chromatogram (TIC)
corresponding to the SEC separation of the monomer and tetramer of the ADH. (b) Native MS spectrum of
the monomer (36.9 kDa) before (left) and after (right) deconvolution with UniDec. (c) Native MS spectrum of the
tetramer (147.5 kDa) before (left) and after (right) deconvolution with UniDec
184 Stéphane Erb et al.

3.9 Homo- Concanavalin A from Jack bean is a lectin commonly used as a

Oligomeric standard in Native MS experiments. Its physiological oligomeric
Concanavalin A state is tetrameric although higher-order oligomers can often be
Analysis on a High- observed at high concentrations. We use it here to describe the
Resolution Q-TOF parameters allowing its quaternary structure assessment with a
Maxis II Q-TOF Maxis II platform.
1. Resuspend 1.5 mg of Concanavalin A from Canavalia ensifor-
mis (Jack bean, Sigma-Aldrich reference no. C5275) in 200 μL
of 10 mM ammonium acetate at pH 6.8 to reach a concentra-
tion around 50 μM for the tetrameric species.
2. Desalt 100 μL of the protein (see Subheading 3.2) on a Vivas-
pin™ 10 kDa cut-off membrane (6–8 cycles at 15,000  g and
+4  C during 8–10 min).
3. Connect the TriVersa Nanomate™ to the mass spectrometer.
Infuse the protein at 10 μM.
4. In the tune page, use the same parameters used for CsI calibra-
tion except for the isCID that should be set at 35 eV.
5. Start a 2-min acquisition in the 1,000–8,000 m/z range.
6. The average smoothed mass spectrum is represented in Fig. 5a.
The corresponding deconvoluted spectrum is obtained after
UniDec processing (Fig. 5b, see Subheading 3.5, step 5) and
allows the identification of Concanavalin A tetrameric species at
102,707 Da.

3.10 High Molecular The use of an Exactive Plus platform is illustrated by the characteri-
Weight Homo- zation of the GroEL chaperone. This 57-kDa protein assembles as a
oligomeric Protein on a 14-mer to form a double toroidal complex, which together with
High-Resolution the GroES co-chaperonin facilitates protein folding in an
Exactive™ Plus EMR ATP-dependent manner [25].
Orbitrap 1. Use the freshly prepared and desalted GroEL protein (see Sub-
heading 2.4).
2. Connect the TriVersa Nanomate™ (see Subheading 3.3) to the
mass spectrometer. Infuse the sample concentrated at 5 μM of
oligomer.
3. In the Tune page, set the following parameters: Positive mode
with a 1,000–20,000 m/z range. Activate the EMR mode and
set the trapping gas pressure at 7 u.a., the ion optics at 4 V
(inter, injection, and bent flatapoles), the S-lens RF voltage at
200 V, the temperature at 250  C, the fragmentation para-
meters at 25 eV and 150 eV for CID and CE values, respec-
tively. Select a nominal resolution of 17,500.
4. Start a 2-min acquisition with a stable TIC and analyze the MS
spectrum.
The resulting averaged mass spectrum is presented in
Fig. 6a (full mass range) and Fig. 6b (zoom between 11,000
 Native Mass Spectrometry 185

Fig. 5 Native MS analysis of Concanavalin A from Canavalia ensiformis protein. (a) Full MS spectrum of
Concanavalin A showing charged states 24+ to 20+ in the m/z range from 4,000 to 5,500 corresponding to
the homo-oligomeric protein (blue circle). (b) Non-denaturing mass spectrum of Concanavaline A after
deconvolution with UniDec highlights the tetrameric state of the protein complex (102,707 Da)

Fig. 6 Native MS analysis of Chaperonin 60 (GroEL) complex. (a) Full MS spectrum of GroEL complex showing
a unique charge states distribution around m/z 12,000. (b) Zoom in the 10,000–14,000 m/z range where
charge states 64+ to 71+ of the protein complex are detected. (c) Associated deconvolution obtained after
data processing using UniDec software in the 785–815 kDa mass range, showing one species corresponding
to the 14-mer homo-complex at 801.6 kDa

and 13,000 m/z). After data processing (see Subheading 3.5,

step 5), a unique species corresponding to the 14-mer complex
is identified (Fig. 6c) at 801.6 kDa.

3.11 Protein The use of a Q-Exactive™ Biopharma Orbitrap platform for the
Oligomeric State determination of protein oligomeric states is exemplified by the
Determination on a analysis of the p8/TTD-A TFIIH subunit and yeast alcohol dehy-
High-Resolution Q- drogenase (ADH). p8 is a 72-amino acid protein shown to form
Exactive™ Biopharma stable homodimers in solution [26]. The ADH is an enzyme
Orbitrap required for the reduction of acetaldehyde to ethanol and is active
as a tetramer [27].
186 Stéphane Erb et al.

3.11.1 The 1. Desalt 20 μL of recombinant p8 [28] at 24 μM as explained in

Trichothiodystrophy Group Subheading 3.1.
A Protein (TTD-A or p8) 2. Connect the TriVersa Nanomate™ to the Q-Exactive™ Bio-
pharma Orbitrap as explained in Subheading 3.3.
3. In Excalibur Tune window, manually set the following
parameters:
Operate the instrument in positive mode with an m/z
range from 200 to 4,000.
4. Intact protein mode: off; HRM mode: off; resolution:
120,000; Micro Scan Count: 1; Max. Ion Time: 100 ms;
AGC target: 1e6, S-Lens RF Level: 40, Capillary Temperature:
275  C, Capillary Voltage: 1.5 kV.
5. Save these parameters as an .mstune file.
Start a 2 min acquisition in the 200–4000 m/z range and
analyze data.
The corresponding averaged raw data is represented in
Fig. 7a. Manual or software-driven (see Subheading 3.5, step
5) deconvolution of the data identifies different species
corresponding to the monomeric (MW ¼ 8,272 Da) and
dimeric (MW ¼ 16,544 Da) p8 protein (Fig. 7b). A close-up
view of the peak at 2,069 m/z (42,000 resolution) shows
isotopic distributions corresponding to both the 4+ charged
state of the monomer and the 8+ charged state of the dimer
(Fig. 7a, inset). The effective resolution ranges from 35,000
(m/z 2,758) to 56,000 (m/z 1,035).

Fig. 7 Native MS analysis of the p8 subunit from the human TFIIH complex (a) Raw MS spectrum of p8
showing charged states 4+ to 9+ corresponding to the monomer (purple) and 6+ to 11+ corresponding to the
dimer (red). The inset shows a close-up view of the 2,069.01 peak, revealing an overlap of the monomeric 4+
and dimeric 8+ charged states. (b) Native MS spectrum of p8 after deconvolution with UniDec, showing the
presence of monomeric (8,272 Da) and dimeric (16,544 Da) species
 Native Mass Spectrometry 187

3.11.2 The Yeast Alcohol 1. Weigh 1.6 mg of alcohol dehydrogenase from Saccharomyces
Dehydrogenase (ADH) Cerevisiae (Sigma-Aldrich). Resuspend in 280 μL of 200 mM
ammonium acetate to reach a tetrameric concentration of
40 μM.
2. Connect the TriVersa Nanomate™ to the Q-Exactive™ Bio-
pharma Orbitrap as explained in Subheading 3.4.
3. In Excalibur Tune window, manually set the following
parameters:
Operate the instrument in positive mode with an m/z
range from 2,500 to 8,000.
Intact protein mode: off; HRM mode: on; resolution:
15,000; Micro Scan Count: 10; Max. Ion Time: 200 ms;
AGC target: 3e6, S-Lens RF Level: 200, Capillary Tempera-
ture: 320  C, Capillary Voltage: 4.4 kV.
Save these parameters as an .mstune file.
4. Start a 2-min acquisition in the 2,500–8,000 m/z range and
analyze data.
The corresponding averaged raw data is represented in
Fig. 8a. Manual or software-driven (see Subheading 3.5, step
5) deconvolution of the data identifies different species
corresponding to the monomeric (MW ¼ 36,879 Da) and
tetrameric (MW ¼ 147,516 Da) ADH protein (Fig. 8b). The
effective resolution ranges from 2,000 (m/z 5,674) to 4,600
(m/z 2,844).

Fig. 8 Native MS analysis of Saccharomyces cerevisiae ADH. (a) Raw MS spectrum of ADH showing charged
states 10+ to 14+ corresponding to the monomer (purple) and 23+ to 28+ corresponding to the tetramer
(red). (b) Native MS spectrum of ADH after deconvolution with UniDec, showing the presence of monomeric
(36,879 Da) and tetrameric (147,516 Da) species
188 Stéphane Erb et al.

4 Notes

1. The optimum ammonium acetate concentration is protein-

dependent and can be optimized by testing concentrations
ranging from 10 mM up to 1 M. If needed, volatile additives
can be added, such as dithiothreitol or tris(2-carboxyethyl)
phosphine for disulfide bond reduction, imidazole, some deter-
gents (see Note 10), or metallic ions in a volatile buffer (cal-
cium acetate, zinc acetate, etc.).
2. Volatility of the buffer is key for native MS. Beware to avoid any
trace of nonvolatile salts such as sodium chloride. Ammonium
acetate resuspended at 200 mM in water is usually at pH 6.9. If
you want to adjust pH below 6.9, do not use HCl but formic
acid or acetic acid instead. Alternatively, you can dilute a com-
mercial 7.5 M ammonium acetate solution (Sigma). Con-
versely, to increase the pH, use liquid NH3 and not NaOH.
3. For smaller volumes, 7 K MWCO Zeba™ Micro Spin columns
(2–12 μL, Thermo Fisher Scientific) can be used.
4. Do not forget to turn off the cooling of the plate after use of
the nanomate. Otherwise, condensation will occur, water will
fill the wells, and the plate will have to be discarded the
next day.
5. The gas pressure and voltage to apply are the main two para-
meters to adjust. The stability of the spray can be visualized in
the Spray Optimization window, by pressing “Graph.” If the
intensity of the electrospray is dropping, incrementally increas-
ing the gas pressure (no more than 0.5 psi) or the Voltage to
apply (no more than 1.80 kV) might help. If the spray is not
coming back, use another nozzle by pressing “Next Nozzle.”
Reducing the gas pressure and Voltage to apply while spraying
usually results in spray loss. The optimum Voltage to apply is
protein-dependent and has to be carefully optimized, as higher
values usually increase the ion transmission (more intense sig-
nal) and buffer removal (better resolved signal) but might
result in complex dissociation. Ticking “Return unused sample
to tray” enables to recover most of the sample after data
acquisition.
6. The backing pressure is a key parameter on Waters Instruments
(LCT, Synapt), which needs to be optimized for each protein
complex. Generally, this value is set around 6 mbar and can be
adjusted if necessary. This is possible by adding a homemade
system with two external valves that reduce the suction of the
primary pump.
 Native Mass Spectrometry 189

7. Reference files for higher m/z ranges can be custom made. In

the case of CsI, increments of +259.81 Da can be directly
added to the initial CsI calibration file.
8. In order to optimally accumulate and transfer heavy ions,
transfer time and prepulse storage time parameters should be
increased compared to small ions analysis. These two para-
meters are linked, but the transfer time has to be higher than
the prepusle storage time.
9. Waters TOF or Q-TOF mass spectrometers can be upgraded
(MS Vision, the Netherlands) by modifying electronics and gas
pressure in order to improve the transmission of high molecu-
lar mass species in non-denaturing conditions. On our
upgraded LCT, the combination of an additional gas arrival
(argon) in the first hexapole housing through a needle valve
and the modification of different electronic parts (e.g., higher
voltages applied to sample and extraction cones) results in an
improvement of high molecular mass transmission. Typically,
the sensitivity for GroEL14 analysis is increased sixfold by
argon admission if the pressure is set around 4  102 mbar.
A micro-Pirani gauge is installed to control the pressure
applied.
10. n-Dodecyl β-D-maltoside can be used at two times the critical
micellar concentration (0.02%) in order to stabilize very labile
soluble complexes. It can also be used at the same concentra-
tion to spray membrane proteins [29, 30].
11. The m/z range on Waters Q-TOF depends on the type of
Quadrupole that is installed: up to 16,000 m/z for a
4,000 m/z quadrupole, up to 32,000 m/z for an 8,000 m/z
quadrupole and up to 128,000 m/z for a 32,000 m/z
quadrupole.
12. The sample cone is a key parameter on Waters Instruments
(LCT, Synapt, etc.) that needs to be optimized for each protein
complex. Similarly, to the voltage applied to the nozzle of the
TriVersa Nanomate™, higher values usually increase the inten-
sity of the signal and provide better resolved peaks by removing
residual buffer. However, high sample cone values might result
in complex dissociation. Here again, an optimum value has to
be found that provides good signal without undermining the
multiprotein complex integrity. A good strategy is to start with
a maximum Sampling Cone voltage (150 V) and progressively
reduce it by 10-V steps.
13. The trap collision energy is another key parameter on Waters
Q-TOF. In the case of membrane proteins or very large multi-
protein complexes (>500 kDa), increasing the trap collision
energy might be helpful or even mandatory in order to ther-
modynamically cool down the ions (large complexes) or
190 Stéphane Erb et al.

remove the remaining detergent molecules. Simultaneously

and incrementally increasing the Trap gas flow (from 2 mL/
min up to 8 mL/min) might improve the transmission of these
particular large or membrane protein complexes.

Acknowledgments

The authors would like to thank Arnaud Poterszman (CAK com-

plex), Albert Weixlbaumer (E. coli RNA polymerase), and Isabelle
Billas Massobrio (ERR complexes) from the IGBMC (Strasbourg,
France) and Virginie Gervais and Alain Milon (p8 protein) from
IPBS (Toulouse, France) for kindly providing samples. This work
was supported by the French Ministry of Research (Investissements
d’Avenir Program, Proteomics French Infrastructure, ANR-10-
INBS-08) and the FEDER (Fonds Européens de Développement
Régional), Toulouse Métropole, and the Région Midi-Pyrénées.

References

1. Ganem B (1991) Detection of noncovalent 8. Andersson FI, Tryggvesson A, Sharon M, Die-

receptor-ligand complexes by mass spectrome- mand AV, Classen M, Best C, Schmidt R,
try. J Am Chem Soc 113:6294–6296 Schelin J, Stanne TM, Bukau B, Robinson
2. Ganem B (1991) Observation of noncovalent CV, Witt S, Mogk A, Clarke AK (2009) Struc-
enzyme-substrate and enzyme-product com- ture and function of a novel type of
plexes by ion-spray mass spectrometry. J Am ATP-dependent Clp protease. J Biol Chem
Chem Soc 113:7818–7819 284(20):13519–13532
3. Katta V, Chait B (1991) Observation of the 9. Sharon M, Mao H, Boeri Erba E, Stephens E,
hemoglobin complex in native myoglobin by Zheng N, Robinson CV (2009) Symmetrical
electrospray-ionization mass spectrometry. J modularity of the COP9 signalosome complex
Am Chem Soc 113:8534–8535 suggests its multifunctionality. Structure 17
4. Nettleton EJ, Sunde M, Lai Z, Kelly JW, Dob- (1):31–40
son CM, Robinson CV (1998) Protein subunit 10. Martinez-Zapien D, Saliou JM, Han X,
interactions and structural integrity of amyloi- Atmanene C, Proux F, Cianférani S, Dock-
dogenic transthyretins: evidence from electro- Bregeon AC (2015) Intermolecular recogni-
spray mass spectrometry. J Mol Biol 281 tion of the non-coding RNA 7SK and
(3):553–564 HEXIM protein in perspective. Biochimie
5. Pinkse MWH, Heck AJR, Rumpel K, Pullen F 117:63–71
(2004) Probing noncovalent protein-ligand 11. Saliou JM, Manival X, Tillault AS,
interactions of the cGMP-dependent protein Atmanene C, Bobo C, Branlant C, Van
kinase using electrospray ionization time of Dorsselaer A, Charpentier B, Cianférani S
flight mass spectrometry. J Am Soc Mass Spec- (2015) Combining native MS approaches to
trom 15(10):1392–1399 decipher archaeal box H/ACA ribonucleopro-
6. Sobott F, Hernández H, McCammon MG, tein particle structure and activity. Proteomics
Tito MA, Robinson CV (2002) A tandem 15(16):2851–2861
mass spectrometer for improved transmission 12. Barrera NP, Di Bartolo N, Booth PJ, Robinson
and analysis of large macromolecular assem- CV (2008) Micelles protect membrane com-
blies. Anal Chem 74(6):1402–1407 plexes from solution to vacuum. Science 321
7. Rose RJ, Damoc E, Denisov E, Makarov A, (5886):243–246
Heck AJ (2012) High-sensitivity Orbitrap 13. Chorev DS, Baker LA, Wu D, Beilsten-
mass analysis of intact macromolecular assem- Edmands V, Rouse SL, Zeev-Ben-
blies. Nat Methods 9(11):1084–1086 Mordehai T, Jiko C, Samsudin F, Gerle C,
 Native Mass Spectrometry 191

Khalid S, Stewart AG, Matthews SJ, 22. Kolesnikova O, Radu L, Poterszman A (2019)
Grünewald K, Robinson CV (2018) Protein TFIIH: a multi-subunit complex at the cross-
assemblies ejected directly from native mem- roads of transcription and DNA repair. Adv
branes yield complexes for mass spectrometry. Protein Chem Struct Biol 115:21–67
Science 362(6416):829–834 23. Fouillen L, Abdulrhaman W, Moras D, van
14. Marcoux J, Robinson CV (2013) Twenty years Dorsselaer A, Poterszman A, Sanglier-Cianfér-
of gas phase structural biology. Structure 21 ani S (2010) Analysis of recombinant phospho-
(9):1541–1550 protein complexes with complementary mass
15. Gan J, Ben-Nissan G, Arkind G, Tarnavsky M, spectrometry approaches. Anal Biochem 407
Trudeau D, Noda Garcia L, Tawfik DS, Sharon (1):34–43
M (2017) Native mass spectrometry of recom- 24. Mohideen-Abdul K, Tazibt K, Bourguet M,
binant proteins from crude cell lysates. Anal Hazemann I, Lebars I, Takacs M,
Chem 89(8):4398–4404 Cianférani S, Klaholz BP, Moras D, Billas
16. van de Waterbeemd M, Tamara S, Fort KL, IML (2017) Importance of the sequence-
Damoc E, Franc V, Bieri P, Itten M, directed DNA shape for specific binding site
Makarov A, Ban N, Heck AJR (2018) Dissect- recognition by the estrogen-related receptor.
ing ribosomal particles throughout the king- Front Endocrinol (Lausanne) 8:140
doms of life using advanced hybrid mass 25. Horwich AL, Fenton WA, Chapman E, Farr
spectrometry methods. Nat Commun 9 GW (2007) Two families of chaperonin: physi-
(1):2493 ology and mechanism. Annu Rev Cell Dev Biol
17. Opalka N, Brown J, Lane WJ, Twist KA, 23:115
Landick R, Asturias FJ, Darst SA (2010) Com- 26. Vitorino M, Coin F, Zlobinskaya O,
plete structural model of Escherichia coli RNA Atkinson A, Moras D, Egly JM,
polymerase from a hybrid approach. PLoS Biol Poterszman A, Kieffer B (2007) Solution struc-
8(9):e1000483 ture and self-association properties of the p8
18. Twist KA, Husnain SI, Franke JD, Jain D, TFIIH subunit responsible for Trichothiody-
Campbell EA, Nickels BE, Thomas MS, Darst strophy. J Mol Biol 368(2):473–480
SA, Westblade LF (2011) A novel method for 27. Raj SB, Ramaswamy S, Plapp BV (2014) Yeast
the production of in vivo-assembled, recombi- alcohol dehydrogenase structure and catalysis.
nant Escherichia coli RNA polymerase lacking Biochemistry 53(36):5791–5803
the α C-terminal domain. Protein Sci 20 28. Gervais V, Muller I, Mari PO, Mourcet A,
(6):986–995 Movellan K, Ramos P, Marcoux J, Guillet V,
19. Marty MT, Baldwin AJ, Marklund EG, Hoch- Javaid S, Burlet-Schiltz O, Czaplicki G,
berg GK, Benesch JL, Robinson CV (2015) Milon A, Giglia-Mari G (2018) Small
Bayesian deconvolution of mass and ion mobil- molecule-based targeting of TTD-A dimeriza-
ity spectra: from binary interactions to polydis- tion to control TFIIH transcriptional activity
perse ensembles. Anal Chem 87 represents a potential strategy for anticancer
(8):4370–4376 therapy. J Biol Chem 293(39):14974–14988
20. Reid DJ, Diesing JM, Miller MA, Perry SM, 29. Marcoux J, Wang S, Politis A, Reading E, Ma J,
Wales JA, Montfort WR, Marty MT (2019) Biggins P, Zhou M, Tao H, Zhang Q,
MetaUniDec: high-throughput deconvolution Chang G, Morgner N, Robinson CV (2013)
of native mass spectra. J Am Soc Mass Spec- Mass spectrometry reveals synergistic binding
trom 30(1):118–127 of nucleotides, lipids and drugs to a multidrug
21. Radu L, Schönwetter E, Braun C, Marcoux J, resistance efflux pump. Proc Natl Acad Sci 110
Kölmel W, Schmitt DR, Kuper J, Cianferani S, (24):9704–9709
Egly JM, Poterszman A, Kisker C (2017) The 30. Laganowsky A, Reading E, Hopper JT, Robin-
intricate network between the p34 and p44 son CV (2013) Mass spectrometry of intact
subunits is central to the activity of the tran- membrane protein complexes. Nat Protoc 8
scription/DNA repair factor TFIIH. Nucleic (4):639–651
Acids Res 45(18):10872–10883
 Chapter 11

Studying Protein–DNA Interactions by Hydrogen/Deuterium

Exchange Mass Spectrometry
Ruzena Filandrova, Daniel Kavan, Alan Kadek, Petr Novak, and Petr Man

Abstract
Protein hydrogen/deuterium exchange (HDX) coupled to mass spectrometry (MS) can be used to study
interactions of proteins with various ligands, to describe the effects of mutations, or to reveal structural
responses of proteins to different experimental conditions. It is often described as a method with virtually
no limitations in terms of protein size or sample composition. While this is generally true, there are,
however, ligands or buffer components that can significantly complicate the analysis. One such compound,
that can make HDX-MS troublesome, is DNA. In this chapter, we will focus on the analysis of pro-
tein–DNA interactions, describe the detailed protocol, and point out ways to overcome the complications
arising from the presence of DNA.

Key words DNA, Hydrogen/deuterium exchange, Protein–DNA binding, Structural mass spec-
trometry, Transcription factor

1 Introduction

Structural proteomics is a rapidly developing field focusing specifi-

cally on protein structural dynamics and characterization of the
architecture of protein macromolecular assemblies. Hydrogen/
deuterium exchange (HDX) ranks as one of the most versatile and
prominent methods in this area. Its biggest advantages are not
being limited by protein size and the ability to study proteins in
their truly native-like environment, at virtually any pH, buffer
composition, protein concentration, or temperature (see Note 1).
In a typical setup, an HDX experiment starts by the dilution of a
protein into a deuterated buffer. At selected time points, aliquots
are taken, and the exchange is quenched (nearly stopped) by rapid
acidification to pH 2.5 and by lowering the temperature to 0
C.
Subsequently, each sample is digested by acidic, nonspecific pro-
teases, and deuteration of the generated peptides is measured by
mass spectrometry (MS) [1–3].

Arnaud Poterszman (ed.), Multiprotein Complexes: Methods and Protocols, Methods in Molecular Biology, vol. 2247,
https://doi.org/10.1007/978-1-0716-1126-5_11, © Springer Science+Business Media, LLC, part of Springer Nature 2021

193
194 Ruzena Filandrova et al.

Albeit very powerful and usually straightforward, HDX-MS has

also some disadvantages, e.g., rather limited spatial resolution and
need for sample dilution as the experiment often starts with
ten-fold dilution into a deuterated buffer. However, due to the
coupling to sensitive mass spectrometric detection, micro- and
submicromolar concentrations are easily managed. Although
HDX in combination with mass spectrometry was introduced
more than two decades ago and there are hundreds of research
articles describing its uses to study protein folding/unfolding,
protein–protein and protein–ligand interactions [4, 5], the number
of reports utilizing HDX for protein–nucleic acid interaction is
rather limited. This is most likely due to the nature of nucleic
acids that complicate HDX-MS analyses. During the quenching
step (lowering the pH) in an HDX-MS workflow, the DNA back-
bone becomes protonated, which often results in its poor solubility
and precipitation [6]. The protein interacting with DNA then tends
to precipitate as well. The presence of DNA can also have adverse
effects on chromatographic separation. Such behavior, however,
depends on the DNA composition and length. Therefore, reports
dealing with larger DNA stretches always utilize technical tricks to
deal with the troublesome DNA while those using very short
oligonucleotides may sometimes be spared these adverse effects.
The first ever attempt to use HDX for protein–nucleic acid
interaction study enabled the determination of a protein–DNA
complex dissociation constant by a combination of HDX and
matrix-assisted laser desorption/ionization mass spectrometry
[7]. However, it took five more years until Sperry et al. in 2008
truly successfully probed structural features of protein–DNA inter-
actions of two biological systems (a protein binding to a telomeric
oligonucleotide and the protease thrombin interacting with an
aptamer). These studies were successfully accomplished thanks to
the incorporation of strong anion exchange trap column prior to
the liquid chromatography-mass spectrometric (LC-MS) analysis
to selectively trap and remove the oligonucleotides [8, 9]. Besides
an ion exchanger, the use of protamine sulfate has also been
reported as beneficial in the HDX-MS analysis of oligonucleotide-
containing samples. Poliakov and coworkers concluded that the
protein–DNA co-precipitation upon the lowering of pH is primar-
ily of electrostatic origin. Thus, inclusion of small, highly basic, and
hence, positively charged protamine improved protein recovery
[10]. Alternatively, in another study, Roberts et al. counteracted
the negative effect of DNA on the formation of peptides, solubility,
and chromatographic separation in HDX by tuning the concentra-
tion of a positively charged denaturing agent (guanidine hydro-
chloride) in their quench buffer [11]. Use of another denaturant
has recently been exploited by Graham et al. to explain structural
and mechanistic aspects of DNA unwinding when guanidine
 HDX-MS of Protein-DNA Complexes 195

hydrochloride was replaced by urea, and an alternative aspartic

protease was used in the quench buffer [12].
In this study, drawing upon the experience gained through our
own work, we present an experimental protocol for the characteri-
zation of protein interactions with DNA (in the form of an oligo-
nucleotide) as used in our group. We have repeatedly observed that
careful optimization and proper control over the quality of the
DNA entering an HDX experiment (amount of dsDNA formed,
purity, salt concentration) are crucial for boosting its ability to bind
to the interaction partner. Ensuring as strong an interaction as
possible in this manner then sometimes largely mitigates the pro-
blems stemming from the DNA presence, as the amount of oligo-
nucleotides used can be kept to a minimum (optimally down to 1:1
molar ratio to the protein), while still providing virtually complete
saturation of the transcription factor (TF). Sometimes, however,
even the best quality of sample is not enough, and the aforemen-
tioned techniques for coping with DNA come into play.
In the following set of protocols, we describe ways to perform
protein–DNA HDX experiments, together with some of the poten-
tial pitfalls, with emphasis on the interaction of transcription factors
with their DNA-response elements. As we have observed that
protein:DNA systems can behave surprisingly differently even
when the proteins as well as DNAs used are similar in size, we will
be using two example systems. Forkhead box protein O4
(FOXO4)/DAF16 [13, 14] serves as an example where short
dsDNA (13 bp) does not interfere significantly with peptide recov-
ery, and only digestion conditions are optimized to reach good
HDX spatial resolution. On the other hand, the TEA domain
family member 1 (TEAD1)/M-CAT [15] represents the opposite
situation when a 15-bp long oligonucleotide has a strong impact on
peptide recovery. These striking differences in behavior stress the
importance of having a vast array of conditions, proteases, and
technical tricks at one’s disposal to fine-tune the HDX protocol.
The procedures we use, as described herein, should provide pro-
spective users with a set of tools to successfully optimize conditions
for HDX and perform their own analyses of a range of TF–DNA
complexes.

2 Materials

1. Protein to be analyzed (see Note 2).

2. Forward and reverse DNA strands (see Note 3).
3. Buffer suitable for the complex (e.g., 20 mM HEPES, pH 7.4,
150 mM NaCl).
4. Gel casting tray and electrophoresis apparatus (Bio-Rad).
196 Ruzena Filandrova et al.

5. Non-denaturing acrylamide gel (12%)—for 10 ml mix the

following: 6 ml H2O, 1 ml 10 concentrated TBE buffer,
80 μl 10% ammonium persulfate (APS), 8 μl TEMED (N,N,
N0 ,N’-tetramethylethylenediamine), 3 ml 40% acrylamide:N,
N0 -methylenebisacrylamide (1:19 mix).
6. DNA loading dye (Thermo Fisher).
7. Protein staining solution (45% methanol, 10% acetic acid,
0.25% Coomassie Brilliant Blue R250).
8. TBE buffer (90 mM Tris–Cl, 90 mM boric acid, 2 mM EDTA,
pH 8.3).
9. Fluorescent DNA stain GelRed (Thermo Fisher).
10. Deuterium oxide (D2O).
11. Quench buffer (500 mM glycine in LC-MS grade water,
pH 2.3 set with concentrated HCl).
12. Urea or Guanidine hydrochloride.
13. Liquid nitrogen.
14. Peltier cooled box or a polystyrene box filled with ice and water
mixture (see Note 4).
15. Two 6-port valves, one of them with injection needle port
(Rheodyne/IDEX).
16. PEEK (polyether ether ketone) tubing (OD 1/1600 , ID—
250 μm and 750 μm) and stainless steel tubing (OD 1/1600 ,
ID—125 μm).
17. UPLC (ultrahigh-performance liquid chromatography) fit-
tings with ferrules (1/1600 ).
18. Hamilton syringe with blunt end needle (100 μl).
19. Timer.
20. Columns with immobilized acidic protease—pepsin, rhizopus-
pepsin (protease type XVIII), nepenthesin-1, nepenthesin-2,
or aspergillopepsin (protease type XIII) immobilized on
POROS™ 20AL (Thermo Fisher Scientific) using previously
described [16, 17] protocols and packed into stainless steel
guard columns (2 mm  20 mm or 1 mm  20 mm, IDEX).
21. Desalting trap column—VanGuard Pre-column (ACQUITY
UPLC BEH C18, 130 Å, 1.7 μm, 2.1 mm  5 mm, Waters).
22. Reversed-phase analytical column (ACQUITY UPLC BEH
C18, 130 Å, 1.7 μm, 1 mm  100 mm, Waters).
23. Loading and desalting pump (1260 Infinity II Quaternary
pump (Agilent Technologies)).
24. Gradient pump (1290 series, Agilent Technologies).
25. LC loading solvent: 0.4% formic acid in water.
 HDX-MS of Protein-DNA Complexes 197

26. LC solvent A (0.4% formic acid, 2% acetonitrile, 97.6% water).

27. LC solvent B (0.4% formic acid, 95% acetonitrile, 4.6% water).
28. Electrospray ionization (ESI) mass spectrometer (any
ESI-equipped MS with reasonable resolution is usable; in our
case, we used an electrospray ionization—Fourier transform
ion cyclotron resonance 15T solariX XR, Bruker Daltonics).
29. Software: DataAnalysis 4.1 (Bruker Daltonics), ProteinScape 4
(Bruker Daltonics), DeutEx (http://peterslab.org/downloads.
php under HDX tools section), mMass (http://mmass.org),
PyMol (https://pymol.org/; v 2.1.0), MSTools (http://
peterslab.org/MSTools/).

3 Methods

3.1 Verification Prior to the HDX-MS experiment, it is advisable to verify whether

of Protein–DNA the DNA duplex is present and that the complex between protein
Complex Formation by and ligand (here DNA) is indeed formed upon mixing of the
a Gel Shift Assay components. In this respect, knowledge of the binding constants
offers an advantage as the experimental setup can be adjusted to
achieve full protein occupancy (complex presence) during the label-
ing step [18]. Here, we use gel shift assay to check DNA duplex
formation and its binding to the studied protein.
1. Dilute both DNA strands in LC-MS purity water and mix them
at equimolar ratio.
2. Heat the mixture at 95
C for 1 min, and let it slowly cool
down to room temperature.
3. Dilute/buffer exchange the protein into a buffer suitable for
the complex which is later used for H/D exchange (20 mM
HEPES, pH 7.4, 150 mM NaCl, see Note 5) to a concentra-
tion at which full occupancy of the protein by DNA is achieved
(see Note 6).
4. Add equimolar amounts of dsDNA to the protein and let the
mixture equilibrate for at least 2 min at a temperature where
both the dsDNA and protein are stable (see Note 7).
5. Prepare following samples: 200 ng of both single-stranded
oligonucleotides, 200 ng of free dsDNA, 5 μg of free protein,
and the equimolar protein–DNA complex equal to 5 μg of the
protein (see Note 8). All components must be prepared in the
same buffer.
6. Add loading buffer to all samples.
7. Prepare non-denaturing polyacrylamide gel by mixing all com-
ponents except APS. Add APS just before pouring the mixture
into a gel casting tray (see Note 9).
198 Ruzena Filandrova et al.

8. Load the samples and run the electrophoresis in TBE buffer at a

constant voltage of 50 V for one gel until the loading dye
reaches half of the gel (see Note 10).
9. Stain the gel with any kind of fluorescent dye for DNA staining,
take a picture, and stain the gel again in protein staining solu-
tion to visualize the protein. Only proteins in complex with
DNA will migrate into the gel. Free protein will normally not
separate as TFs are generally positively charged and interact
strongly with negative DNA.
10. Formation of DNA duplex and protein–DNA complex are
manifested as shifts in the electrophoretic mobility (Fig. 1).
Compare the results from both the staining methods to con-
firm the complex formation—position of the complex band
should be the same for protein and DNA staining.
11. If multiple complex concentrations are tested, select the one
where there is the least free dsDNA (of the same gel mobility as
dsDNA control) visible.

Fig. 1 Gel shift assay. Two oligonucleotides (DNA1 and DNA2) with different
binding affinities are shown. Shifts in electrophoretic mobility in the third and
fifth lane from the left indicate duplex DNA and protein–DNA complex formation,
respectively. Protein control lane stayed empty since no DNA was present. In
both the complex lanes, a band of the same mobility as dsDNA can be observed,
though in the “higher KD complex” lane, this band is much thicker, indicating
lower bound fraction of the protein
 HDX-MS of Protein-DNA Complexes 199

3.2 Optimizing Before starting the H/D exchange, it is crucial to verify how the
Digestion Conditions protein is digested under HDX-MS compatible conditions. Differ-
ent aspects to be considered are protein sequence coverage, peptide
length, and peptide redundancy but also whether the protein
remains soluble after quenching and freezing and how the presence
of DNA affects the whole procedure. The primary goal is to ideally
obtain full sequence coverage with peptides providing good spatial
resolution, i.e., not very short or long ones. Recommendations
regarding optimal length do vary slightly in the literature, but
fragments between 8 and 12 amino acids probably represent the
best standard. Besides the spatial resolution obtainable through the
generated peptides, one can also target nearly amino acid resolution
using suitable fragmentation techniques like electron transfer dis-
sociation [19] or UV photodissociation [20] (see Note 11). In
addition, emphasis should also be put on the generation of large
numbers of overlapping peptides as these can further increase the
spatial resolution [21, 22] and/or provide higher confidence in the
observed changes (see Note 12). To fulfill all these goals, different
proteases, ideally immobilized and packed into a column, should be
tested [17, 23–26]. The columns can be used individually or com-
bined in serial or parallel setting [27]. Column size as well as the
protease density on the POROS resin, which together dictate the
enzyme:protein ratio, can be varied. Other factors affecting the
digestion are flow, temperature, pressure, and the use of denaturing
agents. Time spent on the proteolytic column is primarily deter-
mined by the desalting LC flow. Therefore, higher flow leads to
faster digestion and produces longer fragments and higher redun-
dancy (number of overlapping peptides). Slower flow, on the other
hand, produces more complete digestion, which however may at
times be even detrimental, if the peptides produced are too short.
Typical digestion temperature should be close to 0
C to minimize
deuterium-loss (back-exchange), but it can be raised to 15
C or
even 20
C locally to increase the digestion efficiency. This is usually
achieved either in a dedicated separately temperature-controlled
chamber for the protease column or by placing the column out of
the ice/water bath. However, when tweaking these parameters, one
should always keep in mind that optimal balance between efficient
digestion and minimal H/D back-exchange conditions must be
maintained (hence running the analysis as fast as possible and at
the lowest possible temperature). For proteins not offering satisfac-
tory digestion under mild denaturing conditions provided by the
acidic environment of the quench buffer alone, addition of dena-
turing agents is suggested. Urea or guanidine can be used at quite
high concentrations with all the immobilized proteases; however,
besides rendering the target protein susceptible to digestion, these
agents also affect the protease itself as well as peptide binding to the
reversed phase resin in the trap column. While guanidine was
shown to have mainly a detrimental effect, urea can even enhance
200 Ruzena Filandrova et al.

proteolytic activity [17]. In addition, guanidine is difficult to

remove completely during the subsequent desalting step, and
thus, 4 M urea (final concentration upon quenching) may always
be suggested as the first choice.
When analyzing samples containing nucleic acids, specific pro-
blems may additionally arise upon the quenching of H/D
exchange, as mentioned above. As the pH drops, nucleic acids
tend to precipitate, and it is very likely that the protein of interest
will coprecipitate as well. Therefore, several strategies described in
the Introduction were developed to remove the nucleic acids after
quenching or to reduce their detrimental effect on the HDX-MS
workflow.
1. Prepare the duplex DNA according to the steps 2 and 3 of the
previous chapter (Subheading 3.1) and mix it with the protein
in equimolar ratio and the concentration that will later be used
during D2O labeling. Incubate the mixture for 10–20 min to
ensure binding equilibrium. Also, prepare a sample without
DNA for the comparison of how DNA affects the digestion
pattern. In this step, use normal (H2O, not D2O based)
buffers.
2. Start the LC-MS/MS system—calibrate mass spectrometer,
start analytical gradient pump (Agilent 1290, running at
40 μl/min), loading pump (Agilent 1260, running at 100 μl/
min), and pre-cool the LC setup to 0
C (Fig. 2). If digestion is
carried out at higher temperature, make sure that the protease
column is well conditioned (see Note 13).

Fig. 2 Schematic representation of HDX LC-MS setup. Dimensions of the columns and tubing are shown. The
setup is in position where digestion and desalting are running. Upon switch of the switching valve, the ports
are connected by the gray dashed lines and gradient elution of the desalted peptides from trap column onto an
analytical column is performed
 HDX-MS of Protein-DNA Complexes 201

3. Run a standard protein (e.g., horse heart myoglobin or rabbit

muscle phosphorylase B) to verify the LC separation and pro-
tease column activity. This standard protein used for system
quality control must be different from the protein of interest.
Always keep in mind that the protein must be prepared in an
acidic buffer as the digestion is done using aspartic proteases.
Accidental injection of neutral pH buffer will quickly and
irreversibly inhibit pepsin. Other proteases mentioned here
are more robust in this regard and can tolerate elevated pH
temporarily. Typically prepare 30 pmol of myoglobin in 100 μl
of 250 mM glycine-Cl buffer pH 2.3.
4. Run two blank injections (100 μl of pure quench buffer) to
clean the system and to ensure proper and stable LC conditions
(see Note 14).
5. Mix the sample with the quench buffer to get the sample
volume and concentration as planned for subsequent
HDX-MS experiment (see Note 15). Hence, 50 μl of
500 mM glycine-Cl buffer pH 2.3 with 50 μl of 2 μM protein
solution in 20 mM HEPES, pH 7.4, 150 mM NaCl.
6. Inject the sample directly or subject it to a freeze–thaw cycle if
you plan to collect aliquots by freezing in liquid nitrogen.
7. Wait for desired time (3 min) until digestion and desalting are
finished (see Note 16) and switch the second LC valve so that
the trap column is now in the path of gradient elution and the
peptides will continuously elute on the analytical column.
Together with the switch, also start the gradient (see Note
17) and collect the MS data.
8. Process the LC-MS/MS file using the instrument vendor-
specific software and generate input for the search engine (see
Note 18).
9. Run a database search using the input file from the previous
step. Use a custom-made database containing the sequence of
the protein of interest and the protease(s) used and eventually
other proteins present in the sample. Do not set any digestion
preferences, nor taxonomy filtering. Include possible relevant
modifications (fixed or variable) that may occur in the protein.
Set the mass tolerance on precursor and fragments according to
your instrument performance. Use scoring routines to discard
improbable matches. It is also advisable to run the LC-MS/MS
analysis 3–4 times and only use peptide identifications that
occur in the majority of the analyses.
10. Copy or export the search result (list of identified peptides) to a
spreadsheet editor (e.g., Microsoft Excel) and extract the Pep-
tide Start and End columns (peptide limits). Use this in a
simple text file to visualize the coverage map using DrawMap
202 Ruzena Filandrova et al.

Fig. 3 Sequence coverages of the two studied DNA binding domains. In all instances, immobilized proteases
were used. Table at the right lower corner of each panel shows digestion metrics for the individual conditions
(coverage, number of peptides, average peptide length, and redundancy—how many peptides are on average
covering each residue). (a) DNA binding domain of FOXO4 where initial proteolysis by pepsin (blue bars) was
after optimization replaced by combined digestion with pepsin followed by nepenthesin-1 (red bars). This led
to better spatial resolution and higher redundancy. In case of FOXO4, the presence of oligonucleotides
(DAF-16) caused no problems, and thus, only tuning of the digestion itself was necessary. (b) Examples
from the tuning procedure with TEAD1 DNA binding domain. Pepsin (blue bars) provided full sequence
 HDX-MS of Protein-DNA Complexes 203

script at the MSTools website (http://peterslab.org/

MSTools/DrawMap/DrawMap.php) and eventually use
Excel macro Digestion_metrics.xlsm (http://peterslab.org/
downloads.php under HDX tools section) to extract the diges-
tion metrics.
11. Based on the peptide length, redundancy, coverage, and num-
ber of gaps in the peptide map, try optimizing the digestion by
using alternative proteases or quench buffer additives as
described above (example in Fig. 3) (see Note 19).

3.3 Hydrogen/ The methodology described here represents a manual approach.

Deuterium Exchange Even though nowadays there are robotic solutions available that
can do the sample preparation and injection automatically, they are
quite specific depending on the manufacturer of the robot and the
software controlling it. The manual approach represents a straight-
forward and inexpensive way that can be easily set up in any labora-
tory possessing an LC system and a mass spectrometer (see Note
20).
The actual H/D exchange should be started from equilibrated
solutions in which the complex is fully formed, and the temperature
of all components is stable. The first step in the deuterium labeling
reaction is the dilution of the studied protein into a deuterated
buffer (see Note 21) that has otherwise identical composition as the
initial buffer used for complex pre-formation. After a predeter-
mined time, the exchange is “stopped” by the addition of acidic
quench buffer and by lowering the temperature or eventually by
flash freezing in liquid nitrogen.
The experiments should ideally be done (at least) in triplicate to
allow for statistical validation. There are several ways how to per-
form the replication in HDX-MS. The highest validity is achieved
by doing a biological replication (in the case of recombinant pro-
teins this means using independent protein production batches). A
less significant (though commonly used) level of replication (tech-
nical) is reached by separate deuterium labeling, and the lowest tier
is represented by measuring the same labeling reaction repeatedly
[28]. Replication also allows one to estimate the significance level
above which the differences in deuteration between individual
states may be considered meaningful [29].
ä

Fig. 3 (continued) coverage but lower spatial resolution—on average it had longest peptides. Nepenthesin-
2 (red bars) led to over-digestion, which is indicated by gaps in the sequence, and the identified peptides are
mostly short ones. High activity of the protease resulted in the generation of short peptides and incomplete
sequence coverage. The green bars show the final conditions where aspergillopepsin (protease type XIII) was
utilized in the presence of urea. Addition of denaturing agent was necessary as the protein in the presence of
oligonucleotides precipitated. Addition of urea (optimized to final 2 M concentration) led to much better protein
recovery. In both examples shown here, the advantage of alternative proteases for the digestion of DNA
binding proteins (which are rich in basic residues) is demonstrated through their ability to cleave after the Lys
and Arg residues (highlighted by arrowheads above the sequence)
204 Ruzena Filandrova et al.

Starting from the moment of quenching and throughout the

digestion, desalting, and subsequent analysis, the sample is exposed
to protonated solvents. Even though the minimal exchange condi-
tions are maintained through these steps, the sample still undergoes
deuterium loss. The level of back-exchange is not uniform and
depends on the sequence of individual peptides. However, this
loss can be measured by analyzing a fully deuterated sample and
then using these data, to correct deuteration levels on partially
deuterated peptides [28]. In a typical measurement comparing
two protein states, relative deuterium levels are the desired mea-
sure, and there is no need for back-exchange correction. Thus, in
these cases, this control is often omitted. This is also likely due to
the fact that no clear and universal protocol exists and that proteins
often tend to precipitate or degrade before their full deuteration is
achieved. However, the knowledge of the actual back-exchange is
needed when absolute levels of deuteration are sought (e.g., in
cases where mutated sequences of the same protein are
compared) [30].
1. Prepare the required number of 0.5-ml Eppendorf tubes (for
two conditions, three labeling replicates, and six time points,
this would mean 36 tubes in total) with quench solution—
50 μl of 500 mM glycine-Cl buffer, pH 2.3 (see Note 22).
2. Prepare (bigger) tubes with the protein alone and with the
pre-formed protein–DNA complex. In these, the ten-fold dilu-
tion into deuterated buffer and the labeling reaction will be
performed. At six preselected time points (see Note 23), 50 μl
aliquots will be removed from the reaction and quenched. As
three labeling replicates are to be done, three tubes with pro-
tein alone and three with the protein–DNA complex, each
containing 40 μl of 2 μM protein solution must be prepared
(see Note 24).
3. Prepare 2 ml of deuterated buffer (20 mM HEPES, 150 mM
NaCl, pD 7.4) (see Note 25).
4. Prepare a time schedule that allows efficient pipetting especially
in cases where larger amount of conditions or direct technical
(labeling) replicates are performed at once. This can be easily
done using MSTools script “Experiment planner” (http://
peterslab.org/MSTools/HDExpPlanner/HDExpPlanner.
php). For two experimental conditions, each replicated three
times, and aliquot collection at 20 s, 2 min, 5 min, 20 min, 1 h,
3 h, this may look as shown in Fig. 4.
5. Prepare automatic pipettes with pre-set volumes for H/D mix-
ing (360 μl) and aliquot collection (50 μl).
6. Use a timer that allows countdown followed by a count-up. Set
20 s count-down and start it. Aspirate 360 μl of the deuterated
buffer and when the timer reaches zero (start of the whole
 HDX-MS of Protein-DNA Complexes 205

Fig. 4 Screenshot of HD Experiment planner script (MSTools: http://peterslab.org/MSTools/HDExpPlanner/

HDExpPlanner.php). Exact planning of aliquot collection for two experimental conditions (TF and TF-DNA) and
three labeling replicates. Delays necessary to collect the aliquot or mix the exchange can be varied and are
especially useful for larger number of conditions—here 30 s is left for aliquoting and 1 min for mixing

HDX experiment), add the buffer to the protein. Mix thor-

oughly and prepare the pipette (50 μl) for aliquot collection.
Five seconds before the selected time (20 s), aspirate the
labeled protein and at the precise time according to the experi-
ment plan, add aspirated solution to the tube with the quench
solution. Quickly mix/vortex and flash-freeze the tube in liq-
uid nitrogen. Do this for all the samples according to the time
schedule at times indicated by the timer.
7. Transfer the frozen tubes from liquid nitrogen to a deep freezer
and store them until analysis (see Note 26).
8. Also prepare non-labeled reference samples (a non-deuterated
control) for each condition. This is the same as preparing the
(partially) deuterated samples, except that the buffers used are
made with H2O only.
9. If possible (depends on protein stability), prepare fully deuter-
ated sample (see Note 27). In order to obtain a fully deuterated
control, prepare deuterated buffer containing high concentra-
tion (4–6 M final concentration) of urea or guanidine and
incubate the protein in it overnight at higher (e.g., 37
C)
temperature. Protein concentration should be as high as feasi-
ble, so that it can be lowered before quenching using
206 Ruzena Filandrova et al.

non-denaturing deuterated buffer. This will lower the denatur-

ant concentration which can alter the subsequent workflow
outcomes. From this perspective, using urea is the better
option (see Subheading 3.2). Quench the reaction and eventu-
ally freeze the sample, if this step is included in the workflow.

3.4 Mass 1. Use the LC system as in Subheading 3.2 and perform steps
Spectrometric 2–4. Start all LC pumps, calibrate the mass spectrometer,
Analysis of Deuterated pre-wash, and pre-condition the protease column. Then run
Samples the standard protein followed by two blank injections (see
Note 28).
2. Before the end of previous analysis, take one of the partially
deuterated samples from the freezer and start thawing
it. Depending on the quench buffer composition, it will melt
in 30 s to 1 min. As soon as it is thawed, inject it for analysis.
3. Repeat this for all collected samples (see Note 29).
4. Analyze fully deuterated control exactly as the partially deuter-
ated samples.
5. Finally run two blanks followed by non-deuterated samples for
each experimental condition.
6. Export and/or pre-process all LC-MS acquired data and pre-
pare them for final data processing.

3.5 Interpreting Data Nowadays, there are several different programs that are capable of
from H/D Exchange largely automatic HDX-MS data processing. Waters users rely on
their DynamX suite while others can use HDExaminer from Sierra
Analytics that supports all native file types as well as open source
ones. Another option for Waters and Thermo users is represented
by the HDX Workbench [31] from Omics Informatics. Alterna-
tively, there are software tools freely available from different
research groups [22, 32–35]. These and other programs and their
workflows were recently reviewed in detail by Claesen and Burzy-
kowski and also by Eggertson et al. which in addition offers another
view on HDX-MS data processing [36, 37].
The basic principle which is crucial to the understanding of the
workflow and its requirements is, however, the same. To make this
protocol widely applicable, we will show two possible scenarios.
One, relying on a manual interpretation which is nowadays out-
dated and extremely laborious. However, it demonstrates well the
basic principle which the available programs automate. It can also
be useful for validation purposes and in specific cases (extraction of
EX1/EX2 data). The other workflow presented here employs our
own software called DeutEx (see Note 30).
Data processing in the HDX-MS workflow consists of four
major steps. First is the identification of peptides generated during
the proteolysis step and their temporal localization within an LC
 HDX-MS of Protein-DNA Complexes 207

run. Using the set of coordinates consisting of peptide sequence, its

mass-over-charge (m/z) value(s) (see Note 31), and LC retention
time, individual features can be located in the partially deuterated
samples. In a subsequent step, information about the number of
deuterons carried by individual peptides is extracted. Two major
approaches to deuteration readout rely either on centroid readout
or on deconvolution and theoretical fitting [36]. Finally, this infor-
mation is presented in several graphical ways to visualize the effects
monitored by the HDX-MS experiment.

3.5.1 Workflow A: 1. From an MS/MS search engine, e.g., MASCOT (here utilized
(Largely Automated) Data by Bruker’s ProteinScape), export the search result in the form
Interpretation Using DeutEx of a csv file (see Note 32).
2. Prepare a separate simple text file containing the sequence of
the studied protein in FASTA format.
3. Open DeutEx and go to Analysis>Compose digest file from MS.
New window will appear. In this window, upload your protein
sequence as a FASTA file, then csv file containing the list of
identified peptides from MASCOT and an LC-MS run of a
non-deuterated sample exported into simple text files (each MS
scan corresponding to one file) containing a list of m/z values
and their intensities (see Note 33).
4. Go to Digest, click Create from report and then Extract scan/
charge limits. Each peptide will have its retention limits auto-
matically identified, and extracted ion chromatograms will be
drawn for individual charge states (m/z values). User can man-
ually inspect the assignments and either manually correct or
completely reject them. Once finished, the result should be
exported into a text file that serves as a lookup file for an
automated assignment of HDX data for the deuterated sam-
ples. Close the “Compose digest from MS window.”
5. Go to Analysis>Open Analysis directory. It is advisable to
pre-compile data for each state into a separate folder (which
contains exported LC-MS data in a txt format, with appropriate
fasta sequence and digest file—the list of peptides created in
step 4). However, the data can also be compiled in DeutEx
using the “Compose analysis” function.
6. Once all data are opened, check the settings for mass accuracies
and filtering under the options menu and run the calculation.
Typically, it lasts from seconds to a few minutes, depending on
the number of conditions, number of peptides, and size of the
exported data on a common laptop/office computer. When
calculation is finished, run filtering, which tries to automatically
remove incorrect assignments. Then look through the assign-
ments manually, check the data, and correct any misassign-
ments, if needed.
208 Ruzena Filandrova et al.

7. The basic output from DeutEx is a set of uptake plots that are
displayed either as deuteration percentage or number of
exchanged deuterons as a function of time. The data can also
be exported in simple text form which allows their further
processing in Excel or MSTools (http://peterslab.org/
MSTools/) [38].

3.5.2 Workflow B: 1. In any program that is used for viewing the LC-MS data, open
Manual Data Interpretation a non-deuterated LC-MS analysis.
2. For each peptide, trace an extracted ion chromatogram based
on its m/z (usually included in the search engine report).
Check also other charge states that are visible and enter this
information into an Excel sheet together with the retention
time limits for further use. Sum scans across the LC peak and
export the spectrum as a simple text file (typically two columns
including m/z values and intensity).
3. Open a partially deuterated LC-MS analysis and locate the
deuterated signal corresponding to the peptide information
gathered in the previous step. Repeat the step of summing up
the scans and exporting the spectra into text files. Repeat
iteratively for all peptides. For such a manual approach, it is
advisable to select a minimal set of peptides (preferably short
ones) covering the entire protein without many overlaps (see
Note 34).
4. Once the data are exported to *.txt files, run mMass program
[39] (see Note 35). Go to File>Open and open a file with
non-deuterated data. The spectrum is not by default labeled,
but it is not necessary at this point. Make sure that the correct
way of peak picking is selected. Go to tab Processing>Deisotop-
ing and set “Label envelope tool” to Envelope Centroid option.
Now locate the selected peak (m/z) and zoom to the mass
range covering the whole isotopic pattern of this peptide. Use
“Label envelope tool” (select from tab Tools). Using Shift +
mouse middle wheel scrolling, set the desired charge state—
black dots are changing distance as you scroll up and down. Fit
it to the desired isotopomer distance and now select the num-
ber of dots to cover all visible isotopic peaks from the particular
envelope. This is done by Ctrl + Shift + mouse middle button
scrolling. This tool helps you to localize all isotopes belonging
to the selected peak. Left click and the peak will be labeled.
Average m/z value is written into the spectrum, and the decon-
voluted value (average mass) appears in the right panel (Fig. 5).
Write this number down to an Excel file. Eventually, you can
now repeat this step also for other charge states of the same
peptide. One by one, all m/z values of each peptide from all
deuterated samples are processed using this approach (see
Note 36).
 HDX-MS of Protein-DNA Complexes 209

Fig. 5 Screenshot of mMass window with zoom on the isotopic pattern. Black dots are labels showing which
isotopes will be used for Mavg calculation. The deconvoluted average mass is shown in the right panel

5. When all peptides are processed and the Excel file complete, the
percentage of deuteration or the number of deuterons can be
easily calculated [2] and used for visualization of the results.

3.6 Data There are no set guidelines on how to visualize HDX data, and
Visualization indeed, a multitude of different plots are used throughout the
literature. The first result that is usually obtained is a set of uptake
plots in which the percentage of deuteration or the number of
deuterons is plotted for each peptide as a function of time. How-
ever, from such a display, it is difficult to interpret the data in the
structural context and also to efficiently handle the information
from overlapping peptides. For a more comprehensive view, heat
maps or protection plots [38] are most often utilized. These sum-
marize the information into one picture and present it along the
protein sequence. However, their basic form does not allow one to
visualize information from the overlapping peptides as the data are
plotted using only a subset of selected (usually shortest) peptides
continuously covering the entire sequence. To also include the
redundancy information, peptide midpoints can be calculated
[29] for each peptide, which allows the distinction of closely related
ones (e.g., 14–25, 14–27, 16–25, 16–27). Upon this modification,
210 Ruzena Filandrova et al.

the data can also efficiently be used in mirror (or so-called butterfly)
plots or in bar graphs. The disadvantage of this data display is that
the proportions of the peptides versus the protein sequence are lost.
Probably one of the most attractive ways of data presentation is thus
mapping the observed changes onto an existing high-resolution
structure or protein model. In this case, differences between two
protein states are calculated (preferably for the number of deuter-
ons not for percentages), and the structure is colored according to
the observed changes. Visualization tools are usually included in
the HDX specific software (DynamX, HDExaminer, etc.,); how-
ever, our web-based MSTools suite makes many visualization tools
available for vendor-independent use [38].
Moreover, for structure coloring, PyMol with its scripting
capabilities is unparalleled. In this workflow, the b-factors of pdb
format’s fields in an existing protein structure are hijacked and
replaced with the numbers from an HDX experiment. To load the
data into PyMol, it is necessary to cut each peptide virtually into
individual amino acids and use this as an input for coloring.
1. Prepare a table which contains information about each pep-
tide’s start, end, time of deuteration (in seconds), and the
difference in deuteration between selected states:

Start End Time(s) (TF-DNA—TF)

1 15 20 0.3
1 15 120 0.5
1 15 300 0.6
1 15 1200 0.8
1 15 3600 1.5
1 15 10,800 2.1
8 17 20 0.2
8 17 120 0.4
8 17 300 0.5

Save this table as tab delimited simple text file into a new
folder.
2. Copy Python script HDXPeptideSplitter.py (see Note 37) into
this folder and run it (see Note 38). Command window will
appear, and the script will ask for the name of the input file.
Then it requires information about the time to be parsed
(selected time or all the times in the file) and finally about the
conditions to be extracted (again selected one or all). The script
then creates a set of files depending on the number of condi-
tions and time points in the experiment. Names of the files are
 HDX-MS of Protein-DNA Complexes 211

created using the name of the input file, time, and number of
the column in the input file (e.g., FOXO_20_2.txt means 20-s
time point and second column from the input file which was
named FOXO.txt).
3. Open PyMol and load the pdb file with the structure. Using
“copy” (https://pymolwiki.org/index.php/Copy) command,
get the desired number of copies corresponding to the number
of states and conditions prepared in the previous step. Also
name the objects so that the names correspond to the condi-
tions and times from the dataset.
4. Within the PyMol command line, activate script data2bfactor.
py (see Note 39) by the command “run path/data2bfactor.py”
(where “path” stands for the absolute path to the folder, where
the script is saved on your computer), e.g., run C:\Python27
\Scripts\data2bfactor.py.
5. Reset the b factors to a number that is outside the dataset (e.g.,
99). This will later allow you to color the regions that are not
covered by the peptides in the dataset. Enter command “alter
object_name, b¼99” (where “object_name” stands for the
name of an individual object/structure chosen by the user).
6. Color the structure(s) to some neutral, uniform color using
command “color.” E.g., “color grey70.”
7. Apply your data onto the individual structures. Use command
“data2b_res object_name, path.” For example, data2b_res
FOXO_20_WT-MUT, C:\Data\FOXO_20_2.txt (i.e., second
column of the input file for HDXPeptideSplitter.py script
contained difference between wild-type and mutant). Do this
for all structures/objects in PyMol and all input text files.
8. Now select residues with all negative and positive b values,
respectively. Enter commands “select negative, b<0.0” and
“select positive, b>0.0.”
9. Color these selections using commands “spectrum b, selec-
tion¼negative, minimum¼-X, maximum¼0.0, palet-
te¼blue_white” and “spectrum b, selection¼positive,
minimum¼0.0, maximum¼X, palette¼white_red,” where X is
the highest/lowest number in your dataset. Negative (white-
to-blue gradient) differences are regions where the exchange
was lowered, and positive (white-to-red gradient) are regions
with increased deuteration.
10. Finally, differentiate the regions that were not covered by pep-
tides in the dataset. Run command “color black, (b¼99).”
11. Optionally show a colorbar with the scale using command
“ramp_new colorbar, none, [-X, 0.0, X], [blue, white, red].”
212 Ruzena Filandrova et al.

4 Notes

1. The influence of pH and temperature on the intrinsic chemical

exchange rate must be considered here. If two or more states of
a protein are followed at different pH values or temperatures,
appropriate correction must be applied. Temperature factor can
be calculated from the Arrhenius equation (e.g., comparison of
exchange at 10
C and 37
C will be corrected by a factor of
13.88). The effect of pH is compensated by 10ΔpH value (e.g.,
exchange at pH 7 runs 100 times faster than exchange at
pH 5) [3].
2. Total amount depends on protein size, the LC system used,
and the sensitivity of the mass spectrometer. However, tens
(small protein) to hundreds (large protein) of micrograms are
enough for the comparison of two states and performing trip-
licate analysis.
3. Check the oligonucleotides for possible secondary structure
formation which may affect their binding properties. When
preparing/synthesizing or ordering the oligonucleotides, use
HPLC purification as the last step to obtain desalted DNA
samples.
4. In case where water/ice mixture is used, the valve body should
be properly sealed and isolated from the water (parafilm and
heat shrink tubing) as it prolongs the lifetime of the valve.
5. A wide range of buffer components or concentrations is toler-
ated in H/D exchange workflow as long as they can be
removed during desalting on a reversed phase resin that pre-
cedes separation and mass spectrometric detection. Neverthe-
less, the buffer components should be kept at their lowest
concentration where both the free protein and the complex
are still stable.
6. At least ten times KD [18]. If KD is not known, try multiple
concentrations. Keep in mind that the complex will be diluted
during D2O labeling. Do not use excess of DNA since high
DNA concentration can lead to sample precipitation upon
quenching.
7. Depending on the melting point temperature of the used
dsDNA, cooling might be needed.
8. Try to keep the volume of samples as low as possible to prevent
diffusion from the wells during sample loading—ideally under
5 μl.
9. The concentration of the gel should be adjusted based on
oligonucleotide length.
 HDX-MS of Protein-DNA Complexes 213

10. If possible, cool the electrophoresis apparatus during the run

since the complex or DNA duplex might not be stable at higher
temperatures.
11. It should be considered that the fragmentation of a particular
peptide does not always lead to sufficient number of fragments.
Therefore, the combination of different digestion protocols
and dissociation techniques usually represents the best
solution.
12. Fragments covering the same region of protein and differing in
just few amino acids from the N- or C-terminus should result
in fairly similar H/D uptake plots. Hence, the slope, level of
deuteration, or eventually the differences between the individ-
ual studied states should not be vastly different. If they are, this
might be an indication of a possible data processing error. On
the other hand, if there are multiple individual peptides span-
ning one protein region and all show highly similar trend, the
confidence in the observed changes is significantly increased.
13. Protease column may suffer from autodigestion upon longer
storage. Storing it at conditions that do not affect its activity,
while preventing autodigestion (e.g., citrate buffer, pH 4–5,
4
C), is highly recommended. Also, pre-washing with the
loading solvent before connecting the column in front of the
trap column should be routinely done during an LC system
setup.
14. Specific cleaning solutions containing denaturing agents or
higher concentration of acids can be designed to minimize
carry-over [40]. Good regeneration of the whole system is
likely achieved by combining injection of some other (quality
control) protein together with these washing steps.
15. Quench conditions (quench solution composition, concentra-
tion, quench ratio) should be tuned prior to the experiment
using pure buffers. While protein concentration during the
labeling step is given by the requirements on the protein–
DNA complex formation, the final concentration and sample
volume after quenching depend solely on the requirements of
the mass spectrometer and LC setup (injection volume, trap,
and protease column size + flow). For the MS analysis, it is
worth not considering the true detection limits of the instru-
ment, but staying well above them, as the signal intensity after
labeling decreases due to signal spreading to multiple isotopes
upon deuteration. Keeping the same protein concentration for
LC-MS/MS as for the subsequent LC-MS analyses is crucial as
the protein:protease ratio affects the digestion pattern.
16. Sum the volumes of injection loop, protease column, trap
column, and the connecting tubing and compare it to
the selected flow on loading/desalting pump. Make sure that
214 Ruzena Filandrova et al.

the sample will reach the trap column and several volumes of
the trap column will pass through to achieve efficient desalting.
If guanidine or phosphates are used in sample preparation,
make the desalting slightly longer. This step usually lasts from
1 to 4 min, depending on the flow and volumes mentioned
above.
17. Duration of the gradient and its slope depend on the column
and the LC system used. In general, the gradient should be
fast, lasting just a few minutes. However, during this short
period, the most efficient elution should be achieved. There-
fore, gradient running from 5 or 10% acetonitrile (solvent B)
and ending at 25–35% acetonitrile should be used. In specific
cases, the upper acetonitrile percentage can be raised to 40 or
50% or even higher.
18. For example, here the data are processed in DataAnalysis using
Find>Compounds AutoMS(n) feature and exported into a
MASCOT generic file (*.mgf). Two different algorithms are
used for peak picking (FTMS and SNAP) and thus two *.mgf
files are created for each analysis. The whole procedure can be
automated using Visual Basic macros in Data Analysis. These
macros can be obtained from the authors.
19. If the coverage map shows many gaps and the covered parts are
represented by short peptides, the protein was probably over-
digested. Faster flow (even in hundreds of microliters per
minute) should be tested. One should, however, keep in
mind the trapping efficiency of the trap column at a given
flow rate. Alternatively, another protease may be used. These
steps are also advisable if higher redundancy is desired. In cases
where the peptides are too long, temperature should be
increased or addition of a denaturing agent considered. Use
of alternative aspartic proteases beside pepsin is advantageous
for DNA binding proteins since they tend to be rich in basic
residues. While pepsin exerts nearly no cleavage after Lys, His,
and Arg residues, the other proteases like nepenthesins, asper-
gillopepsin, and rhizopuspepsin cleave at these sites with high
frequency [17, 24, 26]. These proteases may thus yield good
spatial resolution of basic proteins or basic protein regions.
20. It is often stated that the use of robotics leads to higher
reproducibility; however, trained user and careful pipetting
can lead to very similar outcomes. The main advantage in
using automation is the ability to perform the analysis in an
unattended manner and thus using the instrument time more
efficiently. Most importantly, the manual way utilizes freezing
of samples and their subsequent thawing. These steps require
additional 0.5–2 min, during which the sample undergoes
back-exchange or (in the case of low D2O dilution by the
 HDX-MS of Protein-DNA Complexes 215

quench buffer) artificial, nonnative, in-exchange. In contrast,

robotics can set up the mixing of H/D exchange samples
consecutively, so that once the exchange is finished by the
quench solution addition, the sample is directly injected for
analysis. It is also of high importance that some proteins tend
to precipitate during freezing/thawing procedures and here
the use of robotics provides additional advantage.
21. Dilution is often done ten-fold, but higher (20-fold) or lower
(five-fold) ratios are also possible. The key consideration here is
the stability of the complex at a given concentration which is
influenced by the KD of the complex.
22. The amount/volume of the quench solution is given by the
quench ratio. Optimally, ten-fold dilution is used but other
ratios, e.g., 1:1 are also acceptable, if necessary. Quench solu-
tion must be acidic and strong enough to override pH of the
labeling buffer and should make it 2.3–2.5, instead. Diluted
solutions of hydrochloric or phosphoric acid, phosphate buf-
fers, or glycine hydrochloride buffers are typical examples.
Quench solution may also contain denaturing agents or other
components necessary for the solubility of the protein or its
analysis. Only components that are not easily removed by
reversed phase desalting and are not compatible with ESI-MS
should be avoided (e.g., detergents). However, even for non-
ionic detergents, specific workflows allowing their nearly com-
plete removal under HDX-MS conditions have been
described [41].
The tubes used for freezing should be PCR tubes as they
have to be stable at extreme temperature. Normal Eppendorf
microcentrifuge tubes can break or take up some liquid nitro-
gen, which can lead to their explosion during the subsequent
thawing procedure.
23. Preferably run H/D exchange with several time points span-
ning the region from seconds to a couple of hours. For longer
incubation times, a control of protein stability should be per-
formed by incubating the protein for the longest selected time
in the normal buffer and then deuterium labeling it for
the shortest selected period (e.g., 20 s). If the deuteration is
the same as for the sample just deuterated for this time window,
the protein is probably stable. If the deuteration profile differs/
is indicative of protein unfolding, shorter time windows should
be sought or the exchange should be done at lower
temperature.
24. Simple setup requires two protein states: protein alone and
protein in complex with its DNA binding partner. However,
number of conditions to be followed can be much larger, e.g.,
with different protein:DNA ratios or different ligands.
216 Ruzena Filandrova et al.

25. Deuterated buffers must be set to pD (not pH). There is a

simple correction factor compensating the difference of mea-
suring deuteron concentration using normal pH electrodes
pD ¼ pHread + 0.41 [42]. Alternatively, pH reading (clearly
marked as such) can be reported instead of the pD value.
26. Samples should be analyzed as soon as possible, e.g., within a
few days. Do not store them for more than 1 week, even at very
low temperatures.
27. Some proteins are not stable enough to allow the preparation
of a fully deuterated control. For the majority of comparative
HDX-MS experiments, its inclusion is anyway not crucial, but
it can offer valuable insight into the back-exchange levels and
provides good quality control of the system used.
28. Tune the mass spectrometer properly to avoid hydrogen scram-
bling in the ion source [43]. That mainly means lower the
desolvation temperature and possible ion activation during
ion transfer. These low-scrambling parameters may, however,
lead to decrease in signal intensity.
29. It is best not to analyze the samples in any ordered manner,
e.g., sequentially by the conditions or replicates. If the analysis
is stopped or interrupted, recondition the system by running
two or three blanks or a standard protein and two blanks.
Cleaning procedures should be included, and potential carry-
over checked during the analysis. The biggest source of carry-
over is the protease column, which can, however, be washed
during each run [40]. Try to keep the system running contin-
uously so that maximal LC reproducibility is maintained.
30. DeutEx is not yet publicly available but will be offered at our
website: http://peterslab.org/downloads.php under HDX
tools section.
31. It might be beneficial to search for all detected charge states
(m/z values) for each peptide, thus not relying only on those
fragmented during the LC-MS/MS run and identified during
the subsequent search. This will yield more data points which
can be either merged and statistically processed or used sepa-
rately as redundant information supporting the data validity.
32. DeutEx will accept any text form of the input and has auto-
mated detection of the content in the columns. If that fails, the
user can correct it and assign the column headers manually.
33. For better clarity, a short video demonstrating the whole work-
flow is available at http://peterslab.org/downloads/SW/
DeutEx.mp4.
34. In Data Analysis, which is delivered together with MS instru-
ments from Bruker Daltonics, this procedure can be partially
automated using a Visual Basic macro. This macro can be
obtained from the authors upon request.
 HDX-MS of Protein-DNA Complexes 217

35. mMass is a freely available software suitable for processing of

single spectra. It can be obtained from http://mmass.org/.
36. There are also other ways of data processing. Similar to mMass
processing is the use of the MagTran program written by
Zhongqi Zhang. Also, HX-Express2 (http://www.hxms.
com/HXExpress/) can be used for this purpose [32]. The
advantage of HX-Express 2 is its ability to detect and dissect
bimodal isotopic envelopes that are typical for mixed
EX1/EX2 exchange kinetics.
37. Script can be obtained at http://peterslab.org/downloads.php
under HDX tools section.
38. To run Python scripts directly, install python on your PC
(https://www.python.org/downloads/) and include it in the
PATH variable of the operating system (version used here was
2.7.12).
39. Script was written by Robert L. Campbell and is available here:
http://pldserver1.biochem.queensu.ca/~rlc/work/pymol/.

Acknowledgments

Czech Science Foundation projects 16-24309S and 16-20860S are

gratefully acknowledged. Additional support was obtained from
EU/MEYS projects BioCeV (CZ.1.05/1.1.00/02.0109) and
NPU II (LQ1604). R.F. also thanks Charles University Grant
Agency (project 1618218) and SVV260427/2019.

References

1. Katta V, Chait BT, Carr S (1991) Conforma- of solution. Die Makromol Chemie
tional changes in proteins probed by hydrogen- 187:1053–1063
exchange electrospray-ionization mass spec- 7. Ma L, Fitzgerald MC (2003) A new H/D
trometry. Rapid Commun Mass Spectrom exchange- and mass spectrometry-based
5:214–217 method for thermodynamic analysis of
2. Zhang Z, Smith DL (1993) Determination of protein-DNA interactions. Chem Biol
amide hydrogen exchange by mass spectrome- 10:1205–1213
try: a new tool for protein structure elucida- 8. Sperry JB, Wilcox JM, Gross ML (2008)
tion. Protein Sci 2:522–531 Strong anion exchange for studying protein-
3. Bai YW, Milne JS, Mayne L et al (1993) Pri- DNA interactions by H/D exchange mass
mary structure effects on peptide group spectrometry. J Am Soc Mass Spectrom
hydrogen-exchange. Proteins 17:75–86 19:887–890
4. Engen JR, Wales TE (2015) Analytical aspects 9. Sperry JB, Shi X, Rempel DL et al (2008) A
of hydrogen exchange mass spectrometry. mass spectrometric approach to the study of
Annu Rev Anal Chem 8:127–148 DNA-binding proteins: interaction of human
5. Oganesyan I, Lento C, Wilson DJ (2018) Con- TRF2 with telomeric DNA. Biochemistry
temporary hydrogen deuterium exchange mass 47:1797–1807
spectrometry. Methods 144:27–42 10. Poliakov A, Jardine P, Prevelige PE (2008)
6. Sorokin VA, Gladchenko GO, Valeev VA Hydrogen/deuterium exchange on protein
(1986) DNA protonation at low ionic strength solutions containing nucleic acids: utility of
218 Ruzena Filandrova et al.

protamine sulfate. Rapid Commun Mass Spec- resolution by proteolytic fragmentation mass
trom 22:2423–2428 spectrometry analysis. Proc Natl Acad Sci U S
11. Roberts VA, Pique ME, Hsu S et al (2012) A 110:16438–16443
Combining HD exchange mass spectroscopy 23. Cravello L, Lascoux D, Forest E (2003) Use of
and computational docking reveals extended different proteases working in acidic conditions
DNA-binding surface on uracil-DNA glycosy- to improve sequence coverage and resolution
lase. Nucleic Acids Res 40:6070–6081 in hydrogen/deuterium exchange of large pro-
12. Graham BW, Tao Y, Dodge KL et al (2016) teins. Rapid Commun Mass Spectrom
DNA interactions probed by hydrogen- 17:2387–2393
deuterium exchange (HDX) Fourier transform 24. Rey M, Man P, Brandolin G et al (2009)
ion cyclotron resonance mass spectrometry Recombinant immobilized rhizopuspepsin as
confirm external binding sites on the minichro- a new tool for protein digestion in hydrogen/
mosomal maintenance (MCM) helicase. J Biol deuterium exchange mass spectrometry. Rapid
Chem 291:12467–12480 Commun Mass Spectrom 23:3431–3438
13. Boura E, Silhan J, Herman P et al (2007) Both 25. Kadek A, Tretyachenko V, Mrazek H et al
the N-terminal loop and wing W2 of the fork- (2014) Expression and characterization of
head domain of transcription factor Foxo4 are plant aspartic protease nepenthesin-1 from
important for DNA binding. J Biol Chem Nepenthes gracilis. Protein Expr Purif
282:8265–8275 95:121–128
14. Slavata L, Chmelik J, Kavan D et al (2019) 26. Yang M, Hoeppner M, Rey M et al (2015)
MS-based approaches enable the structural Recombinant Nepenthesin II for hydrogen/
characterization of transcription factor/DNA deuterium exchange mass spectrometry. Anal
response element complex. Biomol Ther 9: Chem 87:6681–6687
E535 27. Kadek A, Kavan D, Marcoux J et al (2017)
15. Anbanandam A, Albarado DC, Nguyen CT Interdomain electron transfer in cellobiose
et al (2006) Insights into transcription dehydrogenase is governed by surface electro-
enhancer factor 1 (TEF-1) activity from the statics. Biochim Biophys Acta Gen Subj
solution structure of the TEA domain. Proc 1861:157–167
Natl Acad Sci U S A 103:17225–17230 28. Moroco JA, Engen JR (2015) Replication in
16. Wang L, Pan H, Smith DL (2002) Hydrogen bioanalytical studies with HDX MS: aim as
exchange-mass spectrometry. Mol Cell Proteo- high as possible. Bioanalysis 7:1065–1067
mics 1:132–138 29. Houde D, Berkowitz SA, Engen JR (2011)
17. Kadek A, Mrazek H, Halada P et al (2014) The utility of hydrogen/deuterium exchange
Aspartic protease nepenthesin-1 as a tool for mass spectrometry in biopharmaceutical com-
digestion in hydrogen/deuterium exchange parability studies. J Pharm Sci 100:2071–2086
mass spectrometry. Anal Chem 86:4287–4294 30. Wales TE, Poe JA, Emert-Sedlak L et al (2016)
18. Kochert BA, Iacob RE, Wales TE et al (2018) Hydrogen exchange mass spectrometry of
Hydrogen-deuterium exchange mass spec- related proteins with divergent sequences: a
trometry to study protein complexes. In: comparative study of HIV-1 Nef allelic var-
Methods in molecular biology (Clifton, N.J.). iants. J Am Soc Mass Spectrom 27:1048–1061
Humana Press, New York, NY, pp 153–171 31. Pascal BD, Willis S, Lauer JL et al (2012)
19. Rand KD, Zehl M, Jensen ON et al (2009) HDXWorkbench: software for the analysis of
Protein hydrogen exchange measured at H/D exchange MS data. J Am Soc Mass Spec-
single-residue resolution by electron transfer trom 23:1512–1521
dissociation mass spectrometry. Anal Chem 32. Guttman M, Weis DD, Engen JR et al (2013)
81:5577–5584 Analysis of overlapped and noisy hydrogen/
20. Mistarz UH, Bellina B, Jensen PF et al (2018) deuterium exchange mass spectra. J Am Soc
UV Photodissociation mass spectrometry accu- Mass Spectrom 24:1906–1912
rately localize sites of backbone Deuteration in 33. Lindner R, Lou X, Reinstein J et al (2014)
peptides. Anal Chem 90:1077–1080 Hexicon 2: automated processing of
21. Mayne L, Kan ZY, Sevugan Chetty P et al hydrogen-deuterium exchange mass spectrom-
(2011) Many overlapping peptides for protein etry data with improved deuteration distribu-
hydrogen exchange experiments by the frag- tion estimation. J Am Soc Mass Spectrom
ment separation-mass spectrometry method. J 25:1018–1028
Am Soc Mass Spectrom 22:1898–1905 34. Rey M, Sarpe V, Burns KM et al (2014) Mass
22. Kan Z-Y, Walters BT, Mayne L et al (2013) spec studio for integrative structural biology.
Protein hydrogen exchange at residue Structure 22:1538–1548
 HDX-MS of Protein-DNA Complexes 219

35. Kan ZY, Ye X, Skinner JJ et al (2019) ExMS2: 40. Majumdar R, Manikwar P, Hickey JM et al
an integrated solution for hydrogen-deuterium (2012) Minimizing carry-over in an online
exchange mass spectrometry data analysis. Anal pepsin digestion system used for the H/D
Chem 91:7474–7481 exchange mass spectrometric analysis of an
36. Claesen J, Burzykowski T (2017) Computa- IgG1 monoclonal antibody. J Am Soc Mass
tional methods and challenges in hydrogen/ Spectrom 23:2140–2148
deuterium exchange mass spectrometry. Mass 41. Rey M, Mrazek H, Pompach P et al (2010)
Spectrom Rev 36:649–667 Effective removal of nonionic detergents in
37. Eggertson MJ, Fadgen K, Engen JR et al protein mass spectrometry, hydrogen/deute-
(2020) Considerations in the analysis of hydro- rium exchange, and proteomics. Anal Chem
gen exchange mass spectrometry data. Meth- 82:5107–5116
ods Mol Biol 2051:407–435 42. Glasoe PK, Long FA (1960) Use of glass elec-
38. Kavan D, Man P (2011) MSTools - web based trodes to measure acidities in deuterium oxide
application for visualization and presentation 1,2. J Phys Chem 64:188–190
of HXMS data. Int J Mass Spectrom 43. Guttman M, Wales TE, Whittington D et al
302:53–58 (2016) Tuning a high transmission ion guide
39. Strohalm M, Kavan D, Novak P et al (2010) to prevent gas-phase proton exchange during
mMass 3: a cross-platform software environ- H/D exchange MS analysis. J Am Soc Mass
ment for precise analysis of mass spectrometric Spectrom 27:662–668
data. Anal Chem 82:4648–4651
 Chapter 12

Integrative Mass Spectrometry–Based Approaches

for Modeling Macromolecular Assemblies
Andy M. Lau and Argyris Politis

Abstract
Mass spectrometry (MS)–based strategies have emerged as key elements for structural modeling of proteins
and their assemblies. In particular, merging together complementary MS tools, through the so-called
hybrid approaches, has enabled structural characterization of proteins in their near-native states. Here,
we describe how different MS techniques, such as native MS, chemical cross-linking MS, and ion mobility
MS, are brought together using sophisticated computational algorithms and modeling restraints. We
demonstrate the applicability of the strategy by building accurate models of multimeric protein assemblies.
These strategies can practically be applied to any protein complex of interest and be readily integrated with
other structural approaches such as electron density maps from cryo-electron microscopy.

Key words Structural mass spectrometry, Computational modeling, Protein complexes, Hybrid
approaches, Modeling restraints

1 Introduction

Integrative structural methods has come of age in building 3D

model ensembles of proteins and their complexes [1]. These meth-
ods allow bringing together data sets sourced from diverse struc-
tural methods that otherwise will be reported independently
[2]. Such integration can be achieved through the use of modeling
restraints obtained directly from the experimental data [1, 2]
(Fig. 1). Recent innovative studies have led to the expansion of the
integrative structural toolkit to incorporate a variety of MS-based
methods. These include powerful methods such as chemical cross-
linking [3–6], ion mobility [7, 8], and native MS [9]. Moreover,
these methods can be integrated with other powerful methods such
as X-ray crystallography [10], cryo-EM [11], and hydrogen–deute-
rium exchange (HDX)-MS [11] structural approaches. Impressive
examples of the integrative structural toolkit include the elucidation
of many structural aspects of large complexes such as the protea-
some [5, 6], ATP synthase [12], and nuclear pore complex [13].

Arnaud Poterszman (ed.), Multiprotein Complexes: Methods and Protocols, Methods in Molecular Biology, vol. 2247,
https://doi.org/10.1007/978-1-0716-1126-5_12, © Springer Science+Business Media, LLC, part of Springer Nature 2021

221
222 Andy M. Lau and Argyris Politis

Fig. 1 Integrative modeling with mass spectrometry. Integration of restraints derived from MS-based and cryo-
EM data sets lead to computational structural models that can be directly compared with available crystal
structures. Such comparisons allow for assessing the ability of the methods used to make accurate
predictions

1.1 Native MS Native MS is a powerful technique for interrogating key structural

aspects of proteins and their complexes. It allows examination of
the specificity and selectivity of ligands binding to a protein in
addition to identifying protein stoichiometry and their oligomeric
states [14–16]. The main strengths of native MS are that it can
rapidly provide information on the native state of large proteins and
their complexes, it does not require high concentrations of pure
material, and it can typically deal with heterogeneous samples. As
such, native MS has become an integral component of the struc-
tural biology toolkit complementing traditional structural meth-
ods. Information derived from native MS can be used to model
protein complexes [9] and also integrated with data sets from other
related methods [1, 2].

1.2 Ion Mobility MS Native MS is often hyphenated with ion mobility (IM)-MS. IM-MS
provides information about protein topologies and dynamics in the
gas phase [17, 18]. It is expressed using the collisional cross section
(CCS) which reflects the orientationally average cross section of
proteins as they tumble through the IM cell in the mass spectrom-
eter [19]. CCS from IM-MS is obtained by measuring the arrival
time distributions (ATDs) of different ions [20] and can in turn be
used as shape restraints for interrogating structural models gener-
ated by computational methods [7, 21, 22]. Structural dynamics
and conformational heterogeneity can be inferred from the distri-
bution of mobilities measured by IM-MS [23–25]. Typically, ions
with multiple conformations will result in broad, Gaussian
ATDs [12].

1.3 Chemical Chemical cross-linking (CX)-MS allows covalent linkage of reactive

Cross-Linking amino acid residues enabling information about the protein topol-
and Covalent ogy and their fold [6, 26–29]. CX-MS defines protein subunit
Labelling MS proximities which can be expressed as distance restraints for down-
stream modeling analysis [27, 30]. The information can
 Mass Spectrometry-Based Protein Modelling 223

subsequently be translated into structural restraints, which are then

used by a modeling workflow to generate 3D models of protein
assemblies. Moreover, several computational methods and tools
have been developed to assist the structural interpretation of
CX-MS [4, 6, 31]. Another MS-based labeling strategy that can
be used for modeling protein complexes includes covalent labeling
(CL)-MS. Covalent labeling is a chemical modification technique
that allows the mapping of residue-level surface accessibility. It
relies on the notion that exposed amino acid residues accessible to
the labeling reagents will readily react, while protected (buried)
residues respond to a lesser extent [32]. These differences in reac-
tivity can be associated with conformational changes of proteins
and therefore be used to infer information regarding protein–pro-
tein or protein–ligand interactions [33].

1.4 Hydrogen– HDX-MS has recently gained interest for studying protein confor-
Deuterium mational dynamics and their associations with ligands [15, 34–
Exchange MS 36]. HDX-MS can report on change in the structure and dynamics
of proteins, through monitoring the exchange of protein backbone
amide hydrogens for deuterium in the surrounding labeling buffer.
The exchange rate is dependent on both hydrogen bonding and
solvent accessibility [37–39]. HDX-MS tolerates a broad range of
sample environments and can be used to capture conformational
changes of proteins at peptide-level resolution. Data from two
protein states (e.g., apo and holo) can be compared in a differential
manner, allowing relative changes in protein structure and dynam-
ics to be determined. The resulting significant differences (ΔHDX)
between the two states can in turn be mapped onto 3D models of
proteins to inform on conformational changes [37] (Fig. 1).
Recently HDX-MS has been applied to a range of membrane
proteins and their interactions with lipids [36, 40, 41]. Moreover,
attempts to model data derived from HDX-MS have utilized pro-
tection factors, which can be used as restraints to build model
structures directly from ΔHDX [42].

1.5 Integrating While MS-based approaches offer a wide range of techniques capa-
with Other Structural ble of generating diverse structural information, their integration
Approaches with other powerful structural methods is often desirable. The
most commonly used methods to integrate MS data are X-ray
crystallography and cryo-EM. Traditionally, crystallographic infor-
mation has been the gold standard to either validate the ability of a
computational strategy to build accurate models (benchmarking)
or be used in conjunction with other methods to refine its confor-
mation. The recent resolution revolution of cryo-EM has yield a
game-changing effect in structural biology through its ability to
provide atomic-level resolution of proteins previously thought to
be insuperable (such as dynamic and membrane proteins), thus
significantly expanding the potential number of biological
224 Andy M. Lau and Argyris Politis

structures that can be solved to high resolution. It is therefore

timely to develop methods that can integrate the data-rich
MS-based approaches with cryo-EM.

2 Materials

The following software packages need to be downloaded and

installed:
1. PyMOL (https://pymol.org/2/).
2. Chimera (https://www.cgl.ucsf.edu/chimera/).
3. Integrative Modeling Platform (https://integrativemodeling.
org/).
4. IMPACT (http://impact.chem.ox.ac.uk/).
Webservers accessed as part of this protocol:
5. UniProt (https://www.uniprot.org/).
6. BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi).
7. T-Coffee (http://tcoffee.crg.cat/).
8. RCSB (http://www.rcsb.org/).
9. PDBe (http://www.ebi.ac.uk/pdbe/node/1).
10. EMDB (https://www.ebi.ac.uk/pdbe/emdb/).
All tutorial files can be downloaded from:
11. Github (https://github.com/andymlau/MIMB_Integrative_
Modelling).

3 Methods

All instructions below assume that tutorial files have been down-
loaded and that the required software packages have been installed
and tested for correct functionality.

3.1 Preparing Inputs 1. The IMP platform can be used to generate models of protein
for IMP complexes, informed by experimental techniques which each
contribute modeling restraints. To use stoichiometry, cross-
linking, and cryo-EM restraints, the following files are needed:
Files
initial_model.pdb—contains EM-map aligned model of the
protein complex.
sequences.fasta—contains full sequences of each subunit.
stoichiometry.txt—list of stoichiometric connectivities.
crosslinks.txt—list of experimental cross-links.
 Mass Spectrometry-Based Protein Modelling 225

initial_map.gmm.nX.txt—Gaussian mixture model (GMM)

representation of the electron density map.
topology.txt—contains topological parameters for initializing
each subunit for modeling.
Scripts
modelling.py—script for calling IMP.
calculate_density.sh—bash script for converting density maps
to modeling-compatible maps.
create_gmm.py—IMP script for generating the above maps.
2. The following steps will walk through how to prepare these
files.

3.2 Sourcing 1. Subunits of the complex may be available as crystal or NMR

Structures structures; these may be found at repositories such as RCSB
of Constituent and PDBe. A crystal structure of the tryptophan synthase
Subunits complex has been deposited under PDB 1WBJ (https://www.
rcsb.org/structure/1WBJ). Using PyMOL, enter the “fetch
1WBJ” command into the PyMOL command line to automat-
ically download the file from the RCSB database (see Note 1).
2. PyMOL will display two subunits: alpha (chain A) and beta
(chain B). Display the sequence information by pressing the
“S” button displayed on the bottom right hand side of the
Movies Controls or by entering “set seq_view, on” into the
PyMOL command line.
3. The tryptophan synthase dimer will be coated with ligands and
water molecules which will need to be removed prior to mod-
eling. Type “remove hetatm” into the PyMOL command line to
remove all non-protein atoms.
4. Optionally, type “util.cbc” to enable per-chain coloration to aid
visualization of the subunits. The alpha subunit (chain A) will
be colored green, and beta subunit (chain B) in blue by default.
A screenshot of your PyMOL GUI at this point is shown in
Fig. 2a.
5. With the sequence view on, check the following:
Sequence numbering—check the sequence numbering to
verify that the numbers are correct. Additionally, locate on the
structure the residues involved in cross-linking and verify that
their residue numbers are correct, e.g., check that residue
LYS46 on the cross-link list correctly points to LYS46 on the
structure (see Note 2).
Residue type—only certain residue types will be “visible”
in the experiment depending on the type of cross-linker used,
e.g., BS3 (bis(sulfosuccinimidyl)suberate) is an amine-to-
amine cross-linker only targets lysine and N-terminal amine
226 Andy M. Lau and Argyris Politis

Fig. 2 Visualization of protein structures using PyMOL. (a) Ribbon representation of the PDB 1WBJ tryptophan
synthase. (b) Display of atomic bonds for a selected residue

groups. Check that all cross-linked residues on the model

structure are the correct residue type.
Missing atoms—residue sidechains may be missing atoms
for various reasons, such as poor electron density from protein
crystallography. While checking the residue numbering and
type, also check if any of these sidechains are missing atoms
(see Note 3). By default, proteins opened in PyMOL version
2 and above will be represented in cartoon format. To display
full atomic information, you will need to select the residue of
 Mass Spectrometry-Based Protein Modelling 227

interest, and display it in line or stick format. To do this, select

the residue by clicking on it in the sequence viewer, click on “S”
in the Object Control Panel for “(sele)” and press either “lines”
or “sticks” (Fig. 2b).
6. Next, you will need to identify any missing residues in each of
the subunits. This can be done by aligning the sequence of each
subunit to their canonical sequences (from UniProt) using a
multiple sequence alignment program such as T-Coffee (see
Notes 4 and 5).
7. First, save the 1WBJ PDB file in the FASTA format in PyMOL
by entering both “save initial_model_A.fasta, chain A” and
“save initial_model_B.fasta, chain B” into the PyMOL com-
mand line. This will save the sequences of both chain A (alpha
subunit) and chain B (beta subunit) separately as FASTA files.
8. Access the UniProt entry of the tryptophan synthase alpha
subunit (Accession Number (AN): P00929) and beta subunit
(AN: P0A2K1) at: https://www.uniprot.org/uniprot/
P00929 and https://www.uniprot.org/uniprot/P0A2K1,
respectively.
9. The sequence of each subunit is listed under the “Sequence”
tab of the side bar. Click on the “FASTA” download button to
open the sequence in the FASTA format.
10. Next, we will align the model and canonical sequences using
the T-Coffee webserver. Copy and paste the model and canon-
ical sequences of the alpha chain into T-Coffee and submit.
Repeat for the beta chain. Sample results are shown in Fig. 3a.
11. Regions with differences between theoretical and modeled
sequences will be identified by sequence alignment, including
missing residues (Fig. 3a).
12. If there are many regions of the model with missing coordi-
nates, homology modeling may need to be employed to regen-
erate these missing residues (see Note 6).
13. Finally, combine the “initial_model_A.fasta” and “initial_mo-
del_B.fasta” sequences into a single “initial_model.fasta.”
Since there are two copies of alpha and beta subunits in the
complex, duplicate both the alpha and beta subunit sequences,
and title each subunit as “>TS_alpha_A,” “>TS_beta_B,”
“>TS_alpha_C,” and “>TS_beta_D” (Fig. 3b).

3.3 Determining 1. The native MS spectra of the tryptophan synthase complex can
Complex be used to determine any subcomplexes that build up the full
Stoichiometry Using complex, and the mass of these subcomplexes (Fig. 4).
Native MS 2. A native MS spectra of the complex can be found at [33].
3. Use the subcomplexes to determine the known connections
between the constituent subunits (see Notes 7 and 8). The
228 Andy M. Lau and Argyris Politis

Fig. 3 Sequence analysis. (a) T-Coffee sequence alignment of canonical (AN: P0A2K1) and model sequence
(TS_A_A) of tryptophan synthase beta chain. The second and third residue of TS_A_A (PDB 1WBJ) has been
altered to demonstrate the conservation notation (cons). Asterisk (*) and colon (:) denote full and similar
property conservation, respectively. Only the first line of the sequence has been shown. (b) Alpha and beta
subunit sequence in fasta format
 Mass Spectrometry-Based Protein Modelling 229

Fig. 4 Native MS of tryptophan synthase. (a) Native mass spectra showing four ion distributions of 144, 115,
86, and 29 kDa. (b) Annotated spectra of (a) with assigned subcomplexes

simplest non-monomeric subcomplex has been identified to be

a beta subunit dimer. This indicates a beta–beta connection in
the complex. A single alpha subunit associates with the beta
dimer, forming an alpha–beta–beta subcomplex. This indicates
an alpha–beta connection. As there are no alpha dimers seen,
we can assume that these do not exist. The connectivity of the
complex can therefore be broken down to alpha–beta and beta–
beta subunits.

3.4 Preparation 1. A list of cross-links of the tryptophan synthase complex can be

of Cross-Link List found at [33]. To demonstrate a heterogeneous setup of
for IMP LYS-LYS and ASP/GLU-ASP/GLY cross-links, we have
simulated a set of cross-links for this protocol.
2. The cross-links to be used in the modeling have been prepared
as “crosslinks.txt” and are in the format shown in Fig. 5, with
each cross-link entry listed on a separate line, and each protein
and residue separated by commas. The “prot1,res1,prot2,res2”
header at the top of the file tells the IMP parser how to
interpret the entries.
3. Since there are redundant copies of each alpha and beta subunit
in the experiment, the cross-links must be duplicated for each
alpha–beta and beta–beta pair as determined by native MS
(Fig. 5).

3.5 Preparation 1. Having determined the subunit connections of the complex via
of Stoichiometry List native MS, we will need to convert these into modelable
for IMP restraints.
2. In the same format as the “crosslinks.txt” file, we can include a
pseudo-cross-link between the alpha–beta and beta–beta sub-
units of the complex. As there are two alpha–beta pairs, we will
enforce a pseudo-cross-link between TS_alpha_A-TS_beta_B
and TS_alpha_C-TS_beta_D and the beta–beta connection as
230 Andy M. Lau and Argyris Politis

Fig. 5 Simulated cross-links of the tryptophan synthase complex. Cross-links in

black (3–5 and 12–14) have been duplicated (red) to reflect the twofold
symmetry nature of the full complex. Not all cross-links are shown

Fig. 6 Stoichiometry of the tryptophan synthase complex. Pseudo-crosslinks

between subunits of the complex

TS_beta_B-TS_beta_D. For the residue numbers, provide the

center-most residue identified from the 1WBJ PDB file. The
center-most residues are residue 49 for alpha subunits and
residue 235 for beta subunits. The content of the pseudo-
cross-link file is shown in Fig. 6.
3. The cryo-EM density maps of protein complexes are stored and
accessed from the EMDB database.
 Mass Spectrometry-Based Protein Modelling 231

4. Since no density maps are available for the tryptophan synthase

complex, a simulated map at 14 Å “TS_1WBJ_14A.mrc” has
been generated and included in the protocol files.
5. All subunits will need to be fitted into the map of the density
for the map to restrain the subunit positions during sampling.
For this protocol, the structure has been pre-fitted into the mrc
map (see Notes 9 and 10).
6. The mrc density map will need to be converted into a
modeling-compatible Gaussian Mixture Model (GMM)
format.
7. To do this, have the “initial_map.mrc,” “calculate_density.sh”
and “create_gmm.py” files in the same directory, and open a
terminal window, e.g., Terminal in MacOS/Linux or Bash for
Windows.
8. Enter “bash create_gmm.py initial_map.mrc 6” to automati-
cally begin converting the simulated mrc map into a GMM map
(see Notes 11 and 12). The number at the end refers to the
number of components that the map will possess—the higher
the value, the finer the resolution of the generated map model.
The files “initial_map.gmm.n6.mrc” and “initial_map.gmm.
n6.txt” will be generated. Optionally open the new and origi-
nal maps to visualize the conversion.

3.6 Preparing 1. Open the topology.txt file which has already been filled
Modeling Script (Fig. 7a).
for IMP 2. The topology file contains a “topology_dictionary” for each
subunit. Table 1 explains the function of each heading.
3. In the previous directory, there will be a script called “model-
ling.py” which will call IMP and perform the modeling.
4. “modelling.py” is a python script that will instruct IMP on how
to initialize each subunit and perform random sampling while
restraining their positions and orientations using the stoichi-
ometry, cross-link, and EM restraints.
5. Within “modelling.py,” lines 25–27 under the “Define Input
Files” section will link IMP to the necessary “topology.txt” and
“initial_map.gmm.n6.txt” files which are located in the “./
inputs/” directory (Fig. 7b).
6. The next lines “num_frames” and “num_mc_steps” will set the
number of iterations that the sampling will be performed for
and the number of Monte Carlo steps that are included in one
iteration.
7. “rb_max_trans” and “rb_max_rot” are variables for controlling
the step size of each translation and rotation of each rigid body,
i.e., each subunit.
232 Andy M. Lau and Argyris Politis

Fig. 7 Preparing the IMP modeling script. (a) Content of the topology file topology.txt. (b) Definition of input
files. (c) Examples of rigid body groups

Table 1
Function of headings from the topology file

Heading Function
component_name Name of subunit
domain_name Subunit identifier for modelling
fasta_fn Fasta filename
fasta_id References the subunit sequence listed in specified file
pdb_fn References the PDB file containing all structures
Chain References the chain belonging to the subunit in the PDB specified
residue_range Obtains the atomic position of the residues within range from the PDB
pdb_offset Offsets the PDB by the number specified
bead_size Number of residues to represent using a single bead for coarse-graining
em_residues_per_gaussian Number of components for GMM model generation
 Mass Spectrometry-Based Protein Modelling 233

8. If there are missing residues in the initial subunit, i.e., inherited

from crystal or NMR structures, IMP will replace these missing
residues using flexible beads. The movement of these beads are
controlled by the “bead_max_trans” variable (see Note 13).
9. A list of subunits will need to be given to IMP via a “rigid_bo-
dies” list. In the current format, each subunit will be permitted
translational and rotational movement independently from the
other subunits. Several examples of rigid body groups that can
be set are shown in Fig. 7c.
10. Next, lines 57–102 are needed to set the coarse-grained repre-
sentation of the subunits. Each subunit listed in “topology.txt”
will be replaced by a bead model which will take into account
the “bead_size” parameter as its resolution.
11. Each subunit is automatically given a color for loop on line 74.
12. The ability of each subunit and flexible bead to move is set up
by lines 93–102.
13. Finally, the restraint section “Define Scoring Function Com-
ponents” defines the variables necessary for restraint-guided
modeling.

3.7 Modeling 1. The first restraint is the excluded volume restraint—this pre-
Restraints vents unrealistic overlapping in space through calculating the
volume of each subunit per iteration and avoiding steps which
promote overlap.
2. The excluded volume restraint is computationally expensive to
calculate, and thus, the volume is approximated by bead reso-
lution (by default set to 20) on line 115.
3. Lines 117 and 118 add the restraint to the bead model.
4. Next, we will add two cross-link restraints to our tryptophan
synthase system, xl1 and xl2.
5. xl1 is the stoichiometry restraint which will enforce alpha–
beta–beta–alpha connectivity in our modeling. The xl1
restraint can be adjusted by variables listed in Table 2.
6. The “length” variable passed into xl1 is defined by the approxi-
mate distance between the alpha and beta subunits as measured
from their centers (approximately equal to the sum radius of
alpha and beta subunits). This distance is approximate and
serves to enforce the topology of the system by preventing
the subunits from moving too far away from their designated
binding partners.
7. The xl2 restraint includes cross-links determined for the tryp-
tophan synthase complex. Lines 146–158 are parametrized
identically to xl1; however, the distance threshold “length,”
“slope,” and “label” has been altered.
234 Andy M. Lau and Argyris Politis

Table 2
Key parameters that can be used for the adjustment of modeling restraints

Heading Function
Length Distance threshold to enforce per iteration, set to 38 Å
Slope Linear gradient for preventing the system from expanding too much
Columnmapping Designates the format of the stoichiometry.Txt file
Resolution Sets the bead resolution that the cross-link distance will be measured using
Label Label for modeling output
Csvfile Toggle for stoichiometry.Txt file format to be read as a comma separated values file

8. xl2 is set up with a distance threshold of 35 Å. This distance is

derived from the BS3 cross-linker molecule length (11 Å), two
lysine side chains (2
7 Å) and an additional 10 Å which
permits small-scale subunit translations and rotations and pre-
vents the system from being locked into a particular
conformation.
9. The EM restraint is set by lines 171–178. In the first step, a
GMM map is calculated for each subunit. This map will be
calculated against the GMM of the original complex map. The
EM restraint also has a “slope” variable which will persuade
movement of subunits to remain inside the map volume.
10. Finally, the mc1 variable will run the sampling of the subunit
bead models which have restraints added. Mc1 is an IMP
replica exchange macro which will additionally perform replica
exchange steps on flexible regions of the bead model. These
parameters will only affect flexible beads (see Note 14).
11. “crosslink_restraints” in mc1 adds the two xl1 and xl2
restraints to the replica exchange sampler.
12. A Bayesian scoring function is used to calculate the score of
each model per iteration (see Note 15). “number_of_best_s-
coring_models” controls how many output models are gener-
ated. These are saved within the “global_output_directory”
directory. The top scoring models are saved in the /output/
pdbs folder by default. Model names are updated and ranked
such that model.0.pdb has the highest score of the ensemble.
13. The per-iteration trajectory of the sampling simulation is saved
as /output/rmfs/0.rmf3. This file can be opened and played
using Chimera.
14. The PDB models generated by IMP only contain Ca atoms.
Since there are no missing residues (that have been modeled as
flexible beads), use PyMOL to align the alpha and beta sub-
units to model.0.pdb (see Note 16). You can then save this
atomistic model.
 Mass Spectrometry-Based Protein Modelling 235

15. A PyMOL script has been provided to streamline the conver-

sion between coarse-grained and atomistic models, called
“CG2pdb.py” in the same directory as “modelling.py.”
16. Open a terminal window and direct to the directory of
“CG2pdb.py.” Run the script using “<path to pymol binary-
>-c CG2pdb.py.” The <path to pymol binary> will need to be
changed to the path of PyMOL on your system. The “-c” flag
will suppress the GUI of PyMOL.
17. The directory “output_atomistic” will be made and will con-
tain all 100 best scoring IMP models in atomistic representa-
tion. By default, the CG2pdb.py script will convert from
model.0.pdb to model.99.pdb (100 models). You may alter
the model range within the script by changing “n_first” and
“n_last” variables.
18. At this point, you will have a folder called “output_atomistic”
containing 100 atomistic models of tryptophan synthase that
have been generated using modeling restraints from MS
and EM.

3.8 Calculating 1. The theoretical collision cross section (CCS) of atomic models
the CCS of Generated can be calculated using software such as IMPACT [43] (see
Models Notes 17–19).
2. This value can be compared to experimentally determined CCS
via IM-MS [44].
3. Alternatively, CCS values for complexes may be found in pub-
lished literature [7].
4. The CCS of the tryptophan synthase complex is approximately
7200 Å2.
5. We will use IMPACT to calculate the projection approximation
(PA) of each model.
6. Open a terminal window and direct to the “output_atomistic”
folder.
7. With IMPACT callable from your PATH variable, run “impact
*.pdb -o output_file_CCS.txt.” *.pdb will tell IMPACT to run
through file with the .pdb file extension.
8. When IMPACT has finished, output_file_CCS.txt will display a
list of each file, with the calculated PA and TJM values (see
Note 20).
9. Linear scaling of the calculated PA by an empirical factor of
1.14 has been shown to provide a good approximation of the
true model CCS [45].
10. “output_file_CCS.txt” can be opened using spreadsheet soft-
ware, and a scaling factor of 1.14 can be applied to each value
under “CCS_PA.”
236 Andy M. Lau and Argyris Politis

11. Calculate the CCS % error between the model (CCSmodel)

and experimental CCS (CCSexp) for all tryptophan synthase
models using:

CCSmodel  CCSexp
% error ¼
100
CCSexp
12. The best-fit models will show the lowest % error. This confor-
mation will have been generated using stoichiometry, cross-
linking, and EM restraints and have been filtered using CCS.
13. Finally, evaluate the accuracy of your model against the crystal
structure of the full tetrameric tryptophan synthase complex.
Open PyMOL and use the “fetch 1WBJ, type¼pdb1, multi-
plex¼1” command to automatically download the “biological
assembly” of 1WBJ.
14. 1WBJ will download as two objects in PyMOL, 1WBJ_0001
and 1WBJ_0002 which will each be half of the complex. Both
alpha chains will be labeled as Chain A, and both beta as
Chain B.
15. To align our best-fit model to this complex, we will first need to
combine the two halves of 1WBJ into a single object, with the
same Chain ABCD arrangement.
16. Use the command “alter 1WBJ_0002 & chain A, chain¼‘C’”
and “alter 1WBJ_0002 & chain B, chain¼‘D’” to rename to
chains C and D.
17. Next, use “create 1WBJ_assembled, 1WBJ_*” to create a new
object with all chains of the tryptophan synthase biological
assembly.
18. Align your best fit model to this 1WBJ_assembled object using
either “align <best_model_name> 1WBJ_assembled” or by
clicking on screen in the PyMOL GUI.
19. In the External GUI, the root mean squared deviation
(RMSD) of the fitting will be displayed (Fig. 8). The RMSD
represents the similarity score between the two models. The
lower the RMSD, the more similar the models.

Fig. 8 Scoring RMSD between generated model and crystal structure

 Mass Spectrometry-Based Protein Modelling 237

20. The RMSD value can be improved by either improving the

quality of the restraints (i.e., additional experimental informa-
tion), or by fine tuning the sampling parameters. The number
of iterations can also be increased to maximize the search space
of the program.

4 Notes

1. A comprehensive PyMOL tutorial can be found at: https://

www.mrc-lmb.cam.ac.uk/rlw/text/MacPyMOL_tutorial.
html.
2. The sequence of the protein can be edited using PyMOL
through the command:
alter <selection>, resv += X

where <selection> should be modified to reflect the object

or chain to be altered, and the plus sign should be swapped to
minus if X is to be deducted from the sequence number. When
using this command, <selection> can refer to any selection,
e.g. the object name, segment ID, chain ID or the currently
highlighted selection. For more information, consult the
PyMOL documentation for the Alter command (https://
pymolwiki.org/index.php?title¼Alter&redirect¼no).
3. For cross-linking modeling, residues will need to have the atom
that will be involved in distance measurements. For example,
LYS-LYS cross-links can be measured from Cα, Cβ, or Nζ
atoms. If using the sidechain amide Nζ atom, you will need
to have this atom included in all sidechain models. The Muta-
genesis tool of PyMOL can be used to quickly regenerate all
sidechain atoms, swap between common sidechain rotamers or
switch residue types (https://pymolwiki.org/index.php/
Mutagenesis).
4. FASTA is a text-based format for displaying sequence informa-
tion of proteins or nucleotides (see also https://zhanglab.ccmb.
med.umich.edu/FASTA/). In PyMOL, you can save the
sequence information in FASTA format for any object or selec-
tion using the “save” command (https://pymolwiki.org/
index.php/Save).
5. T-Coffee is a quick and easy-to-use multiple sequence align-
ment tool [46]. A tutorial is available on the webserver website
at: http://tcoffee.crg.cat/apps/tcoffee/tutorial.html.
6. Homology modeling can be performed using the MODEL-
LER software [47]. A detailed tutorial can be found at: https://
salilab.org/modeller/tutorial/.
238 Andy M. Lau and Argyris Politis

7. If your protein of interest undergoes post translational proces-

sing, the canonical sequence reported on UniProt may be
different to the true experimental mass. Additional sequence
information can be found on UniProt for characterized entries.
8. Identified subcomplexes can be used to determine the connec-
tivity based on its internal constituents. Start from the smallest
subcomplex and build up a network of additional subunits.
9. While density maps can be opened in PyMOL, the graphics
viewer Chimera is better suited for visualizing density maps and
possessed advanced functions for fitting structures into
maps, etc.
10. Structures can be fitted into density maps using Chimera’s “Fit
in Map” function. Open both the map and structures in Chi-
mera. “Fit in Map” can be accessed from the Volume Viewer
window when a map file is open, under the “Tools” tab. You
may need to manually move the structure in close proximity of
the map file for better fitting.
11. The calculate_density.sh script can be opened using any text
editor. You will need to replace the “python¼”/anaconda2/
bin/python” line with the path to python for IMP on your
own system.
12. GMMs can be calculated for electron density maps as a sum of
three-dimensional gaussians. The shape of a structure in PDB
format can also be converted into a simulated volume, and
likewise converted into a GMM. The fit between each subunit
GMM can then be rapidly calculated against the GMM of the
complex. Consult the IMP documentation for more informa-
tion on how GMMs are used.
13. The number of residues that are represented by a single bead is
set in the “topology.txt” file under “bead_size.”
14. See IMP documentation at: https://integrativemodeling.org/
talks/dec_2016_workshop/IMP%20software%20introduc
tion%20and%20tutorial.pdf for additional information on rep-
lica exchange simulations.
15. See IMP documentation at: https://integrativemodeling.org/
talks/dec_2016_workshop/Modeling%20of%20the%
20Nup84%20complex.pdf for more information on scoring
functions.
16. Consult the PyMOL documentation for the “Align” function:
https://pymolwiki.org/index.php/Align.
17. CCS can be calculated through four different metrics, these are
the projection approximation (PA), projection superposition
approximation (PSA) [48], exact hard sphere scattering
(EHSS), and trajectory method (TM or TJM). These metrics
vary in accuracy and computational speed see also [19].
 Mass Spectrometry-Based Protein Modelling 239

18. Since proteins and complexes can exhibit conformational var-

iations, the reported CCS may vary depending on these
conformations [49].
19. The CCS of some proteins may not reflect their “native” con-
formations or those calculated from their crystal structures,
due to structural collapse in the gas phase. This phenomenon
commonly affects systems which are intrinsically flexible or
unstructured [50–52].
20. The TJM value reported by IMPACT is an approximation of
the true TJM value [43].

References
1. Alber F et al (2007) Determining the architec- spectrometry: application to the yeast eIF3:
tures of macromolecular assemblies. Nature eIF5 complex. Chem Biol 22:117–128.
450:683–694. https://doi.org/10.1038/ https://doi.org/10.1016/j.chembiol.2014.
nature06404 11.010
2. Ahdash Z, Pyle E, Politis A (2016) Hybrid 10. Russel D et al (2012) Putting the pieces
mass spectrometry: towards characterization together: integrative modeling platform soft-
of protein conformational states. Trends Bio- ware for structure determination of macromo-
chem Sci 41:650–653. https://doi.org/10. lecular assemblies. PLoS Biol 10:e1001244.
1016/j.tibs.2016.04.008 https://doi.org/10.1371/journal.pbio.
3. Bullock JMA, Sen N, Thalassinos K, Topf M 1001244
(2018) Modeling protein complexes using 11. Faull SV et al (2019) Structural basis of Cullin
restraints from crosslinking mass spectrometry. 2 RING E3 ligase regulation by the COP9
Structure 26:1015–1024.e1012. https://doi. signalosome. Nat Commun 10:3814. https://
org/10.1016/j.str.2018.04.016 doi.org/10.1038/s41467-019-11772-y
4. Degiacomi MT, Schmidt C, Baldwin AJ, Ben- 12. Zhou M et al (2014) ATP binding reduces
esch JLP (2017) Accommodating protein conformational flexibility in a rotary ATPase—
dynamics in the modeling of chemical cross- Evidence from ion mobility mass spectrometry.
links. Structure 25:1751–1757.e1755. Nat Chem 6(3):208–215
https://doi.org/10.1016/j.str.2017.08.015 13. Alber F et al (2007) The molecular architecture
5. Lasker K et al (2012) Molecular architecture of of the nuclear pore complex. Nature
the 26S proteasome holocomplex determined 450:695–701. https://doi.org/10.1038/
by an integrative approach. Proc Natl Acad Sci nature06405
U S A 109:1380–1387. https://doi.org/10. 14. Bechara C et al (2015) A subset of annular
1073/pnas.1120559109 lipids is linked to the flippase activity of an
6. Politis A et al (2014) A mass spectrometry- ABC transporter. Nat Chem 7:255–262.
based hybrid method for structural modelling https://doi.org/10.1038/nchem.2172
of protein complexes. Nat Methods 15. Bechara C, Robinson CV (2015) Different
11:403–406 modes of lipid binding to membrane proteins
7. Hall Z, Politis A, Robinson CV (2012) Struc- probed by mass spectrometry. J Am Chem Soc
tural modeling of heteromeric protein com- 137:5240–5247. https://doi.org/10.1021/
plexes from disassembly pathways and ion jacs.5b00420
mobility-mass spectrometry. Structure 16. Ahdash Z et al (2017) Mechanistic insight into
20:1596–1609. https://doi.org/10.1016/j. the assembly of the HerA-NurA helicase-nucle-
str.2012.07.001 ase DNA end resection complex. Nucleic Acids
8. Hansen K et al (2018) A mass spectrometry- Res 45:12025–12038. https://doi.org/10.
based modelling workflow for accurate predic- 1093/nar/gkx890
tion of IgG antibody conformations in the gas 17. Pagel K, Natan E, Hall Z, Fersht AR, Robinson
phase. Angew Chem Int Ed Engl 57 CV (2013) Intrinsically disordered p53 and its
(52):17194–17199 complexes populate compact conformations in
9. Politis A et al (2015) Topological models of the gas phase. Angew Chem Int Ed Engl
heteromeric protein assemblies from mass
240 Andy M. Lau and Argyris Politis

52:361–365. https://doi.org/10.1002/anie. 28. Sinz A (2006) Chemical cross-linking and mass

201203047 spectrometry to map three-dimensional pro-
18. Konijnenberg A et al (2014) Global structural tein structures and protein-protein interac-
changes of an ion channel during its gating are tions. Mass Spectrom Rev 25:663–682
followed by ion mobility mass spectrometry. 29. Sinz A (2018) Cross-linking/mass spectrome-
Proc Natl Acad Sci U S A 111:17170–17175. try for studying protein structures and protein-
https://doi.org/10.1073/pnas.1413118111 protein interactions: where are we now and
19. Gabelica V, Marklund E (2018) Fundamentals where should we go from here? Angew Chem
of ion mobility spectrometry. Curr Opin Chem Int Ed Engl 57:6390–6396. https://doi.org/
Biol 42:51–59. https://doi.org/10.1016/j. 10.1002/anie.201709559
cbpa.2017.10.022 30. Politis A, Schmidt C (2018) Structural charac-
20. Jurneczko E, Barran PE (2011) How useful is terisation of medically relevant protein assem-
ion mobility mass spectrometry for structural blies by integrating mass spectrometry with
biology? The relationship between protein computational modelling. J Proteome
crystal structures and their collision cross sec- 175:34–41. https://doi.org/10.1016/j.jprot.
tions in the gas phase. Analyst 136:20–28. 2017.04.019
https://doi.org/10.1039/c0an00373e 31. Bullock JMA, Thalassinos K, Topf M (2018)
21. Politis A, Park AY, Hall Z, Ruotolo BT, Robin- Jwalk and MNXL web server: model validation
son CV (2013) Integrative modelling coupled using restraints from crosslinking mass spec-
with ion mobility mass spectrometry reveals trometry. Bioinformatics 34:3584–3585.
structural features of the clamp loader in com- https://doi.org/10.1093/bioinformatics/
plex with single-stranded DNA binding pro- bty366
tein. J Mol Biol 425:12. https://doi.org/10. 32. Mendoza VL, Vachet RW (2009) Probing pro-
1016/j.jmb.2013.04.006 tein structure by amino acid-specific covalent
22. Politis A et al (2010) Integrating ion mobility labeling and mass spectrometry. Mass Spec-
mass spectrometry with molecular modelling trom Rev 28:785–815. https://doi.org/10.
to determine the architecture of multiprotein 1002/mas.20203
complexes. PLoS One 5:e12080. https://doi. 33. Schmidt C et al (2017) Surface accessibility and
org/10.1371/journal.pone.0012080 dynamics of macromolecular assemblies
23. Jenner M et al (2011) Detection of a protein probed by covalent labeling mass spectrometry
conformational equilibrium by electrospray and integrative modeling. Anal Chem
ionisation-ion mobility-mass spectrometry. 89:1459–1468. https://doi.org/10.1021/
Angew Chem Int Ed Engl 50:8291–8294. acs.analchem.6b02875
https://doi.org/10.1002/anie.201101077 34. Wang SC et al (2010) Ion mobility mass spec-
24. Jurneczko E et al (2013) Probing the confor- trometry of two tetrameric membrane protein
mational diversity of cancer-associated muta- complexes reveals compact structures and dif-
tions in p53 with ion-mobility mass ferences in stability and packing. J Am Chem
spectrometry. Angew Chem Int Ed Engl Soc 132:15468–15470. https://doi.org/10.
52:4370–4374. https://doi.org/10.1002/ 1021/ja104312e
anie.201210015 35. Hopper JT et al (2013) Detergent-free mass
25. Pacholarz KJ et al (2014) Dynamics of intact spectrometry of membrane protein complexes.
immunoglobulin G explored by drift-tube Nat Methods 10:1206–1208. https://doi.
ion-mobility mass spectrometry and molecular org/10.1038/nmeth.2691
modeling. Angew Chem Int Ed Engl 36. Adhikary S et al (2017) Conformational
53:7765–7769. https://doi.org/10.1002/ dynamics of a neurotransmitter:sodium sym-
anie.201402863 porter in a lipid bilayer. Proc Natl Acad Sci U
26. O’Reilly FJ, Rappsilber J (2018) Cross-linking S A 114:E1786–e1795. https://doi.org/10.
mass spectrometry: methods and applications 1073/pnas.1613293114
in structural, molecular and systems biology. 37. Mistarz UH, Brown JM, Haselmann KF, Rand
Nat Struct Mol Biol 25(11):1000–1008. KD (2016) Probing the binding interfaces of
https://doi.org/10.1038/s41594-018-0147- protein complexes using gas-phase H/D
0 exchange mass spectrometry. Structure
27. Leitner A et al (2010) Probing native protein 24:310–318. https://doi.org/10.1016/j.str.
structures by chemical cross-linking, mass spec- 2015.11.013
trometry, and bioinformatics. Mol Cell Prote- 38. Marcsisin SR, Engen JR (2010) Hydrogen
omics 9:1634–1649. https://doi.org/10. exchange mass spectrometry: what is it and
1074/mcp.R000001-MCP201 what can it tell us? Anal Bioanal Chem
 Mass Spectrometry-Based Protein Modelling 241

397:967–972. https://doi.org/10.1007/ 46. Notredame C, Higgins DG, Heringa J (2000)

s00216-010-3556-4 T-coffee: a novel method for fast and accurate
39. Hebling CM et al (2010) Conformational anal- multiple sequence alignment. J Mol Biol
ysis of membrane proteins in phospholipid 302:205–217. https://doi.org/10.1006/
bilayer nanodiscs by hydrogen exchange mass jmbi.2000.4042
spectrometry. Anal Chem 82:5415–5419. 47. Webb B, Sali A (2016) Comparative Protein
https://doi.org/10.1021/ac100962c Structure Modeling Using MODELLER.
40. Eisinger ML, Dorrbaum AR, Michel H, Curr Protoc Bioinformatics 54:5 6 1–5 6 37.
Padan E, Langer JD (2017) Ligand-induced https://doi.org/10.1002/cpbi.3
conformational dynamics of the Escherichia 48. Bleiholder C, Wyttenbach T, Bowers MT
coli Na(+)/H(+) antiporter NhaA revealed by (2011) A novel projection approximation algo-
hydrogen/deuterium exchange mass spec- rithm for the fast and accurate computation of
trometry. Proc Natl Acad Sci U S A molecular collision cross sections (I). Method.
114:11691–11696. https://doi.org/10. Int J Mass Spectrom 308:1–10. https://doi.
1073/pnas.1703422114 org/10.1016/j.ijms.2011.06.014
41. Martens C et al (2018) Direct protein-lipid 49. Faull PA et al (2009) Gas-phase metalloprotein
interactions shape the conformational land- complexes interrogated by ion mobility-mass
scape of secondary transporters. Nat Commun spectrometry. Int J Mass Spectrom
9:415 283:140–148. https://doi.org/10.1016/j.
42. Claesen J, Politis A (2019) POPPeT: a new ijms.2009.02.024
method to predict the protection factor of 50. Devine PWA et al (2017) Investigating the
backbone amide hydrogens. J Am Soc Mass structural compaction of biomolecules upon
Spectrom 30(1):67–76. https://doi.org/10. transition to the gas-phase using ESI-T-
1007/s13361-018-2068-x WIMS-MS. J Am Soc Mass Spectr
43. Marklund EG, Degiacomi MT, Robinson CV, 28:1855–1862. https://doi.org/10.1007/
Baldwin AJ, Benesch JL (2015) Collision cross s13361-017-1689-9
sections for structural proteomics. Structure 51. Porrini M et al (2017) Compaction of duplex
23:791–799. https://doi.org/10.1016/j.str. nucleic acids upon native electrospray mass
2015.02.010 spectrometry. ACS Cent Sci 3:454–461.
44. Ruotolo BT, Benesch JL, Sandercock AM, https://doi.org/10.1021/acscentsci.
Hyung SJ, Robinson CV (2008) Ion 7b00084
mobility-mass spectrometry analysis of large 52. Scott D, Layfield R, Oldham NJ (2015) Struc-
protein complexes. Nat Protoc 3:1139–1152. tural insights into interactions between ubiqui-
https://doi.org/10.1038/nprot.2008.78 tin specific protease 5 and its polyubiquitin
45. Benesch JL, Ruotolo BT (2011) Mass spec- substrates by mass spectrometry and ion mobil-
trometry: come of age for structural and dyna- ity spectrometry. Protein Sci 24:1257–1263.
mical biology. Curr Opin Struct Biol https://doi.org/10.1002/pro.2692
21:641–649. https://doi.org/10.1016/j.sbi.
2011.08.002
 Chapter 13

Optimization of Sample Preparation for the Observation

of Macromolecular Complexes by Electron (cryo-)
Microscopy
Alexandre Frechard, Grigory Sharov, Maximilien Werderer,
and Patrick Schultz

Abstract
Electron microscopy is a powerful tool for studying the homogeneity and structure of biomolecular
complexes. The small wavelength of electron and the availability of electron optics enable the direct
visualization of macromolecular assemblies in a large range of sizes between 5 and 100 nm. This informs
us about the degree of multimerization or aggregation and provides precise information about their general
shape and dimensions. When combined with sophisticated image analysis protocols, three-dimensional
(3D) information can be gained from 2D projections of the sample, leading to a structural description.
When intermediate steps of a reaction can be imaged, insights into the mode of action of macromolecules
can be gained, and structure–function relations can be established. However, the way the sample is prepared
for its observation within the vacuum of an electron microscope determines the information that can be
retrieved from the experiment. We will review two commonly used specimen preparation protocols for
subsequent single-particle electron microscopy observation, namely negative staining and vitrification.

Key words Single particle electron microscopy, Negative stain electron microscopy, Cryo-EM,
EM sample preparation

1 Introduction

In the last decades, imaging of biological macromolecules by elec-

tron microscopy became a major structural biology method to
describe the shape and size of macromolecules, to investigate
their oligomeric or aggregated state, and, in combination with
single particle image analysis, to describe their structure [1]. Ulti-
mately this approach aims at understanding the mode of action of
molecular assemblies by correlating their structure with their func-
tion. Handling macromolecular assemblies a few nanometers in size
for their observation in the vacuum of a transmission electron
microscope (TEM) while preserving the structural integrity of the

Arnaud Poterszman (ed.), Multiprotein Complexes: Methods and Protocols, Methods in Molecular Biology, vol. 2247,
https://doi.org/10.1007/978-1-0716-1126-5_13, © Springer Science+Business Media, LLC, part of Springer Nature 2021

243
244 Alexandre Frechard et al.

sample is challenging. Over the years, many obstacles were over-

come to reveal biomolecules in atomic details.
For most biological applications, the sample is applied onto
3 mm EM grids that can be made of different material. Continu-
ously carbon-coated grids are used for negative staining where
carbon acts as a support for the sample, whereas perforated
carbon-coated grids can in addition be used in cryo-EM when
adsorption has to be avoided. The carbon surface is generally
hydrophobic and has to be rendered hydrophilic to allow proper
spreading of the sample and adsorption of the biomolecules
[2]. The properties of the carbon surface are altered by glow
discharge, a process where the grids are exposed to an ionizing
plasma formed by applying an electric current through a gas volume
at low pressure. The ionized gas molecules will modify the proper-
ties of the carbon surface of the grids and render it hydrophilic.
We will describe two popular specimen preparation protocols
for subsequent single-particle electron microscopy observation,
namely negative staining and vitrification.

1.1 Negative The most straightforward negative staining approach was first
Staining described by Brenner and Horne in 1959 [3]. This method brings
important preliminary information on sample quality and may be
used on a daily basis to follow and optimize a purification protocol.
Associated with single particle image analysis, it may provide impor-
tant information on homogeneity or oligomerization states. This
information may not be portable for cryo-EM sample preparation
but guaranties a firm starting point. Negative staining consists of
embedding the protein into a matrix of heavy atom salt to increase
the contrast of unstained biomolecules and prevent, to some
extent, their collapse during the drying process when exposed to
the vacuum of the microscope. Stains of heavy metal salts, such as
uranyl acetate, uranyl formate, phosphotungstic acid, and others,
are commonly used, since they strongly scatter electrons and pro-
duce high amplitude contrast. It is important to notice that nega-
tive staining reveals only the surface and the overall shape of protein
molecules and does not provide information about their inner
structure. The resolution that can be achieved after image analysis
of multiple images is limited to 15–20 Å, by the graininess of the
stain and particle distortions during the drying process. This
method is, however, very useful to check particle homogeneity
and to determine initial 3D models. Furthermore, the sample
preparation is easy to perform and to observe at room temperature
in a standard transmission electron microscope. The stained sam-
ple, or more specifically, its heavy metal cast, is more resistant to
radiation damage produced by the electron irradiation, and the
high image contrast favors image interpretation.
 Specimen Optimization for Single Particle cryo-EM 245

1.2 Vitrification The major drawback of negative staining is that the sample is dried
during preparation and introduction into the microscope vacuum.
To overcome this limitation, several options have been explored to
keep the specimen hydrated including the construction of hydrated
chambers around the specimen holder. Attempts were made to
replace water with sugar polymers to functionally replace the shield-
ing properties of water, and remarkable results were obtained for
2D crystals of membrane proteins. Cryogenic methods were tried
early on and proved efficient to reduce radiation damage resulting
from the interaction of electrons with matter by a factor of 2 at
liquid nitrogen temperature. However, freezing water invariably
resulted in ice crystal formation which affects sample distribution
and image quality. Being able to obtain vitreous ice at atmospheric
pressure was a major breakthrough for studying biological macro-
molecules in a fully hydrated state in the electron microscope. The
method consists of flash freezing a thin layer of an aqueous suspen-
sion of the purified biomolecule to form a vitrified sample that can
be transferred and observed in a cryo electron microscope at liquid
nitrogen temperature. A layer of suspension thinner than 100 nm is
generally obtained by removing the excess of a drop of sample
applied onto the electron microscopy grid by blotting with a filter
paper. Alternatively, sample micro droplets can be sprayed onto the
grid or applied by a nanoliter dispensers [4]. Vitrification is gener-
ally performed by plunging the EM grid into liquid ethane cooled
by liquid nitrogen for better heat dissipation. High pressure freez-
ing devices have been developed for the vitrification of up to
300 nm thick sample [5]. This approach, which was rewarded the
Nobel Prize in chemistry in 2017, preserves the hydrated structure
of the molecules at atomic resolution. The vitrification process can
be performed once the sample is adsorbed on a carbon film, but
adsorption can be avoided by using holey carbon films. In this case,
observations are made on the thin layer of vitrified suspension
stretched over the carbon-free holes. However, the sample still
experiences the interaction with the air–water interface which can
be deleterious when the specimen is prone to denaturation or can
result in preferential orientations of the molecules [6].

2 Materials

We list here the equipment needed to prepare purified biological

molecules for their observation in electron microscopy. The nega-
tive staining and vitrification methods require common material
listed in Subheading 2.1 and have their own specificities detailed
in Subheadings 2.2 and 2.3, respectively.
246 Alexandre Frechard et al.

2.1 Common 1. The EM grid, a universal, and versatile specimen support. The
Materials universal support to manipulate EM samples is a thin metallic
and Equipment grid 3 mm in diameter adapted for most electron microscope
holders. EM grids are made of copper, copper/rhodium, or
gold to be conductive and eliminate electron charge accumula-
tion. Square holes of different sizes are separated by grid bars
that absorb electrons. Copper mesh grids with 300 grid squares
per inch are commonly used because they are conductive, stable
under the electron beam, and inexpensive. The grid is generally
covered by a 5–20 nm thin carbon film onto which protein
complexes are adsorbed for subsequent negative staining. To
avoid sample adsorption in cryo-EM, the grids can also be
covered with a self-made carbon film perforated with holes of
various diameter ranging between 0.3 and 10 μm in diameter
[7]. More recently, perforated grids with calibrated circular
holes arranged into regular arrays were developed to serve
automated data acquisition [8]. Such perforated grids can be
covered by a continuous thin carbon film when adsorption
cannot be avoided. Holey carbon films can also be covered by
a single monolayer thick graphene or graphene oxide crystal-
line layer instead of amorphous carbon, thus leading to a less
grainy background [9] (Oxford Instruments, quantifoil micro
tools GMBH).
2. High precision tweezers. To properly manipulate the EM grids
without distortion and contamination, high precision and
sharp tweezers are recommended (Dumont style 5 tweezers,
Fine Science Tools GMBH ref.: 11252-30). Make sure to grab
the grids by the rim to not damage or distort the support.
Standard straight tweezers are sufficient to manipulate grids,
but other types of tweezers can be useful in electron micros-
copy: anti-capillarity tweezers have a wider angle at the tip
when closed, limiting the amount of liquid retained by capillar-
ity. Inverted tweezers are also found useful since they remain
closed until squeezed, limiting the risks of dropping grids when
handling the tweezers.
3. Glow discharge and plasma cleaner devices. These table top
instruments are used to produce a gas plasma to charge the
surface of the grids by an electrostatic potential in order to
produce a hydrophilic surface suitable for protein absorption.
The devices differ by the quality of the vacuum (between 10 2
and 10 6 Torr, or 1.3 and 1.3  10 4 Pa), the way the plasma is
produced, and the mixture of gases that can be used to form
the plasma (ELMO, Cordouan Technologies; Nanoclean
model 1070, Fischione).
4. Filter paper. Different steps of the protocol require blotting the
grids to remove excess of staining solution or water (Macherey-
Nagel, ref.: 202009).
 Specimen Optimization for Single Particle cryo-EM 247

5. Cross-linkers. Chemical cross-linkers such as glutaraldehyde

(Sigma, ref.: G5882) or bis-sulfosuccinimidyl suberate (BS3,
Thermo Fisher Scientific, ref.: 21580) are often used to stabi-
lize fragile sample by forming chemical bonds between proxi-
mal lysine residues. Chemical cross-linkers have a short life time
and may form oligomeric states. They can be aliquoted and
frozen to extend their life time.

2.2 Negative 1. Heavy atom salt solution. Uranyl acetate (Agar Scientific, ref.:
Staining AGR1260A; Polysciences ref.: 21447-25) uranyl formate,
phosphotungstic acid (PTA), and other heavy metals are com-
monly used to embed the biological sample into a shell of
electron-dense material. The heavy atom salt is dissolved in
water at 2% (w/v). Most of the solutions are at low pH, and
only few of them, such as PTA, can be buffered to pH 7.0
without forming aggregates. In this respect, phosphate buffer
present in the sample solution may provoke uranyl acetate
precipitation observed as dark clusters under the microscope.
2. Handling heavy metal salts requires strict safety protections:
use gloves, mask and safety glasses and work under hood when
preparing solutions. Uranyl salts are slightly radioactive, and
specific safety precautions are necessary: store uranyl salts in a
lead protection box, identify the workbench used to work with
uranyl acetate, weigh powders in a secured work station, work
under a hood protected by a Plexiglas screen when handling
concentrated solutions, and finally, make sure to dispose any
contaminated waste according to your lab policy. Radiations
emitted from 2% solutions are shielded by water molecules and
the glass bottle. Depending on national policies, uranyl-based
reagent might be difficult or even illegal to buy. Please make
sure your experiments comply with your local regulations.
3. Note that there are now non-radioactive alternatives to uranyl
acetate: based on methylamine vanadate (NanoVan, Nanop-
robes), methylamine tungstate (Nano-W, Nanoprobes), Ura-
nyLess (TAAB), and UA-Zero (Agar Scientific).
4. Parafilm. The protocol includes several washing steps aiming at
solution replacement. Parafilm may be used to deposit small
solution drops on top of which the EM grid can be placed
(VWR international SAS, ref.: 97949).

2.3 Cryo-EM Preparing the sample for cryo-EM requires more material and
equipment than negative staining. Although automated spray sys-
tems and microfluidic-based devices are emerging, vitrification is
usually performed using a cryo-plunger. Make sure that all the
materials are at your disposal before you start since the protocol
cannot be interrupted.
248 Alexandre Frechard et al.

1. Liquid nitrogen. Liquid nitrogen is used at several steps in the

protocol to refrigerate and liquefy ethane, to store the vitrified
grid and cryo-box, and to transfer the cryo-box to a dedicated
short-term or long-term storage device. Once vitrified, the
specimen has to be kept at liquid nitrogen temperature (boiling
temperature of 196  C under normal atmospheric pressure)
to avoid water devitrification.
2. Plastic beaker. A 1 L polypropylene beaker with a handle to
pour nitrogen (Vitlab, ref.: 442941).
3. Cryo-boxes. These round boxes are used to store the vitrified
samples in liquid nitrogen. They generally hold four EM grids
and have a transparent cover hold in place by a screw. The cover
has a single slit and can be rotated to select one of the four EM
grid slots (Oxford Instruments, ref.: G3539).
4. Screw driver. A screw driver is necessary to fasten and unfasten
the cryo-box cover immersed in liquid nitrogen.
5. Long isolated tweezers. Twenty-centimeter long tweezers are
used to manipulate the cryo-boxes submerged in nitrogen. To
protect the operator from frostbites the tweezers should be
cold-isolated by a protective layer.
6. Liquid nitrogen tank. A 5 L liquid nitrogen container is
required to store over longer periods the cryo-boxes which
contain the grids. The cryo-boxes can be put into Falcon
tubes perforated for liquid nitrogen influx and fixed to a
2–3 mm thick plastic-coated metallic cord which hangs out of
the tank once the tube is immersed. More elaborate systems
have been developed to organize multi-user grid storage, effi-
cient cryo-box retrieval, and automated nitrogen refill (EMS
Cryo Pucks G2).
7. Cryo-plunger. Specimen vitrification is commonly achieved by
forming a thin layer of the sample on the EM grid through
blotting away the excess of sample and subsequent plunging of
the grid into a liquid ethane slush refrigerated by liquid nitro-
gen [5]. Simple, versatile, and portable devices were designed
in which a tweezer holding the grid was fixed on a weighted
friction-reduced rod that moved by gravity to plunge the grid
into the cryogen container. Ethane is liquefied in a 2 mL
aluminum container inserted in a styrofoam container of liquid
nitrogen. Since the boiling temperature of liquid nitrogen is
lower than the melting temperature of ethane ( 183.3  C), the
ethane will tend to solidify. An ethane heating device or a
properly designed ethane tank assures that ethane does not
reach liquid nitrogen temperature and stays as a liquid slush.
Highly automated devices are currently available, providing a
temperature and humidity controlled chamber to avoid
 Specimen Optimization for Single Particle cryo-EM 249

evaporation, a force- and time-controlled blotting and auto-

mated plunging [10] (Vitrobot Mark IV, Thermo Fisher; EM
GP2, Leica).
8. Ethane gas cylinder. A 5 L cylinder containing compressed
ethane gas is required to fill the ethane container. The cylinder
has to be equipped with a pressure reducer to fine tune the gas
flow and with a thin flexible plastic tube (diameter of 2 mm) to
flow the ethane stream into the liquid nitrogen-cooled cryogen
container. The extremity of a pipette tip can be mounted at the
end of the plastic tube for increased precision in driving the
ethane flow.
9. Safety issues. Handling liquid nitrogen and ethane which are
very cold and vaporize into large amounts of gas requires strict
safety protections. Liquid ethane close to its solidification tem-
perature is dangerous and will freezes any hydrated object upon
contact. Care must be taken to avoid spills on skin or eyes. The
use of a face shield is highly recommended. Thick gloves must
be used to manipulate cold objects.

3 Methods

3.1 Negative Negative staining is a fast method to assess the quality of your
Staining sample by checking its stability, homogeneity, and propensity to
aggregate. Key steps of sample preparation are illustrated in Fig. 1
and detailed below.

3.1.1 Preparation of a 1. Glow discharge a carbon-coated EM grid to render the carbon

Negatively Stained Sample support hydrophilic. Place the EM grid in the glow discharging
device, carbon side facing up. The nature of the gas present in
the chamber, the partial gas pressure, the voltage applied, the
glow discharge time, and the geometry of the device will all
affect the ionization of the carbon support. Using air as the
residual gas at a pressure of 1.8  10 1 bar, a current of 2 mA
should be applied for 45 s to assure a suitable hydrophilicity
(Fig. 1a).
2. Deposit 3–5 μL of your sample onto the glow discharged side
of the grid held with tweezers and wait for 1 min for protein
adsorption. The protein concentration should be around
30 μg/mL (Fig. 1b).
3. Place a square of parafilm onto the bench and deposit four
separate 40 μL drops, three drops of water or buffer, and one
drop of phosphotungstic acid at 2%.
4. After adsorption (step 2), place the grid sequentially on top of
the first three droplets of water to remove the material not
adsorbed to the carbon (Fig. 1c). Then remove the excess of
250 Alexandre Frechard et al.

Fig. 1 Negative staining. (a) The EM grid is glow-discharged in a plasma of ionized air to render the support
hydrophilic. (b) 3 μL of the purified biomolecules solubilized in buffer is allowed to adsorb on the EM grid
support for 1 min. (c) The excess of sample and detrimental buffer components are diluted away by several
passes on top of buffer drops. (d) Sample staining: the grid is deposited on the top of a drop of stain—here a
2% of uranyl acetate—for 30 s. (e) The staining solution is blotted away using filter paper. (f) Electron
micrograph of a well-stained sample, particles are well dispersed, show little aggregation and optimal
embedding in the stain layer

water using filter paper and transfer the grid to the last drop
containing the phosphotungstic acid and wait for 30 s
(Fig. 1d).
5. Grab the grid with tweezers, approach the filter paper from the
side of the grid, gently remove the excess of stain by capillarity,
let the grid dry completely for 1 min, and store it in a grid
storage box before observation in the transmission electron
microscope (Fig. 1e).
 Specimen Optimization for Single Particle cryo-EM 251

3.1.2 Observation Ideally the sample should be evenly distributed with little interac-
and Trouble Shooting tions between individual particles (compare Figs. 1f and 3a). The
macromolecules should be included in a layer of stain of the same
thickness as the particle to sustain optimally the sample, to reveal all
parts and to obtain the best contrast. It is often necessary to screen
several places on the grids to find optimal staining. The following
indications may help to improve staining.
1. Poor staining. In a standard bright field transmission electron
microscope, your protein sample should appear as a lighter area
embedded in a darker background. If your sample is poorly
stained, you will only see fingerprints of your protein (Fig. 3b)
or worse, and your sample may be positively stained and appear
as dark spots on a lighter background. You may adjust the glow
discharge conditions to render the support more hydrophilic
and make sure that the staining solution is not removed too fast
by wetting the filter paper beforehand. Adding 0.5% glycerol
may help to reduce the drying speed and improve staining.
2. Excess staining. Your sample may be well embedded, but a
thick layer of stain may be superimposed to the particles, thus
reducing the contrast and obscuring fine details of your sample
(Fig. 3c). Sometimes wiping away the stain faster can help.
Check that your sample buffer does not contain large excess
of sucrose of glycerol which increases its viscosity. In this case,
an additional washing step can be added.
3. Presence of stain crystals. Uranyl salts precipitate at neutral pH
or in contact with phosphate buffer and form small crystals
visible in EM. Make sure that these reagents are absent or
sufficiently dilute in the last staining step and that the pH is
suitable.
4. Damaged/destroyed proteins. Some fragile protein complexes
may be altered during grid preparation and even fall apart
during the dilution, adsorption, or staining steps. Chemical
cross-linkers such as glutaraldehyde or BS3 may stabilize the
sample. As a standard procedure, glutaraldehyde is added at a
final concentration of 0.1–0.5% for 2 min, before adsorption on
the EM grid. Cross-linkers may lead to protein aggregation,
and therefore, cross-linking time and concentration need to be
adjusted. Adding 1% glycerol in the staining solution may help
sustaining fragile samples and prevent their collapse during the
drying step.
5. Small number of particles. When a smaller number of particles
than expected is observed, the sample may form aggregates.
Aggregates are not always detected because of their small inter-
action interface with the carbon support which can lead their
dissociation during the washing or staining steps. Alternatively,
252 Alexandre Frechard et al.

the sample suspension may contain reagents (elution peptides,

detergents, lipids) that compete with the protein for adsorp-
tion. In this case, these reagents have to be removed whenever
possible.

3.2 Vitrification The protocol presented here is adapted for a Vitrobot, but most of
the steps can be adjusted to a handmade plunger or a device from
3.2.1 Preparation
another manufacturer. Key steps are illustrated in Fig. 2 and
of the Vitrified Suspension
detailed below.
1. Prepare four EM grids coated with holey carbon
(non-adsorbed sample) or plain carbon (adsorbed sample)
and glow discharge them as described in Subheading 3.1.1.
2. Prepare the plunging device by adjusting the temperature and
humidity of the chamber.

Fig. 2 Sample vitrification. (a) Styrofoam holder filled with liquid nitrogen, showing the central ethane
container. The ethane gas was liquefied, holding the tube appearing on the left into the cold container. (b)
The grid is loaded on the tweezers attached to the plunger. (c) 3 μL of sample is deposited on the grid which
will be lifted up to the filter paper pads that will be automatically pressed against the grid to blot away the
excess of liquid. (d) The plunger quickly submerges the grid in the liquid ethane. (e, f) The vitrified sample is
transferred into a cryo-box for storage. (g) Electron micrograph of a frozen hydrated sample showing well-
dispersed particles, little aggregation, thin ice, and no ice contamination
 Specimen Optimization for Single Particle cryo-EM 253

3. Cool down the liquid nitrogen and ethane containers with

liquid nitrogen. Once it has reached liquid nitrogen tempera-
ture remove liquid nitrogen from the ethane container and
place it in its holder.
4. Liquefy ethane in the ethane container by introducing a flow of
ethane gas with a polyethylene tubing connected to the ethane
cylinder. The gas flow can be manually adjusted by clamping
the tubing. Place the tube at the bottom of the container with a
weak gas flow until some liquid ethane forms. At that point, the
gas flow may be increased until the liquid level reaches a few
millimeters from the top of the container. Wait for the ethane
to solidify at the rim of the container, which signals that the
ethane is cold enough.
5. Adjust the liquid nitrogen level and place the container under
the plunger.
6. Fix the grid onto the plunger tweezer and transfer it into the
chamber.
7. Apply 3 μL of the sample onto the grid. In the absence of
adsorption, the sample concentration needs to be higher, in
the order of 0.3–1 mg/mL.
8. Blot the grid using the filter paper supplied by the manufac-
turer after having adjusted the blotting time and blotting force.
9. Plunge the grid down into liquid ethane.
10. Transfer quickly the grid into a cryo-box and store it in liquid
nitrogen.

3.2.2 Observation 1. Opaque grid. If your grid is not transparent to electrons, it

and Trouble Shooting generally means that the ice is too thick. Blot time and force can
be increased. Make sure that reagents that increase viscosity
such as glycerol or sucrose are absent. The thickness of the
vitrified layer is of key importance to reach high resolution in
single particle cryo-EM. The thinnest is the better, and small
amounts of nonionic detergents may be added to form even
thinner layers.
2. Poor contrast. Thick ice will reduce image contrast. Some
reagents such as glucose or glycerol may affect the contrast.
Make sure that their concentration is lower than 1%.
3. Ethane contaminants. Long filaments forming aggregates are
sometimes present and originate from impurities present in the
ethane which do not evaporate in the temperature and vacuum
conditions of the microscope (Fig. 3d). Avoid solid ethane
accumulation on the EM grid which forms when transferring
the grid from the ethane to the liquid nitrogen container after
vitrification. Slow retraction of the grid from the container is
the best way to prevent ethane accumulation. Solid ethane may
254 Alexandre Frechard et al.

Fig. 3 Troubleshooting. Negative staining: (a) Electron micrograph showing aggregated particles. (b) Electron
micrograph showing a heavily stained sample in which the details of the particles are lost, and the overall
contrast is weak due to stain accumulation on top of the particle. (c) In this field, the particles are not well-
stained and embedded, and only a small rim of stain surrounds each particle whose structure is poorly
sustained. Sample vitrification: (d) Contaminant filaments arising from residues remaining after ethane
evaporation. (e) Transformed ice image due to warming up of the sample during transfer in the microscope.
(f) Partially molten hexagonal ice crystals (arrows) are spread over the grid surface and may have been
collected during the grid transfer or mounting steps and exposition to humid environment

be gently removed under liquid nitrogen with tweezers. Stor-

age for a few days in liquid nitrogen may detach the solid
ethane from the grid.
4. Crystalline ice. During freezing, parts of the water layer may
not have cooled down fast enough, thus forming hexagonal ice.
This should be restricted to the thicker parts of the suspension
layer. If this problem persists, the plunging speed or depth may
need to be adjusted. Alternatively if the grid heats up during its
manipulation, the vitreous ice will transform into cubic ice and
deteriorate image quality [5] (Fig. 3e). To avoid this problem,
it is imperative to keep the grid cold during the transfer and
mounting steps.
5. Ice contamination. During the grid transfer and mounting
steps, ambient humidity can condense on the grid and form
ice crystals on its surface (Fig. 3f). Such crystals may also be
 Specimen Optimization for Single Particle cryo-EM 255

collected from the liquid nitrogen. Furthermore, ice crystals

loosely attached to the grid may evaporate and re-condense
onto the cold grid surface, thus forming a continuous or speck-
led ice layer sometimes difficult to detect. This continuous
contamination accumulates over time due to water molecules
present in the vacuum of the microscope. Working in a dry
environment, changing liquid nitrogen frequently to avoid ice
crystal accumulation, handling the grids in liquid nitrogen or
very close to the surface, and heating/emptying the containers
between two freezing sessions will reduce ice contaminations.
If contamination builds up rapidly in the microscope, check the
anticontamination devices of your microscope.
6. Damaged sample. During cryo-EM sample preparation, pro-
tein complexes may be damaged when placed in a thin layer of
suspension due to its interactions with the air–water interface.
This may result to partial denaturation, dissociation of subu-
nits, and aggregation. Sample cross-linking with glutaralde-
hyde or BS3 as described in Subheading 3.1.2 generally helps
stabilizing the molecular assemblies. Alternatively, the air–
water interface may be screened with amphiphilic molecules
such as detergents. In this case, the particle concentration may
need to be increased since the air–water interface does not
concentrate the biomolecules any more. The possibility to
adsorb the proteins on a continuous carbon film is worth
exploring to avoid the air–water interface.
7. Lack of particles. It is difficult to predict the protein concentra-
tion needed to produce projection image with densely packed
particles. Recent experiments showed that in most cases as
much as 90% of the particles accumulate at the air–water inter-
face [6]. The number of particles adsorbed on the grid depends
on their propensity to accumulate at the interface. In the
absence of surface accumulation, an extremely rare situation,
particle concentrations of 100 mg/mL (10% of the volume)
may be needed to produce densely populated images. Absence
of particles may therefore also arise from competition for the
surface and high concentrations of detergents or of hydropho-
bic peptides that may shield the surface and prevent surface
accumulation of the sample. In extreme cases, most of the
particles are denatured and form a grainy background. A
cross-linking experiment generally helps sorting out this
problem.
8. The quality of the frozen hydrated grid depends critically on
sample homogeneity and stability. It is essential to give the
highest priority to protein purification to increase the chances
to produce an exploitable grid.
256 Alexandre Frechard et al.

Acknowledgments

We acknowledge the support from the Institut National de la Santé

et de la Recherche Médicale (INSERM), the Centre National pour
la Recherche Scientifique (CNRS), the Ligue contre le Cancer, and
the grant ANR-10-LABX-0030-INRT, a French State fund man-
aged by the Agence Nationale de la Recherche under the frame
program Investissements d’Avenir ANR-10-IDEX-0002-02. We
acknowledge the use of resources of the French Infrastructure for
Integrated Structural Biology FRISBI ANR-10-INBS-05 and of
Instruct-ERIC.

References
1. Cheng Y (2018) Single-particle cryo-EM-how 6. Noble AJ et al (2018) Routine single particle
did it get here and where will it go. Science 361 CryoEM sample and grid characterization by
(6405):876–880 tomography. Elife 7:e34257
2. Dubochet J et al (1971) A new preparation 7. Fukami A, Adachi K (1965) A new method of
method for dark-field electron microscopy of preparation of a self-perforated micro plastic
biomacromolecules. J Ultrastruct Res 35 grid and its application. J Electron Microsc 14
(1):147–167 (2):112–118
3. Brenner S, Horne RW (1959) A negative stain- 8. Russo CJ, Passmore LA (2016) Progress
ing method for high resolution electron towards an optimal specimen support for elec-
microscopy of viruses. Biochim Biophys Acta tron cryomicroscopy. Curr Opin Struct Biol
34:103–110 37:81–89
4. Jain T et al (2012) Spotiton: a prototype for an 9. Palovcak E et al (2018) A simple and robust
integrated inkjet dispense and vitrification sys- procedure for preparing graphene-oxide cryo-
tem for cryo-TEM. J Struct Biol 179(1):68–75 EM grids. J Struct Biol 204(1):80–84
5. Dubochet J et al (1988) Cryo-electron micros- 10. Vos MR et al (2008) The development of a
copy of vitrified specimens. Q Rev Biophys 21 glove-box/Vitrobot combination: air-water
(2):129–228 interface events visualized by cryo-TEM.
Ultramicroscopy 108(11):1478–1483
 Chapter 14

Solubilization and Stabilization of Native Membrane

Proteins for Drug Discovery
Vincent Corvest and Anass Jawhari

Abstract
Membrane proteins (MPs) are stable in their native lipid environment. To enable structural and functional
investigations, MPs need to be extracted from the membrane. This is a critical step that represents the main
obstacle for MP biochemistry and structural biology. Here we describe detergent solubilization screening of
MPs using dot-blot and Western-blot analyses. Good solubilization conditions are ranked for their best
capacity to stabilize MPs using thermal shift assay. The protein functionality is evaluated by radioligand
binding (for G-protein-coupled receptor) and ATPase activity (ABC Transporter) and finally the aggrega-
tion status as well as protein homogeneity are assessed by Native-polyacrylamide gel, chemical cross-linking,
and size exclusion chromatography.

Key words Membrane proteins, Solubilization, Stabilization, Detergent, Native-PAGE, Size exclu-
sion chromatography, Chemical-crosslinking, Function

1 Introduction

Membrane proteins (MP) represent ~30% of human proteins

[1]. They are crucial for cellular physiology as they are directly
involved in cell adhesion, cell–cell communication, signal transduc-
tion, and transport [1]. This is why large majority (~70%) of
therapeutic targets are membrane proteins (MPs) [2]. To better
understand MP function and to enable drug development using
structure-based drug design or fragment-based drug discovery,
MPs need to be extracted from their membrane environment
[3]. MPs extraction and solubilization represent the most critical
step of MPs isolation [4], since depending on the used surfactant/
detergent, they can go through significant structural constraints
which may significantly compromise their structural and functional
integrities. To date, no detergent has been reported which tackles
membrane protein solubilization and stabilization universally.
Therefore, it is essential to screen for the most suitable detergent
[5]. Here we show that solubilization efficiency of different

Arnaud Poterszman (ed.), Multiprotein Complexes: Methods and Protocols, Methods in Molecular Biology, vol. 2247,
https://doi.org/10.1007/978-1-0716-1126-5_14, © Springer Science+Business Media, LLC, part of Springer Nature 2021

257
258 Vincent Corvest and Anass Jawhari

Fig. 1 Screening of membrane proteins solubilization. (a) Large solubilization screening performed in 96-well
plate and monitored by dot-blot. (b) Confirmation of solubilization efficiency by Western blotting. Membrane
protein A and B correspond to typical easy and difficult to solubilize targets, respectively (Adapted from [7])

detergents can be monitored in a 96-well plate dot blot format for

easy and difficult to solubilize MPs (Fig. 1a). To avoid false posi-
tives, Western blot experiments are performed to confirm the pres-
ence of the solubilized proteins of interest at the expected
molecular weight (Fig. 1b). This approach was previously reported
[6, 7]. In the following example, the best stabilizing detergent is
evaluated on its ability to solubilize and stabilize MPs in a Western
blot–based thermal-shift assay (Fig. 2a). This assay relies on the
assumption that unstable heated proteins will aggregate, and after
ultracentrifugation and Western blot, the band intensity
corresponding to the protein will decay proportionally to its insta-
bility [8]. Few examples using this approach were described [8–
10]. To make sure that solubilization and purification steps were
not harmful to the protein, functional assays are performed.
Figure 2b shows two examples, a radioligand binding assay for a
G-protein-coupled receptor (A2AR) and an ATPase activity assay for
an ABC transporter (BmrA). The structural integrity and more
 Solubilization and Stabilization of Membrane Proteins 259

Fig. 2 Membrane protein stability assessment. (a) Western blot based thermostability assay (adapted from
[12]). (b) Functional investigation using radioligand binding and ATPase activity assays for a GPCR and a
Transporter target, respectively. Two examples were previously described [9, 11]

particularly the aggregation status and homogeneity of the MPs are

evaluated by different methods. Native-PAGE helps to ensure that
the protein can migrate inside of the gel, suggesting a
non-aggregation state (Fig. 3a). This is further confirmed by size
exclusion chromatography (SEC, Fig. 3c) which do not show any
protein signal at the void volume. The protein homogeneity and
oligomeric state can also be investigated by SEC, Native-PAGE,
and chemical cross-linking (Fig. 3b).

2 Materials

All solutions are prepared using ultrapure water (18.2 MΩ-cm

conductivity at 25
C) and analytical grade reagents. All chemicals
are obtained from Sigma-Aldrich (unless indicated otherwise).

2.1 Solubilization 1. Bio-Dot® microfiltration apparatus (Bio-Rad).

Screening 2. Phosphate-buffered saline (1 PBS): 25 mM Na2HPO4,
150 mM NaCl, pH 8.0.
260 Vincent Corvest and Anass Jawhari

Fig. 3 Membrane protein behavior in solution. (a) Native-PAGE on purified Matrix 2 ion channel revealed by
western blot. (b) Chemical cross-linking assay using different amounts of glutaraldehyde, analyzed by
SDS-PAGE and revealed by Western blot. (c) Gel filtration (5-150GL) profile and corresponding Western blot
analysis. (Adapted from [13]). A similar approach was also previously reported for the characterization of the
human transcription factor TFIIE [14]

3. Protease inhibitor cocktail (1 PIC): 1 tablet/100 mL buffer

(SigmaFAST EDTA free).
4. Blocking buffer: 1 Roti®-block buffer (Carl Roth).

2.2 Radiobinding 1. MicroBeta filtermat 96-cell harvester (Perkin-Elmer).

Assay 2. MicroBeta2 microplate counter (Perkin-Elmer).
3. PEI-coated GF/B glass fiber filter-bottom 96-well microplate
with BackSeal and TopSeal (Perkin-Elmer).
4. Ultima Gold MV liquid scintillation cocktail (Perkin-Elmer).
5. Wash buffer: 50 mM Tris–HCl, pH 7.4.
6. Radioligand solution: 3.6 μM 3H-CGS21680 (Perkin-Elmer)
in wash buffer.
7. Nonspecific ligand solution: 3.6 μM 3H-CGS21680, 1 mM
CGS21680 in wash buffer.
8. γ-Globulin solution: 0.1% (w/v) γ-globulin in wash buffer.
9. PEG6000 solution: 25% (w/v) PEG6000 in wash buffer.
 Solubilization and Stabilization of Membrane Proteins 261

2.3 ATPase 1. Inorganic phosphate (Pi) standard stock solution: 10 mM

Activity Assay Na2HPO4, pH 8.0.
2. Adenosine 50 -triphosphate (ATP).
3. Sodium orthovanadate (Na3VO4) heated at 95
C.
4. Tris-buffered saline (1 TBS): 50 mM Tris–HCl, 50 mM
NaCl, pH 8.0.
5. Reaction buffer: 50 mM Tris–HCl, 50 mM NaCl, 10 mM
MgCl2, 6 mM NaN3, pH 8.0.
6. Revelation buffer (to be prepared just before use): 45 mM
ascorbic acid, 17.5 mM ammonium molybdate, 15 mM zinc
acetate, pH 5.0.

2.4 Clear-Native- 1. 5 native loading buffer: 250 mM Tris–HCl, 50% glycerol,

PAGE 0.25% bromophenol blue, pH 8.0.
2. CN-PAGE anode buffer: 25 mM imidazole, pH 8.0.
3. CN-PAGE cathode buffer: 50 mM tricine, 0.05%
Na-deoxycholate, 0.01% n-dodecyl β-D-maltoside (DDM,
Carl Roth), 7.5 mM imidazole, pH 8.0.

2.5 Glutaraldehyde 1. Glutaraldehyde (Carl Roth).

Cross-Linking Assay 2. 5 SDS-DTT loading buffer: 250 mM Tris–HCl, 10% sodium
dodecyl sulfate (SDS), 0.5 M dithiothreitol (DTT), 50% glyc-
erol, 0.25% bromophenol blue, pH 8.0.

3 Methods

3.1 Screening of Here we describe a 96-well microplate solubilization screening for

Membrane Proteins membrane proteins using the micro-ultracentrifugation to separate
Solubilization soluble from insoluble fractions. For limited micro-
ultracentrifugation capacity, an alternative using the biotinylated
membranes solubilization and separation method is described
in [7].

3.1.1 Cell Lysis and 1. Resuspend dry cell pellet at 2 mL/g with 1 PBS supplemen-
Membrane Fractionation ted with 1 PIC.
2. Lyse cells on ice using a Bead Beater homogenizer with
0.1 mm diameter glass beads (0.5 mm for yeast) with five cycles
of 30 s lysis followed by 2 min cool down (see Note 1).
3. Centrifuge at increasing speeds at 4
C to collect different
membrane fractions: 1000  g for 5 min (whole cell and cell
debris), 15,000  g for 30 min (internal membranes, ER,
mitochondria, etc.) and 100,000  g for 45 min (plasma
membrane).
262 Vincent Corvest and Anass Jawhari

4. Resuspend membrane fractions at 0.5 mL/g of initial dry cell

pellet with 1 PBS, 1 PIC, 10% (v/v) glycerol.
5. Quantify total membrane proteins of each fraction using the
micro BCA protein assay kit (Thermo Scientific) and determine
the protein of interest localization by SDS-PAGE and
Western blot.
6. Aliquot, flash-freeze in liquid nitrogen and store membrane
fractions at 80
C until use.

3.1.2 96-Well Microplate 1. Prepare 1 mL of 2 stock solutions in 1 PBS, 1 PIC of each

Solubilization detergent or combination of detergents to be tested in a
96-deep well plate (sufficient for 20 tests of 96-well microplate
solubilization) (see Note 2).
2. Dilute the membrane fraction containing the protein of inter-
est at 2 mg/mL in 1 PBS, 1 PIC.
3. Aliquot 50 μL diluted membranes into a 96-well microplate
(see Note 3).
4. Using a multi-channel pipet transfer 50 μL of detergent solu-
tion from the 96-deep well 2 stock solutions plate to the
96-well solubilization microplate containing the membranes.
5. Incubate the 96-well solubilization microplate at 4
C for 2 h
under agitation.
6. Transfer the 100 μL solubilisates into 0.5 mL micro-
ultracentrifuge tubes and ultracentrifuge at 150,000  g for
15 min at 4
C.
7. Transfer the supernatants (soluble fractions) in a 96-well
microplate at 4
C.

3.1.3 Dot-Blot 1. Soak the nitrocellulose membrane in 1 PBS for 10 min at

room temperature.
2. Assemble the Bio-Dot® microfiltration apparatus (Bio-Rad).
3. Lay the membrane on the sealing gasket so that it covers all of
the holes and remove any air bubbles trapped between the
membrane and the sealing gasket (see Note 4).
4. Place the sample template on top of the membrane and finger-
tighten the four screws (see Note 5).
5. Attach a vacuum pump to the flow valve with a waste trap set up
and positioned between the vacuum outlet and the flow valve.
Turn on the vacuum pump (20 kPa pressure) and set the flow
valve to apply vacuum to the apparatus (flow valve position 2).
6. Repeat the four screws tightening process (see Note 6).
7. Set the flow valve to stop vacuum (flow valve position 1) and
turn off the vacuum pump (see Note 7).
 Solubilization and Stabilization of Membrane Proteins 263

8. Apply 100 μL 1 PBS to all 96 wells.

9. Turn on the vacuum pump and set the flow valve to position
2. Gently remove the buffer from the wells. As soon as the
buffer drains from all the wells, set the flow valve to position
1 and turn off the pump.
10. Fill the appropriate wells with 50 μL of supernatant from
solubilization and ensure that there are no air bubbles in the
wells.
11. Allow the samples to filter through the membrane by gravity
for 5 min.
12. Turn on the vacuum pump, set the flow valve to position 2, and
allow complete sample filtration with vacuum.
13. With vacuum applied, loosen the four screws and remove the
sample template.
14. Set the flow valve to position 1, turn off the vacuum pump, and
remove the nitrocellulose membrane.
15. Block the nitrocellulose membrane with 30 mL blocking
buffer and immunodetect the protein of interest using a spe-
cific antibody.
16. Confirm the optimal conditions of solubilization by
SDS-PAGE and Western blot analysis and Western blot–based
thermostability assay.

3.2 Membrane The Western blot–based thermostability assay is adapted from [8].
Protein Stability
1. Solubilize the protein of interest in each detergent condition to
Assessment
be evaluated and in a final volume of 700 μL.
3.2.1 Western Blot– 2. Aliquot 50 μL of solubilisate into 14 PCR microtubes.
Based
3. Using a thermocycler, submit the PCR microtubes to a gradi-
Thermostability Assay
ent of temperature from 25
C to 80
C for 30 min.
4. Immediately transfer to fresh microtubes and centrifuge at
20,000  g for 40 min at 4
C.
5. Analyze the supernatants by SDS-PAGE and Western blot
using an antibody raised against the protein of interest.
6. Quantify the intensity of the immunodetection signal,
corresponding to the protein of interest, plot as function of
temperature, and fit to the Boltzmann equation to determine
the Tm.

3.2.2 Radioligand This protocol is developed for A2A receptor and is adapted from
Binding Assay [9]. A single protein concentration and a single radioligand con-
centration are used to compare the different protein preparations.
This assay is performed in triplicate for each protein sample.
264 Vincent Corvest and Anass Jawhari

1. Prepare the purified membrane proteins at 24 μg/mL in their

own buffer.
2. Using a 96-well microplate with a U-bottom, aliquot 50 μL of
protein sample at 24 μg/mL into two wells (well #1 for the
total signal and well #2 for the nonspecific signal).
3. Add 10 μL of radioligand solution to the well #1 (total signal)
and 10 μL of nonspecific ligand solution to the well #2 (non-
specific signal).
4. Mix by pipetting up and down, and incubate at 4
C for 2 h.
5. Add 60 μL of γ-globulin solution and 120 μL PEG6000 solu-
tion to each well (see Note 8).
6. Mix by pipetting up and down, and incubate at 4
C for
15 min.
7. Using the FilterMate Harvester:
(a) Rinse twice the PEI-coated GF/B filter 96-well micro-
plate with 200 μL of wash buffer.
(b) Transfer the protein samples to the PEI-coated GF/B
filter 96-well microplate.
(c) Wash four times the PEI-coated GF/B filter 96-well
microplate with 200 μL of wash buffer.
(d) Dry the PEI-coated GF/B filter 96-well microplate for
2 min under vacuum.
8. Seal the bottom of PEI-coated GF/B filter 96-well microplate
with the adhesive BackSeal.
9. Add 25 μL of scintillation reagent to each well.
10. Seal the top of PEI-coated GF/B filter 96-well microplate with
the adhesive TopSeal.
11. Incubate the plate for 1 h at room temperature.
12. Top count the CPM for 5 min per well using a microplate
counter. Determine the specific signal by subtracting the non-
specific signal (well #2) from the total signal (well #1).

3.2.3 ATPase This protocol is developed for BmrA transporter and is adapted
Activity Assay from [11].
1. Prepare a range of 20 μL Pi standards at 0 mM, 2 mM, 4 mM,
and 8 mM in 1 TBS from the Pi standard stock solution.
2. Prepare the membrane protein samples at 0.5 mg/mL in their
own buffer.
3. Using a 96-well microplate with a transparent flat-bottom,
prepare 2 series of 10 μL for each Pi standards and protein
samples (series #1 for total signal, series #2 for nonspecific
signal).
 Solubilization and Stabilization of Membrane Proteins 265

4. Add 40 μL of reaction buffer to each well.

5. Add 2 μL of 150 mM ATP to each protein sample well, and
2 μL 1 TBS to each Pi standard well.
6. Add 2 μL of 1 TBS to series #1 well, and 2 μL of hot 6 mM
orthovanadate to series #2 well (see Note 9).
7. Incubate at 37
C for 60 min under gentle agitation.
8. Add 30 μL 10% SDS to stop the reaction.
9. Incubate at room temperature for 10 min under gentle
agitation.
10. Add 180 μL of revelation buffer (freshly prepared).
11. Incubate at 37
C for 20 min under gentle agitation.
12. Measure the absorbance at 620 nm.
13. Calculate for each protein sample the total ATPase activity
(series #1) and the nonspecific ATPase activity (series #2)
using the corresponding standard curve.
14. Determine the specific ATPase activity for each protein sample
by subtracting the nonspecific signal (series #2) from the total
signal (series #1).

3.3 Behavior of 1. Add appropriate volume of 5 native loading buffer to the

Membrane Protein in membrane protein samples leading to 1 final concentration
Solution (see Note 10).
3.3.1 Clear-Native-PAGE 2. Load the sample in a 4–15% Mini-PROTEAN® TGX™ precast
protein gel (Bio-Rad) and run the gel for 70 min at 4
C using
CN-PAGE anode and cathode buffers.
3. Detect the protein by Coomassie staining or Western blot for
an immunodetection (see Note 11).

3.3.2 Glutaraldehyde 1. Prepare a range of glutaraldehyde stock solutions in water: 0%,

Cross-Linking Assay 0.0625%, 0.125%, 0.25%, 0.5%.
2. Aliquot 10 μL of purified membrane protein into five
microtubes.
3. Add 2.5 μL of different glutaraldehyde stock solutions in order
to reach 0%, 0.0125%, 0.025%, 0.05%, and 0.1% final
concentrations.
4. Incubate the samples on ice for 2 h.
5. Stop the crosslinking reaction by adding 3.5 μL of 5
SDS-DTT loading buffer.
6. Analyze the samples by SDS-PAGE and Western blot.
266 Vincent Corvest and Anass Jawhari

3.3.3 Size Exclusion 1. Equilibrate a Superdex™ 200 Increase 5/150 GL column

Chromatography (SEC) (GE Healthcare) with a flow rate of 0.1 mL/min at 4
C with
Analysis 3 column volumes (CV) of eluent buffer corresponding to the
buffer of the membrane protein to be analyzed (see Notes 12
and 13).
2. Inject the membrane protein sample onto the column
(volume  50 μL).
3. Elute the protein at 4
C with 1.5 CV of eluent buffer at
0.1 mL/min.
4. The elution fractions can be analyzed by SDS-PAGE and West-
ern blot to confirm the presence of the membrane protein.

4 Notes

1. As the cell lysate will warm up, due to the high energy mixing
with glass beads, make sure to use an ice-water cooling jacket
with the Bead Beater homogenizer to control the temperature
during the cell disruption. The 2 min cool down step between
two cell lysis cycles is critical to avoid protein denaturation.
2. 2 stock solutions of each detergent or combination of deter-
gents are prepared in 1 PBS, 1 PIC at 20-fold the Critical
Micelle Concentration (CMC) in order to reach a final concen-
tration of 10 CMC in the solubilization tests.
3. Keep one well with no membrane that will serve as negative
control and one well with 2% SDS that will serve as positive
control. An example of 96-well plate solubilization conditions
is described in [7].
4. The membrane should not extend beyond the edge of the
gasket after the Bio-Dot apparatus is assembled.
5. Use a diagonal crossing pattern when tightening the screws to
ensure uniform application of pressure on the membrane
surface.
6. Tightening the four screws while vacuum is applied ensures a
tight sealing, preventing leaking between the wells and con-
tamination between slots.
7. To stop the vacuum, set the flow valve to a position that does
not expose the apparatus to air. That will maintain a slight
depression in the Bio-Dot apparatus preventing leaking
between the wells, while the vacuum pump is off.
8. This step will drastically improve the membrane protein reten-
tion on the GF/B filter.
9. The orthovanadate inhibits ATPase activity, enabling to deter-
mine the nonspecific inorganic phosphate background.
 Solubilization and Stabilization of Membrane Proteins 267

10. The ClearNative gel electrophoresis works without using an

anionic dye. Therefore, this method can be used for the sepa-
ration of membrane proteins prepared with an anionic deter-
gent. For nonionic or zwitterionic detergents, the suitable
method is the BlueNative gel electrophoresis, which uses a
negatively charged Coomassie Brilliant Blue G250 dye that
forms a complex with the protein to enable the gel migration.
11. The choice for the method of protein detection depends on the
amount of protein loaded to the gel. The immunodetection by
Western blot is more sensitive than the Coomassie staining but
requires more steps.
12. For size exclusion chromatography, it is recommended to
degas and filter all solutions through a 0.22 μm filter. However,
some detergents may interact with the filter membrane and be
retained. If in doubt, do not filter the eluent buffer.
13. For size exclusion chromatography, it is recommended to
choose the optimal fractionation range of the column to ensure
better separation between aggregates and non-aggregates for
high molecular proteins.

References

1. Almen MS et al (2009) Mapping the human western blotting. Protein Eng Des Sel 28
membrane proteome: a majority of the human (12):539–542
membrane proteins can be classified according 9. Igonet S et al (2018) Enabling STD-NMR
to function and evolutionary origin. BMC Biol fragment screening using stabilized native
7:50 GPCR: a case study of adenosine receptor. Sci
2. Overington JP, Al-Lazikani B, Hopkins AL Rep 8(1):8142
(2006) How many drug targets are there? Nat 10. Nguyen KA et al (2018) Glycosyl-substituted
Rev Drug Discov 5(12):993–996 Dicarboxylates as detergents for the extraction,
3. Seddon AM, Curnow P, Booth PJ (2004) overstabilization, and crystallization of mem-
Membrane proteins, lipids and detergents: brane proteins. Angew Chem Int Ed Engl 57
not just a soap opera. Biochim Biophys Acta (11):2948–2952
1666(1–2):105–117 11. Matar-Merheb R et al (2011) Structuring
4. Duquesne K, Sturgis JN (2010) Membrane detergents for extracting and stabilizing func-
protein solubilization. Methods Mol Biol tional membrane proteins. PLoS One 6(3):
601:205–217 e18036
5. Eshaghi S (2009) High-throughput expression 12. Hardy D et al (2018) The yin and yang of
and detergent screening of integral membrane solubilization and stabilization for wild-type
proteins. Methods Mol Biol 498:265–271 and full-length membrane protein. Methods
6. Agez M et al (2017) Molecular architecture of 147:118–125
potassium chloride co-transporter KCC2. Sci 13. Desuzinges Mandon E et al (2017) Expression
Rep 7(1):16452 and purification of native and functional influ-
7. Desuzinges Mandon E et al (2017) Novel sys- enza a virus matrix 2 proton selective ion chan-
tematic detergent screening method for mem- nel. Protein Expr Purif 131:42–50
brane proteins solubilization. Anal Biochem 14. Jawhari A et al (2006) Structure and oligo-
517:40–49 meric state of human transcription factor
8. Ashok Y, Nanekar R, Jaakola VP (2015) Defin- TFIIE. EMBO Rep 7(5):500–505
ing thermostability of membrane proteins by
 Part III

Analysis in a Cellular Context

 Chapter 15

Practical Aspects of Super-Resolution Imaging

and Segmentation of Macromolecular Complexes
by dSTORM
Leonid Andronov, Jean-Luc Vonesch, and Bruno P. Klaholz

Abstract
Super-resolution fluorescence microscopy allows imaging macromolecular complexes down to the nano-
scopic scale and thus is a great tool to combine and integrate cellular imaging in the native cellular
environment with structural analysis by X-ray crystallography or high-resolution cryo electron microscopy
or tomography. Here we describe practical aspects of SMLM imaging by dSTORM, from the initial sample
preparation using mounting media, antibodies and fluorescent markers, the experimental setup for data
acquisition including multi-color colocalization and 3D data acquisition, and finally tips and clues on
advanced data processing that includes image reconstruction and data segmentation using 2D or 3D
clustering methods. This approach opens the path toward multi-resolution integration in cellular structural
biology.

Key words Super-resolution microscopy, dSTORM, SMLM, Immunofluorescence, Fluorescence

microscopy

1 Introduction

Fluorescence microscopy is a key technique for the specific obser-

vation of proteins in their native cellular environment. Unfortu-
nately, the resolution of fluorescence microscopy is limited by the
wavelength of light to around 200–300 nm, which limits the
studies to the scale of cellular organelles. The recently emerged
super-resolution fluorescence microscopy allows to overcome this
limit and improves the resolution by about one order of magnitude,
i.e., down to the nanoscopic scale of large macromolecular com-
plexes. Super-resolution microscopy is therefore a powerful tech-
nique that bridges the gap between high-resolution
crystallographic and cryo electron microscopy (cryo-EM) struc-
tures of proteins and their functions in the cellular context [1].
Among the super-resolution techniques, single-molecule local-
ization microscopy (SMLM), which includes techniques such as

Arnaud Poterszman (ed.), Multiprotein Complexes: Methods and Protocols, Methods in Molecular Biology, vol. 2247,
https://doi.org/10.1007/978-1-0716-1126-5_15, © Springer Science+Business Media, LLC, part of Springer Nature 2021

271
272 Leonid Andronov et al.

(direct) stochastic optical reconstruction microscopy ((d)STORM)

[2, 3], photoactivated localization microscopy (PALM) [4], and
point accumulation for imaging in nanoscale topography (PAINT)
[5], has proven to produce the highest resolution even using basic
optical setups and common sample preparation protocols. SMLM
relies on stochastic switching of fluorophores between a fluores-
cently active “on” and a fluorescently inactive “dark” state that can
be controlled by the intensity of the excitation light and composi-
tion of the sample mounting medium. Using appropriate buffer
and imaging conditions, many different fluorophores can be made
suitable for SMLM, including organic fluorophores [6] and fluo-
rescent proteins [7]. SMLM not only provides super-resolution
images but also gives access to properties of individual molecules,
allowing cluster analysis [8], segmentation [9], colocalization esti-
mation [10], molecular counting [11], and probing of stoichiome-
try of macromolecular complexes [12].
In this chapter, we describe how to perform a typical dSTORM
experiment from sample preparation to image acquisition and data
processing, which includes event list generation, image reconstruc-
tion and post-processing of SMLM data that can eventually be
analyzed and segmented by 2D and 3D clustering methods.

2 Materials

2.1 Cell Culture In addition to your favorite adherent cell line, you need:
and Immunolabeling
1. Glass-bottom petri dishes with a diameter of 35 mm for cell
culture (CELLView, Greiner Bio-One).
2. Phosphate-buffered saline (PBS), 1 and 10 times concentrated
(PBS 1
and 10
).
3. Paraformaldehyde (PFA) 4% solution in PBS.
4. 0.1% Triton X-100 in PBS (PBS/Tx).
5. Primary antibody.
6. Secondary antibody coupled to compatible fluorophores
(if primary antibody was not directly labeled).
7. Bovine serum albumin (BSA).
8. Normal goat serum (NGS).
9. Fetal bovine serum (FBS).
Compatible fluorophores:
Alexa Fluor family (Thermo Fisher Scientific): Alexa
647 (the most commonly used fluorophore for SMLM),
Alexa 555, Alexa 532, Alexa 488; Atto 488 (ATTO-TEC
GmbH); Cy3B [13] and others [6].
 Practical Aspects of Super-Resolution Imaging and Segmentation of. . . 273

2.2 Mounting Media: Gloxy [6]: Cysteamine hydrochloride 50 mM, glucose 10% w/v,
Composition glucose oxidase 0.5 mg/ml, catalase 40 μg/ml in TN buffer
and Choice (50 mM Tris (pH 8.0) and 10 mM NaCl).
We use the following stock solutions:
1. Cysteamine hydrochloride (Sigma-Aldrich M6500, 1 M in
H2O, stored at 20  C).
2. Glucose 400 g/l, stored at 20  C.
3. Glucose Oxidase (Sigma-Aldrich G2133, 5 mg/ml, stored at
4  C).
4. Catalase (Sigma-Aldrich C3515, 4 mg/ml, stored at 4  C).
Vectashield/TDE [14]: Vectashield 20% v/v, TDE 70% v/v in
PBS.
1. Vectashield® Antifade Mounting Medium (Vectorlabs
H-1000, stored at +4  C).
2. 2,20 -Thiodiethanol (TDE, Sigma-Aldrich 166782).
Mix 20% v/v Vectashield with 70% v/v TDE and 10% v/v
PBS 10
. For mounting, incubate in solutions with increasing
concentrations of TDE in PBS: 10%, 25%, 50%, 10 min each.
Then replace with the final solution, Vectashield/TDE.
OxEA [15]: Cysteamine hydrochloride 50 mM, Oxy-
Fluor™ 3% v/v, sodium DL-lactate 20% v/v in PBS adjusted
to pH 8–8.5 with 1 M aqueous NaOH.
We use the following stock solutions:
1. Cysteamine hydrochloride (Sigma-Aldrich M6500, 1 M in
PBS, stored at 20  C).
2. OxyFluor™ (Sigma-Aldrich SAE0059, stored at 20  C) [16].
3. Sodium DL-lactate (Sigma-Aldrich L1375, 60% w/w syrup,
stored at +4  C).

2.3 Instrumentation The super-resolution experiments were performed on a Leica SR

GSD system built on the base of the DMI6000 B inverted wide-
field microscope, consisting of the HCX PL APO 100
/1.47 Oil
CORR TIRF PIFOC objective and 1.6
magnification lens that
provide an equivalent pixel size of 100 nm on an EMCCD camera
(Andor iXon3 DU-897U-CS0-#BV); continuous wave fiber lasers
(MPBC Inc., 488 nm 300 mW, 532 nm 1000 mW, 642 nm
500 mW), a diode laser (405 nm 30 mW) and the suppressed
motion (SuMo) sample stage.
274 Leonid Andronov et al.

3 Methods

3.1 Specimen Procedure (adapted from a Leica protocol):

Preparation
1. Seed cells in glass-bottom Petri dishes d ¼ 35 mm (CELLView,
Greiner Bio-One) at a density of ~100 cells/mm2.
2. Aspirate the cell culture medium.
3. Wash briefly 1
with PBS.
4. Fix with 4% PFA in PBS for 20 min.
5. Permeabilize with PBS/Tx twice for 10 min.
6. Incubate with primary antibody overnight at 4  C (see Notes 1
and 2).
7. Wash 3
with PBS/Tx over 2 h each time.
8. Incubate with secondary antibody for 2 h at RT.
9. Wash 3
with PBS/Tx over 2 h each time.
10. Postfix with 4% PFA in PBS for 10 min (see Note 3).
11. Wash briefly 3
with PBS.
12. Store at 4  C in PBS until mounting.
13. Before imaging, replace PBS with an appropriate mounting
medium.
The mounting medium for dSTORM (also known as the imag-
ing buffer) has to be chosen based on the following criteria:
1. Fluorophores used. Alexa-647 works best in Gloxy or
Vectashield-containing media, while OxEA allows to use a
wider range of dyes, e.g., for multi-color imaging.
Vectashield-containing media are usually unsuitable for the
detection of light with wavelengths ≲550 nm and for excita-
tion with wavelengths ≲500 nm due to autofluorescence.
2. Depth of the region of interest (ROI) inside the sample. A ROI
close to the coverslip (<200 nm) can be imaged with total
internal reflection fluorescence microscopy (TIRF) which
allows to reduce background fluorescence. This needs a
mounting medium with a refractive index significantly lower
than that of glass. Therefore, usually water-based media are
used (n  1.33), e.g., Gloxy or OxEA. These media, however,
are less suitable for deeper (more than several μm from the
coverslip) imaging because of spherical aberrations due to the
refraction index mismatch. For such situations, buffers with a
refractive index close to that of glass (n  1.52) are better
suited, e.g., Vectashield/TDE. Using highly inclined and
laminated optical sheet (HiLo) illumination, the sample is
excited with a light beam under an angle smaller than the
critical angle for the given coverglass/specimen interface that
 Practical Aspects of Super-Resolution Imaging and Segmentation of. . . 275

allows to image deeper in the sample as compared to TIRF

microscopy but allows to keep low background intensity
[17]. HiLo illumination can be used with any mounting
medium.
3. Labeling method. Cysteamine- or β-mercaptoethanol-contain-
ing buffers may cleave antibodies which can decrease the spe-
cific labeling and increase background, while TDE-containing
media destabilize phalloidin labeling [18].
4. Possibility to refresh the buffer. Gloxy buffer needs to be
replaced each 2–4 h, OxEA—each 6–12 h, while Vectashield-
based buffers do not need to be replaced.

3.2 Basic SMLM Workflow of acquisition processing:

Experiment A typical acquisition session is composed of the following steps.
1. Preview. It consists of the search of the ROI and focusing using
conventional epifluorescence or TIRF/HiLo illumination with
low laser power (1–5% of maximal power). In the case of the
TIRF/HiLo mode the best direction (azimuth) of laser illumi-
nation to a sample has to be chosen as well. The step is com-
pleted by capturing a wide-field image (Fig. 1a). It is important
that the dSTORM experiment will be carried out under the
same conditions (x, y, z position, epi/TIRF, direction of laser
illumination if TIRF/HiLo) to have a direct comparison with
the classical (epifluorescence) microscopy image and avoid arti-
facts in the super-resolution image.
2. Pumping. Illumination of the sample with high intensity light
in order to bring most fluorescent molecules into the dark
state. At the very beginning, when all the molecules are in the
bright state, due to very strong excitation, the fluorescence is
very intense because all the dyes emit at the same time. After a
little time (in the order of seconds, depending on kinetics of the
fluorophore and excitation intensity), with occupation of the
dark states and depletion of the on state, the intensity of
fluorescence decreases because progressively less molecules
stay in the on state until the number of shining molecules
becomes less than one per diffraction-limited region. At this
time one begins to see the light from each “on” molecule
separately, and it is the time to start the acquisition, the next
step of the experiment.
3. Acquisition. With constant laser intensity, which may be as high
as during pumping or lower, one gets a number of shining
single-molecule events (“blinks”) per frame with certain expo-
sure time (Fig. 1b). This number is progressively dropping with
photobleaching of the molecules, but it can be increased using
“backpumping” (e.g., with a 405-nm laser). With gradual
increase of backpumping laser intensity, it is possible to keep
276 Leonid Andronov et al.

Fig. 1 Example of a typical dSTORM data acquisition cycle (see methods). HeLa cells, β-tubulin marked by
Alexa Fluor-647-conjugated secondary antibodies and mounted in the “Vectashield/TDE” medium. (a) Preview
in HiLo mode. (b) One of the frames of the acquisition (exposure time is 50 ms/frame). (c) Circles indicate the
spots that are selected by the Leica LAS AF software as single-molecule localizations. (d) Final reconstructed
super-resolution image using 17,622 frames with 1,160,875 localization events (collected here over 15 min);
the image is in the histogram mode with 20 nm/pixel, corrected for drift and rendered in SharpViSu [29]. Scale
bars, 2 μm

the number of events per frame at a nearly constant value, until

most of the molecules become bleached and the acquisition is
finished.
4. Post-processing. Although real-time super-resolution image
reconstruction is possible depending on the installed software,
a post-processing step is usually required to improve the result
 Practical Aspects of Super-Resolution Imaging and Segmentation of. . . 277

by optimizing the processing parameters. For example, to

reduce the noise level, it is necessary to increase the threshold
of single-molecule detection, if it was quite low during the
experiment. Vice versa, if the threshold was pretty high, it is
desirable to reduce it in order to get more events and obtain a
better reconstructed image. The frames with undesirable arti-
facts (non-blinking areas, especially in the very first frames)
should be truncated, thus also improving the results.

3.3 Experimental Numerous experimental parameters that can be changed by the

Parameters of dSTORM user are summarized here.
Data Collection
1. Laser power, can be chosen differently for pumping, acquisi-
and Processing tion, and backpumping. Strong excitation intensity during
acquisition increases the speed of “blinking,” reduces the num-
ber of “on” molecules, and ensures faster completion of the
acquisition. However, too strong excitation can also lead to
increase of the background brightness and reduction of the
number of the localized molecules. This parameter has to be
adjusted for each combination of fluorophore-imaging buffer.
Backpumping intensity is normally set to zero for pumping and
for the beginning of acquisition and is turned on with gradually
increasing intensity when the number of events per frame
drops.
2. Frame exposure time. Ideally, it has to be equal to the average
on-time of the fluorophores. If the exposure time is too short,
molecules will appear on many consecutive frames, their signal
to noise ratio (SNR) will be reduced. On the other side, if the
exposure time is too long, the density of localizations per frame
will be higher and the SNR can also be reduced because the
background is acquired during the whole frame exposure time
while the signal from the fluorophores is detected only during
their on-time.
3. Illumination mode: epi-/HiLo/TIRF, azimuth of laser illumi-
nation (when available), affects the penetration depth and the
axial resolution. It is advised to make pumping in
epi-illumination mode to send higher energy to the specimen
and move all the volume of the sample to the dark state; this
provides lower background during the acquisition.
4. EM camera gain for EMCCD detectors. Normally it is set up to
maximum, but to be diminished in case of saturation due to
strong fluorescence and/or high exposure times (i.e., adjust
and calibrate the dynamic range of the camera).
5. Threshold of single-molecular detection in photons/pixel.
Affects the real-time reconstructed super-resolution image
(if available); it can be changed in the post-processing step.
Usually lower threshold increases the noise level, a high
278 Leonid Andronov et al.

threshold reduces the number of events, so a compromise value

should be found. It is convenient to follow which spots on the
images are detected and adjust the threshold in a way that only
the single-molecule spots are detected (Fig. 1c).
6. Pixel size for real-time reconstruction (Fig. 1d), if available.
Commonly set to around 20 nm, and can be reduced for post-
processing (e.g., 5 or 10 nm sampling).

3.4 3D SMLM One of the easiest and most common ways for 3D SMLM imaging
Experiment is realized by the modification of the point spread function (PSF) of
the microscope with an astigmatic aberration.
Calibration. The astigmatic PSF deformation as a function of
the axial position of fluorophores should be calibrated by imaging
fluorescent beads using several known axial positions of the
objective.
1. Install a sample with fluorescent beads with subdiffraction size
in the microscope. For example, multi-color TetraSpeck™
Microspheres, 0.1 or 0.2 μm, can be used.
2. Put in place the cylindrical lens or activate astigmatism in the
adaptive optics module.
3. Focus the microscope in a way that the images of the beads are
closest to circular symmetry.
4. Adjust the imaging parameters (excitation intensity, exposure
time, and electron-multiplying gain of the camera) to get good
SNR. It is preferable to increase first the excitation intensity in
order to keep the exposure time minimal.
5. Image the beads as a Z-stack with a step of 50 nm around
0.6–1 μm of the focus point determined in step 3.
6. Repeat the procedure (steps 1–5) for each spectral channel.
7. Focus the microscope on the beads in one of the spectral
channels. Image the beads through all the channels in this
Z-position. It is necessary to take into account axial chromatic
aberration.
8. Fit all the acquired images with a software for single-molecule
detection, which will be used for the following experiments.
Refer to the manual of the software.
Once the calibration is done, 3D SMLM experiments can
be performed with the conventional SMLM procedure, with
the difference that the maximal density of switched-on mole-
cules on a frame should be reduced to avoid overlap of
stretched PSF spots.

3.5 SMLM Data The first step in SMLM data processing is the determination of the
Processing localization of each individual fluorophore to provide a full event
list from which a super-resolution image can be reconstructed. The
Practical Aspects of Super-Resolution Imaging and Segmentation of. . . 279

single-molecule localization precision and therefore the best attain-

able resolution can be determined by the following formula [19]:

σ2 a2 8πσ 4 b 2
Δx 2 ¼ þ þ 2 2 , ð1Þ
N 12N a N
where Δx is the standard deviation (SD) of the localization error, σ
is the SD of the PSF, N is the number of detected photons, a is the
pixel size, and b is the SD of the background noise. For best
localization precision, it is therefore important to keep a high
number of detected photons from individual fluorophores, low
background noise, small PSF, and a sufficiently small camera pixel
size (usually around 100 nm).
Acquired images have to be processed for the detection of
single molecules (could be done in real time during experiments
depending on microscopy system). This will produce a list of loca-
lizations, from which a super-resolution image can be built or other
information can be extracted using advanced data processing meth-
ods. Most dedicated super-resolution microscopes are supplied
with a software for single-molecule detection, image reconstruc-
tion, and basic post-processing of localization data. In practice,
however, specialized software packages are needed for advanced
data processing.
Single-molecule detection requires first of all a good choice of
the image fitting method. The most popular methods are:
1. Center of mass assessment—the fastest method, but produces
good results only in conditions of low density of localizations
and low homogeneous background. This is a good method for
the initial processing of the data (e.g. during acquisition).
2. Least-square fitting—usually intermediate in speed group of
methods and provides best results when the models for the
PSF or for the noise are hard to determine. Some of these
methods can be used for the localization of overlapping
fluorophores [20].
3. Maximum-likelihood estimation—usually the slowest family of
algorithms, requires a model of the PSF and of the noise. Gives
best results, especially for weak signals, provided that the PSF
shape and the noise model are correctly set [21]. Some of these
methods can be used for localization of overlapping
fluorophores [22].
Even though algorithms for single-molecule detection are
often supplied by the manufacturer of the super-resolution
system, the available tools are often not sufficient, especially
for difficult imaging conditions, such as overlapping localiza-
tions, non-homogeneous background, weak signal, etc. For
such conditions, external software packages such as Thunder-
STORM [23], B-recs [24], SimpleSTORM [25], and others
(reviewed in [26]) can be recommended.
280 Leonid Andronov et al.

3.6 Super-Resolution Unlike in conventional microscopy, the image in SMLM cannot be

Image Reconstruction acquired in a direct optical way, but should be calculated from the
list of localizations, obtained after fitting. Therefore, there are
several methods to calculate an image from SMLM data:
1. Histogram mode. The gray value of a super-resolution pixel
equals to the number of events detected within the pixel’s area.
The fluorescence intensity (number of photons) is not taken
into account. The pixel size chosen by the user strongly affects
the gray levels of the image: in case of too small pixel sizes, the
image would contain only “zeros” and “ones,” i.e., become
binarized. This mode is fast to calculate, but it can produce
noisy and pixilated images, especially for weak density of
localizations.
2. “Gaussian” mode. Every localization is represented as a Gauss-
ian function with the width equivalent to the localization pre-
cision of the fluorophore, taking into account the number of
detected photons. This image is slower to calculate than the
histogram image, but it provides a smooth representation and
allows to decrease pixel size without risk of pixilation. How-
ever, this mode can reduce the resolution of the image [27].
3. Local density mode. The pixel’s gray values are calculated to be
proportional to the local density of fluorophores in the neigh-
borhood of the pixel. The local density can be estimated, e.g.,
using Voronoi diagrams [9] or Delaunay triangulations
[27]. Even though this method is slower than the others (but
still reasonable as in the order of minutes), it provides smooth
images with preserved resolution even for weak signals.

3.7 Post-processing Post-processing of localization data includes methods such as cor-

of Localization Data rection of drift and chromatic aberrations that are essential for most
of experiments and advanced processing methods, such as cluster-
ing and colocalization analysis that are necessary for some experi-
ments, depending on biological question and sample type or
acquisition setup (e.g., multi-color imaging or 3D data
acquisition).
For drift correction, there are two most feasible possibilities:
(1) correction using fiducial markers and (2) correction with cross-
correlation. For the first approach, photostable fiducial markers,
such as fluorescent beads, gold nanoparticles, or quantum dots
[28], need to be incorporated and fixed within the sample, usually
on the surface of the coverslip. The second approach does not need
fiducials and uses properties that are intrinsic to localization data
and rely only on computing [29]. The cross-correlation method
works well for contrasted structures with well-defined shape but
can be less efficient for diffuse structures. It can also be recom-
mended for imaging far from the coverslip because the fiducials
usually can be robustly immobilized only on a glass surface.
 Practical Aspects of Super-Resolution Imaging and Segmentation of. . . 281

For multi-color imaging, chromatic aberrations of the objective

can cause a shift between images, captured with different wave-
lengths. To correct for this, first, the aberrations should be cali-
brated by imaging a multi-color fiducial marker (e.g., beads) with
the spectral channels to be used for the experiments. The shift
detected from the fiducials should be subtracted from the single-
molecule coordinates [30].
If the on-time of fluorophores is higher than the frame expo-
sure time, their images will appear on several consecutive frames.
These localizations can be combined by searching localizations
within a circle of a given radius (usually around 50 nm) around
every localization in following frames. Using this option, it is
possible to increase the number of photons per localization and
therefore the localization precision. An inconvenience is that loca-
lizations of different fluorophores can be combined by error if they
appear on consecutive frame (this is less likely when the densities of
switched-on fluorophores is low). In some cases, it can be also
useful to remove localizations that appear on too many consecutive
frames, eliminating non-blinking regions that can produce artifacts.
Localization event lists can be conveniently analyzed with the
help of histograms as a function of the number of photons per
localization (Fig. 2) and the number of localizations per frame
(Fig. 3). Localizations with too low photon counts can be removed,
thus improving average localization precision. Frames with too
high number of localizations that correspond to very dense

Fig. 2 Typical histogram of the photon count of localizations of the Alexa Fluor-647 dye. The exposure time
was 50 ms/frame, the detection threshold is 60 photons/pixel, 17,622 frames were analyzed producing
1,160,875 events, the average photon count is 2675 photons/event. The analysis was done in the Leica LAS
AF software
282 Leonid Andronov et al.

Fig. 3 Typical histogram of the number of localizations per frame for the Alexa Fluor-647 dye. The curve is
built using blocks of 100 frames each for smoothing reasons. Note increases in the number of localizations
(middle part of the graph) due to an increase in the intensity of the “backpumping” with a 405-nm laser.
Exposure time is 50 ms/frame, detection threshold is 80 photons/pixel; 18,620 frames with 1,161,303 events
were analyzed. The analysis was done in the Leica LAS AF software

localizations that may compromise single-molecule detection can

be removed as well. An additional problem for data quantification
can be relocalization events (fluorophores that appear more than
once during the acquisition); these can be addressed by taking into
account the blinking behavior of fluorophores [31].

3.8 SMLM Data Advanced SMLM data processing techniques include

Segmentation co-localization and cluster analysis. While for classical microscopy
and Cluster Analysis, data, these are commonly done on images, analysis of SMLM
from 2D to 3D images is less optimal, because the image is not the primary data
type, and moreover, they are not built in a unique way (see Sub-
heading 3.6). To get most information from the data, it is preferred
to process directly the localization coordinates.
Cluster analysis or segmentation aims for the detection of
accumulations (clusters) of fluorescently labeled objects, such as
proteins or nucleic acids. These detected clusters should then be
analyzed for their size, density distribution, etc. Two types of
cluster analysis can be distinguished: determination of global prop-
erties of data and detailed analysis of individual clusters. For the first
part of the analysis, the method of choice is Ripley’s K- and
L-functions [32, 33]. It allows to, first, test whether or not the
experimental distribution follows a given pattern, and, second, to
find the characteristic size of clusters [8]. A similar method was also
developed for the estimation of colocalization between two experi-
mental distributions [10], it provides a colocalization value for a
region that reflects the global colocalization extent within this
region.
 Practical Aspects of Super-Resolution Imaging and Segmentation of. . . 283

For detailed cluster analysis, the Density-Based Spatial Cluster-

ing of Applications with Noise (DBSCAN) method [34] can be
used. However, to get optimal results, it needs optimization of its
parameters that can introduce a bias in the results. Recently
emerged Voronoi tessellation-based methods [9, 35–37] allow
not only for an unambiguous local density estimation and visuali-
zation of SMLM data but also for testing of datasets for clustering
via comparison of the Voronoi polygons built on the experimental
data with that of the reference distribution of points and for auto-
mated unbiased segmentation without parametrization [9, 37].
Some of these processing aspects can be found in packages for
microscope control, but in many cases, external tools have to be
used. There are several packages dedicated for processing of locali-
zation data, namely ThunderStorm [23] that includes post-
processing aspects such as image reconstructions in several modes,
filtering and combination of localizations, drift correction, and
estimation of colocalization. ViSP [38] is a convenient tool for
3D visualization and basic segmentation. Lama [39] provides
tools for quality control, resolution estimation, cluster analysis
with Ripley’s functions, DBSCAN and Ordering Points To Identify
the Clustering Structure (OPTICS) [40] algorithms, coordinate-
based colocalization, and estimation of stoichiometry of molecular
complexes [12]. MIiSR [41] is focused on colocalization estima-
tion and cluster analysis using the Ripley’s, DBSCAN, and OPTICS
methods. PALMsiever [42] allows for the combination of localiza-
tions, drift correction, DBSCAN cluster analysis, and different
rendering possibilities. SharpViSu [29] offers comprehensive tools
for the correction of chromatic aberrations and drift, selection and
filtering of localizations, different visualization modes, resolution
estimation and, through ClusterViSu [9], cluster analysis with the
Ripley’s function method and Voronoi-based automated segmen-
tation. Our latest development, 3DClusterViSu, allows to perform
Voronoi-diagram-based segmentation of 3D data [37]. Clearly, if it
can be done for a given experiment, 3D analysis becomes the
method of choice nowadays as compared to 2D analysis because it
avoids artefacts from superposed structures (see details in Supple-
mentary Material in [37]), and it provides direct 3D information
about the molecular organization in the cell, thus also facilitating
the integration with molecular and atomic structures and opening
the path toward cellular structural biology [1].

4 Notes

1. Absence of any blocking reagents and blocking step is possible

with highly specific antibodies. If necessary, BSA 1% or a mix-
ture of decomplemented NGS (5%) and FCS (1%) can be used
before step 6.
284 Leonid Andronov et al.

2. To avoid aggregates, all stock solutions of antibodies should be

centrifuged at 16,000
g for 5 min. All working solutions can
be filtered on 0.22 μm filters.
3. The purpose of the post-fixation of the immune complexes is
twofold: first, avoiding cleavage of disulfide bounds of the
antibodies (destroying the immunoreactivity) in the mounting
buffers that often contain reducing agents like cysteamine or
β-mercaptoethanol. Moreover, cysteamine solutions are also
weakly basic (pH 8–8.5), weakening the bindings. Second,
reducing the mobility of the antibodies on the antigens (off
and on reactions) that may affect the quality of super-resolution
images.

Acknowledgments

We thank Yves Lutz from the Imaging Centre for discussions on the
adaptation of a Leica protocol and his contribution during earlier
published work. This work was supported by CNRS, Association
pour la Recherche sur le Cancer (ARC), Institut National du Can-
cer (INCa), Ligue nationale contre le cancer (Ligue), Agence
National pour la Recherche (ANR), and the USIAS research fel-
lowship program of the University of Strasbourg. The super-
resolution microscope setup was supported by the Alsace Region
and by the French Infrastructure for Integrated Structural Biology
(FRISBI; ANR-10-INSB-05-01) and Instruct-ERIC.

References
1. Orlov I, Myasnikov AG, Andronov L et al performance in localization-based super-reso-
(2017) The integrative role of cryo electron lution imaging. Nat Methods 8:1027–1036
microscopy in molecular and cellular structural 7. Lippincott-Schwartz J, Patterson GH (2009)
biology. Biol Cell 109:81–93 Photoactivatable fluorescent proteins for
2. Rust MJ, Bates M, Zhuang X (2006) Stochastic diffraction-limited and super-resolution imag-
optical reconstruction microscopy (STORM) ing. Trends Cell Biol 19:555–565
provides sub-diffraction-limit image resolu- 8. Owen DM, Rentero C, Rossy J et al (2010)
tion. Nat Methods 3:793–795 PALM imaging and cluster analysis of protein
3. van de Linde S, Löschberger A, Klein T et al heterogeneity at the cell surface. J Biopho-
(2011) Direct stochastic optical reconstruction tonics 3:446–454
microscopy with standard fluorescent probes. 9. Andronov L, Orlov I, Lutz Y et al (2016) Clus-
Nat Protoc 6:991–1009 terViSu, a method for clustering of protein
4. Betzig E, Patterson GH, Sougrat R et al (2006) complexes by Voronoi tessellation in super-
Imaging intracellular fluorescent proteins at resolution microscopy. Sci Rep 6:24084
nanometer resolution. Science 10. Malkusch S, Endesfelder U, Mondry J et al
313:1642–1645 (2012) Coordinate-based colocalization analy-
5. Sharonov A, Hochstrasser RM (2006) Wide- sis of single-molecule localization microscopy
field subdiffraction imaging by accumulated data. Histochem Cell Biol 137:1–10
binding of diffusing probes. Proc Natl Acad 11. Lee S-H, Shin JY, Lee A et al (2012) Counting
Sci U S A 103:18911–18916 single photoactivatable fluorescent molecules
6. Dempsey GT, Vaughan JC, Chen KH et al by photoactivated localization microscopy
(2011) Evaluation of fluorophores for optimal
 Practical Aspects of Super-Resolution Imaging and Segmentation of. . . 285

(PALM). Proc Natl Acad Sci U S A 26. Sage D, Kirshner H, Pengo T et al (2015)
109:17436–17441 Quantitative evaluation of software packages
12. Fricke F, Beaudouin J, Eils R et al (2015) One, for single-molecule localization microscopy.
two or three? Probing the stoichiometry of Nat Methods 12:717–724
membrane proteins by single-molecule locali- 27. Baddeley D, Cannell MB, Soeller C (2010)
zation microscopy. Sci Rep 5:14072 Visualization of localization microscopy data.
13. Cooper M, Ebner A, Briggs M et al (2004) Microsc Microanal 16:64–72
Cy3B™: improving the performance of cya- 28. Glushonkov O, Réal E, Boutant E et al (2018)
nine dyes. J Fluoresc 14:145–150 Optimized protocol for combined PALM-
14. Olivier N, Keller D, Rajan VS et al (2013) dSTORM imaging. Sci Rep 8:8749
Simple buffers for 3D STORM microscopy. 29. Andronov L, Lutz Y, Vonesch J-L et al (2016)
Biomed Opt Express 4:885–899 SharpViSu: integrated analysis and segmenta-
15. Nahidiazar L, Agronskaia AV, Broertjes J et al tion of super-resolution microscopy data. Bio-
(2016) Optimizing imaging conditions for informatics 32:2239–2241
demanding multi-color super resolution locali- 30. Erdelyi M, Rees E, Metcalf D et al (2013)
zation microscopy. PLoS One 11:e0158884 Correcting chromatic offset in multicolor
16. Adler H, Spady G (1997) The use of microbial super-resolution localization microscopy. Opt
membranes to achieve anaerobiosis. J Rapid Express 21:10978–10988
Methods Autom Microbiol 5:1–12 31. Annibale P, Vanni S, Scarselli M et al (2011)
17. Tokunaga M, Imamoto N, Sakata-Sogawa K Quantitative photo activated localization
(2008) Highly inclined thin illumination microscopy: unraveling the effects of Photo-
enables clear single-molecule imaging in cells. blinking. PLoS One 6:e22678
Nat Methods 5:159–161 32. Ripley BD (1976) The second-order analysis of
18. Staudt T, Lang MC, Medda R et al (2007) stationary point processes. J Appl Probab
2,20 -thiodiethanol: a new water soluble mount- 13:255–266
ing medium for high resolution optical micros- 33. Nicovich PR, Owen DM, Gaus K (2017) Turn-
copy. Microsc Res Tech 70:1–9 ing single-molecule localization microscopy
19. Thompson RE, Larson DR, Webb WW (2002) into a quantitative bioanalytical tool. Nat Pro-
Precise nanometer localization analysis for indi- toc 12:453–460
vidual fluorescent probes. Biophys J 34. Ester M, Kriegel H-P, Sander J, Xu X (1996) A
82:2775–2783 density-based algorithm for discovering clus-
20. Holden SJ, Uphoff S, Kapanidis AN (2011) ters in large spatial databases with noise.
DAOSTORM: an algorithm for high- density KDD-96 Proceedings 226–231 (http://www.
super-resolution microscopy. Nat Methods aaai.org)
8:279–280 35. Voronoi G (1908) Nouvelles applications des
21. Abraham AV, Ram S, Chao J et al (2009) paramètres continus à la théorie des formes
Quantitative study of single molecule location quadratiques. J Reine Angew Math
estimation techniques. Opt Express 1908:97–102
17:23352–23373 36. Levet F, Hosy E, Kechkar A et al (2015)
22. Quan T, Zhu H, Liu X et al (2011) High- SR-Tesseler: a method to segment and quantify
density localization of active molecules using localization-based super-resolution microscopy
structured sparse model and Bayesian informa- data. Nat Methods 12:1065–1071
tion criterion. Opt Express 19:16963–16974 37. Andronov L, Michalon J, Ouararhni K et al
23. Ovesný M, Křı́žek P, Borkovec J et al (2014) (2018) 3DClusterViSu: 3D clustering analysis
ThunderSTORM: a comprehensive ImageJ of super-resolution microscopy data by 3D
plug-in for PALM and STORM data analysis Voronoi tessellations. Bioinformatics 34
and super-resolution imaging. Bioinformatics (17):3004–3012
30:2389–2390 38. Beheiry ME, Dahan M (2013) ViSP: represent-
24. Rouault H (2013) Brecs: reconstruction of ing single-particle localizations in three dimen-
dense localization microscopy datasets sions. Nat Methods 10:689–690
https://github.com/hrouault/Brecs 39. Malkusch S, Heilemann M (2016) Extracting
25. Köthe U, Herrmannsdörfer F, Kats I et al quantitative information from single-molecule
(2014) SimpleSTORM: a fast, self-calibrating super-resolution imaging data with LAMA—
reconstruction algorithm for localization LocAlization microscopy analyzer. Sci Rep
microscopy. Histochem Cell Biol 6:34486
141:613–627
286 Leonid Andronov et al.

40. Ankerst M, Breunig MM, Kriegel H-P et al resolution imaging enables the analysis of pro-
(1999) OPTICS: ordering points to identify tein interactions, dynamics and formation of
the clustering structure. In: Proceedings of multi-protein structures. PLoS Comput Biol
the 1999 ACM SIGMOD international confer- 11:e1004634
ence on Management of Data. ACM, 42. Pengo T, Holden SJ, Manley S (2015)
New York, NY, USA, pp 49–60 PALMsiever: a tool to turn raw data into results
41. Caetano FA, Dirk BS, Tam JHK et al (2015) for single-molecule localization microscopy.
MIiSR: molecular interactions in super- Bioinforma Oxf Engl 31:797–798
 Chapter 16

Multicolor FRET-FLIM Microscopy to Analyze Multiprotein

Interactions in Live Cells
Abdullah Ahmed, Jennifer Schoberer, Emily Cooke,
and Stanley W. Botchway

Abstract
The need to describe and understand signaling pathways in live cell is seen as a primary route to identifying
and developing targeted medicines. Signaling cascade is also seen as a complex communication and involves
interactions between multiple interconnecting proteins. Where subcellularly and how different proteins
interact need to be preserved during investigation. Furthermore, these complex events occurring simulta-
neously may lead to a single or multiple end point or cell function such as protein synthesis, cell cytoskele-
ton formation, DNA damage repair, or autophagy. There is therefore a need of real-time noninvasive
methods for protein assays to enable direct visualization of the interactions in their natural environment and
hence overcome the limitations of methods that rely on invasive cell disruption techniques. Förster
resonance energy transfer (FRET) coupled with fluorescence lifetime imaging microscopy (FLIM) is an
advanced imaging method to observe protein–protein interactions at nanometer scale inside single living
cells in real-time. Here we describe the development and use of two-channel pulsed interleave excitation
(PIE) for multiple protein interactions in the mTORC1 pathway. The proteins were first tagged with
multiple color fluorescent protein derivatives. The FRET-FLIM combination means that the information
gained from using standard steady-state FRET between interacting proteins is considerably improved by
monitoring changes in the excited-state lifetime of the donor fluorophore where its quenching in the
presence of the acceptor is evidence for a direct physical interaction.

Key words mTOR, Enzymes, Protein complex, Fluorescence imaging, Confocal microscopy,
Excited-state lifetime, TCSPC, FRET, FLIM GFP-tag, DNA, Three-color

1 Introduction

All cellular processes are driven by the action of proteins, from ion
channels across phospholipid membranes to DNA synthesis and cell
growth. For example, the mechanistic or mammalian target of
rapamycin (mTOR) pathway functions in the coordination of
energy, nutrients, and growth factor availability to regulate key
biological processes including cellular growth, metabolism, and
protein synthesis through the phosphorylation of downstream sub-
strates, ribosomal protein, S6 Kinase 1 (S6K1), and 4E-binding

Arnaud Poterszman (ed.), Multiprotein Complexes: Methods and Protocols, Methods in Molecular Biology, vol. 2247,
https://doi.org/10.1007/978-1-0716-1126-5_16, © Springer Science+Business Media, LLC, part of Springer Nature 2021

287
288 Abdullah Ahmed et al.

protein 1 (4E-BP1/eIF4E) [1–4]. The phosphorylation of these

effectors is to regulate cell growth, aging and adiposity [5], mem-
ory [6], immunity [7], and muscle hypertrophy [8]. The human
mTOR protein exists as a multi-protein complex with Rheb, raptor,
mLST8, PRAS40, and DEPTOR proteins termed mTOR Complex
1 (mTORC1). This complex is the main rapamycin target via the
FKBP12 protein. Furthermore, rapamycin differentially inhibits
S6Ks and 4E-BP1 toward mRNA translation [9]. It has also been
observed that the concentration of rapamycin required for an effect
varies significantly between cell lines with a range of nM-mM
reported [10]. Majority of our understanding of the mTOR signal-
ing pathway and protein interaction has come from classical cell
disruption pull-down assays. Hence, information where within the
cell these assemblies are localized that lead to subsequent down-
stream targets phosphorylation in real time is lost. Cell immunoflu-
orescence staining following cell fixation also has several drawbacks
and limitations such as mislocalization of proteins after fixation,
poor antibody sensitivity and specificity in some cases, as well as
antibodies not reaching their target of interest.
Förster or fluorescence resonance energy transfer (FRET) is an
excellent and powerful technique to determine relevant distances in
biological processes under physiological conditions [11]. It relies
on the non-radiative energy transfer from an excited fluorescent
donor molecule to an acceptor molecule with non-excited fluores-
cence in its close vicinity through a dipole–dipole interaction and
may be used to investigate physical interactions between two or
more small or macro-molecules. Thus molecular scale distances
(1–10 nm) can be measured using energy transfer processes. The
very short distances required for this process to occur (<10 nm)
mean that the molecules, in this case two or more macromolecules
such as proteins, need to be physically close to one another for
FRET to occur. The key general requirements are that the donor
emission spectrum must overlap sufficiently with the acceptor
absorption spectrum while the donor and acceptor dipoles display
a mutual molecular orientation. The efficiency E or the rate of the
energy transfer kT can be described and calculated simply using
Eq. (1):


6
1 R0 R0 6
kT ¼ or E ¼ ð1Þ
τD R R þ R0 6
where τD is the donor excited-state lifetime in the absence of the
acceptor, R is the distance between D (donor) and acceptor, and R0
is the Förster radius. At the Förster radius (R ¼ R0), 50% of the
donor molecules will emit fluorescence while the rest will undergo
energy transfer. Since the energy transfer process is strongly dis-
tance dependent with 1/R6, FRET can be used to measure dis-
tances and examine molecular interactions on a nanometer spatial
 Multicolor FRET-FLIM Microscopy to Analyze Multiprotein Interactions in. . . 289

Fig. 1 Schematic representation of colocalization versus interaction of proteins in a confocal volume of a high
numerical aperture microscope objective

scale. In experimental conditions, values for FRET efficiencies are

typically between 50% and 5%, thus R0  R  112R0 since FRET
pairs are around 5 nm for R0.
However, imaging steady-state FRET under microscopy con-
ditions alone may lead to erroneous and false-positive results since
the protein size (~5 nm and volume) is less than a factor of 50 com-
pared to that of the microscope objective with a diffraction-limited
spot size (~250 nm, Fig. 1). Furthermore, the strict microscopy
condition required to be met to avoid false results during normal
FRET, including instrument error such as thresholding and emis-
sion bleed-through, is rarely accounted for routinely [12].
For multiple acceptors and donors, and going beyond a one to
one protein interaction, the efficiency equation above apparently
becomes less accurate. The probability for FRET coupling for a
donor increases with the number of acceptors. Similarly, the dis-
tance sensitivity of FRET also increases with the number of multiple
potential acceptors for each donor without a real change in molec-
ular distance [13–15]. It has been suggested that in cases where
multiple acceptors modify the FRET coupling, it may be advanta-
geous to work with normalized transfer rates, k0T , as these are
additive while efficiency of transfer E (Eq. 1) is not linear with
multiple acceptor–donor situations [16] and can be described in
Eq. (2) for i number of acceptors (Ai).
P
kD!Ai E
E¼P i so that kD!A ¼ ð2Þ
i kD!Ai þ τ
1 τD ð1  E Þ
D

The efficiency of transfer from two acceptors is therefore:

290 Abdullah Ahmed et al.

Fig. 2 Simplified types of Förster resonance energy transfer (FRET) multiplexing. (a) Standard dual channel
(1:1) FRET process. For example, small two molecular probes or intramolecular FRET sensors co-expressed in
a cell, each containing a donor (D) and acceptor (A) fluorophores that are spectrally separate but emission of
the D overlaps with excitation of A, shown in (c). Both fluorophore needs to be within <10 nm. (b) Sequential
two-step FRET between three fluorophores that form two consecutive FRET donor and acceptor pairs. The
acceptor of the first pair acts as the donor for the second acceptor. In (b1), the donor (D1) is able to transfer
energy to both A1 and A2, so that D interacts with both A1 and A2. But A1 does not interact with A2. (c) In this
triple-fluorophore FRET used to detect multiple interactions in a complex, the overlap between all three does
need to have some spectral overlap. FRET detection between the donors and acceptors can be determined
when they form a donor–acceptor triple complex. (c, d), and (e) are representative excitation and emission
spectra (donors and acceptors pairs) for the fluorescent probes described in the work. The double-headed
arrows indicate the spectral overlap between donor emission and acceptor excitation spectra. (Modified from
Ref 17)

kD!A1 þ kD!A2
E¼ ð3Þ
kD!A þ kD!A þ τ1D
where E is the efficiency, k is the transfer rate, D is the donor, and τ
is the lifetime.
The cumulative FRET efficiencies in a complex multiplexed
acceptor(s)–donor(s) (Fig. 2) further developed by FRET from
single to multiplexed signaling events [17] can be calculated using
the normalized transfer rates related as:

6
0 kT R0 k0
kT ¼ ¼ , E¼ 0 T ð4Þ
Γ þ knr R kT þ 1
where kT is the transferred rate, k0T is the normalized transferred
rate, knr is the non-radiative decay, R is the distance between donor
and acceptor, R0 is the Förster radius, and E is the efficiency.
 Multicolor FRET-FLIM Microscopy to Analyze Multiprotein Interactions in. . . 291

During FRET, the rate of decay is reduced by a quenching

process that depletes the excited state of the donor fluorophore,
i.e., the donor fluorescence lifetime is shortened. Thus, by measur-
ing changes in the excited-state lifetime of the donor at each pixel
making up an image, steady-state FRET can be more accurately
determined. This is described as fluorescence lifetime imaging
microscopy or FRET-FLIM [18–22]. Generally, the two proteins
under investigation are tagged with green fluorescent protein
(GFP) and its variants. In our work, we have used a combination
of blue FP (mTurquoise), yellow FP, and red FP (mCherry) as
donors and acceptors to investigate some of the multiple complex
mTORC1 proteins. We have also applied the FLIM imaging
method to investigate DNA structure as well as protein–protein
interactions in several systems including mammalians, viruses, and
plants [23–27], thus demonstrating wide applicability of FRET-
FLIM. We describe here a multi-color FRET-FLIM microscopy of
two-channel pulsed interleave excitation (PIE) to analyze multi-
protein interactions in live cells.
Measuring the change in donor(s) lifetime in a FLIM experi-
ment provides an elegant way to answer biological questions as well
as protein signaling that may involve multiple proteins at the same
time. It is important to excite only the donor(s) individually, in the
complex, and this is provided by the pulsed interleave excitation
(PIE) method (Fig. 3).

Fig. 3 Schematics of two-channel pulsed interleave excitation fluorescence lifetime imaging microscopy
292 Abdullah Ahmed et al.

2 Materials

General laboratory reagents were purchased as analytical grade

from Sigma-Aldrich (USA), Millipore (UK), Fisher Scientific
(UK) and Thermo Fisher Scientific (UK). Other chemical reagents
and kits (all chemicals included) for the molecular biology were
obtained from Qiagen (Germany), Machinery-Nagel (USA), and
Agilent (UK) unless otherwise stated. All primers were purchased
from Integrated DNA Technologies (IDT, Europe) at a 25 nmole
(0.27 mg) DNA oligo concentration with standard desalting.
FuGENE HD Transfection reagent was purchased from Promega.
All cell culture reagents were bought from Gibco (Thermo Fisher
Scientific).
1. All primers were designed with a Tm (melting temperature)
close to 60  C, although this varied slightly depending on the
GC content. Gene-specific primers with 15 bp extension
homologous to vector ends were designed for In-Fusion™
cloning [28], using an automated primer design tool:
https://www.oppf.rc-harwell.ac.uk/Opiner/, or for fusing
two inserts together using SnapGene® software (from GSL
Biotech; available at snapgene.com). Lyophilized primer stocks
(100 μM) were resuspended with nuclease-free water. Stocks
were diluted 1:10 to make up 10 μM working stocks that were
stored at 20  C.
2. Buffers and media for cloning the fluorescent protein tagged
proteins were prepared at room temperature using ultrapure
water (deionized water, at 18 MΩ-cm at 25  C). The S6K1-
GFPSpark construct was obtained from Sino Biological
(China) and mCherry-raptor and YFP-mTOR from Addgene.
Miniprep and Maxiprep kits were purchased from Qiagen
(UK).
3. HEK293 cell line was purchased from ATCC® to make sure of
origin and purity and free from mycoplasma contamination as
well as passage number. Cell culture dishes with glass bottom
(35 mm) were bought from MatTek.

3 Methods

3.1 Amplification PCRs were set up on ice in 0.2 ml PCR tubes or in an eight-strip
of cDNA by PCR PCR tube (Star Lab) containing both forward (Fwd) and reverse
(Rev) primers at a 10 μM working concentration, Phusion Flash
Mix (Thermo Fisher Scientific), template plasmid DNA
(100–200 ng) containing cDNA gene to be amplified and sterile
water as shown in Table 1. Constituents were gently mixed by
 Multicolor FRET-FLIM Microscopy to Analyze Multiprotein Interactions in. . . 293

Table 1
PCR volumes

Amount for one reaction

Phusion flash master mix (2) 25 μl
Forward primer (10 μM) 1.5 μl
Reverse primer (10 μM) 1.5 μl
Template DNA plasmid 1 μl
Sterile water 21 μl
Total volume 50 μl

Table 2
Parameters of thermal cycles used for PCR

Step Temperature Time Cycle

Initial denaturation 98  C 10 s 1

Denaturation 98 C 1s 29
Annealing 60  C 5s
Extension 72  C 15 s/kbp
Final extension 72  C 2 min 1

Hold 4 C Hold

pipetting up and down. The PCR tubes were placed into a Veriti™
96-well thermal cycler (Thermo Fisher Scientific) running a pro-
gram as listed in Table 2.

3.2 Agarose Gel Upon completion of the PCR, 10 μl of filtered DNA loading buffer
Electrophoresis of PCR (0.25% w/v bromophenol in 30% v/v glycerol) was added to each
Products PCR tube and mixed gently by pipetting up and down. A volume of
40 μl of the PCR–dye mix was then loaded onto a pre-cast 1.0%
agarose/Tris-borate-EDTA (TBE) gel, commonly used for DNA
gel electrophoresis, containing the SYBRSafe stain (Thermo Fisher)
as listed in Table 3. Adjacent to the samples, 5 μl of Hyperlad-
derTM 1 kb (BioLine) was loaded and run in a 10  8 gel tank
(MiniRapide) filled with TBE for 120 min at 70 V using a Power
Pac 300 (Bio-Rad) power supply until the markers were separated
sufficiently. After marker bands were separated, the gel was viewed
with a blue light illuminator (Jencons-PLS) to both visualize and
verify PCR bands at their respected correct base pair lengths.

3.3 DNA PCR Product PCR products were visualized using blue light illumination, excised
Extraction from from the gel with a sterile surgical scalpel and transferred to a 1.5-
Agarose and Clean Up ml tube. The PCR-amplified gene was purified from the gel using a
commercially supplied kit (i.e., Machinery-Nagel).
294 Abdullah Ahmed et al.

Table 3
Parameters of thermal cycles used for PCR

Composition Amount
Agarose (sigma-Aldrich) 0.5 g
TBE (1) (sigma-Aldrich) 50 ml
SYBRSafe (Thermo fisher scientific) 5 μl
Composition of 1.0% TBE agarose gel

Table 4
In-fusion reaction for a two-way fusion

Components Amount for one reaction

Linearized vector (20–100 ng) 1 μl
PCR product insert (10–100 ng) 2 μl
Fusion enzyme 1 μl
5 fusion buffer 2 μl
Sterile water 4 μl
Total 10 μl

3.4 In-Fusion Ligation of the PCR products with linearized pOPIN vectors were
Reaction with pOPIN performed according to the manufacturer’s recommendations
Vectors using a Quick-Fusion cloning kit (Biotool). A two-way fusion
cloning reaction involving one PCR DNA insert and vector was
set up as described in Table 4. For a three-way fusion cloning
reaction involving two PCR DNA inserts and vector, 2 μl of each
insert with 2 μl of vector was added and the final volume adjusted to
10 μl with sterile water.
1. Transform fusion products into competent E. coli (i.e., RecA
strain DH5α) by electroporation or heat shock.
2. Incubate bacteria at 37  C for 30–60 min.
3. Plate different volumes of the bacterial suspension on LB plates
supplemented with appropriate antibiotics for selection.
4. Incubate plates at 37  C overnight.
5. Pick single colonies and grow LB overnight cultures.
6. Purify the recombinant plasmid using a commercially supplied
kit, i.e., the QIAprep Spin Miniprep Kit.
7. Optionally, determine the DNA concentration and purity using
a spectrophotometer, i.e., NanoDrop Microvolume Spectro-
photometer (Thermo Fisher Scientific).
 Multicolor FRET-FLIM Microscopy to Analyze Multiprotein Interactions in. . . 295

3.5 Construction 1. S6K1-mTurqouise2 construct was generated in a similar man-

of S6K1-mTurq2 ner by insertion into a pOPINE-3C-mTurq2 vector, also
and Raptor-YFP provided by the Protein Production UK (PPUK) using the
Plasmid Constructs primers below:
S6K1-mTurq2 Fwd AGGAGATATACCATGAGGCGACG
AAGGAGGCGG.
S6K1-mTurq2 Rev CAGAACTTCCAGTTTTAGATTCA-
TACGCAGGTGCTCTG.
2. Raptor-YFP was generated by in-fusion cloning the full length
raptor ORF from mDsRed-raptor into the pOPINE-3C-YFP
vector, using the primers below:
raptor-YFP Fwd AGGAGATATACCATGGAGTCCGAAAT
GCTGCAATCG.
raptor-YFP Rev CAGAACTTCCAGTTTTCTGACACGCT
TCTCCACCG.
3. Plasmid constructs were verified by reverse PCR screens and
further validated by Sanger sequencing using T7F primers and
appropriate reverse primers at Source Bioscience (UK).

3.6 Cell Lines HEK293 cells were cultured in Minimum Essential Media (MEM)
supplemented with 10% (w/v) fetal bovine serum (FBS), 2 mM L-
glutamine, and 1% (w/v) penicillin–streptomycin. All culture
reagents were acquired from Life Technologies. Cells were incu-
bated at 37  C with 5% CO2 humidified air in T75 culture flasks
(Thermo Scientific).

3.7 Cell Transfection HEK293 was seeded for 24 h at a density of 1  105 or

1.5  105 cells per ml on uncoated 35 mm no 1.5 glass bottom
dishes (MatTek). Cells were transfected with 0.5 μg of plasmid
DNA using FuGENE HD (Promega) transfection reagent.

3.8 Equipment Our two-channel confocal and two-channel FLIM setup was fitted
for Real-Time Imaging with time-correlated single-photon counting cards (TCSPC) from
of Protein–Protein Becker & Hickl (BH SPC 830 or SPC150), and data were analyzed
Interactions Using using the BH SPCImage analysis software v 5.1.
a Confocal 1. Although the two-photon microscopy technique has several
and FRET-FLIM Setup advantages over a one-photon system, it is best to use a
one-photon confocal microscope for the multicolor FRET-
FLIM technique. This is because the ultrafast femtosecond
pulsed laser is likely to excite both the donor(s) and acceptor
(s) at the same time. This needs to be avoided. For the PIE
operation which is essential for three protein interaction stud-
ies, two separate pulsed laser sources are required. It is more
convenient to use two lasers that (1) can vary their repetition
rates, (2) can be synchronized, and (3) can be triggered exter-
nally (Fig. 3).
296 Abdullah Ahmed et al.

2. We have used a BH BDL-SMC-405 pulsed laser with 40 ps

pulse-width synchronized with the NKT supercontinuum laser
operating at 80 MHz. The two lasers were combined using a
fiber combiner (OzOptics, Canada). The supercontinuum laser
operated at the full 80 MHz as the “master” and was used to
trigger the BDL-SMC-405 in a “slave” mode with a 10-ns
delay. It is recommended to run the lasers at 40 MHz, i.e.,
25 ns between pulses to allow a full decay of the fluorescent
protein excited state before the next excitation pulse arrives.
Fluorescence emission was filtered using a 460/60 nm band-
pass filter for CFP or mTurquoise2, or 515/30 nm filter for
GFP (Thorlabs) and focused onto the two hybrid detectors.
3. The laser beams are focused to a diffraction-limited spot using a
high numerical aperture oil or water-immersion objective (e.g.,
Nikon x60 VC NA 1.2) to illuminate specimens on the micro-
scope stage. Fluorescence emission is collected by the same
objective in a standard confocal mode. Line, frame, and pixel
clock signals are generated and synchronized with two external
detectors in the form of two fast hybrid photomultiplier tubes
(PMT) or microchannel plate PMTs (i.e., Hamamatsu
R3809U or HPM-100). The raw FLIM data are generated by
linking these PMTs via two TCSPC PC modules. We used BH
SPC830 or SPC150 cards. A single TCSPC may be used
together with routing electronics, but this would lead to only
half of the intensity from each channel recorded.
4. Prior to FLIM data collection, expression levels of the fluores-
cent protein-tagged proteins of interest are verified by imaging
transfected cells using a confocal microscope. Here, an inverted
Nikon TE2000-U or Ti-E confocal microscope attached to a
Nikon C1 or C2 scanning unit was used with filter sets for GFP
(488 nm excitation) or mDsRed/mCherry/Alexa 555 (543 or
561 nm excitation) and mTurqouise2/Alexa405 (405 nm exci-
tation). A 633-nm interference filter is used in the red channel
to significantly minimize the bleed-through of any GFP or
autofluorescence emission that would otherwise obscure the
mCherry emission. Ideally, equal expression levels for all pro-
teins under investigation are selected to obtain optimal FRET
conditions (Fig. 4). Alternatively, higher levels of the acceptor
protein are desirable.
5. Prior to FLIM image data acquisition, the system performance
needs to be checked and verified. Standard dyes with well-
characterized excitation and emission spectra and excited-
state lifetimes are commonly used (Table 5). We have routinely
used aqueous solutions (~50 μM) of 7-hydroxy-coumarin car-
boxylic acid (7-OH-CCA, 360–405 nm excitation, 420 nm
 Multicolor FRET-FLIM Microscopy to Analyze Multiprotein Interactions in. . . 297

emission), fluorescein at pH 7–10 (488 nm excitation, 515 nm

emission), and rhodamine B (543–561 nm excitation, 590 nm
emission).
6. FLIM images were acquired at 256  256 pixels, although
megapixel sizes (2048  2048) can now be achieved due to
recent advances and improvements in the FLIM acquisition PC
cards and software compared to previous 128  128 pixels
(Fig. 4). Supercontinuum laser coupled with AOTF may be
used to select any excitation and emission wavelengths for both
the confocal and one-photon FLIM operations.

Fig. 4 Live HEK293 cell localization and interaction between the mTORC1-expressed proteins observed by
three-channel confocal and PIE two-channel FLIM. A1: EGPF-S6K1 alone as control with observed lifetime. B1:
The distribution of all lifetime in the image pixel is given in D1. B1–3 Confocal image of cell expression of
mTurg2-S6K1, FYP-Rheb, and mCherry-raptor. B4 and B5 are the two-channel PIE FLIM images of
mTurg2–S6K1 and YFP-Rheb. D2 and D3 are the respective lifetime distributions, indicating a reduction in
lifetime of the mTurg2-S6K1 but not the YFP-Rheb. C1–3 Confocal image of cell expression of mTurg2-S6K1,
FYP-mTOR, and mCherry-raptor. C4 and C5 are the two-channel PIE FLIM images of mTurg2-S6K1 and
YFP-mTOR. D4 and D5 are the respective lifetime distributions, indicating a reduction in lifetime of the
mTurg2-S6K1 but not the YFP-mTOR. E1 and E2 are schematic representation of the physical interaction
between the different proteins within the complex. FLIM of the red channel is not measured but could be
performed in a three-channel PIE setup. Scale bar is 10 μm
298 Abdullah Ahmed et al.

Table 5
Standard solution lifetime calibration for three-colour FRET-FLIM

Excitation Energy
Standard dye % Lifetime (in nm) transfer
7-OH-CCA 100 3.7  0.1 ns 405 N/A
Fluorescein 100 4.0  0.1 ns 488 N/A
Rhodamine b 100 1.7  0.05 ns 543 or 561 N/A
7-OH-CCA + fluorescein 50/50 3.3  0.05 ns 405 Yes
Fluorescein + rhodamine b 50/50 3.1  0.05 ns 488 Yes
7-OH-CCA + 33/33/33 3.3  0.05 ns (Ch1), 404 Yes
fluorescein + rhodamine b 3.1  0.05 ns (Ch2)

Raw FLIM data are analyzed by obtaining excited-state

lifetime values first on a pixel-by-pixel basis, then of a region
of interest, and calculations are made using SPCImage analysis
software.
7. The distribution of lifetime values within the region of interest
is generated and displayed as a curve. Only chi-squared (χ 2)
values between 0.9 and 1.3 are considered a good fit of the data
points. For a pure sample such as the test solutions and donor
FPs (mTurq2 or YFP), it is best to fit the data to a single-
exponential decay. The median, minimum, and maximum
values of the lifetime values from the curve are taken to gener-
ate the range of lifetimes per sample. At least three areas of a
sample dish and at least three independent biological samples
per protein–protein combination are analyzed and the average
of the ranges is taken. The percentage efficiency of energy
transfer (E) from one protein to the other may be determined
using Eqs. (1) and (2):


τDA
E% ¼ 1   100 ð5Þ
τD
where τD and τDA are the measured excited-state lifetime of the
donor and acceptor, respectively.

3.9 Typical Results Typical multichannel FLIM results of multiple protein interactions
are shown in Fig. 4 and changes in lifetime values are given in
Table 6.
 Multicolor FRET-FLIM Microscopy to Analyze Multiprotein Interactions in. . . 299

Table 6
Three-color FRET-FLIM results for multiple mTORC1 interactions

mTORC1 proteins Lifetime (ns) Comment

S6K1-mTurq2 4.3  0.1 Control lifetime
YFP-Rheb 2.8  0.1 Control lifetime
YFP-mTOR ~2.8  0.1 Control lifetime
mCherry-raptor N/A τ not measured
S6K1-mTurq2 + mCherry-raptor 3.3  0.1 Interaction
YFP-Rheb + mCherry-raptor 2.8  0.1 No interaction
YFP-mTOR + mCherry-raptor 2.5  0.1 Interaction
S6K1-mTurq2 + YFP-Rheb + 3.2 (ch 1) + 2.8 Interaction between S6K1 and raptor,
mCherry-raptor (ch 2)  0.1 but no interaction between S6K1
and Rheb or raptor and Rheb.
No direct three-way interaction
between mTOR, raptor, and S6K1
S6K1-mTurq2 + 3.1 (ch1) + 2.5 Strong interaction between S6K1 and
YFP-mTOR + mCherry-raptor (ch 2)  0.1 raptor strong interaction between
raptor and mTOR. Possible slight
direct three-way interaction between
mTOR, raptor, and S6K1

4 Notes

Full personal protective equipment including gloves has to be worn

for the whole procedure to reduce contamination throughout the
preparation. Sterile conditions are required particularly for the cell
culture.

Acknowledgments

This work was supported by a BBSRC iCASE PhD studentship

(BB/L016052/1) to AA. We thank STFC for funding access to the
Central Laser Facility. This work also acknowledges Professor Ray-
mond Owens (PP-UK) for providing support and plasmid vectors
for the cloning work.

References

1. Beretta L, Gingras AC, Svitkin YV et al (1996) 2. Brunn GJ, Hudson CC, Sekulić A et al (1997)
Rapamycin blocks the phosphorylation of Phosphorylation of the translational repressor
4E-BP1 and inhibits cap-dependent initiation PHAS-I by the mammalian target of rapamy-
of translation. EMBO J 15(3):658–664 cin. Science 277(5322):99–101
300 Abdullah Ahmed et al.

3. Gingras AC, Kennedy SG, O’Leary MA et al measurements beyond 100 Å using common
(1998) 4E-BP1, a repressor of mRNA transla- organic fluorophores: enhanced transfer in the
tion, is phosphorylated and inactivated by the presence of multiple acceptors. J Biomed Opt
Akt(PKB) signaling pathway. Genes Dev 12 17:11006. https://doi.org/10.1117/1.JBO.
(4):502–513 17.1.011006
4. Holz MK, Ballif BA, Gygi SP et al (2005) 15. Walczewska-Szewc K, Bojarski P, d’Auria S
mTOR and S6K1 mediate assembly of the (2013) Extending the range of FRET—the
translation preinitiation complex through Monte Carlo study of the antenna effect. J
dynamic protein interchange and ordered Mol Model 19:4195–4201. https://doi.org/
phosphorylation events. Cell 123(4):569–580 10.1007/s00894-013-1810-3
5. Arif A, Terenzi F, Potdar AA et al (2017) EPRS 16. Koushik SV, Blank PS, Vogel SS (2009) Anom-
is a critical mTORC1–S6K1 effector that influ- alous surplus energy transfer observed with
ences adiposity in mice. Nature 542:357–361 multiple FRET acceptors. PLoS One 4(11):
6. Caccamo A, Branca C, Talboom JS et al (2015) e8031. https://doi.org/10.1371/journal.
Reducing ribosomal protein S6 kinase 1 expres- pone.0008031
sion improves spatial memory and synaptic 17. Bunt G, Wouters FS (2017) FRET from single
plasticity in a mouse model of Alzheimer ‘s to multiplexed signaling events. Biophys Rev
disease. J Neurosci 35:14042–14056 9:119–129. https://doi.org/10.1007/
7. Kim SY, Baik KH, Baek KH et al (2014) S6K1 s12551-017-0252-z
negatively regulates TAK1 activity in the toll- 18. Wolfgang B (2015) Advanced single photon
like receptor. Mol Cell Biol 34:510–521 counting applications. Springer, New York, pp
8. Marabita M, Baraldo M, Solagna F et al (2016) 1–624. https://doi.org/10.1007/978-3-319-
S6K1 is required for increasing skeletal muscle 14929-5
force during hypertrophy article S6K1 is 19. Botchway SW, Scherer KM, Hook S et al
required for increasing skeletal muscle force (2015) A series of flexible design adaptations
during hypertrophy. Cell Rep 17:501–513 to the Nikon E-C1 and E-C2 confocal micro-
9. Choo AY, Yoon SO, Kim SG et al (2008) scope systems for UV, multiphoton and FLIM
Rapamycin differentially inhibits S6Ks and imaging. J Microsc 258(1):68–78. https://doi.
4E-BP1 to mediate cell-type-specific repression org/10.1111/jmi.12218
of mRNA translation. Proc Natl Acad Sci U S A 20. Scully AD, Macrobert AJ, Botchway S et al
105(45):17414–17419. https://doi.org/10. (1996) Development of a laser-based fluores-
1073/pnas.0809136105 cence microscope with subnanosecond time
10. Choo AY, Blenis J (2009) Not all substrates are resolution. J Fluoresc 6(2):119–125. https://
treated equally implications for mTOR, doi.org/10.1007/BF00732051
rapamycin-resistance and cancer therapy. Cell 21. Suhling K, Siegel J, Phillips D et al (2002)
Cycle 8:567–572. https://doi.org/10.4161/ Imaging the environment of green fluorescent
cc.8.4.7659 protein. Biophys J 83(3589–3595):2002
11. Förster T (1948) Zwischenmolekulare Ener- 22. Marcu L, French PMW, Elson DS (2014)
giewanderung und Fluoreszenz [Intermolecu- Fluorescence lifetime spectroscopy and imag-
lar energy migration and fluorescence]. ing—principles and applications in biomedical
Annalen der Physik (in German). 437: 55–75. diagnostics. CRC Press, Boca Raton, Florida,
B För s t e r T (1946) Energiewanderung und pp 1–570. SBN 9781439861677—CAT#
Fluoreszenz. Naturwissenschaften K12851
33:166–175? (Translation: by Klaus Suhling). 23. Estandarte AK, Botchway S, Lynch C et al
J Biomed Opt 17:011002 (2016) The use of DAPI fluorescence lifetime
12. Pawley JB (2006) Handbook of biological con- imaging for investigating chromatin condensa-
focal microscopy, 3rd edn. Springer, New York, tion in human chromosomes. Sci Rep 6:31417.
NY, pp 1–985. https://doi.org/10.1007/ https://doi.org/10.1038/srep31417
978-0-387-45524-2_8. ISBN 978-0-387- 24. Yadav RB, Burgos P, Parker AW et al (2013)
25921-5 mTOR direct interactions with Rheb-GTPase
13. Fábián ÁI, Rente T, Szöllősi J et al (2010) and raptor: sub-cellular localization using fluo-
Strength in numbers: effects of acceptor abun- rescence lifetime imaging. BMC Cell Biol 14:3.
dance on FRET efficiency. ChemPhysChem https://doi.org/10.1186/1471-2121-14-3
11:3713–3721. https://doi.org/10.1002/ 25. Stubbs CD, Botchway SW, Slater SJ et al
cphc.201000568 (2005) The use of time-resolved fluorescence
14. Maliwal BP, Raut S, Fudala R et al (2012) imaging in the study of protein kinase C loca-
Extending Förster resonance energy transfer lisation in cells. BMC Cell Biol 6(1):22
 Multicolor FRET-FLIM Microscopy to Analyze Multiprotein Interactions in. . . 301

26. Jeshtadi A, Burgos P, Stubbs CD et al (2010) interactions between Golgi tethering factors
Interaction of poxvirus intracellular mature and small GTPases in plants. Traffic 10
virion proteins with the TPR domain of kinesin (8):1034–1046. https://doi.org/10.1111/j.
light chain in live infected cells revealed by two- 1600-0854.2009.00930.x
photon-induced fluorescence resonance energy 28. Nettleship JE, Assenberg R, Diprose JM et al
transfer fluorescence lifetime imaging micros- (2010) Recent advances in the production of
copy. J Virol 84(24):12886–12894. https:// proteins in insect and mammalian cells for
doi.org/10.1128/JVI.01395-10 structural biology. J Struct Biol 172
27. Osterrieder A, Carvalho CM, Latijnhouwers M (1):55–65. https://doi.org/10.1016/j.jsb.
et al (2009) Fluorescence lifetime imaging of 2010.02.006
 Chapter 17

Context-Specific and Proximity-Dependent Labeling

for the Proteomic Analysis of Spatiotemporally Defined
Protein Complexes with Split-BioID
Cinthia Amaya Ramirez, Stefanie Egetemaier, and Julien Béthune

Abstract
Proximity-dependent labeling techniques such as BioID and APEX2 allow the biotinylation of proteins
proximal to a protein of interest in living cells. Following streptavidin pulldown and mass spectrometry
analysis, this enables the identification of native protein–protein interactions. Here we describe split-BioID,
a protein-fragment complementation assay that increases the resolution of BioID. Using this technique,
context-specific protein complexes can be resolved.

Key words BioID, Protein-fragment complementation assay, Protein–protein interactions, Biotin,

Proteomics

1 Introduction

In all living organisms, thousands of proteins mediate most cellu-

lar functions. To do so, proteins usually do not act alone. Rather,
they assemble with other proteins to build dynamic macromolec-
ular complexes that can remodel according to the exact functions
that need to be exerted. The protein–protein interactions (PPI)
involved in such complexes are often deregulated in disease and
are thus promising targets for therapeutics [1]. Identifying and
characterizing PPI networks are hence of prime importance when
trying to understand how a protein of interest (POI) works. To
complement classical pulldown approaches in which the POI is
isolated by affinity purification (AP) and co-purifying proteins
identified by mass spectrometry (MS) analysis, proximity-
dependent biotinylation techniques were recently introduced
that allow the labeling of proteins vicinal to the POI in living
cells. Two such techniques are currently available: APEX2- and
BioID-mediated labeling [2, 3]. The former relies on an engi-
neered peroxidase that activates a biotin-phenol substrate that is

Arnaud Poterszman (ed.), Multiprotein Complexes: Methods and Protocols, Methods in Molecular Biology, vol. 2247,
https://doi.org/10.1007/978-1-0716-1126-5_17, © Springer Science+Business Media, LLC, part of Springer Nature 2021

303
304 Cinthia Amaya Ramirez et al.

fed to the cells, whereas the latter uses an abortive variant of a

protein biotin ligase (BPL) that activates native biotin. In both
the cases, the activated substrate diffuses around the enzyme and
leads to the biotinylation of proximal proteins in an estimated
range of approximately 10 nm for BioID [4]. When fused to a
POI, both enzymes thus mediate the biotinylation of nearby
factors including interacting proteins. These biotinylated proteins
can then be efficiently isolated by streptavidin pulldown and
identified by MS. As opposed to AP-MS, APEX2- and BioID-
MS do not aim at purifying assembled protein complexes by
pulling one of its components but rather identify proteins that
were marked within cells because they were in close proximity to
the POI. In proximity-dependent labeling approaches, it thus
does not matter whether interacting proteins are still associated
during the pulldown procedure, making these techniques more
powerful than AP-MS for detecting transient interactions [5] or
PPI that depend on intact or poorly soluble cellular structures
[6]. Yet, a limitation to both approaches is that they cannot easily
resolve the remodeling and/or maturation of protein complexes.
In the common situation in which a POI is part of several distinct
complexes depending on the cellular context, both AP- and
APEX2/BioID-MS will identify all possible PPI but will not
assign individual interactions to specific context-dependent com-
plexes. We and others have recently introduced split-BioID assays
[7, 8] in which BirA*, the BPL of BioID, is split into two inactive
and poorly interacting fragments that can reassemble into an
active enzyme when fused with two interacting proteins. Provided
a POI is known to interact with a given factor within a specific
protein complex, applying split-BioID to these two proteins
allows the labeling of additional components of that specific
assembly, ignoring any additional PPI the POI may have within
other protein complexes (Fig. 1). The general strategy of a split-
BioID experiment is outlined in Fig. 2. Briefly, the coding
sequences for two putative interacting POI are cloned in frame
with the two BirA* fragments, termed NBirA* and CBirA*, and
transfected in mammalian cells. Biotinylation of proximal proteins
is induced by the addition of excess biotin in the growth medium.
Biotinylated proteins are then isolated by streptavidin pulldown
and can be analyzed either by immunoblotting or MS. Of note
the protocol described here is virtually identical to a classical
BioID procedure, the only difference being that in BioID the
POI is fused to the full BirA* enzyme.
 Context-Specific Proximity-Dependent Proteomics with Split-BioID 305

Fig. 1 Principle of split-BioID. Protein A interacts with protein B within Complex I or with protein C within
Complex II. Fusing the BirA* fragments NBirA* and CBirA* to protein A and protein B, respectively, leads to the
reassembly of an active enzyme upon dimerization of these two proteins and the specific labeling of further
components of complex

Fig. 2 Outline of the main steps in a typical split-BioID experiment

306 Cinthia Amaya Ramirez et al.

2 Materials

Prepare all solutions using ultrapure water (such as mQ water from

Millipore or equivalent) and analytical grade reagents. When pre-
paring stock solutions and reagents, care should be taken to avoid
keratin contamination (essentially: always use gloves, single-use
plastic ware, and do not touch yourself).

2.1 Preparation 1. DNA restriction enzymes.

of Plasmids 1 2. T4 DNA ligase or DNA ligation kit.
3. Polymerase chain reaction (PCR) reagents.
4. Miniprep/Midiprep kit for plasmid DNA purification.
5. Transformation competent E. coli cells.

2.2 Transfection 1. Cell line of interest (see Note 1).

of Mammalian Cells 2. Cell culture medium (see Note 2).
3. Transfection reagent (see Note 3).

2.3 1. Biotin stock solution: 50 mM in neutralized ammonium

Proximity-Dependent hydroxide (see Note 4). Store at 20  C.
Biotinylation 1
2. Doxycycline stock solution: 10 mg mL in 70% ethanol (see
Note 5). Store at 20  C.

2.4 Streptavidin 1. Streptavidin-coupled magnetic beads (see Note 6). Store at

Pulldown 4  C.
2. Magnet for 1.5-mL tubes.
3. Lysis buffer II: 50 mM Tris–HCl, 500 mM NaCl, 5 mM
EDTA, 0.4% SDS, 1 mM DTT, protease inhibitor cocktail,
pH adjusted to 7.4. Prepare freshly (see Note 7).
4. Equilibration buffer: 50 mM Tris–HCl, 150 mM NaCl, 0.05%
Triton X-100, 1 mM DTT, adjusted to pH 7.4.
5. Wash buffer 1: 2% SDS in water. Prepare freshly (see Note 7).
6. Wash buffer 2: 50 mM Hepes, 1 mM EDTA, 500 mM NaCl,
1% Triton X-100, 0.1% sodium-deoxycholate, pH adjusted to
7.4. Prepare freshly (see Note 7).
7. Wash buffer 3: 10 mM Tris–HCl, 250 mM LiCl, 1 mM EDTA,
0.5% NP-40, 0.5% sodium-deoxycholate, pH adjusted to
8. Prepare freshly (see Note 7).
8. Wash buffer 4: 50 mM Tris–HCl, 50 mM NaCl, 0.1% NP-40,
pH adjusted to 7.4. Prepare freshly (see Note 7).
9. Elution buffer: 10 mM Tris–HCl, 2% SDS, 5%
β-mercaptoethanol, 2 mM biotin, pH adjusted to 7.4. Store
at 20  C.
 Context-Specific Proximity-Dependent Proteomics with Split-BioID 307

10. Bradford reagent.

11. 25G needles.
12. 1-mL syringes.
13. Rotating wheel for 1.5-mL tubes.

2.5 1. Lysis buffer I: 50 mM Tris–HCl pH 7.4, 150 mM NaCl, 2 mM

SDS-Polyacrylamide EDTA, 0.5% NP-40, 0.5 mM DTT, protease inhibitor cocktail.
Gel Electrophoresis 2. Stacking gel buffer: 1 M Tris–HCl, pH 6.8.
and Western Blotting
3. Resolving gel buffer: 1.5 M Tris–HCl, pH 8.8.
4. Thirty percent acrylamide/bisacrylamide solution (37.5:1).
5. Running buffer: 25 mM Tris–HCl, 192 mM glycine,
0.1% SDS.
6. Sample loading dye (5): 250 mM Tris–HCl, pH 6.8, 5%
β-mercaptoethanol, 0.02% bromophenol blue, 10% SDS, 30%
glycerol.
7. Low-fluorescence PVDF membranes (see Note 8).
8. Trans-Blot Turbo Transfer System and reagents (see Note 9).
9. Infrared dye-conjugated streptavidin (see Note 8). Store at
4  C.
10. Precast protein gels and reagents (see Note 10).
11. Coomassie staining solution: 0.02% (w/v) Coomassie blue
stain G-250, 5% (w/v) aluminum sulfate-(14–18)-hydrate,
10% (v/v) ethanol (96%), 2% (v/v) orthophosphoric acid
(85%). Store at room temperature.
12. Destaining solution: 10% (v/v) ethanol (96%), 2% (v/v) ortho-
phosphoric acid (85%).
13. Sterile 15-cm bacterial dishes.
14. Sterile scalpels.

3 Methods

3.1 Cloning of Two 1. Design primers to amplify the two genes to be tested, so that
Genes of Interests restriction sites are included that allow subcloning in split-
in Split-BioID Plasmid BioID plasmids (see Note 11).
2. Perform PCR to amplify both genes, run the reactions on a 1%
agarose-TAE gel and cut slices of agarose gel that encompass
the amplified DNA. Extract and purify the amplified DNA
using a standard DNA gel extraction kit.
3. Digest amplified DNA and acceptor plasmid(s) with the rele-
vant restriction enzymes. Typically 2 μg of plasmid is mixed
with 1 μL of each restriction enzyme in a total volume reaction
of 20 μL, whereas 25 μL of the purified PCR products is mixed
308 Cinthia Amaya Ramirez et al.

with 1 μL of each restriction enzyme in a total volume reaction

of 30 μL. DNA digestion with restriction enzymes is typically
performed at 37  C for at least 1 h (see Note 12).
4. Run the plasmid restriction reactions on a 1% agarose-TAE gel
and cut slices of agarose gel around the digested plasmid back-
bone bands. Extract and purify the digested DNA using a
standard DNA gel extraction kit. Digested PCR products can
be directly purified using a PCR or reaction clean up protocol
usually included in gel extraction kits.
5. Ligate digested plasmids and PCR products using T4 DNA
ligase. Typically, a total of 50–100 ng DNA (plasmid + insert
with a three- to fivefold molar excess insert/plasmid) is mixed
with ligation buffer and DNA ligase in a total of 10 μL and
incubated at room temperature for 5–10 min.
6. Use 3 μL of the ligation reaction to transform 50 μL of com-
petent E. coli cells. Typically bacteria and DNA are gently
mixed and incubated on ice for 30 min. Transformation is
then induced by heat shock at 42  C for 30–45 s. The cells
are then placed back on ice. After 2 min, 250 μL of pre-warmed
(37  C) LB medium is added and the cells incubated on a
shaking block at 37  C for 1 h at 800 rpm (see Note 13).
Thereafter, 100 μL of the incubation is spread on a 10 cm
bacterial LB-agar plate containing the appropriate antibiotic
selection. The plates are incubated overnight at 37  C.
7. On the next day, two to four antibiotic-resistant colonies are
picked to inoculate 3 mL of LB medium containing the appro-
priate antibiotic in 15 mL tubes. The tubes are then incubated
for 12–16 h at 37  C, 180 rpm in a shaker-incubator.
8. On the next day, plasmids are isolated by Miniprep and their
sequences verified by Sanger sequencing analysis.
9. Correct plasmids can be re-amplified and purified with a Mid-
iprep protocol to generate a larger stock of sequence-verified
split-BioID plasmids.

3.2 Small-Scale 1. Seed a six-well cell culture plate with the cells to transfect so
Testing of Proximity- that they reach 50–70% confluence on the following day. Incu-
Dependent bate overnight at 37  C, 5% CO2. For HeLa cells, typically
Biotinylation 80–150  103 cells are seeded per well in a volume of 2 mL (see
in Living Cell Note 14).
2. On the next day, remove the medium from the six-well plate
and replace with 2 mL of fresh medium per well.
3. Prepare the transfection mixes for all conditions to be tested,
usually the pair of proteins of interest and a negative control (see
Note 15). When using PEI as a transfection reagent for HeLa
cells, we typically use 6 μg PEI diluted in 100 μL DMEM
Context-Specific Proximity-Dependent Proteomics with Split-BioID 309

(without serum) mixed with 3 μg plasmid diluted in 100 μL

DMEM (without serum) for each well to transfect. The trans-
fection mixes (200 μL per well) are incubated for 5–15 min at
room temperature (RT) and then added drop wise to their
corresponding wells. The cells are then incubated overnight
at 37  C, 5% CO2.
4. On the next day, remove the medium from each well and
replace by 2 mL of DMEM containing 50 μM biotin and
200 ng mL 1 doxycycline if using inducible plasmids under
the control of a tet-responsive element (see Note 16). Incubate
the cells for at least 16 h at 37  C, 5% CO2.
5. On the following day, wash cells in each well with 1 mL ice-cold
PBS and then scrap the cells in 100 μL of lysis buffer I with
disposable cell scrappers.
6. Transfer each sample of scrapped cells to 1.5 mL tubes, resus-
pend briefly by pipetting and centrifuge at 14,000  g for
10 min at 4  C to clear the lysates.
7. Transfer the supernatants to fresh 1.5-mL tubes and place
on ice.
8. Determine the protein concentration of each sample using a
standard Bradford assay.
9. Prepare samples for SDS-PAGE by mixing an equal amount
(15–30 μg protein content) of each lysate with 5 SDS-loading
buffer to yield a total of 30 μL SDS-PAGE sample in 1
SDS-loading buffer (volume are adjusted with lysis buffer I).
Denature the samples at 95  C for 5 min (see Note 17).
10. Load the samples (20 μL/lane) and molecular weight markers
on an SDS-polyacrylamide gel with a percentage allowing the
appropriate resolution of the different fusion proteins.
11. After electrophoresis, transfer the fractionated proteins to an
immunoblotting membrane (see Note 9).
12. Following protein transfer, block the membrane with 5% dry
milk in PBS for 30 min at RT.
13. To detect biotinylated proteins, incubate the membrane in a
50-mL falcon tube with 5 mL of conjugated streptavidin
diluted 1:15,000 in PBS containing 2% bovine serum albumin
(BSA) and 0.1% Tween-20 for 30 min at RT on a roller-shaker.
14. Wash the membrane three times with PBS containing 0.1%
Tween-20 for 10 min at RT (see Note 18) and then proceed
with detection of chemoluminescence or fluorescence signals
(Fig. 3).
15. If detection of the fusion proteins is desired, incubate the
membrane in a 50-mL falcon tube with 5 mL of anti-Myc
(to detect the fusion to NBirA*) and anti-FLAG (to detect
310 Cinthia Amaya Ramirez et al.

Fig. 3 Typical Western blot for a small-scale split-BioID validation experiment.

Biotinylated proteins were detected with conjugated streptavidin. Endogenous
biotinylated proteins are detected in non-transfected cells. A similar pattern is
observed for the GFP/TNRC6C negative control. Additional biotinylated proteins
(mainly the fusion proteins themselves) are detected when split-BioID is applied
to the Ago2 and TNRC6C, which are known to directly interact

the fusion to CBirA*) diluted at their optimal concentrations

(see Note 19) in PBS containing 2% bovine serum albumin
(BSA) and 0.1% Tween-20 for 1 h at RT on a roller-shaker (see
Note 20).
16. Wash the membrane three times with PBS containing 0.1%
Tween-20 for 10 min at RT.
17. Incubate the membrane in a 50-mL falcon tube with 5 mL of
conjugated secondary antibodies directed against the anti-Myc
and -FLAG primary antibodies diluted at their optimal con-
centrations (see Note 21) in PBS containing 2% bovine serum
albumin (BSA) and 0.1% Tween-20 for 30 min at RT on a
roller-shaker.
18. Wash the membrane three times with PBS containing 0.1%
Tween-20 for 10 min at RT (see Note 18) and then proceed
with detection of chemoluminescence or fluorescence signals.

3.3 Large-Scale 1. For each condition to be tested, seed three to four 10-cm cell
Proximity-Dependent culture dishes with the cells to transfect, so that they reach
Biotinylation approximately 70% confluence on the following day. Incubate
for Proteomics Studies overnight at 37  C, 5% CO2. For HeLa cells, typically 8  105
cells are seeded per dish in a volume of 10 mL (see Note 14).
Context-Specific Proximity-Dependent Proteomics with Split-BioID 311

2. On the next day, remove the medium from the dishes and
replace with 10 mL of fresh medium.
3. Prepare the transfection mixes for all conditions to be tested,
usually the pair of proteins of interest and a negative control (see
Note 15). When using PEI as a transfection reagent for HeLa
cells, we typically use 12 μg PEI diluted in 150 μL DMEM
medium (without serum) mixed with 6 μg plasmid diluted in
150 μL DMEM medium (without serum) per plate. The trans-
fection mixes are incubated for 5–15 min at room temperature
(RT) and then dispatched drop wise to their corresponding
dishes (300 μL per 10 cm plate). The cells are then incubated
overnight at 37  C, 5% CO2.
4. On the next day, transfer the cells from each dish to 15 cm
plates. To do so, the cells are first washed with 7 mL PBS and
then incubated with 1.5 mL of a trypsin-EDTA solution for
5 min at RT. DMEM (3.5 mL) is then added and the cells
resuspended by pipetting and transferred to 15 cm plates filled
with 20 mL DMEM containing enough biotin to reach a final
concentration of 50 μM for a total volume of 25 mL, and
enough doxycycline for a concentration of 200 ng mL 1 doxy-
cycline if using inducible plasmids under the control of a
tet-responsive element (see Note 16). Incubate the cells for at
least 16 h at 37  C, 5% CO2.
5. On the following day, cells in each 15 cm dish are washed twice
with 20 mL PBS and then scrapped in 1.5 mL PBS using
disposable cell scrappers.
6. Scrapped cells are then transferred to 1.5 mL tubes and har-
vested by centrifugation at 1200  g for 5 min at 4  C.
7. Cell pellets can then be snap frozen in liquid nitrogen and if
desired stored at 80  C until further processing.
8. To prepare cell lysates, the frozen cell pellets are first resus-
pended in 1 mL lysis buffer II at room temperature and passed
ten times (five strokes) through 25G needle attached to a 1-mL
syringe (see Note 21).
9. The lysates are then sonicated in a cold ultrasonic bath for
4 cycles of 30 s at high intensity with 30 s pause time between
each cycle (see Note 22).
10. Add Triton X-100 to the sonicated lysates to a final concentra-
tion of 2% (Typically 100 μL of a 20% Triton X-100 solution is
added to 900 μL of lysate) and then enough volume of 50 mM
Tris–HCl, pH 7.4 solution, so that the final NaCl concentra-
tion is adjusted to 150 mM (typically 2.3 mL is added to 1 mL
of Triton X-100 adjusted lysates) (see Note 23).
312 Cinthia Amaya Ramirez et al.

11. Distribute the adjusted lysates (total volume approximately

3.3 mL) in three 1.5-mL tubes and clear by centrifugation at
16,000  g for 10 min at 4  C.
12. Pool the supernatants from each condition in 15-mL tubes.
13. Determine the protein concentration of each sample using a
standard Bradford assay. Adjust volumes of each sample, so that
they all have equal concentration and keep 10% aside as Input
sample for later analysis. Typically 3–3.5 mg of protein content
is used for a streptavidin pulldown.

3.4 Streptavidin 1. Pre-equilibrate the total amount of streptavidin-coupled beads,

Pulldown typically 200 μL per pulldown (see Note 24) by transferring the
for Proteomics Studies beads to a 1.5-mL tube and wash them once with 1 mL equili-
bration buffer.
2. Dispatch the equilibrated beads in enough 1.5-mL tubes to
accommodate the starting amount of 3–3.5 mg protein sample
per conditions (usually two to four tubes are needed).
3. Place the tubes on a magnetic rack, remove the supernatant,
and for each condition, resuspend the beads with equal
amounts of cleared lysates.
4. Incubate the bead/lysate mixture overnight at 4  C on a rotat-
ing wheel.
5. On the next day, place the tubes on a magnetic rack, wait until
the beads have stuck to the side of the tubes and gently aspirate
the supernatants using a capillary tip connected to a vacuum
pump. From now on, the following steps are performed at RT.
6. In each tube, resuspend the beads with 200 μL of wash buffer
1 and pool beads from the same conditions in a single 1.5-
mL tube.
7. Incubate for 8 min on a rotating wheel.
8. Place the tubes on a magnetic rack, wait until the beads have
stuck to the side of the tubes and gently aspirate the super-
natants using a capillary tip connected to a vacuum pump or
manually with a pipette.
9. Wash the beads once more with 1 mL of wash buffer 1 as in
steps 7 and 8.
10. Following the same procedure as in step 9, wash the beads
twice with 1 mL of wash buffer 2, then wash buffer 3, then
transfer to fresh 1.5-mL tubes and proceed with wash buffer 4.
11. After the final wash, briefly centrifuge the samples on a table
micro-centrifuge, place back on the magnetic rack, and remove
any remaining buffer with a pipette.
12. Elute captured biotinylated proteins (see Note 25) by adding
30 μL of elution buffer to each sample. Incubate at 98  C,
 Context-Specific Proximity-Dependent Proteomics with Split-BioID 313

800 rpm for 15 min on a thermomixer, then immediately

remove the beads on the magnetic rack and transfer the eluted
material in a fresh 1.5-mL tube.

13. Eluted samples can be stored at 20 C until further
processing.

3.5 SDS-PAGE 1. Prior to sample preparation for MS analysis, assess the success
and Sample of the split-BioID experiment by Western blot. To do so,
Preparation for MS prepare input and elution blot samples as follows: equal protein
Analysis amounts of input samples mixed with 5 SDS loading buffer in
a total volume of 28 μL (load 25 μL per lane), 5 μL of each
elution sample mixed with 1.25 μL of 5 SDS loading buffer
(load 6 μL per lane). Western blot analysis is performed as
described in Subheading 3.2, steps 10–18 (Fig. 4a).
2. To prepare sample for MS analysis, we recommend using com-
mercial precast gels and reagents (see Note 10). Prepare sam-
ples for SDS-PAGE as follows: 18.75 μL elution samples mixed
with 4 sample buffer. Load the samples on a 4–20%
SDS-polyacrylamide gel and run the electrophoresis until the
running front migrate 2–4 cm into the gel.

Fig. 4 Typical Western blot and protein gel for a split-BioID experiment. (a) Biotinylated proteins are shown for
input and eluted samples from a streptavidin pulldown experiment. GIGYF2 and 4EHP are direct binding
partners. Ago2 and 4EHP do not interact directly and serve as a negative control pair. (b) The elution sample of
a split-BioID experiment was run on a protein gel. The region to be excised and processed for MS analysis is
indicated. (Panel (b) is adapted from Fig. 6 of Schopp and Béthune 2018 [15] under a creative commons
license(CRC 3.0, by,nc,nd)). Asterisks indicate endogenous biotinylated proteins
314 Cinthia Amaya Ramirez et al.

3. Stain the gel with colloidal Coomassie Brillant Blue G250

staining as follows (see Note 26): wash the gel three times for
10 min with ultrapure water in a 15-cm bacterial dish. Incubate
the gel with Coomassie staining solution for 2–12 h (until
bands for streptavidin and some biotinylated proteins are
clearly visible) at RT. Optionally destain for 1 h at RT with
destaining solution. Rinse the gel twice briefly with ultrapure
water.
4. For each sample, excise the whole lane with a clean scalpel,
excluding the streptavidin band (typically the main visible band
running at approximately 17 kDa), and transfer to 1.5 mL tube
(Fig. 4b).
5. The excised lanes can be sent to a proteomics facility for in-gel
trypsin digestion and MS analysis (see Note 27).

4 Notes

1. We have tested split-BioID in HeLa cells, in principle it should

work in any mammalian cell type and more generally in any cell
type in which BioID has been demonstrated to work.
2. Any standard cell culture medium appropriate for the chosen
cell line is fine. Most media will already contain biotin as it is an
essential vitamin. However, split-BioID like BioID requires the
addition of an excess (50 μM) of exogenous biotin to work.
3. Any transfection reagent known to promote high transfection
efficiency and low toxicity for the chosen cell line is fine.
4. Biotin has a good solubility in ammonium hydroxide or
NaOH. To make a stock solution of biotin, dissolve it at
50 mg mL 1 (corresponds to approximately 200 mM) in 2 M
ammonium hydroxide or 2 M NaOH. Once biotin is
completely dissolved, dilute this solution to 50 mM biotin in
500 mM Hepes, pH 7.4, then adjust the pH to 7.4 with HCl.
Sterilize by passing through a 0.22-μM filter, then prepare
aliquots in 1.5-mL safe lock tubes and store the resulting
1000 stock solution at 20  C. We have used up to 2-year-
old stock solutions prepared and stored this way without appar-
ent loss of activity.
5. Doxycycline can be dissolved at 10 mg mL 1 in 70% ethanol
and stored in a screw cap microtube at 20  C in the dark. The
high percentage alcohol ensures that the stock solution is ster-
ile. We have used up to 1-year-old stock solutions prepared and
stored this way without apparent loss of activity.
6. We routinely perform streptavidin pulldowns with magnetic
Dynabeads MyOne Streptavidin C1 beads. Many other types
of beads, including sepharose beads [5] and NeutrAvidin- [7]
Context-Specific Proximity-Dependent Proteomics with Split-BioID 315

or StrepTactin- [9]coupled beads, have also been used success-

fully by others. Some beads are provided as pre-coated with
BSA to decrease nonspecific protein binding. We recommend
not using pre-coated beads, as BSA will later be a major source
of contamination for the MS analysis.
7. We usually prepare lysis and wash buffers freshly from stock
solutions (1 M Tris–HCl, pH 7.4 or 8, 5 M NaCl, 500 mM
EDTA pH 8, 20% SDS, 20% Triton X-100, 20% NP-40, 10%
sodium deoxycholate, 5 M LiCl, 1 M DTT).
8. Low-fluorescence PVDF membranes are required for
fluorescence-based detection of IR dyes-conjugated secondary
antibody. From all the PVDF membranes we tested, only
Immobilon®-FL PVDF membrane from Millipore gave satis-
factory results. Alternatively nitrocellulose membranes typically
yield low autofluorescence. If using chemoluminescence-based
detection, any immunoblotting membrane is fine. Detection of
biotinylated proteins with IR dye- or HRP-conjugated strepta-
vidin works equally well.
9. Ultrafast transfer protocols (7 min) allowed by the Transblot
Turbo system from Bio-Rad or equivalent devices is sufficient
for the detection of biotinylated proteins with conjugated
streptavidin. Any other transfer method is also fine and may
yield better results when the detection of the fusion proteins is
needed, especially for large-size proteins.
10. The use of precast gels and commercial reagents such as load-
ing buffer is optional but recommended to avoid excessive
keratin contamination when running SDS-PAGE for MS
samples.
11. We have provided split-BioID plasmids (see https://www.
addgene.org/Julien_Bethune/) that allow the bidirectional
expression of the NBirA* and CBirA* fusions in various orien-
tations and under the control of the same tet-responsive ele-
ment [8]. Of important note, these plasmids should be used in
tet-on or tet-off cell lines. The two fusion proteins can be
expressed from the same plasmid or from two separate plas-
mids. When testing a pair of protein for the first time, we
recommend trying all possible iterations of the fusion proteins
(NBirA* and CBirA* appended at the N- or C- terminus of
each protein of interest, and protein 1 appended to NBirA* or
CBirA*). We did observe that certain combination work sig-
nificantly better than others.
12. If later many of the resulting isolated plasmids prove to be the
wrong constructs, a dephosphorylation can be included. Typi-
cally we digest the plasmids as indicated, and after 1 h, add
alkaline phosphatase buffer and enzyme and let the reaction
316 Cinthia Amaya Ramirez et al.

proceed for another hour at 37  C. Alkaline phosphate and

restriction enzymes are removed in the following gel
extraction step.
13. When using ampicillin selection, the 1 h incubation step at
37  C can be omitted and the cells directly spread on selection
plates.
14. These numbers are indicative and must be adapted to each cell
line, including HeLa cells from different lab origin.
15. As a negative control, we recommend pairing each the two
tested proteins with an unrelated protein. Whereas using
NBirA*-GFP as an unrelated protein worked fine for us, we
recommend not using CBirA*-GFP as an additional control
protein as it always yielded considerable background biotinyla-
tion in our hands. This is likely due to the very high expression
level of CBirA*-GFP that probably leads to spontaneous
re-association of CBirA* with NBirA* [8]. Other unrelated
CBirA* fusions worked fine for us. A powerful negative control
for MS analysis is a split-BioID or BioID experiment (or a pool
of experiments) applied to two interacting proteins that are
unrelated to the proteins of interest.
16. Doxycycline is necessary if using tet-responsive element–regu-
lated plasmids in tet-on cell lines.
17. Samples may be stored at 20  C at that point.
18. If using fluorescence-based detection and Tween-20 as a
source of autofluorescence, a fourth short wash of the mem-
brane with plain PBS is recommended before detection.
19. We had good experience with the 9E10 (dilution 1:1000) and
M2 (dilution 1:500) mouse monoclonal antibodies directed
against myc and FLAG, respectively.
20. If detection of the fusion proteins proves to be challenging,
incubation of the primary antibodies overnight at 4  C may
yield stronger signals.
21. Cell lysates can be very viscous. If needed the syringe can be
filled the first few times without the needle.
22. These parameters worked well for a Bioruptor plus sonification
device from Diagenode and are given as guidance when using
another sonication bath.
23. We have tended to observe less efficient binding to the
streptavidin-coupled beads at much higher salt concentrations.
24. These numbers are for Dynabeads MyOne Streptavidin C1
beads and may need to be adjusted when using other types of
beads.
25. We chose to elute biotinylated proteins from the beads rather
than performing on-bead trypsin digestion. The latter is the
 Context-Specific Proximity-Dependent Proteomics with Split-BioID 317

most widely used in published BioID experiments but diges-

tion of coupled streptavidin, which is in large excess to the
captured biotinylated proteins, yields a large amount of Strep-
tavidin peptides that might affect the sensitivity of the MS
analysis. On the other hand, elution of biotinylated proteins
requires very harsh denaturing conditions that induce signifi-
cant release of streptavidin from the beads. We address this by
running the eluted sample on a gel and cutting the lane above
the streptavidin bands (running at 17 kDa). Whereas this pro-
cedure allows minimizing streptavidin contamination it has the
disadvantage of excluding small molecular weight proteins
(MW < approximately 20 kDa) from the MS analysis.
StrepTactin-coupled beads were described to allow efficient
capture of biotinylated proteins and elution in mild conditions
using an excess of free biotin [9, 10]. This procedure may thus
be the method of choice to minimize streptavidin contamina-
tion and allow analyzing the full range of captured biotinylated
proteins.
26. In our hands this procedure that has been previously described
by Dyballa and Metzger [11] consistently produces very sensi-
tive staining with a fast and easy protocol. Other sensitive and
MS-compatible staining procedures can also be applied.
27. Standard MS analysis protocols similar to those used to analyze
sample from AP approaches are fine. More elaborate protocols
that also aim at enriching and identifying biotinylated peptides
were also described [9, 12–14].

Acknowledgments

This work was supported by the excellence initiative by the German

federal and state governments: DFG-EXC81, cluster of excellence
CellNetworks.

References
1. Scott DE, Bayly AR, Abell C et al (2016) Small mammalian cells. J Cell Biol 196(6):801–810.
molecules, big targets: drug discovery faces the https://doi.org/10.1083/jcb.201112098
protein-protein interaction challenge. Nat Rev 4. Kim DI, Birendra KC, Zhu W et al (2014)
Drug Discov 15(8):533–550. https://doi. Probing nuclear pore complex architecture
org/10.1038/nrd.2016.29 with proximity-dependent biotinylation. Proc
2. Lam SS, Martell JD, Kamer KJ et al (2015) Natl Acad Sci U S A 111(24):E2453–E2461.
Directed evolution of APEX2 for electron https://doi.org/10.1073/pnas.1406459111
microscopy and proximity labeling. Nat Meth- 5. Lambert JP, Tucholska M, Go C et al (2015)
ods 12(1):51–54. https://doi.org/10.1038/ Proximity biotinylation and affinity purification
nmeth.3179 are complementary approaches for the interac-
3. Roux KJ, Kim DI, Raida M et al (2012) A tome mapping of chromatin-associated protein
promiscuous biotin ligase fusion protein iden- complexes. J Proteome 118:81–94. https://
tifies proximal and interacting proteins in doi.org/10.1016/j.jprot.2014.09.011
318 Cinthia Amaya Ramirez et al.

6. Morriswood B, Havlicek K, Demmel L et al interactions and subcellular localizations. Nat

(2013) Novel bilobe components in Trypano- Commun 9(1):1188. https://doi.org/10.
soma brucei identified using proximity- 1038/s41467-018-03523-2
dependent biotinylation. Eukaryot Cell 12 11. Dyballa N, Metzger S (2009) Fast and sensitive
(2):356–367. https://doi.org/10.1128/EC. colloidal coomassie G-250 staining for proteins
00326-12 in polyacrylamide gels. J Vis Exp 30. https://
7. De Munter S, Gornemann J, Derua R et al doi.org/10.3791/1431
(2017) Split-BioID: a proximity biotinylation 12. Kim DI, Cutler JA, Na CH et al (2018) Bio-
assay for dimerization-dependent protein inter- SITe: a method for direct detection and quan-
actions. FEBS Lett 591(2):415–424. https:// titation of site-specific biotinylation. J
doi.org/10.1002/1873-3468.12548 Proteome Res 17(2):759–769. https://doi.
8. Schopp IM, Amaya Ramirez CC et al (2017) org/10.1021/acs.jproteome.7b00775
Split-BioID a conditional proteomics approach 13. Mackmull MT, Klaus B, Heinze I et al (2017)
to monitor the composition of spatiotempo- Landscape of nuclear transport receptor cargo
rally defined protein complexes. Nat Commun specificity. Mol Syst Biol 13(12):962. https://
8:15690. https://doi.org/10.1038/ doi.org/10.15252/msb.20177608
ncomms15690 14. Udeshi ND, Pedram K, Svinkina T et al (2017)
9. Opitz N, Schmitt K, Hofer-Pretz V et al (2017) Antibodies to biotin enable large-scale detec-
Capturing the Asc1p/receptor for activated C tion of biotinylation sites on proteins. Nat
kinase 1 (RACK1) microenvironment at the Methods 14(12):1167–1170. https://doi.
head region of the 40S ribosome with quanti- org/10.1038/nmeth.4465
tative BioID in yeast. Mol Cell Proteomics 16 15. Schopp IM, Béthune J (2018) Split-BioID—
(12):2199–2218. https://doi.org/10.1074/ proteomic analysis of context-specific protein
mcp.M116.066654 complexes in their native cellular environment.
10. Liu X, Salokas K, Tamene F et al (2018) An J Vis Exp 134:57479. https://doi.org/10.
AP-MS- and BioID-compatible MAC-tag 3791/57479
enables comprehensive mapping of protein
 Chapter 18

A Directed Evolution System for Lysine Deacetylases

Martin Spinck, Maria Ecke, Damian Schiller, and Heinz Neumann

Abstract
Lysine acetylation is a ubiquitous modification permeating the proteomes of organisms from all domains of
life. Lysine deacetylases (KDACs) reverse this modification by following two fundamentally different
enzymatic mechanisms, which differ mainly by the need for NAD+ as stoichiometric co-substrate.
KDACs are often found as catalytic subunit in protein complexes involved in cell cycle regulation,
chromatin organization and transcription. Their promiscuity with respect to sequence context and type
of lysine acylation convolutes the network of functional and physical connections.
Here we present an efficient selection method for KDACs in E. coli, which allows for the creation of acyl-
type specific KDAC variants, which greatly facilitate the investigation of their physiological function. The
selection system builds on the incorporation of acylated lysines by genetic code expansion in reporter
enzymes with essential lysine residues. We describe the creation of KDAC mutant libraries by saturation
mutagenesis of active site residues, the isolation of individual mutants from this library using the selection
system, and their biochemical characterization with acylated firefly luciferase.

Key words Genetic code expansion, Non-canonical amino acids, Lysine deacetylases, Directed
evolution

1 Introduction

Lysine acetylation was first discovered on histone proteins and

linked to transcriptional regulation by neutralizing the positive
charge of lysine, thereby loosening the DNA–histone interaction
[1]. The reversible nature of this process was first suggested in 1977
due to a spike in histone acetylation observed upon n-butyrate
treatment [2]. Twenty years later, the responsible enzyme, histone
deacetylase 1 (HDAC1), was isolated by affinity chromatography,
and HDAC2 was independently identified by sequence homology
to yeast Rpd3 as a transcription factor [3, 4]. This elicited interest in
this enzyme family and led to the discovery of 16 further human
KDACs [5].
KDACs are divided into two families (histone deacetylases and
sirtuins) and four classes (I–IV) based on their reaction mechanism
and sequence homology. Classes I, II, and IV consist of histone

Arnaud Poterszman (ed.), Multiprotein Complexes: Methods and Protocols, Methods in Molecular Biology, vol. 2247,
https://doi.org/10.1007/978-1-0716-1126-5_18, © Springer Science+Business Media, LLC, part of Springer Nature 2021

319
320 Martin Spinck et al.

deacetylases that use a catalytic zinc ion to hydrolyze the peptide

bond, while class III enzymes, the sirtuins, use NAD+ in a unique
mechanism for cleavage [6, 7]. While histones were the first
reported substrates of KDACs, thousands of other proteins and
several new modifications have now been identified in mammals
[8], plants [9, 10], fungi [11] and bacteria [12]. Among the
identified proteins are important cellular components, for example,
p53 as substrate of Sirt1 [13] and HDAC1 [14] and α-tubulin of
HDAC6 [15].
In addition to the large number of protein substrates, it was
found that various lysine modifications (butyryl, crotonyl, malonyl,
etc.) can be reversed by KDACs [8]. While the function of many of
these modifications is still unclear, it has been shown that they
possess their own set of effector proteins such as readers [16],
writers [17], and eraser [18].
Due to the promiscuous activity of KDACs to various sub-
strates, regulation plays an important role, which in the case of
KDACs often means that they are part of larger protein complexes,
for example the transcriptional regulators CoREST [19], mSIN3
[20], NuRD [21], or NCoR/SMRT [22]. While most substrate
proteins have long been known, the role of individual modifications
and their connection to particular KDACs are still unclear. Pro-
blems arise from the high substrate overlap of the different KDACs,
the vast number of modified proteins and the questionable signifi-
cance of many acylations, since the modification reaction can also
take place non-enzymatically on proteins [23].
Here we describe a powerful bacterial selection system to iso-
late KDACs by their activity and selectivity. Using genetic code
expansion, we create an acylation-responsive bifunctional auxotro-
phic marker, Ura3 K93ac, to couple cell growth to KDAC activity.
Applying the same general concept, we develop a straightforward
luciferase assay to detect various deacylase activities of KDACs. The
created system is exemplified by the identification and characteriza-
tion of acyl-type selective E. coli CobB variants from a library of
30 million mutants. This approach allows access to various new
biochemical tools, for example, KDACs with specific deacylase
activity. This may open new, systematic approaches to deconvolute
the complex interactions of KDACs and lysine acylation.

2 Materials

2.1 Strains
and Plasmids (Table 1)

2.2 Creation of KDAC 1. Expand High Fidelity Polymerase (11732650001, 3.5 U μL
1,
Mutant Libraries by Sigma Aldrich).
Inverse PCR 2. 3 M NaOAc pH 5.2 (ITW Reagents).
Table 1
Strains and plasmids

Cell line Manufacturer Genotype

Efficiency™ DH10B™ Thermo fisher Scientific F
mcrA Δ(mrr-hsdRMS-mcrBC) φ80lacZΔM15 ΔlacX74 recA1 endA1
competent cells araD139 Δ(ara, leu)7697 galU galK λ
rpsL nupG/pMON14272/pMON7124
DH10B ΔcobB ΔpyrF Spinck et al. 2020 [24] [DH10B] pyrF cobB
BL21 (DE3) Novagen fhuA2 [lon] ompT gal (λ DE3) [dcm] ΔhsdS λ DE3 ¼ λ sBamHIo ΔEcoRI-B
int::(lacI::PlacUV5::T7 gene1) i21 Δnin5
pPylT-URA3 Spinck et al. 2020 [24] scURA3 cloned in pPylT
pPylT-URA3K93TAG-PylS Spinck et al. 2020 [24] PylS cloned cistronically in pPylT-URA3K93TAG
pPylT-URA3K93TAG-AcKRS3 Spinck et al. 2020 [24] pPylT-URA3K93TAG-AcKRS3
pBK-His6-CobB Spinck et al. 2018 [25] pBK-His6-CobB
pFluc Spinck et al. 2018 [25] FlucK529TAG-His6 cloned in pCDF-PylT
pBK-AcKRS3opt Spinck et al. 2018 [25] Encodes AcKRS3 with mutations for improved PylT binding
pBK-PylS Neumann et al. 2008 [26] Encodes M. barkeri pyrrolysyl-tRNA synthetase
A Directed Evolution System for Lysine Deacetylases
321
322 Martin Spinck et al.

3. 70% ethanol (Thermo Fisher Scientific).

4. 100% ethanol.
5. Primers (100 μM, PAGE purified, Sigma Aldrich, Note 1).
6. dNTPs (10 mM, Thermo Fisher Scientific).
7. QIA Quick PCR Purification Kit.
8. Bsa I-HF (20 U μL
1, NEB).
9. Dpn I (20 U μL
1, NEB).
10. T4 DNA Ligase (5 U μL
1, Thermo Fisher Scientific).
11. 10 T4-Ligase-Buffer (Thermo Fisher Scientific).
12. Hind III (10 U μL
1, Thermo Fisher Scientific).
13. Electroporation Cuvette (2 mm, sterile, VWR).
14. Recovery Medium (Invitrogen).
15. Thermocycler (SensoQuest).
16. Centrifuge 5424R (Eppendorf).
17. Vortex Genie 2 (Scientific Industries).
18. Thermomixer comfort (Eppendorf).
19. NanoDrop (ND-1000, Peqlab).
20. Electroporation System GenePulser Xcell (BioRad).

2.3 Selection 1. M9 Medium: 42.5 g/l Na2HPO4·2H2O, 15.0 g/l KH2PO4,

of Lysine Deacetylases 2.5 g/l NaCl, 5.0 g/l NH4Cl, 1 mM MgSO4, 0.4% glucose,
0.1 mM CaCl2 in sterile ddH2O.
2. LB medium: 1% tryptone, 0.5% yeast extract, 1% NaCl.
3. LB agar plates: LB medium, 2% agar, 15 μg/mL tetracycline
(tet), and 50 μg/mL kanamycin (kan).
4. Positive selection agar plates: M9 medium, 2% agar, 0.2% arab-
inose, 1% glycerol, 4.1 mg/mL tryptophan, 16 mg/mL valine,
16 mg/mL leucine, 16 mg/mL isoleucine, 50 μg/mL kan,
15 μg/mL tet, 10 mM N(ε)-acetyl-L-lysine (or 1 mM of any
other modified lysine).
5. Negative selection agar plates: M9 minimal medium, 2% agar,
0.2% arabinose, 1% glycerol, 4.1 mg/L tryptophan, 16 mg/L
valine, 16 mg/L leucine, 16 mg/L isoleucine, 50 μg/mL kan,
15 μg/mL tet, 10 mM N(ε)-acetyl-L-lysine (or 1 mM of any
other modified lysine), 0.1% 5-fluoroorotic acid, 0.1 mM
uracil.
6. Phosphate-buffered saline (PBS): 137 mM NaCl, 2.7 mM KCl,
10 mM Na2HPO4, 1.8 mM KH2PO4.
7. Petri dishes (150  20 mm, Greiner Bio-One).
8. 1 M in H2O. H-Lys(Ac)-OH (BACHEM).
9. 50 mg/mL kanamycin (Applichem).
 A Directed Evolution System for Lysine Deacetylases 323

10. 15 mg/mL tetracyclin (Sigma Aldrich).

11. 10% in DMSO 5-Fluoroorotic acid (Zymo Research).
12. 25 mM in H2O Uracil (Applichem).
13. 96 deep well block (AXYGEN).
14. 10 mm cuvettes (Sarstedt).
15. Plasmid Isolation Kit (Roti-Prep Plasmid MINI, ROTH).
16. Incubator IPP55 (Memmert).
17. Multichannel pipette (10–100 μL Research plus, Eppendorf).
18. Photometer (Biowave CO8000 Cell Density Meter, WPA).
19. Centrifuge 5810 R (Eppendorf).

2.4 KDAC Assay 1. 20% L(+)-Arabinose (Roth).

Based on Acylated 2. 50 mg/mL spectinomycin.
Firefly Luciferase
3. 1 M IPTG (Roth).
4. 2 M Nicotinamide (Sigma Aldrich).
5. Pierce™ Protease Inhibitor Mini Tablets, EDTA Free (Thermo
Fisher Scientific).
6. 200 mM Phenylmethylsulfonylfluorid (PMSF, Roth) in
isopropanol.
7. 1 M Dithiothreitol (DTT, AppliChem).
8. Lysozyme (SERVA).
9. 1 mg per purification DNase I (Roche).
10. HisPur™ Ni-NTA Resin (Thermo Fisher Scientific).
11. CobB wash buffer: 20 mM HEPES, pH 7.5, 200 mM NaCl,
20 mM DTT, 1 mM PMSF supplemented with lysozyme
(0.1 mg/mL), and DNAse (~0.1 mg).
12. Elution buffer (CobB wash buffer supplemented with 200 mM
imidazole).
13. Gel filtration buffer (20 mM HEPES, pH 7.5, 100 mM NaCl,
10 mM DTT).
14. KDAC reaction buffer: 25 mM Tris, pH 8, 137 mM NaCl,
2.7 mM KCl, 1 mM MgCl2, 1 mM DTT, 1 mg/mL BSA.
15. 2 Luciferase Buffer: 40 mM Tricine, pH 7.8, 200 μM EDTA,
7.4 mM MgSO4, 2 mM NaHCO3, 34 mM DTT, 0.5 mM
ATP, 0.5 mM luciferin (Thermo Fisher Scientific).
16. 96-well PCR plates (clear, semi-skirted, Thermo Fisher
Scientific).
17. 80 mM NAD+, in 200 mM Tris pH 8 (Sigma Aldrich).
18. 96-well plate (Microfluor® 2, black, flat bottom, Thermo
Fisher Scientific).
324 Martin Spinck et al.

19. FLuc wash buffer: 20 mM Tris–HCl, 10 mM imidazole,

200 mM NaCl, 10 mM DTT, 2 mM PMSF, 0.5 Roche
Protease Inhibitor cocktail, 20 mM nicotinamide, pH 8.0,
0.1 mg/mL Lysozyme, DNase I.
20. FLuc-store buffer: 20 mM Tris pH 8, 50 mM NaCl,
10 mM DTT.
21. Microfluidizer (110S, Hyland SCIENTIFIC).
22. Poly-Prep® Chromatography Columns (9 cm, 2 mL bed vol-
ume, Bio-Rad).
23. Amicon Ultra-15 (10 kDa, 30 kDa, Merck).
24. Superdex 200 column (HiLoad 26/600, GE Healthcare).
25. NanoDrop (ND-1000, Peqlab).
26. Microtiter plate reader (FLUOstar Omega, BMG Labtech).

3 Methods

The following three sections outline the necessary steps for the
creation, selection, and characterization of a KDAC mutant library.
We are using E. coli KDAC CobB as an example. CobB is a multi-
functional enzyme able to reverse various lysine modifications such
as acetylation, crotonylation, propionylation or butyrylation. In
this example, we decided to remove its decrotonylation activity by
directed evolution.

3.1 Creation of KDAC The active site residues of sirtuins were randomized by three rounds
Mutant Libraries by of enzymatic inverse PCR [27]. Positions in this example were
Inverse PCR chosen based on the crystal structure of CobB (pdb-file: 1S5P
[28]) close to the bound acetyl-lysine head group (~7 Å radius,
see Fig. 1).

3.1.1 Inverse PCR Primers contain randomized bases (NNK) at the sites of interest
and recognition sites for Bsa I-HF to allow scar less ligation for
plasmid circularization (Fig. 2).
1. Set up several PCRs in parallel as indicated in Table 2 (see Notes
2 and 3):
2. Run PCR according to the program detailed in Table 3.
3. Add 1 μL of Dpn I (20 U μL
1, NEB) to each 50 μL PCR.
Incubate for 1 h, 37  C.
4. Purify PCR product by QIA Quick PCR Purification Kit or by
Gel purification kit (see Fig. 3).
5. Bsa I-HF digestion of the PCR products for 2 h at 37  C
(Table 4).
6. Purify PCR product by QIA Quick PCR Purification Kit and
elute in 50 μL.
 A Directed Evolution System for Lysine Deacetylases 325

Fig. 1 Active site of the CobB crystal structure (1S5P [28]) with bound peptide.
The acetyl-lysine (AcK) and NAD+ (placed with chimera from 2H4F [29]) are
shown, the selected atom to place the sphere for the selection of the active site
residues is indicated (black circle). Five amino acids (orange) were chosen
because of their distance to the indicated atom and to the NAD+. The codons
of all five amino acids were replaced by NNK codons in three rounds of
mutagenesis

3.1.2 Ligation 1. Digested PCR product is pooled and ligated overnight at 16  C

(see Note 4).

Ligation(scale to adjust
DNA concentration <10 ng/μl)
ddH2O 325 μL
Ligation buffer (10) 50 μL
Digested PCR product 100 μL

1
T4 ligase (5 U μL ) 25 μL
Total volume 500 μL

2. Precipitate ligation reaction with ethanol by addition of 1/10

vol. of 3 M NaOAc, pH 5.2–500 μL of ligation reaction and
2 vol. of 100% ethanol.
3. Let the solution stand on ice for 15 min.
4. Centrifuge at 19,000  g for 45 min at 4  C (see Note 5).
5. Remove the supernatant.
6. Add 1 mL of 70% ethanol.
7. Vortex shortly and let stand on ice for 10 min.
8. Centrifuge at 19,000  g for 45 min at 4  C.
9. Repeat the last two steps one more time.
10. Remove the ethanol.
326 Martin Spinck et al.

Fig. 2 Workflow of library creation. Primers are designed to overlap with 4 bp

(green box) flanked by Bsa I sites (dark red box) on the 50 -side, so that the
overlap region will become single-stranded DNA after digestion. Several NNK
codons (red cross) can be introduced by a single primer pair. Amplifying the
target plasmid in a PCR yields a linear product (black) containing the mutations,
the Bsa I site on both sides of the duplicated 4 bp overlap. The template is
destroyed using Dpn I, single-strand overhangs created by Bsa I digest and the
product ligated

11. Evaporate leftover ethanol (37  C or RT) (see Note 6).

12. Dissolve all pellets in 10–40 μL ddH2O (pooling all ligation
reactions).
(a) First, pipette the droplet above the DNA pellet (the pellet
may be invisible).
 A Directed Evolution System for Lysine Deacetylases 327

Table 2
Composition of the inverse PCR mix

ddH2O 41 μL
Buffer expand high fidelity (10) 5 μL
Volume of each primer (100 μM) 0.5 μL
dNTP mix (10 mM) 1 μL

1
Template (~400 ng μL ) 1 μL

1
Expand polymerase (3.5 U μL ) 1 μL
Total volume 50 μL

Table 3
Cycles for the inverse PCR

Time (min:sec) Temp. # Passes


Denat. 1:00 95 C 1
Denat. 0:20 95  C 10
Anneal. 0:30 65  C 10
1  C/cycle
Elong. 5:00 68  C 10

Denat. 0:20 95 C 25
Anneal. 0:30 55  C 25

Elong. 5:00 68 C 25

(b) Then slowly pull down the droplet with the pipette tip.
(c) When the round droplet touches the DNA pellet, it will
spread and run to the bottom of the tube slowly.
(d) Repeat this process several times until there is no DNA left
(droplet stays round).
(e) Transfer the solution to the next tube.
13. Determine DNA concentration by NanoDrop.
14. Test ligation efficiency by Hind III digest of 100 ng of DNA in
a 10-μL reaction (see Note 7 and Fig. 3).

3.1.3 Electroporation 1. E. coli MAX Efficiency™ DH10B™ Competent Cells (Invitro-

gen) are thawed on ice.
2. 4 μL ligated PCR product are mixed with 40 μL Competent
Cells in a 1.5-mL reaction tube.
3. Incubate 1 min on ice.
4. Transfer 40 μL to the bottom of a 0.2 cm electroporation
cuvette (BioRad).
328 Martin Spinck et al.

Fig. 3 Examples for each step are from the creation of various libraries. (a) Example for a primer pair,
important sequence features are highlighted. (b) Examples for PCR products ranked from suitable to
unsuitable for subsequent steps. Troubleshooting recommendations are indicated. Start a library with the
least efficient PCR, because the required number of transformants increases while performance usually
declines in successive rounds of mutagenesis. (c) Efficiency of the ligation should be tested before
electroporation. A circular vector will give a single band (1) when digested with a single cutting restriction
enzyme, if the ligation is incomplete, it will result in two extra bands with different intensities depending on the
ligation efficiency (2, 3). No ligation results in just a double band (4). Reason for incomplete ligation can be the
presence of side products capping the linear DNA (compare PCRs labeled # and * in b). Transformation of 3 or
4 did not yield any or sufficient numbers of colonies, while transformation of 2 was still sufficient
 A Directed Evolution System for Lysine Deacetylases 329

Table 4
Restriction digest of PCR product (scale to match volume of PCR product)

PCR product 30 μL
ddH2O 11 μL
Cut smart buffer (10) 5 μL
Bsa I-HF (20 U μL
1) 4 μL
Total volume 50 μL

5. Dry the cuvette with a paper towel and place it in the electro-
poration system.
6. Electroporate using E. coli electroporation settings: 2500 V,
25 μF, 200 Ω.
7. Immediately add 1 mL recovery medium (RT).
8. Transfer the reaction into a 1.5 mL reaction tube.
9. Incubate at 37  C for 1 h.
10. Pool the transformations in a volume of 250 mL LB medium
containing 50 μg/mL kan and plate a dilution series (10
4 to
10
7) on LB-kan plates. Incubate both overnight at 37  C, the
flask shaking at 200 rpm.
11. Estimate transformation efficiency from the colony count of
the dilution series.
12. Isolate plasmid DNA from the pool and from at least 12 indi-
vidual clones and sequence the KDAC ORF.
13. If transformation efficiency exceeds three times the diversity of
the library and the majority of individual sequences are unique
and without frameshifts continue to the next step.
14. Use the plasmid preparation isolated from the pool as template
in the next round of mutagenesis.

3.2 Selection The mutant libraries generated above contain a large number of
of Lysine Deacetylases different mutants due to the numerous possible combinations of
randomized positions (205 different proteins coded for by 325
codon combinations if five positions are mutated). Most of the
mutations will be detrimental to protein function or folding, ren-
dering the mutants inactive. Hence, an efficient selection system is
required to identify the mutants of interest in a large background of
inactive ones. Therefore, we designed a bacterial selection system
that allows parallel screening of each mutant in a single step [24]. We
created an E. coli strain without deacetylase activity and impaired
uracil biosynthesis by inactivating the genes for CobB and PyrF
using CRISPR/Cas9 and Lambda Red Recombineering [30]. We
complemented the uracil auxotrophic strain with Ura3 (orotidine
50 -phosphate decarboxylase), which contains an amber codon in
330 Martin Spinck et al.

Fig. 4 Outline of the bacterial KDAC selection system. (a) Specific mutants can be enriched by transforming
E. coli expressing acylated Ura3 with the library. Each individual cell will then perform the demodification
reaction and only produce active Ura3 if the encoded KDAC mutant has activity for the attached modification.
The active Ura3 can then be used to either remove or enrich the active mutants depending on the selection
plate composition. (b) Phenotyping. DH10B ΔcobB ΔpyrF expressing Ura3 K93ac or K93boc and either CobB or
HDAC8 were grown on positive (–Ura) or negative (+5-FOA/Ura) selection conditions. Both KDACs are able to
cleave AcK and hence grow only under positive selection. The situation is inversed when BocK is incorporated
which neither KDAC can cleave. (c) Example for a positive selection of the CobB active site mutant library in
the presence or absence of AcK or CrK. Approximately 108 cells were plated in each case (library diversity
approximately 3.4  107). Number of colonies gives an estimate of surviving mutants and is used as a
reference for the required transformation efficiency in subsequent selection rounds

place of the codon for an essential lysine in its active site (K93)
[31]. Suppression of this codon with an acylated lysine by genetic
code expansion [26] produces an enzyme that can be activated by
co-expression of a KDAC. This allows for positive (in the absence of
uracil) and negative (in the presence of 5-fluoroorotic acid [32] and
uracil) selection of its activity (Fig. 4).
 A Directed Evolution System for Lysine Deacetylases 331

3.2.1 Preparation 1. Grow an overnight culture of DH10B ΔcobB ΔpyrF pPylT-

of Electrocompetent Cells URA3K93TAG-PylS or DH10B ΔcobB ΔpyrF pPylT-
(E. coli ΔpyrF ΔcobB URA3K93TAG-AcKRS3 (5 mL LB medium containing
pURA3K93TAG-PylS/ 15 μg/mL tet).
AcKRS3) 2. Inoculate 100 mL LB with 1 mL overnight culture and incu-
bate at 37  C for 2.5–3 h until the OD600 reaches 0.3.
3. Harvest the cells by centrifugation at 2500  g for 5 min at
4  C and remove the medium.
4. Suspend the cells with 30–40 mL ice-cold, autoclaved, particle
free ddH2O.
5. Centrifuge at 2000  g for 5 min at 4  C and discard the
supernatant.
6. Repeat steps 4 and 5 once again (see Note 8).
7. Centrifuge at 2000  g, remove remaining water.
8. Suspend cells in 200 μL ddH2O.
9. Keep the cells on ice until further use.

3.2.2 Selection of KDACs 1. Electroporate freshly prepared electrocompetent DH10B

ΔcobB ΔpyrF pPylT-URA3K93TAG-PylS or DH10B ΔcobB
ΔpyrF pPylT-URA3K93TAG-AcKRS3 with 1 μg of the CobB
mutant library.
2. Follow steps 3–7 of Subheading 3.1.3.
3. After the recovery time, add all transformed cells to 100 mL LB
medium containing 50 μg/mL tet and 50 μg/mL kan. Prepare
a small-scale serial dilution down to a factor of 10
7 and plate
them on appropriate LB agar plates. Incubate media and plates
at 37  C overnight.
4. Estimate the number of transformants from the colony count
of plates. It should exceed three times the number of all possi-
ble mutants in the library.
5. Inoculate 50 mL LB containing 50 μg/mL tet and 50 μg/mL
kan with 1 mL of overnight culture and incubate at 37  C until
the OD600 exceeds 0.3 (see Note 9).
6. Harvest the cells by centrifugation at 500  g for 5 min at 4  C
and remove the medium.
7. Resuspend the cells in sterile PBS. For first positive selection,
plate out aliquots of 8  108 cells on each positive selection
agar plate and incubate at 37  C for 2–3 days. For negative
selection and all successive selections, plate aliquots of
8  105 cells on each negative selection plate and incubate at
37  C for 16 h (see Note 10).
8. Add ~3 mL sterile PBS onto the plate and scrape the colonies
from the agar with the help of a sterile cell spreader. Distribute
332 Martin Spinck et al.

the cells into aliquots of 500 μL and isolate the plasmid DNA
using a commercial kit. Pool plasmid DNA for the next round
of selection.
9. Repeat steps 1–8 until all selection rounds are completed.
10. Transform DH10B cells with the plasmid DNA obtained by
the selection via heat shock. Plate the cells on agar plates
containing 50 μg/mL kan and incubate them overnight at
37  C.
11. On the next day, prepare a 96 deep well block by filling each
well with 1 mL LB medium containing 50 μg/mL kan. Inocu-
late each well by picking single colonies of the transformed
DH10B cells. Incubate the block overnight at 37  C.
12. Harvest the cells by centrifuging the 96 well block for 10 min
at 2700  g and remove the media.
13. Isolate the plasmid DNA of each well and sequence the KDAC
ORF (see Note 11).

3.2.3 6-AU Assay This step is optional, but can help filter out the most active variants
(Optional) from a pool of mutants. Therefore, repeat the last selection step of
Subheading 3.2.2 with the addition of 6-azauridine (6-AU) to the
agar plates (0.01–2 mM, at least 10 different concentrations). This
can be done by plating the pool of mutants or by replicating single
colonies from a 96-well plate onto the selection plates. 6-AU is an
inhibitor of Ura3, which requires cells to produce more active
Ura3, increasing the stringency of the positive selection assay.

3.3 KDAC Assay Similar to Ura3, the firefly luciferase (FLuc) also contains an essen-
Based on Acylated tial lysine (K529) in its active site. When modified, the luciferase is
Firefly Luciferase completely inactive and therefore can be used to monitor the
activity of enzymes capable of reversing the modification [25]. We
use this to assay the activity and selectivity of the isolated mutants.

3.3.1 Preparation 1. Transform E. coli BL21 DE3 with plasmids pFluc and
of FLucmod pBK-AcKRS3opt or pBK-PylS depending on the desired
modification.
2. Grow the cells in LB medium (50 μg/mL spectinomycin and
50 μg/mL kan, 5 mM N(ε)-acetyl-lysine or 1 mM other AS
and 20 mM nicotinamide) at 37  C to and OD600 of 1.0.
3. Shift cells to 30  C and induce protein expression by the
addition of 1 mM IPTG.
4. After 4 h at 30  C harvest cells by centrifugation, wash with
PBS, and suspend in Ni-wash buffer.
5. Lyse cells by passing them three times through a microfluidizer.
6. Centrifuge lysate at 20,000  g for 20 min and transfer the
supernatant to a falcon tube.
 A Directed Evolution System for Lysine Deacetylases 333

7. Supplement the supernatant with 500 μL Ni-NTA-beads.

8. Incubate the beads for 2 h with agitation at 4  C.
9. Wash the beads with 50 mL FLuc-wash buffer and elute bound
proteins in 2 mL FLuc-wash buffer supplemented with
200 mM imidazole.
10. Exchange buffer to FLuc-store buffer using a centrifugal filter.
11. Use the Fluc as deacetylase substrate (see Note 12).

3.3.2 Purification 1. Transform E. coli ΔpyrF ΔcobB with pBK-His6-CobB and

of CobB pPylT-Ura3 by a method of choice (see Note 13).
2. Setup 20 mL overnight culture in LB medium (containing
15 μg/mL tet, 50 μg/mL kan) from a single colony.
3. Inoculate 1L LB medium (containing 15 μg/mL tet, 50 μg/
mL kan) with 10 mL overnight culture.
4. Grow to OD 0.3–0.5 at 37  C and 200 rpm.
5. Decrease temperature to 30  C and induce protein expression
by addition of arabinose (0.2% final).
6. Harvest cells by centrifugation (6000  g, 20 min) after 16 h of
expression.
7. Wash the cell pellet with PBS and store at
20  C.
8. Suspend the cell pellet in 25 mL CobB wash buffer supplemen-
ted with lysozyme (0.1 mg/mL) and DNase (~0.1 mg).
9. Lyse cells by passing through a microfluidizer 2–3 times.
10. Pellet cell debris by centrifugation and decant supernatant into
a Falcon tube.
11. Add 500 μL Ni-NTA resin to the supernatant, incubate for 1 h
with agitation.
12. Pour the suspension into a plastic column with frit.
13. Wash with 50 mL CobB wash buffer.
14. Elute protein in 2 mL Elution buffer.
15. Concentrate fractions using a centrifugal filter and exchange
buffer by diluting twice with gel filtration buffer.
16. Apply the concentrated fractions to a Superdex 200 column
equilibrated with gel filtration buffer (2.5 mL/min, 4  C).
17. Collect 4 mL fractions while monitoring protein concentration
at 280 nm.
18. Analyze protein containing fractions by SDS-PAGE, pool, and
concentrate by ultrafiltration.
19. Flash freeze protein in liquid nitrogen and store at
80  C.
334 Martin Spinck et al.

3.3.3 Characterization For testing a large number of mutants for a desired property,
of Isolated Mutants KDACs partially purified by Ni-NTA (Subheading 3.3.2, steps 1–
14) or even cell lysates can be used if mutants can be expressed in an
E. coli ΔcobB strain (e.g., DH10B ΔcobB ΔpyrF). The assay is
exceptionally sensitive, so that even small contaminations with
KDACs such as E. coli CobB will produce high background signals.
All isolated mutants are expressed, and activity for different mod-
ifications is assessed. Eventually, the desired KDACs should be
purified to homogeneity by Ni-NTA affinity and gel filtration chro-
matography for full characterization.
1. Prepare cell lysate (Subheading 3.3.2, steps 1–10), Ni-Eluate
(steps 1–14), or fully purified KDACs (steps 1–18) of all
mutants and a wild-type control.
2. Estimate protein concentration by SDS-PAGE or nano drop.
3. Set up reactions in 96-well PCR plates in triplicate for every
lysine modification as following: 0.2 mg/mL KDAC, 2 mM
NAD+, 10 μL Luciferase eluate (1:125 to 1:8000 dilution,
depending on the signal strength, at least 100-fold over back-
ground) in KDAC reaction buffer to a final volume of 50 μL.
4. Incubate at 30  C for 30 min with 50  C lid temperature in a
thermocycler.
5. Transfer 40 μL of the reaction mixture to a 96-well microtiter
plate.
6. Add 40 μL of 2 Luciferase Buffer.
7. Measure luminescence using a Microplate reader.
8. Normalize activity to the signal of the wild-type.
9. Mutants with enhanced selectivity and/or activity are selected
for further analysis.
10. Perform steps 1–8 with various concentrations of mutant and
wild-type enzyme (between 1 nM and 16 μM).
11. Activity can be compared in the linear region of the obtained
curve for each individual substrate.

4 Notes

1. The primer needs to be as defined as possible. Prime fragments

can lead to PCR artifacts and increase side products. Ultimately
this decreases library quality and increases the effort in
creating it.
2. Total volume is adjusted to required diversity. For example,
4  107 mutants approximately 10–16 PCRs are required,
depending on the yield and ligation efficiency.
 A Directed Evolution System for Lysine Deacetylases 335

3. Side product formation severely decreases the efficiency of

ligation, requiring more PCRs for sufficient DNA amounts
after gel purification.
4. High concentrations of DNA will lead to the formation
of concatemers, scale volume to adjust DNA concentration
<10 ng/μl.
5. Longer centrifugation times improve DNA recovery.
6. Do not overly dry the DNA.
7. More than 50% of the DNA should be circular for efficient
electroporation.
8. If the cells continue to stick to the wall of the tube and do not
get resuspended easily, add one more washing step.
9. The cells tend to grow very slowly at this step. An OD600 of
1 would be ideal, but values as low as 0.3 are also sufficient if
the growth takes too much time.
10. The default procedure for screening is one round of positive
selection in the beginning, then one round of negative selec-
tion and finally one more round of positive selection with the
same conditions as in the first round. If there are only very few
colonies growing on the selection plates, the screening can be
stopped after one or two rounds. In this case, the single colo-
nies can be picked directly for plasmid DNA isolation, and step
7 can be skipped.
11. The 96-well block can be stored at
20  C, and the isolation of
plasmid DNA can be continued on another day.
12. Titrate dilution series to find the optimal working concentra-
tion for your deacetylase.
13. Due to low transformation efficiency, it is recommended to
successively transform the plasmids. A stock of E. coli DH10B
ΔcobB ΔpyrF competent cells with pPylT-Ura3 plasmid can be
stored at
80  C. The pPylT-Ura3 plasmid is required to
provide sufficient AraC protein for the induction of CobB
expression. Since CobB isolates are initially encoded on mini-
mal pBK plasmids, this method avoids an intermediate
recloning.

Acknowledgments

The authors are grateful for the financial support by the

Max-Planck Institute of Molecular Physiology and the German
Research Foundation (DFG) [NE1589/5-1 to H.N.].
336 Martin Spinck et al.

References
1. Allfrey VG, Faulkner R, Mirsky AE (1964) regulatory enzyme CobB in Escherichia coli.
Acetylation and methylation of histones and Mol Cell Proteomics 12:3509–3520. https://
their possible role in the regulation of rna syn- doi.org/10.1074/mcp.M113.031567
thesis. Proc Natl Acad Sci U S A 51:786–794 13. Vaziri H et al (2001) hSIR2(SIRT1) functions
2. Riggs MG, Whittaker RG, Neumann JR, as an NAD-dependent p53 deacetylase. Cell
Ingram VM (1977) n-Butyrate causes histone 107:149–159
modification in HeLa and friend erythroleu- 14. Luo J, Su F, Chen D, Shiloh A, Gu W (2000)
kaemia cells. Nature 268:462. https://doi. Deacetylation of p53 modulates its effect on
org/10.1038/268462a0 cell growth and apoptosis. Nature
3. Taunton J, Hassig CA, Schreiber SL (1996) A 408:377–381. https://doi.org/10.1038/
mammalian histone deacetylase related to the 35042612
yeast transcriptional regulator Rpd3p. Science 15. Hubbert C et al (2002) HDAC6 is a
272:408 microtubule-associated deacetylase. Nature
4. Yang WM, Inouye C, Zeng Y, Bearss D, Seto E 417:455–458. https://doi.org/10.1038/
(1996) Transcriptional repression by YY1 is 417455a
mediated by interaction with a mammalian 16. Andrews FH et al (2016) The Taf14 YEATS
homolog of the yeast global regulator RPD3. domain is a reader of histone crotonylation.
Proc Natl Acad Sci U S A 93:12845–12850 Nat Chem Biol 12:396–398. https://doi.
5. Seto E, Yoshida M (2014) Erasers of histone org/10.1038/nchembio.2065
acetylation: the histone deacetylase enzymes. 17. Sabari BR et al (2015) Intracellular crotonyl-
Cold Spring Harb Perspect Biol 6:a018713. CoA stimulates transcription through p300-
https://doi.org/10.1101/cshperspect. catalyzed histone crotonylation. Mol Cell
a018713 58:203–215. https://doi.org/10.1016/j.
6. Sauve AA (2010) Sirtuin chemical mechan- molcel.2015.02.029
isms. Biochim Biophys Acta 18. Wei W et al (2017) Class I histone deacetylases
1804:1591–1603. https://doi.org/10.1016/ are major histone decrotonylases: evidence for
j.bbapap.2010.01.021 critical and broad function of histone crotony-
7. Lombardi PM, Cole KE, Dowling DP, Chris- lation in transcription. Cell Res 27:898–915.
tianson DW (2011) Structure, mechanism, and https://doi.org/10.1038/cr.2017.68
inhibition of histone deacetylases and related 19. Meier K, Brehm A (2014) Chromatin regula-
metalloenzymes. Curr Opin Struct Biol tion: how complex does it get? Epigenetics
21:735–743. https://doi.org/10.1016/j.sbi. 9:1485–1495. https://doi.org/10.4161/
2011.08.004 15592294.2014.971580
8. Tan M et al (2011) Identification of 67 histone 20. Zhang Y, Iratni R, Erdjument-Bromage H,
marks and histone lysine crotonylation as a new Tempst P, Reinberg D (1997) Histone deace-
type of histone modification. Cell tylases and SAP18, a novel polypeptide, are
146:1016–1028. https://doi.org/10.1016/j. components of a human Sin3 complex. Cell
cell.2011.08.008 89:357–364. https://doi.org/10.1016/
9. Finkemeier I, Laxa M, Miguet L, Howden AJ, S0092-8674(00)80216-0
Sweetlove LJ (2011) Proteins of diverse func- 21. Xue Y et al (1998) NURD, a novel complex
tion and subcellular location are lysine acety- with both ATP-dependent chromatin-remo-
lated in Arabidopsis. Plant Physiol deling and histone deacetylase activities. Mol
155:1779–1790. https://doi.org/10.1104/ Cell 2:851–861
pp.110.171595 22. Wen YD et al (2000) The histone deacetylase-3
10. Wu X et al (2011) Lysine acetylation is a wide- complex contains nuclear receptor corepres-
spread protein modification for diverse proteins sors. Proc Natl Acad Sci U S A 97:7202–7207
in Arabidopsis. Plant Physiol 155:1769–1778. 23. Simithy J et al (2017) Characterization of his-
https://doi.org/10.1104/pp.110.165852 tone acylations links chromatin modifications
11. Henriksen P et al (2012) Proteome-wide anal- with metabolism. Nat Commun 8:1141.
ysis of lysine acetylation suggests its broad reg- https://doi.org/10.1038/s41467-017-
ulatory scope in Saccharomyces cerevisiae. Mol 01384-9
Cell Proteomics 11:1510–1522. https://doi. 24. Spinck M, Neumann-Staubitz P, Ecke M, Gas-
org/10.1074/mcp.M112.017251 per R, Neumann H (2020) Evolved, selective
12. Colak G et al (2013) Identification of lysine erasers of distinct lysine acylations. Angew
succinylation substrates and the succinylation
 A Directed Evolution System for Lysine Deacetylases 337

Chem Int Ed Engl 59:11142–11149. https:// 29. Hoff KG, Avalos JL, Sens K, Wolberger C
doi.org/10.1002/anie.202002899 (2006) Insights into the Sirtuin mechanism
25. Spinck M, Ecke M, Sievers S, Neumann H from ternary complexes containing NAD+
(2018) Highly sensitive lysine deacetylase and acetylated peptide. Structure 14
assay based on acetylated firefly luciferase. Bio- (8):1231–1240
chemistry 57:3552–3555. https://doi.org/ 30. Pyne ME, Moo-Young M, Chung DA, Chou
10.1021/acs.biochem.8b00483 CP (2015) Coupling the CRISPR/Cas9 sys-
26. Neumann H, Peak-Chew SY, Chin JW (2008) tem with lambda red Recombineering enables
Genetically encoding N(epsilon)-acetyllysine in simplified chromosomal gene replacement in
recombinant proteins. Nat Chem Biol Escherichia coli. Appl Environ Microbiol
4:232–234. https://doi.org/10.1038/ 81:5103–5114. https://doi.org/10.1128/
nchembio.73 aem.01248-15
27. Stemmer WP, Morris SK (1992) Enzymatic 31. Appleby TC, Kinsland C, Begley TP, Ealick SE
inverse PCR: a restriction site independent, (2000) The crystal structure and mechanism of
single-fragment method for high-efficiency, orotidine 50 -monophosphate decarboxylase.
site-directed mutagenesis. BioTechniques Proc Natl Acad Sci U S A 97:2005–2010.
13:214–220 https://doi.org/10.1073/pnas.259441296
28. Zhao K, Chai X, Marmorstein R (2004) Struc- 32. Boeke JD, Trueheart J, Natsoulis G, Fink GR
ture and substrate binding properties of cobB, (1987) 5-Fluoroorotic acid as a selective agent
a Sir2 homolog protein deacetylase from in yeast molecular genetics. Methods Enzymol
Escherichia coli. J Mol Biol 337:731–741. 154:164–175
https://doi.org/10.1016/j.jmb.2004.01.060
 INDEX

A Double strand break (DSB) repair ........................ 44, 126

Double-stranded DNA donors ...................................... 40
Affimers................................................................. 105–120 DSTORM ............................................................. 271–284
Analytical ultracentrifugation (AUC) ................ 155, 157,
159, 161, 162, 167 E
Artificial binding proteins.................................... 105–120
EM sample preparation........................................ 244, 255
B Endogenous host .............................................................. 7
Enzymes.............................................................21, 23, 24,
Baculovirus insect cell expression................................... 18 27, 35, 41, 47, 62, 147, 150, 178, 185, 199, 304,
BioID ................................................................... 304, 314,
306–308, 315, 316, 319, 320, 324, 330, 332, 334
316, 317 Excited-state lifetime .................................. 288, 291, 298
Biolayer interferometry (BLI) ..........................59, 61, 62, Expression in bacteria ....................................................... 3
64, 70, 74, 127
Biosensors ................................................................ 61, 65, F
70, 75, 146
Biosimilars ....................................................................... 79 Fab .....................................................................60, 62, 70,
Biotin ................................................................... 111, 146, 74, 75, 77–79, 89–102
304, 306, 309, 311, 314, 317 Fluorescence ................................................. 6, 40, 43, 45,
54, 55, 91, 93, 128, 129, 131, 132, 135, 137,
C 138, 141, 145, 149, 150, 156, 168, 274–276,
279, 288, 291, 296, 307, 309, 310, 315, 316
Chemical-crosslinking....................................75, 259, 260 Fluorescence imaging ............................. 54, 55, 291, 296
Cloning .............................................................. 17–37, 45,
Fluorescence lifetime imaging microscopy
50, 55, 62, 63, 292, 294, 295, 299, 307, 308 (FLIM)..............................................291, 295–298
Complex purification ........................................... 178, 303 Fluorescence microscopy ....................................... 52, 271
Computational modelling .................................. 159, 222, Fluorescent proteins..........................................6, 89, 101,
223, 233, 238
272, 291, 292, 296
Confocal microscopy ........................................... 295, 296 Förster resonance energy transfer
CRISPR/Cas9.................................................. 39–56, 328 (FRET)..............................................288–291, 296
Cryo-electron microscopy (cryo-EM) ................. 78, 223,
Functions ............................................................. 3, 39, 55,
224, 230, 244, 246–248, 253, 255, 271 60, 89, 91, 105, 131, 157, 160, 179, 207–209,
Crystallography ..................................................... 77, 174, 230, 232–234, 238, 243, 257, 263, 271, 278,
223, 226
279, 281, 283, 287, 303, 320, 328
Fv fragments .................................................77–79, 84, 89
D
Detergents .........................................................6, 7, 9, 12, G
13, 15, 137, 141, 188, 190, 215, 252, 253, 255, Genetically encoded fluorescent antibody ..................... 78
257, 258, 262, 263, 266, 267 Genetic code expansion ................................................ 320
Directed evolution ............................................... 319–335
Genome editing .............................................................. 51
DNA ..................................................................18, 20, 21, GFP-tag .............................................................. 43, 45, 52
23–25, 27, 30, 32–37, 39–56, 62–65, 67, 73, 75,
85–88, 91, 93–96, 116, 125–141, 146, 148, 150, H
180, 181, 193–217, 287, 291–295, 306–308,
319, 322, 326, 327, 329, 332, 335 Hetero-association ..............................156, 160, 163, 167
DNA polymerase ...............................................23, 25, 30, HIV reverse transcriptase..................................... 147, 148
34, 42, 145–152 Homology based cloning .........................................34, 55

Arnaud Poterszman (ed.), Multiprotein Complexes: Methods and Protocols, Methods in Molecular Biology, vol. 2247,
https://doi.org/10.1007/978-1-0716-1126-5, © Springer Science+Business Media, LLC, part of Springer Nature 2021

339
 MULTIPROTEIN COMPLEXES: METHODS AND PROTOCOLS
340 Index
Hybrid approaches ........................................................ 126 Protein interactions...........................................18, 39, 61,
Hydrogen/deuterium exchange ......................... 193–217 105, 126, 155, 156, 195, 288, 289, 291, 295,
296, 303
I Protein-protein interaction (PPI) ....................... 303, 304
Proteomics...........................................190, 193, 303–317
IgG................................................ 60, 62, 70, 73, 75, 101
Immunodiagnostics ........................................................ 79
R
Immunofluorescence .................................................... 288
Inhibitor .............................................................. 9, 11, 12, Recombinant antibodies ........................... 60, 66, 77–102
43, 60, 66, 105, 109, 115, 117, 260, 306, 307, Recombinant antibody formats...................................... 70
323, 324, 332 Research reagents .......................................................... 105
Restriction free (RF) .........................................18, 24, 32,
K 34, 35, 184, 186, 187
Kallikrein-related protease 7 (KLK7)......................60–62,
S
66, 69, 70, 75
Kinetics .............................................................. 59–62, 65, ScFv-Fc ........................................................ 60, 62, 70, 73
70, 73, 75, 138, 140, 145–153, 159, 160, 168, Secretory expression........................................................ 78
217, 275 Sedimentation coefficients.................................. 156, 160,
162, 163, 169
L Sedimentation velocity (SV) ................................... 79, 85,
91, 155–164, 168
Lysine deacetylases ............................................... 319–335
Sequence and ligation independent cloning
M (SLIC) .................................................... 18, 23–25,
30, 32, 34, 55, 62
Mammalian target of rapamycin (mTOR).......... 287, 288 Single-chain variable fragment format
Membrane protein complexes ........................... 3–15, 190 (scFv).............................................. 60, 61, 63, 102
Membrane proteins................................................. 3, 4, 6, Single-molecule localization microscopy
7, 77, 78, 189, 223, 245, 257–267 (SMLM)................................................... 271, 272,
Microcalorimetry........................................................... 130 275, 277–279, 281, 283
Modelling restraints .................................... 221, 224, 235 Single particle electron microscopy ............................. 244
Multiprotein complexes ............................ 17–37, 39, 189 Size exclusion chromatography (SEC) .................. 10, 13,
15, 89, 93, 99, 127, 133, 156, 161, 162, 164,
N
181–183, 259, 266, 267
Native-PAGE................................................259–261, 265 Solubilization .............................................................9, 12,
Negative stain electron microscopy ............................. 245 257–259, 261–263, 266
Non-canonical amino acids .......................................... 319 Stabilization ....................................................84, 257–267
Non-covalent mass spectrometry ............... 173, 174, 181 Stoichiometries..................................................6, 61, 126,
Non-homologous endjoining (NHEJ)............... 126, 127 127, 133, 135, 136, 140, 141, 156, 173, 174,
Nucleo-protein complexes............................................ 125 181, 222, 224, 230, 233, 236, 272, 283
Structural mass spectrometry ............................. 174, 193,
O 194, 209, 221–223, 239
Subcomplexes .............................................. 226, 229, 238
Oligomeric states.................................184, 185, 247, 259
Super-resolution microscopy ............................... 126, 271
P SwitchSENSE technology ................................... 145, 146

Panning....................................... 106, 107, 111–118, 120 T

Phage display ..........................60, 61, 106, 107, 111–116
Therapeutic antibodies .......................................... 79, 100
Phage ELISA ................................................110, 116–120
Thermodynamic parameters ................................ 126, 133
Polyproteins................................................. 78, 79, 89, 93
Thermophoresis ................................................... 125–142
Protein complexes .............................................12, 17, 39,
Three-color.................................................................... 299
59, 126, 155, 181, 188, 189, 223, 224, 230, 246,
Timecorrelated single-photon counting cards
251, 255, 303–317, 320
(TCSPC) ................................................... 295, 296
Protein-DNA binding......................................... 181, 203,
Transcription factor (TF)............................ 181, 195, 210
214, 215
Transient gene expression.........................................59–75
Protein-fragment complementation assay ................... 303
Translocon ..................................................5–8, 11, 13–15
Protein heterogeneity .......................................... 155–169